pgj r ; ( Published by The Palaeontological Association • London Price £38-00 alaeontology VOLUME 41 • PART 5 • SEPTEMBER 1998 THE PALAEONTOLOGICAL ASSOCIATION (Registered Charity No. 276369) The Association was founded in 1957 to promote research in palaeontology and its allied sciences. •COUNCIL 1997-1998 President : Professor E. N. K. Clarkson, Department of Geology and Geophysics, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW Vice-Presidents : Dr R. M. Owens, Department of Geology, National Museum and Gallery of Wales, Cardiff CF1 3NP Dr P. Doyle, Department of Earth Sciences, University of Greenwich, Grenville Building, Pembroke, Chatham Maritime, Kent ME4 4AW Treasurer: Dr T. J. Palmer, Institute of Geography and Earth Sciences, University of Wales, Aberystwyth, Dyfed SY23 3DB Membership Treasurer: Dr M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL Institutional Membership Treasurer : Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ Secretary: Dr M. P. Smith, School of Earth Sciences, University of Birmingham, Birmingham B15 2TT Newsletter Editor : Dr S. Rigby, Department of Geology and Geophysics, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW (co-opted) Newsletter Reporter: Dr P. Pearson, Geology Department, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ Marketing Manager : Dr A. King, English Nature, Northminster House, Peterborough PEI 1UA Publicity Officer : Dr M. A. Purnell, Department of Geology, University of Leicester, University Road, Leicester LEI 7RH Editors Dr D. M. Unwin, Geology Department, University of Bristol, Wills Memorial Building, Queens Road, Bristol BS8 1RJ Dr R. Wood, Department of Earth Sciences, University of Cambridge, Downing Street, Cambridge CB2 3EQ Dr D. A. T. Harper, Geologisk Museum, Copenhagen University, 0ster Voldgade 5-7, 1 350 Copenhagen K, Denmark Dr A. R. Hemsley, Department of Earth Sciences, University of Wales College of Cardiff, Cardiff CF1 3 YE Dr J. Clack, University Museum of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ Dr B. M. Cox, British Geological Survey, Keyworth, Nottingham NG12 5GG Dr D. K. Loydell (Technical Editor), Department of Geology, University of Portsmouth, Burnaby Building, Burnaby Road, Portsmouth POl 3QL Other Members : Dr M. J. Simms, Department of Geology, Ulster Museum, Botanic Gardens, Belfast BT9 5 AB Mr F. W. J. Bryant, 27, The Crescent, Maidenhead, Berkshire SL6 6AA Overseas Representatives Argentina : Dr M. O. Mancenido, Division Paleozoologia invertebrados, Facultad de Ciencias Naturales y Museo, Paseo del Bosque, 1900 La Plata. Australia : Dr K. J. McNamara, Western Australian Museum, Francis Street, Perth, Western Australia 6000. Canada : ProfessorS. H. Williams, Department of Earth Sciences, Memorial University, St John’s, Newfoundland A1B 3X5. China: Dr Chang Mee-mann, Institute of Vertebrate Palaeontology and Palaeoanthropology, Academia Sinica, P.O. Box 643, Beijing. Dr Rong Jia-yu, Nanjing Institute of Geology and Palaeontology, Chi-Ming-Ssu, Nanjing. France: Dr J.-L. Henry, Institut de Geologie, Universite de Rennes, Campus de Beaulieu, Avenue du General Leclerc, 35042 Rennes Cedex. Iberia: Professor F. Alvarez, Departamento de Geologia, Universidad de Oviedo, C /. Jesus Arias de Velasco, s/n. 33005 Oviedo, Spain. Japan : Dr I. Hayami, University Museum, University of Tokyo, Hongo 7-3-1, Tokyo. New Zealand: Dr R. A. Cooper, New Zealand Geological Survey, P.O. Box 30368, Lower Hutt. Scandinavia: Dr R. Bromley, Fredskovvej 4, 2840 Holte, Denmark. USA : Professor A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66044. Professor N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403. Professor M. A. Wilson, Department of Geology, College of Wooster, Wooster, Ohio 44961. Germany : Professor F. T. Fursich, Institut fur Palaontologie, Universitat, D8700 Wurzburg, Pleicherwall 1 MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1998 are: Institutional membership . . £90-00 (U.S. $175) Student membership .... £10-00 (U.S. $20) Ordinary membership . . . £28-00 (U.S. $50) Retired membership .... £14-00 (U.S. $25) There is no admission fee. Correspondence concerned with Institutional Membership should be addressed to Dr J. E. Francis, Department of Earth Sciences, The University, Leeds LS2 9JJ. 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 M. J. Barker, Department of Geology, University of Portsmouth, Burnaby Road, Portsmouth POl 3QL. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1998 will receive Palaeontology, Volume 41, Parts 1-6. Enquiries concerning back numbers should be directed to the Marketing Manager. Non-members may subscribe, and also obtain back issues up to five years old, at cover price through Blackwell Publishers Journals, P.O. Box 805, 108 Cowley Road, Oxford OX4 1FH, UK. For older issues contact the Marketing Manager. US Mailing: Periodicals postage paid at Rahway, New Jersey. Postmaster: send address corrections to Palaeontology, c/o Mercury Airfreight International Ltd, 2323 E-F Randolph Avenue, Avenel, NJ 07001, USA (US mailing agent). Cover: coalificd terminal sporangia from the Lower Devonian of the Welsh Borderland containing permanent tetrads (far left) and dyads. Similar spores found dispersed in Ordovician rocks are considered the earliest evidence for embryophytic life on land (from left to right, NMW94.76G.1; NMW96.1 1G.6; NMW97.42G.4. All x 45). lib* aries THE SIGNIFICANCE OF A NEW NEPHROPID LOBSTER FROM THE MIOCENE OF ANTARCTICA by RODNEY M. FELDMANN and J. ALISTAIR CRAME Abstract. The nephropid lobster, Hoploparia gazdzicki sp. nov., is described from Early Miocene glacio- marine sedimentary rocks of King George Island, South Shetland Islands, Antarctica. Such an occurrence considerably extends the stratigraphical range of a widespread lobster genus that reached its acme in the Late Cretaceous. The previous youngest records were from the Eocene of western Europe, and it would appear that, by the Early Miocene, the genus may have become a relict in relatively cold and deep waters in Antarctica. Although the full phylogenetic implications of this extension to the stratigraphical range are not yet apparent, there are some important palaeoecological ones. This occurrence can be taken as a further indication that certain benthic decapods were able to survive the onset of glacio-marine conditions in Antarctica. Perhaps other factors, such as the availability of food, habitat space, or decline in seasonal temperature fluctuation, ultimately controlled the decline of this major benthic group in the Southern Ocean. The fossil record of decapod crustaceans in Antarctica is remarkably robust in rocks ranging in age from Late Jurassic through to Eocene (Feldmann and Tshudy 1989). However, there are currently only two known occurrences of fossil decapods on the continent in post-Eocene rocks and there are only a few living pelagic decapods known from the region today. The decapods are certainly one of the key benthic groups to be grossly under-represented in the living Antarctic marine fuana and their demise has often been linked in a general way to Cenozoic climatic deterioration (Clarke and Crame 1989; Arntz et al. 1997). The two post-Eocene records of decapods in Antarctica are those of the homolodromiid crab, Antarctidromia inflata Forster, from the Cape Melville Formation (CMF) on King George Island, South Shetland Islands (Forster et al. 1985; see below), and an extremely fragmentary specimen of a palinurid lobster from the Pliocene of the Vestfold Hills, Princess Elizabeth Land (Feldmann and Quilty 1997). Although such a sparse record may be due to the restricted onshore occurrences of Cenozoic marine sedimentary rocks in Antarctica, it is also thought to reflect a very real decline in taxonomic diversity (Clarke and Crame 1989). In this context, it is particularly significant that a new species of fossil lobster has been collected from the Lower Miocene at Cape Melville, King George Island (Text-fig. 1). It is even more noteworthy that this lobster, Hoploparia gazdzickii sp. nov., represents a significant upward extension in the stratigraphical range of a lineage whose acme was reached in the Late Cretaceous, and whose youngest known representative, prior to this discovery, was from the Eocene of the northern hemisphere. It is the purpose of this paper to describe this new species and to speculate on the implications of this occurrence for both decapod evolution and biogeography in the high southern latitudes. GEOGRAPHICAL AND STRATIGRAPHICAL SETTING The three specimens to be described here were collected from a sequence of glacio-marine sedimentary rocks which is virtually unique to the Cape Melville peninsula (Text-fig. 1). Collectively, the sequence comprises the CMF, which is in turn a component of the Moby Dick Group, King George Island Supergroup (Birkenmajer 1987). Estimated to be between 175 and 200 m thick, the CMF is in sharp contact at its easternmost extremity with underlying columnar- jointed basalts/andesites of the Sherratt Bay Formation. On the western margins of the peninsula, [Palaeontology, Vol. 41, Part 5, 1998, pp. 807-814) © The Palaeontological Association 808 PALAEONTOLOGY, VOLUME 41 superficial deposits Penguin Island Gp 62°00’ Cape Melville Fm Destruction Bay Fm :;>>] Sherratt Bay Fm P.2702. 205, 224B 62°01’S' text-fig. 1. Locality and geological sketch map for Cape Melville, King George Island, South Shetland Islands. Localities of type specimens shown. Geological information based on Birkenmajer (1987, fig. 5). the CMF is underlain by a further sedimentary unit, the Destruction Bay Formation (Birkenmajer 1987; Text-fig. 1). The CMF is overlain to the west by sub-Recent volcanic rocks assignable to the Penguin Island Volcanic Group. The predominant lithology within the CMF is a pale grey/green/brown-weathering mudstone to silty mudstone bearing conspicuous, small to very large lonestones. Occasionally, the matrix coarsens to fine- or even medium-grained sandstone, and in places there are irregular seams and lenses of very coarse- to pebbly-sandstones. The lonestones comprise an extremely wide range of igneous, metamorphic and sedimentary lithologies, and are undoubtedly glacial dropstones. Some of them are of only local (i.e. northern Antarctic Peninsula) origin, but others, such as archaeocyath-bearing limestones, ripple cross-laminated red sandstones, and pink-weathering polymict conglomerates, indicate a source region as far distant as the Transantarctic Mountains (i.e. at least some 2000 km to the south). The CMF varies from a subhorizontal structural disposition to a gentle east or north-east dip of about 5°. It is cut by a prominent north-west-south-east trending dyke swarm, which is in turn cut by a later series of north-north-east-south-south-west trending normal faults. Two of these andesitic-basaltic dykes have been dated radiometrically (K-Ar) at 20 Ma, and this puts a minimum age constraint on the entire Moby Dick Group (Birkenmajer 1990). In addition, a tuff from close to the base of the Destruction Bay Formation has been dated (K-Ar) at 23 Ma, and both brachiopods and foraminifera from the same unit have strong Lower Miocene affinities. A consensus of radiometric and palaeontological age determinations indicates that the CMF is best regarded as Early Miocene (Birkenmajer 1987, 1990). All three specimens were collected from the plateau surface extending along the top of the peninsula (Text-fig. 1). The holotype, P.270 1.82, comes from the south-western corner, close to the FELDMANN AND CRAME: MIOCENE LOBSTER 809 moraine material associated with the edge of the ice cap; the two paratypes (P.2702.205, 224B) were collected from the eastern slopes of ‘Crab Creek’. Such locations indicate a stratigraphical position within approximately the uppermost 75 m of the CMF, and a close association with a rich benthic marine invertebrate assemblage dominated by infaunal bivalves, gastropods, solitary corals and crabs. As might be expected in such a mud-rich environment, the bivalve assemblage is dominated by deposit-feeding nuculids and nuculanids. Other taxa include limopsids, several small heteroconchs, and a comparatively large number of anomalodesmatans. The prolific crab remains range from disarticulated chelae and incomplete carapaces to whole, articulated specimens associated with burrow structures. They have been assigned to just one taxon, Antarctidromia inflata Forster (Forster et al. 1985, 1987). The solitary corals have been identified as Flabellum rariseptatum (Roniewicz and Morycowa 1987), and the common gastropods include a medium-large volutid, at least two types of buccinid, a large turrid ( Austrotoma ), and several forms of naticid. Brachiopods, echinoids, scaphopods, bryozoans and large foraminiferans are also present, and overall the assemblage has a relatively deep-water, outer-shelf aspect. SYSTEMATIC PALAEONTOLOGY Order decapoda Latreille, 1803 Infraorder astacidea Latreille, 1803 Family nephropidae Dana, 1852 Genus hoploparia McCoy, 1 849 Type species. Astacus longimanus G. B. Sowerby, 1826, by subsequent designation of Rathbun, 1926. Hoploparia gazdzickii sp. nov. Text-figures 2-3 Derivation of name. The trivial name recognizes the significant contributions of Andrzej Gazdzicki, Polish Academy of Sciences, Warszawa, to the study of the geology and palaeontology of King George Island. Types. The holotype, P.2701.82, and two paratypes, P.2702.205 and 2702. 224B, are deposited in the collections of the British Antarctic Survey, Cambridge, England. Description. Moderate to small sized (for genus) carapace more than twice as long as high, with diminutive cephalic spines, and with well defined groove pattern. Dorsal margin biconvex with postcervical groove crossing midline behind midlength. Posterior margin incomplete, convex. Ventral margin smoothly convex with narrow marginal rim and furrow. Frontal margin broken but with shallow, rimmed orbital margin. Rostrum not preserved. Two weak spine rows developed on dorso-anterior portion of cephalic region. Rostral spine row with four spines of which anteriormost is largest. Supraorbital spine row with three spines increasing in size anteriorly. Single, prominent antennal spine. Carapace grooves narrow, deeply incised, distinct. Cervical groove (e of Text-fig. 3) originates at point about one-third total height from midline, becoming narrower and better defined ventrally; curving anteriorly in smooth arc terminating abruptly against nearly straight, anteriorly-inclined antennal groove (b). Gastroorbital groove (d) an indistinct depression. Postcervical groove (c) with straight dorsal segment crossing midline behind midpoint of carapace, weakly convex-forward midsection inclined at about 45° to dorsum, and short anterior section curving toward cervical groove. Branchiocardiac groove (a) smoothly convex forward, coalescing with postcervical groove at midsection and merging with deeply incised, tightly curved hepatic groove (bj) defining presumed position of adductor testis muscle insertion (/) which is swollen and bears several fine granules. Region of mandibular external articulation (co) broadly and subtly swollen. Branchiostegite with very fine, uniformly spaced setal pits overall and few very fine pustules along ventral margin. Abdomen with well differentiated tergal and pleural surfaces separated by distinct convex-downward ridge (Text-fig. 3). Terga with coarsely punctate irregularly undulating axial regions and transversely ovoid, irregular 810 PALAEONTOLOGY, VOLUME 41 text-fig. 2. Hoploparia gazdzickii sp. nov. A, left lateral view of holotype, P.2701.82, showing nearly complete cephalothorax and abdomen, b, left lateral view of paratype, P.2702.205, showing incompletely preserved abdomen, c, right lateral view of paratype, P.2702.244B, showing crushed, incomplete cephalothorax. d, right lateral view of abdomen of holotype. E, frontal view of holotype showing mandibles (arrow) and fragments of maxillipeds. All x 1-5. swellings laterally. Posterior rim of each tergum smooth, elevated. Pleura smoother than terga, domed medially, swollen at posterodorsal corner where somites articulate. Pleuron of first somite small, triangular, anteriorly, directed; that of second somite larger than any of the others, broadly obovate with posteriorly directed acute spine. Remaining pleura lanceolate with acute tips directed slightly toward posterior. Single pit situated at midpoint of pleura 3-5. Telson margins not preserved; axis depressed, bounded by two broadly elevated longitudinal ridges diverging slightly toward posterior. Uropods large, elongate, oval, with diaresis. Mandibles strongly inflated, occlusal surface of right mandible overlaps that on left. First pereiopods not known. Proximal elements of pereiopods 2-5 of uniform size, long, narrow, cylindrical. Measurements. Dimensions of carapace, in mm, are given on Text-figure 3. FELDMANN AND CRAME: MIOCENE LOBSTER 811 text-fig. 3. Line drawings of Hoploparia gazdzickii sp. nov. showing the positions of the carapace features, orientation of measurements taken, in millimetres, and details of morphology of the ab- domen (composite drawing based primarily on the holotype) a, branchiocardiac groove; b, antennar groove; bp hepatic; c, postcervical; d, gastroorbital; e, cervical; to, position of mandibular external articulation; /, inferred position of 'adductor testis' muscle attachment. Remarks. Hoploparia, along with other nephropid lobsters, was recently subjected to a cladistic analysis (Tshudy and Babcock 1997) which tested morphological characters used to define genera as well as the affinities of included genera. On the basis of this, and previous works, representatives of the genus Hoploparia may be distinguished from those in the closely related genus Homarus in several ways. With reference to the specimens described above, the development of a cervical groove that extends dorsal to the level of the gastroorbital groove, possession of a postcervical groove that is strongly developed throughout and that extends toward the cervical groove, projection of the branchiocardiac groove ventral to the presumed attachment site of the adductor testis muscle merging with the hepatic groove to intercept the cervical groove, and development of strong ornament on the abdomen resulting in clear demarcation of the tergal and pleural regions, are all characters that permit confident assignment to Hoploparia. Other distinguishing features, including the nature of the rostrum and the conformation of the chelae (Glaessner 1969), cannot be used because they are not preserved on the available material. Species within the genus are distinguished on the basis of carapace ornament, relative degrees of development of the carapace grooves, morphology of the chelae, and details of the ornament on the abdomen. The combination of the characters exhibited by Hoploparia gazdzickii sp. nov. clearly distinguishes it from previously described species. It possesses an antennal groove that is nearly straight, instead of smoothly curved, and that is steeply inclined in an anterodorsal direction; this feature is unique. In addition, the carapace of the new species is nearly devoid of nodes, spines, or other ornament. In this regard, it more closely resembles species of Homarus. The only distinctive carapace ornament is that of the two rows of spines on the cephalic region and a large antennal spine, characters exhibited by all, or nearly all, species within the genus. Those spines, however, are diminutive in H. gazdzickii sp. nov. Finally, although the carapace is nearly smooth, the abdomen is heavily ornamented. The type species of the genus, Hoploparia longimanus (G. B. Sowerby), from the Upper Cretaceous of England, possesses rows of nodes just posterior to the cervical and postcervical grooves, pustulose ornament on the cephalic region, an antennar groove that is nearly parallel to the ventral carapace margin, and highly ornamented abdominal pleura; none of these characters is evident on H. gazdzickii. Two species of Hoploparia have been described previously from Antarctica. Hoploparia stokesi (Weller 1903), has been collected from numerous sites on Snow Hill, Seymour, James Ross, and Vega islands in rocks ranging from Campanian through to Paleocene (Feldmann and Tshudy 1989). 812 PALAEONTOLOGY, VOLUME 41 Individuals within this species can be distinguished readily from H. gazdzickii. Hoploparia stokesi tends to be much larger, exhibits a more granular carapace, more strongly developed spines on both the mandibular articulation and the adductor testis region, a prominent spine on the ridge separating the abdominal pleura from the terga, a nodose surface on the telson, and keeled uropods. Hoploparia antarctica Wilckens, 1907, is known from the Campanian of James Ross Island as well as the Campanian-Maastrichtian of southern and central Patagonia, Argentina. This latter species bears a prominent row of antennal spines, moderate to weakly developed intermediate and branchial carinae on the branchiostegite, nearly smooth tergal surfaces, and strongly inflated borders on the abdominal pleura. DISCUSSION Prior to this study, the youngest occurrences of Hoploparia were those in the Eocene of Europe; there are, for example, references to Hoploparia sp. from both Germany (Ebert 1887) and Italy (Ristori 1889). The overall pattern of distribution for the genus would appear to be one of origin in the Early Cretaceous, at least by the Hauterivian (Feldmann 1974), and possibly as early as the Berriasian-early Valanginian in the Americas (Aguirre Urreta 1989). By the Late Cretaceous, the genus was distributed world-wide: in the epicontinental seaways of North America, the Atlantic Ocean basin, Europe, Madagascar, South America and Antarctica. The geographical range of the genus declined significantly during the Paleogene, after which it was thought to have vanished, either by true extinction or by giving rise to one or more of the modern nephropid genera (Aguirre Urreta 1989). However, it is now apparent that the genus persisted as a relict in the Antarctic region at least into the Early Miocene. There remain unanswered questions regarding the origins of the modern nephropid genera. Although that topic is not directly relevant to the present work, this new discovery of a Miocene Hoploparia does raise the question of whether it formed the rootstock of at least some of the modern genera or whether it was a contemporary of genera which had arisen earlier. Metanephrops Jenkins, 1972 is reported to have arisen at least by the Late Cretaceous in the form of Metanephrops jenkinsi Feldmann, 1989, from the James Ross Basin, Antarctica. Tshudy and Babcock (1997) concluded that the genera most closely allied to Hoploparia arose in the Early Cretaceous. Difficulty in testing the relationships of the other nephropids using palaeontological evidence arises because most of the modern nephropids are inhabitants of outer shelf and slope habitats (Holthuis 1974), and these are not well represented in the fossil record. There are also unresolved questions regarding the palaeoecological implications of the occurrence of Hoploparia in the CMF fauna. Certainly the preponderance of occurrences of the genus throughout its geological history have been in inner shelf, moderate to high energy settings (Aguirre Urreta 1989). In almost all instances too, biotic associations indicate normal marine settings. As fossil occurrences of Hoploparia range from as far north as Greenland to as far south as Antarctica, this could be taken to represent original water temperatures ranging from cool-temperate to subtropical. Nevertheless, the outer limit of the bathymetric range has never been adequately constrained, and these records from the Lower Miocene of Antarctica indicate that the genus also inhabited moderately deep waters. A comparison can be made here with modern Homarus Weber, which is known to occur at virtually all depths on the continental shelf and, before active capture by humans, was observed in tide pools in intertidal settings (Herrick 1911). Finally, it is striking how two quite different benthic decapod genera, Antarctidromia and Hoploparia , co-occur in the CMF. This is without doubt a glacial deposit, for the dropstones could only have been deposited from very large icebergs originating along the southernmost margins of the Weddell Sea (present day Ronne Ice Shelf)- Thus, at least two distinct decapod taxa were able to survive cold-water (glacial) conditions and it may be that there was no simple link between the onset of glaciation and the extinction of many benthic marine taxa (Clarke and Crame 1989). Perhaps the extinctions were phased over a long period of time, or factors other than low temperature per se were of paramount importance. There is a growing volume of evidence to suggest that the ability to cope with oligotrophic conditions may be just as important to survival within the FELDMANN AND CRAME: MIOCENE LOBSTER 813 present-day Southern Ocean benthos as the ability to withstand near-freezing conditions (Arntz et al. 1997). Acknowledgements. RMF’s work at the British Antarctic Survey was supported by NSF grant OPP 9526252. JAC thanks S. M. Redshaw for helping to collect the specimens during the 1995 field season. REFERENCES aguirre urreta, m. b. 1989. The Cretaceous decapod Crustacea of Argentina and the Antarctic Peninsula. Palaeontology , 32, 499-552. arntz, w. e., gutt, j. and klages, m. 1997. Antarctic marine biodiversity 3-14. In battaglia, b., Valencia, j. and walton, D. w. H. (eds). Antarctic communities : species, structure and survival. Cambridge University Press, Cambridge, 464 pp. birkenmajer, k. 1987. Oligocene-Miocene glacio-marine sequences of King George Island (South Shetland Islands), Antarctica. Pa/aeontologia Polonica, 49, 9-36. — 1990. Geochronology and climatostratigraphy of Tertiary glacial and interglacial successions on King George Island, South Shetland Islands (West Antarctica). Zentralblatt fiir Geologic und Paldontologie, 1, No. 1-2, 141-151. clarke, A. and crame, J. A. 1989. The origin of the Southern Ocean marine fauna. 253-268. In crame, j. a. (ed.) Origins and evolution of the Antarctic biota. Geological Society, London, Special Publication, 47, 322 pp. dana, J. D. 1852. Crustacea. United States Exploring Expedition during the years 1838, 1839 , 1840, 1841 , 1842 under the command of Charles Wilkes, U.S.N., 13. C. Sherman, Philadelphia, 1620 pp. ebert, T. 1887. Beitrag zur Kenntnis der tertiaren Dekapoden Deutschlands. Jahrbuch der ( Koniglich ) Preussischen Geologischen Landesanstalt und Bergakademie, Berlin, 262-271. feldmann, r. m. 1974. Hoploparia riddlensis, a new species of lobster (Decapoda: Nephropidae) from the Days Creek Formation (Hauterivian, Lower Cretaceous) of Oregon. Journal of Paleontology , 48, 587-593. — 1989. Metanephrops jenkinsi n. sp. (Decapoda: Nephropidae) from the Cretaceous and Paleocene of Seymour Island, Antarctica. Journal of Paleontology, 63, 64-69. — and quilty, p. g. 1997. First Pliocene decapod crustacean (Malacostraca: Palinuridae) from the Antarctic. Antarctic Science , 9, 56-60. — and tshudy, d. m. 1989. Evolutionary patterns in macrurous decapod crustaceans from Cretaceous to early Cenozoic rocks of the James Ross Island region, Antarctica, 183-195. In crame, j. a. (ed.). Origins and evolution of the Antarctic Biota. Geological Society, London, Special Publication 47, 322 pp. forster, r., gazdzicki, a. and wrona, r. 1985. First record of a homolodromiid crab from a Lower Miocene glacio-marine sequence of West Antarctica. Neues Jarbuch fiir Geologic und Paldontologie, Monatshefte, 8, 399-404. 1987. Homolodromiid crabs from the Cape Melville Formation (Lower Miocene) of King George Island, West Antarctica. Pa/aeontologia Polonica , 49, 147-161. glaessner, m. f. 1969. Decapoda. R399-R651, R626-R628. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R. Arthropoda 4. Geological Society of America and University of Kansas Press, Lawrence, 651 pp. herrick, F. h. 191 1. Natural history of the American lobster. Bulletin of the U.S. Bureau of Fisheries, 29, 1-408. holthus, l. b. 1974. The lobsters of the superfamily Nephropidae of the Atlantic Ocean (Crustacea: Decapoda). Bulletin of Marine Science, 24, 723-884. jenkins, R. J. F. 1972. Metanephrops, a new genus of late Pliocene to Recent lobsters (Decapoda, Nephropidae). Crustaceana, 22, 161-177. latreille, p. A. 1802-1803. Histoire naturelle, generate et particulere, des crustaces, des arachnides, des myriapodes et des insectes: Volume 3. F. Dufart, Paris, 468 pp. McCOY, F. 1849. On the classification of some British fossil Crustacea with notices of new forms in the University Collection at Cambridge. Annals and Magazine of Natural History, Series 2, 4, 161-179, 330-335. rathbun, m. j. 1926. The fossil stalk-eyed Crustacea of the Pacific slope of North America. Bulletin of the U.S. National Museum, 138, 1-155. ristori, g. 1889. Crostacei Piemontesi del miocene inferiore. Bollettino della Societa geologica italiana , 7, 397-412. roniewicz, e. and morycowa, e. 1987. Development and variability of Tertiary Flabellum rariseptatum (Scleractinia), King George Island, West Antarctica. Palaeontologia Polonica, 49, 83-103. 814 PALAEONTOLOGY, VOLUME 41 sowerby, G. b. 1826. Description of a new species of Astacus, found in a fossil state at Lyme Regis. Zoological Journal , 2, 493^194. tshudy, d. and babcock, l. e. 1977. Morphology-based phylogenetic analysis of the clawed lobsters (family Nephropidae and the new family Chilenophoberidae). Journal of Crustacean Biology, 17, 253-261. weller, s. 1903. The Stokes collection of Antarctic fossils. Journal of Geology, 11, 413^119. wilckens, o. 1907. Die Lamellibranchiaten, Gastropoden U.S.W. der oberen Kreide Siidpatagoniens. Berichte der Naturforschended Gesellschaft zu Freiburg i. Br., 15, 97-166. RODNEY M. FELDMANN Department of Geology Kent State University Kent, Ohio 44242, USA Typescript received 2 October 1997 Revised typescript received 6 February 1998 J. ALISTAIR CRAME British Antarctic Survey High Cross, Madingley Road Cambridge CB3 OET, UK NEW PYGOCEPHALOMORPH CRUSTACEANS FROM THE PERMIAN OF CHINA AND THEIR PHYLOGENETIC RELATIONSHIPS By ROD S. TAYLOR, SHEN YAN-BIN and FREDERICK R. SCHRAM Abstract. Members of the malacostracan order Pygocephalomorpha are among the most characteristic elements in nearshore marine and freshwater communities in the Carboniferous and Permian of Europe and North America. A new family of pygocephalomorph Eumalacostraca, Tylocarididae, with two new monospecific genera, is described from China, where it occurs in the Early Permian Tungtzeyen Formation of Fujian, and in the Late Permian Lungtan Formation of Hunan. The descriptions of Fujianocaris bifurcatus gen. et sp. nov. and Tylocaris asiaticus gen. et sp. nov. are based on dorsally preserved isolated carapaces, some showing incomplete abdominal details, but with no complete tail fans. Opinions on the affinities of Pygocephalomorpha to other malacostracans have varied but they are generally regarded as a separate order of ‘mysidacean' peracarids. Hitherto the phylogeny of the group has not been considered, and the current family level taxonomy remains rather artificial. A cladistic analysis of fossil and Recent 'mysidacean' and pygocephalomorph crustaceans is presented here which outlines the affinities within the group and holds promise for an eventual natural taxonomy of the Pygocephalomorpha. Little work has been done on the palaeobiology and taxonomy of fossil Crustacea in China, especially with respect to global biogeography (Shen 1983), with the exception of extensive taxonomic work on conchostracans, which range from the Devonian to the Cretaceous (Shen 1978, 1981, 1984, 1990; Zhang et al. 1990). Palaeobiological research has increased in China recently due to the discovery of such important localities as the Cambrian Lagerstdtte at Chengjiang (e.g. Chen et al. 1995fl, 19956), and a result has been the discovery of new crustaceans in Early Permian strata in south-east China. This paper describes the new taxa Fujianocaris bifurcatus and Tylocaris asiaticus , both apparently belonging to the Pygocephalomorpha. Whilst these new species are only the second reported discovery of Pygocephalomorpha in China (see Shen 1983), members of this order have long been recognized elsewhere as one of the most prominent and striking crustacean groups in late Palaeozoic nearshore marine and freshwater communities, in particular from North America and Europe. However, determination of the phylogenetic affinities of this enigmatic group has remained problematical. Prestwich (1840) was the first to describe a pygocephalomorph, a carapace from the British Coal Measures; he named it A pus dubius , and believed that its affinities might be with the notostracan phyllopods. Later, Huxley (1857) described Pygocephalus cooperi , also from the British Coal Measures; in this specimen the ventral aspect of the thorax is preserved, but he did not compare it with A. dubius. Salter (1861) realized that the carapace described by Prestwich was not a phyllopod, and erected the genus Anthrapalaemon to accommodate it and some newly discovered carapace specimens that he ascribed to another species, A. grossarti. No-one appreciated at that time that these various taxa had affinities to one another. Indeed, there persisted in the literature an unnatural dichotomous taxonomy: fossils preserving a dorsal view of the carapace were placed in Anthrapalaemon , while those preserving the ventral aspects of the thorax bore the name Pygocephalus. The confusion increased when Woodward ( 1 879) applied the generic name Necroscilla to separate abdomina and Salter (1863) placed a tail fan in a separate genus Diplostylus. The generic name Anthrapalaemon became widely employed for any large lobster-like carapace. [Palaeontology, Vol. 41, Part 5, 1998, pp. 815-834, 2 pis] © The Palaeontological Association 816 PALAEONTOLOGY, VOLUME 41 text-fig. 1. Localities from which Fujianocaris bifurcatus and Tylocaris asiaticus have been collected, indicated by arrowheads. Shaded areas represent the Lower Permian. Peach (1883) erected a separate genus, Pseudogalathea , for some distinctly ridged forms, and did the same for some other Scottish taxa that he segregated under the genus Tealliocaris. Brooks (1962) made a major contribution towards resolving the taxonomy of this group. He proposed Pseudotealliocaris , for some distinctly decorated taxa, recognized the synonymy of Pygocephalus and Anthrapalaeomon (former is senior synonym), confirmed the taxonomic status of the North American species Anthracaris gracilis, erected Mamayocaris for another North American species, and made some assumptions about the supposed higher taxonomic affinities of the pygo- cephalomorphs. Brooks suggested that one should not compare pygocephalomorphs with phyllopods, schizopods or decapods, as had been done in the past, but placed them in a distinct order, Eocarida, with various other Palaeozoic forms. Finally, Schrain (1974c/, 19746, 1979) imposed some order on the species level taxonomy in the group, especially among the British faunas, clarified the issues of thoracopod anatomy that had coloured Brooks’ interpretation of the higher taxonomy, and performed a cladistic analysis that advanced a clear hypothesis about the possible TAYLOR ET AL.\ PERMIAN CRUSTACEANS 817 higher affinities of the pygocephalomorphs. In addition, Schram (1978) also recognized another genus in the Permian of Russia, Jerometichenoria. All these discoveries focused largely on ‘northern hemisphere’ taxa from Laurentia. Nevertheless, another important source of pygocephalomorphs occurs in ‘southern hemisphere’, essentially Gondwanan, localities. Broom (1931) described a South African species, Notocaris tapscotti , and Clarke (1920) first recognized a Brazilian form, Paulocaris pachecoi. Later, Beurlen (1931, 1934) expanded on the South American fauna with his erection of Liocaris and Pygaspis, both again from Brazil. Brooks (1962) synonymized both of these genera with Paulocaris , but they have since been resurrected by other authors (e.g. Pinto 1971), reflecting the taxonomic confusion that has marked the history of this group. Unfortunately, these Gondwanan taxa are based on rare and poorly preserved material, making definitive taxonomic assignments difficult. Brooks (1969) set these poorly known, southern hemisphere forms aside as a separate family, Notocarididae, but its only diagnostic character, reduced abdomen flexed under the thorax, clearly does not apply to all southern forms and may merely be an artefact of preservation. The Brazilian Pygaspis bear a regular, large, posteriorly directed abdomen (Pinto 1971), and the supposed diagnostic flexure under the thorax is also present on many specimens of northern hemisphere pygocephalomorphs. In the course of this work, we noted similarities between our two new genera and the Scottish Carboniferous genus Pseudogalathea. However, there are palaeobiogeographical implications arising from this, with phylogenetically highly derived animals, with many apomorphic carapace features, arriving at disparate parts of the Palaeozoic world. Whilst we could not preclude this possibility, it caused us to re-examinee the total array of anatomical information that could be derived from fossil pygocephalomorphs and possible near relatives, and we performed a cladistic analysis to test more rigorously our initial conclusions on the affinities of the Chinese taxa. The material used in this study was obtained from the Permian of Fujian Province, south-east China (Text-fig. 1). Most specimens were collected from a coal mine in the village of Changta, Nanjing County, in the third member of the Early Permian Tungtzeyen Formation (one specimen has also been reported from an equivalent horizon at Longtan village, Yongdin County, Fujian (Zhu 1990, pi. 21, fig. 15)). One specimen was found at each of the following: Xihushan, Longyan County, Fujian Province, Early Permian Tungtzeyen Formation; Shitangpu Village, Lukou Town, Zhuzhou City, Hunan Province, Late Permian Lungtan Formation; and an undetermined locality from the Permian of Fujian. This last specimen, due to its uncertain provenance is not considered further. All specimens are deposited in the Nanjing Institute of Geology and Palaeontology (NIGP), Academia Sinica. SYSTEMATIC PALAEONTOLOGY Class malacostraca Latrielle, 1802 Order pygocephalomorpha Beurlen, 1930 Family tylocarididae fam. nov. Diagnosis. Carapace with large falciform rostrum with central groove and papillated margin; prominent mid-dorsal keel, with posterior bifurcation merging with posterior carapace margin; well-developed cervical and rostro-gastric ridges, surrounding papillated rostral ridge; antero-lateral and posterio-lateral spines present; heavily thickened carapace margin. Abdomen with medial and one set of lateral ridges on tergites; elongate telson with finely bifurcated tip; endopods and exopods with serrate margins and no diaresis associated with exopod. Genus fujianocaris gen. nov. Derivation of name. From Fujian Province, China. Type species. Fujianocaris bifurcatus. 818 PALAEONTOLOGY, VOLUME 41 Diagnosis. Carapace with prominent mid-dorsal keel, bifurcated at both anterior and posterior ends, and pair of prominent lateral keels; carapace margin, rostrum, cervical ridge and keels decorated with papillations, slightly smaller on the carapace margin and rostrum; remainder of carapace smooth; cervical ridges well developed, with shallow cervical grooves; large falciform rostrum with central groove present. Fujianocaris bifurcatus sp. nov. Plate 1, figures 1-5; Text-figures 2b, 3a Derivation of name. From the posterior bifurcations of the telson and the mid-dorsal keel of the carapace. Holotype. NIGP 126323 A/B; part and counterpart of a carapace and associated abdomen (PI. 1, fig. 1). Paratypes. NIGP 126324A-2, 3, NIGP 126327, NIGP 126328, NIGP 126329-1, 2, NIGP 126330-1, 2, NIGP 12633 IB-2, NIGP 126332A, B, NIGP 126333A/B, NIGP 126334-2, NIGP 126335, NIGP 126336A-1, 2/B- 1, 2. Horizon and locality. No. 25 coal bed, third member of Lower Permian Tungtzeyen Formation, Xiangshuping, Changta coal mine, Nanjing County, Fujian Province (Text-fig. 1). Diagnosis. As for genus. Description. The carapace appears to have been heavily sclerotized. A prominent mid-dorsal keel is present, extending two-thirds to three-quarters its length to the posterior margin (NIGP 126328; PI. 1, fig. 4). This keel is continuous with a greatly thickened ridge along the posterior, lateral and anterior margins of the carapace. At the immediate anterior end of the medial keel is a pair of antero-laterally directed cervical ridges (in some specimens, these appear to be almost continuous with the keel (NIGP 126330-1 ; PI. 1, fig. 3)). These ridges are curved slightly outwards and extend approximately one-third of the distance to the lateral margins of the carapace; they appear to run parallel to what seems to be a set of very shallow cervical grooves (NIGP 126329- 1, 2). At its posterior end, the medial keel terminates in a pair of mid-lateral, posteriorly directed spines (NIGP 126323B, NIGP 126331B-2). A pair of lateral keels extends from just posterior of the cervical grooves to the posterior carapace margin, approximately midway between the medial keel and the lateral margin of the carapace (NIGP 126328). These lateral keels extend for approximately the same distance as the medial one. Papillations decorate all keels, more heavily on the medial, and the thickened posterior and lateral carapace margins (NIGP 126330-1; PI. 1, fig. 3). No branchiostegal spines are present. The rostrum is long, approximately one-quarter the length of the carapace, and curves slightly ventrally. It is semicircular in cross section, with a dorsal, central groove (NIGP 126330-1, NIGP 126334-2, NIGP 126329- 1, 2), and originates from a triangular rostral ridge anterior to the cervical ridges (NIGP 126330-1). The rostral margin and rostral ridge are papillated. There emerges from this rostral ridge a pair of papillated antero-lateral gastric ridges, running approximately parallel to the antero-lateral margin of the carapace (NIGP 126336B-1, NIGP 126329-1, 2). These ridges are wider and more robust laterally than mid-dorsally. At their lateral extent they turn posteriorly, adjacent to the termination of the cervical ridges (NIGP 126330-1, 2). Short, EXPLANATON OF PLATE 1 Figs 1-5. Fujianocaris bifurcatus gen. et sp. nov. 1-2, NIGP 126323A, holotype. 1, x 4; 2, tail fan (small arrow = telson, large arrows = endopods, tailless arrows = expods); x 11. 3, NIGP 126329-1; x 11. 4, NIGP 126328, x 5-5. 5, NIGP 126322, x 5. 1-3 from the Early Permian Tungtzeyen Formation, Changta, Nanjing County, Fujian Province; 4 from the same formation, Xihushan, Longyan County, Fujian Province; 5 from the Late Permian Lungtan Formation, Shitangpu, Lukou, Zhuzhou, Hunan Province. PLATE 1 TAYLOR et al., Fujianocaris 820 PALAEONTOLOGY, VOLUME 41 text-fig. 2. a, Tylocaris asiaticus gen. et sp. nov. ; NIGP 126324A; partial tail fan; x 14-5. B, Fujianocaris bifurcatus gen. et sp. nov.; NIGP 126334; abdomen and tail fan; x 12-5. Both specimens from the Early Permian Tungtzeyen Formation, Changta village, Nanjing County, Fujian Province (small arrow = telson; large arrows = endopods; tailless arrows = exopods). rounded, antero-lateral and long, postero-lateral spines are present (NIGP 126323B, NIGP 126327, NIGP 126331B-2, NIGP 126334A-2, 3). A set of broad optic notches is located between the rostrum to the antero- lateral spine. The abdomen is short, slightly less than one-half the length of the carapace (NIGP 126323A, NIGP 126334- 2). Four abdominal segments are exposed (the first two shielded under the carapace), each possessing well- developed pleura with posteriorly-directed processes. Each abdominal tergite possesses a mid-dorsal triangular boss (best developed on the last two pleomeres) as well as a pair of small lateral ridges (NIGP 126332B). The length of the segments remains constant whilst the width decreases markedly in the series, such that the sixth abdominal segment is approximately one-half the width of the third (NIGP 126323A, NIGP 126327, NIGP 126334-2). The telson is narrow and very long, with a length c. 2-5 times that of the last abdominal segment (NIGP 126323A, NIGP 126334-2) (Text-fig. 2b). The telson possesses a longitudinal, medial ridge (PI. 1 , fig. 2), whilst its distal terminus appears to form a small fork (NIGP 1 26323A). The uropods consist of lobate exopods and endopods, the latter with medially serrate margins (NIGP 126323A). No diaresis is noted on the exopods, nor are statocysts visible. Remarks. The sole specimen (PI. 1, fig. 5) collected from the Late Permian Lungtan Formation of Shitangpu village, Hunan Province is of particular interest. It is included here in Fujianocaris bifurcatus , despite some small differences from other members of this species. In most aspects (e.g. the cervical and gastric ridges and the mid-dorsal keel) it is like other specimens of F. bifurcatus , but TAYLOR ET AL.. PERMIAN CRUSTACEANS 821 table 1. Measurements in millimetres of Fujianocaris bifur catus. Specimen Rostrum length Carapace length Abdomen length Abdominal segment width Telson length 1 2 3 4 126322 > 1-7 12-2 126323A > 1-3 9-3 5-8 8-5 7-4 5-9 4-5 6-7 126327 > 3-8 15-3 11-8 10-5 9-8 80 6-3 126328 2-8 12-2 126329-1 > 1-8 8-0 126329-2 1-6 6-6 126330-1 > 30 140 126330-2 7-8 126331B-2 13-3 126332A 2-7 10-5 6-7 5-8 4-7 126334-2 3-75 5-3 4-75 3-2 2-7 > 2-0 126336-1 30 9-5 126336-2 1-7 6-8 it lacks the lateral keels, that are characteristic of this species. This specimen is preserved such that there is little contrast between it and the surrounding matrix, making it difficult to determine whether all relevant details of the carapace have been preserved, or whether the absence of these keels is an artefact of preservation. Due to the overall similarities between this specimen and the Early Permian F. bifurcatus, it is considered for the time being as an unusual member of this taxon rather than a separate species. Genus tylocaris gen. nov. Derivation of name. From the Greek tylos, knob, referring to the presence of numerous papillations over the carapace. Type species. Tylocaris asiaticus. Diagnosis. Carapace with prominent mid-dorsal keel, bifurcated at posterior end, with gastric and cardiac ridges anterior to, and hepatic ridges flanking the anterior end; small papillations highly concentrated on mid-dorsal keel and carapace margin, and more loosely distributed over remainder of carapace; cervical and cardiac ridges well developed; rostrum falciform with central goove; telson long and narrow, with elongate medial ridge and small fork on terminus; a pair of pits on the dorsal surface of each endopod and exopod. Tylocaris asiaticus sp. nov. Plate 2, figures 1-3; Text-figures 2a, 3b Derivation of name. From its discovery in Asia. Holotvpe. NIGP 126324 A-l/B; part and counterpart of an incomplete carapace and its associated abdomen (PI. 2,' fig. 1). Paratypes. NIGP 126325, NIGP 126326A/B, NIGP 126331 A/B-l, NIGP 126334-1. 822 PALAEONTOLOGY, VOLUME 41 Horizon and locality. No. 25 coal bed, third member of Lower Permian Tungtzeyen Formation, Xiangshuping, Changta coal mine, Nanjing County, Fujian Province (Text-fig. 1). Diagnosis. As that for genus. Description. The carapace was probably not heavily sclerotized in life, as suggested by wrinkling of some specimens (NIGP 126326B; see PI. 2, fig. 2). It posseses a very prominent mid-dorsal ridge, extending two- thirds the length of the carapace from the cervical groove to the posterior margin. This median ridge forks posteriorly and is continuous with a thickened ridge along the posterior margin of the carapace. At the point at which these ridges merge, there is a set of tiny, posteriorly directed processes (NIGP 126326B). The posterior thickened ridge continues along the lateral and anterior margins of the carapace (NIGP 126324A-1). Flanking the anterior end of the median ridge is a pair of highly arched hepatic ridges, with concave surfaces facing inwards (PI. 2, fig. 2). Immediately anterior to these is a fine cervical groove, which is in turn adjacent to a pair of antero-laterally directed cervical ridges (NIGP 126324A-1/B, NIGP 126331B-1). A pair of broad optic notches is present between the rostrum and a set of tiny, rounded antero-lateral spines (NIGP 126324-1 /B, NIGP 126326B). Papillations are densely concentrated on the medial keel and the posterior and lateral carapace margins (NIGP 126324A-1, NIGP 126325), and this ornament is also distributed over the central portion of the carapace, becoming less densely aggregated near the lateral margins (PI. 2, figs 1-2). No branchiostegal serrations on the lateral margins are present. The rostrum is long, one-quarter to one-third the length of the carapace. It is slightly falciform, is an extension of the papillated mid-dorsal rostral ridge (NIGP 126331A/B-1), and possesses papillations along its margin. A pair of narrow, weakly developed, antero-lateral ridges emerges from the anteriormost region of the rostral ridge. These extend posteriorly and laterally from the rostral ridge to the cervical groove (NIGP 126331B-1). The carapace bears a pair of short, rounded, antero-lateral spines lateral to the optic notch and a pair of well-developed postero-lateral spines (NIGP 126324B, NIGP 126326B). One specimen (NIGP 126324A/B; PI. 2, fig. 1) possesses what appear to be dislocated, regularly segmented antennal fragments near the anterior end of the carapace. The abdomen is approximately the same length as the carapace. Five abdominal segments are exposed, which possess posteriorly pointed pleura. Segment width decreases whilst length increases distally along the abdominal series, such that the sixth abdominal segment is approximately one-half the width but twice the length of the second segment (NIGP 126324A-1). Each of the tergites bears a broad, triangular medial ridge, as well as a pair of narrow, longitudinal lateral ridges (NIGP 126325; PI. 2, fig. 1). The elongate and narrow telson appears to terminate in a finely forked tip (NIGP 126331A). It is longer by approximately one-third than the final abdominal tergite, and carries a narrow medial keel running its entire length. Two specimens each show what may be a single caudal furca, occurring at approximately the middle (NIGP 126331 A) and near the end (NIGP 1 26323 A- 1) of the telson. A pair of lobate uropods, possibly distally pointed, are present, the endopod possessing serrate margins (NIGP 126324A-1 ; Text-fig. 2a). A diaresis is not visible on the exopods. One specimen (NIGP 126325; PI. 2, fig. 3) exhibits a pair of small pits along the dorsal midline of the exopods and endopods. Statocysts are not seen. Remarks. There is one anomalous specimen (NIGP 126326 A/B; PI. 2, fig. 2), which possesses, immediately anterior to the cardiac groove, two sets of three well-developed spines/nodes instead of cardiac ridges, with spine/node size decreasing antero-laterally. It it slightly deformed, but appears to be considerably wider (length/width c. 0-8) than the others (length/width c. 1-3 in undeformed specimens). Despite these differences, with the small sample it is considered here to be an unusual member of this taxon, and is perhaps an example of sexual dimorphism; more material might demonstrate that it is a different species. EXPLANATION OF PLATE 2 Figs 1-3. Tylocaris asiaticus gen. et sp. nov. 1, NIGP 126324A; x 5; 2, NIGP 126326A; x 5; 3, NIGP 126325; x 10-75 (arrows = pits). All from the Early Permian Tungtzeyen Formation, Changta village, Nanjing County, Fujian Province. PLATE 2 TAYLOR et al. , Ty locar is 824 PALAEONTOLOGY, VOLUME 41 table 2. Measurements in millimetres of Tylocaris asiaticus. Specimen Rostrum length Carapace length Abdomen length Abdominal segment width Telson length 1 2 3 4 5 126324 A- 1 ~ 16 14-0 7-0 5-7 50 4-2 ~ 6-5 126325 9-2 5-4 4-6 3-8 2-7 ~ 5-3 126326 2-7 7-6 126331 A-l 2-9 80 6-3 3-3 2-7 1-8 126334-1 2-7 9-6 METHODS A data matrix based on 33 morphological characters from 31 taxa (Table 3) was created using MacClade 3.01. Taxa were chosen based on several criteria. All 18 known pygocephalomorphs were included, with most data derived from the literature. Some information on British pygo- cephalomorphs was obtained from examination of material at the Hunterian and Kelvingrove museums in Glasgow, the National Museum of Scotland in Edinburgh, and the British Geological Survey in Key worth. PAUP 3.1.1 was used to perform a cladistic analysis of this matrix. Heuristic searches were the only practical option, due to the large size of the matrix and the high number of unknowns within it. After an initial unweighted analysis of the matrix, a successive reweighting option was employed, in which the unweighted matrix underwent an heuristic search and was then reweighted using the rescaled consistency index (RCI). This was in turn followed by another heuristic search, and so on until there was no further reduction in the minimum tree lengths obtained. This method provided a set of the most parsimonious trees for a matrix in which the most ‘important’ characters are granted the highest influence on the outcome of the analysis (see Table 4). Representative recent mysidacean and lophogastrid taxa were included in this analysis, as well as all known fossil mysid forms, to determine whether the new Chinese species were more closely associated to the similar mysidacean/lophogastrid forms than to the pygocephalids. A hypothetical ancestor was used as an outgroup, scored with zeros for all character states - a so-called Lundberg rooting. Whilst such a procedure is not regarded as an ideal solution to the outgroup problem, it proved useful in this analysis as there was no clear choice in the selection of an outgroup: the most obvious choice would be the mysids and lophogastrids, but since these taxa were actually included in the analysis, their use as outgroups would heavily bias the results. It is important to note that several alternative options were explored in these cladistic analyses, including the exclusion of certain ‘problematical’ taxa (i.e. N. tapscotti, and both Pygaspis species) whose positions appeared to be very unstable, the ordering of selected characters, the treatment of the lophogastrid and mysid taxa as outgroup taxa with the exclusion of the hypothetical ancestor from the analysis, and so forth. In each of these cases, the resolution of the tree as well as the consistency index (Cl) were reduced, suggesting that the set of trees described here, whilst far from perfect, is probably the best possible based on the currently available information. It is hoped that current work being done in South America by Professor Pinto and his associates (Pinto, pers. comm.) on some of the less well-known pygocephalomorph species, such as P. pachecoi , will provide more information on some of the more problematical taxa. This may, in turn, greatly improve the resolution and informational content of analyses of this difficult group. CHARACTERS To arrive at our cladistic analysis we assembled a list of 33 features based largely on carapace and tail fan morphology. The commonly incomplete pygocephalomorph specimens forced us to focus TAYLOR ET AL.: PERMIAN CRUSTACEANS 825 text-fig. 3. Reconstructions of: A, Fujicinocaris bifurcatus', b, Tylocaris asiaticus. Scale bars represent 5 mm. on these parts of the exoskeleton, which are those most often preserved and thus provide the majority of the taxonomic characters that are used to define genera and species. The characters and observations on them are listed below, and they include both binary and multi-state features. 1. Hepatic spines absent (0) or present (1). These spines constitute a frequently encountered set anterior to the cervical grooves. 2. Gastric spines absent (0) or present (1). This set of spines characterizes only the monotypic genus Anthracaris. 826 PALAEONTOLOGY, VOLUME 41 table 3. Data matrix used in the phylogenetic analysis (see Methods and Table 1 for information regarding the identity of the characters). Character 000000000 1 1 1 1 1 1 1 1 1 12222222222333 3 Taxon 12 3 456789012345678901234567890123 Hypothetical ancestor Lophogaster intermedius Gnathophausia longispina Paralophogaster glaber Eucopia unguiculata Chalaraspidum ala t urn Ceratolepis hamata Neognathophausia ingens Peachocaris strongi Schimperella beneckei Mysis flexuosa Pygocephalus cooper i Pygocephalus dubius Pygocephalus aisenvergi Tealliocaris woodwardi Pseudogalathea macconochiei Fujianocaris bifurcatus Tylocaris asiaticus Chaocaris chinensis Anthracaris gracilis Pseudo teal l iocaris caudafimbriata Pseuiotealliocaris etheridgei Pseudotealliocaris palinscari Jerometichenoria grandis Mamayocaris jepseni Mamayocaris jaskoski Notocaris tapscotti Paulocaris pachecoi Liocaris Pygapsis brasiliensis Pygaspis ginsburghi 000000000000000000000000000000000 000100010000005001001000100000000 01 10010200000060010121 1 1 102230200 01100002000000500?010002102000000 000001020000006001010002100000000 001001011000005000002000707300000 01000002000000600100200110?000100 001 10102000700601 101 1001 107230100 0 0 0 0 0 0 0 2 0 0 0 0 ? 7 7 7 7 7 0 0 1 0 0 I 7 0 0 0 0 0 0 0 0 000000001000004000011002107000000 000000000000003001000001100000000 011010010000221010012000001000100 01 101 1010020221010012000001000100 011000010020221010012001001000100 010001001020? 1210001 1001001230100 001 101000101 1 1 1000012000077300200 001101110111774101172 1 10217200111 001101170121776101172112117000111 0?110010012????????????????121200 1 1 1021000000121010012000001000200 01 10? 1000000 7 01 100772100701210200 01 10010010001 12100012010001210200 000001001000221 10001 1 100107230100 011000001030777777777777777000100 011010020000121000012000001000000 01 10100200001 I 101001 1000001000000 0 0 0 0 0 0 0 ? 0? 0 0 1 7 1 0 0 0 0 ? 7 0 0 ? 0? 7 0 0 0 0 0 0 001001010001777777772007072000100 000000020000???????? 107000200 0. 100 0000000000102210000? 100001 1000000 0 0 0 0 0 0 0 0 0 0 0 0 ? 7 7 7 7 7 7 ? 2 0 1 7 0 ? 1 0 0 0 0 0 0 3. Anterolateral spine absent (0) or present (1). These spines can mark the lateral extent of the optic notch on the anterior margin of the carapace. 4. Postero-lateral ‘process’ absent (0) or present (1). These variably developed spines can be found at the postero-lateral aspects of the carapace. 5. Branchiostegal spines/serrations absent (0), only on the anterior carapace margin ( 1 ) or along the entire carapace margin (2). These distinctive features can ornament either the anterior or the entire lateral margins of the carapace. 6. Mid-dorsal ridge/keel (extending between the cervical groove and posterior carapace margin) absent (0) or present (1). This forms the most prominent component of a complex series of possible grooves and ridges on the carapace of mysidacean-like pericarids. 7. Medio-lateral spines absent (0) or present (1). A set of spines on the posterior margin of the carapace just lateral to the mid-dorsal ridge or keel. 8. Cervical groove whole (0), split ( 1 ) or strongly posteriorly directed (2). This is the principal groove TAYLOR ET AL.. PERMIAN CRUSTACEANS 827 table 4. Final results of the Rescaled Consistency Index (RCI) reweighting of the characters used in this analysis. Character Final weight 1 Hepatic spines 1000 2 Gastric spines 133 3 Anterolateral spine 97 4 Posterolateral ‘process’ 200 5 Branchiostegal spines/serrations 444 6 Mid-dorsal keel 58 7 Medio-lateral spines 1000 8 Cervical groove 111 9 Cervical constriction 100 10 Marginal thickening 1000 11 Carapace papillations 200 12 Branchiostegal inflation 1000 13 Telson lobe number 389 14 Telson spine 267 15 Telson 1/w ratio 300 16 Telson medial ridge 400 17 Telson terminal process 400 18 Telson terminus 429 19 Uropod margin 1000 20 Uropod diaresis 100 21 Abdominal pleurae 81 22 Abdominal medial keel 63 23 Abdominal lateral keel 63 24 Length of sixth abdominal segment 127 25 Abdominal posterior narrowing 389 26 Abdominal segments visible 250 27 Sternal field 571 28 Primary lateral keels 286 29 Secondary lateral keels 563 30 Tertiary lateral keels 1000 31 Rostral keel 156 32 Cervical ridge 1000 33 Rostro-gastral ridge 1000 on the carapace of these crustaceans and stands in contrast to the more complex series of grooves seen on the carapace of decapod eucarids. 9. Constriction of carapace margin at cervical groove absent (0) or present (1). 10. Massive thickening of carapace margin absent (0) or present (1). This forms distinctive structures along the margin. 11. Surface papillations on the carapace absent (0), restricted to specific regions of carapace (1), covering entire carapace (2) or merged to form texture/sculpturmg (3). A multi-state feature typically useful in distinguishing between pygocephalomorph species. 12. Branchiostegal inflation absent (0) or present (1). It is difficult to categorize just what this feature represents. It is well developed in several genera. One could assume it bears some relationship to the possible development of gills in the branchiostegal chamber, but this cannot be easily confirmed in the fossils. It might also bear some relationship to streamlining necessary to facilitate surface flow over the thoracic region of the body. 828 PALAEONTOLOGY, VOLUME 41 13. Telson lobe/furca number zero (0), one pair (1) or two pairs (2). This and the following five characters often form a most coherent set of features for generic diagnoses in the order. 14. Telson spine absent (0), rounded (1) or pointed (2). 15. Telson length/width ratio < 0-5 (0), 0-5 1—1-0 (1), 1-01 — 1 -5 (2), 1-5 1—2-0 (3), 2-01-2-5 (4), 2-5 1—3 0 (5), > 3-01 (6). 16. Telson medial ridge absent (0) or present (1). 17. Telson terminal process absent (0) or present (1). 18. Telson terminus whole (0) or forked (1). 19. Uropod margins straight (0) or serrate (1). 20. Uropod diaresis absent (0) or present (1). 21. Abdominal pleurae absent (0), gently rounded (1) or angular (2). Insofar as they are preserved, decorative features of the abdomen (here and in the succeeding characters) can help to delineate species. 22. Abdominal medial keel/ridge absent (0) or present (1). 23. Abdominal lateral keels absent (0) or one pair (1). 24. Length of sixth abdominal segment same as fifth (0), slightly longer than fifth ( 1 ) or much longer than fifth (2). 25. Abdominal posterior narrowing: none (0) slight (1) or great (2). 26. Abdominal segments visible: six (0) or one or two covered (1). This feature actually reflects the degree of posterior development of the carapace. Typically the carapace covers only the thorax, but in some instances it extends backwards to cover the anterior part of the abdomen. 27. Sternal field narrow (0), wide and triangular (1) or wide and rectangular (2). This feature is not always evident, unless the ventral part of the thorax is preserved. It appeared (e.g. Schram 1986) that essentially only two forms of thoracic sternite field prevailed: narrow, with little development of sternites; or triangular, with narrow sternites anteriorly and wider ones posteriorly. In examination of some of the pygocephalomorphs from Brazil, it became clear that the observations of Pinto (1971) concerning wide anterior sternites on the thorax to form a more rectangular field have great value. Whilst this feature is unknown in many pygocephalomorph genera at present, we suspect that as more information becomes available this may prove to be a very important character for sorting higher relationships in the group. 28. Primary lateral keels absent (0), medio-lateral (1), gastro-lateral (2) or postero-lateral (3). 29. Secondary lateral keels absent (0), free (1), postero-lateral (2), close to lateral margin (3) or ‘fused’ with lateral margin (4). 30. Tertiary lateral margin absent (0) or present (1). 31. Rostral keel absent (0), not reaching cervical groove (1) or reaching cervical groove (2). 32. Cervical ridge absent (0) or present (1). 33. Rostro-gastral ridge absent (0) or present (1). RESULTS For the initial, unweighted analysis, a total of 30 most parsimonious trees with a length of 129 steps was found, with a Cl of 0-411. These trees, whilst showing some trends for specific groups in the analysis did not provide sufficient resolution to deduce relationships for all taxa involved, and thus we employed the use of the reweighting methods discussed above. This successive weighting regime provided a total of 15 most parsimonious trees of length 132, with a Cl of 0-402. A 50 per cent, majority rule tree for these trees is shown in Text-figure 4. Several interesting relationships emerged from this analysis. First, the recent and fossil mysids plus the recent lophogastrids form a distinct (if somewhat confused) clade, even when not specifically treated as an outgroup in the analysis. Within the pygocephalomorph ‘ingroup’, several distinct clades are evident which show considerable overall support for some of the taxonomic divisions outlined by Brooks (1962). As seen in the tree in Text-figure 4, the three species of Neognathophausia ingens TAYLOR ET AL.: PERMIAN CRUSTACEANS 829 2 2 3 -c "tc -5 -5 >■ C a a a ■£• a -c < ■« a ^ £. 8 ? U O O £ sj S Cl s ^ c > > > c 5 £ £ £ ^ § c text-fig 4. Strict consensus tree of the 15 trees obtained from an analysis incorporating successive reweighting of a pygocephalomorph and lophogastrid data set, using a hypothetical ancestor as outgroup. Character state changes are plotted (see Methods section for further details). Pygaspis ginsburghi 830 PALAEONTOLOGY, VOLUME 41 Pygocephalus form a monophyletic group with Anthracaris and both species of Mamayocaris. This closely reflects Brooks’ (1962) taxonomic scheme, in which Anthracaris and Mamayocaris are included with Pygocephalus in the family Pygocephalidae. One major disagreement with Brooks (1969) is the unification of Tealliocaris with the three species of Pseudotealliocaris to form a monophyletic clade; he had placed the latter in Pygocephalidae and the former in Tealliocarididae. Thus, Brooks’ generic distinction between Tealliocaris and Pseudotealliocaris may be an unnatural taxonomic separation. His familial separation of these genera is certainly suspect. Jerometichenoria is united by this analysis with this tealliocaridid clade, suggesting that the family Jerometi- chenoriidae, proposed by Schram (1978), may also be unnecessary. A close relationship seems to exist between the three Chinese forms, Fujianocaris and Tylocaris and the Carboniferous Chaocaris, and the British Pseudogalathea , and this may also extend to the problematical South American genera Paulocaris and Liocaris. This is perhaps the most interesting relationship to emerge from this analysis, as it could indicate taxonomic and palaeobiogeographical relationships between these geographically widely separated taxa. This result also contrasts with Brooks’ (1962) interpretation, in that his placement of Pseudogalathea with Tealliocaris in Tealliocarididae is not supported by this analysis. In addition, Brooks placed Paulocaris in Notocarididae with Notocaris, another association that does not appear to be supported by this analysis. Both the Pygaspis species and N. tapscotti occur basally in the pygocephalomorph ‘clade’, with no clear associations to any of the three major pygocephalomorph clades expressed in the analysis. We hope that a more adequate understanding of the anatomy of the southern hemisphere species will resolve the polychotomies in this part of the tree, and allow us to address definitively the issues of pygocephalomorph classification. DISCUSSION Age Beds of the Tungtzeyen Formation containing Fujianocaris bifurcatus also contain several other taxa, including plants, conchostracans, bivalves, brachiopods, gastropods, ammonoids, fusilinids and crinoids. These taxa collectively are the basis for the assignment of an Early Permian age for the Tungtzeyen Formation (Sheng et al. 1982). Morphology At the outset of this study, it was assumed that the two new species were members of the extinct Carboniferous/Permian order Pygocephalomorpha, based on overall morphology and similar time ranges. However, we also considered that they might be related to Recent mysids or, more likely. Recent Lophogastrida. There is a great number of morphological similarities between the latter and the Pygocephalomorpha. They have both been considered as sub-orders of the Mysida, and were elevated to the status of separate orders by Schram (1984). The main distinguishing characters for the Pygocephalomorpha are the presence of a triangular field of sternites on the ventral surface of the thorax and the development of a complex tail fan, including at least one pair of caudal furcae associated with the telson (Schram 1986). Since none of the specimens described here shows either ventral preservation or a complete fail fan, these unfortunately could not be used. Important characters that distinguish these new species from the morphologically similar pygocephalomorph Pseudogalathea are: the complex cervical and rostro-gastric ridges; the medial ridge of the telson in the tylocaridids; and the highly elongated postero-lateral spines of Pseudogalathea. Pygocephalus , another pygocephalomorph to which F. bifurcatus and T. asiaticus could be compared (Brooks 1962, 1969), is distinguishable from the tylocaridids by the absence of antero-lateral serrations on the carapace margin, the presence of a medial ridge on the telson, and the presence of carapace ridges. A third pygocephalomorph genus, Chaocaris, occurs in China (Shen 1983) and has several similarities to the tylocaridids, but is distinguished by its possession of a set of mid-lateral carapace keels, the absence of a medial carapace keel, and an elongate, narrow rostral TAYLOR ET AL.\ PERMIAN CRUSTACEANS 831 ridge extending from the anterior end of the carapace to the cervical ridge. The taxonomic placement of Chaocaris with the pygocephalomorphs is uncertain, as this taxon is based on a single carapace. T. asiaticus and F. bifurcatus also show similarities to mysidacean species known from the fossil record, in particular Schimperella beneckei and Peachocaris strongi. S. beneckei can be distinguished by its six exposed abdominal somites and its possession of a truncate telson that is shorter than its associated uropods (Hessler 1969), and P. strongi by its rounded abdominal pleurae, the exposure of all six abdominal somites, and the presence of large, rounded postero-lateral lappets on the carapace (Brooks 1962). One important morphological character that suggested to us a possible relationship between F. bifurcatus and T. asiaticus and the lophogastrids instead of the pygocephalomorphs is the apparent presence of a bifurcation at the terminal end of the telson, resembling a pair of terminal spines. This is a common occurrence in the order Lophogastrida but is generally absent among pygocephalo- morphs. This is, however, an uncertain character at best, due to the usually poor nature of preservation of the tail fan in these animals. Associated faunas and ecology Specimens of Tvlocaris asiaticus and Fujianocaris bifurcatus were collected from three different localities in south-east China: most are from the Early Permian Tungtzeyen Formation at Changta, Fujian, which comprises alternating thin beds of grey to dark grey, fine-grained quartz sandstone and siltstone, interbedded with mudstone and coal beds. The accompanying flora and fauna includes plants (Gigantonoclea fukiensis, Sphenophyllum sino-coreanum , Pecopteris ( Rajahia ) rigida , P. belitelioides, Sphenopteris tenuis. Aster ophyllites longifolius, Lobatannularia lingulata, Giganto- pteris dictyophylloides, Compsopteris sp., and Cordaites sp.), bivalves (Bakevellia ceratophaga, Wikingia elegans, Vosellina aff. yunnanensis, Astartella cf. ambiensis, Stuchburia sp. and Palaeoneilo sp.), brachiopods ( Cathaysia sp., Neoplicatifera sp., Linoproductus sp., Lingula sp. and Pygnochonetes sp.), gastropods ( Cyclozyga sp., Baylea sp. and BeUerophon sp.), ammonoids ( Altudoceras sp. and Schouchangoceras sp.), crinoids ( Cyclocylicus quinquelobus) and unidentified insect wing fragments. The flora at these south-east Chinese localities may represent the Late Palaeozoic Cathaysian flora (see Zhang and He 1985). The gigantopterids probably represent tropical woody climbers, carried to the site of deposition by streams or winds (Yao 1983). The brachiopod Lingula and the bivalves B. ceratophaga , Stuchburia sp. and V. aff. yunnanensis are all euryhaline forms, which lived in shallow marine settings. These floral and faunal characters, along with the lithological characteristics, suggests deposition in a nearshore marine environment, with possible repeated deepening cycles. This high-salinity environment may be largely responsible for the relative scarcity of specimens and their general incompleteness, as such shallow water fully marine faunas are rarely preserved in the fossil record (Schram 1981; Briggs and Clarkson 1989). It is perhaps due to the highly sclerotized nature of the carapace of F. bifurcatus that it is preserved in such high numbers, in comparison with T. asiaticus. There is considerable generic ( Sphenophyllum , Pecopteris , Sphenopteris and Asterophyllites ) and some specific overlap (A. longifolius) between this south-east Chinese flora and that of the Late Carboniferous Mazon Creek Essex assemblage, which has been interpreted as a nearshore marine fauna (Janssen 1965; Pfefferkorn 1979; Schram 19796). Whilst not closely related geologically during the Permian (Scotese and McKerrow 1990; Ziegler et ai, pers. comm.), southern China and continental southern North America were both located near the equator and probably shared tropical environments, which might account for the similar floras. The single specimen of F. bifurcatus from the Lower Permian at Xihushan, Fujian Province was found in dark grey mudstones, with no associated faunal or floral elements. There is a lack of data for this section, due to little collecting having been done. Based on its lithology, this unit is assumed to have been deposited in a coastal marine environment, similar to that inferred for the better known Lower Permian at Changta, Xiangshuping. 832 PALAEONTOLOGY, VOLUME 41 A further single specimen of F. bifurcatus was collected from an exposure of the Upper Permian at Shitangpu, Hunan Province where the Lungtan Formation is composed of yellowish to dark grey thin-bedded mudstone. It also contains brachiopods ( Spinomarginifera pseudosintanensis , Spinomarginifera sp., Leptodus tenuis, Martinia sp., Punctospirifer sp., Oldhamina sp., Haydenel/a sp., and Gubleria sp.) and bivalves ( Schizodus sp., Palaeoneilo sp., Nuculopsisl sp. and Stutchburial sp.). The absence of terrestrial or freshwater plant material indicates a system isolated from freshwater runoff. The relative abundance of brachipods, seemingly preserved in situ , suggests a quiet marine environment, as does the presence of exclusively fine-grained sediments, which were probably deposited in a nearshore marine or paralic setting, possibly lagoonal or a protected bay (Wang 1985; Zhang 1992). The occasional presence of carapaces of F. bifurcatus and T. asiaticus on closely associated bedding planes suggests that these two species lived in the same or closely associated communities. It is difficult to establish their role within these communities, however, as their preservation is insufficient to discern such features as mouthpart anatomy and thus insight into feeding type. These eumalacostracans may represent low-level carnivores, as suggested for seemingly similar forms by Schram (1981), but only the collection and description of further material, with better or new morphological details, can answer this question. Palaeobiogeography The placement of these new Chinese taxa into the order Pygocephalomorpha presents some new and difficult questions about Palaeozoic palaeobiogeography. Based on this order alone, there is evidently some palaeobiogeographical relationship between central North America, South America, South Africa, Great Britain, and, tentatively, southern China during the Carboniferous and Permian. However, it is not known whether this is due to similar ecological conditions, or to a true biogeographical connection. Whilst Permian maps (e.g. Scotese and McKerrow 1990) show similar latitudinal positions for several areas in which pygocephalomorphs occur (i.e. North America, Great Britain), there is no physical connection between these regions and the land masses destined to make up China. However, the same can also be said for the taxa found in such areas as South Africa and Brazil, which were not closely related palaeogeographically to North America and Great Britain. Thus, the question of historical biogeography for the order Pygocephalomorpha is a difficult one, regardless of the taxonomic position of the tylocaridids. One possibly important trend can be seen in the temporal distribution of the pygocephalomorphs. All of the 1 1 known Carboniferous species occur in close association with Laurentia. Conversely, of the six known Permian species, five have a Gondwanan distribution (the exception being Mamayocaris jepseni , which is Laurentian in origin). Thus, there was a general shift in the distribution of the pygocephalomorphs from the Laurentian to the Gondwanan coastal margins over the Carboniferous to the Permian, with the exception of isolated populations which remained in Laurentian waters throughout the Permian. This concurs with the observations of Schram (1977), who discussed malacostracan crustacean distributions during the Palaeozoic and the Triassic. He suggested a restriction to Laurentian waters during the Late Palaeozoic for the malacostracans, followed by an expansion of their distribution to other parts of the world with the formation of the Pangaean supercontinent during the Permian. The new information provided by the tylocaridid pygocephalomorphs clearly supports this observation. It is difficult to draw further conclusions about Palaeozoic palaeobiogeography, especially with respect to the pygocephalomorph crust- aceans, from the data as it currently stands. The same biogeographical problems exist, however, in the alternative hypothesis, in which the tylocaridids might be members of the Lophogastrida rather than Pygocephalomorpha. Little is known about the fossil record of the lophogastrids with the exception of the Carboniferous species Peachocaris strongi from North America and the Triassic species Schimperella beneckei from Alsace, France. Thus, the same problematical issue arises: trying to draw connections between the closely related North American and European regions to the distant Chinese land masses. TAYLOR ET AL.: PERMIAN CRUSTACEANS 833 Acknowledgements. This research was made possible by a grant (no. 750.195.17) from ‘de Stichting Geologisch, Oceanografisch en Atmosferisch Onderzoek’, Nederlandse Organisatie voor Wetenschappelijk Onderzoek (The Department of Geological, Oceanographic and Atmospheric Research, the Dutch Organization for Scientific Research). Thanks are also extended to Dr Ronald Jenner III, for his valuable assistance and comments on the manuscript. REFERENCES beurelen, K. 1930. Vergleichende Stammesgeschichte Grundlagen, Methoden, Probleme unter besonderer Beruksichtigung der hoheren Krebse. Fortschriften der Geologie und Palaeontologies 8, 317-464. 1931. Crustacean reste aus den Mesosaurierschichten (Unterperm) von Brasilien (Sao Paulo). Palaeontologische Zeitschrift, 13, 35-50. 1934. Die Pygaspiden, eine neue Crustaceen (Entomostracen) Gruppe aus den Mesosaurier-fuhrenden Iraty-Schichten Brasiliens. Palaeontologische Zeitschrift , 16, 122-138. Briggs, D. E. G. and clarkson, E. N. K. 1989. Environmental controls on the taphonomy and distribution of Carboniferous malacostracan crustaceans. Transactions of the Royal Society of Edinburgh: Earth Sciences, 80, 293-301. brooks, h. k. 1962. The Palaeozoic Eumalacostraca of North America. Bulletins of American Paleontology, 44, 163-338. 1969. Eocarida. R341-R345. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R. Arthropoda 4. Geological Society of America and University of Kansas Press, Lawrence. broom, R. 1931. On the Pygocephalus- like crustacean of the South African Dwyka. Proceedings of the Zoological Society of London, 571-573. chen jun-yuan, zhu mao-yan and zhou gui-qing 1995. The Early Cambrian medusiform metazoan Eldonia from the Chengjiang Lagerstatte. Acta Palaeontologica Polonica, 40, 213-244. zhou gui-qing and ramskold, L. 1995. The Cambrian lobopodian Microdictyon sinicum. Bulletin of the National Museum of Natural Science, 5, 1—93. Clarice, J. M. 1920. Crustacea from the Permian of Sao Paulo, Brazil. Bulletin of the New York State Museum, 219/220, 135-137. hessler, r. r. 1969. Pericarida. R360— R393. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part R. Arthropoda 4. Geological Society of America and University of Kansas Press, Lawrence. huxley, T. h. 1857. Description of a new crustacean ( Pygocephalus cooperi ) from the Coal Measures. Quarterly Journal of the Geological Society, London, 13, 363-369. janssen, R. E. 1965. Leaves and stems from fossil forests. Popular Science Series 1. Illinois State Museum. latrielle, p. a. 1802. Histoire naturelle generate et particuliere, des crustaces at des insects. 3. F. Dufort, Paris, 486 pp. peach, B. N. 1883. Further researches among the Crustacea and Arachnida of the Carboniferous rocks of the Scottish Border. Proceedings of the Royal Society of Edinburgh, 30, 511-529. pfefferkorn, h. w. 1979. High diversity and stratigraphic age of the Mazon Creek flora. In nitecki, m. h. (ed.). Mazon Creek fossils. Academic Press, New York, 581 pp. pinto, i. D. 1971. Reconstituigao de Pygaspis Beurlen, 1934 (Crustacea - Pygocephalomorpha). Sua Posigao Sistematica, seu significado e de outros fosseis para o Gondwana. Anais da Academia Brasil Ciencias, 43 (supplemento). prestwich, j. 1840. On the geology of the Coalbrookdale. Transactions of the Geological Society, London, 5, 413-495. salter, j. w. 1861 . On some higher Crustacea of the British Coal Measures. Quarterly Journal of the Geological Society, London , 17, 528-533. 1863. On some fossil Crustacea from the Coal-measures and Devonian rocks of British North America. Quarterly Journal of the Geological Society, London, 19, 76. schram, f. R. 1974a. Mazon Creek caridoid Crustacea. Fieldiana: Geology, 30, 9-65. - — 19746. Convergences between Late Paleozoic and modern caridoid Malacostraca. Systematic Zoology, 23, 323-332. - 1977. Paleozoogeography of Late Paleozoic and Triassic malacostracans. Systematic Zoology, 26, 367-379. — 1978. Jerometichenoria grandis n. gen., n. sp. (Crustacea: Mysidacea) from the Lower Permian of the Soviet Union. Journal of Paleontology, 52, 605-607. 1 979a. British Carboniferous Malacostraca. Fieldiana: Geology, 40, 1-129. 834 PALAEONTOLOGY, VOLUME 41 — 19796. The Mazon Creek biotas in the context of a Carboniferous faunal continuum. 159-190. In nitecki, m. h. (ed.). Mazon Creek fossils. Academic Press, New York, 581 pp. — 1981. Late Paleozoic crustacean communities. Journal of Paleontology , 55, 126-137. 1984. Relationships within eumalacostracan crustaceans. Transactions of the San Diego Society of Natural History , 20, 301-312. scotese, c. R. and McKERROW, w. s. 1990. Revised world maps and introduction. 1-21. In McKerrow, w. s. and scotese, c. R. (eds). Palaeozoic palaeogeography and biogeography. Memoir of the Geological Society, 12, 1^135. SHENG JINZHANG, JIN YUGAN, RUI LIN, ZHANG LINXIN, ZHENG ZHOUQUAN, WANG YUJING, LIAO ZHOUTING and zhao jiamin 1982. [Correlation chart of the Permian in China with explanatory text.] 153-170. In [Stratigraphic correlation chart in China with explanatory text.] Science Press, Beijing. [In Chinese.] shen yan-bin 1978. Leaid conchostracans from the Middle Devonian of south China with notes on their origin, classification and evolution. Papers for the International Symposium on the Devonian System, 1-15. — 1981. Cretaceous conchostracan fossils from eastern Shandong. Acta Palaeontologica Sinica, 20, 518-526. — 1983. A new pygocephalomorph genus (Eumalacostraca) of Lower Carboniferous from Anhui. Acta Palaeontologica Sinica, 22, 663-668. — 1984. Occurrence of Permian leaid conchostracans in China and its palaeogeographical significance. Acta Palaeontologica Sinica, 23, 505-513. — 1990. A new conchostracan genus from Lower Permian Tungziyan Formation, Fujian. Acta Palaeontologica Sinica, 29, 309-314. wang hungzhen 1985. Atlas of the palaeogeography of China. Cartographic Publishing House, Beijing, 300 pp. [In Chinese and English.] woodward, H. 1879. On Necroscilla wilsoni a supposed stomatopod crustacean from the Middle Coal Measures, Cossall, near Ilkeston. Quarterly Journal of the Geological Society, London, 35, 551-552, pi. 26. yao zhaoqi 1983. Ecology and taphonomy of gigantopterids. Bulletin of the Nanjing Institute of Geology and Palaeontology, Academia Sinica, 6, 63-84. zhang shanzhen and he YUANLiANG 1985. Late Palaeozoic palaeophytogeographic provinces in China and their relationships with plate tectonics. Palaeontologia Cathayana, 2, 77-86. zhang wentang, shen yanbin and niu shaowu 1990. Discovery of Jurassic conchostracans with well- preserved soft parts and notes on its biological significance. Palaeontologia Cathayana, 5, 311-352. zhang yangyu 1992. Sedimentary environments and its evolution of Tungziyan Formation in Fujian. Coal Field Geology of China , 4, 20-25. zhu tong 1990. The Permian coal-bearing strata and palaeobiocoenosis of Fujian. Geological Publishing House, Beijing, 127 pp. ROD S. TAYLOR FREDERICK R. SCHRAM Institute for Systematics and Population Biology University of Amsterdam P.O. Box 94766 1090 GT Amsterdam The Netherlands e-mail taylor@bio.uva.nl e-mail schram@bio.uva.nl shen yan-bin Nanjing Institute of Geology and Palaeontology Academia Sinica 39 East Beijing Road, Nanjing The People’s Republic of China Typescript received 16 April 1997 210008 Revised typescript received 18 September 1997 e-mail LPSNIGP@nanjing.jspta.chinamail.sprint.com TH RE E-DIMENSION ALLY MINERALIZED INSECTS AND MILLIPEDES FROM THE TERTIARY OF RIVERSLEIGH, QUEENSLAND, AUSTRALIA by IAN J. DUNCAN, DEREK E. G. BRIGGS and MICHAEL ARCHER Abstract. An assemblage of three-dimensionally preserved insects and millipedes from the late Oligocene/ early Miocene limestones of Riversleigh (north-west Queensland) augments a sparse Tertiary insect record from Australia. The fauna includes four species of Coleoptera, one of Trichoptera represented only by the larva, and a myriapod. The arthropods are uncompacted and have been replicated in calcium phosphate. Early phosphatization has preserved original structures such as the overlapping layers and helicoidal pore canals of the procuticle, and wrinkles in the arthrodial membrane. The most remarkable preservation is of the ocular apparatus. The hexagonal lenses and their rhabdom emplacements are preserved in the Coleoptera. The trichopteran larva displays an unusual form of compound eye, consisting of large, separate circular lenses. Where the cornea has been lost, an irregular lattice of ‘cups’ is exposed. This is the first example of this ‘ schizochroal’-type eye reported in a fossil insect. Bacteria and fungi associated with the decay of the insects are themselves mineralized. Despite the great success of insects through time in terms of both abundance and diversity, surprisingly little is known about the fossil insects of Australia. The majority of remains so far uncovered consist of wings or indeterminate fragments. Only rarely are wings attached to bodies and even when this occurs, most fossils are preserved in a crushed and distorted condition (Riek 1970a). Approximately 350 fossil species in 19 orders have been described from Australia, covering a period extending from the Upper Permian to the Pliocene (for reviews see Riek 1970a, 19706; Jell and Duncan 1986). However, this record includes only three significant Tertiary insect deposits besides those at Riversleigh (Text-fig. 1): Redbank and Dinmore, southern Queensland (Riek 1967) and Vegetable Creek, NSW (Riek 1954). The information provided by these sites is limited by the quality of the fossil material. The Lower Tertiary Redbank assemblage is dominated by Coleoptera and Homoptera, but Blattodea, Hemiptera-Heteroptera, Neuroptera, Mecoptera and Diptera are also represented (Riek 1970a). Insects are much rarer in the Dinmore assemblage, with single wings of an orthopteran, an isopteran, a homopteran and an odonatan comprising the total fauna (Riek 1970a). The Vegetable Creek fauna is restricted to immature aquatic insects, predominantly Ephemeroptera and Diptera. Only beetle elytra have been recovered from other Tertiary sites (Riek 1970a). The remarkable three-dimensionally preserved insects from Riversleigh represent an important addition to this Tertiary record. THE RIVERSLEIGH FAUNA Riversleigh’s major importance is the data that it provides on the Tertiary mammals of Australia (Archer and Bartholomai 1978; Archer and Clayton 1984), nearly trebling the species of this age previously recorded for the entire continent (Archer et al. 1994a, 19946, 1995). The insects were first discovered in the course of processing the vertebrate-packed limestone in search of small bones and teeth. Acetic acid digestion may have introduced a bias, as specimens which are not phosphatized would have been destroyed. Specimens which preserve organics were recovered from other Riversleigh sites (D. A. Arena, pers. comm.). Coleopteran specimens are rare. At least four different species, each belonging to a different family, have been found. Fragmentary larval specimens of a (Palaeontology, Vol. 41, Part 5, 1998, pp. 835-851] © The Palaeontological Association 836 PALAEONTOLOGY, VOLUME 41 text-fig. 1 . Location map. Left, detail of Upper site, Godthelp’s Hill and geology of the surrounding area. Right, locations of Riversleigh and principal Tertiary insect-bearing Lagerstatten in eastern Australia (after Megerian 1992). single species of Trichoptera are much more common. A myriapod, and a previously documented isopod (Archer et al. 1994 a), complete the arthropod fauna as presently known. Locality and stratigraphy Although Riversleigh was originally thought to represent a single, distinct mid Miocene assemblage (Tedford 1967), fossils have now been recovered from over 100 different sites, ranging in age from late Oligocene to near present day (Archer et al. 1989, 1994a, 19947>, 1995). For a detailed discussion of the stratigraphy, see Megerian (1992). The arthropods reported here were recovered from one locality, the Upper Site of Godthelp's Hill (Text-fig. 1). Arthropods are also known from other Riversleigh sites (Camel Sputum and Dunsinane sites). At the Upper Site, the vertebrate fauna is particularly diverse, with almost twice the number of marsupial species as any surviving Australian ecosystem, as well as a diverse range of birds, reptiles and amphibians (Archer et al. 1989, 1994a). The Upper Site is interpreted as a former shallow (c. 1 m deep), lime-rich pool in a rain forest (Archer et al. 1994a), where levels of calcium carbonate were sufficient to result in the precipitation of a thin peripheral crust. DUNCAN ET AL.: TERTIARY INSECTS AND MILLIPEDES 837 text-fig. 2. Micro-environments of preservation in a trichopteran larva (QM F34592). A, cuticular fragment from within the tail assemblage displaying distinct chevron structures arranged in parallel rows; structural detail is lost to the lower right; x 65000. b, cuticle a few micrometres distant to ‘a’; the structures have been replicated by plates of calcium phosphate such that the gross structure is retained while the detail is lost; x 65000. c, ‘nodular’ surface of the dorsal cuticle; x 25000. d, replacement of the nodules within a few micrometres of ‘c’ with plates of calcium phosphate; x 30000. Specimens are gold coated. Accumulations of piles of crystalline shards, interspersed with animal remains, are common in many of the Riversleigh deposits. The crystal crust may have given the appearance of firmness, only to give way under the weight of an animal that strayed onto it, which then drowned. The Upper Site limestone is characterized by black, iron-rich bands that may reflect periods of anaerobic conditions, where an absence of oxygen inhibited scavengers, contributing to the lack of disarticulation of the vertebrate remains. Wrinkled sheets, interpreted as algal mats, have also been recovered from acid residues (Archer et al. 1994a). The trichopteran larvae are found in tube-like extensions of this mat-like material. The range of vertebrates found at the Upper Site suggests that all the surrounding micro-environments are represented, from tree tops (many possums) to the forest floor (wynyariids, macropopoids, perameloids, etc.), and the water itself (frogs) (Archer et al. 1994 a). TAPHONOMY The most striking feature of the Riversleigh arthropods is their preservation in three dimensions. However, only the more recalcitrant tissues have survived; the internal soft tissues have decayed. 838 PALAEONTOLOGY, VOLUME 41 text-fig. 3. Ornamentation of cuticle of trichopteran larva (QM F34587). A, prescutum with irregular pattern of pits; a spiracle is also present; x 4500. b, interface between the scutum, showing sites of hair emplacement in its upper half, and the scutellum; the scutellum is more regularly patterned than the prescutum; x 5000. c, scutum and scutellum, overlapping the next segment (to the right); x 4000. D, limb of first thoracic segment, surrounded by arthrodial membrane; x 700. e, arthrodial membrane; x 1750. f, close-up of hair emplacement; x 2250. text-fig. 4. Underlying arrangement of microfibrils within the cuticle of coleopteran species C (QM F34582). a, longitudinal arrangement of microfibrils; x 13000. b, longitudinal and cross sectional arrangement of microfibrils; x 32500. The overlapping layers that make up the cuticle are preserved, as are the ocular framework, rhabdom and individual lenses of the eye. Microprobe analysis confirms that the specimens are preserved in carbonate-fluorapatite. Some specimens are infilled with detrital matter that appears to have been phosphatized contemporaneously. DUNCAN ET AL.\ TERTIARY INSECTS AND MILLIPEDES 839 text-fig. 5. Canals within the cuticle of the trichopteran larva (QM F34591 ). a, series of parallel canals exposed in tail segment; x 9000. B, helicoidal pore canals (cf. Bouligand 1965) in a cuticle fragment; x 27500. Cuticle Structures less than 1 //m in dimensions are preserved on the surface of the cuticle (Text-fig. 3). The distinctive arrangement of the microfibrils (Text-figs 4, 9a) is evident in section, as is the helicoidal structure observed by Bouligand (1965) and Neville et al. (1969) in living insects (Text-fig. 5b). The orientation of the crystals follows that of the structural proteins and other biomolecules within the cuticle, emphasizing the high fidelity of replication by calcium phosphate. Parallel canals can be discerned within the tail segment of one of the larvae (Text-fig. 5a). The cuticle of the trichopteran larva displays distinct chevron structures arranged in parallel rows (Text-fig. 2a) just a few micrometres distant from a point where the cuticle has been replaced by plates of phosphate that replicate the gross structure but obliterate the detail (Text-fig. 2b). A similar phenomenon is evident where the nodular patterning of the larval cuticle (Text-fig. 2c) is replaced within a few micrometres by plates of phosphate (Text-fig. 2d). This indicates that the conditions under which decay and mineralization took place varied on a sub-millimetre scale (see Martill 1988; Briggs and Kear 1994). Eye Perhaps the most striking evidence of the fidelity of preservation occurs in the eyes (Text-fig. 6; Duncan and Briggs 1996). The insect compound eye is composed of arrays of ommatidia - the basic visual unit (Snodgrass 1935; Text-fig. 6f). Each ommatidium is composed of a dioptric apparatus and rhabdom, isolated from the next ommatidium by pigment cells; the dioptric apparatus of lens and crystalline cone controls the focusing of light. The cuticle of this apparatus is composed predominantly of chitin, which differs ultrastructurally from that surrounding it (Neville 1970). The rhabdom contains the visual pigments that trigger impulses in the optic nerve (Snodgrass 1935). The entire ocular apparatus of the Riversleigh specimens is replicated in calcium phosphate. The eyes exhibit varying degrees of alteration through decay and diagenesis. Where the lenses (diameter 30 jum) are preserved, the characteristic hexagonal-packing arrangement of the compound eye is apparent (Text-fig. 6a, i). With the deflation of an individual lens through decay, crystal aggregates are evident in the interior. Where the dioptric apparatus is lost, the rhabdom emplacement is exposed (Text-fig. 6b, d, k). In cross section the rhabdoms can be discerned 840 PALAEONTOLOGY, VOLUME 41 text-fig. 6. Compound eyes, a, i, coleopteran species A; b-d, k, coleopteran species B; G, j, coleopteran species C; f, h, trichopteran larva, a, lenses displaying characteristic hexagonal packing (QM FI 6648); x 2500. b, framework of pigment cells revealed by loss of dioptric apparatus (QM F34583); x 7000. c, walls of rhabdomal emplacement lined with bacteria (QM F34583); x 20000. D, framework of the eye (QM F34583); x 1500. e, schematic of ommatidium of apposition eye. Abbreviations: C, corneal lens; CC, crystalline cone; PYPC, primary pigment cells; SPC, secondary pigment cells; RC, retinal cells; R, rhabdom. f, ‘schizochroal- type’ eye (QM F34584); x 2750, G, cross section through rhabdomal emplacement showing ommatidial cups (QM F34582); x 2000. h, close-up of ‘schizochroal-type’ eye, showing ommatidial cups (QM F34584); x2750. i, complete eye (QM F16648); x 720. J, complete eye (QM F34582); x 2500. k, complete eye (QM F34583); x480. DUNCAN ET A L. : TERTIARY INSECTS AND MILLIPEDES 841 text-fig. 7. Fossil fungi and bacteria, a, fungal hyphae criss-crossing the interior of head and thorax of larva (QM F34584); x 1750, b, cross section through single fungal strand (QM F34584); x 9000. c, fungal colony covering sternite of larva (QM F34592); x 225000. d, bacterial mat covering surface of coleopteran wing (QM F34595); x 600. extending radially from the surface of the eye (Text-fig. 6g, j). With the rhabdoms stripped away, the ommatidial cups, the concave receptacles of the ocular apparatus, are evident. The pigment cells that once lined the rhabdom interior of one specimen have been destroyed by bacteria, which are now preserved as rod-like protuberances from the walls (Text-fig. 6c). Bacteria are also preserved enshrouding the hind wings of a beetle (Text-fig. 7d). Mineralized bacteria have been reported from other localities (e.g. Messel; Wuttke 1983). The trichopteran larva displays an unusual form of compound eye consisting of large, separate circular lenses rather than the more usual closely packed hexagonal type (Text-fig. 6f, h). Where the cornea and dioptric apparatus have been lost, an irregular arrangement of ‘cups’ is exposed. This ‘schizochroal-type’ eye occurs in only a small number of living insects (Kinzelbach, 1967 ; Clarkson 1979; Paulus 1979; see Horvath et al. 1997 for a review), where it is thought to maximize light reception. Caterpillars of the Lepidoptera, the sister group of Trichoptera, normally have six isolated biconvex lenses, widely distributed on each side of the head. Coleopteran and megalopteran larvae can have up to six well-separated stemmata on each side of the head, the Neuroptera and Raphidioptera up to seven, and the Strepsiptera five. Larvae of most Mecoptera have dispersed faceted eyes consisting of 30-35 typical ommatidia. Fungi A trichopteran larva preserves fungal hyphae in the head capsule and trunk (Text-fig. 7a). Individual strands criss-cross the interior of the insect displaying a simple lateral dichotomous 842 PALAEONTOLOGY, VOLUME 41 branching (Text-fig. 7a). In cross section (Text-fig. 7b) an individual strand shows a structureless core (12 //m in diameter) surrounded by a layer displaying a distinct radial pattern (17 //m broad). The outer layer represents a crystalline overgrowth around the original fungal strand, which is presumably represented by the core which has a diameter (12 //m) comparable to that of modern hyphae (10 //m). Extant hyphae consist of an outer wall of hemicellulose or chitin around a cavity, the strands forming a filamentous system (Talbot 1971). An unusual fungal growth is also noted covering the external surface of a larval segment, where a number of strands appear to radiate from a central point, each joined by short lengths to form a distinctive meshwork (Text-fig. 7c). Environmental conditions The exceptional preservation of the Riversleigh insects raises several questions regarding both the rate and mechanism of mineralization. Most models for the preservation of non-mineralized tissues require rapid burial, anoxicity, or both, in order to preclude scavenging (Seilacher et al. 1985). The presence of a surface crust and algal mats at the Upper Site at Riversleigh would have inhibited circulation and promoted anoxicity. Only the more recalcitrant tissue (i.e. cuticle, or calcified cuticle in the case of the myriapod) is preserved and this, coupled with the presence of bacteria within the rhabdom emplacements and fungal hyphae in the head capsule of one of the specimens, suggests that decay proceeded for some time prior to mineralization. The fungal strands criss-cross the interior of some specimens indicating that the fungus colonized the carcass after the internal tissue was lost through decay. It is clear that the limited number of arthropod species recovered from this site cannot reflect the total diversity of this rain forest environment. Those taxa (and life stages) that are preserved, and survived the acid digestion, are probably the more readily phosphatized elements of the biota. Their higher preservation potential may reflect the original biochemistry of the cuticle. Other examples of three- dimensionally preserved insects Three-dimensional preservation normally relies on sufficiently early mineralization to prevent collapse through decay, and to protect the fossil from overburden-induced compaction. Thus insects preserved as organic remains are rarely three-dimensional, except in conservation traps ( sensu Seilacher el al. 1985) such as amber (Poinar and Hess 1982; Henwood 1992, 1993; Grimaldi et al. 1994) and asphalt (Miller 1983; Stock 1992; Stankiewicz et al. 1997). The insects of the Oligocene Bembridge Marls, Isle of Wight, England (Jarzembowski 1980) are an exception. Here they are preserved essentially as a void left by the decayed internal tissues, lined with the cuticle, which is represented by a micrometre-thick, highly altered, organic layer. Most three-dimensionally preserved insects occur in early diagenetic concretions, such as the siderite nodules that are known from a variety of Carboniferous sites (see Bolton 1905; Woodward 1907; Heyler 1980; Baird et al. 1985a, 19856), notably at Mazon Creek in north-eastern Illinois (Richardson 1956; Johnson and Richardson 1966; Nitecki 1979; Baird et al. 1985a). These Carboniferous insects rarely preserve ultrastructural details of the cuticle (see Baird et al. 1985a) in contrast to those preserved in Tertiary concretions. Calcareous nodules from the Miocene of Barstow, California (Palmer 1957) exhibit micrometre-scale replication of the cuticle and internal tissue by a suite of minerals including quartz, apatite, celestite, gypsum and zeolite (Park 1995). The concretions of the Eocene London Clay, England, which are composed of pyrite, apatite or calcite (Britton 1960; Allison 1988), have yielded various beetles (Britton 1960), and a pyritized maggot (Rundle and Cooper 1971) which preserves surface details of the cuticle, but not the internal tissues. Phosphalic concretions from the mid Tertiary Dunsinane Site at Riversleigh, preserve insects (D. A. Arena, pers. comm.). DUNCAN ET AL. : TERTIARY INSECTS AND MILLIPEDES 843 The Riversleigh insects are exceptional in that phosphatization of the cuticle has led to three- dimensional preservation without the formation of a concretion. Phosphatized insects have also been reported from the Eocene Quercy Phosphorites of France (Handschin 1944), and the Oligocene fissure fillings of Ronheim, Germany (Hellmund and Hellmund 1996), but in both cases crystallization is coarser than at Riversleigh and less detail is preserved. A similar style of preservation in calcite is known from the Miocene volcanic deposits of Rusinga and M’fwangano Islands, Lake Victoria, Kenya (Leakey 1952, 1963) but only the gross morphology of taxa with thickened cuticles, such as millipedes and beetles, is preserved. Calcified millipedes are also known from Holocene cave deposits in the West Indies (Donovan and Veltkamp 1994), but the cuticle is likely to have been biomineralized in life. SYSTEMATIC PALAEONTOLOGY The morphological terminology and classification used here is that of The Insects of Australia (C.S.I.R.O. 1992). An open nomenclature is employed, as identification to the lowest taxonomic level is impossible due to the incomplete nature of the specimens. The specimens of Coleoptera lack appendages, including wings and elytra. In some cases only the head and thorax have been recovered. Thus identification must be based primarily upon the emplacement of the coxae and features of the head. No complete specimen of the trichopteran larva has been found. Details are often obscured by debris adhering to the ventral surface which cannot be removed without damage to the specimen. The only myriapod specimens are undifferentiated segments. The specimens are held in the Queensland Museum, Brisbane, to which the abbreviation QM refers. Phylum ARTHROPODA Superclass hexapoda Latreille, 1825 Class insecta Linne, 1758 Subclass pterygota Brauer, 1885 Division endopterygota Sharp, 1899 Order coleoptera Linne, 1758 Suborder polyphaga Emery, 1886 Superfamily curculionoidea Latreille, 1802 Family curculionidae Latreille, 1802? Coleopteran species A Text-figures 6a, i, 8a Material. QM F16648, QM F34585, incomplete adults with only damaged head and prothorax present. Description Head. The head is large and produced forward into a rostrum, which is longer than broad. The compound eyes are large, bulging and situated dorsolaterally at the base of the rostrum. Thorax. The prothorax is broad, half as long as wide, with gently convex lateral margins. The anterolateral corners of the pronotum project to form protective ‘shoulders’ about the head. The prosternum is bounded laterally by concave sternopleural sutures. The posterior margin of this plate curves round and between the fore coxae. The first pair of coxae are contiguous and meet along the midline. They are globular in shape and incorporate a lateral facing concavity to accommodate the femur. Dimensions. Maximum length of head and prothorax: 5 mm. 844 PALAEONTOLOGY, VOLUME 41 text-fig. 8. Coleoptera. A, species A, undetermined curculionid (QM FI 6648); x 100. b, species B, undetermined polyphagan (QM F34583); x 120. c, species C, undetermined histerid (QM F34582); x 120. D, species D, undetermined ommatid (QM F34595); x 60. Remarks. This species is referred to Curculionidae on the basis of its stout rostrum, large eyes toward the rostral base and contiguous, projecting, fore coxae. The incompleteness of the specimens prevents a more detailed interpretation. Superfamily and Family indet. Coleopteran species B Text-figures 6b-d, k, 8b Material. QM F34583, an incomplete adult, with only head and prothorax intact; QM F34586, an incomplete adult, consisting only of the pronotum. Description. The body is highly convex in cross section. Head. The head is hooded by the pronotunr (Text-fig. 8b) and is all but concealed from above. The anterior margin is gently convex. The large, bulbous compound eyes are ventrolateral in position, and approach the anterior margin (Text-fig. 6k). The mouth is hypognathous. Thorax. In plan view, the prothorax is a longitudinally elongate semicircle. The anterior margin is convex. On the ventral surface the sternopleural suture of the prosternum runs from the lateral margins of the head to the DUNCAN ET AL.\ TERTIARY INSECTS AND MILLIPEDES 845 coxae of the first limbs. These sutures mark the lateral margins of the prosternum, which is bounded anteriorly by the head and posteriorly by the transverse suture of the mesosternal plate. The plate rises to an elevated process between the first pair of limbs. The sternopleural sutures of the mesothorax form a gently curved semicircular outline. The thorax slopes rapidly from the sternopleural sutures to the lateral margin of the prothorax. Dimensions. Maximum length of head and pronotum: 5 mm. Remarks. The specimens show a number of characters that support assignment to the Polyphaga: notopleural sutures are absent on the prothorax, the ventral portion of the notum (hypomeron) is joined directly to the sternum on each side along the notosternal suture, and the pleuron is reduced and concealed. Insufficient detail is preserved to allow a more detailed taxonomic assignment. Superfamily hydrophyloidea Latreille, 1802 Family histeridae Latreille, 1802 Coleopteran species C Text-figures 4a-b, 6g, j, 8c Material. QM F34582, an almost complete adult, with only mid and hind legs missing. Description. The outline of the body is a near perfect oval. Head. The head is small (less than one-eighth body length) and sub-rectangular in outline, almost half as long as wide. It is sunk deeply into the pronotum and is concealed when viewed from above. The eyes are flattened and occupy the entire lateral margin of the head, approaching the anterior margin (Text-fig. 6j). The mouth is hypognathous. Thorax. A distinct pronotum, narrower than the meso- and metathorax, hoods the head. When viewed from above it appears rectangular in outline, and extends laterally beyond the head. The outline of the pronotum tapers gently from the posterior to the anterior margin. On the anterior margin of the prosternum is a raised median process which becomes a ridge running the length of the prosternum, decreasing in height as it does so. The first pair of limbs immediately flanks this ridge. The fore coxae, although partially obscured by the encrusted tibia, appear both large and transverse. Fore trochantins appear absent. The pronotum and the mesonotum are united along a transverse suture. The mesosternum is bound laterally by the coxae of the second pair of limbs, which appear to open laterally. Its anterior margin is marked by the boundary between the pronotum and the mesonotum, its posterior by the metasternal transverse anterior suture. The metasternal surface is divided along the midline by the longitudinal suture. Both the fore and hind coxae incorporate a concavity to accommodate the femur. Abdomen. The abdomen tapers gently posteriorly, forming a rounded pygidium. The elytra are truncate leaving the propygidium and pygidium exposed. There are five ventrites. Dimensions. Maximum length of beetle: 5 mm. Remarks. The ovoid body shape, truncate elytra exposing two complete tergites, and head all but concealed by the pronotum, are indicative of superfamilies Hydrophyloidea and Staphylinoidea. A median metasternal suture is unknown in the latter. The separation of the mid-coxae by more than the width of one coxa, and the wider separation of the hind coxae, indicate that the species belongs to the family Histeridae, and not Hydrophilidae. 846 PALAEONTOLOGY, VOLUME 41 text-fig. 9. Cuticle and wing of ommatid (coleopteran species D) (QM F34595). a, break in cuticle reveals distinct alignment of microfibrils; x 9000. B, terrace-like pattern of cuticle between wings on ventral surface; x 6000. c, close-up of wing; x 6000. Suborder archostemata Kolbe, 1908 Superfamily cupidoidea Latreille, 1802 Family ommatidae Newman, 1839 Coleopteran species D Text-figures 7d, 8d, 9a-c Material. QM F34595, an almost complete adult, missing head and prothorax, with the wings covering much of the dorsal surface of the body (Text-fig. 8d). Description Thorax. The lateral margins of the mesothorax are parallel for much of their length, but begin to converge gently towards the anterior. The mesothoracic coxae are contiguous, globular in shape, with a posterior-facing concavity. A median suture divides the metasternum. The meta-coxae are also globular and adjoin the anterior margin of the metasternum, the margin of which curves between and around them. The lateral metasternal sutures diverge gently from the mesothoracic coxae, so that the metasternum increases in width posteriorly. The metathoracic coxae are larger than the mesothoracic but are not contiguous. The cuticle displays distinct lineation (Text-fig. 9a). The dorsal surface is almost entirely shrouded by the exposed hind wings. They slope from the anterior ‘shoulders' toward the midline. The gap separating them decreases in width posteriorly. It extends one-quarter of the length of the thorax, at which point the wings meet. The cuticle within the gap displays a distinct terracing (Text-fig. 9b). At the anterior margin is a small pinnacle, posterior of which is a narrow ridge which runs the length of the gap. The remains of the wings shrouding the dorsal surface show traces of venation (Text-fig. 9c), but the detail is obscured by a coating of mineralized bacteria (Text-fig. 7d). Abdomen. The abdomen consists of five segments, and has a distinct blunt appearance, the lateral margins tapering gently to a rounded pygidium. The first ventrite, which is the largest, curves around and between the metathoracic coxae. The other four are smaller, and all are of similar length. Dimensions. Maximum length of specimen, 7 mm. Remarks. This species is assigned to the suborder Archostemata on the basis of the metathoracic trochantins. The presence of five ventrites indicates that the species belongs to either Cupedidae or Ommatidae. The lack of grooves on the ventral surface to accommodate the legs precludes assignment to the former, and indicates that this species belongs to the latter. DUNCAN ET AL.. TERTIARY INSECTS AND MILLIPEDES 847 text-fig. 10. Trichopteran larva, a, composite image of head, thorax and upper abdomen (QM F34587); x 80. b, ventral view of head (QM F34593); x 100. c-d, tail assemblage (QM F34594); x 140. Order trichoptera Kirkby, 1815 Superfamily and Family indet. Text-figures 2a-d, 3a-f, 5a-b, 6f, h, 10a-d Material. QM F34584, QM F34587-QM F34594, all of which are incomplete larval stages. Description. In cross section the dorsal surface is strongly curved, whilst the ventral surface is flattened. Head. The head is globular in shape with a slightly flattened anterior margin, and is broader than long. It is marked by two large ventrolateral antennal sockets, which protrude downwards. The scape of the antenna is large, and circular in cross section. The lateral epicranial sutures arise at the posterior margins of the head, pass around the sockets of the antennae on the lateral side and converge to form the median suture. The general outline of the suture is that of an inverted ‘ Y \ The epicranium is patterned by a random arrangement of setae emplacements. Five ocelli form a semicircle about a central ocellus towards the anterior margin of the head. The clypeal region, which is divided into two equal segments, is slightly produced and evenly convex over the entire margin. The sutura frontoclypealis is dorsally convex. The labrum is short and gently tapered. The mandibles are curved and opposable with a double saw-toothed edge. The ventral surface of the cranium is covered by a bilaterally symmetrical labium. Flanking this is a pair of gently convex maxillae (Text-fig. 10b). Anterior to this, also flanking the labium, is a small, flattened eye of ‘ schizochroal’-type (Text-fig. 6f, h). Thorax. The thorax consists of three segments, with the pronotum being the largest (Text-fig. 10a). All are much wider than long, with the cuticle of the dorsal surface more heavily sclerotized than that of the ventral surface and the abdomen. There is a distinct ridge around the periphery of the pronotum. There are three distinct units to each tergite. The anterior prescutum (Text-fig. 3a) is relatively narrow but increases in thickness dorsally, forming a ‘saddle-like’ feature and patterned by an irregular arrangement of pits (Text- fig. 3b). The scutum, the largest of the three units, is patterned by a random arrangement of raised setae emplacements. The posterior scutellum is wider than the prescutum, although similarly patterned (Text- fig. 3c). The scutellum of each segment overlies the prescutum of that behind. The individual tergites are 848 PALAEONTOLOGY, VOLUME 41 separated by intersegmental membranes. A large spiracle is present on the lateral surface of the pronotum. There is a slight bulging of each segment laterally, just above the first pair of limbs. The limbs themselves are robust and decrease in size posteriorly. The base of each is protected by a coxal collar (Text-fig. 3d). The adjacent arthrodial membrane is distinctly patterned (Text-fig. 3e-f). Abdomen. The abdomen bears at least nine tergites, although no complete specimen exists. There is a slight swelling of the abdomen about the fifth-last tergite and it tapers gently posterior to this. The scutellum of each tergite overlaps the prescutum of the following one, as with the thorax. The last few tergites have a distinctive appearance, the penultimate segment bearing two posteriorly projecting conical ‘horns’, or terminal prolegs, which are fused to the dorsal surface (Text-fig. 10c-d). The final tergite terminates in a hemispherical ‘bulb’ (although this may be an extrusion of internal tissue, since this ‘bulb’ is absent from a second specimen). A tube-like spiracle projects posteriorly from the ventral surface of each abdominal somite, although poor preservation prevents confirmation that such a spiracle is present on segment 1 . The cuticle of the abdomen, including that of the tergite with the ‘horns’, is ornamented in a similar fashion to the thorax. The only difference is the presence of small ‘fang-like’, posteriorly projecting barbs on the prozonite, which decrease in size towards the ventral surface, where they are absent. On the ventral surface are small, longitudinally aligned setae. However, their arrangement is obscured by encrusting material. Dimensions. Almost complete specimens suggest an overall length of 50-70 mm. Remarks. The orders Trichoptera and Lepidoptera are united in the informal rank Amphies- menoptera. Accordingly, the larvae possess many similarities. However, the presence of a pair of projecting conical ‘horns’, or terminal abdominal prolegs, allows the Riversleigh species to be assigned with some confidence to the order Trichoptera. More detailed taxonomic assignment would require information on the position of setae or spicules on the abdomen. Class myriapoda Latreille, 1796 Order julida Brandt, 1833 Family indet. Material. QM F34596 (two specimens), mid sections of the trunk composed of four and five segments respectively. Description. The trunk consists of a number of leg-bearing rings. The sclerites of each ring are fused together and to the pleurotergal arch to form a completely cylindrical sclerite (Monozonian condition). The prozonite of one ring is overlapped by the metazonite of that preceding it. The two zonites are separated by a distinct suture. Each ring carries two pairs of limbs and constitutes a diplosegment. The coxal openings are small and project laterally. The limbs are slender. Dimensions. Length of single ring 3 mm, diameter 4 mm. Remarks. The presence of only disarticulated segments precludes assignment beyond the level of suborder. Acknowledgements. The research at Riversleigh was supported by the Australian Research Grants Scheme (Archer); the National Estate Grants Scheme, Queensland (Archer); the University of New South Wales; the Commonwealth Department of Environment, Sports and Territories; the Queensland National Parks and Wildlife Service; the Commonwealth World Heritage Unit; ICI Australia Pty Ltd; the Australian Geographic Society; the Queensland Museum; the Australian Museum; the Royal Zoological Society of N.S.W.; the Linnean Society of N.S.W.; Century Zinc Pty Ltd; Mount Isa Mines Pty Ltd; Surrey Beatty and Sons Pty Ltd; the Riversleigh Society Inc.; and private supporters including E. Clark, M. Beavis, M. Dickson, S. and J. Laverack and S. and D. Scott-Orr. We thank P. A. Jell for access to the Riversleigh arthropods and D. A. Arena for supplying a preprint of a paper on the Dunsinane Site and for stratigraphical and other information. Skilled preparation of the specimens was carried out by A. Gillespie. Useful discussion was DUNCAN ET AL.\ TERTIARY INSECTS AND MILLIPEDES 849 provided by E. N. K. Clarkson, P. R. Wilby and A. C. Neville. E. A. Jarzembowski and P. Hammond reviewed an earlier draft. Additional SEM assistance was provided by Chris Jones of The Natural History Museum (London). IJD is a NERC research student. REFERENCES allison, p. a. 1988. Taphonomy of the Eocene London Clay biota. Palaeontology , 31, 1079-1110. archer, M. and bartholomai, a. 1978. Tertiary mammals of Australia: a synoptic review. Alcheringa , 2, 1-19. — and clayton, G. (eds). 1984. Vertebrate zoogeography and evolution in Australasia. Hesperian Press, Perth, 136 pp. godhelp, H., hand, s. j. and megerian, d. 1989. Fossil mammals of Riversleigh, northwestern Queensland: preliminary overview of biostratigraphy, correlation and environmental change. Australian Zoologist , 25, 29-65. hand, s. J. and godhelp, h. 1994a. Riversleigh, the story of animals in ancient rainforests of inland Australia. Second edition. Reed Books, Chatswood, NSW, 264 pp. - 19946. Patterns in the history of Australia’s mammals and inferences about palaeohabitats. 80-103. In hill, R. (ed.). History of the Australian vegetation: Cretaceous to Recent. Cambridge University Press, Cambridge, 340 pp. - 1995. Tertiary environmental and biotic change in Australia. 77-90. In vrba, e. s., denton, g. H., partridge, T. c. and burckle, L. h. (eds). Paleoclimate and evolution, with emphasis on human origins. Yale University Press, New Haven, 265 pp. baird, G. c., sroka, s. c., shabica, c. w. and beard, T. L. 1 985a. Mazon Creek-type fossil assemblages in the U.S. midcontinent Pennsylvanian: their recurrent character and paleoenvironmental significance. Philo- sophical Transactions of the Royal Society, Series B, 311, 87-99. -shabica, c. w., anderson, i. l. and richardson, E. s. ir 19856. The biota of a Pennsylvanian muddy coast: habitats within the Mazonian Delta Complex, northeastern Illinois. Journal of Paleontology , 59, 253-281. bolton, h. 1905. Notes on the geological horizon and palaeontology of the ‘Soapstone Bed' in the Lower Coal Measures, near Colne, Lancashire. Geological Magazine, (5), 2, 433 — 444. bouligand, y. 1965. Sur une architecture torsadee repandue dans de nombreuses cuticules d'arthropodes. Compte rendu hebdomadaire des seances de FAcademie des Sciences, Paris, 261, 3665-3668. brandt, x. 1833. Ueber die inder Regentschaft Algier beobachteten Myriopoden. In wagner, m. (ed.). Alger III. Leopold Voss, Leipzig, 135 pp. brauer, f. 1885. Systematische-zoologische Studien. Sitzungsberichte der (K.) Akademie der Wissenschaften in Wien, mathematische-naturwissenschaftlichen Klasse, 91, 237 — 43 1 . briggs, d. e. G. and rear, A. J. 1994. Decay and mineralization of shrimps. Palaios, 9, 431-456. britton, E. b. 1960. Beetles from the London Clay (Eocene) of Bognor Regis, Sussex. Bulletin of the British Museum, Natural History (Geology), 4, 27-50. Clarkson, E. N. k. 1979. The visual system of trilobites. Palaeontology, 22, 1-22. COMMONWEALTH SCIENTIFIC AND INDUSTRIAL RESEARCH ORGANIZATION (C.S.I.R.O) 1992. The insects of Australia. Melbourne University Press, Melbourne, 1029 pp. donovan, s. K. and veltkamp, c. I. 1994. Unusual preservation of late Quaternary millipedes from Jamaica. Lethaia , 27, 355-362. duncan, I. J. and briggs, d. e. g. 1996. Three-dimensionally preserved insects from the Tertiary of Riversleigh, Australia. Nature, 381, 30-31. emery, c. 1886. Ueber Phylogenie und Systematik der Insekten. Biologisches Zentralblatt , 5, 648-656. grimaldi, d. a., bonwich, E., delannoy, m. and doberstein, s. 1994. Electron microscope studies of mummified tissues in amber fossils. American Museum Novitates, 3097. 1-31. handschin, E. 1944. Insekten aus den Phosphoriten des Quercy. Schweizer Palaontologische Abhandlungen , 64, 1-23. hellmund, m. and hellmund, w. 1996. Phosphoritisierte Insekten- und Annelidenreste aus der mittelohgozanen Karstpaltenfiillung ‘Ronheim 1’ bei Harburg (Bayern, Siiddeutschland). Stuttgarter Beitrage zur Naturkunde , Series B, 241, 1—21. henwood, a. 1992. Exceptional preservation of dipteran flight muscle and the taphonomy of insects in amber. Palaios, 1, 203-212. — 1993 Recent plant resins and the taphonomy of organisms in amber: a review. Modern Geology, 19, 35-59. 850 PALAEONTOLOGY, VOLUME 41 heyler, D. 1980. Fossiles du Stephanian de Montceau-les-Mines (France). Bulletin du Societe Histoire Naturelle, Autun , 94, 3-6. horvath, G. and clarkson, E. n. K. 1993. Computational reconstruction of the probable change of form of the corneal lens and the maturation of optics in the post-ecdysial development of the Devonian Trilobite Phacops rana milleri Stewart 1927. Journal of Theoretical Biology , 160, 343-373. — and pix, w. 1997. Survey of modern counterparts of schizochroal trilobite eyes: structural and functional similarities and dilferences. Historical Biology , 12, 229-264. jarzembowski, E. a. 1980. Fossil insects from the Bembridge Marls, Palaeogene of the Isle of Wight, southern England. Bulletin of the British Museum , Natural History (Geology), 33(4), 237-293. jell, p. A. and duncan, p. m. 1986. Invertebrates, mainly insects, from the freshwater, Lower Cretaceous, Koonwarra Fossil Bed (Korumburra Gp), South Gippsland, Victoria. Memoir of the Association of Australasian Palaeontologists, 3, 111-205. Johnson, r. g. and richardson, e. s., jr 1966. A remarkable Pennsylvanian fauna from the Mazon Creek area, Illinois. Journal of Geology, 74, 626-631. kinzelbach, r. 1967. Zur Kopfmorphologie der Facherfliigler (Strepsiptera, Insecta). Zoologische Jahrbiicher Anatomie, 84, 559-684. kirkby, w. 1815. An introduction to entomology. Longman, London, 512 pp. kolbe, h. j. 1908. Mem System der Coleopteran. Zeitschrift fur wissenschaftliche Insektbiologie, 4, 116-123. latreille, p. a. 1802. Histoire naturelle, generate et particuliere, des crustaces et des insectes. 3. Dufart, Paris, 467 pp. — 1825. Families naturelles du regne animal, exposees succinctement et dans un ordre analytique, avec T indication de leurs genre. Dufart, Paris, 570 pp. leakey, l. s. b. 1952. Lower Miocene invertebrates from Kenya. Nature, 169, 1020-1021. 1963. Adventures in the search for man. National Geographic, 123(1), 132-152. linne, C. von 1758. Sv sterna Naturae Regna Tria Naturae: secundum classes, or dines, genera, species, cum characteribus differentiis synonymis, locis. 10th edition, revised. Laurentius Salvius, Holmiae, 824 pp. martill, d. m. 1988. Preservation of fish in the Cretaceous of Brazil. Palaeontology, 31, 1-18. megerian, D. 1992. Interpretation of the Miocene Carl Creek Limestone, northwestern Queensland. The Beagle, 9, 219-248. miller, s. E. 1983. Late Quaternary insects of Rancho La Brea and McKittrick, California. Quaternary Research, 20, 90-104. mills, J. s., white, r. and gough, L. J. 1984. The chemical composition of Baltic amber. Chemical Geology, 47, 15-39. Neville, a. c. 1970. Cuticle ultrastructure in relation to the whole insect. 17-39. In Neville, a. c. (ed.). Insect ultrastructure. Symposium of the Royal Entomological Society of London, 5. — thomas, M. G. and zelazny, b. 1969. Pore canal shape related to molecular architecture of arthropod cuticle. Tissue and Cell, 1, 183-200. nitecki, M. H. (ed.) 1979. Mazon Creek fossils. Academic Press Inc., New York, 581 pp. palmer, a. r. 1957. Miocene arthropods from the Mojave desert, California. Professional Papers of the U.S. Geological Survey, 294, 237-80. park, L. E. 1995. Geochemical and paleoenvironmental analysis of lacustrine arthropod-bearing concretions of the Barstow Formation, southern California. Palaios, 10, 44-57. paulus, H. F. 1979. Eye structure and the monophyly of the Arthropoda. 299-383. In gupta, a. p. (ed.). Arthropod phytogeny. Van Nostrand Reinhold, New York, 330 pp. poinar, G. o. Jr 1992. Life in amber. Stanford University Press, Stanford, California, 349 pp. and hess, R. 1982. Ultrastruture of 40-million-year-old insect tissue. Science, 215, 1241-1242. richardson, e. s. Jr 1956. Pennsylvanian invertebrates of the Mazon Creek area, Illinois. Fieldiana Geology, 12, 1-76. riek, e. f. 1954. A re-examination of the Upper Tertiary mayflies described by Etheridge and Olliff from the Vegetable Creek tin-field. Records of the Australian Museum, 23, 159-160. - 1967. Further evidence of Panorpidae in the Australian Tertiary (Insecta: Mecoptera). Journal of the Australian Entomological Society, 6, 71-72. - 1970a. Fossil history. 168-187. In c.s.i.r.o. Division of Entomology. The Insects of Australia. 3rd edition. Melbourne University Press, Carlton, Victoria. 19706. Origin of the Australian insect fauna. Second Gondwana Symposium - 1970, 593-598. rundle, A. j. and cooper, j. 1971. Occurrence of a fossil insect larva from the London Clay of Herne Bay, Kent. Proceedings of the Geologists' Association, 82, 293-295. DUNCAN ET AL.: TERTIARY INSECTS AND MILLIPEDES 851 seilacher, a., reif, w.-E. and westphal, F. 1985. Sedimentological, ecological and temporal patterns of fossil Lagerstatten. Philosophical Transactions of the Royal Society of London, Series B, 311, 5-23. sharp, d. 1899. Some points in the classification of the Insecta Hexapoda. In Proceedings of the fourth International Congress of Zoology, Cambridge, England, 246-249. Snodgrass, R. E. 1935. Principles of insect morphology. McGraw Hill, New York, 667 pp. STANKiEwicz, b. a., briggs, d. e. g., evershed, p. r. and duncan, i. j. 1997. Chemical preservation of the insect cuticles from the asphalt deposits of California. Geochimica et Cosmochimica Acta, 61, 2247-2252. stock, c. h. 1992. Rancho La Brea. A record of Pleistocene life in California. Natural History Museum of Los Angeles County, Science Series, 37. 79 pp. talbot, p. h. b. 1971. Principles of fungal taxonomy. Macmillan, London, 274 pp. tedford, r. h. 1967. Fossil mammal remains from the Tertiary Carl Creek Limestone, northwestern Queensland. Bulletin of the Bureau of Mineral Resources, Australia, 92, 217-36. woodland, b. a. and stenstrom, r. c. 1979. The occurrence and origin of siderite concretions in the Francis Creek Shale (Pennsylvanian) of northeastern Illinois. 69-104. In nitecki, m. h. (ed.). Mazon Creek fossils. Academic Press Inc., New York, 581 pp. woodward, h. 1907. Further notes on the Arthropoda of the British Coal Measures. Geological Magazine, (5), 4, 539-549. wuttke, m. 1983. ‘ Weichteil-Erhaltung’ durch lithizierte Mikroorganismen bei mitteleozanen Vertebraten aus den Olschiefern der ‘ Grube-Messel ' bei Darmstadt. Senckenbergiana Lethaea, 64, 509-527. IAN J. DUNCAN DEREK E. G. BRIGGS Department of Earth Sciences University of Bristol Wills Memorial Building Queen’s Road Bristol BS8 1HQ, UK e-mail ian.j.duncan@bristol.ac.uk D.E.G.Briggs@bristol.ac.uk MICHAEL ARCHER Vertebrate Palaeontological Research Laboratory School of Biological Sciences University of New South Wales Sydney 2052, Australia Typescript received 24 February 1997 Revised typescript received 23 June 1997 EFFACED STYGINID TRILOBITES FROM THE SILURIAN OF NEW SOUTH WALES by d. j. holloway and p. d. lane Abstract. Eight species of illaenimorph trilobites belonging to five genera of the Styginidae are described from limestones of the mid-late Wenlock to Ludlow Mirrabooka Formation and its stratigraphical equivalents in the Orange district. New South Wales. The morphology of illaenimorph ( = effaced) stygimds is discussed; the term ‘omphalus’ is introduced for the socketed, tubercle-like projection present in some genera on the interior of the cranidium at or in front of the anterior end of the axial furrow. Amongst other characters, the gross convexity of the exoskeleton, the form of the rostral plate, the presence of the omphalus, the form of the thorax, and possibly the form of the hypostome are deemed most useful for generic diagnosis; characters used for discrimination at a lower taxonomic level include the proportions of the exoskeleton, the degree of effacement, the pattern of cranidial muscle scars, the size and position of the eye, and the character and disposition of sculpture. New taxa are Excetra iotops gen. et sp. nov., Lalax olibros gen. et sp. nov., L. lens gen. et sp. nov., Rhaxeros synaimon sp. nov. and R. trogodes sp. nov. Bumastus (Bumastella) Kobayashi and Hamada is raised to generic status and its diagnosis emended; specimens from New South Wales are assigned to the type species B. spicula , which is considered to be synonymous with five other Japanese species assigned to three different genera by Kobayashi and Hamada. Bumastus is tentatively recorded on the basis of a single rostral plate; the genus is otherwise known with certainty only from Laurentia and eastern Avalonia. Meraspid transitory pygidia of Bumastella and Lalax from New South Wales are up to eight times larger than those of other styginids with well documented ontogenies; transitory pygidia of large size are known also in some other Silurian effaced styginids, and it is suggested that the phenomenon may result from neoteny. The assumption that sexual maturity in trilobites coincided with the meraspid-holaspid transition is refuted. The effaced styginids from New South Wales show strong faunal affinity with those from the Upper Wenlock or Lower Ludlow of Japan. Trilobites are abundant in Silurian limestones west of the city of Orange in central western New South Wales. Few of the species have been described, although they constitute the most diverse and best preserved Silurian trilobite faunas known from Australia. Although at least nine major groups of invertebrates are present in the limestones, the largely disarticulated elements of trilobite exoskeletons are predominant, and most of them are effaced (illaenimorph) forms belonging to the Styginidae. Faunas of this type also occur in lithologically similar relatively pure limestones of Silurian age elsewhere in the world, and were named the ‘Styginid-Cheirurid-Harpetid Assemblage’ by Thomas and Lane (1998, p. 447, figure 36.1-36.2), who described the lithology in which it occurs, and its stratigraphical and geographical distribution. The reasons for the dominance of effaced trilobite elements in such assemblages are not known; large and small elements of trilobites occur together, and with large and small specimens of other invertebrate groups (ostracodes and brachiopods, respectively with dimensions as little as 2 mm and up to 50 mm), so that hydrodynamic sorting seems not always to be a factor. In some cases it might have been; some ‘nested’ occurrences of effaced cranidia and/or pygidia were noted, for example of Rhaxeros trogodes from locality PL 1996. The effect upon the aspect of the association as preserved of discarded exuviae cannot be assessed; however, the commonness of this type of association in rocks ranging from Ordovician to Permian indicates that effaced trilobites might have been the dominant forms in life. In the present paper, the effaced styginids from only the mid to late Wenlock and Ludlow limestones of the Orange district sequences are described. [Palaeontology, Vol. 41, Part 5, 1998, pp. 853-896, 8 pis) © The Palaeontological Association 854 PALAEONTOLOGY, VOLUME 41 text-fig. 1. A, South-eastern Australia; approximate area of Text-figure 1b indicated by square, b, general area from which the trilobites were collected. c-D, location of fossiliferous localities. STRATIGRAPHY The Silurian sequence in the area between Borenore and Molong, 20-30 km west-north-west of Orange (Text-fig. 1 ), is characterized by marked and complex changes in lithofacies which, together with the geographically restricted nature of previous geological mapping, has led to the recognition of a variety of stratigraphical units in different parts of the area (Text-fig. 2). In the western part of the area, near 'Mirrabooka' homestead, the lowermost Silurian unit is the Boree Creek Formation (Sherwin 1 97 1 z/, p. 210), an impure limestone up to 60 m thick (Sherwin and Pickett, in Pickett 1982, p. 141) which unconformably overlies late Ordovician andesitic HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 855 volcanics. The Boree Creek Formation was divided by Sherwin (1971a) into lower and upper limestone units, informally named Limestones A and B respectively, separated by a calcareous tuffaceous sandstone referred to as the ‘tuffaceous trilobite bed’. A possible disconformity between Limestone A and the ‘tuffaceous trilobite bed’ was considered by Sherwin and Pickett (in Pickett 1982, p. 140) to be probably of only very local significance, and not to represent a significant break. These authors also expressed doubts about the lateral extent of the lithological subdivisions of the Boree Creek Formation; observations by DJH suggest that the subdivisions can be recognized only locally, that the lithology of the ‘tuffaceous trilobite bed’ is only the result of decalcification of the limestone, and that trilobites are not restricted to the middle part of the formation. Conodonts from the Boree Creek Formation were correlated by Bischoflf (1986, p. 36; text-fig. 8) with the latest Llandovery to early Wenlock amorphognathoides and ranuliformis biozones. Disconformably overlying the Boree Creek Formation is the Mirrabooka Formation. This consists of 500 m of predominantly fine sandstones and siltstones but also includes several lenticular limestone bodies that were referred to by Sherwin (1971a) as Limestones C, H and I. The following graptolite species from near the base of the Mirrabooka Formation were recorded by Pickett (1982, p. 147): Pristiograptus meneghini, Monoclimacis cf. flemingi, M. cf.flumendosae, Dendrograptus sp. and Dictyonema sp. - which are considered to represent a mid to late (but not latest) Wenlock age. The age indicated by an assemblage (Bohemograptus bohemicus , Lobograptus ‘ scanicus ’, Linograptus posthumus and Dictyonema sp.) from about the horizon of Limestone I (Text-fig. 2), in the upper part of the formation is equivocal; correlation with the Ludlow scanicus and leintwardinensis biozones was suggested (Pickett 1982, p. 147), but Dr D. Loydell (pers. comm.) suggests that although the Ludlow is indicated, the named species are not all known to occur stratigraphically together elsewhere, and re-examination of the fauna should be undertaken. Overlying the Mirrabooka Formation with apparent conformity are up to 400 m of shales and siltstones that are commonly olive green or red; they were assigned by Sherwin (1971a, p. 219) to the Wallace Shale, the type locality of which lies some 25 km to the south. In the ‘Mirrabooka’ area the formation also contains pods of limestone up to 250 m long, and a boulder bed and exotic blocks of Ordovician sediments and volcanics. The graptolites Monograptus cf. ultimus and M. bouceki , indicative of the Upper Ludlow, occur in the lower part of the formation (Byrnes, in Pickett 1982, p. 154), and the upper part is considered to extend into the Devonian. In the eastern part of the area, just west of Borenore, the Rosyth Limestone (Walker 1959, p. 42) is equivalent to at least part of the Boree Creek Formation. Problems with the definition of the Rosyth Limestone were discussed by Pickett (1982, p. 161), who restricted the name to the lowermost part of the sequence, which consists of richly fossiliferous limestones and calcareous shales with a thickness of 50 m to more than 100 m. Overlying these strata, apparently disconformably, are 100 m of unnamed lithic arenites, shales and bedded limestones. These are in turn overlain disconformably by the 600 m thick massive white, grey and red limestones of the Borenore Limestone, which is equivalent to the Mirrabooka Formation and probably the lowermost Wallace Shale. Bischoff (1986, p. 38 ; text-fig. 8) reported conodonts of the early Wenlock ranuliformis Biozone in the lower part of the Borenore Limestone; it is possible, however, that the specimens did not come from this formation but from the underlying unnamed limestone unit mentioned above. To the north of ‘Mirrabooka’, towards Molong, the Mirrabooka Formation grades into the Molong Limestone (Adrian 1971, p. 193), which directly overlies late Ordovician andesitic volcanics with unconformity, and is unconformably overlain by Late Devonian sandstones. Several discrete limestone bodies occurring to the south of the main outcrop area of the Molong Limestone in the vicinity of ‘ Mirrabooka’ homestead were referred to by Sherwin (1971a) as Limestones D-G and J; these were considered by Sherwin (1971a, p. 212) to be southern extremities of the Molong Limestone, but Pickett (1982, p. 147) assigned them to the Mirrabooka Formation. The stratigraphical range of the Molong Limestone is uncertain, but the upper part is believed to be equivalent to part of the Wallace Shale and the lower part may contain equivalents of the Rosyth Limestone (Pickett 1982, p. 148). 856 PALAEONTOLOGY, VOLUME 41 text-fig. 2. General stratigraphy of the collection area; modified from Pickett (1982, fig. 18). Abbreviations: Lst. = limestone; Tuff. = tuffaceous trilobite bed of Sherwin (1971a); LLAND. = Llandovery; PR = Pn'doli. TRILOBITE FAUNAS The earliest record of trilobites from the area was by de Koninck (1876) who identified Illaenus wahlenbergi Barrande?, Bronteus partschi Barrande and Harpes ungula Sternberg from Borenore Caves, in strata now assigned to the Borenore Limestone. Etheridge (1909) assigned specimens from a similar horizon in the same area to his earlier established species Illaenus johnstoni. Etheridge and Mitchell (1917) described the new species Bronteus angusticaudatus from the Borenore Limestone south of Borenore Caves, and also B. mesembrinus and B. molongensis from Timestone-beds adjacent to Molong’, a locality that could refer either to the Molong Limestone or to the Early Devonian Garra Formation. Other trilobites to have been recorded from the Borenore Limestone are Calymene , Encrinurus and Sphaerexochus (Siissmilch 1907; Campbell et al. 1974); Dun (1907, p. 265, pi. 40, fig. 7) also identified Phacops , but the specimen illustrated is an encrinurid pygidiuin. From the Silurian sequence below the Borenore Limestone just to the east of Borenore Caves, Fletcher (1950) described the new species Dicranogmus bartonensis (= Trochurus bartonensis; see Thomas and Holloway 1988, p. 221), Encrinurus borenorensis ( = Batocara borenorense\ see Edgecombe and Ramskold 1992, p. 259, and Holloway 1994, p. 255) and Phacops macdonaldi (= Ananaspis macdonaldi ; see Holloway 1980, p. 63), as well as Phacops crossleii Etheridge and Mitchell. Also present in Fletcher’s collections, although not recorded by him, is a species of Youngia (Holloway 1994, p. 244). The precise locality from which the material was collected is unknown, but the occurrence of some of the same species in the Boree Creek Formation suggests that the material came from a similar stratigraphical level. From the Boree Creek Formation, Sherwin (1971a, fig. 8) listed the trilobites Bumastus, HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 857 Decoroproetusl, Ananaspis macdonaldi, Batocara borenorense, Trochurus bartonensis and Dicran- urus, and Sherwin (1971b) described Acernaspisl oblatus (a junior synonym of Ananaspis macdonaldi ; see Holloway 1980, p. 64). Work in progress indicates that the trilobite fauna of this formation includes at least 24 other genera belonging to the families Styginidae, Proetidae, Scharyiidae, Brachymetopidae, Cheiruridae, Staurocephalidae, Calymenidae, Lichidae and Odonto- pleuridae. The most abundant trilobite faunas in the overlying Mirrabooka Formation occur in limestones H and I, from which Sherwin (1971a, fig. 8) listed Scutellum, Bumastus sp. B, Bumastus sp. C, Kosovopeltis , Decoroscutellum cf. molongensis (Etheridge and Mitchell), Decoroproetusl and Cheirurus. Apart from the effaced styginids that are the subject of the present work, preliminary investigations on the remaining trilobites indicate that the fauna is at least as diverse as that of the underlying Boree Creek Formation, with representatives of the same families. In addition to the trilobites, other invertebrates present in the Mirrabooka Formation include brachiopods, gastropods, bivalves, rostroconchs, ostracodes, ?stromatoporoids, corals, and pelmatozoan debris. From siltstones just above the base of the Wallace Shale near ‘Mirrabooka’ homestead, Sherwin (1968) described the trilobite Denckmannites rutherfordi , and also recorded Encrinurus mitcheUi (Foerste) and an indeterminate odontopleurid. MATERIALS, METHODS AND LOCALITIES The material is preserved as largely dissociated exoskeletal elements in indurated limestones, which vary from predominantly white to pale grey, pink or, in the case of the Borenore Limestone, red. The cuticle of the trilobites is invariably present and usually adheres preferentially to the internal mould. The matrix is commonly sparry calcite, although sugary-textured and micritic patches occur; mechanical exposure of the exoskeletal elements by Vibrotool is normally relatively easy since the matrix readily parts from the outer surface of the cuticle. Removal of the cuticle from the internal mould, which is necessary to expose features of its internal surface, is correspondingly difficult. Localities. The trilobites were collected from the following horizons and localities; the localities are marked on Text-figure 1. The PL prefix to locality numbers refers to the Museum of Victoria invertebrate fossil locality register. Grid references apply to the Molong 8631-1 & IV and Cudal 8631-11 & III 1 :50,000 topographic sheets (1st edition) published by the Central Mapping Authority of New South Wales. 1. Limestone H, about middle of Mirrabooka Formation: PL1989, GR FD74502330; PL1996, GR FD74502350. 2. Limestone I, upper half of Mirrabooka Formation: PL1991, GR FD75452250; PL 1992, GR FD75802245; PL1993, GR FD75552265; PL 1988, GR FD75452155 (correlation with Limestone I tentative). 3. Limestone J, equivalent to the uppermost part of the Mirrabooka Formation: PL1998, GR FD73502480. 4. Borenore Limestone, ?lower half : PL448, GR FD80351865; PL3301, GR FD80501905; PL3302, GR FD804190; PL3303, GR FD805189; PL3304, GR FD804188. 5. Molong Limestone, horizon indeterminate: PL1995, GR FD74202580. 858 PALAEONTOLOGY, VOLUME 41 text-fig. 3. Palaeogeographical map for the Ludlow, showing distribution of Bumastella (stars), Rhaxeros (solid circles) and Lalax (triangles); base map modified from Scotese and McKerrow (1990). Repository. All illustrated material is housed in the invertebrate palaeontological collections of the Museum of Victoria, Melbourne (NMV). PALAEOBIOGEOGRAPHICAL IMPLICATIONS The most significant palaeobiogeographical pattern to emerge from our study is the close affinity of the effaced styginids from New South Wales with those from the Late Wenlock or Early Ludlow limestones of Mt Yokokura, Japan (Kobayashi and Hamada 1974, 1984, 1986, 1987). Two of the genera, Bumastella and Rhaxeros , are known only from eastern Australia and Japan (Text-fig. 3); however, Leioscutellum Wu, 1977, from the Llandovery of China, may be a senior synonym of Rhaxeros. Bumastella is represented in Australia and Japan by the same species, B. spicula (Kobayashi and Hamada, 1974), although it has been recorded from Japan under a number of different names (see below). Rhaxeros synaimon and R. trogodes, described herein from New South Wales, also possibly occur in Japan (see below). Lalax occurs in both New South Wales {L. olihros sp. nov., L. lens sp. nov.) and Japan (e.g. L. kattoi (Kobayashi and Hamada, 1984)) but is more widely distributed, being known also from Bohemia (e.g. L. bouchardi (Barrande, 1846)), Norway (e.g. L. inflatus (Kiaer, 1908)), Estonia (‘ Illaenus (Bumastus) barriensis' of Holm 1886; see below), Kazakhstan (L. bandaletovi (Maksimova, 1975)), the United Kingdom (e.g. L. xestos (Lane and Thomas, 1 978c/)), and the eastern United States (e.g. L. chicagoensis (Weller, 1907)). Work in progress on other elements of the trilobite faunas of the Orange district supports the HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 859 close affinity with Japan, suggesting the existence of a distinctive eastern Gondwanan fauna during the Silurian. SEGMENTAL VARIATION IN POSTERIOR TAGMATA OF BUMASTELLA AND LALAX Many of the dissociated posterior tagmata of Bumastella spicula , Lalax olibros and L. lens from New South Wales include in their anterior part various numbers of fused segments that appear to be unreleased thoracic segments. These specimens thus have the morphology of meraspid transitory pygidia, but there are two aspects of the specimens that are unusual : ( 1 ) they are much larger than meraspid transitory pygidia of other trilobites; and (2) unlike meraspid transitory pygidia of other trilobites, some specimens of Bumastella spicula from New South Wales show no correlation between the number of fused segments and size. Tables 1-2 and Text-figures 4-5 show the size of specimens from New South Wales and number of fused segments present. In Lalax olibros the number of fused segments decreases progressively from five to zero as the specimen size increases up to a maximum width of c. 9 mm; all specimens larger than this lack fused segments. In L. lens the smallest known specimen has just one fused segment, and all other specimens, with maximum widths greater that 7 mm, lack fused segments. In relatively small specimens of Bumastella spicula the number of fused segments generally decreases with increasing specimen size up to a maximum width of about 15 mm, at which size no fused segments are present; there are, however, several exceptions to this pattern (see Table 2). The smallest specimen of B. spicula has five fused segments whereas a slightly larger one has six, the posteriormost segment having a pleural furrow like the segments in front but lacking an interpleural furrow posteriorly. Most specimens of B. spicula with maximum widths of 15 mm or more (1 1 out of 17 specimens) lack fused segments; the remainder have from one to three fused segments (mostly two), there being no correlation between the number of fused segments and specimen size. In all three species from New South Wales, the pygidium behind the fused segments is identical in form to that of pygidia lacking them. This indicates that the general reduction in the number of fused segments with increasing specimen size in Lalax olibros and at least the smaller specimens of Bumastella spicula (maximum widths up to 15 mm) is due to the progressive release of segments into the thorax rather than to effacement. Release of segments into the thorax is also demonstrated by some specimens in which the anteriormost segment is only partially fused (e.g. PI. 6, fig. 7). Hence the fused segments are protothoracic segments, and the specimens are meraspides according to the definition of Whittington (1957, 1959). table 1. Numbers of fused thoracic segments in posterior tagmata of Lalax olibros sp. nov. and L. lens sp. nov., arranged in order of increasing maximum width. For both species, all specimens larger than those listed lack fused thoracic segments. Species Registration no. Maximum width (mm) Fused segments Figured herein Lalax olibros PI 44944 4-3 5 PI. 5, fig. 20 PI 448 10 4-8 4 PI. 5, fig. 14 PI 44808 6-2 2 PI. 5, fig. 13 PI 44809 6-4 2 - PI 448 16 6-8 2 - PI 44807 7-6 2 - PI 44806 c. 9 1 PI. 5, fig. 14 PI 44844 c. 9 0 - PI 44805 9-2 0 - Lalax lens PI 448 36 6-9 1 PI. 6, fig. 7 P144837 7-1 0 - 860 PALAEONTOLOGY, VOLUME 41 table 2. Numbers of fused thoracic segments in posterior tagmata of Bumastella spicula from New South Wales, arranged in order of increasing maximum width. All known measurable specimens are listed. Registration no. Maximum width (mm) Fused segments Formation Figured herein P145038 4-7 5 Borenore Lst PI. 2, fig. 8 PI 44943 5-2 6 Mirrabooka Fm PI. 2, fig. 12 PI 44942 9-3 3 Mirrabooka Fm - PI 44940 c. 9-5 3 Mirrabooka Fm - PI 45045 c. 9-5 1 Borenore Lst - PI 44939 100 4 Mirrabooka Fm PI. 2, fig. 15 PI 44941 c. 10 3 Mirrabooka Fm - P144938 10-5 1 Mirrabooka Fm PI. 2, fig. 17 PI 44969 c. 13 1 Molong Lst - PI 45039 c. 13 1 Borenore Lst - PI 44968 14-6 1 Molong Lst - P145036 15-0 0 Borenore Lst - P144937 15-3 0 Mirrabooka Fm - PI 45042 c. 15-5 0 Borenore Lst - P145037 15-7 0 Borenore Lst - PI 45043 17-5 0 Borenore Lst - P145027 17-8 2 Mirrabooka Fm - P145035 19-5 0 Borenore Lst PI. 2, fig. 5 P144936 20-4 2 Mirrabooka Fm PI. 2, figs 3, 6 PI 44930 210 3 Mirrabooka Fm PI. 2, figs 9-10 PI 44965 c . 22 0 Molong Lst - P144935 c. 23 2 Mirrabooka Fm PI. 2, fig. 18 PI 44970 24-3 0 Molong Lst - PI 44934 26-6 0 Mirrabooka Fm PI. 2, figs 1-2 PI 44964 c. 27 0 Molong Lst - P144933 c. 30 2 Mirrabooka Fm - P144932 c. 32 1 Mirrabooka Fm PI. 2, fig. 13 P144931 c. 60 0 Mirrabooka Fm - Further circumstantial evidence supports the conclusion that pygidia of Bumastella spicula with fused segments belong to meraspides. It is provided by two specimens having articulated thoracic segments attached: NMV P144943 (PI. 2, fig. 12) has four segments in the thorax and six fused segments in the pygidium; and NMV P144930 (PI. 2, figs 9-10) has seven segments in the thorax and three fused segments in the pygidium. With the possible exception of Dysplanus, which was diagnosed as having only nine thoracic segments, all known styginids, effaced or not, for which there is information available have ten thoracic segments in the holaspis. It seems likely, therefore, that NMV PI 44943 and NMV PI 44930 exhibit the total number of segments, some thoracic and some still fused, that are going to satisfy the completion of the holaspid thorax (i.e. they are degree 4 and degree 7 meraspides respectively). Of course, because these specimens lack cephala, it is possible that their thoraces are incomplete anteriorly; however, the progressive and marked increase in width (tr.) of the articulating facets on the anteriormost segments present suggests to us that these are from the front of the thorax, and that no segments are missing. The apparently random occurrence of fused segments in pygidia of Bumastella spicula more than 15 mm wide requires further discussion. It might be suggested that, although the fused segments in pygidia less than 15 mm wide are protothoracic segments, those in larger pygidia were added after the release of segments into the thorax had ceased (i.e. after the holaspid stage had been attained). Segments are known to be added to the holaspid pygidium in some trilobites (e.g. Shumardia , Dionide\ see Whittington 1957, p. 442), although the process is unknown in Styginidae. If this process HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 861 Maximum Width (mm) text-fig. 4. Plot of number of fused thoracic segments in posterior tagmata of Lalax olibros (circles) and L. lens (squares) versus maximum width of specimen; all specimens larger than those plotted lack fused segments. text-fig. 5. Plot of number of fused thoracic segments in posterior tagmata of Bumastella spicula from New South Wales versus maximum width of specimen. had occurred in B. spicula , it would mean that specimen NMV P 144930 (discussed above), with seven segments in the thorax and three fused segments in a pygidium 21 mm wide, is not a meraspis displaying the complete number of postcephalic segments, as we have suggested, but a holaspis with the first three thoracic segments broken off. We consider this to be unlikely, in view of the similarity (except for size) between such specimens and smaller transitory pygidia of the species. Other evidence suggesting that the fused segments in the larger pygidia were not added in the holaspid stage is: (1) the fused segments occur at the front of the pygidia, not at the back where new segments are formed; and (2) the number of fused segments in specimens wider than 15 mm does not show a general increase with increasing specimen size. It is notable that specimens of Bumastella spicula more than 15 mm wide with fused segments are known only from the Mirrabooka Formation (locality PL 1989; see Table 2) and not from localities in the Borenore and Molong limestones. This suggests that the apparently random occurrence of fused segments in large specimens may reflect either the presence of more than one species within the sample, or the influence of environmental factors that affected the rate of segment release into the thorax in individuals at that locality. However, we can detect no other morphological evidence to support the first possibility, nor is there any evidence (lithological, faunal or taphonomic) that environmental conditions at PL 1989 differed from those at other localities where B. spicula is found. In other styginids for which relatively complete ontogenies are known, meraspid transitory pygidia are up to a little more than 3 mm wide (Table 3). Somewhat larger transitory pygidia, up to a little more than 4 mm wide (excluding marginal spines) in a specimen with one protothoracic segment, were recorded in Kosovopeltis borealis (Poulsen) by Ludvigsen and Tripp (1990, text-fig. 862 PALAEONTOLOGY, VOLUME 41 4f, pi. 4, fig. 5). The largest transitory pygidia of Lalax olibros and L. lens are almost two to three times the size of the largest transitory pygidium of Kosovopeltis (Table 1), whereas the largest transitory pygidium of Bumastella spicula is eight times larger (Table 2; however, the smallest pygidium of B. spicula without fused segments is three to four times larger than the largest transitory pygidia of K. borealis). The smallest transitory pygidia of L. olibros and B. spicula known, with five fused segments (i.e. meraspid degree 5 if there are ten segments in the holaspid thorax), are between two and four times larger than degree five transitory pygidia of Scutellum calvum and Dentaloscutellum hudsoni (Chatterton 1971, figs 5a, 7). table 3. Range in width of transitory pygidia, excluding marginal spines, in some styginids (meraspid degrees 0-9). Species Age Width of transitory pygidia (mm) Reference Failleana calva Ordovician 0-7-2-7 Chatterton (1980, pi. 5, figs 10, 20-21, 30-31) Kosovopeltis svobodai Silurian 0-6-3-2 Kachka and Saric (1991, fig. 6) Scutellum calvum Devonian 1-0-19 Chatterton (1971, fig. 7) Dentaloscutellum hudsoni Devonian 1 -0-2-3 Chatterton (1971, fig. 5a) Meraspid transitory pygidia of large size are also known in other Silurian effaced styginids. A transitory pygidium of Lalax bouchardi from Bohemia was figured by Snajdr (1957, pi. 10, fig. 7); the specimen, which has two protothoracic segments, is c. 5-8 mm wide. Transitory pygidia of effaced styginids from Arkansas have also been observed by one of us (DJH); widths of these specimens are 6-3 mm for a specimen of Lalax with one protothoracic segment, 4—4-2 mm for specimens of Illaenoides with five protothoracic segments, and 5-8 mm for a specimen of Illaenoides with three protothoracic segments. This evidence indicates that large meraspides are not unique to the species under discussion here, and in fact the phenomenon may be widespread amongst effaced Silurian styginids. The evolution of such forms with large meraspides from an ancestor or ancestors with normal-sized meraspides may be the result of neoteny (reduced rate of morphological development). Assuming the same rates of growth and moulting in ancestor and descendent, the latter must have undergone a greater number of moults in order to reach a larger size at the same stage of ontogeny. Hence the rate of morphological change (in this case, release of protothoracic segments into the thorax) must have been delayed at each moult in the descendent. If the delay was cumulative at each moult, then throughout ontogeny the descendent form would have fallen further and further behind its ancestor in release of segments into the thorax, accounting for the very large size of some of the meraspides. Meraspid transitory pygidia comparable in size to those of Bumastella spicula are known in the nileid Illaenopsis harrisoni from the Arenig of South Wales. Fortey and Owens (1987, p. 197) reported that in this species, which they considered to have the largest merapides of any trilobite, transitory pygidia with one protothoracic segment reach a width of 20 mm, which is two-thirds the size of the largest known transitory pygidia of B. spicula. Neoteny was also cited by Fortey and Owens (1987, p. 197) as a possible mechanism for the development of the giant meraspid transitory pygidia of I. harrisoni. A question raised by such giant meraspides is whether they have any bearing on the attainment of sexual maturity (i.e. whether they were biologically adult, although not morphologically complete). From study of the many well-documented and relatively complete trilobite ontogenies known there is no direct evidence for the onset of sexual maturity, and to no organs or structures of the exoskeleton of any trilobite can be ascribed a sexual function. However, sexual maturity is commonly assumed, unjustifiably, to have coincided with attainment of the holaspid stage (e.g. HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 863 McNamara 1986, p. 124), although this was defined by Whittington (1957, 1959) in purely morphological terms by the acquisition of the full complement of thoracic segments. It is recognized that protaspid, meraspid and holaspid stages are not developmentally homologous amongst all trilobites (Hughes and Chapman 1995, p. 349), suggesting that sexual maturity may have occurred at different growth stages in different taxa. Whittington (1957, p. 445) noted that some trilobites may increase in length 30- to 40-fold during the holaspid stage (e.g. holaspides of Isotelus gigas range from 8-9 mm to 400 mm long, whilst those of Paradoxides range from 13-5 mm to over 400 mm). This very large increase in size leads us to suspect that sexual maturity was attained later than the meraspid-holaspid transition in such forms. Some of the meraspid transitory pygidia of Bumastella spicula from New South Wales are amongst the largest specimens of the species known. Assuming that specimens in our collections are representative of the full size attained by individuals of the species, some of these meraspides must have been mature. If so, the release of protothoracic segments into the thorax must have continued in those individuals after maturity, although perhaps at a much slower rate in respect to the moult rate than prior to maturity. If none of the meraspides of B. spicula was mature, and assuming that sexual maturity is related to size, it is reasonable to conclude that holaspides smaller than the largest meraspis were also immature. This would mean that the breeding population of the species is virtually unrepresented amongst our collections, a possibility that we believe to be unlikely. We consider, therefore, that sexual maturity and attainment of the holaspid stage are unrelated in B. spicula, and that they may be unrelated in many other trilobite species. SYSTEMATIC PALAEONTOLOGY Remarks. Because of the great to extreme effacement of the forms described below, it has not been possible to produce brief generic diagnoses. As diagnosed, genera are distinguished on combinations of characters, all of which are therefore listed. However, it is our belief that the gross convexity of the exoskeleton, the form of the rostral plate, the presence or lack of the ‘omphalus' and ‘anterolateral internal pit’ (see terminology below), and the form of the thorax (width of axis, distance between axial furrow and fulcrum) are of paramount importance in diagnosing genera of effaced styginids. Of possible importance also is the form of the hypostome, which is all too often unknown in the present material and in previously described species of effaced styginids. We consider that the proportions of the exoskeleton, the degree of effacement, the pattern of cranidial muscle scars, and the size and position of the eye (which can, however, vary greatly during ontogeny; see Bumastella spicula below) have less taxonomic value, and may be of more use in diagnosing species. Descriptions commence with the gross morphology of the exoskeleton, followed by descriptions of sculpture, muscle scars and ontogeny where these headings are appropriate. In view of the significant differences in appearance between testiferous and exfoliated specimens of the same species, all descriptions are of the external surface of the exoskeleton, unless otherwise stated. Terminology. Muscle scars on the glabella are numbered GO, Gl, etc. from the posterior forward. The feature associated with the cephalic axial furrow often referred to as the ‘lateral muscle impression’ is here termed the ‘lunette’. The ‘holcos’ (Helbert and Lane, in Helbert et al. 1982, p. 132) is the concave zone parallel to and near the lateral and posterior margins of some styginid pygidia. Anteriorly, the holcos is deflected adaxially to unite with an oblique depression running subparallel to the posterior edge of the articulating facet; examination of the ontogenetic development of Failleana (Chatterton 1980, pi. 5; Ludvigsen and Chatterton 1980, pi. 1) indicates that this oblique depression is the pleural furrow on the anterior segment of the pygidium. On the interior of the cranidium of some effaced styginids is a raised boss, commonly with a median depression (often figured as a pit with a central swelling on the internal mould), at which the axial furrow may terminate anteriorly, as in Cybantyx (see Lane and Thomas 1978a, text-fig. 40, Paracybantyx (see Ludvigsen and Tripp 1990, pi. 1, figs 5-9) and Lalax (see PI. 5, fig. 10); the 864 PALAEONTOLOGY, VOLUME 41 A B C text-fig. 6. Relationship of eye ridge to cephalic muscle impressions and omphalus in various styginids. a, Cybantyx insignis (Hall, 1867), based on UC 9902 in the Field Museum of Natural History, Chicago, b, Lalax chicagoensis (Weller, 1907), based on holotype, UC 9910; anterolateral pit not distinguishable on specimen, c, Bumastella spicula (Kobayashi and Hamada, 1974), based on NMV PI 44904. d, Raymondaspis reticulata Whittington, based on Whittington (1965, pi. 56, figs 1, 3, 6). Abbreviations: aip = anterolateral internal pit; ( = lunette; o = omphalus. term ‘omphalus’ (latinized from the Greek for navel) is introduced for this structure. The omphalus may be reflected on the exterior of the exoskeleton as a pit or as a small patch devoid of sculpture. Even when the axial furrow cannot be recognized anteriorly, because of a high degree of effacement, the omphalus may be clearly defined and is useful in indicating the transverse extent of the glabella, as in Lalax gen. nov. Other effaced styginids in which the omphalus occurs are Dysplanus Burmeister, 1843, Failleana Chatterton and Ludvigsen, 1976, Litotix Lane and Thomas, 1978a, Opsypharus Howells, 1982 and Platillaenus Jaanusson, 1954. Situated just in front of the omphalus, and usually slightly adaxial or abaxial to it, there may be a small pit on the interior of the cranidium (appearing as a node on internal moulds), as in Lalax (see PI. 5, figs 3, 7, 9-10; Lane and Thomas 1978a, pi. 3, fig. 14a) and also Cybantyx-, this pit is here termed the anterolateral internal pit. The omphalus was described by Ludvigsen and Chatterton (1980, p. 476) as a ‘socketed pit' (illustrated as a pitted tubercle in their pi. 1, fig. h) and considered by them to be one of the diagnostic features of Bumastinae (Raymond, 1916; as conceived by Jaanusson 1959). Of the genera normally assigned to the subfamily, however, the omphalus is absent in the type species of Bumastus , B. barriensis Murchison, 1839, and in Goldillaenus Schindewolf, 1924, Illaenoides Weller, 1907 and Thomastus Opik, 1953. More recently described effaced styginid genera in which the omphalus is absent are Bumastella Kobayashi and Hamada, 1974, Excetra gen. nov., Ligiscus Lane and Owens, 1982, Meitanillaenus Chang, 1974, Ptilillaenus Lu, 1962 and Rhaxeros Lane and Thomas, 1980. From its position, the omphalus is not the homologue of the ‘fossula’, which is defined as lying at the anterior edge of the eye ridge. In Cybantyx insignis and Lalax chicagoensis, the omphalus is situated well in front of the anterior end of the eye ridge (Text-fig. 6a-b). In Bumastella spicula (described below) the eye ridge is directed towards a point between G2 and G3 (Text-fig. 6c); Bumastella lacks an omphalus, but in other effaced styginids in which the omphalus is present it is situated level with or in front of G3. Hence, although no fossula is present in these forms or the others under discussion here, the omphalus lies morphologically anterior to where such a structure would be expressed. Chatterton and Ludvigsen (1976, p. 39) stated, however, that in Failleana calva the omphalus is situated where ‘ ...a very shallow furrow that is posteriorly continuous with palpebral furrow joins axial furrow’ (i.e. at the posterior edge of the eye ridge); this statement is not HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 865 supported by their illustrations (Chatterton and Ludvigsen 1976, pi. 6, figs 6, 39) which show the omphalus lying well in front of the weak furrow defining the anterior edge of the eye ridge. The omphalus seems to be homologous with the pit (as expressed on the exterior of the exoskeleton) that lies in the axial furrow at or just behind the junction with the lateral border furrow in some non-effaced Ordovician styginids, including species of Stygina Salter, 1853, Raymondaspis Pribyl, in Prantl and Pribyl, 1949 (see Text-fig. 6d) and Turgicephalus Fortey, 1980, and also in Theamataspis Opik, 1937 (see Skjeseth 1955, pi. 3, fig. 1 ; Whittington 1965, pi. 56, figs 3, 7, pi. 59, figs 1, 5-8; Fortey 1980, pi. 6, figs 1-2, 5, pi. 7, figs 1-3, pi. 9, figs 1-3, 5). In Platillaenus , the omphalus (‘ Vordergrube’ of Jaanusson 1954, pi. 3, fig. 6) also lies at the junction of the axial and lateral border furrows. Functionally, the omphalus was apparently involved with the attachment of the hypostome to the cranidium. Jaanusson (1954, p. 548, text-fig. 1, pi. 3, figs 2, 4) showed that in Dysplanus centrotus the anterior wing process of the hypostome projects well forward above the cephalic doublure and is in close proximity with the omphalus. Chatterton and Ludvigsen (1976, p. 39, pi. 6, figs 2, 1 1, 16) described a protuberance similar in form to the omphalus on the inner (dorsal) margin of the librigenal doublure of Failleana calva; this librigenal protuberance faces the omphalus and the two are adjacent in a complete cephalon. Although they were unsure of its function, Chatterton and Ludvigsen (1976, p. 39) suggested that the librigenal structure may have been associated with ligament or muscle attachment to the anterior wing of the hypostome, or with the omphalus. Orientation. In convex to highly convex trilobites, it is difficult to indicate the exact orientation of specimens for description and photography. As applied herein to cephala and cranidia, ‘dorsal view’ indicates that the posterior margin adaxial to the fulcrum is vertical, 'palpebral view’ that the upper edge of the visual surface (or palpebral suture) is horizontal, and ‘plan view' that the maximum sagittal length is being shown. For pygidia, ‘dorsal view’ indicates that the anterior margin adaxial to the articulating facet is vertical, and ‘plan view’ that the lateral and posterior margins are horizontal (in plan view a little less than the maximum sagittal length of the pygidium is shown). As convexity varies during ontogeny of individual species, and between species, ‘plan’ and ‘palpebral’ views are not necessarily identical at all stages of ontogeny and in different species, and in other cases the two views may coincide. Suborder illaenina Jaanusson, 1959 Family styginidae Vogdes, 1890 Remarks. The family name is used here in the emended sense of Lane and Thomas (1983, p. 156), who recognized no subfamilial divisions because of the inability at present to recognize phyletic lines of development. Ludvigsen and Tripp (1990, p. 8) retained the division of Styginidae into Stygininae, Scutelluinae and Bumastinae ‘as an aid to grouping the large number of genera in the family’ but did not indicate which characters could be used as a basis for this subdivision. We consider, however, that the only justification for the recognition of supraspecific taxa is evolutionary relationship. Although also rejecting Ludvigsen and Tripp’s concept of styginid subfamilies as taxa of convenience, Adrain et al. (1995, p. 726) recognized Scutelluinae and Bumastinae as phylogenetic entities that ‘may very likely prove to be monophyletic’, but no evidence to support this statement was offered. Nielsen (1995, pp. 295, 320) recently used Stygininae and Bumastinae as subdivisions of Styginidae without any discussion on the characters he regarded as diagnostic. Genus bumastella Kobayashi and Hamada, 1974 Type species. By original designation; Bumastus ( Bumastella ) spiculus Kobayashi and Hamada, 1974 from the Upper Wenlock (or Lower Ludlow), Gomi, Yokokura-yama, Kochi Prefecture, Shikoku, Japan. Emended diagnosis. Cephalon extremely convex (sag., exsag., tr.), almost hemispherical; in dorsal profile, posterior margin transverse medially, deflected posteroventrally abaxial to fulcrum towards 866 PALAEONTOLOGY, VOLUME 41 broadly rounded genal angle. Axial furrow shallow and poorly defined close to posterior margin, not impressed farther forwards. Eye small, in palpebral view placed with posterior edge in transverse line with posterior edge of lunette, in lateral view placed at half height of cephalon; visual surface borne on steep, concave band of librigena. Anterior branch of facial suture gently convergent forwards, posterior branch gently divergent backwards. Rostral plate sub-triangular, lacking posterior flange, convexity in sagittal line greatest anteriorly. Thorax with gently convex axis narrowing markedly backwards; articulating furrows not impressed; pleurae with horizontal portion adaxial to fulcrum widening (tr.) in more posterior segments, as wide (tr.) as or wider than gently downturned portion abaxial to fulcrum. Pygidium wider than long (sag.), much less convex than cephalon, with strongly developed holcos. Remarks. The above diagnosis is an attempt to characterize the genus allowing for the considerable morphological changes which are seen during ontogeny. These changes, which are outlined below in the section on ontogeny, are partly the reason that we consider the type species to be synonymous with two other species of Bumastella and one of Bumastus erected by Kobayashi and Hamada (1974) at the same time, and with at least one (and possibly another) form described as possible species of Illaenoides by the same authors in a later work (see synonymy of Bumastella spicula herein). Kobayashi and Hamada (1974, p. 50) erected Bumastella as a subgenus of Bumastus , and diagnosed it as having a 'narrower axial lobe which is clearly separated in thorax from pleurae by pronounced axial furrows. Eyes are very large in comparison with those of Stenopareia Holm, 1886 (see Owen and Bruton 1980, pi. 2, fig. 1 1). Short genal spines are present in the type species.’ We consider that, apart from the effacement, Bumastella has little similarity with Bumastus and is not closely related to it. In our opinion, Bumastella has more similarities to Illaenoides and Thomastus (for comparison with the latter see Sandford and Holloway 1998). The similarities with Illaenoides (type species I. triloba Weller, 1907, p. 226, pi. 17, figs 6-9, pi. 19, figs 12-14) from the Niagaran (upper Llandovery or Wenlock) of Illinois include the extreme convexity of the cephalon, the glabella that narrows weakly forwards from the posterior cephalic margin to the lunette, the small eyes, the subparallel anterior branch and weakly divergent posterior branch of the facial suture, and the strong holcos. Illaenoides differs from Bumastella in that the eyes are even smaller and are situated farther forwards, with the posterior margin in front of the lunette in palpebral view; the posterior cephalic margin is deflected less strongly backwards abaxial to the fulcrum; the rostral plate is lenticular in outline rather than triangular, with connective sutures that are distinctly sigmoidal rather than almost straight; the pleurae on the posteriormost thoracic segments are wider abaxial to the fulcra than adaxially; and the pygidium is longer. In view of the differences, even from closely related forms, we consider it necessary to raise Kobayashi and Hamada’s taxon to generic status. Stratigraphical range and distribution. Late Wenlock to ?early Ludlow; Japan and New South Wales. EXPLANATION OF PLATE 1 Figs 1-21. Bumastella spicula (Kobayashi and Hamada, 1974); locality PL 1989, Mirrabooka Formation, unless otherwise indicated. 1-3, 5, NMV P144906; cephalon, anterior and palpebral views; x2; ventral view; x 2-5; lateral view; x 2. 4, NMV PI 44928; rostral plate, ventral view; x 4. 6, NMV PI 44953; locality PL1995, Molong Limestone; small cephalon, oblique view; x 5. 7-8, NMV P145028; locality PL448, Borenore Limestone; cephalon, lateral and palpebral views; x L75. 9, NMV PI 44907; cephalon, palpebral view; x 3. 10, 13, NMV P144905; cephalon, palpebral and dorsal views; x 1-75. 1 1, 14, 21 , NMV P144904; largest cephalon, dorsal, anterior and oblique views; x F75. 12, NMV P144908; small cephalon, anterior view; x 5. 15-16, 18, NMV P144910; smallest cephalon, palpebral, oblique and lateral views; x 6. 17, 19, NMV P144951 ; locality PL1995, Molong Limestone; small cephalon, palpebral and oblique views; x4-5. 20, NMV P144950; locality PL1995, Molong Limestone; small cephalon, oblique view; x4-5. PLATE 1 HOLLOWAY and LANE, Bumastella PALAEONTOLOGY, VOLUME 41 868 Bumastella spicula (Kobayashi and Hamada, 1974) Plate 1, figures 1-21; Plate 2, figures 1-15, 17-18; Text-figure 6c 1909 Illaenus Johnstoni Etheridge; Etheridge, p. 1, figs 1-2 [non Etheridge 1896, p. 33, pi. fig. 3], 1974 Bumastus glomerosus Kobayashi and Hamada [partim], p. 47, pi. 1, figs 3-6, 8 [non fig. 7 = Rhaxeros subquadratus ] ; text-fig. 2a. 1974 Bumastus ( Bumastella ) spiculus Kobayashi and Hamada, p. 51, pi. 2, fig 3; text-fig. 2d. 1974 Bumastus ( Bumastella ) bipunctatus Kobayashi and Hamada, p. 51, pi. 2, figs 4—9; ?pl. 3, fig. 1 ; text-fig. 2e. 1974 Bumastus (Bumastella) aspera Kobayashi and Hamada [partim], p. 52, pi. 3, figs 3-5 [non fig. 6 = Rhaxeros cf. synaimon ]; text-fig. 2f, cephalon only [pygidium = Rhaxeros cf. synaimon ]. 1985a Bumastus glomerosus ; Kobayashi and Hamada [partim\ , p. 345. 1985a Bumastus ( Bumastella ) spiculus ; Kobayashi and Hamada, p. 345. 1985a Bumastus ( Bumastella ) bipunctatus', Kobayashi and Hamada, p. 345. 1985a Bumastus ( Bumastella ) aspera', Kobayashi and Hamada [partim ], p. 345. 1985a Illaenoides (?) magnisulcatus Kobayashi and Hamada [nom. nud .], p. 345. ?1985a Illaenoides (?) abnormis Kobayashi and Hamada [nom. nud.], p. 345. 1986 Illaenoides (?) magnisulcatus Kobayashi and Hamada, p. 452, pi. 90, fig. 5. ?1986 Illaenoides (?) abnormis Kobayashi and Hamada, p. 453, pi. 90, fig. 6. «o«1987 Bumastus glomerosus; Kobayashi and Hamada, p. 110, figs 1a, 2.1a-d [= Lai ax l sp.]. Holotype. University of Tokyo Museum No. 7345; from the Upper Wenlock (or Lower Ludlow); Gomi, Yokokura-yama, Kochi Prefecture, Shikoku, Japan. Other material. Approximately 17 cephala, 19 cranidia, 18 librigenae, one rostral plate, six thoracopyga and 26 pygidia (including meraspid transitory pygidia), from PL448, PL1988, PL1989, PL1995, and PL3301- PL3304. Description. Cephalon almost semicircular in lateral and dorsal views, more than semicircular in anterior and palpebral views; in palpebral view, sagittal length 85 per cent, of maximum transverse width which is level with posterior edge of eye. Axial furrow forming shallow notch in posterior cephalic margin, rapidly dying out just in front of posterior margin but in some specimens very faintly discernible as far forward as lunette, towards which it converges weakly; width of glabella at posterior margin slightly more than half maximum cephalic width. From posterior end of axial furrow, a short (tr.), indistinct furrow is directed laterally and slightly forwards, just in front of narrow (tr.), horizontal portion of posterior fixigenal margin adaxial to fulcrum. Lunette weakly expressed, rather more than 50 per cent, of length of visual surface, situated with anterior edge level with cephalic midlength (sag.) in palpebral view. Fixigena adaxial to palpebral lobe slightly inflated above general transverse convexity (in anterior view); palpebral lobe narrow (tr.), sloping less steeply abaxially than adjacent part of fixigena, length c. 20 per cent, of sagittal length of cephalon in palpebral view, placed at 100-150 per cent, its own length from posterior margin; palpebral furrow shallow and poorly defined, weakly curved in palpebral view. Eye ridge (faintly visible in one specimen under the microscope but not in photographs; see Text-fig. 6c) very narrow, running anteromedially from close to front of palpebral lobe towards midway between G2 and G3, but dying out before reaching line of these impressions. Anterior section of facial suture subparallel to sagittal line posteriorly, curving adaxially anteriorly to cut cephalic margin in line (exsag.) with abaxial edge of lunette; palpebral section of suture very gently curved; posterior section diverging gently backwards, almost straight for most of its course but deflected slightly more strongly abaxially near posterior margin. Librigena gently convex exsagittally and weakly convex transversely; visual surface gently convex dorso-ventrally, subparallel sided with rounded anterior and posterior margins; lenses very small and very numerous. Cephalic doublure steeply inclined, more convex medially than laterally. Rostral plate with maximum width a little more than twice sagittal length, and 50 per cent, of width of cranidium at palpebral lobes. Anterior margin broadly rounded; lateral margins converging backwards at c. 110-120°, slightly more strongly near anterior margin than farther back; posterior margin transverse or weakly convex forwards. Hypostome unknown. Number of thoracic segments unknown. In a specimen with seven thoracic segments articulated with a meraspid transitory pygidium having three protothoracic segments (PI. 2, figs 9-10), thoracic axis is 80 per cent. HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 869 as wide (tr.) posteriorly as anteriorly, and horizontal portion of pleurae adaxial to fulcrum is twice as wide posteriorly as anteriorly; overall, thorax becomes slightly wider backwards. Axial furrow broad and shallow. Anterior segments with pleurae flexed strongly backwards at fulcrum, and with large articulating facets occupying most of segmental length ; more posterior segments successively less strongly flexed backwards at fulcrum but curving forwards slightly distally, with successively smaller articulating facets. Pygidium more convex sagittally than transversely, semi-elliptical in outline, in plan view sagittal length 70 per cent, of maximum width, which is at posterior outer angle of articulating facet. Anterior margin gently arched forwards across axis, between anterior ends of holcos; anterior width of axis 50 per cent, of maximum pygidial width; short (sag., exsag.) articulating half ring defined by absence of sculpture and by faint articulating furrow abaxially. Articulating facet short (exsag.), occupying c. 60 per cent, of width (tr.) of anterior pleural margin; anterior edge of facet with small process situated closer to abaxial than adaxial extremity. Anteriormost pleural furrow strongly developed, joining in a broad curve with holcos which dies out quite suddenly at c. 66 per cent, of maximum pygidial length. Doublure occupying c. 20 per cent, of sagittal length of pygidium, slightly less convex medially than anteriorly but more steeply inclined. Sculpture. Cephalon everywhere covered by dense, small and indistinct pits; especially at posterior margin, and for a little way forward on glabella, this pattern of pits has very narrow, irregular, anastomosing furrows superimposed. Near to anterior margin of cranidium, a few weakly developed terrace ridges are present. Lateral and anterior margins of cephalon bear a very narrow but distinct marginal thread. Rostral plate with ten non-anastomosing terrace ridges present on ventral surface; most anterior of these ridges run parallel to anterior margin, medial ones become transverse, and posterior few ridges curve convex backwards to become subparallel to posterior margin. Thorax and pygidium bear packed indistinct pits like those of cephalon. In addition, distinct terrace ridges are present on and behind articulating facets of thoracic pleurae and pygidium, forming a chevron pattern on lateral parts of pygidium abaxial to the holcos (PI. 2, fig. 13). Muscle scars. Glabella bears four pairs, distinguished by weak wrinkling of exterior of exoskeleton in some specimens. GO and G1 elongated, of similar size and in line exsagittally ; GO situated less than its own length from posterior cephalic margin, extending forwards almost level with posterior edge of lunette in palpebral view; G1 with posterior margin just in front of transverse line through back of lunette and anterior margin opposite front of eye in palpebral view. G2 smallest, sub-circular, slightly more abaxially placed than Gl, equidistant from Gl and G3. G3 also sub-circular, slightly larger than G2 (but much smaller than GO and Gl) and placed slightly farther abaxially, about twice as far from anterior cephalic margin as from G2. Pygidia have two or more pairs of relatively small, sub-circular and poorly defined muscle scars situated anteriorly close to the sagittal line; another pair of larger, sub-circular scars, situated farther from the sagittal line just in front of the pygidial midlength (sag.), is defined by a reticulate appearance of the exterior of the exoskeleton. Surrounding the region of the paired muscle scars laterally and posteriorly, and extending backwards as far as 75 per cent, of the sagittal pygidial length, is a broad, arcuate band of scattered, small (c. 01-0-2 mm diameter), circular pits, possibly representing muscle attachment sites on the interior of the exoskeleton (PI. 2, fig. 13) The smallest meraspid transitory pygidium (PI. 2, fig. 8), with five protothoracic segments, has a pair of elliptical scars impressed on the interior of the exoskeleton, either side of a small, raised (on internal mould) sub-triangular area apparently representing the axis. These scars are probably the posterior pair described above. Ontogeny. The above description is based on some of the largest exoskeletal elements from New South Wales, which differ from the smallest growth stages in a number of respects. Morphological changes that occur during ontogeny (apart from the change in number of protothoracic segments in meraspid transitory pygidia; see above) are as follows. 1 . The genal spine gradually deceases in length and finally disappears. In the smallest cephalon, with a width of 5 mm (PI. 1, figs 15-16, 18; librigena 2-5 mm maximum length) the genal spine is 1-6 mm long; on a cephalon 8-2 mm wide (PI. 1, figs 17, 19; librigena 4-5 mm long) it is 1-4 mm long; on a librigena about 9-5 mm long (PI. 2, fig. 14) it is developed only as a very short, thorn-like point (the tip of which is broken off); and on a librigena 10-2 mm long (PI. 2, fig. 11) it is represented by a very small swelling. 2. The cranidium changes from slightly wider than long to slightly longer than wide in plan view. 3. The length of the visual surface changes from about one-third to about one-tenth of the length of the cephalon in plan view. 870 PALAEONTOLOGY, VOLUME 41 4. The eye changes in position from less than one-half its own length, to more than its own length from the posterior edge of the cephalon. 5. Overall convexity of the cephalon increases (compare PI. 1, figs 7, 18), and the cephalic outline in plan view changes from semicircular to circular. 6. Sagittal length of the pygidium increases slightly relative to width. 7. Overall convexity of the pygidium increases. 8. The holcos and anteriormost pleural furrow are very weak in the smallest meraspid transitory pygidia (PI. 2, figs 8, 12) and become more distinct in larger specimens. In some other effaced styginids, for example Bumastus barriensis (see Lane and Thomas 1978a, pi. 4, fig. 6) and Failleana calva (see Ludvigsen and Chatterton 1980, pi. 1, figs s-v), a genal spine is present in small specimens but is absent later in ontogeny. Remarks. Study of the ontogenetic changes in Bwnastella collected from a single locality (PL 1989), discussed above, led us to the conclusion that those Kobayashi and Hamada species we have synonymized, each of which is based on only one or a few specimens (mostly cranidia), all from the Yokokura limestone of Mt Yokokura (possibly the same locality; see Kobayashi and Hamada 1985a, p. 345), represent different stages in the ontogeny of a single species. The smallest form available in our collections has the morphology of B. spicula (the type species), which is followed by specimens in order of increasing size having the form of B. bipunctata , B. aspera, ‘ IllaenoidesV magnisulcatus, possibly ‘7.?’ abnormis , and finally the largest morph 1 Bumastus' glomerosus which Kobayashi and Hamada (1974, p. 47) noted ‘is the largest illaenid species in the Yokokura fauna’. The Japanese material, of which we have examined plaster casts, is not as well preserved as that from New South Wales, so that it has not been possible to compare all details, such as sculpture. However, based on a comparison of general proportions and convexity of cranidia during growth, we believe that the latter material is conspecific. We consider that the pygidium Kobayashi and Hamada (1974) assigned to Bumastella aspera , and one of the pygidia they assigned to ‘ Bumastus' glomerosus , do not belong to Bumastella but to two different species of Rhaxeros (see discussions of R. synaimon and R. trogodes). Three cephala from the Borenore Limestone at Borenore Caves were referred by Etheridge (1909) to his species Illaenus johnstoni , originally based (Etheridge 1896) on material from the Ordovician EXPLANATION OF PLATE 2 Figs 1-15, 17-18. Bumastella spicula (Kobayashi and Hamada, 1974); locality PL1989, Mirrabooka Formation, unless otherwise indicated. 1-2, NMV P144934; pygidium, dorsal and lateral views; x 2. 3, 6, NMV P144936; transitory pygidium with two protothoracic segments, lateral and dorsal views; x 2-25. 4, NMV P145041 ; locality PL3301, Borenore Limestone; librigenal doublure, ventral oblique view; x 3. 5, NMV P145035; locality PL448, Borenore Limestone; pygidium and posteriormost thoracic segment, dorsal view; x 2-25. 7, NMV P144924; librigena, oblique view; x 4. 8, NMV P145038; locality PL448, Borenore Limestone; transitory pygidium with five protothoracic segments, dorsal view; x 8. 9-10, NMV P144930; transitory pygidium with three protothoracic segments and seven articulated thoracic segments, dorsal and lateral views; x 2-25. II, NMV P144920; librigena, oblique view; x 4. 12, NMV P144943; transitory pygidium with six protothoracic segments and four articulated thoracic segments, dorsal view; x 7. 13, NMV PI 44932; transitory pygidium with one protothoracic segment and with last thoracic segment articulated; detail showing sculpture, oblique view; x4. 14, NMV P144962; locality PL1995, Molong Limestone; librigena, oblique view; x4. 15, NMV P144939; transitory pygidium with four protothoracic segments, dorsal view; x 4. 17, NMV P144938; transitory pygidium with one protothoracic segment, dorsal view; x 4. 18, NMV P144935; transitory pygidium with two protothoracic segments, latex cast in ventral view; x 2. Figs 16, 19. Bumastella sp.; NMV PI 44972; locality PL 1989, Mirrabooka Formation; pygidium, lateral and dorsal views; x2-25. PLATE 2 HOLLOWAY and LANE, Bumastella 872 PALAEONTOLOGY, VOLUME 41 of Tasmania. The specimens were deposited in the former Mining and Geological Museum in Sydney, but Dr I. Percival of the Geological Survey of New South Wales has advised us that they were transferred in the 1930s to the Australian Museum, where there is now no record of their existence (Mr R. Jones, pers. comm.). Nevertheless, Etheridge’s illustrations of one of the cephala clearly show that it belonged to Bumastella. Other specimens of Bumastella collected by us in the vicinity of Borenore Caves are indistinguishable from B. spicula from the Mirrabooka Formation and the Molong Limestone farther to the west, and we consider them to be conspecific. A tiny pygidium from the Borenore Limestone at Borenore Caves was tentatively assigned to Illaenus wahlenbergi Barrande by de Koninck (1876). The small size of the specimen (3 mm long by 2 mm wide), and de Koninck’s description of it as having ‘four segments of the thorax ... connected to it’, suggest that it may have been a meraspid. It was not illustrated and has since been destroyed by fire, so its identity is indeterminate, but the fact that it was longer than wide suggests that it did not belong to Bumastella spicula. Bumastella sp. Plate 2, figures 16, 19 Material. A single pygidium from PL1989. Remarks. This pygidium differs from similarly sized pygidia of Bumastella spicula , including those from the same locality, and apparently belongs to a separate species. The differences from B. spicula include a more elongate outline, much lower convexity, a narrower axis anteriorly, a narrower (tr.) articulating facet and a correspondingly wider anterior pleural margin adaxial to the facet, a shallower holcos that does not extend as far backwards, finer pitting on the exterior of the exoskeleton, and several prominent ridges around the lateral and posterior margins instead of a single one. This is the only other species-group form of Bumastella known. Genus bumastus Murchison, 1839 Type species. By monotypy; Bumastus Barriensis Murchison. 1839; Barr Limestone Member of the Coalbrookdale Formation; Flay Head lime works. Great Barr, West Midlands Metropolitan County, UK. Other species. B. danielsi (Miller and Gurley, 1893), B. graftonensis (Meek and Worthen, 1870), B. ioxus (Hall, 1867). Diagnosis. See Lane and Thomas 1978u, p. 11. Remarks. The taxonomic problems that effacement in trilobites has caused historically are well illustrated by this genus. Since 1839, many species of Ordovician and Silurian effaced trilobites have been referred to Bumastus which, until the early part of the twentieth century, was almost universally considered to be a subgenus of Illaenus (e.g. Barrande 1852; Burmeister 1843; Salter 1867; Holm 1886; Vogdes 1890; Weller 1907). Most of these species have since been assigned to other genera. We consider that only the three species listed above, in addition to the type species, can be assigned with confidence to Bumastus. Stratigraphical range and distribution. ?Late Llandovery to earliest Ludlow; USA (Arkansas, Illinois and Oklahoma) and UK (Welsh Borderland and West Midlands). Bumastusl sp. Plate 4. figures 17-18 Material. A single rostral plate from PL 1995. HOLLOWAY AND LANE. EFFACED SILURIAN TRILOBITES 873 Remarks. This rostral plate cannot be assigned to any of the other effaced styginid species known from PL 1995. The specimen is tentatively assigned to Bumastus because it has an upwardly flexed flange posteriorly, in this respect resembling the rostral plates of B. barriensis (see Lane and Thomas 1978a, pi. 2, fig. 1 b— c) and B. cf. ioxus (Hall, 1867). The rostral plates of those species differ from the present specimen, however, in that the line along which the flange is flexed upwards is strongly convex backwards rather than transverse, and the flange itself is more concave (sag.) and is smooth instead of bearing well developed terrace ridges. Genus excetra gen. nov. Derivation of name. Latin, referring to the fanciful resemblance of the cephalon, when viewed anterolaterally, to the head of a snake; gender feminine. Type species. Excetra iotops gen. et sp. nov. Diagnosis. Cephalon strongly convex in transverse profile, in sagittal profile gently convex in posterior half and strongly convex in anterior half. Axial furrow subparallel to sagittal line immediately behind lunette but diverging backwards closer to posterior cephalic margin, diverging gently forwards in front of lunette and dying out just in front of glabellar midlength (sag.); omphalus and anterolateral internal pit absent. GO large, elliptical, not reaching axial furrow, situated less than its own length from posterior margin; G1 longer (exsag.) than GO, kidney-shaped, extending close to axial furrow anteriorly; G2 comma-shaped; G3 small, transverse. Posterior fixigenal margin with well-developed articulating flange bounded anteriorly by furrow that is flexed backwards distally. Eye small, of low convexity, with posterior edge transversely opposite midlength of lunette; socle absent. Posterior branch of facial suture weakly diverging backwards; anterior branch strongly diverging. Genal angle broadly rounded. Librigenal doublure with flattened facet on posterior edge. Connective suture converging backwards across anterior part of doublure and diverging backwards across posterior part, meeting inner edge of doublure close to outer end of hypostomal suture. Rostral plate gently inflated medially; posterolaterally with acute, upwardly curved projections abaxial to strongly transversely arched hypostomal suture. Thoracic axis parallel-sided, comprising about half segmental width (tr.) ; articulating furrows present on axial rings; axial furrow distinct; pleurae steeply downturned abaxial to narrow (tr.), horizontal proximal portion. Pygidium moderately to strongly convex, with articulating half ring defined by well impressed articulating furrow; anteriormost pleural furrow very weakly defined adaxially, holcos absent; articulating facet rather weakly defined posteriorly. Remarks. The cephalon of Excetra resembles that of Ligiseus , known from the type species, L. arcanus Lane and Owens, 1982 (p. 47, pi. 3, figs 4—8; fig. 3) from the uppermost Llandovery or lowest Wenlock of western North Greenland, and L. smithi Adrain, Chatterton and Blodgett, 1995 (p. 726, figs 2. 1-2.2, 2.4-2.15, 3.13, 3.15-3.16) from the upper Llandovery of Alaska. The similarities include the moderate convexity of the cephalon, the axial furrow that is subparallel immediately behind the lunette and posteriorly divergent farther backwards, the absence of the omphalus, the articulating flange on the posterior cephalic margin bounded in front by a distinct furrow, the medially inflated rostral plate, and the thorax with relatively narrow (tr.), subparallel- sided axis, deep axial furrow, and articulating furrows on each of the axial rings (see Adrain et al. 1995). The cephalon of Ligiseus differs from that of Excetra in that the eye is much larger and is situated farther back, with its anterior edge opposite the front of the lunette; the axial furrow is more distinct in front of the lunette; GO and G1 both extend laterally to the axial furrow, GO is situated farther from the posterior cephalic margin and G1 is sub-quadrate rather than kidney- shaped; G2 is ovate rather than comma-shaped; the anterior branch of the facial suture is more divergent; the genal angle has a short, broad spine; the connective suture apparently converges backwards across the entire doublure instead of diverging across the posterior part; and the rostral 874 PALAEONTOLOGY, VOLUME 41 plate apparently lacks upturned projections posterolaterally. The pygidium of Ligiscus, very poorly known in the type species but well documented in L. smithi , is not similar to that of Excetra, being much less convex and having a distinct axis and well-defined pleural ribs and furrows. Excetra iotops sp. nov. Plate 3, figures 1-19; Plate 4, figures 1-10, 13 Derivation of name. Combination of Greek ‘iota’ - small, and ‘ops’ - eye. Holotype. Cephalon NMV P 1 447 1 3 (PI. 3, figs 1-3); from PL1989. Paratypes. Cephala NMV P144712, NMV P144717; cranidia NMV PI 447 1 4-P 1 447 1 6, P144718, P144721, P144723-P 144725, P144727, P144730, P144733; librigenae NMV P144736-P144737; rostral plates NMV P144739, P144903; incomplete thorax NMV P144751; pygidium with attached thoracic segment NMV P144731 ; pygidia NMV P144734, P144741-P144747, P144749-P144750, P144752, P144754; all from PL1989. Other material. Two fragmentary cephala, five cranidia and five pygidia from PL1989. Diagnosis. As for the genus. Description. Cephalon c. 80 per cent, as long as wide (sag.) in dorsal view, widest just in front of genal angle; anterior and lateral margins uniformly curved; posterior margin of glabella gently convex backwards. Glabella gently convex (tr.) in posterior half, slightly more than half maximum width of cephalon at posterior margin, width at lunette c. 75 per cent, of posterior width. Median pit present on glabellar interior opposite posterior edge of GO. Lunette sub-circular, situated more than its own length from posterior edge of cephalon. Eye situated more than twice its own length from posterior cephalic margin; palpebral lobe bounded adaxially by weak furrow. Anterior and posterior branches of facial suture meeting cephalic margin approximately on same exsagittal line as palpebral suture; posterior branch with gentle sigmoidal curve; anterior branch almost straight just in front of eye, where it diverges at about 40° to sagittal axis, and broadly curved anteriorly. Posterior articulating flange strongly downturned at mid-width (tr.); articulating furrow deeper than axial furrow. Cephalic doublure expanding and greatly increasing in convexity anteromedially towards abaxial end of hypostomal suture; facet on posterior edge of doublure not extending abaxially as far as genal angle. Median part of hypostomal suture gently arched in transverse profile, convex backwards in ventral profile; lateral part of suture deflected posterolaterally and dorsally. Narrowest (tr.) part of rostral plate situated on transverse line through median part of hypostomal suture. Thoracic axial rings strongly arched (tr.), decreasing very slightly in length sagittally; articulating furrows short (sag., exsag.) and shallow. Inner part of pleurae (about 40 per cent, of transverse width) with very short (exsag.) articulating flange on anterior edge; outer part of pleurae gently convex (tr.), with pointed tips. Pygidium 1 10-120 per cent, as wide as long (sag.) in plan view. Articulating half ring slightly less than half maximum pygidial width, gently convex (sag., exsag.). Adaxial part of anterior pleural margin with narrow (tr.) EXPLANATION OF PLATE 3 Figs 1-19. Excetra iotops gen. et sp. nov.; locality PL1989, Mirrabooka Formation. 1-3, NMV P144713, holotype; cephalon, dorsal, anterior and lateral views; x4-5. 4, NMV P 1 447 1 7 ; cephalic doublure, ventral view; x 4. 5-6, NMV PI 44903; rostral plate, ventral and posterior views; x 3-5. 7, NMV PI 44730; cranidium, palpebral view; and NMV P 14473 1 ; pygidium with posteriormost thoracic segment (see PI. 4, fig. 6), oblique view; x 4. 8-10, NMV P144727; smallest cranidium, palpebral, anterior and lateral views; x 8. I I 12, NMV P144723; cranidium, palpebral and lateral views; x 6. 13, NMV P 1 447 1 8 ; cranidium, palpebral view; x 5. 14-15, 18-19, NMV P144712; largest cephalon, palpebral, lateral and oblique views; x 3; and detail showing muscle scars on interior of glabella and fixigena; x 5. 16-17, NMV P144737; librigena, oblique dorsal and oblique ventral views; x 5. PLATE 3 HOLLOWAY and LANE, Excetra 876 PALAEONTOLOGY, VOLUME 41 articulating flange similar to, but weaker than, that on cephalon; abaxial to flange, anterior edge of articulating facet is weakly deflected forwards in plan view. Doublure steeply inclined, increasing slightly in width toward sagittal line where it is c. 40 per cent, of pygidial length in plan view; outer part of doublure gently convex, inner part gently concave and bearing about six pairs of weak radial furrows. Sculpture. Fine, dense pits extend over external surface of cephalon and pygidium. Marginal band of subparallel terrace ridges present on anterior and lateral parts of cephalon, and curving inwards for a short distance adaxial to genal angle; this band very narrow (one to two ridges wide) posterolaterally, widest anteromedially (ten to thirteen ridges wide). Upturned part of cephalic doublure abaxial to hypostomal suture with terrace ridges more widely spaced than on narrow outer portion of doublure; terrace ridges on outer portion of doublure diverge adaxially onto rostral plate. Pygidium with terrace ridges present anterolaterally on dorsal surface, where they are largely restricted to articulating facet, running subparallel to pygidial margin on front of facet and curving posterolaterally farther back ; one or two ridges extend along pygidial margin behind facet to about 75 per cent, pygidial length from anterior in plan view. Terrace ridges on outer part of pygidial doublure more closely spaced than on inner part of doublure. Muscle scars. Glabellar muscle scars faint on external surface; on interior slightly raised (i.e. impressed on internal mould), sharply delimited and with a weak dendritic pattern on GO and Gl. GO longer (exsag.) than wide, length equal to that of eye; Gl with posterior edge opposite front of lunette and anterior edge opposite front of axial furrow, extending closer to axial furrow anterolaterally than does GO ; G2 broader abaxially than adaxially, not extending as far abaxially as Gl, anterior edge at about 25 per cent, glabellar length from anterior in palpebral view; G3 close to G2 and extending farther abaxially. Interior of fixigena with numerous, small, raised scars, except on lunette (PI. 3, fig. 9). Pygidia with a pair of weakly impressed, sub-circular or exsagittally elongated scars (slightly raised on interior), situated either side of sagittal line a short distance behind articulating furrow (PI. 4, fig. 5); in some specimens, these scars joined to articulating furrow by faint, anteriorly diverging furrows (PI. 4, fig. 8). Lateral and posterolateral to these paired scars, interior of pygidium has numerous, mostly small, raised scars, some of which are arranged in six or more radial rows (PI. 4, fig. 5) that are reflected on external surface of some specimens as extremely faint furrows; larger, radially elongated scars present towards adaxial ends of some rows (PI. 4, fig. 9). Ontogeny. The two smallest cranidia (sagittal length 3-3 mm in palpebral view; PI. 3, figs 8-10) show distinct differences from the largest ones on which the preceding description is based. The differences are as follows. 1 . The axial furrow is deeper, especially anteriorly where it extends almost to the cranidial margin, diverging quite strongly forwards from a point transversely opposite the front of the palpebral lobe. 2. The glabella is narrower in its posterior half (in relation to the sagittal length of the cranidium and the width across the palpebral lobes). 3. The occipital furrow is present as a shallow depression that is longer (sag., exsag.) than the occipital ring and contains muscle scar GO laterally. 4. The occipital ring is slightly inflated medially and bears two median tubercles: a larger one on the posterior edge of the ring and a smaller one just in front. 5. Gl is more distinct on the exterior of the exoskeleton. 6. The fixigena behind the front of the palpebral lobe is not as steeply declined abaxially (compare PI. 3, figs 2, 9). 7. There is a shallow depression on the front of the fixigena, running subparallel to and close to the anterior cephalic margin. 8. The palpebral lobe is relatively longer, the palpebral furrow is more distinct, and there is a weak eye ridge directed anteromedially from the front of the palpebral lobe. 9. The anterior and posterior branches of the facial suture diverge more strongly from either end of the palpebral lobe. 10. The lunette is larger and extends farther back, to about its own length from the posterior edge of the cephalon. Remarks. The position of the paired muscle impressions on the anteromedian part of the pygidium of Excetra iotops, and the fact that in some specimens they are joined to the articulating furrow by a faint, anteriorly diverging furrow, suggest that they are homologous with the pits at the posterior end of the furrow that divides the pygidial axis longitudinally in some non-effaced styginids (see HOLLOWAY AND LANE: EFFACED SILURIAN TRILOBITES 877 Planiscutellum kitharos Lane and Thomas, 1978a, pi. 6, fig. 4b). Also probably homologous are the smooth, ovate areas on the posterolateral part of the pygidial axis of Meroperix ataphrus , described and illustrated by Lane (1972, p. 345, pi. 60, fig. 4b). This evidence suggests that the pygidial axis of Excetra iotops is very short, as in other styginids. Genus lalax gen. nov. Derivation of name. Greek ‘frog’, alluding to the protuberant eye and palpebral area; gender masculine. Type species. Lalax olibros gen. et sp. nov. Other species. L. bandaletovi (Maksimova, 1975); L. bouchardi (Barrande, 1846) (= Bumastus praeruptus Kiaer, 1908; see Helbert et al. 1982, p. 133); L. chicagoensis (Weller, 1907); L. clairensis (Thomas, 1929); L. hornyi (Snajdr, 1957); L. inflatus (Kiaer, 1908) (= Bumastus phrix Lane and Thomas, 1978a; see Helbert 1984, p. 134); L. kattoi (Kobayashi and Hamada, 1984); L. lens sp. nov.; L. xestos (Lane and Thomas, 1978a); L.l sakoi (Kobayashi and Hamada, 1984); L.l transversalis (Weller, 1907). Diagnosis. Cephalon strongly convex (sag.), curvature in sagittal plane subtending more than 90°, height in lateral profile greater than or equal to sagittal length. Omphalus and anterolateral internal pit present. Axial furrow diverging moderately behind and immediately in front of lunette, dying out anteriorly behind omphalus. Eye large, situated less than its own length from posterior cephalic margin; socle not strongly convex (tr.). Posterior branch of facial suture strongly diverging backwards; anterior branch diverging moderately forwards. Genal angle broadly rounded. Rostral plate sub-triangular, gently convex (sag., exsag.) over anterior 70 per cent, and gently concave in posterior part, without upturned posterior flange; connective suture meeting hypostomal suture close to sagittal line; vincular furrow present across posterior edge of doublure. Thorax with very wide, gently arched axis comprising 60-70 per cent, segmental width (tr.); axial furrow weak; fulcrum situated very close to axial furrow; pleurae abaxial to fulcrum almost continuous in slope with lateral part of axial rings. Pygidium moderately convex (sag., exsag.), lenticular in dorsal view, maximum width just in front of midlength; anteriormost pleural furrow and holcos very weak or not defined. Terrace ridges present over most of dorsal surface of cephalon and pygidium. Remarks. Bumastus is most easily distinguished from Lalax by its rostral plate that is lenticular rather than triangular in outline in ventral view and has a vertical, concave (sag.) posterior flange, and by the connective suture meeting the hypostomal suture a little farther from the sagittal line. Distinguishing the genera may be difficult in the absence of information on the rostral plate, but in Bumastus GO and G1 are confluent rather than separate (compare Lane and Thomas 1983, text-fig. 2a, d); the omphalus is absent, although the anterolateral internal pit may be present; the visual surface is longer (exsag.) and narrower (tr.), with upper and lower margins parallel over almost their entire length; and the socle is more convex (tr.) and is separated from the visual surface by a deeper furrow. In the presence of the omphalus and the anterolateral internal pit, and the outline of the rostral plate, Lalax is similar to Cybantyx (see Lane and Thomas 1978a, pi. 5, figs 1-8). Cybantyx differs from Lalax in having a narrow, upturned anterior and lateral cephalic border; the lunette is situated slightly farther forwards, with its anterior edge slightly in front of the anterior edge of the eye in palpebral view; the axial furrow is more distinct in front of the lunette, extending as far as the omphalus; the anterior branch of the facial suture converges weakly in front of the palpebral lobe instead of diverging; the posterior part of the rostral plate is not concave (sag.); and the pygidium is longer. Litotix Lane and Thomas, 1978a, with type and only known species L. armata (Hall, 1865; see annotation of this reference below; Lane and Thomas 1978a, pi. 4, figs 8-18; text-fig. 4a-e) resembles Lalax in the convexity and proportions of the cephalon and pygidium, and in the presence of the omphalus. The rostral plate of Litotix is unknown, but the weak sagittal carina on the 878 PALAEONTOLOGY, VOLUME 41 cephalon, the axial furrow that extends anteriorly to the omphalus, the absence of the anterolateral internal pit, and the spinose genal angle are differences from Lalax. Two species are assigned to the genus with question. L. sakoi (see Kobayashi and Hamada 19856, pi. 30, fig. 1) is known only from a cranidium, but may be synonymous with L. kattoi (see Kobayashi and Hamada 19856, pi. 30, fig. 5), which is from the same locality and appears to differ only in its larger size; also possibly belonging to the same species is the pygidium assigned to Bumastus glomerosus by Kobayashi and Hamada (1987, p. 110, fig. 1a, 2 . 1 a— 3400 determinable bivalves; Liljedahl 1984), but still unusually numerous compared with most Palaeozoic chiton occurrences. A fairly low energy depositional environment is evident from the generally small degree of wear and breakage. The chiton assemblage includes Thairoplax pelta , TP afif. peltal, Plectrochiton tegulus , Alastega lira , Heloplax papilla , Enetoplax decora , Arctoplax ornata, head B and head/tail C, as well as a large number of C. actinis (Cherns 1998). The preservation of silicified sclerites shows little distinction in crystal size between surface and inner layers, superficially a mosaic of fairly small, inward growing quartz crystals (Schmitt and Boyd 1981, cf. Pattern 1, evident in C. actinis from this locality; Cherns 1998). Evidence for rapid precipitation of chalcedony and fine quartz suggests relatively high silica concentrations. Details of surface features are well preserved, indicating early cementation of the micritic carbonate matrix, before delayed precipitation into dissolution cavities. Preserved shell thickness and convexity indicate replacement before significant compaction. Some of the silicified material is from Klintebys-1, also in the Halla Formation. These specimens are more coarsely crystalline and beekitized, fairly worn, with loss of surface detail. Fragmented specimens typically have sealed edges. The preservation suggests coarser surrounding sediment, a higher energy environment producing more fragmentation, and again delayed precipitation into cavities. Coarser replacement might also indicate slower precipitation and lower silica concen- trations than at Mollbos-1 (Schmitt and Boyd 1981). As with Mollbos-1, the preservation of the convexity of shells means that replacement preceded compaction. The Klintebys collections yielded only A. lira , including head and tail sclerites, together with Chelodes spp. (Cherns 1998). Gen. A from Angvards-4 in the Hamra Formation shows similar coarse beekitization. TERMINOLOGY AND MEASUREMENTS The Silurian chitons are paleoloricates, lacking the sutural laminae for insertion beneath the adjacent plate which characterize neoloricate, and hence living chitons (e.g. Smith 1960). The measurements and terminology for paleoloricate sclerites with a posterior apex, ventral apical area and mixoperipheral growth follow those outlined by Cherns (1998, text-fig. 2). Most chitons, paleoloricates and neoloricates, have a ventral extension of the outer dorsal shell layer, the tegmentum, to form the apical area (Smith and Toomey 1964). In contrast, several of the paleoloricate chitons described here have a dorsal apex and holoperipheral growth style, typically found only in tail sclerites of Recent neoloricate chitons. These sclerites have a prominent raised dorsal apex, or mucro, which corresponds on the ventral surface to a deep subapical cavity, and text-fig. 1. Map of Gotland showing the geological succession of Silurian (upper Llandovery to Ludlow) strata, all localities for chitons in existing Riksmuseum collections (italics) and new silicified collections (bold italics). (For discussion of the locality Atlingbo, see Cherns 1998.) 942 PALAEONTOLOGY, VOLUME 41 growth lines anterior field i i i i i A posterior field lateral field subapical cavity groove in thickened ventral surface thickened pad lateral cavity B text-fig. 2. Diagram of holoperipheral sclerites of Heloplax papilla gen. et sp. nov., to illustrate terminology used in description of A, dorsal and B, ventral surfaces. Intermediate sclerites have a prominent mucronate sub- central apex, elevated anterior and posterior shell fields, and transverse flexure through the depressed lateral fields. A line of coarser granular ornament delimits the lateral from posterior shell fields. The sculpted ventral surface of a thickened sclerite has additional, smaller lateral cavities as well as the deep, anteriorly slanting subapical cavity, and a pattern of longitudinal and oblique furrows and pads around the cavities. Scale bar represents 1 mm. lateral field subapical B text-fig. 3. Diagram of holoperipheral sclerites of Enetoplax decora gen. et sp. nov. to illustrate terminology used in description of a, dorsal and B, ventral surfaces. Intermediate sclerites have an anteriorly displaced mucronate apex, elevated short anterior shell field, long, gently curved and elevated posterior field, gentle transverse flexure through short, depressed lateral fields; lateral and posterior fields without distinct separation. The ventral surface of thickened sclerites has a deep, anteriorly slanting subapical cavity, nearer to the anterior margin and associated with shallower development of pads and longitudinal furrows than in Heloplax papilla (Text-fig. 2). Scale bar represents 1 mm. dorsal shell with anterior, lateral and posterior fields. Terminology for such plates in two genera is shown in Text-figures 2-3. In the systematic descriptions below, figured specimens are indicated by an asterisk against specimen numbers. CHERNS: SILURIAN POLYPLACOPHORA 943 SYSTEMATIC PALAEONTOLOGY Class polyplacophora de Blainville, 1816 Subclass paleoloricata Bergenhayn, 1955 Remarks. Suprageneric classification requires revision encompassing other Palaeozoic chitons, and will be considered in a wider review. Genus gotlandochiton Bergenhayn, 1955 Type species. G. interplicatus Bergenhayn, 1955 (p. 15, pi. 1, fig- 6; pi. 2, fig. 4 (reconstruction)), by original designation, from the Upper Wenlock, Silurian, of Gotland, Sweden. Diagnosis (emended from Bergenhayn 1955). Intermediate sclerites broad, arched, with straight, deep, trapezoidal side slopes; jugum rounded, jugal angle close to perpendicular; shell fields not evident. Wide anterior margin gently convex to transverse; anterolateral corners rounded, anterolateral margins short, parallel to slightly divergent; posterolateral margins longer, straight, tapering rapidly across triangulate posterior shell to posterior apex; apical angle close to perpendicular. Ornament of low rounded ridges and narrow grooves parallel to growth lines. Apical area apparently broad and short. Remarks. Gotlandochiton was erected by Bergenhayn (1955) to include four new species described from Gotland. The original generic diagnosis stated that the form of the intermediate sclerites resembled that of most living chitons, with distinct shell fields and with jugal or complete coverage (across the following sclerite). The new family Gotlandochitonidae Bergenhayn, 1955 had a more discrete diagnosis of intermediate sclerites wider than long, variable in shape within the genus, and with weak but distinct shell fields. In the Treatise on invertebrate paleontology (Smith 1960, p. 150), relatively small size was noted as a family character. Bergenhayn (1955) distinguished this genus from Chelodes on the basis of sclerites that were wider than long, and with distinct shell areas/fields. In the type species, the broad sclerite has straight side slopes flexed across the jugum, but shell fields are not evident. However, Bergenhayn (1955) also erected and included three other species within Gotlandochiton '. G. laterodepressus, G. troedssoni and G. birhombivalvis. Of these, the first two, in which central and lateral shell fields are developed, have been synonymized with C. gotlandicus (Cherns 1998), and Chelodes is now recognized as including species that have distinct shell fields (Cherns 1998). For Chelodes , Cherns (1998) noted that sclerites only consistently become longer than wide with increasing size, so that as a criterion particularly for smaller sclerites this is of limited value. G. birhombivalvis is transferred here to the new genus Thairoplax, described below. Smith and Toomey (1964) noted that Gotlandochiton should display clearly defined shell areas, and suggested an ‘apical area less than L5 mm wide, extending across the entire posterior margin or present mainly in the vicinity of the valve apex’ (p. 18). G. hand Smith, in Smith and Toomey, 1964, from the lower Ordovician of southern Oklahoma, USA, has broad, rectangular flexed sclerites which have a very narrow band-like ventral apical area and some ventral transverse thickening (Smith and Toomey 1964). The emended generic diagnosis above does not include shell fields, the approximately straight posterior margin in G. hand compares with a triangulate posterior shell in both Gotlandochiton interplicatus and Thairoplax gen. nov., and the very short apical area is also apparently different in form from those of both genera. Gotlandochiton interplicatus Bergenhayn, 1955 Plate 1, figure la-d v* 1955 Gotlandochiton interplicatus Bergenhayn, p. 15, pi. 1 fig. 6; pi. 2 fig. 4 [reconstruction]. 944 PALAEONTOLOGY, VOLUME 41 1960 Gotlandochiton interplicatus Bergenhayn; Smith, p. 150, fig. 34, 4 [reconstruction, Bergenhayn 1955], 1975 Gotlandochiton interplicatus Bergenhayn; Van Belle, p. 125. 1977 Gotlandochiton interplicatus Bergenhayn; Sirenko and Starobogatov, p. 31. 1987 Gotlandochiton interplicatus Bergenhayn; Smith and Hoare, p. 34. Material and locality. Holotype RM Mo6012*, intermediate sclerite, with fragment of adjacent plate; Klints Othem (= Spillings 1-2, Laufeld 1974; Jaanusson 1986), Gotland; Slite Formation, Slite g. Upper Wenlock (Homerian). Diagnosis. As for the genus. Description (emended from Bergenhayn 1955). Flolotype an intermediate sclerite in limestone matrix obscuring ventral surface, right posterolateral edge broken ; fragment of anterior left portion of second, more posterior sclerite partially covered only by apex. Broad (width 12-3 mm), arched sclerite, wider than long (length 11-3 mm), with straight and deep, trapezoidal side slopes (PI. 1, fig. lb-c), jugal ridge rounded, jugal angle 98°. Shell fields not evident. Anterior margin wide and gently convex, rounding into short anterolateral margins that are parallel to slightly divergent. Posterolateral margins longer, straight, tapering rapidly to posterior apex, apical angle 94°. Maximum width of sclerite near posterolateral corners, behind mid-length. Dorsal surface fairly worn, but ornament of low rounded ridges and incised narrow grooves ( = ribs of Bergenhayn 1955, p. 16) parallel to growth lines, i.e. to anterior and anterolateral margins, crossing posterolateral margins (PI. 1, fig. la-c). Granular sculpture identified by Bergenhayn (1955, p. 16; PI. 1, fig. lc) is a patchy, replacement fabric. In lateral profile, jugal ridge slightly convex (PL 1, fig. lb-c). In transverse profile, shell flexed across rounded jugal ridge (PI. 1, fig. Id), height/length 0-38. Remarks. On the basis of the limited overlap of the holotype onto a second sclerite, with coverage apparently confined to the jugal area, Bergenhayn (1955) deduced that the apical area was narrow and restricted to the apex. He commented that this would leave triangular areas of the body wall exposed laterally between plates (Bergenhayn 1955, p. 15). However, growth lines transect the posterolateral margins onto the ventral apical area (Cherns 1988), which would have spanned the breadth between posterolateral corners and was thus apparently wide, close to the maximum width. The broken right posterolateral margin, curved from apex to posterolateral corner (PI. 1, fig. lc, cf. the straight left margin in fig. la-b may represent breaking away of the apical area here. Its length (= median length; Cherns 1998, text-fig. 2) and shape of the anterior margin are unknown, the former at most the length from the posterolateral corners, and thus less than half the length of the sclerite. The apparently wide apical area suggests a greater degree of overlap of sclerites than is apparent in the specimen, where the plates may have separated during preservation. EXPLANATION OF PLATE 1 Fig. 1. Gotlandochiton interplicatus Bergenhayn, 1955; RM Mo6012, holotype; intermediate sclerite; Klints Othem, Gotland; Slite Group, Upper Wenlock (Homerian). la, dorsal view, showing also fragment of following sclerite; lb, left lateral view, showing also fragment of following sclerite; lc, right lateral view, broken posterolateral margin; Id, anterior view. All x3. Figs 2-5. Thairoplax birhomhivalvis (Bergenhayn, 1955); intermediate sclerites. 2, holotype, RM Mo6031; Visby, Gotland; ?Lower Visby Formation, Upper Llandovery (Telychian); ventral external mould. 3, RM Mo6023; Kalens kvarn, Gotland; lower Hogklint Formation, Lower Wenlock (Sheinwoodian). 3a, dorsal view; 3b, left lateral view; note coarse replacement fabric; 3c, anterior view. 4, RM Mo6024a; Kalens kvarn, Gotland; lower Hogklint Formation, Lower Wenlock (Sheinwoodian); dorsal view, posterior shell broken. 5, RM Mo6024b; Kalens kvarn, Gotland; lower Hogklint Formation, Lower Wenlock (Sheinwoodian); left lateral view, partly embedded sclerite. All x 3. Fig. 6. gen. A indet.; RM M0I6O.O6I; Angvards-4, Gotland; Hamra Formation, Upper Ludlow (upper Ludfordian); broken intermediate sclerite. 6a, dorsal view; 6b, ventral view; 6c, left lateral view; 6d, posterior view. All x 3. PLATE 1 CHERNS, Silurian chitons 946 PALAEONTOLOGY, VOLUME 41 A silicified sample from Angvards-4, from the younger, late Ludlow Hamra Formation, yielded one partial sclerite (gen. A, below) that has a broad form and distinctive shallow ridge-and-groove ornament somewhat similar to that of G. inter plicatus, and an unusual, transverse anterior margin to the apical area. Gotlandochiton birhombivalvis Bergenhayn, 1955, in which sclerites are not notably wide, has a V-shaped ventral apical area (Bergenhayn 1955, pi. 1, fig. 7). Until more material becomes available to verify this ventral feature in G. interplicatus , the genus should be restricted to the type species. G. birhombivalvis is therefore transferred to the new genus Thairoplax. gen. A indet. Plate 1, figure 6a-d Material, locality and horizon. One partial silicified (beekitized) intermediate sclerite, RM M0I6O.O6I*, Gotland (RN 631953 164607), Hamra Formation, Upper Ludlow (Ludfordian). Description. One broken and beekitized intermediate sclerite, the left side without the apex. Sclerite apparently wider than long, slightly cordate, with jugal flexure and straight side slope, without shell fields. Side slope fairly deep, trapezoidal (PI. 1, fig. 6c). Broad, transverse, slightly rounded anterior margin having only shallow and narrow median embayment, rounding into only slightly convex to straight anterolateral margin parallel to jugal ridge, rounding sharply into straight posterolateral margin tapering steeply towards posterior apex. Anterolateral and posterolateral margins apparently roughly equal in length. Strong ornament of rounded ridges and incised grooves parallel to growth lines. Maximum width at posterolateral corners, well behind mid- length. Ventral apical area broad, triangulate, with raised transverse anterior margin across breadth of shell to posterolateral corners. Apical length/length in specimen 0-28, estimated for complete sclerite possibly c. 0-3. Ventral surface smooth. Remarks. The broad shell has strong ornament of rounded ridges and narrow incised grooves parallel to growth lines that is fairly similar to that of the holotype of G. interplicatus , and unlike that of all other Gotland chitons. The transverse anterior margin is slightly embayed medially, by comparison with the smoothly rounded anterior in G. interplicatus , and the jugal angle on the broken sclerite appears to be slightly greater at c. 105°. The side slope is less deep, the antero- and posterolateral margins appear roughly equal in length, i.e. longer, and shorter and steeper, respectively. No shell fields are apparent. The broad apical area would imply complete coverage across the adjacent sclerite; its transverse anterior margin is unique among the Gotland chiton fauna. The specimen is notably younger than G. interplicatus , from the Upper Ludlow as opposed to Upper Wenlock. Because it is only a single specimen, from a stratigraphical horizon that has previously yielded only Chelodes gotlandicus Lindstrom, 1884, it has been left under open nomenclature. Genus thairoplax gen. nov. Derivation of name. From the Greek thairos , hinge, and p/ax, plate, to indicate the flexed form of sclerites. Type species. Thairoplax pelta gen. et sp. nov.; Upper Wenlock, Silurian, Gotland, Sweden. Diagnosis. Arched intermediate sclerites flexed longitudinally across jugum, jugal angle per- pendicular to slightly obtuse, side slopes straight, trapezoidal; anterior margin transverse; posterolateral shell triangulate, posterior apex pointed, acute. Ventral apical area broad, V-shaped, less than one-third of length. Shell becoming medium to large size, typically longer than wide, thicker medially, tapering outwards; ventral surface smooth, without localized thickening. Ornament of fine growth lines. Shell fields weak. Broad central and narrower lateral fields. CHERNS: SILURIAN POLYPLACOPHORA 947 Remarks. Thairoplax gen. nov. is similar to Chelodes in having fairly large, wedge-shaped sclerites with a posterior apex, but differs in its marked jugal flexure between straight trapezoidal side slopes, and in limited ventral thickening, greater medially, in contrast to the thick to massive, sculptured ventral surface in larger sclerites of Chelodes. Thairoplax differs from Gotlandochiton , which is similar in having flexed intermediate sclerites with straight side slopes, in having sclerites at least as long as wide, a transverse anterior margin, a more acute apex, and a distinctly V-shaped apical area. Paleochiton Smith, 1964 (P. kindbladensis Smith, in Smith and Toomey, 1964) and Kindblado- chiton Van Belle, 1975 (K. arbucklensis (Smith, in Smith and Toomey, 1964); Van Belle 1975) are monospecific early Ordovician genera from southern Oklahoma, USA, which have broadly rectangular intermediate sclerites, but both have transverse posterior margins, in contrast to the extended, triangulate posterior portion of the sclerite in Thairoplax. In Paleochiton the ventral apical area is short and bandlike. Kindbladochiton has a ventral transverse thickened ridge lacking in the smooth ventral surface of Thairoplax. Kluessendorf (1987) described chiton morphotype A from the middle-upper Silurian of Wisconsin, USA, for a flexed shell with flat side slopes and parallel lateral margins, which on general form and V-shaped anterior margin of the apical area could belong within Thairoplax (see discussion of T. birhombivalvis below). Thairoplax pelta gen. et sp. nov. Plate 2, figures 1-3 Derivation of name. From the Greek pelte, shield, to describe the shape of sclerites. Material , locality and horizon. Seven intermediate sclerites, Mollbos-l, Gotland, Halla Formation, Upper Wenlock (Homerian); holotype RM Mol59.901*, isolated plate with dorsal small bryozoan encrustation; syntypes RM Mol59.937, 159.952*, 159.972, 159.973*, 160.019, 160.026. Diagnosis. Shield-shaped intermediate sclerites flexed slightly obtusely across jugum, rounded jugal ridge, side slopes long; sclerites elongate. Ventral apical area short, apical length/length < 0-2. Description. Shield-shaped intermediate sclerites of medium size (mean length 15-3 mm, s.d. = 2-86, n = 6; holotype length 121 mm), elongate, mean length/width 1-4 (s.d. = 0-09, n = 6; holotype F38). Flexed and thickened slightly medially across rounded jugal ridge, tapering across flat, trapezoidal side slopes towards lateral margins; mean jugal angle 103° (s.d. = 3-8, n = 7; holotype 108°). Anterior part of shell roughly rectangular, long, with anterior margin close to transverse, very slightly embayed to convex, median length/length 099 (s.d. = 0-02, n = 6; holotype 1-0), anterolateral corners rounded. Straight anterolateral margins more than half the length of the sclerite, almost parallel, maximum width towards anterior, in front of midlength; posterolateral corners distinct; shorter, straight posterolateral margins tapering rapidly to acute posterior pointed apex, mean apical angle 69° (s.d. = 5-6, n = 7 ; holotype 70°). Dorsal sculpture of fine growth lines parallel to anterior and anterolateral margins, crossing onto ventral apical area (e.g. PI. 2, fig. 3a, c); parallel, prominent larger growth steps (e.g. PI. 2, figs a, c). The latest growth lines may be entirely ventral, forming a narrow ventral rim, e.g. holotype (PI. 2, fig. lb). Weak dorsal radial folds give poor definition of broad central and narrow posterolateral areas (PI. 2, figs la, 3a). Ventral apical area short, narrow band tapering across to posterolateral corners, to markedly V-shaped (PI. 2, figs b), mean apical length/length 0-18 (s.d. = 0-02, n = 5; holotype 0-10). Apical area ornamented with growth lines, anterior margin slightly raised above smooth ventral surface. Lateral profile (PI. 2, figs c) shows straight jugal ridge, roughly parallel anterolateral margin, transverse anterior margin and rapidly tapering, shorter posterolateral margin. Transverse section (PI. 2, figs d ) V-shaped posteriorly, becoming more rounded anteriorly, side slopes straight and tapering. Mean height/length 0-34 (s.d. = 0 05, n = 7; holotype 0-37). 948 PALAEONTOLOGY, VOLUME 41 Remarks. T. pelta differs from the similar-aged (late Wenlock) Gotlandochiton interplicatus in having elongate sclerites, weakly defined shell areas with a broad triangular central field, a bandlike to strongly V-shaped ventral apical area, and ornament of only fine growth lines. T. pelta is found among collections from Mollbos-l dominated quantitatively by C. actinis , from which it is easily distinguished by its flexed, shield-shaped, non-massive form. Thairoplax^. aff. peltal Plate 3, figure 4 Material, locality and horizon. Mollbos-l, Gotland, Halla Formation, Upper Wenlock (Homerian); one isolated intermediate sclerite, RM Mol60.020*. Description. Arched, medium sized intermediate sclerite, similar length to width, flexed across jugum, side slopes straight; length 121 mm, length/width 104, height/length 0-42, jugal angle 96°. Anterior margin transverse, slightly embayed, rounding through fairly long, gently convex and divergent anterolateral margins; maximum width at posterolateral corners, behind midlength. Posterolateral margins straight, similar length to anterolateral margins, tapering rapidly across triangulate posterior shell to pointed apex, apical angle 85°. Ventral apical area broad, V-shaped, tapering outwards, apical length/length 0-3 1 . Ventral surface smooth, not greatly thickened, but with rounded, low triangular transverse ridge between posterolateral corners, tapering anteriorly, posteriorly extending as narrow median pad flanked by low furrows towards anterior rim of apical area (PI. 3, fig. 4b). Dorsal weak low radial jugal fold across fairly narrow anterior embayment; fine growth lines. Remarks. This specimen occurred in a sample with T. pelta and C. actinis , and is broadly similar in its flexed form with straight side slopes and V-shaped apical area to T. pelta. However, it differs in having rounded anterolateral margins, slightly tapering anteriorly and of similar length to the posterolateral margins, in having a relatively long apical area (apical length/length 0-31, cf. mean 0T8 for T. pelta), and particularly in the localized ventral thickening into a low transverse ridge. It may represent an anterior intermediate sclerite of T. pelta , tapering towards a typically small head sclerite, but the pattern of ventral thickening, unseen otherwise in T. pelta, may indicate that it does not belong within this species (or genus?). Thairoplax birhombivalvis (Bergenhayn, 1955) Plate 1, figures 2-5 v* 1955 Gotlandochiton birhombivalvis Bergenhayn, p. 18, pi. 1, fig. 7; pi. 2, fig. 6 [reconstruction]. 1977 Gotlandochiton birhombivalvis Bergenhayn; Sirenko and Starobogatov, p. 31. 1987 Gotlandochiton birhombivalvis Bergenhayn; Smith and Hoare, p. 15. Material, locality and horizon. Four intermediate sclerites; holotype RM Mo6031*, external mould of ventral surface, Visby, Gotland, Lower Visby Formation (Visby a), Upper Llandovery (Telychian); syntypes RM Mo6023* and 6024* (two specimens), Kalens kvarn (= Kolens kvarn), Visby, Gotland, Hogklint Formation, Lower Wenlock (Sheinwoodian). EXPLANATION OF PLATE 2 Figs 1-3. Thairoplax pelta gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, upper Wenlock (Homerian); intermediate sclerites. 1, holotype (with small bryozoan encrustation on left anterior), RM Mol59.901 ; la, dorsal view; lb, ventral view; lc, right lateral view; Id, posterior view. 2, RM Mol59.973; 2a, dorsal view, showing prominent growth increment; 2b, ventral view; 2c, right lateral view; 2d, posterior view. 3, RM Mo 159.952; 3a, dorsal view, showing prominent growth increment, broad central shell field; 3b, ventral view; 3c, right lateral view; note weak fold defining broad central shell field, fine growth ornament; 3d, posterior view. All x 3. PLATE 2 CHERNS, Thairoplax 950 PALAEONTOLOGY, VOLUME 41 Diagnosis (emended from Bergenhayn 1955, p. 18). Intermediate sclerites with deep side slopes, jugal ridge sharp to more rounded, jugal angle almost perpendicular; anterolateral margins slightly divergent, posterolateral margins longer, maximum width at posterolateral corners, apical angle acute. Ventral apical area tapering rapidly outwards, apical length/length c. 0-3. Description. The holotype is a distorted, arched, ventral external mould, showing flat, deep trapezoidal side slopes diverging almost perpendicularly (87°) across a sharp jugal ridge, and with a broad, V-shaped apical area that tapers rapidly outwards from its central flexure, apical length/length 0-31. Triangulate posterior shell to pointed apex, apical angle 71°, straight posterolateral margins, longer than anterolateral margins, bordered by apical area (PI. 1, fig. 2). RM Mo6023 (PI. 1, fig. 3) shows only the dorsal surface, strongly arched (height/ length 0-58) across a jugal flexure (86°) sharper posteriorly but becoming more rounded anteriorly, and with weakly defined, broad triangular central shell field expanding from apex to outside anterolateral corners on deep trapezoidal side slopes. Sclerite fairly large (length 13-9 mm), slightly longer than wide (length/width 107), not greatly thickened. Anterior margin straight across flexed central area, anterolateral corners sharply rounded into straight to only slightly convex lateral margins, and longer, straight posterolateral margins tapering rapidly to apex; apical angle 73°. Growth lines, or possibly slightly stronger ornament of very shallow ridges and grooves, follow anterior and anterolateral margins, crossing onto the ventral apical area at posterolateral corners behind mid- length; larger growth increments indicated by spaced narrow ridges/grooves (PI. 1, fig. 3b). Two worn and coarsely replaced chiton specimens on RM 6024, both partially embedded in limestone, come from the same locality as RM Mo6023, from a slightly younger horizon than the holotype. One has the apex broken off, but shows part of the dorsal surface arched strongly across the jugal ridge, as in RM Mo6023, and appears from growth lines to have a transverse anterior margin and straight anterolateral margins (PI. 1, fig. 4). The other specimen shows the left dorsal surface including the apex, with a long straight posterolateral margin tapering to a pointed apex (PI. 1, fig. 5). Bergenhayn (1955) noted a granular sculpture preserved on one sclerite, presumably RM Mo6023, where coarse sparite replacement of the shell has produced an apparent surface pattern. Remarks. Bergenhayn (1955, p. 18) based this species, which he considered very distinct, on the four intermediate sclerites described above from Visby and Kalens kvarn, Visby, from the Upper Llandovery to Lower Wenlock. Specific characters, of two rhomboid shaped sides hinged along the midline at a jugal angle of c. 90°, an ornament of very low, evenly spaced growth lines (= larger growth increments), and total coverage across the adjacent plate (indicated by a broad apical area), he noted as combined with an absence of shell fields. Despite the lack of shell fields, he still assigned the species to Gotlandochiton on the basis that the shell form and, more questionably, complete coverage did not belong within Chelodes. In comparison with T. pelta, the older species T. birhombivalvis, based on more limited, non- isolate material, has deeper, shorter side slopes, a more perpendicular flexure, longer posterolateral margins across the triangulate posterior shell, and a longer V-shaped apical area. Intermediate sclerites are broader and less elongate. EXPLANATION OF PLATE 3 Figs 1-3. Plectrochiton tegulus gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). I, RM Mo 160.032, holotype; intermediate sclerite. la, dorsal view; lb, ventral view, note ?weak transverse ridge between anterolateral corners; lc, left lateral view; Id, posterior view. All x 5. 2, RM Mol59.942; intermediate sclerite. 2a, dorsal view; 2b, ventral view; 2c, left lateral view; 2d, posterior view. All x 5. 3, RM Mo 159.900; intermediate sclerite. 3a, dorsal view; 3b, ventral view; 3c, left lateral view; 3d, posterior view. All x 4. Fig. 4. Thairoplaxl aff. pelta!, Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian); RM Mo 160.020, intermediate sclerite. 4a, dorsal view showing weak narrow jugal fold; 4b, ventral view ; note low transverse ridge anterior to apical area, with tapering extension beneath apical area rim; 4c, left lateral view; 4d, posterior view. All x 3. PLATE 3 CHERNS, Plectrochiton, Thairoplaxl 952 PALAEONTOLOGY, VOLUME 41 Kluessendorf (1987) compared Morphotype A, an incomplete specimen showing the ventral surface, from the Racine Dolomite (Wenlock/Ludlow) of Wisconsin, USA, with T. birhombivalvis, on the basis of a flexed form with flat side slopes and parallel lateral margins. The elongate form and apparently short apical area are more similar to T. pelta , although there are insufficient diagnostic characters to allow close comparison. plectrochiton gen. nov. Derivation of name. From the Greek plektron, a tool for plucking a stringed instrument, to describe the triangulate shape of sclerites. Type species. P. tegulus gen. et sp. nov., from the Upper Wenlock, Silurian of Gotland, Sweden. Diagnosis. Broad and short, small low-arched triangulate sclerites, wider than long; transverse to gently convex anterior margin, rounded anterolateral corners, tapering straight posterolateral margins to broad, pointed posterior apex, apical angle almost perpendicular. Ornament of fine growth lines; no shell fields, jugal angle obtuse, c. 125°. Apical area approximately one-third of length, wide, tapering outwards to anterolateral corners, V-shaped to concave anterior margin. Ventral surface smooth, concave, triangulate to lozenge-shaped. Remarks. The small, broad and only gently arched, triangulate form of intermediate sclerites in Plectrochiton gen. nov., without shell fields and with only fine growth line ornament, is distinct from other genera of Palaeozoic chitons (e.g. Smith and Toomey 1964, p. 17). By comparison with other Gotland chitons, Chelodes Davidson and King, 1874 has commonly large, elongate, wedge- to heart-shaped intermediate sclerites, in some species with shell fields. Gotlando chiton Bergenhayn, 1955 and Thairoplax gen. nov. have medium to large, flexed sclerites with straight trapezoidal side slopes. The triangulate form distinguishes Plectrochiton gen. nov. from the roughly rectangular sclerites of Ordovician Paleochiton Smith, in Smith and Toomey, 1964, and Kindbladochiton Van Belle, 1975, and from the Ordovician-Cretaceous Ivoechiton Bergenhayn, 1955. Plectrochiton tegulus gen. et sp. nov. Plate 3, figures 1-3 Derivation of name. From the Latin tegulus , a tile, to describe the very low-arched form. Material, locality and horizon. Eight intermediate sclerites from Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian); holotype RM Mo 160.062*, isolated plate, syntypes 159.865-159.866, 159.874, 159.900*, 159.936, 159.942*, 160.009. Diagnosis. As for the genus. Description. Small and low-arched, short broad triangulate intermediate sclerites that are wider than long, without shell fields. Mean length 5-9 mm (s.d. = 2-2, n = 8; holotype 7 1 mm), mean length/width 0-87 (s.d. = 0 04, n = 6; holotype 0 86), mean jugal angle 125° (s.d. 3-8, n = 8; holotype 123°). Wide anterior margin straight to gently convex, with rounded anterolateral corners, no anterolateral margins, tapering straight posterolateral margins to broad, pointed posterior apex, mean apical angle 88° (s.d. = 91, n = 8; holotype 91°). Maximum width across anterolateral corners, well anterior of midlength. Ornament of fine growth lines parallel to anterior margin, transecting posterolateral margins onto ventral apical area. Apical area with mean apical length/length 0-29 (s.d. = 008, n = 6; holotype 0-35), wide, with a slightly raised anterior margin V- shaped to rounded and concave anteriorly (PI. 3, figs lb, 2b, 3b), tapering outwards along posterolateral margins to anterolateral corners. Ventral surface smooth, concave, triangulate to lozenge-shaped, may have slight transverse thickening across between anterolateral corners (PI. 3, fig. lb). CHERNS: SILURIAN POLYPLACOPHORA 953 Lateral profile triangular, fairly shallow, with flat to slightly convex dorsal surface, gently convex anterior margin (PL 3, figs lc, 2c, 3c). Transverse section shallow, shell thicker medially, tapering laterally (PI. 3, figs Id, 2d, 3d). Remarks. P. tegulus gen. et sp. nov. is distinguished from small sclerites of C. actinis Cherns, 1998, which co-occur in samples from Mollbos, by the absence of anterior invagination, lower length to width ratio, and shallower transverse profile with a more obtuse jugal angle. The slightly elevated anterior rim to the apical area may indicate muscle attachment along this margin (Cherns 1998). The fairly small sized, triangulate sclerites could represent head or tail sclerites, although they lack features commonly found in such sclerites of chitons, such as distinct, commonly radiate, ornament of head sclerites, and a prominent mucro in tail sclerites (e.g. Smith 1960; Hyman 1967). For C. actinis , ovoid, ornamented plates that co-occur with the intermediate sclerites have been described as head sclerites (Cherns 1998). alastega gen. nov. Derivation of name. From the Latin ala , wing, and Greek stege, roof, to described the winged form of the sclerites. Type species. A. lira gen. et sp. nov. from the Upper Wenlock, Silurian of Gotland, Sweden. Diagnosis. Small arched sclerites, flexed across jugum, with triangulate, pointed posterior apex; slightly elevated and rounded, broad triangulate jugal shell field flattening anteriorly; apical angle nearly perpendicular, jugal angle slightly obtuse; ornament of shallow rounded ridges and furrows, growth lines, stronger on lateral fields. Intermediate sclerites small, wide and short, strongly arched and winged; jugal ridge rounded anteriorly, broad anterior embayment, side slopes deep and straight, triangulate posterior shell to apex. Ventral apical area short V-shaped band tapering across long posterolateral margins. Transverse ventral thickening forming V-shaped ridge, tapering outwards. Tail sclerites as long as wide, lower arched, more triangulate and weakly trilobed, shallower anterior embayment; jugal field elevated and rounded, side slopes shallower; apical area short, V- to U-shaped anterior margin, tapering across long posterolateral margins, ventral surface with transverse V-shaped thickening. Head sclerites small, elongate, ovoid, low arched; distinct rounded triangulate jugal field, posterior pointed apex; ventral surface smooth, concave, apical area not known. Remarks. Small, short and wide, winged and strongly arched intermediate sclerites with distinct transverse ventral thickening at around mid-length parallel to the short V-shaped apical area are characteristic of Alastega gen. nov. Ivoechiton (I. oklahomensis Smith, in Smith and Toomey, 1964; I. calathicolus Smith, in Smith and Toomey, 1964) and Kindbladochiton (K. arbucklensis Smith, in Smith and Toomey, 1964) from the lower Ordovician of Oklahoma, USA, have intermediate sclerites wider than long, with a transverse thickening across the ventral surface of sclerites, and with a posterior margin swept back from the apex or transverse (Smith and Toomey 1964). Alastega gen. nov. differs from Ivoechiton in having defined shell fields, and from both in its long straight posterolateral margins tapering across the triangulate posterior shell to a pointed apex, with a corresponding V-shaped ventral apical band. Alastega lira gen. et sp. nov. Plate 4; Text-figure 4 Derivation of name. From the Latin lira , plough ridge, to describe the dorsal ornament. Material locality and horizon. Mollbos- 1, Gotland, Halla Formation, Upper Wenlock (Homerian); 21 isolated sclerites (including four tail sclerites); holotype RM Mo 159.845*, intermediate sclerite; RM Mol 59.827, 954 PALAEONTOLOGY, VOLUME 41 159.846-159.847, 159.848*, 159.852, 159.876-159.882, 159.893-159.894, 159.917, 159.949, 159.987; tail sclerites 159.826*, 159.849*, 159.883, 160.010. Klintebys-1, Gotland, Halla Formation, Upper Wenlock (Homerian); 21 isolated sclerites (including two head, three tail sclerites); 160.036-160.037, 160.039, 160.041, 160.043-160.046, 160.048, 160.050-160.053, 160.057, 160.058*, 160.059, head sclerites 160.047, 160.060*, tail sclerites 160.038, 160.040, 160.049. Diagnosis. As for the genus. Description. Intermediate sclerites (PI. 4, figs 1-2; Text-fig. 4d-g) small, strongly arched and winged, flexed across rounded jugum, side slopes straight and deep. Mean length 3-6 mm (s.d. = 2-0, n = 28; holotype 2-7 mm), smaller sclerites much wider than long, becoming less so with growth, mean length/width 0-78 (s.d. = 018, n = 20; holotype 0-66). Jugal ridge flattening anteriorly over slightly elevated and rounded, broad triangulate jugal field, mean jugal angle 96° (s.d. = 8, n = 32; holotype 97°). Anterior margin broad, rounding through shallow median embayment across jugal field, mean median length/length 0-88 (s.d. = 0 06, n = 27; holotype 0-85). Strongly rounded anterolateral corners into short, slightly convex, divergent anterolateral margins, maximum width at posterolateral corners. Posterolateral margins longer, straight, tapering rapidly across triangulate posterior to pointed broad apex; mean apical angle 88° (s.d. = 12, n = 26; holotype 89°). Ornament of shallow rounded ridges and furrows, and growth lines, sinuate parallel to anterior and anterolateral margins, stronger on lateral fields (PI. 4, fig. la, c; Text-fig. 4d, f). Ventral surface with short, V- shaped apical area as slightly raised band across posterolateral margins, tapering outwards to posterolateral corners, mean apical length/length 0T7 (s.d. = 006, n = 15; holotype 0T9). Ventral surface smooth, with transverse thickened triangular ridge around midlength, V-shaped, thickest medially, tapering towards and flattening anteriorly and posteriorly (PI. 4, fig. lb, e, 2a-b; Text-fig. 4e), becoming relatively more posterior with increased size of sclerite. Lateral profile (PI. 4, fig. lc. Text-fig. 4f) gently convex dorsally, weak radial fold elevating low jugal area, deep side slopes to posterolateral corners, steep straight posterolateral margins, sinuate shallowing anterolateral to anterior margins. Transverse profile strongly arched across jugal flexure rounding anteriorly, side slopes straight, tapering outwards (PI. 4, figs ld-e, 2b; Text-fig. 4g), mean height/length 0-67 (s.d. = 0T8, n = 28; holotype 0-78). Thickened ventral ridge producing longitudinal flexure of ventral surface into two inclined planes, particularly evident in smaller specimens; inclined posterior profile with V-shaped ventral surface, angular ventral flexure (PI. 4, fig. 2b), inclined anterior profile with lunate ventral surface (PI. 4, fig. le). Tail sclerites (PI. 4, fig. 3; Text-fig. 4a-c) roughly as long as wide, mean length 3 6 mm (s.d 1-2, n = 5), mean length/width 104 (s.d. = 0-27, n = 3), lower arched, more triangulate and weakly trilobed. Elevated, rounded and broad triangulate jugal field, narrower, less convex lateral fields, mean jugal angle 101° (s.d. = 9, n = 5). Anterior margin more shallowly embayed across jugal field, mean length/length 0-93 (s.d. = 0-06, n = 4), strongly rounded anterolateral corners into very short anterolateral margins, long straight posterolateral margins tapering across triangulate posterior to pointed posterior apex, mean apical angle 88° (s.d. = 27, n = 5). Ornament of growth lines, low ridges/furrows poorly preserved. Ventral short apical area with V- to U- shaped elevated margin, tapering outwards across posterolateral margins (PI. 4, fig. 3b; Text-fig. 4b), mean apical length/length 0-2 1 (s.d. = 004, n = 3). Ventral surface smooth, flexed across broad transverse triangulate thickened ridge, greatest medially, V-shaped, around midlength. Longitudinal profile showing shallower side slopes, radial fold elevating jugal field, less steep posterolateral margins (PI. 4, fig. 3c; Text- fig. 4c). Anterior transverse profile lower arched, side slopes straight, mean height/length 0 42 (s.d. = 0 05, n = 5); inclined anterior ventral surface lunate, inclined posterior ventral surface with more angular flexure. EXPLANATION OF PLATE 4 Figs 1-3. Alastega lira gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). 1, RM Mo 159.845, holotype; intermediate sclerite. la, dorsal view; note rounded ridge-and-furrow ornament; lb, ventral view, showing V-shaped transverse ridge; lc, right lateral view; Id, posterior view; le, anterior, slightly tilted view to show lunate anterior surface of transverse ridge. 2, RM Mol59.848; intermediate sclerite. 2a, ventral view, showing V-shaped transverse ridge further anterior than in lb; 2b, posterior, slightly tilted view to show V-shaped posterior surface of transverse ridge. 3, RM Mo 159.849; tail sclerite. 3a, dorsal view, note elevated fold of central shell field ; 3b, ventral view, V-shaped transverse ridge well in front of apical area; 3c, right lateral view; 3d, anterior view. All x 15. PLATE 4 jTp' j> ., r ' Wr Silfe Sgp$£ Swjr ^v>-,*- ■*■ •- MUH^* Mpi.:$-v-.,%>-'. :'V'/'*U;!,,^ ^ijfD J|P h '. ■: ML .jwbs CHERNS, Alastega 956 PALAEONTOLOGY, VOLUME 41 text-fig. 4. Alastega lira gen. et sp. nov. a-c, RM Mo159.826; M611bos-l, Gotland; Halla Formation, Upper Wenlock (Homerian); tail sclerite, fragmented on left side. A, dorsal view, showing elevated fold of central shell field, b, ventral view, transverse ridge well in front of apical area, c, right lateral view, showing elevated central shell field. All x 20. d-g, RM Mol60.058; Klintebys-1, Gotland; Halla Formation, Upper Wenlock (Homerian); beekitized intermediate sclerite. d, dorsal view; note elevated shell central field, rounded ridge- and-furrow ornament. E, ventral view; note V-shaped transverse ridge. F, left lateral view, showing well developed ornament. G, posterior view. All x 5. h-k, RM Mo 160.060; Klintebys-1, Gotland; Halla Formation, Upper Wenlock (Homerian); head sclerite. h, dorsal view, showing elevated central field, rounded ridged ornament on lateral fields. I, ventral view, j, lateral view, showing elevated fold of central field, k, posterior view. All x 7. Head sclerites (Text-fig. 4h-k) known only from two specimens, both beekitized, one poorly preserved. Small, ovoid, low arched, elongate, with distinct rounded, triangulate jugal field; mean length 4-6 mm (s.d. = 0-57, n = 2), mean length/width F33 (s.d. = 0-40, n = 2), mean jugal angle 95° (s.d. = 1, n = 2). Anterior margin transverse to gently convex across jugal field, rounding into long, convex anterolateral margins, short straight posterolateral margins tapering rapidly across triangulate posterior to pointed apex; apical angle c. 88°. Ornament of low rounded ridges and furrows, growth lines, particularly developed on lateral fields. Ventral surface smooth, concave, deepest towards posterior, apical area unknown but probably very short. Lateral profile shallow, low fold elevating jugal field. Posterior profile low arched, mean height/length 0-29 (s.d. = 0 07, n = 2), short straight side slopes. CHERNS: SILURIAN POLYPLACOPHORA 957 Remarks. The thickened transverse ventral ridge characteristic of this genus associates the short, wide, high arched intermediate sclerites with relatively longer, more triangulate, lower arched tail sclerites. In particular, the lunate shape of the anterior ventral surface of the thickening is distinctive. In addition, in both these types of sclerite and in the elongate, ovoid and shallow arched head sclerites the dorsal surface has a triangulate, slightly elevated and rounded jugal field, and distinctive ornament of shallow rounded ridges and furrows, preserved better on the flatter lateral fields. All have almost perpendicular apical angles and slightly obtuse jugal angles. The ovoid elongate shape and shallow form of the head sclerites show similarities to those described recently for the large chiton C. actinis (Cherns 1998, text-fig. 4), although the A. lira sclerites are much smaller, with coarser ridged ornament, and a more distinct and rounded jugal field. The material of A. lira comes from Mollbos-1 and Klintebys-1, both from the Late Wenlock Halla Formation. All specimens from Mollbos-l are small but include some that are well preserved. Many of those from Klintebys-1 are poorly preserved and beekitized, but they include also larger examples, of both intermediate and tail sclerites. The size difference is notable, with several 7-9 mm long, compared with the means (including these) of 3-6 mm, and unfortunately the sclerites preserve poor detail. However, they do share the general form and characteristic ventral ridge in both intermediate and tail plates, and co-occur in samples with the more typical, small specimens; hence they are treated here as the same species. In intermediate sclerites, the ventral thickening becomes relatively more posterior as the shell lengthens, and its anterior lunate surface develops shallow sculpting to enhance lateral pads (Text- fig. 4e, cf. PI. 4, fig. lb). The ventral thickening in longer (i.e. larger) intermediate sclerites produces a natural balance, and presumed life position, with gentle anterior tilt of the jugal field, leaving the ventral anterior surface horizontal and the posterior surface behind the ridge elevated slightly. The intermediate sclerites show variation in the breadth of anterior embayment and degree of divergence of anterolateral margins. These features may relate to different positions of plates along the animal, in particular to narrowing of the broad intermediate sclerites anteriorly towards the elongate head sclerite. heloplax gen. nov. Derivation of name. From the Greek helos , nail, stud, and plax , plate, to describe the rounded form of sclerites. Type species. H. papilla gen. et sp. nov. from the Upper Wenlock (Silurian) of Gotland, Sweden. Diagnosis. Small, transversely elongate, ovoid intermediate sclerites with subcentral elevated mucronate apex, rounded margins. Fairly broad, vaulted triangulate anterior field; broader triangulate posterior field, concave becoming flattened to convex posteriorly, elevated; depressed lateral fields across transverse flexure. Maximum shell width slightly posterior of mid-length, at posterolateral corners. Dorsal concentric growth lines; distinct granular ornament, coarser anteriorly and laterally, also coarsening outwards; quincunx pattern but with line of larger granules demarcating posterior field. Ventral surface with small deep median subapical cavity, oblique towards anterior. Ventral thickening leading to strongly sculpted surface around subapical area. ?Tail sclerites smaller, relatively broad, vaulted convex anterior and posterior fields, coarser line of ornament within lateral fields. Remarks. Heloplax gen. nov. differs from all other paleoloricate chitons, except Enetoplax gen. nov. and Arctoplax gen. nov. described below, in having small ovoid intermediate plates with a dorsal mucronate apex, and concentric holoperipheral growth (Text-fig. 2). They thereby lack the ventral apical area of sclerites with a posterior apex, representing extension of the dorsal outer tegmentum onto the ventral surface, found in at least most paleoloricate chitons, and in neoloricate chitons except for tail sclerites (e.g. Pterochiton spatulatus, Pedanochiton discomptus; Smith and Toomey 1964; Debrock et al. 1984; Hoare 1989). In Heloplax , the sclerites are vaulted and convex anteriorly, and in most the concave post-apical posterior field becomes elevated. This shell morphology 958 PALAEONTOLOGY, VOLUME 41 suggests that these are intermediate sclerites, which can become imbricated, and not tail plates. Variations in morphology suggest that both intermediate and tail plates are represented, the latter being vaulted and convex both anteriorly and posteriorly. From the Lower Carboniferous (Mississippian) of Utah, Hoare (1989) described as a tail plate a single small sclerite of generally similar configuration to Heloplax , but with sutural laminae and hence a neoloricate chiton. Heloplax papilla gen. et sp. nov. Plates 5-6 Derivation of name. From the Greek papilla , bud, nipple, to describe the granular ornament. Material , locality and horizon. Mollbos-l, Gotland, Halla Formation, Upper Wenlock (Homerian); 25 isolated sclerites; holotype RM Mo 159.832*, intermediate sclerite; intermediate sclerites RM Mo 159.828*, 159.829*, 159.830-159.831, 159.833, 159.867*, 159.884, 159.891-159.892, 159.896, 159.898, 159.912, 159.920, 159.954, 159.968, 159.997, 160.011, 160.017; Ttail sclerites RM Mol59.834, 159.851*, 159.886-159.887, 159.889, 159.984. Diagnosis. As for the genus. Description. (Text-fig. 2). Small broad intermediate sclerites, transversely ovoid, with rounded margins; mean length 3 6 mm (s.d. = 0-3, n = 16; holotype 3 7 mm), mean length/width 0 70 (s.d. = 0 05, n = 16; holotype 0-67). Prominent sub-central pointed dorsal apex is elevated and mucronate; mean 0-48 (s.d. = 0 06, n = 17; holotype 0-46) of length from anterior. Maximum width only slightly posterior of apex at posterolateral corners, mean 0-54 (s.d. = 0 04, n = 16; holotype 0 57) of length from anterior. Concentric growth lines about apex, some more distinct growth increments forming low ridges towards outer part of dorsal surface (e.g. PI. 6, fig. la, d). Dorsal surface with vaulted, triangulate, transversely convex anterior and posterior fields, depressed, concave to flattened lateral fields. Anterior field fairly broad, arched, smoothly rounded and convex, elevating to apex, posterior field broader, concave behind apex, flattening and becoming convex to margins. Lateral fields rounding and becoming concave between anterior and posterior fields, most depressed close to apex, flattening outwards. Longitudinal profile strongly convex anteriorly to prominent apex, concave to flattened or becoming convex posteriorly, elevated (PI. 5, figs lc, 2c; PI. 6, figs lc, 2c). In transverse profile, gently convex, arched, deeper laterally. Granular dorsal ornament, coarser in anterior and lateral fields, and coarsening outwards, generally slightly finer granulation across posterior field. Quincunx arrangement of granules, but with band of coarser granules from apex to posterolateral corners demarcating posterior field (e.g. PI. 5, figs la, b, e, 2a-b). Anterior and lateral fields not differentiated by ornament. Ornament becoming less well defined in thickened sclerites (e.g. PI. 6, fig. la, c). Ventral surface smooth, with gentle convex transverse flexure corresponding to lateral depressed fields of the dorsal surface. Small deep median cavity beneath apex, slanted obliquely towards anterior, circular to ovoid. EXPLANATION OF PLATE 5 Figs 1-2. Heloplax papilla gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian); intermediate sclerites. 1, RM Mol59.832, holotype. la, dorsal view; note sub-central mucronate apex, line of coarser granules across depressed lateral shell fields; x 15. lb, ventral view, showing subapical cavity, flanking lateral depressions; x 15. lc, oblique ventral view, showing transverse flexure of sclerite across lateral fields, behind subapical cavity, obliquely directed lateral depressions, deep subapical cavity slanting anteriorly; x 1 5. Id, left lateral view; note line of coarser granules, elevation of posterior shell field behind mucronate apex; x 15. le, detail of lateral ornament, showing line of coarser granules; x 25. 2, RM Mol 59.867. 2a, dorsal view; note holoperipheral quincunx granular ornament coarsening outwards; x 15. 2b, ventral view; note deep subapical cavity, flanking oblique lateral furrows, transverse llexure of sclerite; x 15. 2c, right lateral view; note line of coarser granules across lateral field, strongly elevated posterior field; x 14. 2d, detail of granular ornament from anterior edge, with new growth increment inserted from below; x 50. PLATE 5 CHERNS, Heloplax 960 PALAEONTOLOGY, VOLUME 41 with smooth margins; mean 0-32 (s.d. = 0-05, n= 17; holotype 0-30) of length from anterior. Ventral thickening leading to surface sculpting around sub-apical cavity, and development of an additional, flanking lateral pair of more anterolaterally directed, smaller shallower cavities beneath apical region; subapical cavity bordered anteriorly by thickened pad, posteriorly by shallow longitudinal furrow flattening outwards, lateral cavities in shallow anterolateral furrows (PI. 6, fig. lb, d; also less thickened sclerites in PI. 5, figs lb-c, 2b). Anterior surface and flexed region becoming strongly sculpted and thickened, posterior surface with shallow tapering median furrow, shallower outer rim to sclerite. ?Tail sclerites (PI. 6, fig. 3) smaller and relatively broader than intermediate, mean length 2-3 mm (s.d. = 0-14, n = 6), mean length/width 0-55 (s.d. = 0 03, n = 5). Ovoid, smoothly rounded with elevated sub-central mucronate apex more anterior of maximum width; apex mean 040 (s.d. = 008, n = 6) of length from anterior, maximum width mean 0-54 (s.d. = 0-04, n = 6) of length from anterior. Vaulted convex anterior and posterior fields, without posterior elevation, lateral fields narrow, depressed (PI. 6, fig. 3c). Transverse profile low arched, broad, deeper laterally. Granular ornament, coarsening outwards, with line of coarser granules from apex across lateral fields (PI. 6, fig. 3a). Ventral surface with deep, slanting median cavity anterior of transverse flexure beneath apex, mean 0-33 (s.d. = 0 04, n = 6) of length from anterior. Ventral surface otherwise smooth, becoming thickened and sculpted particularly in subapical ad flexed region to give shallow median longitudinal furrow and anterolateral furrows flanking subapical cavity. Remarks. By comparison with intermediate sclerites, the ovoid, rounded ?tail sclerites are smaller, relatively broad, and have a gently convex longitudinal profile, both anterior and posterior fields being vaulted. The mucronate apex is a little more anterior, ventrally the subapical pit is similarly situated. Both have outward coarsening granular ornament, but in the ?tail sclerites the characteristic line of larger granules lies within the lateral fields rather than at their posterior limit. The ?tail sclerites are distinct mainly on size, shape, and the longitudinal profile lacking post-apical concavity to posterior elevation. On the material available, and because of the line of larger granules among the lateral ornamentation of both types of sclerite, and fairly similar form overall, the smaller sclerites are proposed as possible tail plates for this species. They might otherwise represent anterior intermediate plates, but are less likely to represent head plates because those typically differ more in morphology from the other plates in chitons. No head sclerites are identified for H. papilla. Two small end sclerites from Mollbos, described below (pp. 966, 968), may include the appropriate head sclerite for this species. enetoplax gen. nov. Derivation of name. From the Greek enete, brooch, and Latin plax, plate, to describe the form. Type species. E. decora gen. et sp. nov. from the Upper Wenlock (Silurian) of Gotland, Sweden. Diagnosis. Small, transversely elongate, ovoid to sub-triangular intermediate sclerites with elevated mucronate apex displaced anteriorly from centre; rounded margins. Strongly vaulted short and fairly narrow triangulate anterior field to pointed apex, elevated by low folds; depressed shallow lateral fields flattening outwards across weak transverse flexure, rounding into broad long triangulate posterior field, concave becoming flattened to convex outwards. Broadening behind apex to maximum width just posterior of mid-length. Dorsal concentric growth lines; granular ornament, coarsening outwards, coarser on anterior area, finer to coarse on lateral to posterior areas. Ventral surface with small deep median subapical cavity near anterior margin, slanted anteriorly. Ventral thickening leading to only weak sculpting of surface. ?Head sclerite round, with elevated mucronate apex close to anterior, vaulted short anterior field, gently convex and long posterolateral field, granular ornament; ventral surface concave, shallow subapical cavity. Remarks. Enetoplax gen. nov. is similar to Heloplax gen. nov. (above) in having small ovoid intermediate sclerites with a dorsal mucronate apex and holoperipheral growth. It differs in that sclerites are less vaulted and flexed, have the apex more anterior, and a correspondingly shorter and narrower anterior field, a longer, shallower and broader posterior field, and maximum width further CHERNS: SILURIAN POLYPLACOPHORA 961 displaced posteriorly behind the apex (Text-fig. 3). Ventrally the subapical cavity is more anterior, and transverse flexure is weak. Ventral thickening leads to limited surface sculpting around the subapical area, by contrast to the strong development of this surface in Heloplax. Granular ornament in both genera coarsens outwards, and is more developed on the anterior field which in Enetoplax is delimited by low radial folds. Heloplax has a distinct line of coarser granules radiating from the apex across the lateral fields. Both Enetoplax and Heloplax sclerites occur in samples from Mollbos-1, both together and separately. Although the intermediate sclerites are broadly similar, and different from all other chitons described, they are easily distinguished morphologically, and are separated here at generic level. Enetoplax decora gen. et sp. nov. Plates 7-8 Derivation of name. From the Latin decoris , adorned, to describe the ornamented plates. Material , locality and horizon. Mollbos-I, Gotland, Halla Formation, Upper Wenlock (Homerian); 40 isolated sclerites; holotype RM Mol59.999*, intermediate sclerite; intermediate sclerites RM Mol59.835* (?tail), 159.836*, 159.837-159.839, 159.840 (Ttail), 159.841-159.84, 159.850, 159.868, 159.885, 159.888, 159.890, 159.897, 159.913-159.916, 159.921, 159.923, 159.924 (?tail), 159.955-159.956, 159.960, 159.969, 159.974, 159.983, 160.000*, 160.001, 160.016, 160.025, 160.028-160.029, 160.033-160.035; ?head sclerite RM Mol59.998*. Diagnosis. As for the genus. Description. (Text-fig. 3). Small, transversely elongate, intermediate sclerites with rounded, convex margins; mean length 3-3 mm (s.d. = 0 5, /; = 38; holotype 3-3 mm), length/width 0-68 (s.d. = 0 07, n = 33; holotype 0-49). Elevated mucronate apex anterior of centre, mean 0-33 (s.d. = 0 06, /; = 38 ; holotype 0 36) of length from anterior; ovoid to sub-triangular, broadening behind apex to maximum width slightly posterior of midlength within lateral areas, mean 0-54 (s.d. = 0 06, n = 38; holotype 0-60) of length from anterior. Strongly vaulted, short triangulate anterior field to elevated pointed apex; transversely convex, fairly narrow, elevated by low radial folds above lateral fields, low median radial fold or smoothly rounded, convex (e.g. PI. 7, fig. la, c-d; PI. 8, fig. la, c). Weakly defined triangulate posterior field, long and expanding from apex, shallow becoming slightly elevated; gently convex to flattened transversely, broadening outwards, broader than anterior field. Lateral fields gently concave, depressed, flattening outwards, rounding into posterior field, more clearly bounded against low folds of elevated anterior field. In longitudinal profile, convex short anterior field to apex, concave becoming flattened to slightly convex along long posterior field (PI. 7, figs lc, 2c; PI. 8, figs lc, 2d, 3c). Transverse profile shallow, broad, deeper laterally (PI. 7, fig. Id). Growth lines, concentric about apex, more distinct on outer part of dorsal surface, increments forming low ridges (e.g. PI. 7, fig. la, c-d). Granular ornament coarsening outwards, quincunx pattern, coarser and more prominent across vaulted anterior area (e.g. PI. 8, fig. 1), finer to coarse laterally and posteriorly (e.g. PI. 7, figs la, 2c). Ventral surface smooth, with small, deep, round to ovoid median cavity with smooth margins near anterior margin, slanting towards anterior from beneath apex (e.g. PI. 7, fig. 2e); mean 0-22 (s.d. = 0 05, n = 38) of length from anterior. Ventral thickening leading to distinct shallow rim outside thickened surface (e.g. PI. 7, fig. 2b, e), a pair of very shallow longitudinal furrows flanking median pad behind cavity, lateral pads (e.g. PI. 7, fig. lb), but relatively little sculpting of surface. Gently convex transverse flexure behind cavity, but low curvature across ventral surface. ?Tail sclerites not clearly distinct, and hence not separated from the remainder of sclerites for biometrics, but possibly represented by three relatively flatter specimens with apex and, particularly, ventral cavity somewhat more anterior (e.g. PI. 8, fig. 2). Ovoid to sub-triangular, rounded, with elevated mucronate apex towards anterior, mean 0-27 (s.d. = 0 03, n = 3) of length from anterior; maximum width near mid-length, mean 0-51 (s.d. = 0-13, n = 3) of length from anterior. Very short, narrow, convex anterior field, elevated by low folds from lateral fields, low median radial fold; much longer, broader, concave to flattened posterior field, with slightly lobed or convex posterior margin (PI. 8, fig. 2a, c). Lateral fields shallowly depressed, concave to flattened. Ventral surface with anteriorly slanting, round to ovoid deep subapical pit close to anterior margin. 962 PALAEONTOLOGY, VOLUME 41 mean 0T8 (s.d. = 0 03, n = 3) of length from anterior (PI. 8, fig. 2b, d). Becoming thickened, distinct shallower outer rim, otherwise smooth. Dorsal coarse granular anterior ornament, concentric growth lines. ?Head sclerite (PI. 8, fig. 3) fairly poorly preserved, limiting biometric measurements. Small, round, with elevated mucronate apex close to anterior. Very short, convex anterior field to apex; long, gently convex posterolateral fields. Dorsal concentric growth lines, granular ornament (poorly preserved). Ventral surface smooth, concave, shallow subapical cavity. Remarks. Tail sclerites are only tentatively proposed for Enetoplax, differing from intermediate sclerites in being lower arched, with the apex, and particularly the subapical cavity, closer to the anterior margin, and the longer posterior field shallower and flatter. By comparison, possible Heloplax tail sclerites are notably smaller than intermediate sclerites, relatively broad and have convex, vaulted anterior and posterior fields and thus a convex longitudinal profile. The round head sclerite is associated with Enetoplax, rather than Heloplax , because of the apex near to the anterior, and long posterolateral field. arctoplax gen. nov. Derivation of name. From the Latin arcto, compress, and Greek plax. plate, to describe the pinched form of the sclerites. Type species. A. ornata gen. et sp. nov. from the Upper Wenlock (Silurian) of Gotland, Sweden. Diagnosis. Small, high arched, elongate spatulate intermediate sclerites with strong constriction to sub-central mucronate apex, triangulate shell fields defined by low radial folds and furrows. Anterior and posterior shell fields vaulted, convex, lateral fields depressed; lateral and posterior fields with triangulate, angled faces; anterior field with weaker triangulate facies, rounded medially. Jugal area flat posteriorly, rounded and downward sloping anteriorly; side slopes deep, steep, tapering anteriorly, height greatest posteriorly; maximum width at anterolateral corners, but tapering little along lateral margins; anterior margin transverse to convex, rounding anterolaterally, posterior margin transverse. Ventral surface smooth; deeply concave transversely, narrowing through apical constriction, small deep narrow subapical cavity. Anterior profile rounded, low arched, posterior profile deeper, high arched, with angular corners between faces. Fine granular dorsal ornament, concentric growth lines. Remarks. The vaulted spatulate form with sub-central mucronate apex is comparable in its holoperipheral growth, and hence lack of a ventral apical area, to Heloplax gen. nov. and Enetoplax EXPLANATION OF PLATE 6 Figs 1-3. Heloplax papilla gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). 1, RM Mo 159.828; well thickened intermediate sclerite. la, dorsal view, note line of coarser granules, concentric growth lines; lb, ventral view, showing sculpted thickened ventral surface: deep, anteriorly slanting subapical cavity with anterior and posterior thickened pads, flanking smaller lateral cavities at base of anteriorly directed, expanding oblique depressions, transverse flexure of sclerite, median longitudinal furrow across posterior field, lc, oblique ventral view showing posterior, tapering extension of furrows across lateral cavities, tapering bifurcation of pad behind anteriorly slanting subapical cavity. Id, right lateral view, showing posterior elevation behind mucronate apex, line of coarser granules across lateral field, holoperipheral granular ornament. 2, RM Mol59.829; intermediate sclerite. 2a, dorsal view; note outward coarsening granular ornament, growth lines; 2b, ventral view, showing deep subapical cavity; 2c, left lateral view. 3, RM Mol59.851; ?tail sclerite. 3a, dorsal view; note line of coarser granules across depressed lateral fields, mucronate sub-central apex; 3b, ventral view, showing transverse flexure behind slanting subapical cavity, median longitudinal furrows across posterior field, flanked by thickened pads; 3c, left lateral view; note lack of elevation of posterior field, holoperipheral growth lines. All x 15. PLATE 6 1 ‘ > V ; ■ A ^SsBKKSbS^ tfefl f ; \.V;f*«-v.; ®r :. 1 ll|l| :■;,- 1 1 - S I ,■■ S iff 1 :-V'^r;-K lllli3i ■* Tl5 v :-; r+r ^ f„v A'sA^Sl ■ CHERNS, Heloplax 964 PALAEONTOLOGY, VOLUME 41 gen. nov., and different from other chitons except for some tail sclerites. At Mollbos-1, the sclerites occur not only singly but as several within samples, including some plates of variable size with similar patterns of larger growth increments which probably belonged to the same individual. This would support an interpretation as intermediate plates, and not tail plates of another chiton. Although the ventral subapical cavity in Arctoplax compares with similar features in the small ovoid intermediate sclerites of Heloplax and Enetoplax , the form of Arctoplax sclerites is clearly distinct. The sub-central apex and highly unusual form mean that the anterior-posterior orientation is somewhat equivocal. The normal imbrication of chiton plates produces some overlap of the posterior apical area across the anterior edge of the following plate. In Arctoplax , the plates lack an apical area and their form precludes overlap; the deeper, high arched end is interpreted here provisionally as posterior (e.g. Text-fig. 7). One problematical Ordovician genus Llandeilochiton Bergenhayn, 1955, based on a single small specimen from the Llandeilo of southern Scotland (L. ashbyi Bergenhayn, 1955), has a rectangular, flexed form with a sub-central apex, and apparently folds delimiting shell areas, but with a marked jugal furrow (Bergenhayn 1955, pi. 2, fig. 12). The specimen has not been examined, and its chiton affinities have been questioned (Smith and Hoare 1987). Arctoplax ornata gen. et sp. nov. Plate 9 Derivation of name. From the Latin orno, decorate, referring to the fine granular ornament. Material, locality and horizon. Mollbos-1, Gotland, Halla Formation, Upper Wenlock (Homerian); 15 isolated intermediate sclerites (including one anterior and three posterior fragments); holotype RM Mo 159.856*, intermediate sclerite;RM Mol59.853-159.855, 159.857, 159.899*, 159.904, 159.911, 159.925, 159.944, 159.948, 159.986, 159.996*, 160.002, 160.030. Diagnosis. As for the genus. Description. Small and high vaulted, elongate spatulate sclerites, mean length 7-0 mm (s.d. = 0-72, n = 11; holotype 7-7 mm), mean length/width 1 -54 (s.d. = 0-19, n = 9; holotype 1-35). Strong constriction to a sub- central mucronate apex, mean 0-51 (s.d. = 0 04, n = 11; holotype 0-58) of length from anterior, low angular to rounded radial folds and furrows from apex defining shell fields. Anterior and posterior fields broad, vaulted, convex, with anterior field shallowing from apex, posterior field deepening; steep lateral fields depressed. Lateral and posterior fields folded into triangulate, angled faces, anterior field with weaker, triangulate faces anterolaterally, becoming rounded. Low radial folds elevating anterior and posterior fields, posterior jugal field, central part of lateral fields. Maximum width at anterolateral corners but tapering little along lateral margins; maximum height at posterolateral corners; mean height/length 0-51 (s.d. = 008, n = 11; holotype 048). Anterior field moderately arched, sloping outward, anterior margin rounded, more transverse medially, convex anterolateral corners through steeper radial fold and flexure (PI. 9, figs le, 2e). Long, slightly concave to convex, lateral margins, deepening posteriorly; lateral fields sloping steeply outwards below constricted apex, two triangulate faces flanking central low fold (PI. 9, figs c). Squarish posterolateral EXPLANATION OF PLATE 7 Figs 1-2. Enetoplax decora gen. et sp. nov.; Mollbos-1, Gotland; Halla Formation, Upper Wenlock (Homerian); intermediate sclerites. 1, RM Mol59.999, holotype. la, dorsal view; note anteriorly displaced mucronate apex, prominent growth increments, holoperipheral growth; lb, oblique ventral view; note anteriorly slanting subapical cavity near anterior margin, thickened medial pad flanked by longitudinal furrows; lc, right lateral view; lateroposterior shell field flat; Id, anterior view showing mucronate apex, prominent growth increments. 2, RM Mo 160.000. 2a, dorsal view; 2b, ventral view; note thickened surface inside marginal rim; 2c, left lateral view; 2d, detail of granular ornament from right hand side, x 50; 2e, details of subapical cavity; note smooth, anteriorly slanting walls, cavity within thickened surface, inside marginal rim, x 50. All except 2d-e x 15. PLATE 7 CHERNS, Enetoplax 966 PALAEONTOLOGY, VOLUME 41 corners through fold and acute flexure, posterior margin straight, transverse; posterior field strongly vaulted across elevated flat triangulate jugal field, flanked either side by two steep, flat triangulate faces angled outwards and downwards respectively (PI. 9, figs, Id, 2d). Fine and even, granular dorsal ornament, evident only on some sclerites, coarsening and best preserved towards outer parts of shell across anterior and lateral areas (PI. 9, figs la, c, f, 3a, c-d). Growth lines concentric about apex, larger growth increments more evident towards outer part of dorsal surface. Ventral surface smooth, concave and strongly folded around a longitudinal axis, deepest posteriorly in triangulate jugal area, apical constriction flanked by depressed side slopes, shallowing anteriorly. Small, deep narrow subapical cavity. All shell areas expanding away from apical constriction, slight corrugation of surface reflecting radial folds. Remarks. Granular ornament is well preserved on a few sclerites (holotype, PL 9, fig. 1, and PI. 9, fig. 3), but not evident on several other, possibly more thickened, specimens (PI. 9, fig. 2). Those ornamented sclerites are relatively broad and shallow, with concave lateral margins, while the more thickened sclerites are narrower and deeper, slightly convex laterally. However, differences in ornament may be preservational, and on the basis of the fairly limited material available, variation is regarded here as intraspecific. Tail sclerites have not yet been recognized for A. ornata , although the head sclerite may be among the two described below. head B indet. Text-figure 5a-e Material , locality and horizon. Mollbos-l, Gotland, Halla Formation, Upper Wenlock (Homerian); one isolated sclerite, RM Mol59.825*. Description. Small, vaulted, convex semicircular sclerite (with some damage along the posterior edge) with elevated sub-central mucronate apex. Short, fairly broad and high; length 14 mm, length/width 0-58, height /length 10. Flat, elevated triangulate face within posterior area, steep, outward sloping, broad convex anterolateral area to rounded semicircular anterolateral margin, shallowing slightly near margin. Slight depressed radial flexure to posterolateral corners, rounding strongly into transverse, arched posterior margin (Text-fig. 5b), posterior shell arched and elevated (Text-fig. 5d), posterior margin transverse. Fine granular dorsal ornament; concentric growth lines more distinct near margins, where a new growth increment secreted from beneath is particularly evident posteriorly (Text-fig. 5e). Ventral surface deeply concave, smooth, deepest subapically and across median posterior face. Remarks. The small, semicircular and convex form is typical of chiton end plates, and the transverse arched margin and steepness of the outward sloping face suggest that this is a head sclerite. The elevated sub-central mucronate apex, flat triangulate medial posterior face, arched and elevated posterior, and fine granular ornament across the whole shell, are features which associate it most closely with A. ornata (Text-fig. 7). However, the size, apical morphology and ornament are also comparable to Heloplax , but somewhat less so to Enetoplax for which a head sclerite is already tentatively recognized. All three genera are found in other samples from the same horizon. EXPLANATION OF PLATE 8 Figs 1-3. Enetoplax decora gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). 1, RM Mol59.936; intermediate sclerite. la, dorsal view; note coarser granular ornament on anterior shell held ; 1 b, ventral view ; lc, right lateral view, posterolateral shell field gently concave, becoming elevated. 2, RM Mo 159.835; ?tail sclerite. 2a, dorsal view; note trilobed, scalloped posterior margin; 2b, ventral view; subapical cavity close to anterior, thickened surface inside marginal rim; 2c, oblique ventral view, showing thickened surface inside marginal rim, medial pad inside weak longitudinal furrows. 3, RM Mo 159.998; ?head sclerite. 3a, dorsal view; note mucronate apex close to anterior margin; 3b, ventral view; weak subapical cavity, surface with attached grains; 3c, left lateral view; posterolateral shell field gently convex; 3d, anterior view, showing mucronate apex, convex transverse profile. All x 15. PLATE 8 CHERNS, Enetoplax 968 PALAEONTOLOGY, VOLUME 41 ?head or tail C indet. Text-figure 5f-i Material , locality and horizon. Mollbos-l, Gotland, Halla Formation, Upper Wenlock (Homerian); one isolated sclerite, RM Mol59.824. Description. Small, transversely ovoid sclerite, convex becoming arched posteriorly, with dorsal shallow, rounded longitudinal furrows and ribs radiating posteriorly from a weakly elevated ?dorsal apex close to anterior; length 2-5 mm, length/width 0 86, height/length 0 88. Anterior area very short, arched, semicircular margin with distinct rim; strongly rounded posterolateral corners, posterior margin also semicircular, more arched (Text-figs 5f, 5h). Ventral surface smooth, concave, deepest centrally, shallowing outwards (Text-fig. 5g). Concentric growth, ?some weak granular ornament; thickened rim of anterolateral margin showing several growth increments. Remarks. The fairly poorly preserved sclerite is broken along one edge and across the ventral surface, removing part of a later shell increment and hence the part of the dorsal margin. An apparently continuous marginal rim on an earlier shell layer suggests concentric growth around a dorsal apex, with additional shell increments added as complete layers from the ventral side, leading to significant thickening. The small size, arched convex form and semicircular margins suggest that this is an end plate, but anterior and posterior orientation are somewhat equivocal, only partly dependent upon which end plate it represents. A weak elevation near the unbroken margin apparently represents the apex, from which shallow rounded ribs and furrows, not well preserved, radiate slightly. Positioned with the apex uppermost, both margins remain moderately arched (Text-fig. 5h-i), yet if either margin is positioned flat the other margin becomes steep and raised. Radial ornament from the mucro is fairly common in both head and tail plates of neoloricate chitons (e.g. Smith 1960). The holoperipheral growth and ovoid shape of this small sclerite are comparable to Heloplax and Enetoplax, although the ribbed ornament is different. The mode of growth is also similar to the larger, spatulate Arctoplax, but different from all other Gotland chitons. MUSCULATURE IN GOTLAND CHITONS The musculature of Chelodes actinis was discussed from sculpting of the ventral surface in thickened sclerites (Cherns 1998). In living chitons, the complex musculature (e.g. Hyman 1967) between valves and from valves into the body wall leaves no evident insertion sites in the inner shell layer, the ventral hypostracum. The function of the various sets of muscles acting on the valves lies in drawing the plates together and holding them against the body. Sutural laminae and insertion plates, formed of the middle shell layer, the articulamentum, which is absent from paleoloricate chitons, provide physical articulation of plates and attachment to the mantle. Plates are commonly embedded in and partially covered by the mantle. EXPLANATION OF PLATE 9 Figs 1-3. Arctoplax ornata gen. et sp. nov.; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian); intermediate sclerites. 1, RM Mol59.856, holotype. la, dorsal view; note sub-central mucronate apex, radial folds within lateral and posterior shell fields, prominent growth increments; x 7. lb, ventral view, showing narrow subapical cavity; x 7. lc, left lateral view; x 7. Id, posterior view; x 7. le, anterior view; x 7. If, detail of dorsal granular ornament; x 15. 2, RM Mo 159.899. 2a, dorsal view; 2b, ventral view; 2c right lateral view, note prominent growth increments; 2d, posterior view; 2e, anterior view. All x 7. 3, RM Mo 159.996; broken left and right posterolateral margins. 3a, dorsal view; x 7. 3b, ventral view; note narrow subapical cavity; x 7. 3c, right lateral view; x 7. 3d, detail of granular ornament; x 15. PLATE 9 CHERNS, Arctoplax 970 PALAEONTOLOGY, VOLUME 41 text-fig. 5. a— E, head B indet.; RM Mo 159.825; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). a, dorsal view; note sub-central mucronate apex, holoperipheral growth, b, left lateral view, c, oblique anterior view, showing elevated triangular posterior field. D, posterior ventral view, e, posterior view. All x 20. f— i, ?head or tail C indet.; RM Mol59.826; Mollbos-l, Gotland; Halla Formation, Upper Wenlock (Homerian). F, dorsal view, showing radial ridged ornament, thickened posterior margin with growth increment. G, ventral view; note broken growth increment beneath outer shell, h, right lateral view, showing prominent growth increment at posterior, broken anteriorly. I, posterior view. All x 15. The elongate shield-shaped, flexed intermediate sclerites of Thairoplax pelta sp. nov. lack localized ventral thickening, but T.l aff. pelta ? has a low triangulate transverse thickened ridge in front of, and extending medially towards, the apical area (PI. 3, fig. 4b). Intermediate sclerites of Plectrochiton tegulus sp. nov. also show a slight thickening across the equivalent area, although closer to the anterior of these short triangulate sclerites (PI. 3, fig. lb). The small, winged intermediate and tail sclerites of Alastega lira sp. nov. have well-developed ventral triangulate transverse ridges. In all the genera, the apical area typically has a slightly raised rim above the smooth ventral surface. Other paleoloricate chitons which have thickened transverse ventral ridges include Kindbladochiton and the Ordovician Ivoechiton spp. (Smith and Toomey 1964). The physical effect of these ridges, which taper outwards, is to elevate slightly the posterior shell, the apical area CHERNS: SILURIAN POLYPLACOPHOR A 971 B text-fig. 6. Reconstructions of A, Enetoplax and b, Heloplax as chitons, shown in left lateral view with normal plate orientations. Sclerites did not overlap but may have become imbricated through muscular rotation to elevate further the posterior shell fields. Scale bars represent 1 mm. extending above the following sclerite. The ridge would also provide muscle attachment sites. C. actinis sclerites lack a similar marked thickening in front of the apical area, although developing shallow furrows flanking a medial pad towards the apical area similar to that of T.l aff. peltal (Cherns 1998, pi. 4, fig. 3b). These furrows apparently represent muscle insertion sites, additional to the apical area rim that might serve for muscle attachment. None of the Gotland chitons has marked cavities extending beneath the apical area, as described for the Ordovician Chelodes whitehousei and Cambrian Matthevia spp. (Runnegar et al. 1979). The three holoperipheral genera have mucronate sclerites with a prominent subapical cavity presumed to function for muscle insertion. In both Heloplax and Enetoplax (Text-figs 2-3), these small, deep cavities are anteriorly slanting, in front of a transverse flexure of the shell; in the former, the subapical cavity becomes flanked in some sclerites by additional, shallower, anterolaterally directed cavities (PI. 5, fig. lb; PI. 6, fig. lb). The subapical cavities are circular to transversely ovoid, with smooth margins, tapering into the valve and fairly deep, less even at the basal insertion point (PI. 7, fig. 2d). Although the main insertion directions for the ventral cavities appear to be anterior and anterolateral, shallow furrows across the posterior field suggest that each also housed posteriorly directed muscles (PI. 6, fig. lb— c; PI. 7, fig. lb). In Heloplax , the posterior field of intermediate sclerites becomes elevated; in Enetoplax , a longer field curves more shallowly (Text- fig. 6). Thus the shell flexure serves a similar function to the thickened transverse ridges described above, for other genera; although the sclerites did not overlap, they could probably be drawn down on to the body, drawn closer, rotated and imbricated (cf. Matthevia reconstruction in Runnegar et al. 1979, text-fig. 1). The tail plates for both Heloplax and Enetoplax lack posterior elevation, the posterior fields being convex to flat (Text-fig. 6). Arctoplax sclerites have a deep, tapering, rather slit-like sub-central subapical cavity (PI. 9, fig. lb), but in addition their radial angled folds and furrows across the high-arched, elongate shell might 972 PALAEONTOLOGY, VOLUME 41 text-fig. 7. Reconstruction of Arctoplax as a chiton in slightly oblique, left lateral view, with the rounded shallower end of sclerites anterior, the squarer and deeper shell field posterior. The head sclerite has been suggested as head B indet. There is no overlap of plates, but muscular rotation of plates may have produced imbrication, elevating the posterior field. Scale bar represents 1 mm. have provided additional attachment surfaces or muscle tracks into the body wall. The subapical cavity may thus have anchored longitudinal and radial muscles, whereas those in Heloplax and Enetoplax apparently housed primarily anteriorly and anterolaterally directed muscle blocks. If the function of angled radial faces was connected primarily with the musculature, then by comparison their stronger development at the deeper end, interpreted here tentatively as posterior, might indicate the reverse orientation as more appropriate. However, the deep, long side slopes and marked apical constriction produced lateral cover for a narrow elongate body, where the incongruent, arched anterior and posterior profiles precluded overlap of sclerites. With the more rounded, shallower end as anterior (Text-fig. 7), some rotation of the shell behind the subapical muscle insertion site, to imbricate sclerites as proposed for the other genera, might elevate slightly the deeper, more arched and angular end. Runnegar et al. (1979) interpreted two deep ventral pits in tall conical valves of the late Cambrian Matthevia variabilis Walcott, 1885 as muscle insertion sites for dorso-ventral muscles from shell to foot. A comparable single pit was present in the younger late Cambrian M. xvalcotti Runnegar, Pojeta, Taylor and Collins, 1979, and a shallow pit in the early Ordovician Hemithecella expansa Ulrich and Bridge, 1941. A concavity beneath the apical area in some early Ordovician C. white- housei Runnegar, Pojeta, Taylor and Collins, 1979 was interpreted as equivalent, and Matthevia as a primitive chiton. By contrast, Stinchcomb and Darrough (1995), in discussing Cambrian- Ordovician Hemithecella , argued that ventral pits were not chiton features and that Matthevia , Hemithecella and some C. whitehousei should be considered as multi-plated problematical molluscs distinct from chitons. THE NATURE OF THE SCLERITOME The majority of the Gotland chitons described above represent isolated sclerites from silicified samples of late Wenlock age. Recent, neoloricate chitons typically have eight plates, of which seven are secreted in the late trochophore stage of ontogeny, the eighth, tail valve somewhat later (Okuda 1947; Hyman 1967). Limited records of articulated fossil chitons indicate a similar valve complement also in older neoloricate and in paleoloricate chitons (e.g. Rolfe 1981; Hoare and Mapes 1989). The only articulated plates among the Gotland material are the type specimen of G. interplicatus , which shows a fragment of a second, following sclerite (PI. 1, fig. 1). The six similar intermediate sclerites in chitons vary in size along the body, but the head particularly, and also tail plates, can be very different (e.g. Smith 1960). The plates have posterior apices and mixoperipheral growth, with the outer tegmentum extending ventrally to form the apical area, except for some tail plates which have a dorsal mucronate apex and holoperipheral growth. In Recent chitons, the apical area is usually a narrow band, but in some early chitons may be much longer (e.g. Chelodes raaschi\ Kluessendorf 1987). Fine growth lines are preserved on most of the Gotland chiton sclerites, but also larger, stepped growth increments where similar patterns of growth can be used to associate plates from the same individual as well as indicating intra-individual variation (e.g. C. act inis ; Cherns 1998, pi. 6; PI. 7, figs 1-2). Such occurrences for all three of the new genera described here which show holoperipheral CHERNS: SILURIAN POLYPLACOPHORA 973 growth are important in the interpretation of these plates as intermediate and not tail sclerites. These and the more typical plates with mixoperipheral growth have highly comparable growth patterns. Recent chitons mostly range in size up to c. 0-05 m, but can be much larger ( Cryptochiton stelleri , 0-33 m; Hyman 1967), and usually have a flattened, ovoid foot. Overlap of the physically interlocking plates, relating to the length of the apical area, is generally quite small, and plates are normally broad and short. The Gotland chiton assemblage includes several with elongate rather than broad plates. Bergenhayn (1955) used formulae to estimate the length and elongation of Gotland Chelodes and Gotlandochiton , based on sclerite length, width, and length of apical area, reconstructing large sclerites with relatively long apical areas in Chelodes spp. as belonging to very elongate animals, G. interplicatus with its broad sclerite rather less so at about three times as long as wide (Bergenhayn 1955, pi. 2). Of the chitons described here, T. pelta has long sclerites with a short apical area, little overlap between plates, and hence must have been an elongate animal, and Aretoplax without overlap of plates was long but also narrow (Text-fig. 7). The remaining Gotland genera - Plectrochiton , Alastega , Enetoplax and Heloplax - had smaller, short and broad sclerites, more similar in shape to Recent chitons, the last two of these genera with no overlap of plates (Text- fig. 6). The stratigraphical evidence suggests that the late Wenlock sequences for the Gotland chiton localities represent very nearshore facies, associated with rocky shorelines (Cherns 1996). Those facies relationships indicate an ecology similar to that of most Recent chitons. Acknowledgements. I thank Dr L. Jeppsson (Lund University) for providing the silicified collections of Gotland chitons, and for his kind hospitality, also Dr L. Liljedahl for providing some of those silicified samples; Professor M. G. Bassett (National Museum of Wales, Cardiff) for discussion and critical review of the manuscript, and for providing darkroom facilities at NMW ; and Mrs G. Evans and Mrs L. Norton (NMW) for drafting Text-figures 2-3 and 6-7. Professor R. D. Hoare (Bowling Green State University, Ohio, USA) kindly read and criticized the manuscript. I am grateful to the Palaeozoology Section, Naturhistoriska Riksmuseum, Stockholm for loan of museum specimens. REFERENCES bergenhayn, J. R. M. 1943. Preliminary notes on fossil Polyplacophoras from Sweden. Geologiska Foreningens i Stockholms Forhandlingar , 65, 297-303. — 1955. Die fossilen Schwedischen Loricaten nebst einer vorlaufigen Revision des Systems der ganzen Klasse Loricata. Lunds Universitets Arsskrift, Nya Forhandlingar , Avdelningen 2, 51 (8), 1 — 46, pis 1-2. 1960. Cambrian and Ordovician loricates from North America. Journal of Paleontology , 34, 168-178. blainville, H. M. D. de 1816. Prodrome d’une nouvelle distribution systematique du regne animal. Bulletin des Sciences, Societe Philomathique de Paris , 105-124. cherns, l. 1996. Silurian chitons as indicators of rocky shores? - new data from Gotland. 41 . In Johnson, m. e. and brett, c. e. (eds). The James Hall Symposium: Second International Symposium on the Silurian System, Program and Abstracts. University of Rochester, N.Y., 1 14 pp. — 1998. Chelodes and closely related Polyplacophora (Mollusca) from the Silurian of Gotland, Sweden. Palaeontology, 41, 545-573. davidson, t. and king, w. 1874. On the Trimerellidae, a Palaeozoic family of the palliobranchs or Brachiopoda. Quarterly Journal of the Geological Society , London, 30, 124-172, pis 12-19. debrock, m. d., hoare, r. d. and mapes, r. h. 1984. Pennsylvanian (Desmoinesian) Polyplacophora (Mollusca) from Texas. Journal of Paleontology , 58, 1 1 17-1135. hoare, r. d. 1989. Mississippian polyplacophoran (Mollusca) from Utah. Journal of Paleontology, 63, 252. — and mapes, R. H. 1989. Articulated specimen of Acutichiton allynsmithi (Mollusca, Polyplacophora) from Oklahoma. Journal of Paleontology, 63, 251. hyman, l. h. 1967. The Invertebrata : Mollusca 1. Volume 6. McGraw-Hill, New York, vii + 792 pp. jaanusson, v. 1986. Locality designations in old collections from the Silurian of Gotland. Swedish Museum of Natural History, Stockholm 19 pp. kluessendorf, j. 1987. First report of Polyplacophora (Mollusca) from the Silurian of North America. Canadian Journal of Earth Sciences, 24, 435-441. 974 PALAEONTOLOGY, VOLUME 41 laufeld, s. 1974. Reference localities for palaeontology and geology in the Silurian of Gotland. Sveriges Geologiska Undersokning, Series C, 705, 1-172. and jeppsson, l. 1976. Silicification and bentonites in the Silurian of Gotland. Geologiska Foreningens i Stockholms Forhandlingar , 97, 207-222. liljedahl, l. 1984. Silurian silicified bivalves from Gotland. Sveriges Geologiska Undersokning , Series C, 804, 1-82. lindstrom, G. 1884. On the Silurian Gastropoda and Pteropoda of Gotland. Kongliga Svenska Vetenskaps- Adakamiens Handlingar , 19(6), 1-250, pis 1—21. okuda, s. 1947. Notes on the post-larval development of the giant chiton. Cryptochiton stelleri (Middendorff). Journal of Faculty of Science Hokkaido University , Series 6, Zoology , 9, 261-215. rolfe, w. d. i. 1981. Septemchiton — a misnomer. Journal of Paleontology, 55, 675-678. runnegar, b., pojeta, j. Jr, taylor, M. E. and collins, D. 1979. New species of the Cambrian and Ordovician chitons Matthevia and Chelodes from Wisconsin and Queensland: evidence for the early history of polyplacophoran mollusks. Journal of Paleontology, 53, 1374-1394. schmitt, J. G. and boyd, d. w. 1981. Patterns of silicification in Permian pelecypods and brachiopods from Wyoming. Journal of Sedimentary Petrology, 51, 1297-1308. sirenko, v. I. and starobogatov, ya. i. 1977. On the systematics of Paleozoic and Mesozoic chitons. Paleontological Journal, 1977(3), 285-294. [K sistematike paleozoyskikh i mesozoyskikh khitonov, Paleontologischeskii Zhurnal, 1977(3), 30-41]. smith, A. G. 1960. Amphineura. 14 1-176. In moore, r. g. (ed.). Treatise on invertebrate paleontology. Volume I. Mollusca 1. Geological Society of America and University of Kansas Press, Lawrence, Kansas, xxiii + 351 pp. — and hoare, R. D. 1987. Paleozoic Polyplacophora: a checklist and bibliography. Occasional Papers of the California Academy of Sciences, 146, 1-71. — and toomey, D. F. 1964. Chitons from the Kindblade Formation (Lower Ordovician), Arbuckle Mountains, Southern Oklahoma. Circular of the Oklahoma Geological Survey, 66, 1^41, pis 1-8. stinchcomb, b. l. and darrough, g. 1995. Some molluscan Problematica from the Upper Cambrian-Lower Ordovician of the Ozark Uplift. Journal of Paleontology , 69, 52-65. ulrich, E. o. and bridge, j. 1941. Hemithecella expansa. 19-20; pi. 68, fig. 6. In butts, c. Geology of the Appalachian Valley in Virginia, Part 2. Bulletin of the Virginia Geological Survey, 52, 1-271. van belle, r. a. 1975. Sur la classification des Polyplacophora: 1. Introduction et classification des Paleoloricata, avec la description de Kindbladochiton nom. nov. (pour Eochiton Smith, 1964). Informations de la Societe Beige de Malacologie, Serie 4, 5, 121-131, pi. 1. walcott, c. d. 1885. Notes on some Paleozoic pteropods. American Journal of Science , 30, 17-21. LESLEY CHERNS Department of Earth Sciences Cardiff University Box 914, Cardiff CF1 3YE, UK Typescript received 6 June 1997 Revised typescript received 10 November 1997 NEW SILURIAN NEOTAXODONT BIVALVES FROM SOUTH WALES AND THEIR PHYLOGENETIC SIGNIFICANCE by V. ALEXANDER RATTER and JOHN C. W. COPE Abstract. The arcoidean bivalves, Trecanolia acincta gen. et sp. nov. and Uskardita mikraulax gen. et sp. nov., are described from the Wenlock of South Wales. These bivalves are accommodated within the new family Frejidae, alongside the closely related Silurian genera Freja Liljedahl and Alytodonta Cope. The frejids are characterized by an amphidetic, chevron-shaped duplivincular ligament, and a ventrally diverging dental arrangement of pseudocardinals and pseudolaterals. The group provides further evidence that the superfamily Arcoidea evolved from an early Ordovician ancestor, such as Catamarcaia Sanchez and Babin. The frejids represent an early diversification of the arcoideans, previously unknown in the Palaeozoic. The subclass Neotaxodonta was proposed by Korobkov (1954) to distinguish the arcoid and limopsoid bivalves from the nuculoids. The two groups share a superficially similar taxodont dentition that had long caused them to be taxonomically linked, following an initial proposal by Douville (1912). The subclass Palaeotaxodonta, proposed by Korobkov (1954) for the nuculoids, was accepted in the bivalve volumes of the Treatise (Cox et al. 1969-71). However, the Neotaxodonta - which removed the superfamilies Arcoidea and Limopsoidea from the pterio- morphian bivalves -was not. Taylor et al. (1969, 1973) argued that the shell microstructure of the superfamilies Arcoidea and Limopsoidea provided good grounds for separating them from other pteriomorphians, and Cope (1995) proposed that the Neotaxodonta be recognized as distinct from the Pteriomorphia. As removal of these two superfamilies from the Pteriomorphia, as a restricted order Arcoida, left many former arcoids within the Pteriomorphia without ordinal status. Cope (1996) assigned these to the order Cyrtodontida. The neotaxodonts have a well established Mesozoic to Recent history, but their Palaeozoic origin and subsequent diversification is contentious. Many previous phylogenetic schemes considered the cyrtodontid pteriomorphians as the probable ancestors of the arcoid neotaxodonts (e.g. Newell 1954, 1965; Cox 1960; Vogel 1962; Cox et al. 1969-71 ; Pojeta 1971, 1978; Waller 1978, 1990). This view was reinforced by the known fossil record, as rich Ordovician faunas of cyrtodonts were known, but the earliest undisputed arcoids were of Devonian age. Possible earlier forms included two poorly known Ordovician species referred to Parallelodon. Babin (1966) showed that P. antiquus Barrois, 1891 was based on a poorly preserved specimen and that Barrois’ figures were highly interpretative. Furthermore, Glyptarca primaeva Hicks, 1873 (included in the Treatise within Parallelodon) was described by Carter (1971) and Cope (1996) as a palaeoheterodont. The earliest undoubted neotaxodont form included in the Treatise was therefore the Devonian Parallelodon. Recently this view has been challenged (Cope 1997a, 19976). Sanchez and Babin (1993) described the genus Catamarcaia from the Middle Arenig of Argentina (see Text-fig. 6). This genus combines continuous dentition (i.e. lacking an edentulous area on the hinge-plate) with a duplivincular ligament; they considered it as a pteriomorphian with palaeo- heterodont affinities. However, Cope (1997a) concluded that the continuous dentition of Catamarcaia was not characteristic of Ordovician pteriomorphs, but was more typical of neotaxodonts. Furthermore, the dental arrangement of Catamarcaia is similar to that of the palaeo- IPalaeontology, Vol. 41, Part 5, 1998, pp. 975-991, 1 pi.) © The Palaeontological Association 976 PALAEONTOLOGY, VOLUME 41 heterodont Glyptarca. This suggests that the neotaxodonts could have evolved directly from the palaeoheterodonts. Cope (1997 6) carried these arguments further and figured a Silurian neotaxodont, Alytodonta gibbosa , showing characters intermediate between Catamarcaia and the Wenlock genus Freja Liljedahl, 1984. Herein we describe two new genera of neotaxodonts from the Wenlock Series of South Wales and assess their phylogenetic implications. SYSTEMATIC PALAEONTOLOGY All linear measurements are given in millimetres (mm), measured with Vernier callipers under a binocular microscope. Abbreviations used for dimensions, angles and statistics are: a° = angle of obliquity, AL = anterior length, CH = cardinal area height, CL = cardinal area length, L = length, H = height, OL = oblique length, S.D. = standard deviation, U° = umbonal angle, UH = umbonal height, and W = width. The following abbreviations are used to describe preservation: I = internal, E = external, R = right, L = left, A = articulated, C = complete valve, P = slightly fragmented valve, F = fragment, X = recrystallized calcite shell; the letters are combined to describe the preservation of each specimen (e.g. an ILC is a complete internal mould of a left valve). The specimens are housed in the collections of the British Geological Survey, Keyworth (BGS). Subclass neotaxodonta Korobkov, 1954 Order arcoida Stoliczka, 1871 Superfamily arcoidea Lamarck, 1809 Family frejidae fam. nov. Type genus. Freja Liljedahl, 1984 (p. 36), here designated. Diagnosis. Equivalved, subequilateral to inequilateral, moderately inflated arcoids, with a sub- orbicular to ovoid outline; ligament external, duplivincular, chevron-shaped and positioned on a cardinal area; hinge line straight ; dentition of simple pseudocardinal teeth on the central hinge area, bounded anteriorly and posteriorly by pseudolaterals, elongated obliquely to hinge line; adductor muscle scars anisomyarian; external prosopon of concentric growth lines only. Remarks. This new family is established to accommodate Silurian representatives of the superfamily Arcoidea. These include the genera Freja Liljedahl, 1984, Alytodonta Cope, 19976, Trecanolia gen. nov. and Uskardita gen. nov. (Text-fig. 1). The frejids are characterized by a sub-circular to ovate outline and a straight hinge line. The dentition consists of a series of small, simple pseudocardinal teeth along the central area of the hinge plate, which are terminated by anterior and posterior pseudolateral teeth that are elongated at an oblique angle to the hinge line. The earlier genera, namely Alytodonta and Uskardita , have much shorter anterior pseudolaterals that are more oblique to the hinge line than later representatives. Trecanolia has more modified pseudolateral dentition; the undersides of the ventral teeth have cardinal denticles developing at irregular intervals. All four genera have a chevron-shaped duplivincular ligament between the hinge plate and the beak. Liljedahl (1984, pp. 36—37) was unsure of the systematic position of Freja at the superfamily and family level. At that time, the newly described genus had a unique hinge construction and gross shell morphology. The prosocline outline, straight hinge line and anisomyarian adductors were typical in the cyrtodontoideans, yet the hinge plate and ligament closely resembled the Mesozoic and Cenozoic arcids. With the recent discovery of closely related genera from the Llandovery and Wenlock of Britain, it is possible to view the morphology of Freja as typical of this family of Silurian arcoids and it is herein considered as the type genus. Cope (19976, pp. 740-741) erected the new genus Alytodonta , which was based on a single specimen from the Lower Llandovery at Girvan, Ayrshire. Alytodonta displays the duplivincular RATTER AND COPE: SILURIAN NEOTAXODONT BIVALVES 977 text-fig. 1. Hinge details of the four frejid genera. A, Alytodonta gibbosa Cope, 19976; Natural History Museum L 49858; modified from Cope 19976, text-fig. 4. B, Uskardita mikraulax gen. et sp. nov.; BGS GSM 22187. c, Freja fecunda Liljedahl, 1984; Geological Survey of Sweden Type 3367 ; modified from Liljedahl 1984, fig. 21. d, Trecanolia acincta gen. et sp. nov.; BGS DEX 2869. Scale bars represent 4 mm. ligament, subovate outline and continuous dentition that characterize the Frejidae. In his description. Cope (19976) placed Alytodonta within the family Parallelodontidae. However, this group contains predominantly Mesozoic and Cenozoic genera that have hinge plates, shell outlines and external prosopon that are easily distinguished from the Frejidae. The parallelodontids have a more inequilateral dental arrangement, chiefly consisting of short, oblique anterior teeth and very long posterior pseudolaterals that are almost parallel with the hinge line. Furthermore, their shells are usually antero-posteriorly elongated and ornamented with both concentric growth lines and radial ridges and gutters. The family Frejidae cannot accommodate Catamarcaia , which displays a duplivincular ligament and a composite dentition of actinodontoid laterals and some small taxodont teeth (Sanchez 1995). The combination of these characters is unique to this bivalve and has resulted in previous uncertainty regarding its sytematic position. It is clearly a neotaxodont but, as yet, cannot be readily accommodated in any existing family. Stratigraphical range. Lower Llandovery to Upper Wenlock of Britain, Upper Wenlock of Gotland (Sweden), and possibly the Wenlock of Wisconsin, USA. Genus trecanolia gen. nov. Derivation of name. From the Welsh tre (= settlement), and canoi (= middle) - the Welsh equivalent of ‘Middleton’, an allusion to the type locality. Type species. Trecanolia acincta sp. nov., by monotypy. Diagnosis. Equivalved, inequilateral, moderately inflated, opisthocline, sub-orbicular to oblique shells with a narrow, straight hinge plate, gently rounded posterior and ventral margins and a narrow, strongly convex anterior extremity; umbones elevated above hinge line; beaks orthogyre. Dentition of 11-18 simple, cardinal teeth proximal to umbonal area, terminated by two short anterior pseudolaterals and two longer posterior pseudolaterals; both sets elongated normal to hinge plate; underside of ventral anterior pseudolateral has three to six cardinal denticles; ventral side of posterior pseudolaterals has up to seven taxodont denticles, juvenile specimens may lack the posterior denticles; subumbonal edentulous space present only in juveniles. Adductor scars anisomyarian; anterior adductor scar ovate and supported by shallow myophoric buttress; 978 PALAEONTOLOGY, VOLUME 41 A B Length (mm) text-fig. 2. a, bivariate scatter diagram of dimensions measured on Trecanolia acincta gen. sp. nov. b, angles and measurements taken on T. acincta. posterior scar larger, but less deeply impressed. Ligament duplivincular and chevron-shaped, located between hinge and beak. External prosopon unknown. Remarks. Trecanolia is distinguished from other frejids by the oblique and posteriorly elongate shell shape and by the presence of large denticles on the underside of the ventral anterior and posterior pseudolaterals (Text-fig. 3). Trecanolia is most closely related to Freja\ both genera have numerous cardinal teeth and pseudolaterals oriented normal to the hinge plate, similar anisomyarian adductors, and orthogyre beaks. However, Freja has an orbicular, subequilateral shell outline, a gently convex anterior margin, less elongate posterior pseudolaterals and lacks the unusual denticles. Uskardita is easily distinguished from Trecanolia by the large cardinal area, more conspicuous duplivincular ligament, oblique and shorter anterior cardinal teeth and the more upright umbo. Alytodonta is more distantly related, as the three posterior pseudolaterals are elongate and rather pterioid in appearance and the beaks are prosogyrate. Trecanolia acincta gen. et sp. nov. Plate 1, figures 1-10; Text-figures 2-3 v. 1978 Bivalve gen. et sp. nov.; Squirrel and White, pi. 3, figs 12-14 (?) 1996 Leiopteria cf. undata (Hall, 1852); Watkins, fig. 2e. Derivation of name. From the Latin acinctus (= well equipped); referring to the numerous teeth and denticles. EXPLANATION OF PLATE 1 Figs 1-10. Trecanolia acincta gen. et sp. nov.; overflow cut from the old lake in Middleton Hall Estate, near Middleton Hall Lodge, 2995 metres S68° W of Dryswlyn Station, Carmarthenshire; grid reference SN 5265 1880; probably Upper Wenlock. 1, BGS DEX 2869A, holotype; internal mould of left valve; x 5. 2, latex cast of holotype; x 5. 3-4, BGS DEX 2834B; internal mould of left valve and latex cast; x 7. 5, 9, BGS DEX 2869B; hinge detail of latex cast and internal mould of right valve; x 6. 6-7, BGS DEX 2843 A; internal mould of right valve and latex cast; x 6. 8, BGS DEX 2848; dorsal view of conjoined valves, internal mould; x 4. 10, BGS DEX 2880; internal mould of left valve; x 5. PLATE 1 RATTER and COPE, Trecanolia 980 PALAEONTOLOGY, VOLUME 41 Holotvpe. BGS DEX 2869A, a complete internal mould of a left valve. The specimen has been tectonically distorted resulting in a shortened antero-posterior axis. No undeformed specimens have been collected. Paratypes. BGS DEX 2869B (ILC), BGS DEX 2823A (IRF), BGS DEX 2834A, B, C, D, E (IRC, ILC, IRF, ILC and IRP respectively), BGS DEX 2839A (ILC), BGS DEX 2841 and 2847A (ILF and ELF - part and counterpart), BGS DEX 2847B, C, D (IRF, ILF and IRP respectively), BGS DEX 2843A (IRP), BGS 2848 (IAF), DEX 2880A (ILP). All specimens are from the type locality and horizon. Type locality and horizon. Overflow cut from old lake in Middleton Hall Estate, near Middleton Hall Lodge, 2995 metres S68° W of Dryswlyn Station, Carmarthenshire, west Wales; grid reference SN 5265 1880. Probably of late Wenlock age (Squirrel and White 1978). Measurements BGS DEX AL L H W OL AL/L H/L OL/L U° ■a ttf) cd C G i_ "™ ^ O w _ Ui , , (D G £ ~ 5 n £ o 5 N 3* u* cd ^ £ r£ C/5 "So W C/5 cd 42 c/5 G a> s JD 13 O ^ • a 0 « C/5 c/5 Y c <3 C to o c o s <3 . cd *U aT]3 22 ' S3 G '* o 42 C/5 JL> • «-N l-( •a g 1 * G? J2 1 g 2 o G O 3 co 6 (U ^ G 'G X> cu ^ C/5 ^ w rn a ; E •— 42 MANNIK: SILURIAN CONODONTS 1007 Several other taxa, originally also described as species of Pterospathodus, appear to belong to some other genera. P. posteritenuis (Uyeno and Barnes 1983, pi. 2, figs 1-11, 14—18) is, most probably, identical to Pranognathus tenuis (Mannik and Aldridge 1989, text-fig. 5). The apparatus of P. cadiaensis of Bischoff (1986) was studied in detail by Wang and Aldridge (1996), and reidentified as Gamachignathus macroexcavatus. P. retroramus of McCracken (1991, pi. 4, figs 24—25; pi. 5, figs 1-5, 8) is most probably related to Astropentagnathus, not to Pterospathodus. P. eopennatus lineage P. eopennatus sp. nov. 1971 p 1971 vp .1972 v. 1972 .1975 non 1975 1978 1978 1978 1978 .1979 .1979 .1979 .1979 1980 p? 1983 .1985 1985 v. 1986 v. 1986 ? 1986 p? 1986 ? 1987 1988 v. 1989 1990 1990 1990 1990 p 1990 v. 1990 ? 1991 ? 1996 ? 1996 Spathognathodus celloni Walliser, 1964; Schonlaub, p. 44, pi. 2, figs 1-5. Carniodus carinthiacus Walliser, 1964; Schonlaub, p. 46, pi. 3, fig. 6 (non figs 7-8 [= P. a. amorphognathoides ]). Neoprioniodus costatus paucidentatus Walliser, 1964; Aldridge, p. 193, pi. 5, fig. 21 (non fig. 20 [indet.]). Ozarkodina adiutricis Walliser, 1964; Aldridge, p. 198, pi. 5, figs 2-3. Llandoverygnathus celloni (Walliser, 1964); Aldridge, pi. 1, figs 20-21. Llandoverygnathus celloni (Walliser, 1964); Schonlaub, p. 53, pi. 1, figs 18-19 ([ = Aulacognathus ? sp.]). Neoprioniodus costatus paucidentatus Walliser, 1964; Miller, pi. 2, fig. 12. Exochognathus brevialatus (Walliser, 1964); Miller, pi. 3, figs 7-8. Ozarkodina adiutricis Walliser, 1964; Pickett, pi. 1. fig. 27. Neospathognathodus pennatus (Walliser, 1964); Pickett, pi. 1, figs 24—25. Carniodus sp. Aldridge, p. 12, pi. 1, fig. 8. Llandoverygnathus celloni (Walliser, 1964); Aldridge, pi. 1, figs 9-10. Llandoverygnathus pennatus (Walliser, 1964); Aldridge, pi. 1, fig. 11. Llandoverygnathus sp. 12 Aldridge, pi. 1, figs 12-15. Pterospathodus celloni (Walliser, 1964); Aldridge, fig. 1. simple cone element, group ‘c’ Uyeno and Barnes, p. 26, pi. 8, figs 6-7, 9-12, 18? (non fig. 19 [indet.]). Pterospathodus celloni (Walliser, 1964); Aldridge, p. 80, pi. 3.1, figs 25-26. Pterospathodus celloni (Walliser, 1964); Qiu, pi. 1, figs 1-2. Pterospathodus celloni (Walliser, 1964); Bischoff, p. 194, pi. 28, figs 34-39; pi. 29, figs 1-8. Pterospathodus pennatus (Walliser, 1964); Bischoff, p. 200, pi. 30, figs 12-14, 23-30. Pterospathodus pennatus (Walliser, 1964); Jiang et al., pi. 4, fig. 3. Spathognathodus celloni Walliser, 1964; Jiang et al., pi. 4, figs 5, 17 (non fig. 6 [indet.]). Pterospathodus celloni (Walliser, 1964); Over and Chatterton, p. 2, fig. 1. Pterospathodus celloni (Walliser, 1964); Qiu, pi. 1, fig. 3 [cop. Qiu 1985, pi 1, fig. 1]. Pterospathodus, celloni-morph Mannik and Aldridge, text-fig. 3 A. Carniodus sp. Armstrong, pi. 8, fig. 5. Pterospathodus celloni (Walliser, 1964); Armstrong, p. 118, pi. 19, figs 6-14. Pterospathodus pennatus pennatus (Walliser, 1964); Armstrong, p. 119, pi. 19, figs 15-17. Pterospathodus celloni (Walliser, 1964); Uyeno, p. 65, pi. 3, figs 1-7, 13-14; pi. 11, figs 25-30. Pterospathodus cf. P. celloni (Walliser, 1964); Uyeno, pi. 1 1, figs 18-20 (non pi. 3, figs 8-10 [ = Astropentagnathus ? sp.]). Pterospathodus celloni (Walliser, 1964); Mannik and Viira, pi. 17, figs 18, 21. Pterospathodus celloni (Walliser, 1964); McCracken, p. 109, pi. 4, figs 4-11. Carniodus carnulus Walliser, 1964; Wang and Aldridge, pi. 4, fig. 12. Pterospathodus celloni (Walliser, 1964); Wang and Aldridge, pi. 5, figs 2-3. Derivation of name. In reference to the morphological similarity and postulated direct evolutionary relationship to P. pennatus. Holotype. Pa element Cn 7879, Nurme core, sample M-889, int. 30-20-30-30 m; Plate 1, figure 19. Type horizon and locality. Lower part of the Velise Formation, Adavere Regional Stage, Telychian; Nurme core, interval 14-00-30-30 m. 1008 PALAEONTOLOGY, VOLUME 41 text-fig. 4. For caption see opposite. MANNIK: SILURIAN CONODONTS 1009 Diagnosis. Pa element morphologically highly variable, with a pennate inner lateral process. Pb2 element without anterior process. Posterior processes of S elements evenly denticulated. Processes of carniciform element undenticulated. Description. Pa element is represented by eight main morphs. Morph 7a-elements relatively long (up to 16-18 denticles). Sinistral elements (PI. 1, figs 18-19; PI. 2, fig. 35; Text-figs 4p, 5m, r-s), as a rule, without lateral process. The denticles are lower in the middle of the blade and relatively higher at each end. Dextral elements (PI. 1, figs 10, 20, 22; PI. 2, figs 34, 40; Text-figs 4x, 6a) differ by having a pennate lateral process and have higher denticles only at the anterior end of the blade. Both sinistral and dextral elements possess a short, triangular, generally undenticulated outer lateral lobe or process (a few specimens have a single denticle on it). Basal cavity deep and wide under posterior part of element. The lower line of the denticle roots turns steeply down in the posterior part of the blade. Occurrence. Appears in the A. irregularis Subzone, is quite rare in the A. kuehni Subzone, but becomes common in the A. tuber culatus ssp. nov. 2 Subzone. Morph lb (PI. 1, fig. 21; Text-figs 4n-o, q, 6e-f, j-k)- similar to morph la but differs from it by having considerably shorter denticles and a higher base. Rare dextral specimens may possess a bifurcated lateral process (Text-fig. 4n). Occurrence. The same as for morph la. Morph 2a - short elements with distinctly higher denticles on the posterior part (sinistral element - PI. 2, figs 32, 37, 39; Text-figs 5l, 6i, n, t) or anterior part (dextral element - PI. 2, figs 23, 36; Text-figs 5j-k, 6g-h, m) of the blade. Lateral process better developed on the dextral element. Short triangular undenticulated lateral process is common on the outer side on both elements. Denticles are relatively tall. Occurrence. Appears in the uppermost part of the A. irregularis Subzone and reaches the A. tuberculatus ssp. nov. 3 Subzone. Dominates in the A. kuehni Subzone but is less common below and above that interval. Morph 2b (Text-fig. 5o-p) - similar to morph 2a; differs from it by having shorter denticles and higher base. Occurrence. The same as for morph 2a, but morph 2b is less common. Morph 3 (PI. 1, figs 16-17; Text-figs 4a-b, g, 5d, g-i, 7h-i, q)- dextral and sinistral forms almost identical. Elements relatively long with narrow, tall denticles. Denticle roots (white matter) almost reaches the base line on the distal parts of the blade. The denticles are somewhat lower near the cusp but higher on the distal parts of the blade. A denticulated lateral process is common on dextral, but rare on sinistral elements. Basal cavity shallow and narrow. Occurrence. Relatively rare but occasionally present from the A. irregularis Subzone to A. tuberculatus ssp. nov. 3 Subzone. Morph 4 (PI. 2, figs 33, 38, 41; Text-fig. 6l, op, u) - relatively short, anterior denticles highest proximally, decreasing gradually in height in the distal direction. Posterior denticles are considerably shorter. The sharp decrease in the height of denticles just behind the cusp is the most characteristic feature of this morph. Intermediates between morphs 4 and 2 have been found. Occurrence. A. tuberculatus ssp. nov. 2 Subzone. text-fig. 4. Pterospathodus eopennatus ssp. nov. 1 ; Astropentagnathus irregularis Subzone, a-b, g, Pa element, morph 3. OF, h-m, Pa element, morph 5. N-o, Q, Pa element, morph lb. p, x, Pa element, morph la. R-w, Pa element, morph 6. y-z, a1— F1, j1, M1, Pbj element, w^d2, Pb2 element, g1-!1, o\ element; k1-!,1, n\ p1, Pc element, q'-u1, carnuliform element, morph a. v1, modified carnuliform element. E2, G2, carniciform element. F2, i2, SCj element. H2, J2, Sc2 element, k2, Sb2 element. L2, Sa element. M2, Sbj element. Scale bar represents 1 mm. 1010 PALAEONTOLOGY, VOLUME 41 Morph 5 - both sinistral (PI. 1, figs 25-27, 33; Text-figs 4f, k-m, 5b-c, f) and dextral (PI. 1, figs 28(7), 30, 35-36; Text-figs 4c-e, h-j, 5a, e) forms relatively short with higher denticles close to the ends of the blade. Those on the anterior blade are slightly higher than on the posterior. Basal margin almost straight in lateral view (PI. 1, figs 26, 30; Text-fig. 4h-m) or slightly convex (PI. 1, fig. 28(7); Text-fig. 4c-f). Inner denticulated lateral process well developed on both forms. Shorter outer lateral process better developed (often bearing one or two denticles) on the sinistral forms. Occurrence. Upper part of the A. irregularis Subzone and the lowermost part of the A. kuehni Subzone. Morph 6 (PI. 1, figs 23(7), 24, 29, 34; Text-fig. 4r-w) - similar to morph 5 but differs by being relatively shorter, possessing higher denticles (particularly on dextral elements - PI. 1, figs 29, 34; Text-fig. 4r-s, u) and having poorly developed lateral process(es). Occurrence. A. irregularis Subzone. EXPLANATION OF PLATE 1 Figs 1-9, 1 1-15. Pterospathodus eopennatus ssp. nov. 2. 1, Cn 7861 ; outer lateral view of sinistral Pb, element. 2, Cn 7862; outer lateral view of dextral Pb2 element. 3, Cn 7863; posterior view of dextral Sb2 element. 4, Cn 7864; outer lateral view of sinistral modified carnuliform element. 5, Cn 7865; outer lateral view of dextral curved element, morph a. 6, Cn 7866; inner lateral view of dextral Sc2 element. 7, Cn 7867; outer lateral view of sinistral carnuliform element, morph a. 8, Cn 7868; posterior view of sinistral Sb, element. 9, Cn 7869; inner lateral view of dextral carniciform element. 1 1, Cn 7870; outer lateral view of dextral Pb, element. 12, Cn 7871 ; inner lateral view of dextral Sc3 element. 13, Cn 7872; inner lateral view of dextral Sc, element. 14, Cn 7873; outer lateral view of dextral Pc element. 15, Cn 7874; outer lateral view of dextral Pb, element. Figs 1 and 1 1 from Nurme core, sample M-900, int. 22-65-22-80 m; figs 2-3, 5, 8 and 14 from Nurme core, sample M-907, int. 1 7-50—17-60 m; figs 4, 6-7, 9, 12-13 and 15 from Viki core, sample M-8, int. 168-60-168-80 m. Figs 10, 16-46. Pterospathodus eopennatus ssp. nov. 1. 10, Cn 7875; inner lateral view of dextral Pa element, morph la. 16, Cn 7876; inner lateral view of sinistral Pa element, morph 3. 17, Cn 7877; inner lateral view of dextral Pa element, morph 3. 18, Cn 7878; inner lateral view of sinistral Pa element, morph la. 19, Cn 7879; inner lateral view of sinistral Pa element, morph la. 20, Cn 7880; upper view of dextral Pa element, morph la. 21, Cn 7881 ; inner lateral view of sinistral Pa element, morph lb. 22, Cn 7882; inner lateral view of dextral Pa element, morph la. 23, Cn 7883; inner lateral view of sinistral Pa element, morph 6(?). 24, Cn 7884; inner lateral view of sinistral Pa element, morph 6. 25, Cn 7885; inner lateral view of sinistral Pa element, morph 5. 26, Cn 7886; inner lateral view of sinistral Pa element, morph 5. 27, Cn 7887; upper view of sinistral Pa element, morph 5. 28, Cn 7888; inner lateral view of dextral Pa element, morph 5(7). 29, Cn 7889; inner lateral view of dextral Pa element, morph 6. 30, Cn 7890; inner lateral view of dextral Pa element, morph 5. 31, Cn 7891 ; outer lateral view of dextral Pc element. 32, Cn 7892; outer lateral view of dextral Pc element. 33, Cn 7893; upper view of sinistral Pa element, morph 5. 34, Cn 7894; inner lateral view of dextral Pa element, morph 6. 35, Cn 7895 ; upper view of dextral Pa element, morph 5. 36, Cn 7896 ; upper view of dextral Pa element, morph 5. 37, Cn 7897; outer lateral view of sinistral Pc element. 38, Cn 7900; lateral view of symmetrical Pb2 element. 39, Cn 7899; outer lateral view of dextral Pb2 element. 40, Cn 7898; outer lateral view of sinistral Pb2 element. 41, Cn 7901 ; outer lateral view of sinistral Pb, element. 42, Cn 7902; outer lateral view of sinistral Pb, element. 43, Cn 7903; outer lateral view of dextral Pb, element. 44, Cn 7904; inner lateral view of sinistral Sc2 element. 45, Cn 7905; inner lateral view of dextral Sc2 element. 46, Cn 7906; inner lateral view of sinistral carniciform element. Figs 10, 18, 20 and 21 from Valgu section, sample M-882; figs 16-17 from Nurme core, sample M-891, int. 29T0-29-20m; figs 19, 22 and 32 from Nurme core, sample M-889, int. 30-20-30-30 m; figs 23-24, 34 and 4 1 —43 from Viki core, sample M-954, int. 183-17-183-32 nr; figs 25, 28 and 35 from Viki core, sample M-960, int. 181-81—181-91 m; fig. 26 from Viki core, sample M-956, int. 1 82-90—1 83-04 m; figs 27, 31 and 37 from Viki core, sample M-962, int. 1 8 1 -29— 181 -40 m; fig. 29 from Viki core, sample M-958, int. 1 82-22—1 82-30 m; figs 30, 33 and 36 from Nurme core, sample M-890, int. 29-50-29-60 m; figs 38-40, 44-46 from Nurme core, sample M-903, int. 2040-20-50 m. All x 50. PLATE 1 MANNIK, Pterospathodus 1012 PALAEONTOLOGY, VOLUME 41 text-fig. 5. For caption see opposite. MANNIK: SILURIAN CONODONTS 1013 Remarks. In all morphs the sinistral and dextral forms of the Pa element are morphologically different from each other. The pennate inner lateral process tends to be more common on the dextral element. In this paper two populations, evolutionarily connected and stratigraphically following each other, are described as subspecies of P. eopennatas : P. eopennatus ssp. nov. 1 and P. eopennatus ssp. nov. 2. P. eopennatus can be recognized world-wide (see synonymy). However, a revision of collections is needed to identify subspecies. Pterospathodus eopennatus ssp. nov. 1 Plate 1, figures 10, 16-46; Text-figures 4, 5a-c, e-f Material. Several hundred to over a thousand of each of Pa, Pbx and Pc elements; tens to hundreds of Pb2, M, Scx, Sc2, Sb, Sa and carnuliform elements; few tens of carniciform elements. Diagnosis. P. eopennatus with Pa element represented by morphs la, lb, 3, 5 and 6; the last two are found only in this taxon. Pbx element relatively long, arched in lateral view. Remarks. The relative abundance of morphs varies in different parts of the studied area. Morphs 1 a and 1 b dominate faunas in the continental part of Estonia whereas morph 6 is the most abundant in the Viki core from the western part of the island of Saaremaa (Text-fig. 1). However, all morphs described above have been found in all studied sections from this interval. Occurrence. A. irregularis Subzone and the lowermost part of the A. kuehni Subzone. Pterospathodus eopennatus ssp. nov. 2 Plate 1, figures 1-9, 11-15; Plate 2, figures 23, 32-41; Text-figures 5d, g-v\ 6 v. 1998 Pterospathodus sp. nov. e Mannik and Malkowski, pi. 1, figs 19, 23-25, 27. Material. Several hundred Pa and Pbx elements; tens to hundreds of Pb2, Pc, M, Scx and Sc2 elements; a few tens of Sc3, Sb, Sa, carnuliform and carniciform elements. Diagnosis. P. eopennatus with the Pa element represented by morphs la, lb, 2a, 2b, 3 and 4. Pb, element short and triangular in lateral view. Remarks. Pbx element dominated by forms which are relatively short and almost triangular in lateral view (PI. 1, figs 1,11; Text-figs 5t-u, c'-d1, 6q-r). Rare specimens with a longer anterior process (PI. 1, fig. 15; Text-fig. 5q) may be found in the lower part of the P. eopennatus ssp. nov. 2 range but become dominant in the upper part (in the A. tuberculatus ssp. nov. 2 Subzone). Other elements are morphologically identical in both subspecies. Two chronological populations are recognized in the range of P. eopennatus ssp. nov. 2. 1. The population in the A. kuehni Subzone (Text-fig. 5d, g-v1) is dominated by morph 2a; morphs la and lb are rare and morph 3 can be found occasionally. 2. The population in the A. tuberculatus ssp. nov. 2 Subzone (Text-fig. 6) is dominated by morphs la and lb; morphs 2a and 2b occur but are rare. Morph 4 is restricted to this subzone. Also, the oldest Sc3 elements (PI. 1, fig. 12; Text-fig. 6P-K1) found so far come from this population. Occurrence. A. kuehni and A. tuberculatus ssp. nov. 2 subzones. text-fig. 5. Pterospathodus eopennatus ssp. nov.; Aulacognathus kuehni Subzone, a-c, e-f. Pterospathodus eopennatus ssp. nov. 1, Pa element, morph 5. d, g-v1, Pterospathodus eopennatus ssp. nov. 2. D, g-i. Pa element, morph 3. j-l. Pa element, morph 2a. m-n, r-s. Pa element, morph la. o-p, Pa element, morph 2b. q, t-u, c1— d1, Pbx element, v, e1, Mx element, w-z, Pb2 element, a'-b1, Scx element, f1, k1, Pc element, h1, Sb2 element. G1, r1, Sbx element. P-J1, Sc2 element. H-m1, s1, carnuliform element, morph a. N1, curved element, morph a. o1, Sa element, p'-q1, modified carnuliform element. Tx-v\ carniciform element. Scale bar represents 1 mm. 1014 PALAEONTOLOGY, VOLUME 41 text-fig. 6. For caption see opposite. MANNIK: SILURIAN CONODONTS 1015 P. amorphognathoides lineage Pterospathodus amorphognathoides Walliser, 1964 sensu nov. Diagnosis. The Pa element of P. amorphognathoides is characterized by an inner lateral process, pennate in the older and bifurcated in the younger forms; with (in younger populations) or without (in older ones) a basal platform; may or may not possess a triangular to semiquadrate lateral lobe or short, usually undenticulated process on the outer side of element. Pb2 element with anterior and posterior processes. Posterior processes of the S elements unevenly denticulated; within the row of short narrow denticles a few larger ones are randomly situated. Processes of the carniciform element bear up to four or five tiny denticles. Remarks. P. amorphognathoides is represented by a morphologically variable sequence of closely related populations including several evolutionarily connected successive subspecies : P. a. angulatus , P. a. lennarti ssp. nov., P. a. lithuanicus and P. a. amorphognathoides (see below). The changes in the morphology of the elements in the apparatus at the boundary between the P. eopennatus and P. amorphognathoides lineages are relatively sharp and took place at the level corresponding to one of the main events in the evolution of Telychian conodont faunas (Mannik 1995). v.* 1964 1975 1981 ? 1981 ? 1981 .1982 .1982 1982 ? 1983 p? 1983 p 1983 1983 ? 1985 1987 1987 1987 ? 1988 v. 1989 v. 1989 Pterospathodus amorphognathoides angulatus (Walliser, 1964) Plate 2, figures 1-22, 24-31; Text-figures 7-8 Spathognathodus pennatus angulatus Walliser, p. 79, pi. 14, figs 19-22. Llandoverygnathus pennatus (Walliser, 1964); Aldridge, pi. 1, figs 24-25. Pterospathodus pennatus procerus (Walliser, 1964); Uyeno and Barnes, pi. 1, fig. 23. Pterospathodus celloni (Walliser, 1964); Uyeno and Barnes, pi. 1, figs 20-21. Carniodus carnulus Walliser, 1964; Uyeno and Barnes, pi. 1, figs 18-19. Pterospathodus pennatus angulatus (Walliser, 1964); Aldridge and Mohamed, pi. 2, figs 8-11. Pterospathodus pennatus pennatus (Walliser, 1964); Aldridge and Mohamed, pi. 2, fig. 12. Pterospathodus celloni (Walliser, 1964); Aldridge and Mohamed, pi. 2, fig. 7. Carniodus carnulus Walliser, 1964; Uyeno and Barnes, p. 16, pi. 5, figs 1-10 (figs 2-3 [= cop. Uyeno and Barnes 1981, pi. 1, figs 18-19]). Pterospathodus celloni (Walliser, 1964); Uyeno and Barnes, p. 24, pi. 5, figs 17-18, 20-24 (non fig. 18 [indet.]; figs 20-22 [= cop. Uyeno and Barnes 1981, pi. 1, figs 20-21]). Ozarkodina polinclinata (Nicoll and Rexroad, 1968); Uyeno and Barnes, p. 22, pi. 5, fig. 19 (non figs 11-16 [= O. polinclinata ]). Pterospathodus pennatus procerus (Walliser, 1964); Uyeno and Barnes, p. 24, pi. 8, figs 1-3 [fig. 1 [= cop. Uyeno and Barnes 1981, pi. 1, fig. 23]). Pterospathodus pennatus pennatus (Walliser, 1964); Aldridge, p. 80, pi. 3.1, fig. 27. Pterospathodus celloni (Walliser, 1964); An, p. 202, pi. 33, figs 8-10. Neoprioniodus triangularis paucidentatus Walliser, 1964; An, pi. 35, figs 21-22. Pterospathodus pennatus (Walliser, 1964); Dumoulin and Harris, fig. 4 N. Pterospathodus pennatus procerus (Walliser, 1964); Qiu, pi. 1, figs 5-7 [cop. Qiu 1985, pi. 1, figs 5, 8-9]). Pterospathodus , angulatus-morph Mannik and Aldridge, text-fig. 3 B. Pterospathodus , pennatus-morph Mannik and Aldridge, text-fig. 3C. text-fig. 6. Pterospathodus eopennatus ssp. nov. 2; Apsidognathus tuberculatus ssp. nov. 2 Subzone, a-d. Pa element, morph la. e-f, j-k, Pa element, morph lb. g-i, m-n, t. Pa element, morph 2a. L, o-p, u, Pa element, morph 4. q-r, w, Pbj element, s, z-E1, Pb2 element, v, x-Y, H1, Pc element, f'-g1, Mt element. L1— N1, Sct element. P-K1, Sc3 element, o1, iP-v1, Sc2 element, p1, Sb2 element. Q1— R1, Sb3 element, s1-^, carniciform element, w1, Sa element, x'-z1, carnuliform element, morph a. a2-b2, carnuliform element, morph b(?). c2-d2, modified carnuliform element. e2-f2, curved element, morph a. Scale bar represents 1 mm. 1016 PALAEONTOLOGY, VOLUME 41 v. 1998 Pterospathodus cf. amorphognathoides angulatus (Walliser, 1964); Mannik and Malkowski, pi. 1, figs 20-22. Material. Several hundred of each of the P, M, Sc and carnuliform elements; tens to hundreds of Sb, Sa and carniciform elements. Diagnosis. P. amorphognathoides with elements without platform. Pa element long, with pennate inner lateral process and lower denticles in the middle part of the blade. Remarks. Pa element of P. a. angulatus is similar to the morphs la and lb of the Pa element of P. eopennatus ssp. nov. 2 (PI. 2, figs 34-35; Text-fig. 6a-f, j-k) but differs by having a very long blade with at least 20, usually even more denticles on mature specimens. The Pa element of P. a. angulatus is represented by two morphs, one of them with tall denticles (PI. 2, figs 1, 3-4, 9; Text-figs 7a-b, d, 8b-c, f-h) and the other with short denticles (PI. 2, fig. 12; Text-figs 7c, G, k, 8a). The ends of lateral processes on Sb elements lack bifurcation. In the lower part of the range of P. a. angulatus morphs 2 and 3 of the Pa, but also extremely rare specimens of the Pb2 and the carniciform elements typical of P. eopennatus ssp. nov. 2 occur occasionally. P. a. angulatus is the oldest representative of the P. amorphognathoides lineage and was evolutionarily followed by P. amorphognathoides lennarti ssp. nov. explanation of plate 2 Figs 1-22, 24-31. Pterospathodus amorphognathoides angulatus (Walliser, 1964). 1, Cn 7907 ; inner lateral view of dextral Pa element, morph a. 2, Cn 7908; outer lateral view of dextral Pb2 element. 3, Cn 7909; inner lateral view of sinistral Pa element, morph a. 4, Cn 7910; upper view of dextral Pa element, morph a. 5, Cn 7911; outer lateral view of sinistral Pb2 element. 6, Cn 7912; outer lateral view of sinistral Pbt element. 7, Cn 7913; outer lateral view of sinistral Pc element. 8, Cn 7914; outer lateral view of sinistral Pb2 element. 9, Cn 7915; inner lateral view of sinistral Pa element, morph a. 10, Cn 7916; outer lateral view of dextral Pbx element. 1 1, Cn 7917; outer lateral view of dextral Pbx element. 12, Cn 7918; inner lateral view of sinistral Pa element, morph b. 13, Cn 7919; outer lateral view of sinistral carnuliform element, morph a. 14, Cn 7920; outer lateral view of sinistral carnuliform element, morph a(?). 15, Cn 7921; outer lateral view of sinistral modified carnuliform element. 16, Cn 7922; inner lateral view of sinistral Sc3 element. 17, Cn 7923; inner lateral view of sinistral SCj element. 18, Cn 7924; inner lateral view of dextral element. 19, Cn 7925; inner lateral view of sinistral MT element. 20, Cn 7926; inner lateral view of sinistral Sc2 element. 21, Cn 7927; inner lateral view of dextral Sbj element. 22, Cn 7928; inner lateral view of dextral Sc2 element. 24, Cn 7929; inner lateral view of dextral carniciform element. 25, Cn 7930; outer lateral view of sinistral curved element, morph a. 26, Cn 7931 ; outer lateral view of sinistral curved element, morph b. 27, Cn 7932; posterior view of dextral Sbj element. 28, Cn 7933; posterior view of dextral Sb2 element. 29, Cn 7934; lateral view of Sa element. 30, Cn 7935; outer lateral view of dextral Sb, element. 31, Cn 7936; posterior view of Sa element. Figs 1, 3^1, 6-7, 9, 1 1, 15 and 24 from Nurme core, sample M-1051, int. 1 2-70 — 1 2-85 m; figs 2, 5 and 10 from Uulu-330 core, sample M-1298, int. 1 39-84—1 39-92 m; figs 8 and 12 from Nurme core, sample M-1050, int. 1 3-30—1 3-40 m; figs 13-14, 16-22 and 25-31 from Velise-Korgekalda section, sample VE-2. Figs 23, 32-41. Pterospathodus eopennatus ssp. nov. 2. 23, Cn 7937; inner lateral view of dextral Pa element, morph 2a. 32, Cn 7938; inner lateral view of sinistral Pa element, morph 2a. 33, Cn 7939; upper view of dextral Pa element, morph 4. 34, Cn 7940; inner lateral view of dextral Pa element, morph la. 35, Cn 7941 ; inner lateral view of sinistral Pa element, morph la. 36, Cn 7942; inner lateral view of dextral Pa element, morph 2a. 37, Cn 7943; inner lateral view of sinistral Pa element, morph 2a. 38, Cn 7944; upper view of dextral Pa element, morph 4(?). 39, Cn 7945; inner lateral view of sinistral Pa element, morph 2a. 40, Cn 7946; inner lateral view of dextral Pa element, morph la. 41, Cn 7947; inner lateral view of dextral Pa element, morph 4. Figs 23, 34-35 and 37-38 from Nurme core, sample M-907, int. 1 7-50—17-60 m; figs 32 and 36 from Viki core, sample M-l 1, int. 1 75-60—1 75 80 m; figs 33, 39 and 41 from Nurme core, sample M- 900, int. 22-65-22-80 m; fig. 40 from Viki core, sample M-8, int. 168-60-168-80 m. All x 50. PLATE 2 MANNIK, Pterospathodus 1018 PALAEONTOLOGY, VOLUME 41 text-fig. 7. Pterospathodus amorphognathoides angulatus (Walliser, 1964); Apsidognathus tuberculatus ssp. nov. 3 Subzone, a-b, d. Pa element, morph a. c, G, K., Pa element, morph b. H-l, Q, Pa element, morph 3. e-f, j, p, Pb, element. L-o, R-u, Pb2 element, v-w. Pc element. x-Y, Sc2 element, z, a1, M, element. B1— D1, SCj MANNIK: SILURIAN CONODONTS 1019 Occurrence. A. tuberculatus ssp. nov. 3 and P. a. angulatus subzones. P. a. angulatus can be recognized world- wide (see synonymy). p. 1972 ? 1972 ? 1972 .1985 v. 1986 vp. 1986 Pterospathodus amorphognathoides lennarti ssp. nov. Plate 3, figures 21^46; Text-figure 9 Pterospathodus amorphognathoides Walliser, 1964; Aldridge, p. 208, pi. 3, fig. 18 (non figs 17, 19 [ = P. a. amorphognathoides ]). Neoprioniodus costatus costatus Walliser, 1964; Aldridge, p. 193, pi. 5, fig. 22. Ozarkodina gaertneri Walliser, 1964; Aldridge, p. 200, pi. 5, fig. 7. Pterospathodus pennatus subsp. nov. Aldridge, p. 81, pi. 3.1, fig. 28. P. celloni (Walliser); Nakrem, fig. 6a. Carniodus carnulus Walliser, 1964; Nakrem, fig. 61 ( non fig. 7c,/[= P. a. lithuanicus ]). Derivation of name. In honour of Dr Lennart Jeppsson, an expert on Silurian conodonts. Material. Several hundred to a thousand carnuliform elements; several hundred P, M, Sc, carniciform and curved elements; many tens to hundreds of Sb and Sa elements. Holotype. Dextral Pa element Cn 7968, Pahapilli core, sample M-1520, int. 47-30-47-20 m (Yelise Formation, northern Saaremaa, Estonia); Plate 3, figure 21. Type horizon and locality. Middle part of the Velise Formation, Adavere Regional Stage; Pahapilli core, interval 46-60-50-30 m. Diagnosis. P. amorphognathoides with elements without platform. The first denticle on the bifurcated lateral process of the Pa element is situated away from the main row of denticles and is connected with the last one by a narrow high ridge. Remarks. The most characteristic feature separating the Pa element of P. amorphognathoides lennarti ssp. nov. from that of younger subspecies (e.g. P. a. lithuanicus ; PI. 4, figs 29-32, 34-35; Text-fig. 10a-b, d-e) is the deep groove between the main row of denticles and the first one on the inner lateral process (Text-fig. 9a, d-e). As a rule, that denticle is connected with the main row by a narrow high ridge (PI. 3, figs 21-22, 25, 27; Text-fig. 9a, d-e). The configuration (the direction of the branches) of the lateral process is highly variable. The distal part of the posterior branch is usually turned parallel to the anterior one (PI. 3, fig. 27). The Pa element of P. amorphognathoides lennarti ssp. nov. is represented by two morphs. Morph 1 has tall denticles and a relatively low base (Text-fig. 9a, c, g) whereas morph 2 is characterized by short denticles and a higher base (Text-fig. 9e). The denticulation of morph 1 tends to be more irregular and includes overgrown denticles (Text-fig. 9g). A sub-triangular outer lateral lobe/short process is also characteristic, better developed on the sinistral element. P. amorphognathoides lennarti ssp. nov. is a direct descendant of P. a. angulatus. These two taxa differ mainly in the lack (or extremely rare occurrence) of bifurcation on the inner lateral process in P. a. angulatus and in the development of an outer lateral lobe on the Pa elements of P. a. lennarti ssp. nov. The apparatus of P. a. lennarti ssp. nov. differs from earlier taxa in the presence of short element, e'-f1, Sc3 element, g-i1, t1, carniciform element, j1, Sbj element, k1, Sb2 element, l1, curved element, morph a. M1, curved element, morph b. n1, modified carnuliform element, o'-s1, carnuliform element, morph a. Scale bar represents 1 mm. 1020 PALAEONTOLOGY, VOLUME 41 text-fig. 8. Pterospathodus amorphognathoides angulatus (Walliser, 1964); P. a. angulatus Subzone, a. Pa element, morph b. b-c, f-h, Pa element, morph a. E, n, t-u, Pbj element, d, i-m, r-s, z, Pb2 element, o-p, Pc element. Q, v, Mj element, w-x, Sc2 element. Y, SCj element. A1, Sbj element, b1, Sb2 element. c^-D1, carnuliform MANNIK: SILURIAN CONODONTS 1021 and modified short morphs of the carnuliform element (PI. 3, figs 24, 33; Text-fig. 9v-w, lx-m\ v1— w1). Occurrence. The P. amorphognathoides lennarti Subzone. On Gotland P. a. lennarti ssp. nov. has been found in loose pebbles from Sjalso (collection of L. Jeppsson - sample G88-637LJ). P. a. lennarti ssp. nov. has also been found in Carnic Alps (Seewarte section - collection of H. P. Schonlaub, sample 195/1-2, and probably in the Cellon section - collection of O. H. Walliser, one fragment in sample 10 H/J), in Great Britain (Aldridge 1972, pi. 3, fig. 18, Ticklerton 2 section; 1985, pi. 3.1, fig. 28, loc. 23 - uppermost Purple Shales of small stream 850 m south-west of Ticklerton, Shropshire) and Norway (Nakrem 1986, fig. 6 a, Malmoyakalven section, Vik Fm„ 42-5 m). Pterospathodus amorphognathoides lithuanicus Brazauskas, 1983 sensu novo Plate 3, figures 1-20, Plate 4, figures 21, 28-35; Text-figure 10 v.* 1983 Pterospathodus amorphognathoides lithuanicus Brazauskas, p. 60, figs 1-7. v. 1986 Pterospathodus amorphognathoides W alliser, 1964; Nakrem, fig. 6b— d, e(?), f—g, i. v.? 1986 Pterospathodus pennatus pennatus (Walliser, 1964); Nakrem, fig. 6 h. Material. Several hundreds to a thousand of each of the Pa, Pb and carnuliform elements; many hundreds of the Pc, M, S, carniciform and curved elements. Emended diagnosis. P. amorphognathoides without basal platform. The first denticle on the bifurcated inner lateral process is situated close to the main row of denticles. Remarks. Brazauskas (1983) described only the Pa element of the apparatus. Here P. a. lithuanicus is considered to include the complete set of elements of the Pterospathodus apparatus. Morphologically the most distinct element in this apparatus is the Pa element (PI. 4, figs 21, 29-32, 34, 45; Text-fig. 10a-b, d-e). The Pbx (PI. 3, figs 15, 17; Text-fig. 10c, F, o, u) and Pb2 (PI. 3, fig. 1 ; PI. 4, fig. 28; Text-fig. 10g-l, p) elements can also be quite easily separated from those of the older subspecies, as they possess a weak lateral basal thickening lacking on the corresponding elements of P. a. lennarti ssp. nov. As a rule, the sinistral Pa element (PI. 4, figs 31-32; Text-fig. 10b, e) possesses a distinct rounded or triangular lateral lobe on the outer side of the element. This structure is almost absent on the dextral element (PI. 4, figs 29-30; Text-fig. 10a, d). In the apparatus of P. a. lithuanicus a new modification of curved element, morph c, appears (PI. 3, fig. 16; Text-fig. IOy^z1). Occurrence. In the P. a. lithuanicus Subzone. Outside the Baltic (Estonia, Lithuania) P. a. lithuanicus has so far been illustrated only from Norway (Nakrem 1986, fig. 6 d,e(?),g-i, Malmoyakalven section, Vik Fm.). However, it is most probable, that after revision of collections from other regions of the world this taxon will be recognized to have a much wider distribution. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964 Plate 4, figures 1-20, 22-27 ; Plate 5; Text-figures 11-15 v.* 1964 Pterospathodus amorphognathoides Walliser, p. 67, pi. 15, figs 9-15. v. 1964 Ozarkodina gaertneri Walliser, p. 57, pi. 27, figs 12-19. v. 1964 ICarniodus carinthiacus Walliser, p. 3l, pi. 27, figs 20-26. v. 1964 Carniodus carnuius Walliser, p. 32, pi. 27, figs 27-38; pi. 28, fig. 1 element, morph b. E1— H1, Sc3 element. P-j1, curved element, morph a. K1, curved element, morph b. iP-m1, carniciform element, n1, modified carnuliform element. cP-T1, carnuliform element, morph a. Scale bar represents 1 mm. PALAEONTOLOGY, VOLUME 41 1022 v. 1964 Carniodus carnus Walliser, p. 34, pi. 28, figs 2-7. v. 1964 Carniodus carnicus Walliser, p. 32, pi. 28, figs. 8-11. vp. 1964 Neoprioniodus subcarnus Walliser, p. 51, pi. 28, figs 13-18 (non fig. 12 [= P. celloni]). v. 1964 Neoprioniodus triangularis triangularis Walliser, p. 52, pi. 28, figs 25-30. v. 1964 Neoprioniodus costatus costatus Walliser, p. 48, pi. 28, figs 36-41. v. 1964 Roundya latialata Walliser, p. 71, pi. 31, figs 11-14. .1966 Ozarkodina gaertneri Walliser, 1964; Spasov and Filipovic, p. 44, pi. 1, figs 1-2. .1966 Carniodus carinthiacus Walliser, 1964; Spasov and Filipovic, p. 38, pi. 1, fig. 3. .1966 Pterospathodus amorphognathoides Walliser, 1964; Spasov and Filipovic, p. 48, pi. 1, figs 4-5. .1966 Neoprioniodus subcarnus Walliser, 1964; Spasov and Filipovic, p. 42, pi. 1, figs 8-9. .1966 Neoprioniodus costatus costatus Walliser, 1964; Spasov and Filipovic, p. 42, pi. 1, figs 10-11. .1966 Carniodus carnus Walliser, 1964; Sapsov and Filipovic, p. 40, pi. 1, figs 12-13. 1966 Roundya brevialata Walliser, 1964; Spasov and Filipovic, p. 49, pi. 1, fig. 14. .1966 Carniodus carnulus Walliser, 1964; Spasov and Filipovic, p. 39, pi. 1, fig. 15. .1966 Carniodus carnicus Walliser, 1964; Spasov and Filipovic, p. 38, pi. 1, fig. 16. EXPLANATION OF PLATE 3 Figs 1-20. Pterospathodus amorphognathoides lithuanicus Brazauskas, 1983. 1, Cn 7948; outer lateral view of dextral Pb2 element. 2, Cn 7949; outer lateral view of sinistral Sbj element. 3, Cn 7950; inner lateral view of dextral Sc2 element. 4, Cn 7951 ; inner lateral view of sinistral Sc2 element. 5, Cn 7952; outer lateral view of sinistral curved element, morph a. 6, Cn 7953; inner(?) lateral view of dextralf?) modified carnuliform element, short morph. 7, Cn 7954; outer lateral view of sinistral curved element, morph b. 8, Cn 7955; posterior view of dextral Sbt element. 9, Cn 7956; inner lateral view of sinistral Mj element. 10, Cn 7957; inner lateral view of dextral Sc2 element. 11, Cn 7958; posterior view of sinistral Sbj element. 12, Cn 7959; inner lateral view of dextral Sc3 element. 13, Cn 7960; posterior view of sinistral Sb., element. 14, Cn 7961 ; outer lateral view of sinistral Pc element. 1 5, Cn 7962 ; outer lateral view of dextral Pbt element. 1 6, Cn 7963 ; outer lateral view of dextral curved element, morph c. 17, Cn 7964; outer lateral view of dextral Pbt element. 18, Cn 7965; inner lateral view of sinistral Sc3 element. 19, Cn 7966; outer lateral view of dextral carnuliform element, short morph. 20, Cn 7967; inner lateral view of dextral carniciform element. Figs 1, 5-8, 10, 13 and 18-20 from Viki core, sample M-976, int. 1 49-95—1 50 08 m; figs 2 and 1 1 from Viki core, sample M-362, int. 15F87-152 00 m; figs 3-4, 9 and 14-15 from Viki core, sample M-979, int. 148-75-148-85 m; fig. 12 from Viki core, sample M-972, int. 151 -25—1 5 1 -40 m; fig. 16 from Viki core, sample M-971, int. 151 -54 — 1 5 1-64 m; fig. 17 from Viki core, sample M-367, int. 146-70-146-80 m. Figs 21-46. Pterospathodus amorphognathoides lennarti ssp. nov. 21, Cn 7968; upper view of dextral Pa element. 22, Cn 7969; upper view of sinistral Pa element. 23, Cn 7970; outer lateral view of dextral curved element, morph b. 24, Cn 7971 ; inner lateral view of sinistral modified carnuliform element, short morph. 25, Cn 7972; upper view of dextral Pa element. 26, Cn 7973; outer lateral view of sinistral Pb2 element. 27, Cn 7974; upper view of dextral Pa element. 28, Cn 7975; inner lateral view of dextral Sc3 element. 29, Cn 7976; outer lateral view of sinistral Pc element. 30, Cn 7977; inner lateral view of sinistral Sc2 element. 31, Cn 7978; inner lateral view of dextral M, element. 32, Cn 7979; inner lateral view of dextral Sc3 element. 33, Cn 7980; outer(?) lateral view of dextral(?) modified carnuliform element, short morph. 34, Cn 7981 ; outer lateral view of dextral Sb2 element. 35, Cn 7982 ; outer lateral view of dextral Pbt element. 36, Cn 7983 ; outer lateral view of sinistral Pb: element. 37, Cn 7984; inner lateral view of sinistral Sc3 element. 38, Cn 7985; inner lateral view of sinistral Sc3 element. 39, Cn 7986; inner lateral view of sinistral element. 40, Cn 8080; outer lateral view of sinistral curved element, morph a. 41, Cn 7987; posterior view of Sa element. 42, Cn 7988; outer lateral view of dextral Pbt element. 43, Cn 7989; inner lateral view of dextral Sc2 element. 44, Cn 7990; posterior view of dextral Sb1 element. 45, Cn 7991 ; lateral view of symmetrical?) carnuliform element, morph b. 46, Cn 7992; outer lateral view of dextral modified carnuliform element. Fig. 21 from Pahapilli core, sample M-1520, int. 47-20-47-30 m; figs 22-25, 33, 35, 42^13 and 45^46 from Viki core, sample M-966, int. 1 53-88—1 54-05 m; figs 26-27 from Uulu-330 core, sample M-1070, int. 137-70-137-85 m; figs 28, 34, 40-41 and 44 from Uulu-330 core, sample M- 1301, int. 1 38 05—138- 1 5 m; figs 29-31 and 39 from Viki core, sample M-967, int. 153-60-153-72 m, figs 32 and 37-38 from Viki core, sample M-360, int. 1 53-35—1 53-50 m; fig. 36 from Viki core, sample M-968, int. 1 53 05-1 53-20 m. All x 50. PLATE 3 MANNIK, Pterospathodus 1024 PALAEONTOLOGY, VOLUME 41 text-fig. 9. For caption see opposite. MANNIK: SILURIAN CONODONTS 1025 p. 1968 Ozarkodina gaertneri Walliser, 1964; Igo and Koike, p. 14, pi. 1, figs 5-6 (non figs 7-9 [= P. pennatus procerus]). 1968 Pterospathodus amorphognathoides Walliser, 1964; Igo and Koike, p. 16, pi. 2, figs 12-13. p. 1968 Carniodus sp. A Igo and Koike, p. 8, pi. 3, fig. 3 (non fig. 2 [ = P. p. procerus ?]). p. 1968 Neoprioniodus spp. Igo and Koike, p. 14, pi. 3, fig. 4 (non fig. 24 [= P. p. procerusl]). .1968 Ozarkodina gaertneri Walliser, 1964; Nicoll and Rexroad, p. 49, pi. 2, figs 12-14. .1968 Ozarkodina neogaertneri Nicoll and Rexroad, p. 50, pi. 2, figs 15-16. 1968 Pterospathodus amorphognathoides Walliser, 1964; Nicoll and Rexroad, p. 56, pi. 3, figs 1-5, 6?, 7. .1968 Carniodus carinthiacus Walliser, 1964; Nicoll and Rexroad, p. 24, pi. 5, figs 1-2. 1968 Carniodus carnicus Walliser, 1964; Nicoll and Rexroad, p. 25, pi. 5, fig. 3. .1968 Carniodus carnulus Walliser, 1964; Nicoll and Rexroad, p. 25, pi. 5, figs 4-5. p. 1968 Carniodus carnus Walliser, 1964; Nicoll and Rexroad, p. 26, pi. 5, figs 6, 8 (non fig. 7 [indet.]). .1968 Neoprioniodus subcarnus Walliser, 1964; Nicoll and Rexroad, p. 41, pi. 5, fig. 10. .1968 Neoprioniodus costatus Walliser, 1964; Nicoll and Rexroad, p. 40, pi. 5, figs 15-16. .1968 Neoprioniodus triangularis Walliser, 1964; Nicoll and Rexroad, p. 42, pi. 5, fig. 17. 1969 Carniodus carnulus Walliser, 1964; Schonlaub, pi. 1, fig. 5. .1969 Pterospathodus amorphognathoides Walliser, 1964; Schonlaub, pi. I, fig. 8. .1969 1 Carniodus carinthiacus Walliser, 1964; Schonlaub, pi. 1, fig. 12. .1969 Ozarkodina gaertneri Walliser, 1964; Schdnlaub, pi. 1, fig. 15. 1969 Pterospathodus amorphognathoides Walliser, 1964; Drygant, p. 49, pi., fig. 6. .1969 Carniodusl carinthiacus Walliser, 1964; Drygant, p. 54, pi., fig. 5. ? 1969 Neoprioniodus subcarnus Walliser, 1964; Drygant, p. 53, pi., figs 12-14. .1970 Pterospathodus amorphognathoides Walliser, 1964; Manara and Vai, p. 494, pi. 62, fig. 15. .1970 Ozarkodina gaertneri Walliser, 1964; Manara and Vai, p. 487, pi. 62, fig. 17. .1971 Pterospathodus amorphognathoides Walliser, 1964; Schonlaub, p. 45, pi. 2, figs 6-12. p. 1971 Carniodus carinthiacus Walliser, 1964; Schonlaub, p. 46, pi. 3, figs 7-8 (non fig. 6 [= P. eopennatus ]). .1971 Neoprioniodus subcarnus Walliser, 1964; Rexroad and Nicoll, pi. 1, fig. 11. .1971 Pterospathodus amorphognathoides Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 20—21. .1971 Ozarkodina gaertneri Walliser, 1964; Rexroad and Nicoll, pi. 2, fig. 22. .1971 Ozarkodina neogaertneri Nicoll and Rexroad, 1968; Rexroad and Nicoll, pi. 2, fig. 23. .1972 Ozarkodina gaertneri Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 1-3. .1972 Pterospathodus amorphognathoides Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 4—7. .1972 Carniodus carnulus Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 8-11. .1972 Carniodus carnus Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 12-13. .1972 Carniodus carinthiacus Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 1-3. 1972 Carniodus carnicus Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 4, 5?. .1972 Neoprioniodus subcarnus Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 6-7. .1972 Neoprioniodus costatus Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 8-11. .1972 Neoprioniodus triangularis Walliser, 1964; Rexroad and Nicoll, pi. 2, figs 12-13. 1972 Exochognathus brevialatus (Walliser, 1964); Rexroad and Nicoll, pi. 2, figs 21?, 22. .1972 Ozarkodina neogaertneri Nicoll and Rexroad, 1968; Rexroad and Nicoll, pi. 2, fig. 34. p. 1972 Ozarkodina gaertneri Walliser, 1964; Aldridge, p. 200, pi. 5, fig. 5 (non fig. 7 [= P. amorphognathoides lennarti ssp. nov or P a. lithuanicus]). .1972 Carniodus carinthiacus Walliser, 1964; Aldridge, p. 168, pi. 5, figs 8-10. .1972 Carniodus carnicus Walliser, 1964; Aldridge, p. 168, pi. 5, fig. 11. .1972 Carniodus carnulus Walliser, 1964; Aldridge, p. 169, pi. 5, figs 12-14. .1972 Carniodus carnus Walliser, 1964; Aldridge, p. 169, pi. 5, figs 15-16. text-fig. 9. Pterospathodus amorphognathoides lennarti ssp. nov. A, C-D, G?, Pa element, morph a. E, Pa element, morph b. b, f, Pb, element, h-l, Pb2 element, n-o. Pc element, p-q, Mj element, r-t, x, Sc3 element, u, Sc2 element, v-w, modified carnuliform element, short morph, y, g1, Scx element, z, Sa element, a1, Sb, element, b1-^, curved element, morph a. d1, curved element, morph b. e^f1, P-k1, carniciform element. iP-P, modified carnuliform element. P-M1, v'-w1, carnuliform element, short morph. rP-iP, carnuliform element, morph a. Scale bar represents 1 mm. 1026 PALAEONTOLOGY, VOLUME 41 .1972 Neoprioniodus subcarnus Walliser, 1964; Aldridge, p. 195, pi. 5, fig. 17. p. 1972 Pterospathodus amorphognathoides Walliser, 1964; Aldridge, pi. 3, figs 17, 19 ( non fig. 18 [= P. a. lennarti ssp. nov.]). .1974 Pterospathodus amorphognathoides Walliser, 1964; Aldridge, fig. 1 E-F. .1975 Pterospathodus amorphognathoides Walliser, 1964; Aldridge, pi. 1, figs 22-23. .1975 Carniodus carnulus Walliser, 1964; Aldridge, pi. 1, figs 3-4, 8-9. .1975 Exochognathus latia/atus (Walliser, 1964); Aldridge, pi. 3, fig. 15. 1975 Neoprioniodus costatus costatus Walliser, 1964; Aldridge, pi. 3, fig. 17. 1975 Distomodus triangularis (Walliser, 1964); Aldridge, pi. 3, fig. 19. 1975 Pterospathodus amorphognathoides Walliser, 1964; Saladzius, pi. 2, figs 7, 8(?). 1975 Carniodus carinthiacus Walliser, 1964; Saladzius, pi. 1, fig. 5. 1975 Neoprioniodus costatus Walliser, 1964; Saladzius, pi. 1, fig. 9. 1975 Neoprioniodus subcarnus Walliser, 1964; Saladzius, pi. 1, fig. 11. 1975 Neoprioniodus triangularis Walliser, 1964; Saladzius, pi. 1, fig. 12. 1975 Neoprioniodus triangularis triangularis Walliser, 1964; Saladzius, pi. 1, fig. 13. .1976 Carniodus carnulus Walliser, 1964; Barrick and Klapper, p. 68, pi. 1, figs 1-2, 6-8, 12-14. . 1976 Pterospathodus amorphognathoides Walliser, 1964; Barrick and Klapper, p. 82, pi. 1, figs 4, 9-1 1, 16. 1976 Pterospathodus amorphognathoides Walliser, 1964; Kuwano, pi. 2, fig. 2. .1976 Pterospathodus amorphognathoides Walliser, 1964; Miller, fig. 822. 1976 Ozarkodina gaertneri Walliser, 1964; Miller, fig. 820. .1977 Pterospathodus amorphognathoides Walliser, 1964; Cooper, p. 1065, pi. 2, figs 3, 6. 1977 Neoprioniodus subcarnus Walliser, 1964; Liebe and Rexroad, pi. 1, fig. I. EXPLANATION OF PLATE 4 Figs 1-4. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 2. 1, Cn 7993; upper view of dextral Pa element. 2, Cn 7994; upper view of sinistral Pa element. 3, Cn 7995; outer lateral view of sinistral Pbj element. 4, Cn 7996; outer lateral view of dextral Pbj element. Figs 1 and 3^4 from Viki core, sample M-375, int. 1 37-95—1 38- 1 0 m; fig. 2 from Viki core, sample M-374, int. 1 38-95—1 39- 1 0 m. Figs 5-20, 22-27. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 1. 5, Cn 7997; outer lateral view of dextral Pb3 element. 6, Cn 7998; upper view of dextral Pa element. 7, Cn 7999; outer lateral view of dextral Pb., element. 8, Cn 8000; upper view of sinistral Pa element. 9, Cn 8001 ; outer lateral view of dextral Pb3 element. 10, Cn 8002; outer lateral view of sinistral Pbj element. 11, Cn 8003; outer lateral view of dextral Pb2 element. 12, Cn 8004; outer lateral view of dextral Pa element. 13, Cn 8005; outer lateral view of dextral carnuliform element, short morph. 14, Cn 8006; outer lateral view of sinistral curved element, morph a. 15, Cn 8007; inner lateral view of sinistral Sc3 element. 16, Cn 8008; inner lateral view of sinistral Sc2 element. 17, Cn 8009; inner lateral view of dextral Sc3 element. 1 8, Cn 8010; inner lateral view of sinistral Sc3 element. 19, Cn 801 1 ; lateral view of Sa element. 20, Cn 8012; inner lateral view of dextral carniciform element. 22, Cn 8013; outer lateral view of sinistral curved element, morph b. 23, Cn 8014; outer lateral view of dextral carnuliform element, morph a. 24, Cn 8015; inner lateral view of dextral Mj element. 25, Cn 8016; posterior view of dextral Sb2 element. 26, Cn 8017; outer lateral view of dextral Pc element. 27, Cn 8018; posterior view of dextral Sb, element. Figs 5-6 and 8-10 from Viki core, sample M-367, int. 1 46-70-1 46-80 m; figs 7, 11-17, 19-20 and 22-27 from Viki core, sample M-368, int. 145-40- 1 45-55 m; fig. 18 from Viki core, sample M-369, int. 144-45-144-50 m. Figs 21, 28-35. Pterospathodus amorphognathoides lithuanicus Brazauskas, 1983. 21, Cn8019; inner lateral view of dextral Pa element. 28, Cn 8020; outer lateral view of sinistral Pb2 element. 29, Cn 8021 ; upper view of dextral Pa element. 30, LO 7736t; upper view of dextral Pa element. 31, LO 7737t; upper view of sinistral Pa element, juvenile specimen. 32, Cn 8022; upper view of sinistral Pa element. 33, Cn 8023; outer lateral view of sinistral carnuliform element, morph a(7). 34, Cn 8024; outer lateral view of dextral Pa element. 35, Cn 8025; outer lateral view of sinistral Pa element. Figs 21 and 34-35 from Viki core, sample M-979, int. 148-75—1 48-85 m; figs 28 and 33 from Viki core, sample M-976, int. 149-95—1 50 08 m; figs 29 and 32 from Uulu-330 core, sample M-1302, int. 1 37-44—1 37-50 m; figs 30-31 from Sjalso section (Gotland), sample G88- 637LJ. All x 50. PLATE 4 MANNIK, Pterospathodus PALAEONTOLOGY, VOLUME 41 text-fig. 10. For caption see opposite. MANNIK: SILURIAN CONODONTS 1029 .1977 Carniodus carinthiacus Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 2. .1977 Carniodus cumulus Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 3. .1977 Carniodus carnicus Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 4. .1977 Pterospathodus amorphognathoides Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 9. .1977 Ozarkodina gaertneri Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 10. .1977 Exochognathus latialatus (Walliser, 1964); Liebe and Rexroad, pi. 1, fig. 16. .1977 Distomodus triangularis (Walliser, 1964); Liebe and Rexroad, pi. 2, fig. 29. .1977 Neoprioniodus costatus Walliser, 1964; Liebe and Rexroad, pi. 2, figs 30-31. 1977 Exochognathus brevialatus (Walliser, 1964); Liebe and Rexroad, pi. 2, fig. 38. 1978 Neoprioniodus costatus costatus Walliser, 1964; Miller, pi. 2, figs 10-1 1 .1978 Apparatus ‘C’ Walliser, 1964; Miller, pi. 4, figs 8-11. .1980 Carniodus cumulus Walliser, 1964; Helfrich, pi. 1, figs 1-6. 1980 Pterospathodus amorphognathoides Walliser, 1964; Helfrich, pi. 2, figs 17-19. 1980 Pterospathodus celloni (Walliser, 1964); Helfrich, pi. 2, fig. 30. .1981 Pterospathodus amorphognathoides Walliser, 1964; Nowlan, pi. 7, fig. 6. .1981 Pterospathodus amorphognathoides Walliser, 1964; Uyeno and Barnes, pi. 1, fig. 24. .1982 Pterospathodus amorphognathoides Walliser, 1964; Aldridge and Mohamed, pi. 2, figs 13-16. .1982 Carniodus carnulus Walliser, 1964; Aldridge and Mohamed. pi. 2, figs 17-24. .1983 Carniodus carnulus Walliser, 1964; Mabillard and Aldridge, pi. 2, figs 13-14. .1983 Pterospathodus amorphognathoides Walliser, 1964; Mabillard and Aldridge, pi. 2, figs 25-27 . .1983 Pterospathodus amorphognathoides Walliser, 1964; Nowlan, fig. 4K [cop. Nowlan 1981, pi. 7, fig. 6], .1983 Pterospathodus amorphognathoides Walliser, 1964; Uyeno and Barnes, p. 24, pi. 8, fig. 24 [cop. Uyeno and Barnes 1981, pi. 1, fig. 24], .1983 Pterospathodus amorphognathoides Walliser, 1964; Barrick, fig. 18 M. .1984 Carniodus ? carinthiacus Walliser, 1964; Drygant, p. 83, pi. 3, figs 8-11. 1984 Neoprioniodus subcarnus Walliser, 1964; Drygant, p. 84, pi. 3, figs 14-17. .1984 Pterospathodus amorphognathoides Walliser, 1964; Drygant, p. 109, pi. 7, figs 13-16. 1984 Ozarkodina gaertneri Walliser, 1964; Drygant, p. 1 10, pi. 7, figs 22-27 . (?) 1984 Carniodus carnicus Walliser, 1964; Drygant, p. 82, pi. 3, figs 12-13. v. 1985 Pterospathodus amorphognathoides Walliser, 1964; Nehring-Lefeld, p. 635, pi. 1, figs 3-8. 1985 Carniodus carnulus Walliser, 1964; Kleffner, pi. 2, figs 26-28. 1985 Pterospathodus amorphognathoides Walliser, 1964; Kleffner, pi. 1, fig. 3; pi. 2, figs 29-31. v. 1985 Carniodus carnulus Walliser, 1964; Nehring-Lefeld, p. 632, pi. 2, figs 1—10. (?) 1985 Pterospathodus amorphognathoides Walliser, 1964; Yu, p. 24, pi. 2, fig. 9. 1986 Pterospathodus amorphognathoides Walliser, 1964; Jiang et al., pi. 4, figs 1-2. p. 1987 Pterospathodus amorphognathoides Walliser, 1964; Over and Chatterton, pi. 4, figs 1-2 ( non fig. 3 [= P. rhodesil]). .1987 Carniodus carnulus Walliser, 1964; Kleffner, fig. 57-7. .1987 Pterospathodus amorphognathoides Walliser, 1964; Kleffner, fig. 55-9, 11. .1987 Pterospathodus pennatus procerus (Walliser, 1964); Kleffner, fig. 510. .1987 Pterospathodus amorphognathoides Walliser, 1964; An, p. 201, pi. 33, figs 1-3. p. 1987 Pterospathodus pennatus procerus (Walliser, 1964); An, p. 202, pi. 33, figs 4, 7 (non figs 5-6 [ = P. p. procerus]). p. 1987 Exochognathus brassfieldensis (Branson and Branson, 1947); An, pi. 35, fig. 18 (non fig. 17 [indet.]). 1988 Pterospathodus amorphognathoides Walliser, 1964; Qiu, pi. 1, fig. 8 [cop. Qiu 1985, pi. 1, fig. 3]. .1989 Pterospathodus amorphognathoides Walliser, 1964; Mannik and Aldridge, text-fig. 1 G-L. v. 1989 Pterospathodus amorphognathoides Walliser, 1964; Mannik and Aldridge, text-fig. 3 D-E. text-fig. 10. Pterospathodus amorphognathoides lithuanicus Brazauskas, 1983. a-b, d-e, Pa element, c, F, o, u, Pb1 element, g-l, p, Pb2 element, m-n, Sc2 element. Q, v. Pc element, r-s, Mt element. T, Sc2 element. w-Y, carnuliform element, short morph. z-D1, Sc3 element. E1— G1, Sbt element. hP-l1, carnuliform element, morph a. m1, modified carnuliform element, short morph, n1, modified carnuliform element!?), o1, modified carnuliform element, p1, Sb2 element. Q1, Sa element. Rx-s\ carniciform element, t'-u1, carnuliform element, morph b. v'-w1, curved element, morph b. x1, curved element, morph a. yx-z\ curved element, morph c. Scale bar represents 1 mm. 1030 PALAEONTOLOGY, VOLUME 41 p. 1990 Pterospathodus pennatus procerus (Walliser, 1964); Uyeno, p. 66, pi. 3, fig. 18 ( non figs 19-20 [ = P. p. procerus]). v. 1990 Pterospathodus amorphognathoides Walliser, 1964; Mannik and Viira, pi. 17, figs 28, 32. .1991 Carniodus carnulus Walliser, 1964; Klefifner, fig. 5 27-28. .1991 Pterospathodus amorphognathoides Walliser, 1964; Kleffner, fig. 621, 26-27. .1992 Pterospathodus amorphognathoides Walliser, 1964; Barca et al., pi. 10, figs 7-10. .1992 Carniodus carnulus Walliser, 1964; Barca et al., pi. 10, figs 11-12. .1996 Pterospathodus amorphognathoides Walliser, 1964; Wang and Aldridge, pi. 5, fig. 9. v. 1998 Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Mannik and Malkowski, pi. 1, figs 10, 14-17. Material. Several hundreds to thousands of all elements. Emended diagnosis. P. amorphognathoides with distinct basal platform/platform ledges of various configurations and dimensions on all elements. Remarks. The size and shape of the basal platform of the Pa element is highly variable, evidently due to evolutionary changes (see below). The Pa element may or may not possess a triangular lateral lobe/short usually undenticulated process on the outer side of the element. Occasionally, additional EXPLANATION OF PLATE 5 Figs 1-4, 10, 16. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 5. 1, Cn 8026; upper view of dextral Pa element. 2, Cn 8027; upper view of dextral Pa element. 3, LO 7738t; upper view of simstral Pa element. 4, Cn 8028; outer lateral view of dextral carnuliform element, short morph. 10, Cn 8029; outer lateral view of dextral modified carnuliform element. 16, Cn 8040; inner lateral view of sinistral Sc3 element. Figs 1-2 from Viki core, sample M-391, int. 115-45-1 1 5-60 m; fig. 3 from Overstekvarn 2 section (Gotland), sample G88-635LJ ; figs 4 and 10 from Viki core, sample M-995, int. 1 13-80-1 13-95 m; fig. 16 from Viki core, sample M-390, int. 1 1 8-40— 1 18-50 m. Figs 5-9, 11-15, 17-23, 35. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 4. 5, Cn 8030; outer lateral view of sinistral Pb2 element. 6, Cn 8031; inner lateral view of sinistral Sc2 element. 7, Cn 8032; outer lateral view of dextral Pb, element. 8, Cn 8033; inner lateral view of dextral carniciform element. 9, Cn 8034; outer lateral view of dextral modified carnuliform element, short morph. 11, Cn 8035; inner lateral view of sinistral Sct element. 12, Cn 8036; upper view of dextral Pa element. 13, Cn 8037; upper view of sinistral Pa element. 14, Cn 8038; inner lateral view of dextral Sc3 element. 15, Cn 8039; outer lateral view of sinistral carnuliform element, morph a. 17, Cn 8041 ; outer lateral view of sinistral modified carnuliform element. 18, Cn 8042; lateral view of symmetrical carnuliform element, morph a. 19, Cn 8043; posterior view of sinistral Sb2 element. 20, Cn 8044; posterior view of dextral Sb2 element. 21, Cn 8045; outer lateral view of dextral curved element, morph b. 22, Cn 8046; posterior view of dextral Sbt element. 23, Cn 8047; outer lateral view of sinistral Pc element. 35, Cn 8048; outer lateral view of sinistral modified carnuliform element, short morph. Figs 5, 8-9, 1 1, 14-15, 17-20 and 22 from Viki core, sample M-386, int. 1 23-25— 1 23-45 m; fig. 6 from Viki core, sample M-385, int. 124-60—124-75 m; tigs 7, 12-13, 21, 23 and 35 from Viki core, sample M-384, int. 1 25-60—125-75 m. Figs 24-34, 36-40. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 3. 24. Cn 8049; upper view of sinistral Pa element. 25, Cn 8050; inner lateral view of sinistral SCj element. 26, Cn 8051 ; outer lateral view of sinistral Pbj element. 27, Cn 8052; upper view of sinistral Pa element. 28, Cn 8053; outer lateral view of dextral curved element, morph a. 29, Cn 8054; inner lateral view of dextral SCj element. 30, Cn 8055; inner lateral view of sinistral element. 31, Cn 8056; outer lateral view of sinistral Pb2 element. 32, Cn 8057; upper view of dextral Pa element. 33, Cn 8058; inner lateral view of dextral Sc2 element. 34, Cn 8059; outer lateral view of dextral carnuliform element, morph a. 36, Cn 8060; outer lateral view of sinistral modified carnuliform element. 37, Cn 8061 ; outer lateral view of dextral Pb2 element. 38, Cn 8062; inner lateral view of sinistral Sc3 element. 39, Cn 8063; outer lateral view of sinistral carnuliform element, short morph. 40, Cn 8064; outer lateral view of dextral Sb2 element. Figs 24, 26, 28, 31-33, 36 and 38-39 from Viki core, sample M-378, int. 1 34-80—1 34-90 m; figs 25 and 29 from Viki core, sample M-380, int. 131-85-132-00 m; figs 27, 30, 34, 37 and 40 from Viki core, sample M-381, int. 1 30-45—1 30-55 m. All x 50. PLATE 5 MANNIK, Pterospathodus 1032 PALAEONTOLOGY, VOLUME 41 text-fig. 11. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 1. A-c, f, Pa element, d-e, g, j-k, q, Pbx element, h-i, n-p, Pb2 element, l-m, Pc element, r-t, f1, Sc3 element, u-v, Mj element, w, Sb2 element, x, Sbj element. Y, a1, h1, carniciform element, z, Sa element. B1, modified carnuliform MANNIK: SILURIAN CONODONTS 1033 lateral denticle(s) may occur at the distal part of the process. Although the fauna is dominated by elements with a bifurcated lateral process, pennate elements may be found quite often. These are easily separated from pennate elements of P. p. procerus on the basis of the configuration of the basal platform and cavity. Based on changes in the morphology of the Pa element, five main temporal populations can be recognized in P. a. amorphognathoides (Pis 4—5; Text-figs 3, 11-15). Population 1 (PI. 4, figs 5-20, 22-27 ; Text-fig. 1 1). The Pa element possesses a narrow but distinct platform (PI. 4, figs 6, 8; Text-fig. 11a-c, f). This population also includes rare specimens almost indistinguishable from the elements of P. a. lithuanicus. However, in this interval the Pa elements similar to those of P. a. lithuanicus possess a short narrow ledge on the outer side of the distal part of the posterior process (PI. 4, fig. 12). This structure is missing (Pi. 4, figs 29-32, 35), or is very rare (PI. 4, fig. 34), on the Pa elements of P. a. lithuanicus. It is possible that the presence of such Pa elements in Population 1 is evidence that this population forms an evolutionary link between P. a. lithuanicus and P. a. amorphognathoides. Population 1 is included in P. a. amorphognathoides because the subspecies boundary is drawn at the appearance of the new character (platform), not at the point where it was found in all individuals. It is also evident that the initial appearance of platform ledges on the Pa element started on the outer side of the distal part of the posterior process. Characteristic for Population 1 are at least three morphs of Pbx elements. 1. Relatively large and long elements without platform ledges (PI. 4, figs 5, 10; Text-fig. 11j-k), resembling those of P. a. lithuanicus (compare PI. 3, figs 15, 17; Text-fig. 10c, F, o, u). 2. Small straight elements with a distinct high cusp and narrow platform-ledges (PI. 4, fig. 9; Text-fig. 1 1d-e). 3. Relatively large straight elements with a short cusp, high denticles on the long anterior process and low denticles on the shorter posterior process (Text-fig. lie). This morph is almost identical to the holotype of Walliser’s ‘ Ozarkodina gaertneri' (Walliser 1964, pi. 27, fig. 14). In the collections studied, this type of element is extremely rare, occurring only in few samples and represented, as a rule, by one or two specimens. Rare elements of this type can be found also in younger populations (Text-fig. 13v). Morphs 1 and 2 dominate in Population 1, the former being more abundant in older strata and morph 2 in younger strata. Population 2 (PI. 4, figs 1-4; Text-fig. 12). The morphologically highly variable platform is widest on the outer proximal side of the Pa element and narrows gradually distally. Population 3 (PI. 5, figs 24-40; Text-fig. 13). The platform on the Pa elements reaches its maximum size in this population. The edges of it, particularly on the posterior process, are strongly undulating and partly turned up. Many specimens possess a triangular or semiquadrate outer lateral lobe/short process which may or may not bear denticle(s) (PI. 5, fig. 27; Text-fig. 12a, c). In Population 3 platform ledges become well developed on most of the elements of the apparatus. Population 4 (PI. 5, figs 5-9, 1 1-15, 17-23 ; Text-fig. 14). On typical Pa elements of this population the platform is wide on the posterior process and becomes rapidly narrow on the anterior process forming a distinct ‘bulge’ on the outer side of the element, anterior of the point where the inner bifurcated process joins the main blade. Just behind this ‘bulge’ the platform is widest and possesses an upturned, undulating edge. The structure described above is distinct on the dextral element (PI. 5, fig. 12; Text-fig. 14a, c) but less well developed on the sinistral one (PI. 5, fig. 13; Text-fig. 14b). Population 5 (PI. 5, figs l^t, 10, 16; Text-fig. 15). Most characteristic for this population is a Pa element with the platform widest proximally on the outer side of the element and narrowing equally towards the posterior and anterior ends. Many elements also possess an additional denticle between the main row of denticles and that on the bifurcating lateral process (Text-fig. 15e, g). Very characteristic of this, the youngest population of P. a. amorphognathoides, is the strongly arched (in lateral view) modified carnuliform element (PI. 5, fig. 10; Text-fig. lScd-p1)- Also, in some sections, a few Pa elements with an extremely wide platform (Text-fig. 15c) were found in the uppermost part of the range of Population 5. element. c'-D1, Sct element, e1, i\ Sc2 element. G1, carnuliform element, short morph. P-N1, carnuliform element, morph a. o1, carnuliform element, morph b(?). p1, curved element, morph c. q1, curved element, morph a. R1, curved element, morph b. Scale bar represents 1 mm. 1034 PALAEONTOLOGY, VOLUME 41 text-fig. 12. For caption see opposite. MANNIK: SILURIAN CONODONTS 1035 The elements of P. a. amorphognathoides in Population 3 are quite similar to those of P. rhodesi (see below), differing from them mainly by the less well developed platform/platform ledges. P. a. amorphognathoides is abundant and dominates conodont faunas in the open shelf carbonate- terrigeneous facies. Towards the basin it becomes rare and is ecologically replaced by P. pennatus procerus. The populations listed above will in the future probably allow the recognition of several stratigraphically useful subdivisions in the P. a. amorphognathoides Zone. However, further studies are needed. Occurrence. P. a. amorphognathoides , Lower Pseudooneotodus bicornis and Upper Ps. bicornis zones. P. a. amorphognathoides has been recognized world-wide (see synonymy) except for a few regions: Severnaya Zemlya (Mannik 1983) and the Sub-Polar Urals (Melnikov, pers. comm.). It is extremely rare in eastern Canada (Gaspe Peninsula; Nowlan 1983). P. a. amorphognathoides is evidently also missing in several other regions (e.g. Greenland - Armstrong 1990; Alaska - Savage 1985; some regions in north-western Canada - McCracken 1991 ; and Australia - Bischoff 1986) where it is replaced by P. rhodesi. P. pennatus lineage As was noted above the P. pennatus lineage probably appeared, together with the P. amorphognathoides lineage, at the end of the P. eopennatus Zone. Both lineages evidently originated from the same ancestral taxon: P. eopennatus ssp. nov. 2 (Text-fig. 3; Mannik 1995). However, some data suggest also another possibility. Morph 5 of P. eopennatus ssp. nov. 1 from the A. irregularis- A. kuehni subzones is morphologically very similar to P. p. pennatus of Walliser (1964, pi. 14, figs 23-26). In the Cellon section P. p. pennatus is found together with P. celloni in strata considerably younger than the known range of morph 5 in Estonia. Also, the morphologies of elements and co-occurrences of taxa in Cellon indicate that the oldest P. celloni fauna described from that section is no older than the latest P. eopennatus , but most probably comes from the earliest P. celloni chron (Mannik 1996). In Estonia, in open shelf environments morph 5 disappeared during the end -irregularis event. That event caused considerable changes in conodont faunas. Several taxa became extinct or disappeared temporarily and P. eopennatus ssp. nov. 1 was replaced by P. eopennatus ssp. nov. 2. However, it cannot be excluded that P. eopennatus ssp. nov. 1 survived the end -irregularis event somewhere in offshore regions and gave rise to the P. p. pennatus - P. p. procerus lineage. The morphological similarities between the Pa element of P. p. pennatus and morph 5 of P. eopennatus ssp. nov. 1 suggest a possibility that these two taxa are directly connected. It is possible that the two ecologically restricted lineages might have appeared already at the end of the A. irregularis chron. However, no data about the deeper basin lineage is yet available from the interval between the ranges of P. eopennatus ssp. nov. 1 and P. p. pennatus. Also, P. p. pennatus itself has not been found in Estonia. Therefore, in this paper P. pennatus is described as a descendant of P. eopennatus appearing in the sequence at the same time as P. amorphognathoides. Pterospathodus pennatus pennatus (Walliser, 1964) *1964 Spathognathodus pennatus pennatus Walliser, p. 79, pi. 14, figs 23-26; pi. 15, fig. 1. (?)1968 Neospathognathodus pennatus (Walliser, 1964); Nicoll and Rexroad, p. 47, pi. 2, fig. 5. Remarks. P. p. pennatus has not been identified in Estonia. In Cellon P. p. pennatus has the same text-fig. 12. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 2. a-g, Pa element, h-k, n-q, Pb2 element, l-m, u-v, Pbj element. R, Y, Pc element. s-T, Mj element, w-x, g'-h1, Sc3 element, z, Sb2 element, a1, SbT element, b1, Sa element. cP-D1, Sc2 element. e\ SCj element. F1, carniciform element. P-J1, carnuliform element, short morph. iP-l1, carnuliform element morph a. M1— N1, modified carnuliform element, o1, ?curved element, morph c. p1, curved element, morph a. q1, curved element, morph b. Scale bar represents 1 mm. 1036 PALAEONTOLOGY, VOLUME 41 text-fig. 13. For caption see opposite. MANNIK: SILURIAN CONODONTS 1037 range as P. celloni (Walliser 1964). In this section P. p. pennatus is very rare in the lowermost sample studied (10 B) which is dominated by elements of P. a. angulatus. In sample 10 H/J, I have identified a fragment probably belonging to P. amorphognathoides lennarti ssp. nov. together with P. p. pennatus and P. celloni. In the uppermost sample with P. p. pennatus (sample 10 J), a few Pa and Pb elements (Walliser 1964, pi. 15, fig. 1; pi. 27, fig. 5) occur which are morphologically almost identical to those of P. p. procerus, indicating a close relationship between these two subspecies. The morphologically distinct Pb^ Pb2, Pc, Scx, Sc2 and Sb2 elements, which occur in Estonia with the Pa elements of P. celloni, are considered to belong to the apparatus of the latter species (see below). However, ‘ Neoprioniodus triangularis tenuirameus' (Walliser 1964, pi. 28, figs 22-24; = M, element; fig. 21 cannot be identified without direct study of the specimen) and ‘ Carniodus eocarnicus ’ (Walliser 1964, pi. 28, figs 19-20; = SCj element) may belong either to P. celloni or P. p. pennatus - they all are found together in the same strata. Here these elements are tentatively assigned to P. celloni (see synonymy). The specimen illustrated by Walliser (1964, pi. 28, fig. 12) as "N. subcarnus' is considered to be the Sc2 element of P. celloni. The majority of the other Carniodus- elements illustrated by Walliser (1964) possess distinct platform ledges and evidently belong to P. a. amorphognathoides (see above). Occurrence. P. celloni Zone. v. 1964 1966 p. 1968 p. 1968 1968 .1968 p. 1968 .1968 .1968 p. 1968 1969 .1976 v. 1979 .1983 1984 1984 1985 v. 1985 1985 vp. 1986 .1987 Pterospathodus pennatus procerus (Walliser, 1964) Plate 6, figures 1-25, 27-35; Text-figure 16 Spathognathodus pennatus procerus Walliser, p. 80, pi. 15, figs 2-8. Spathognathodus pennatus procerus Walliser, 1964; Spasov and Filipovic, p. 50, pi. 1, fig. 6. Spathognathodus pennatus procerus Walliser, 1964; Igo and Koike, pp. 18-19, pi. 2, figs 8-10 (non fig. 1 1 [indet.]). Ozarkodina gaertneri Walliser, 1964; Igo and Koike, p. 14, pi. 1, figs 7-9 (non figs 5-6 [= P. a. amorphognathoides ]). Neoprioniodus costatus paucidentatus Walliser, 1964; Igo and Koike, p. 12, pi. 3, figs 16-17. Neoprioniodus triangularis tenuirameus Walliser, 1964; Igo and Koike, p. 13, pi. 3, figs 18-19. Carniodus sp. Algo and Koike, p. 8, pi. 3, fig. 2 (non fig. 3 [= P. a. amorphognathoides ]). Carniodus sp. B Igo and Koike, p. 8, pi. 3, fig. 20. Roundval sp. C Igo and Koike, p. 17, pi. 3, figs 25-28. Neoprioniodus spp. Igo and Koike, p. 14, pi. 3, fig. 24 (non fig. 4 [= P. a. amorphognathoides ]). Spathognathodus pennatus procerus Walliser, 1964; Drygant, p. 50, pi., figs 2-3. Pterospathodus pennatus pennatus (Walliser, 1964); Barrick and Klapper, p. 86, pi. 1, fig. 19. Pterospathodus pennatus procerus (Walliser, 1964); Jeppsson, p. 235, fig. 1X1-8. Pterospathodus pennatus procerus (Walliser, 1964); Savage et al., fig. 2 A-F. Pterospathodus pennatus procerus (Walliser, 1964); Stouge and Bagnoli Stouge, p. 109, pi. 2, figs 14-17. Pterospathodus pennatus procerus (Walliser, 1964); Drygant, p. 108, pi. 7, figs 17-20. Pterospathodus pennatus procerus (Walliser, 1984); Yu, pi. 1, figs 1-2. Pterospathodus pennatus pennatus (Walliser, 1964); Nehring-Lefeld, p. 637, pi. 1, figs 1-2. Pterospathodus pennatus procerus (Walliser, 1974); Savage, p. 714, fig. 4 A-K. Pterospathodus procerus (Walliser, 1964); Bischoff, p. 204, pi. 29, figs 9-10, 15-30 (non figs 13-14 [indet.]); pi. 30, figs 1-2 (non figs 3-11 [= P. rhodesi ]). Pterospathodus pennatus procerus (Walliser, 1964); Over and Chatterton, pi. 4, fig. 4. text-fig. 13. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 3. a-d, f, j. Pa element, e, i, n-o, v, Pbj element, g-h, k-m, Pb2 element, p-q. Pc element. R, w, SCj element, s, z, element. T-u, Y, Sc3 element, x, Sa element, a1, Sc2 element, b1, l1, carnuliform element, short morph. cWd1, m1, modified carnuliform element. E1, Sb1 element, f1, Sb2 element. G1, modified carnuliform element, short morph. h\ curved element, morph a. I1, curved element, morph b. J1, carniciform element, k1, modified carnuliform element, short morph. N1, carnuliform element, morph b. o1, carnuliform element, morph a. Scale bar represents 1 mm. 1038 PALAEONTOLOGY, VOLUME 41 text-fig. 14. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 4. a-d. Pa element. E, G, I-J, Q-s, Pb2 element. F, H, K, Pbt element. L, z, element, m-n, SCj element, o-p, Sc2 element. T, Sbj element, u-v, Sb2 element, w, Y, Pc element, x, carniciform element, a^b1, e1, Sc3 element. C^-D1, carnuliform element, short morph. F1 i1 , carnuliform element, morph a. j1, carnuliform element, morph b(?). k1, m1, curved element, morph b. L1, curved element, morph a. isP-o1, modified carnuliform element. Scale bar represents 1 mm. MANNIK: SILURIAN CONODONTS 1039 text-fig. 15. For caption see opposite. 1040 PALAEONTOLOGY, VOLUME 41 p. 1987 Pterospathodus pennatus procerus (Walliser, 1964); An, p. 202, pi. 33, figs 5-6 ( non figs 4, 7 [ = P. a. amorphognathoides ]). v. 1990 Pterospathodus procerus (Walliser, 1964); Mannik and Viira, pi. 17, figs. 29. p. 1990 Pterospathodus pennatus procerus (Walliser, 1964); Uyeno, p. 66, pi. 3, figs 19-20 ( non fig. 18 [ = P. a. amorphognathoides ]). .1991 Pterospathodus procerus (Walliser, 1964); McCracken, p. 109, pi. 4, figs 12-23. p? 1991 Carniodus carnulus Walliser, 1964; McCracken, p. 108, pi. 3, figs 13-14 ( non figs 6-12, 15 [= P. rhodesi ]). .1992 Pterospathodus pennatus procerus (Walliser, 1964); Nehring-Lefeld, pi. 3, figs 1-2. v. in press Pterospathodus pennatus procerus (Walliser, 1964); Mannik and Malkowski, pi. 1, figs 6-7, 11-13, 18. Material. Many tens to about a hundred of the Pa and Pbx elements; few to a few tens of all other elements. Remarks. The apparatus of P. p. procerus is well represented in several samples in the studied collections. Pa, Pbj, Pb,, Pc, Mls M2, Sc1? Sc2, Sc3, Sb1? Sb2, Sa, carnuliform morphs a and b, and a possible carniciform element are recognized. The M2 element (PI. 6, figs 1,3; Text-fig. 16w-z) has so far been found only in the P. p. procerus apparatus. The Sc2 element seems to be represented by two morphs, one without and the other with denticles on the basal part of the anterior edge of the cusp (PI. 6, figs 9, 22 and 7, 25 respectively; Text-fig. lbp1-^ and Rl-s\ v^w1 respectively). Probable carniciform element (Text-fig. lbx^A2, n2-o2) of this apparatus possesses a considerably taller cusp than its possible homologues in other apparatuses of Pterospathodus (Text-figs 4-15). The data available allow the recognition of ecological replacement of P. a. amorphognathoides by P. p. procerus towards offshore environments (Mannik 1992). In open shelf environments P. p. procerus is extremely rare or completely absent in the P. a. amorphognathoides Zone, although in many regions (Estonia - Mannik 1992; Gotland - Jeppsson 1979, Jeppsson and Mannik 1993; Britain - Mannik and Aldridge 1989, Aldridge et al. 1993) P. p. procerus has been identified from a short interval above the last P. a. amorphognathoides. This appearance of P. p. procerus in open shelf environments was probably connected with the Ireviken Event, with the extinction of P. a. amorphognathoides creating a vacant niche. Occurrence. P. a. amorphognathoides to Upper P. p. procerus zones (Jeppsson 1994, 1997) in deeper basin environments ; Lower and Upper P. p. procerus zones in open shelf facies. P. p. procerus has been found in most known sequences world-wide (see synonymy). v.* 1964 v. 1964 vp. 1964 vp. 1964 v. 1964 v. 1964 v.? 1964 ? 1968 ? 1968 1971 1971 Pterospathodus celloni (Walliser, 1964) Plate 6, figures 26, 36-54; Text-figure 17 Spathognathodus celloni Walliser, p. 73, pi. 14, figs 3-16. Ozarkodina adiutricis Walliser, p. 54, pi. 27, figs 1-10. Carniodus eocarnicus Walliser, p. 34, pi. 28, fig. 20 ( non fig. 19 [= P. amorphognathoides]). Neoprioniodus subcarnus Walliser, p. 51, pi. 28, fig. 12 ( non figs 13-14 [= P. a. amorphognathoides ]). Neoprioniodus triangularis tenuirameus Walliser, p. 53, pi. 28, figs 21-24. Neoprioniodus costatus paucidentatus Walliser, p. 48, pi. 28, figs 31-35. Roundya brevialata Walliser, p. 69, pi. 31, figs 8—10. Ozarkodina adiutricis Walliser, 1964; Nicoll and Rexroad, p. 48, pi. 2, fig. 8. Neospathognathodus celloni (Walliser, 1964); Nicoll and Rexroad, p. 45, pi. 2, figs 1^4. Neospathognathodus celloni (Walliser, 1964); Rexroad and Nicoll, pi. 1, figs 2—4. Ozarkodina adiutricis Walliser, 1964; Rexroad and Nicoll, pi. 1, fig. 5. text-fig. 15. Pterospathodus amorphognathoides amorphognathoides Walliser, 1964; Population 5. A— J, Pa element, k-m, p-r, Pb2 element, n-o, Pbj element. s-T, Mt element, u-v, Sc2 element, w, E1, Sc3 element, x, h1, SCj element, b1, carniciform element, c1, Sa element. D1, Sb2 element. F1, carnuliform element, short morph. G1, n1, curved element, morph a. P-j1, carnuliform element, morph b. K1— L1, carnuliform element, morph a. M1, curved element, morph b(7). cfi-p1, modified carnuliform element. Scale bar represents 1 mm. MANNIK: SILURIAN CONODONTS 1041 1972 Ozarkodina adiutricis Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 15-16. 1972 Spathognathoides celloni Walliser, 1964; Rexroad and Nicoll, pi. 1, figs 17-19. 1976 Pterospathodus celloni (Walliser, 1964); Barrick and Klapper, p. 82, pi. 1, figs 3, 5. 1977 Ozarkodina adiutricis Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 11. 1977 Spathognathodus celloni Walliser, 1964; Liebe and Rexroad, pi. 1, fig. 12. 1989 Pterospathodus celloni (Walliser, 1964); Mannik and Aldridge, text-fig. 1.4 -F. ? 1994 Pterospathodus celloni (Walliser, 1964); Watkins et a/., pi. 10, figs 1-4. Material. Many tens of Pa and Pb elements; a few Pc, SCj and Sc2 elements. Remarks. P. celloni is very rare in the Estonian collections, which mainly represent a proximal carbonate-terrigeneous facies. It is mostly represented by Pa and Pbj elements, with the Pb2 element quite common. However, the identification of the Pb2 element among co-occurring juvenile Pb2 elements of P. amorphognathoides lennarti ssp. nov. and P. a. lithuanicus is quite problematical. Of the other elements of P. celloni , only the Scx, Sc2 and extremely rare specimens of the Sb2 have been identified. Sc15 Sc2 and Sb2 elements of P. celloni are morphologically almost identical to their homologues in P. eopennatus apparatuses (PI. 1, figs 3, 6, 13, 44—45 ; Text-figs 4f2, h2-k2, Sa'-b1, h1— j1 ; bL1-?1, u'-v1). The Sb9 is also almost identical to its homologue in P. p. procerus (PI. 6, figs 10-11, 17-19, 27; Text-fig". 16m1). Some peculiar carniodiform elements occur together with P. celloni. They are here identified as carniciform(?) (PI. 6, fig. 50; Text-fig. 17L-N1) and carnuliform (PI. 6, figs 42, 51-52, 54; Text-fig. Hcd-i2) elements of P. celloni apparatus. It is also probable that at least some of the elements described by Walliser (1964) as ‘A. triangularis tenuirameus' and ‘A. costatus paucidentatus' represent, accordingly, the Mj and Pc elements of P. celloni (see synonymy). P. celloni seems to have been more abundant in deeper basin environments (graptolite-bearing facies) and was very rare in open shelf regions. Although its origin needs further investigation it is evident that P. celloni was closely related to the P. pennatus lineage. Occurrence. From the uppermost part of the P. a. angulatus Subzone to the P. a. lithuanicus Subzone in open shelf facies. The range in deeper facies needs further studies but may be longer. The distribution of P. celloni in other regions needs further studies, with revision of collections. However, based on the published data (see synonymy) it seems quite probable that P. celloni can be recognized in most known Telychian sequences. Pterospathodus rhodesi (Savage, 1985) .1984 Pterospathodus n. sp. A Stouge and Bagnoli Stouge, p. 109, pi. 1, figs 1-6. .1984 Carniodus carnulus Walliser, 1964; Stouge and Bagnoli Stouge, p. 108, pi. 1, figs 11-19. .* 1985 Pterospathodus amorphognathoides rhodesi Savage, p. 714, fig. 3 A-T. .1985 Carniodus carnulus Walliser, 1964; Savage, p. 714, fig. 2D-N. 1985 Xainzadontus dewukaxiaensis Yu, p. 25, pi. 1, fig. 14. 1985 Ozarkodina gaertneri Walliser, 1964; Yu, pi. 1, fig. 8. 1985 Roundya triangularis Yu, p. 25, pi. 1, fig. 13. v. 1986 Carniodus carnulus Walliser, 1964; Bischoff, p. 177, pi. 5, figs 18-34; pi. 6, figs 1-37. v. 1986 Pterospathodus procerus (Walliser, 1964); Bischoff, p. 204, pi. 30, figs 3-11. v. 1986 Pterospathodus latus Bischoff, p. 197, pi. 30, figs 15-18, 31; pi. 31, figs 1-14. v 1986 Pterospathodus amorphognathoides Walliser, 1964; Bischoff, p. 186, pi. 30, figs 19-22; pi. 31, figs 15-39. .1987 Pterospathodus pennatus rhodesi (Savage, 1985); Over and Chatterton, p. 21, pi. 4, figs 5-6. p. 1987 Pterospathodus amorphognathoides Walliser, 1964; Over and Chatterton, pi. 4, fig. 3 ( non figs 1-2 [ = P. a. amorphognathoides ]). .1989 Pterospathodus rhodesi (Savage, 1985); Mannik and Aldridge, text-fig. 4 A-B. .1990 Carniodus carnulus Walliser, 1964; Armstrong, p. 68, pi. 4, figs 1 4 — 29. .1990 Pterospathodus amorphognathoides Walliser, 1964; Armstrong, p. 115, pi. 19, figs 1-5. 1042 PALAEONTOLOGY, VOLUME 41 .1990 Pterospathodus pennatus rhodesi (Savage, 1985); Armstrong, p. 120, pi. 20, figs 6-16. p. 1991 Carniodus carnulus Walliser, 1964; McCracken, p. 108, pi. 3, figs 6-12, 15 (non figs 13-14 [ = P. p. procerus]). .1991 Pterospathodus rhodesi (Savage, 1985); McCracken, p. 109, pi. 5, figs 6-15. Remarks. In some regions (Australia, Alaska, Greenland, Tibet). P. a. amorphognathoides is replaced by, or co-occurs with, P. rhodesi , all elements of which are characterized by extremely wide platform/platform ledges (see illustrations listed in synonymy). The origin of P. rhodesi is not known. In general, the morphology of the elements and the architecture of the apparatus suggest that P. rhodesi is closely related to the P. amorphognathoides lineage. However, the co-occurrence of P. a. amorphognathoides and P. rhodesi in some sections (e.g. in the southern Mackenzie EXPLANATION OF PLATE 6 Figs 1-25, 27-35. Pterospathodus pennatus procerus (Walliser, 1964). 1, LO 7739t; inner lateral view of dextral M2 element. 2, LO 7740t; inner lateral view of dextral M, element. 3, LO 7741 1; inner lateral view of sinistral M2 element. 4, LO 7742t; inner lateral view of sinistral M, element. 5, Cn 8065; outer lateral view of dextral Pb2 element. 6, LO 7743t; outer lateral view of sinistral Pb2 element. 7, LO 7744t; inner lateral view of sinistral Sc2 element. 8, Cn 8066 ; upper view of sinistral Pa element. 9, LO 7745t ; inner lateral view of dextral Sc2 element. 10, LO 7746t; posterior view of dextral Sb, element. 11, LO 7747t; outer lateral view of dextral Sb2 element. 12, LO 7748t; posterior view of dextral Sb! element. 13, LO 7749t; posterior view of sinistral Sb, element. 14, Cn 8067; outer lateral view of sinistral Sc, element. 15, LO 7750t; inner lateral view of sinistral Sc, element. 16, Cn 8068; outer lateral view of sinistral Pb, element. 17, LO 775 1 1 ; posterior view of sinistral Sb2 element. 18, LO 7752t; outer postero-lateral view of sinistral Sb2 element. 19, LO 7753t; inner antero-lateral view of sinistral Sb2 element. 20, LO 7755t; outer lateral view of dextral Pc element. 21, LO 7755t; inner lateral view of dextral curved)?) element. 22, LO 7756t; inner lateral view of dextral Sc2 element. 23, LO 7757t; inner lateral view of dextral Sc., element. 24, LO 7758t; inner lateral view of sinistral Sc3 element. 25, LO 7759t; inner lateral view of dextral Sc2 element. 27, LO 7760t; outer lateral view of sinistral Sb2 element. 28, LO 7761 1 ; outer lateral view of dextral carnuliform element, morph a. 29, LO 7762t; outer lateral view of dextral carnuliform element, morph a. 30, LO 7763t ; inner lateral view of sinistral carnuliform element, morph b. 31, LO 7764t; outer lateral view of sinistral curved element. 32, LO 7765t; inner lateral view of sinistral carnuliform element, morph a. 33, LO 7766t; outer lateral view of sinistral carnuliform element, morph a. 34, LO 7767t; posterior view of Sa element. 35, LO 7768t; posterior view of Sa element. Figs 1-4, 7, 10-12, 15, 17-21, 25, 27-29, 31-33 and 35 from Nygardsbackprofilen-1 section (Gotland), sample G93-977LJ; figs 5, 8, 14 and 16 from Ohesaare core, sample M-935, int. 348-40-348-60 m; figs 6, 9, 13, 22-24, 30 and 34 from Nygardsbackprofilen-1 section (Gotland), sample G93-978LJ. Figs 26, 36-54. Pterospathodus celloni (Walliser, 1964). 26, Cn 8069; outer lateral view of dextral Pb2 element. 36, LO 7769t; upper view of sinistral Pa element. 37, Cn 8070; outer lateral view of dextral Pc element. 38, Cn 8071 ; inner lateral view of dextral Sc, element. 39, LO 7770t; inner lateral view of sinistral Sc2 element. 40, Cn 8072; outer lateral view of dextral Pb, element. 41, Cn 8073; inner lateral view of dextral Pa element. 42, Cn 8074; inner lateral view of dextral carnuliform)?) element. 43, LO 7771 1; inner lateral view of dextral M, element. 44, Cn 8075; inner lateral view of sinistral Sc, element. 45, LO 7772t; inner lateral view of sinistral Pa element. 46, LO 7773t; outer lateral view of sinistral Pb, element. 47, LO 7774t; inner lateral view of dextral Sb, element. 48, LO 7775t; outer lateral view of sinistral Sb2 element. 49, LO 7776t; inner lateral view of dextral Pa element. 50, Cn 8076; inner lateral view of dextral carniciform(?) element. 51, Cn 8077; inner lateral view of dextral carnuliform)?) element. 52, Cn 8078; inner lateral view of sinistral carnuliform element. 53, LO 7777t; outer lateral view of dextral Sb2 element. 54, Cn 8079; inner lateral view of sinistral carnuliform(?) element. Figs 26 and 44 from Viki core, sample M-967, int. 153-60-153-72 m; fig. 36 from Nar core (Gotland), int. 351-60-351-65 m; fig. 37 from Viki core, sample M-360, int. 1 53-35—1 53-50 m; figs 38 and 41 from Uulu-330 core, sample M-1070, int. 137-70-137-85 m; figs 39, 43, 45^46 and 48^19 from Sjalso section (Gotland), sample G88-637LJ; fig. 40 from Uulu-330 core, sample M-1301, int. 138 05-138 1 5 m; figs 42, 50-52 and 54 from Viki core, sample M-976, int. 1 49-95—1 50-08 m; figs 47 and 53 from Nar core (Gotland), int. 361-30-361-40 m. All x 50. PLATE 6 MANNIK, Pterospathodus 1044 PALAEONTOLOGY, VOLUME 41 text-fig. 16. For caption see opposite. j MANNIK: SILURIAN CONODONTS 1045 Mountains, Northwest Territories of Canada; Over and Chatterton 1987) indicates that, most probably, they belong to separate species. The Pa element of P. rhodesi is represented at least by two morphs : morph 1 with a bifurcated inner lateral process (Mannik and Aldridge 1989, text-fig. 4 A); and morph 2 with a pennate inner lateral process (Mannik and Aldridge 1989, text-fig. 4 B). P. rhodesi has not been identified in the Baltic or in other parts of Europe. Occurrence. P. a. amorphognathoides Zone (only?). ORIGIN The ancestry of Pterospathodus is uncertain. P. eopennatus appears widely without a direct antecedent. Bischoff (1986) described a presumed direct ancestor of P. celloni (= P. eopennatus in this paper) from New South Wales, Australia, which he called P. cadiaensis. Elements assigned to this taxon by Bischoff are also known from several other regions (Norway - Aldridge and Mohamed 1982, pi. 1, fig. 33; Severnaya Zemlya - Mannik 1983, fig. 4y; Northern Urals- Melnikov, pers. comm.; South China - Aldridge, pers. comm.). Well preserved material from the last region revealed that, in reality, these elements belong to a Gamachignathus apparatus, and P. cadiaensis was reidentified as G. macroexcavatus (Wang and Aldridge, 1996). The new data presented herein indicate an increased level of similarity between the Pterospathodus and Pranognathus apparatuses (Mannik and Aldridge 1989, text-fig. 5; Uyeno and Barnes 1983, pi. 2, figs 1-11, 14-18, identified as belonging to Pterospathodus). However, although the elements of Pranognathus can be readily homologized with those of Pterospathodus (Pa, Pbj, Pc, M15 SCj, Sc2, Sb15 Sb, and Sa can be recognized in Pranognathus) there are still too many significant morphological differences between these two apparatuses to allow us to consider Pranognathus as a direct ancestor of Pterospathodus. In Pranognathus , the Pa and Pb: elements (Mannik and Aldridge 1989, text-fig. 5^4-7, U-X) possess deep wide cavities, the Pc element (Mannik and Aldridge 1989, text-fig. 5 J-K, Y-Z) has three long processes, the Mt element (Mannik and Aldridge 1989, text-fig. 5 L-M) has a denticulated inner lateral process and a large number of denticles on the anterior and posterior processes, and the anterior process of the Sc2 element (Mannik and Aldridge 1989, text-fig. 5N) is densely denticulated. Also, several elements known from Pterospathodus (Pb2, Sc3, canuliform, carniciform and curved), have not been identified in Pranognathus. However, as the ranges of Pterospathodus and Pranognathus are separated by a considerable time interval, it cannot be excluded that these two taxa are related to each other. Hence, the origin of Pterospathodus remains unknown and needs further studies. CONCLUSIONS 1 . Elements previously referred to Carniodus do not belong to a separate genus, but formed part of the apparatus of Pterospathodus. 2. The structure of the Pterospathodus apparatus was evidently much more complicated than previously considered. It consists of at least 14 (15) elements: Pa, Pb,, Pb2, Pc, Mt, M2 (so far recognized only in the P. p. procerus apparatus), Scx, Sc2, Sc3, Sbj, Sb2, Sa, carnuliform element with five morphs, curved element with three morphs and carniciform element. 3. In the Telychian, at least two distinct Pterospathodus lineages: P. a. anguIatus-P. a. amorphognathoides , and P. p. pennatus - P. p. procerus existed and evolved separately, the former in the open shelf carbonate-terrigeneous facies and the latter in deeper basin graptolite-bearing facies. text-fig. 16. Pterospathodus pennatus procerus (Walliser, 1964). a-d, Pa element, e-f, Pb, element. G-o, Pb, element, p-t, a'-b1, Pc element, u-v, c1-D1, Mx element, w-z, M2 element, e1-*1, o\ Sct element. G1— K1, Sc3 element. U-M1, Sbj element, n1, t1, Sb, element, p^s1, vMv1, Sc, element, u1, Sa element. x1-^2, carniciform element. b2-c2, carnuliform element, morph b. d2-m2, carnuliform element morph a. n2-o2, ?carniciform element. Scale bar represents 1 mm. 1046 PALAEONTOLOGY, VOLUME 41 text-fig. 17. Pterospathodus celloni (Walliser, 1964). a-j, m-n. Pa element, k-l, o-s, Pbj element, t, Sb2 element. u-A1, Pb2 element, b^e1, SCj element, f'-k1, Sc2 element, id-N1, carniciform(?) element, o1-!2, carnuliform(?) element. Scale bar represents 1 mm. Both probably originated from a common ancestral taxon at the end of the P. eopennatus Zone, although they may have appeared in the upper part of the A. irregularis Subzone. 4. Evolution was more rapid, and the morphological variation within each population greater, in open shelf environments. MANNIK: SILURIAN CONODONTS 1047 5. Three main intervals of evolution are recognized in the Pterospathodus sequence. The boundaries between them are marked by distinct morphological changes in the elements. 6. P. celloni was restricted to deeper basin environments, and was closely related to the P. pennatus lineage. 7. The appearance of P. celloni and P. p. procerus in open shelf environments occurred during times when the usual conditions there were partially (or completely - Ireviken Event) altered causing temporary disappearances or the final extinction of several lineages. 8. The evolutionary steps in the Pterospathodus lineage, together with changes shown by the rest of the Telychian conodont fauna, provide excellent potential for high resolution stratigraphy (Mannik 1995, 1996). Acknowledgements. Richard J. Aldridge, Howard A. Armstrong, Gunther C. O. Bischoff, Antanas Brazuaskas, Lennart Jeppsson, Hans A. Nakrem, Hans P. Schonlaub and Otto H. Walliser kindly gave me free access to their collections of conodonts. Lennart Jeppsson and Richard J. Aldridge read the manuscript critically and suggested many improvements. Claes Bergman and Fredrik Jerre assisted with the SEM, Gennadi Baranov made the prints and Kaie Ronk the drawings of conodonts. The research was carried out in the Institute of Geology, Tallinn (financed by the Institute of Geology and The Estonian Science Foundation), and in the Department of Historical Geology and Palaeontology, Lund University, Lund (financed by The Swedish Natural Science Research Council). My sincere thanks to everybody. REFERENCES aldridge, r. j. 1972. Llandovery conodonts from the Welsh Borderland. Bulletin of the British Museum [Natural History ), Geology Series , 22, 127-231, pis 1-9. — 1974. An amorphognathoides Zone conodont fauna from the Silurian of the Ringerike area, south Norway. Norsk Geologisk Tidsskrift, 54, 295-303. — 1975. The stratigraphic distribution of conodonts in the British Silurian. Journal of the Geological Society. London , 131, 607-618, 3 pis. — 1979. An upper Llandovery conodont fauna from Peary Land, eastern North Greenland. Rapport Gronlands Geologiske Undersogelse , 91, 7-23, pis 1—2. — 1980. Notes on some Silurian conodonts from Ireland. Journal of Earth Sciences of the Royal Dublin Society , 3, 127-132. — 1985. Conodonts of the Silurian System from the British Isles. 68-92. In higgins, a. c. and Austin, r. l. (eds). A stratigraphical index of conodonts. Ellis Horwood, Chichester, 263 pp. — ieppsson, L. and dorning, k. j. 1993. Early Silurian episodes and events. Journal of the Geological Society, London. 150, 501-513. — and mohamed, I. 1982. Conodont biostratigraphy of the early Silurian of the Oslo Region, 109-120, 2 pis. In worsley, d. (ed.). Subcommission on Silurian Stratigraphy. Field Meeting, Oslo Region, 1982. Paleontological Contributions from the University of Oslo, 278, 175. an tai-xiang 1987. [The Lower Palaeozoic conodonts of South China.] Beijing University Press, Beijing, 238 pp., 35 pis. [In Chinese.] Armstrong, h. a. 1990. Conodonts from the Upper Ordovician-Lower Silurian carbonate platform of North Greenland. Gronlands Geologiske Undersogelse, 159, 1-151, pis 1-23. barca, s., ferretti, a., massa, p. and serpagli, E. 1992. The Hercynian Arburese Tectonic Unit of SW Sardinia. New stratigraphic and structural data. Rivista Italiana di Paleontologia e Stratigrafia, 98, 1 19-136, pis 1-4. brazauskas, a. z. 1983. Conodont zones of Lithuanian Llandovery facies. Nauchnye Trudy Vysshih Uchebnyh Zavedenij Litovskoj SSR, Geologija , 4, 41-66. [In Russian, with English and Lithuanian summaries], barrick, J. E. 1983. Wenlockian (Silurian) conodont biostratigraphy, biofacies, and carbonate lithofacies, Wayne Formation, Central Tennessee. Journal of Paleontology, 57, 208-239. and klapper, g. 1976. Multielement Silurian (late Llandoverian-Wenlockian) conodonts of the Clarita Formation, Arbuckle Mountains, Oklahoma, and phylogeny of Kockelella. Geologica et Palaeontologica, 10, 59-100, pis 1-4. bischoff, G. c. o. 1986. Early and middle Silurian conodonts from midwestern New South Wales. Courier Forschungsinstitut Senckenberg, 89, 337, pis 1-34. 1048 PALAEONTOLOGY, VOLUME 41 cooper, b. j. 1977. Toward a familial classification of Silurian conodonts. Journal of Paleontology, 51, 1057-1071. drygant, d. m. 1969. [Konodonty Restevskogo, Kitajgorodskogo i Mukshinskogo gorizontov Silura Podolii.] Paleontologicheskij Sbornik, 6 (1), 49-54, 1 pi. [In Russian]. - 1984. [Korrelyaciya i konodonty silurijskih-nizhnedevonskih otlozhenij Volyno-Podolij.\ Naukova Dumka, Kiev, 192 pp., 16 pis. [In Russian], dumoulin, A. and Harris, A. G. 1987. Off-platform Silurian sequence in the Ambler River Quadrangle. Geologic Studies in Alaska by the U.S. Geological Survey during 1987, 35-38. helfrich, c. T. 1980. Late Llandovery-early Wenlock conodonts from the upper part of the Rose Hill and the basal part of the Mifflintown formations, Virginia, West Virginia, and Maryland. Journal of Paleontology, 54, 557-569, 2 pis. igo, H. and koike, T. 1968. Ordovician and Silurian conodonts from the Langkawi Island, Malaya, Part 2. Geology and Palaeontology of Southeast Asia, 4, 1-21, pis 1-3. JIANG WU, ZHANG FANG, ZHOU XIYUN, XIONG JIANFEI, DAI JINYIE and ZHONG DUAN 1986. [Conodonts - palaeontology .] Southwestern Petroleum Institute, Sichuan, 264 pp, pis 265-287. [In Chinese], jeppsson, l. 1979. Conodonts. 225-248. In jaanusson, v., laufeld, s. and skoglund, r. (eds). Lower Wenlock faunal and floral dynamics - Vattenfallet section, Gotland. Sveriges Geologiska Undersokning Avdhandlingar och Uppsatser, Cl 62, 1-294. - 1987. Lithological and conodont distributional evidence for episodes of anomalous oceanic conditions during the Silurian. 129-145. In aldridge, r. j. (ed.). Palaeobiology of conodonts, Ellis Horwood, Chichester, 180 pp. - 1994. A new standard Wenlock conodont zonation. 133. In schonlaub, h. p. and kreuzer, l. h. (eds). IUGS Subcommission on Silurian Stratigraphy - Field Meeting Eastern + Southern Alps, Austria 1994 in mem. H. Jaeger. Berichte der Geologischen Bundesansta.lt, 30, 1-156. - 1997. A new latest Telychian, Sheinwoodian and early Homerian (early Silurian) Standard Conodont Zonation. Transactions of the Royal Society of Edinburgh: Earth Sciences. 88, 91-114. - and mannik, p. 1993. High-resolution correlations between Gotland and Estonia near the base of the Wenlock. Terra Nova, 5, 348-358. klaamann, e. 1990. Locality 8:3. Valgu outcrop. 181-182. In kaljo, d. and nestor, h. (eds). Field Meeting, Estonia 1990. An excursion guidebook. Tallinn, 209 pp. kleffner, m. a. 1985. Conodont biostratigraphy of the stray ‘Clinton' and ‘Packer Shell- (Silurian, Ohio subsurface) and its bearing on correlation. 219-229, 2 pis. In gray, j., maslowski, a., McCullough, w. and shafer, w. e. (eds). The new Clinton collection - 1985. Ohio Geological Society, Columbus, Ohio, 243 pp. - 1987. Conodonts of the Estill Shale and Bisher Formation (Silurian, southern Ohio): biostratigraphy and distribution. Ohio Journal of Sciences, 87 (3), 78-89. 1991. Conodont biostratigraphy of the upper part of the Clinton Group and the Lockport Group (Silurian) in the Niagara Gorge Region, New York and Ontario. Journal of Paleontology, 65, 500-511. kuwano, y. 1976. Finding of Silurian conodont assemblages from the Kurosegawa tectonic zone in Shikoku, Japan. Memoirs of the National Science Museum, 9, 17-22, 2 pis. [In Japanese with English summary]. liebe, r. m. and rexroad, c. b. 1977. Conodonts from Alexandrian and early Niagaran rocks in the Joliet, Illinois, area. Journal of Paleontology, 51, 844-857, pis 1-2. mabillard, j. e. and aldridge, R. J. 1983. Conodonts from the Coralliferous Group (Silurian) of Marloes Bay, south-west Dyfed, Wales. Geologica et Palaeontologica, 17, 29-43, pis 1-4. manara, c. and vai, G. b. 1970. La sezione e i conodonti del costone sud del M. Rauchkofel (Paleozoico, Alpi Carniche). Giornale di Geologia, 36, 441-503, pis 59-63. McCracken, a. d. 1991. Silurian conodont biostratigraphy of the Canadian Cordillera with a description of new Llandovery species. 97-127, 5 pis. In orchard, m. j. and McCracken, a. d. (eds). Ordovician to Triassic conodont paleontology of the Canadian Cordillera. Bulletin of the Geological Survey of Canada, 417, 1-335. miller, r. h. 1976. Revision of upper Ordovician, Silurian, and lower Devonian stratigraphy, southwestern Great Basin. Bulletin of the Geological Society of America, 87, 961-968. - 1978. Early Silurian to early Devonian conodont biostratigraphy and depositional environments of the Hidden Valley Dolomite, southeastern California. Journal of Paleontology, 52, 323-344, pis 1-4. mannik, p. 1983. Silurian conodonts from Severnaya Zemlya. Fossils and Strata, 15, 111-119, 1 pi. 1992. [Upper Ordovician and lower Silurian conodonts in Estonia.] Unpublished Ph.D. thesis. University of Tartu, Estonia. [In Russian]. 1995. The evolution of selected conodont lineages and an improved zonation for the Telychian (late Llandovery). 52-54. In brock, g. a. (ed.). First Australian Conodont Symposium (AUSCOS-I) and the MANNIK: SILURIAN CONODONTS 1049 Boucot Symposium, 18-21 July, 1995, Abstracts and Programme. Special Publications of the Macquarie University Centre for Ecostratigraphy and Palaeobiology (MUCEP), 1, 1-108. — 1996. Telychian (Early Silurian) conodont Pterospathodus : evolution and taxonomy. 35. In dzik, j. (ed.). Sixth European Conodont Symposium (ECOS VI), Abstracts. Institut Paleobiologii PAN, Warszawa, 70 pp. and aldridge, R. J. 1989. Evolution, taxonomy and relationships of the Silurian conodont Pterospathodus. Palaeontology, 32, 893-906. and malkowski, k. 1998. Silurian conodonts from the Goldap core, Poland. Palaeontologia Polonica , 58, 139-149. and viira, v. 1990. Conodonts. 84—89, pis 16-18. In kaljo, d. and nestor, h. (eds). Field Meeting, Estonia 1990. An excursion guidebook. Tallinn, 209 pp. nakrem, H. a. 1986. Llandovery conodonts from the Oslo Region, Norway. Norsk Geologisk Tiddskrift, 66, 121-133. nehring-lefeld, m. 1985. Conodonts of the amorphognathoides Zone (Silurian) from eastern part of the Podlasie Depression. Kwartalnik Geologiczny, 29 (3/4), 625-652, pis 1-6. [In Polish with English and Russian summaries]. — 1992. Biostratigraphy of the Old Paleozoic carbonates in the Zawiercie area (NE margin of the Upper Silesian Coal Basin). Geological Quarterly, 36, 171-198, pis 1-5. nicoll, r. s. and rexroad, c. b. 1968. Stratigraphy and conodont paleontology of the Salamonie Dolomite and Lee Creek Member of the Brassfield Limestone (Silurian) in southeastern Indiana and adjacent Kentucky. Bulletin of the Department of Natural Resources of the Geological Survey of Indiana, 40, 1-73, pis 1-7. nowlan, g. s. 1981. Late Ordovician - early Silurian conodont biostratigraphy of the Gaspe Peninsula -a preliminary report. 257-291, 7 pis. In lesperance, p. j. (ed.). Subcommission on Silurian Stratigraphy, Ordovician-Silurian Boundary Working Group. Field Meeting, Anticosti-Gaspe, Quebec 1981, Vol. 2: stratigraphy and paleontology. University of Montreal, 321 pp. 1983. Early Silurian conodonts of eastern Canada. Fossils and Strata, 15, 95-110. over, d. j. and chatterton, d. e. 1987. Silurian conodonts from the southern Mackenzie Mountains, Northwest Territories, Canada. Geologica et Palaeontologica, 21, 1^19, pis 1-8. pickett, J. 1978. Silurian conodonts from Blowclear and Liscombe Pools, New South Wales. Journal and Proceedings of the Royal Society of New South Wales, 111, 35-39. qiu hongrong 1985. Silurian conodonts in Xizang (Tibet). Bulletin of the Institute of Geology of the Chinese Academy of the Geological Science, 11, 23-38, pis 1-2. [In Chinese with English summary]. 1988. Early Palaeozoic conodont biostratigraphy of Xizang (Tibet). Symposium on Stratigraphy and Paleontology, 185-202, 6 pis. rexroad, c. R. and nicoll, r. s. 1971. Summary on conodont biostratigraphy of the Silurian System of North America. Memoir of the Geological Society of America, 127, 207-225, pis 1-2. — 1972. Conodonts from the Estill Shale (Silurian, Kentucky and Ohio) and their bearing on multielement taxonomy. Geologica et Palaeontologica, SB1, 57-74, pis 1-2. saladzius, v. 1975. Conodonts of the Llandoverian (lower Silurian) deposits of Lithuania. 219-225, pis 1-2. In grigelis, a. a. (ed.). The fauna and stratigraphy of Palaeozoic and Mesozoic of Baltic and Byelorussia. ‘Minds’, Vilnius, 249 pp. [In Russian, with English summary]. savage, n. m. 1985. Silurian (Llandovery-Wenlock) conodonts from the base of the Heceta Limestone, southeastern Alaska. Canadian Journal of Earth Sciences, 22, 711-727. — potter, a. w. and wyatt, g. g. 1983. Silurian and Silurian to early Devonian conodonts from West Central Alaska. Journal of Paleontology , 57, 873-875. schonlaub, h. p. 1969. Das Palaozoikum zwischen Bischofalm und Hohem Trieb (Zentrale Karnische Alpen). Jahrbuch der Geologischen Bundesanstalt, 112, 265-320, pis 1-2. — 1971. Zur Problematic der Conodonten-Chronologie an der Wende Ordoviz/Silur mit besonderer Beriicksichtigung der Verhaltnisse im Llandovery. Geologica et Palaeontologica, 5, 35-57, pis 1-3. 1975. Conodonten aus dem Llandovery der Westkarawanken (Osterreich). Verhandlungen der Geologischen Bundesanstalt, 2-3, 45-65, pis 1-2. spasov, h. and filipovic, i. 1966. The conodont fauna of the older and younger Palaeozoic in southeastern and northwestern Bosnia. Geoloski Glasnik, 11, 33-53, pis 1-3. stouge, s. and bagnoli stouge, g. 1984. An upper Llandovery conodont fauna from Eastern Hall Land, North Greenland. Bolletino della Societd Paleontologica Italiana, 23, 103-112, pis 1-2. sweet, w. c. 1981. Morphology and composition of elements. W5-W20. In robinson, r. a. (ed.). Treatise on invertebrate paleontology. Part W. Miscellanea. Supplement 2. Conodonta. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 202 pp. 1050 PALAEONTOLOGY, VOLUME 41 1988. The Conodonta. Morphology, taxonomy, paleoecology, and evolutionary history of a long-extinct animal phylum. Oxford Monographs on Geology and Geophysics, 10, 1-212. uyeno, T. t. 1990. Biostratigraphy and conodont faunas of Upper Ordovician through Middle Devonian rocks, eastern Arctic Archipelago. Bulletin of the Geological Survey of Canada , 401, 1—2 1 0, pis 1-20. and barnes, c. R. 1981. A summary of Lower Silurian conodont biostratigraphy of the Jupiter and Chicotte formations, Anticosti Island, Quebec. 173-184, 1 pi. In lesperance, p. j. (ed.). Subcommission on Silurian Stratigraphy, Ordovician-Silurian Boundary Working Group. Field Meeting, Anticosti-Gaspe, Quebec 1981, Vol. 2: stratigraphy and paleontology. University of Montreal, 321 pp. 1983. Conodonts of the Jupiter and Chicotte formations (lower Silurian), Anticosti Island, Quebec. Bulletin of the Geological Survey of Canada, 355, 1-48, pis 1-9. walliser, o. h. 1964. Conodonten des Silurs. Abhandlungen des Hessischen Landesamtes fur Bodenforschung, 41, 1-106, pis 1-32. wang chen-yuan and aldridge, r. j. 1996. Conodonts. 46-55, pis 1-3. In chen xu and rong jia-yu (eds). [Telychian ( Llandovery ) of the Yangtze Region and its correlation with British Isles.] Science Press, Beijing, 157 pp. [In Chinese]. watkins, R., kuglitsch, j. j. and McGee, p. E. 1994. Silurian of the Great Lakes Region, Part 2: paleontology of the Upper Llandovery Brandon Bridge Formation, Walworth County, Wisconsin. Milwaukee Public Museum Contributions in Biology and Geology, 87, 1-71, pis 1—4. yu hongjin 1985. Conodont biostratigraphy of Middle-Upper Silurian from Xainza, Northern Xizang (Tibet). Contribution to the geology of the Qinghai-Xizang (Tibet) Plateau, 16, 15-31, pis 1-3. [In Chinese with English summary], peep mannik Institute of Geology Tallinn Technical University Estonia pst. 7 Typescript received 3 July 1997 EE0001 Tallinn, Estonia Revised typescript received 25 March 1998 e-mail mannik@gi.ee APPENDIX List of localities cited in text (see also Text-fig. 1). A. Exposures 1. Valgu-a drainage canal; studied section corresponds to Point 1 in Klaamann (1990, p. 181). 2. Velise-Korgekalda - low cliff on the left bank of the Paardu (on some maps indicated as Velise) River; c. 2-5 km west-south-west of a bridge across the river in Velise village, central Estonia. 3. Sjalso - shore exposure; c. 4-3 km west of Vaskinde church, Gotland, Sweden. 4. Overstekvarn 2 - several small exposures in the southern brook about 200 m north-west of Overstekvarn; c. 4-45 km (south-)south-west of Lummelunda church, Gotland, Sweden. 5. Nygardsbackprofilen 1 - a brook section at the mouth of Nygardsbacken (backen = brook) and shore section south-west of it; c. 2-97 km north-west of Vasterhedje church, Gotland, Sweden. B. Cores 1. The Nurme, Uulu-330, Viki and Ohesaare cores are housed in the Sarghaua field station (central Estonia), Institute of Geology, Tallinn Technical University. 2. The Pahapilli core is housed in the Turja field station (southern Saaremaa, Estonia), Geological Survey of Estonia. 3. The Nar core is housed in the Geological Survey of Sweden, Uppsala, Sweden. ON PREDATOR DETERRENCE BY PRONOUNCED SHELL ORNAMENT IN EPIFAUNAL BIVALVES by HYWEL M. I. STONE Abstract. Laboratory experiments, undertaken to determine the effectiveness of pronounced shell ornament in epifaunal bivalves against predatory shell boring by subtropical muricid gastropods and extraoral feeding by asteroids, suggest that natural and artificial spines deter muricid predators from attacking ornamented areas of the bivalve shell but do not have a similar effect upon predatory asteroids. These findings are discussed in relation to the extant and often highly spinose cementing bivalve families Spondylidae and Chamidae. The adaptive radiation of the Muricidae in the Albian may have resulted in selection for highly ornamented epifaunal bivalve taxa in shallow, warm water environments where the epifaunal habit renders sessile prey particularly vulnerable to attack by roving durivorous predators. The ability to produce spines, however, was already apparent in ancestral Pectinoida in the late Palaeozoic. It is concluded that the pronounced shell ornament of the free valves of warm water cemented epifaunal bivalve taxa is functional against shell boring muricids. Other hypothesized functions are discussed briefly. For many years it has been argued that shell ornament in bivalved molluscs is directly related to the mode of life of the animal. Thus, shallow infaunal taxa display sculptures that are interpreted as acting as aids to burrowing or as stabilizers for life within soft substrata (Stanley 1970), and cementing epifaunal taxa produce commarginal lamellae or spines on the Tower’ valve as aids to attachment to hard substrata (Stenzel 1971). There have been various hypotheses for the function of spines and commarginal lamellae on the ‘upper’ free valve of cementing pleurothetic epifauna. It has been suggested, for example, that ornament serves to increase the effective strength of the shell and thus defend it against predators (Vermeij 1987). It has also been proposed that ornament acts to attract the growth of epibionts to the shell. The latter may discourage predation by visually camouflaging the shell or chemically masking secreted metabolites that may act as cues to potential predators (Vance 1978; Feifarek 1987). Ornament may also provide protection for sensory outposts of mantle tissue (by analogy with brachiopod soft part morphology) (Rudwick 1965; Stenzel 1971), or act to deter rasping by certain grazers which may over time erode and weaken the shell. Finally, both spines and commarginal lamellae may act as direct defences against predation (Kauffmann 1969; Vermeij 1987). It is this last hypothesis on which the present study is focused. Historically, however, both the nature of such defences, and the identity of the predators which they may deter, have been far from clear. As Harper and Skelton (1993a) have pointed out, much of the literature concerning shell ornament in bivalves has been based upon anecdotal evidence. The great majority of shell ornament studies have concentrated on infaunal taxa and their adaptations to life within soft substrata (e.g. Stanley 1970, 1981, 1988; Wilson 1979; Watters 1993). However, some work has shown that predation by naticid gastropods may be an important selective force operating on the functional design of certain shallow infaunal bivalve groups, especially in the Indo-Pacific (Ansell and Morton 1983). The effects of shell ornament on other predatory methods, such as smothering by naticids (Ansell and Morton 1987), foraging by crabs, asteroids and birds (Carter 1968; Ansell 1969), and fish predation, have received little attention. Very little evidence has been presented concerning the possible anti-predatory effects of epifaunal bivalve shell sculpture in relation to a variety of predatory methods, including shell boring by muricid gastropod predators and extraoral feeding by asteroids, both subjects of the present study. Muricids are known to exert heavy predation pressure on organisms in many shallow warm (Palaeontology, Vol. 41, Part 5, 1998, pp. 1051-1068] © The Palaeontological Association 1052 PALAEONTOLOGY, VOLUME 41 temperate, subtropical and tropical rocky shores and coral reefs (Taylor 1976, 1978). In these environments, spinose epifaunal prey taxa may be an important component of the intertidal and subtidal megafauna. Jackson (1977), for example, in his studies of Jamaican reefs, stated that more than 50 per cent, of spondylid, dimyid and chamid mortalities may be caused by shell boring muricid gastropods. Recently, a possible defensive role for the pronounced shell ornament of the intertidal oyster Saccostrea cucullata (Born) has been suggested by Taylor (1990), who noted that in Hong Kong the marginal spines of this species may hinder edge-boring by muricids. Preliminary experimental observations of Harper and Skelton (1993a) on the same species also suggested that spines are directly effective at deterring muricid predatory activity; they stressed the potential importance of the rise of the shell boring muricids in the early Cretaceous as a factor influencing bivalve defensive traits. Apart from studies such as that of Harper (1994), little work has been published on bivalve predation by subtropical and tropical extraoral feeding asteroids. Two epifaunal cementing bivalve families, Spondylidae and Chamidae, are of particular interest in the present analysis of the direct inhibitory effect of shell ornament on predators. Although unrelated, both have a wide distribution in warm temperate, subtropical and tropical shallow waters, particularly on rocky shores and reefs (Zavarei 1973; Bernard 1976). Stanley (1970) noted that the spinose ornamentation of these groups, particularly the spondylids, may serve as a defensive adaptation. He considered reef fishes potentially important but did not demonstrate experimentally a link between epifaunal bivalve ornamentation and predation. Likewise, Logan (1974) has suggested that the spines of Spondylus americanus Hermann, a species living on the Bermuda Platform, perform a protective function, but did not state which predators would be deterred. This study presents evidence, through manipulative experiments, that spines on bivalve shells may directly hinder a variety of subtropical muricid gastropod predators, but do not have a similar effect against a subtropical extraoral feeding asteroid. The significance of these findings is discussed. The terminology used to describe shell surface structures and ornamentation has seldom been rigorously applied in the literature. A ‘spine’ is here defined as a calcareous projection, perpendicular, oblique or subparallel to the general surface of the shell, that arises as a result of the combination of radial and concentric elements of shell growth (Cox 1969), and whose height is greater than that of the shortest dimension of its base. For structures where the basal diameter is equal to or greater than the height, the term ‘node’ is better applied. A ‘lamella’ is defined as a calcareous projection, perpendicular, oblique or subparallel to the general surface of the shell, that arises with varying degrees of radial expression but always commarginal with the commissure, and whose height is greater than that of the basal thickness. For structures where the basal thickness is equal to or greater than the height, the term ‘commarginal ridge’ is better applied. MATERIALS AND METHODS To determine the effectiveness of pronounced shell ornamentation in bivalves against predation by shell boring muricids and an extraoral feeding asteroid, a series of experiments was conducted on living animals in flow-through Perspex aquaria at the Swire Institute of Marine Science, Cape d’Aguilar, Hong Kong, from May to July 1995. Over this period, aquarium seawater temperatures reflected those in the field, with an average of 25 °C + 2 °C. Bivalve prey were offered as a simple choice between ornamented and unornamented individuals (Experiments 1 to 3) or with varying degrees of shell ornamentation (Experiment 4). Different forms of artificial sculptural attachments were produced to form spinose shells in Experiments 1, 3 and 4, and the effects of natural ornament were observed in Experiment 2. For prey bivalves, the mytilid Perna viridis (Linnaeus) was collected intertidally from a pier in the Tolo Channel (New Territories), the mytilid Septifer virgatus (Wiegmann) intertidally from the exposed eastern shore of Cape d’Aguilar (Hong Kong Island), and the chamid Chama reflexa Reeve sublidally from Hoi Ha Wan (eastern New Territories). In shallow subtidal and low intertidal STONE: PREDATOR DETERRENCE BY BIVALVES 1053 text-fig. 1. Diagrammatic representation of artificial ‘ spines ’ attached to mussel shells in Experiments 1 and 3. A— B, lateral and marginal views of polyethylene ‘spines’ attached to Septifer virgatus offered to Thais luteostoma as prey in Experiment 1 . C-D, lateral and marginal views of ‘ spines ’ attached to Perna viridis cut from mussel shell, offered to Coscinasterias acutispina as prey in Experiment 3. Scale bar represents 1 mm (a-b) or 2 mm (c-D). areas around many rocky shores in Hong Kong, C. reflexa cements to large boulders, often on their undersides or in crevices (pers. obs.). At the collection site, they were often found aggregated into dense clumps in crevices in the vertical walls of a small jetty. The shell boring muricid predators Thais luteostoma Holten, T. clavigera Kuster and Morula musiva Kiener, and the extraoral feeding asteroid predator Coscinasterias acutispina Stimpson were all collected from sheltered sites in Lobster Bay (Cape d’Aguilar). The asteroids form an important component of the epifauna at the collection site and may feed opportunistically (Harper 1994). The muricid Chicoreus microphyllus Lamarck was collected from Hoi Ha Wan. C. micro- phyllus , T. luteostoma , C. acutispina and C. reflexa were all gathered subtidally, whilst T. clavigera , M. musiva , P. viridis and S. virgatus were found in the intertidal zone. Apart from in Experiment 4, all bivalve prey were distributed in a random manner on the base of each tank. For the duration of the experiments the byssate S. virgatus and P. viridis were not allowed to attach themselves to the floor or walls of each tank in order to reduce the possibility of variation in strength of attachment and to ensure relative homogeneity amongst predator-prey interactions. All predator sizes were recorded. In order to reduce any bias in the observations resulting from death from other possible causes, only those bivalves whose flesh had been completely removed from the shell by the predator were recorded as ‘consumed’ in the following experiments. Any prey that showed any signs of physical or behavioural deterioration were immediately discarded. For each experiment, predators were initially starved for one week prior to use. All experiments were run for six weeks and aquaria were checked daily. Three replicates were undertaken for each study apart from Experiment 4, with six replicates. Statistical analyses were calculated using the chi square test. 1054 PALAEONTOLOGY, VOLUME 41 text-fig. 2. Diagrammatic representation of artificial polyethylene ‘spines’ attached to the lateral surface of Perna viridis shells offered to Morula musiva as prey in Experiment 4. a, lateral and posterior views of unornamented mussels dotted with cyanoacrylate as a control. B, lateral and posterior views of 1 mm high ‘spines’, c, lateral and posterior views of 2 mm high ‘spines’. D, lateral and posterior views of 4 mm high ‘spines’. Scale bar represents 5 mm. o O B C Artificial spines Artificial spines were manufactured and attached to prey bivalve shells in Experiments 1, 3 and 4. Both the form and arrangement of such spines varied according to the experiment, as described below. Apart from in Experiment 4, the process of adhering spines or dotting unornamented prey with epoxy in the laboratory lasted about one hour per specimen at room temperature. All prey, ornamented and unornamented, underwent the same period of emersion. In the aquarium, the muricid Thais luteostoma appeared to be an obligate edge-borer of the prey Septifer virgatus , probably the result of alternative prey not being available. The artificial spines required for Experiment 1, therefore, were fixed only along the commissure of both valves, including the area of emergence of the byssus. Since the bivalve prey in this experiment were of relatively small size, and for ease of manipulation, the artificial spines were cut from polyethylene strips to a length of 5 mm, with basal dimensions of 2 x 2 mm. These were fixed to the bivalves with epoxy resin ( Araldite Rapid - Ciba Geigy) at 2 mm distances along the commissure so that they both radiated outwards from the margin and interdigitated by passing through the plane of the commissure at a low angle (Text-fig. 1a-b). The artificial spines were maintained at a length of 5 mm in all cases. As a control for the use of the epoxy, unornamented prey were dotted with Araldite at the same sites as artificial spine attachment in ornamented individuals. Most spines remained secure for the duration of each relevant experiment. If any became dislodged, prey were replaced by newly prepared individuals. Sculptural augmentation in Experiment 3 consisted of adhering artificial spines all around the commissure of the prey shell, in a similar arrangement to that of Experiment 1, except that, because the prey used in this instance were generally larger than S. virgatus , it was possible to cut the spines from dead Perna viridis shells with a craft saw, instead of from polyethylene strips. Spines produced in this way were cut to a length of approximately one-third of the length of the shell along the axis of greatest growth, and were attached with epoxy resin. Their natural curvature resulted in interdigitation across the plane of the commissure (Text-fig. 1c-d). This arrangement was chosen for two reasons: (1) the resulting spines were particularly sturdy when fixed and (2), as the mussels are orthothetic in life, a roving predator is likely to come into contact with the margins of the valves, rather that the sides, particularly on sense mussel beds. As a control, epoxy resin was also applied to the shells of unornamented prey so that any differences in asteroid feeding behaviour by, for example, the presence of adhesive chemically masking metabolic cues secreted by the mussels, would apply to both ornamented and unornamented prey alike. In Experiment 4, artificial spines were cut at varying lengths and arranged over the whole surface of the free unattached valve of artificially cemented P. viridis prey. For ease of manipulation, spines were cut from polyethylene strips and attached with cyanoacrylate (Superglue-Loctite) rather than epoxy resin. This provided effective underwater strength of adhesion. The spines were spaced at STONE: PREDATOR DETERRENCE BY BIVALVES 1055 2 mm from each other on the free valve surface, mirroring the basal dimensions of the projections themselves (2x2 mm) (Text-fig. 2). As a control, the five unornamented bivalves were dotted with cyanoacrylate at 2 mm intervals, thus mimicking the spine distributions on ornamented individuals. All treatments covered the entire free valve from the umbo to the posterior margin. The process of attaching the spines to the shells in the laboratory lasted about ten minutes for each specimen. The use of epoxy resin and cyanoacrylate in manipulative experiments is far from new. Harper (1991), for example, in her work on the effects of cementation on predation, has shown that daubing the shells of Mytilus edulis with epoxy reveals no apparent inhibitory effect on crushing predation by the crabs Cancer pagarus and Carcinus maenas or extraoral predation by the asteroid Asterias rubens. For the work presented here, a pilot study of drilling by Thais clavigera revealed no statistically significant difference in consumption between numbers of Pema viridis whose shell had been dotted with either epoxy or cyanoacrylate and those that had been left untouched. Experiment 1 This experiment was designed to show whether artificial spines had an inhibitory effect on shell boring by Thais luteostoma. Five T. luteostoma were offered 20 Septifer virgatus as prey. Ten of the latter were unornamented, and ten had artificial ‘spines’ attached to both valves all around the commissure. Any differences in the positions of bore holes in ornamented and unornamented prey were noted, and observations were made on the possible inhibitory effects of the spines themselves. All bivalves eaten were replaced with ones of similar size and state of ornamentation. Experiment 2 This experiment was designed to show if the natural ornamentation of the ‘upper’ right valve or commissural edge of the prey bivalve Chama reflexa inhibits boring by the muricids Chicoreus microphyllus and Thais clavigera. Two C. microphyllus and ten T. clavigera were offered 20 C. reflexa as prey in each tank. The numbers of each muricid species used merely reflected availability and no attempt at making any inferences about differences in the number of bore holes made by the two muricids was made. Ten bivalves were presented with their natural shell ornament intact, and ten had been filed to a smooth, unornamented finish. The natural sculpture consisted of either very small spines (up to F5mm long) covering the ‘upper’, right valve, or a crenulate marginal lamella projecting some 2 mm from the commissure, or both. For prey which had been filed smooth, ornament was removed only from the right valve of each specimen. The left valves of all ornamented and unornamented specimens, which were covered by commarginal series of attachment lamellae except at the site of the attachment scar, were left untouched. No attempt was made to cement artificially the prey in the tanks and the bivalves were offered in random distributions and orientations. A range of prey sizes was presented and consumed individuals were replaced by ones of similar size and state of ornamentation. As in Experiment 1, the areas of boring were noted. Exact positions of bore holes are not displayed graphically, however, because of the variable shell morphology of C. reflexa and hence the difficulty of projecting such positions onto a standard graphic template. Experiment 3 This experiment was designed to show if artificial spines deter predation by the extraoral feeding asteroid Coscinasterias acutispina. Ten individuals of C. acntispina were offered 40 Pema viridis as prey, of which 20 were ‘ornamented’ and 20 ‘unornamented’. A random size range of prey was presented, and all bivalves consumed were replaced by individuals of similar size and ornament. Numbers of consumed individuals were recorded as well as the feeding behaviour of C. acutispina in order to determine any inhibitory effect that the spines may confer. Only animals with three or 1056 PALAEONTOLOGY, VOLUME 41 more primary arms were selected for use in the study and were chosen at random for each replicate from a holding tank. As noted by Harper (1994), there is a tendency for C. acutispina to undergo fissiparity, and this may be a result of stress caused by aquarium confinement or by other factors, such as high levels of prey consumption. Any daughter asteroids produced in this way were immediately removed so that only ten starfishes were present in each replicate at any one time. Experiment 4 This was essentially a prey choice study and was designed to show if increasingly spinose prey bivalves had a concomitant increasingly inhibitory effect on shell boring by the muricid Morula musiva. Twenty Perna viridis were cemented artificially to a Perspex sheet in an aquarium and presented to five M. musiva predators. Attachment to Perspex was effected by attaching each shell with epoxy resin by either the right or left valve. Both the choice of the valve of attachment and the orientation of each fixed mussel on each sheet were determined at random. Each replicate sheet was slotted vertically into a Perspex frame at the bottom of an aquarium tank, providing individual predator compartments with partitioning sheets of cemented prey separating each compartment. This design was considered satisfactory as it allowed the valves of the mussels to gape normally with apparently little detrimental effect upon feeding and respiration. Of the 20 P. viridis on each sheet partition, five were unornamented, five were ornamented with 1 mm long artificial spines, five with 2 mm long spines and five with 4 mm long spines. All artificial spines were applied to the free valve in each case and, as in Experiment 1, the spines were cut from polyethylene strips, with a basal area of 2 x 2 mm. Artificial attachment of prey to the Perspex sheets did not result in any apparent detrimental effect to normal valve gaping. Moreover, despite this treatment, all the prey in this experiment continued to produce copious byssus threads, suggesting that they were behaving normally. These threads were scraped away from their attachment to the sheet to reduce the possibility of predators being caught and immobilized by them, as has been reported for other mytilids (e.g. Petraitis 1987; Wayne 1987; Day et al. 1991), which may have biased the results. In each replicate, consumed prey were replaced in the same manner as the experiments above until five or more of each particular ornament type were eaten. Subsequently, all remaining mussels of the same type were removed, leaving a reduced amount of sculptural possibilities to be tackled by M. musiva. This process was repeated until termination of the experiment when prey preferences could be determined. Consumed prey were also analysed for position of bore holes. RESULTS Experiment 1 Thais luteostoma versus Septifer virgatus with artificial marginal spines. The null hypothesis that equal numbers of ornamented and unornamented prey should be consumed is rejected using the chi square test at a 95 per cent, confidence level (Table 1). Very few spinose bivalves were tackled and, of those that were, there was no evidence of any predator behavioural modification resulting in changes in bore hole positioning. All bore holes, for both spiny and non-spiny prey, were made at the commissural edge and the vast majority located antero-ventrally in the area of the emergence of the byssus (Text-fig. 4a). A byssal gape is present to a greater or lesser degree in all mytilids and, when not permitted to attach to a substrate, is potentially a particularly vulnerable area because it may allow direct entry to the body cavity and metabolites leached from the gape may act as chemical cues to potential predators. The method of experimentation described above is justified because it is even more likely that spines would provide effective defence if the mussels were allowed to attach normally by the byssus and become orthothetically orientated, as in life. This is especially true when shells aggregate side to side in beds and a potential predator would be presented with ornamented dorsal margins rather than the unornamented sides of the valves. The results show that, in the aquarium, Thais luteostoma is an obligate edge-borer on Septifer virgatus and that artificial marginal spines attached to prey shells effectively inhibit predation by this STONE: PREDATOR DETERRENCE BY BIVALVES 1057 table 1. Numbers of unornamented and artificially ornamented Septifer virgatus consumed by the muricid gastropod Thais luteostoma in three replicate experiments. Statistical analysis using chi square. Assuming a confidence level of 95 per cent., the null hypothesis that equal numbers of ornamented and unornamented prey are eaten is rejected for all replicates. Replicate A B C Spiny prey consumed 2 3 3 Non-spiny prey consumed 36 34 38 Overall total 38 37 41 Predator size (mm) 36-2 36-5 32-7 P 0 05 g> 0-05 >> 0 05 Right valves bored 10 5 11 Left valves bored 10 17 9 P ?> 0 05 < 001 > 0 05 muricid. No attempt was made to bore the sides of any valves and no incomplete bore holes were recorded. It is interesting to note that most boring resulted in damage to both valves either side of the thin byssal gape in any one individual, but the periostracum often remained intact along the commissure (Text-fig. 4a). This provides additional evidence to that already published that the thick periostracum of many mytilids, including that of S. virgatus at a thickness of 60 //m (Harper 1997), is difficult to penetrate by gastropod borers and may constitute a defensive adaptation (Harper and Skelton 19936). An inspection of those spinose mussels whose defences had been overcome by T. luteostoma showed that one or more spines had been dislodged, suggesting that if all spines had remained completely intact, there would have been even less predation of ornamented individuals. In only one prey individual were all the spines intact after a successful predatory attack (Text- fig. 4b). Aquarium observations revealed that, after making contact with the prey, the muricids would hold the mussels with the foot and orientate them so that the byssal gape was innermost, presumably so that the radula and Accessory Boring Organ could be applied in this area. Physical manipulation of prey by muricid gastropods has not been previously described. Many ornamented and unornamented prey were held in this way, suggesting that chemical stimuli, which have been shown to be a causal factor in attracting muricid gastropods to potential victims (Kohn 1961 ; Pratt 1974; Carriker 1981 ; Cross 1983; Williams et al. 1983), had not been impaired by the experimental technique, and the rejection of spinose prey was a result of behavioural responses provoked by unfavourable subsequent tactile stimulus. Large numbers of unornamented prey were consumed, despite being dotted with epoxy, and therefore it is concluded that the presence of spines alone is responsible for inhibiting predatory activity. 1058 PALAEONTOLOGY, VOLUME 41 table 3. Number of bore holes made in different areas of the shell of Chama reflexa by the muricid gastropods Chicoreus microphyllus and Thais clavigera in three replicate experiments. Replicate A B C Valve Left Right Left Right Left Right Umbonal 0 0 1 Median (ornamented) — 0 — 0 — 0 Median (unornamented) 1 10 1 5 0 10 Inter-lamellar 2 — 4 — 4 — Area of attachment 7 — 12 — 5 — Total 10 10 17 5 9 11 text-fig. 3. Generalized representation of the bivalve Chama reflexa offered to Chicoreus microphyllus and Thais clavigera in Experiment 2. a, right valve with crenulate marginal lamella, divided into three areas: (1), umbonal; (2), median (ornamented); (3), median (unornamented). Dotted area indicates spinose valve surface. b, left valve, divided into two areas: (4), inter-lamellar; (5), area of attachment. Hatched area indicates flat attachment scar. Scale bar represents 2 mm. Experiment 2 Chicoreus microphyllus and Thais clavigera versus Chama reflexa. The results are summarized in Tables 2 and 3. There was no significant difference between the numbers of ornamented and unornamented prey bored, and the null hypothesis that equal numbers of each should be consumed cannot be rejected with chi square analysis at a 95 per cent, confidence level. Upon closer examination, however, some important observations may be made. Table 3 shows the number of times specified areas of the shell were bored successfully for both left and right valves. Five areas are described (Text-fig. 3). For the right, or ‘upper’ valve, these are: umbonal, median (ornamented) and median (unornamented) (Text-fig. 4c). For the left, or Tower' valve these are: inter-lamellar and area of attachment. Of the ornamented C. reflexa prey, the sculpture on certain areas of the surface of the right valve may be less expressed in some individuals than in others. Bore holes made at these sites are described as median (unornamented). The natural ornament may also have been eroded from the umbonal region during ontogeny, resulting in possible reduced defence in this area (Text-fig. 3a). For the left valve, boring was not observed to have occurred through the STONE: PREDATOR DETERRENCE BY BIVALVES 1059 attachment lamellae of the left valve, but only in the commarginal inter-lamellar spaces between adjacent elements (Text-fig. 3b). Table 3 reveals that on no occasion was a bore hole successfully made in a truly ornamented area of a prey shell. The majority of bore holes were either made in the area of attachment of the left valve or unornamented areas of the right valve. In the chamids’ natural habitat, boring of the left valve is extremely unlikely to occur unless the bivalve has become forcibly detached, and it is not surprising that the muricids should attack unornamented areas of left valves in the aquarium, in particular the potentially weak area of attachment. However, although the majority of left valves were thinnest in this area, the fact that other areas of both left and right valves were bored, including inter-lamellar areas and the relatively thick right valves, suggests that the degree of ornament expression is the primary determinant of the choice of bore hole site rather than valve thickness. It seems reasonable to suggest that pronounced marginal ornament exhibited by potential prey will additionally afford a degree of protection from boring near the valve edges, in much the same way as the artificial marginal spines of S. virgatus in Experiment 1. It may be concluded that the shell spines of lamellae of C. reflexa offer a degree of defence against boring by muricids. table 4. Numbers of unornamented and artificially spinose Perna viridis consumed by the extraoral feeding asteroid Coscinasterias acutispina in three replicate experiments. Statistical analysis using chi square. Assuming a confidence level of 95 per cent., the null hypothesis that equal numbers of ornamented and unornamented prey are consumed cannot be rejected. Replicate A B C Ornamented prey consumed 25 17 46 Unornamented prey consumed 28 24 41 Overall total 53 41 87 P >> 0 05 ^ 0 05 > 0 05 Experiment 3 Coscinasterias acutispina versus Perna viridis. The results are summarized in Table 4. They reveal that there was no significant difference in the numbers of spinose and non-spinose mussel prey consumed. The null hypothesis that equal numbers of each should be eaten could not be rejected using chi square analysis at a 95 per cent, confidence level. The larger number of bivalves eaten in replicate C can probably be attributed to the larger mean size of the starfishes available for this replicate. In the aquarium, the feeding behaviour of C. acutispina is typical of other extraoral feeding asteroids. It assumes a hunched posture over the prey bivalve so that the stomach lobes can be extruded into the shell after prising the valves apart. For mussel prey, the stomach lobes may be extruded into the body cavity through the narrow byssal gape. This has been reported for other genera (Feder 1955; Lavoie 1956; Carter 1968). In contrast to the experiments with muricid borers presented above, C. acutispina showed no statistically significant preference for the non-spinose mussels offered. In general, the predatory behaviour of the starfish did not differ when it was confronted by ornamented or unornamented prey. Such behaviour was characterized by an initial encounter with the prey and subsequent manipulation so that the ventral margins, and in particular the byssal gape, were orientated towards the oral region of the predator. In this experiment, the spines of some of the artificially ornamented mussels had been removed by the predators. It is, therefore, probable that the ornament was removed during the process of manipulation or possibly prising of the valves, although the latter was not directly observed. Attached marginal spines were up to one-third the length of the axis of greatest growth in any one individual and, because they 1060 PALAEONTOLOGY, VOLUME 41 text-fig. 4. Scanning electron micrographs of muricid bore holes in prey bivalve shells, a, hole bored by Thais luteostoma through posterior extremity of the byssal gape of Septifer virgatus', x 43. B, single occurrence of a bore hole made by Thais luteostoma between intact artificial ‘spines' (seen left and right in picture) attached to Septifer virgatus', x 33. c, bore hole made by Chicoreus microphyllus in median area of a left valve of Chama reflexa lacking macro-ornament; x 20. D, bore hole made by Morula musiva in an unornamented area between artificial ‘spines’ attached to the lateral surface of a Perna viridis shell; x 45. Material in author’s possession. interdigitated across the plane of the commissure in those mussels whose spines remained intact throughout the feeding process, and would therefore still provide an effective barrier, it is likely that prising the valves apart a short distance would be ineffective in itself. Thus, the stomach lobes may have been extruded to at least the same length as the spines themselves in order to reach the body cavity. The above evidence reveals that C. acutispina is adept at dealing with different types of prey and has no difficulty in overcoming the artificially defended bivalves in this experiment. Experiment 4 Morula musiva versus Perna viridis. The results of this choice experiment are summarized in Table 5. The muricid predators chose unornamented mussels as a first choice in three out of the six replicates and as a second choice in two replicates. Highly ornamented prey with 4 mm long spines were chosen only in third place in two replicates. Complete consumption of the other sculptural types, 1 mm and 2 mm spines, was more variable. The results, therefore, suggest that prey with less expressed ornament are preferred to those that are highly spinose (Text-fig. 4d). STONE: PREDATOR DETERRENCE BY BIVALVES 1061 table 5. Results of Experiment 4, in which the muricid Morula musiva was offered a choice of Perna viridis prey: unornamented (NS), with 1 mm long artificial spines, with 2 mm artificial spines or with 4mm long artificial spines in six replicate experiments. Unornamented prey were chosen as a first choice in three out of six replicates, and as a second choice in two replicates. Replicate A B C D E F First choice 1 mm 2 mm NS 2 mm NS NS Second choice NS — — NS 1 mm — Third choice 4 mm — — 4 mm 2 mm — Fourth choice — — — 1 mm — — Total prey consumed 21 11 12 20 22 10 text fig. 5. Positions of all bore holes made by Morula musiva in right (a, c, e, g) and left (B, D, F, H) valves of the mussel Perna viridis. Filled circles: holes made in non-attached ‘free' valves. Hollow circles: holes made in artificially attached valves. Umbones are towards the centre of the page, ahb, holes made in valves without artificial ornament but dotted with cyanoacrylate as a control, c-d, holes made in valves with 1 mm long artificial ‘spines’ attached, e-f, holes made in valves with 2 mm long artificial ‘spines’ attached, g-h, holes made in valves with 4 mm long artificial ‘spines’ attached. Scale bar represents 5 mm. Since it was necessary to allow the mussels to function as near normally as possible in this experiment, all prey were cemented for only a portion of the relevant valve so that the valve margins were raised clear of the artificial substrate. This permitted some boring in the attached valve, especially in the area of the umbo. Table 6 reveals that the percentage of bore holes made in attached valves increases from 9-7 per cent in non-spinose mussels to 40-0 per cent, in prey with 4 mm spines. This is additional evidence of the deterrent value of increasing shell ornament. The positions of all bore holes, for both attached and free valves, are shown in Text-figure 5. It is inferred from this experiment, therefore, that the primary effect of pronounced spines on the shells of epifaunal bivalves is that of hindering direct access to the general shell surface. 1062 PALAEONTOLOGY, VOLUME 41 table 6. Bore hole analysis of the four types of artificially ornamented Perna viridis drilled by the muricid Morula musiva. 1 mm 2 mm 4 mm Ornament type Non-spiny spines spines spines Percentage bore holes in attached valve 9-7 27-3 37-5 400 DISCUSSION The biotic changes associated with the so-called Mesozoic Marine Revolution (M.M.R.) have been identified by Vermeij (1977, 1978, 1987) as a probable causal factor in the subsequent evolution of strengthening armour in gastropods and bivalves. The M.M.R. was characterized by an increase in the importance of a host of predatory methods and resulted not only in an increase in the development of prey armour in potentially vulnerable molluscan epifaunal taxa but also an expansion of groups escaping predators in other ways, such as those colonizing the infaunal realm (Stanley 1968; Vermeij 1978, 1987). Predatory methods, and the potential consequences of the M.M.R. for molluscan prey, have been assessed by Vermeij (1977, 1987) and Harper and Skelton (1993u). Epifaunal bivalves are particularly vulnerable to three predatory strategies: (1) 'insertion and extraction’, by extraoral feeding asteroids, prising gastropods, some birds and fish, octopods, arthropods and crustaceans; (2) ‘crushing’, by many crustaceans, fish, arthropods and some molluscs; and (3) ‘boring’, by muricid and naticid gastropods and some cephalopods. It has been proposed that these three strategies became dramatically more important as a result of the M.M.R. (Vermeij 1977, 1987). The experiments presented here were undertaken to examine the effects of two such predatory strategies: shell boring by muricid gastropods and extraoral feeding by asteroids. Both predatory groups have often been held responsible for severe predation impact on epifaunal molluscs, particularly bivalves, in subtropical-tropical and temperate shallow water environments respectively (Galtsoff and Loosanoff 1939; Hancock 1955, 1958; Taylor 1976, 1978), and thus they are particularly suitable for a study of the anti-predatory effects of epifaunal bivalve shell ornament. Cemented bivalves in particular may present themselves as potential sitting targets to muricid gastropods (Harper and Skelton 1993 a). Of the principal epifaunal cemented taxa, warm-water oysters, spondylids and chamids display a high degree of morphological plasticity, associated with the habit of cementation, and an ability to construct shells with pronounced spines and/or commarginal lamellae. The results presented here suggest that such ornament in prey taxa is itself directly effective in deterring boring by muricid gastropods. For both chamid and mytilid prey, unornamented areas of the shell are preferentially bored over ornamented areas in each of the three experiments conducted with muricid predators. Where behavioural adaptation seems to have resulted in a high degree of bore hole site specialization, for example in relation to edge-boring by Thais luteostoma in Experiment 1, any subsequent site modification that might have been expected when confronted by artificially ornamented prey is lacking. Thus, T. luteostoma does not attack the unornamented valve surfaces of Septifer virgatus prey and, rather than choosing new bore hole sites in mussels with marginal ornament, this predator takes advantage of the preferential site in unornamented prey (in this case the byssal gape). As new shell ornament of potential prey is formed at the commissure, it is less likely to have been eroded than sculpture formed earlier in ontogeny which cannot be repaired, and predators which attack the valve margins are likely to be at a competitive disadvantage in comparison with those that choose the bore hole site more liberally. This may help to explain why edge-boring of bivalves by muricid gastropods in the field has only rarely been reported (e.g. Morton 1994). STONE: PREDATOR DETERRENCE BY BIVALVES 1063 In contrast to the muricid predators, the subtropical extraoral feeding asteroid Coscinasterias acutispina is not measurably deterred by artificial spines fixed to mussel prey, even when such spines are positioned around the entire commissure, including the byssal gape. It is through a commissural gape that the stomach lobes of the asteroid are extruded. The gape may be formed mechanically by prising the valves of the prey apart a small distance or it may be a natural element of shell growth. Prising was not directly observed in the current experiment and, as prey were orientated with the byssal gape nearest the oral region of the predator, it was concluded that the stomach lobes were being extruded through this region. In a pilot study, a group of predators were removed from prey during feeding with their everted stomach lobes clearly visible. Despite the presence of large artificial marginal spines on Perna viridis , C. acutispina is clearly adept at dealing with such ‘ difficult ’ prey. It is conceivable that, if spines had been arranged over the general surface of the valves, including the commissure, manipulation of prey by the asteroid might have been more problematical. However, artificial ornament strong enough to withstand such handling could not be attached successfully on the valve surface. Moreover, as stated previously, it is the commissure of the prey bivalve that is most often heavily defended, as this is where new shell material is laid down, and thus it is the ability to extrude stomach lobes through a heavily defended commissure that is of greatest interest. Although naturally cemented bivalves do not present extra-oral feeding asteroids with natural gapes through which stomach lobes can be extruded, it is likely that prising the valves apart a small distance will create a temporary gape, and thus a potentially vulnerable area. In Experiment 3, valve prising was not directly observed but may not have been required for extra-oral feeding because of the presence of the byssal gape in the Perna viridis prey offered. The methodology of the present study is therefore considered satisfactory. It has been stated by some authors, such as Vermeij (1987), that extraoral feeding asteroids are not in general a common component of the benthic megafauna on tropical and subtropical rocky shores. In contrast, muricid gastropods are often abundant and likely to exert strong predatory pressure (Taylor 1976, 1977, 1978). In these areas, therefore, the threat from muricids is potentially far more important than that of extraoral feeding asteroids. The impact of the extraoral method of attack, however, may be particularly severe in cooler waters, where asteroids often aggregate in large numbers and may devastate local epifaunal bivalve populations (Galtsoff and Loosanoff 1939 ; Hancock 1955, 1958). However, in cooler waters highly ornamented bivalve taxa are rare or non- existent (Nicol 1964, 1965; Harper and Skelton 1993a). It seems highly doubtful, therefore, that the presence of marginal ornament in epifaunal bivalves is functional against the extraoral method of attack, and this proposal is strengthened by the results of the experiment presented here. To determine whether pronounced shell ornament in cementing epifaunal bivalve families, such as the unrelated Spondylidae and Chamidae, arose as an adaptation to the threat posed by muricid gastropod borers, it is necessary to demonstrate agreement between the timing of the onset of the adaptive radiation of muricids in the Albian (Taylor et al. 1980, 1983) on the one hand and the appearance of highly ornamented potential prey taxa on the other. Harper and Skelton have documented the numbers of spinose and non-spinose species of the family Spondylidae from the Jurassic to the present day, and show a general increase in the percentage of spinose species over time (Harper and Skelton 1993a, fig. 3). My own unpublished data concur with their results. Despite this increase, however, it is apparent that the rise of spinose ornamentation in Bivalvia predates the adaptive radiation of the muricid gastropods. The ability to construct ornament was already evident in some late Palaeozoic Pectinoida, such as members of the family Pseudomonotidae. It has been proposed that the Spondylidae most probably arose from this family (Newell and Boyd 1970). In their work on Permian ostreiform Bivalvia, Newell and Boyd suggested that the morphological series: Pseudomonotis (Pseudomonotis)-Pseudomonotis ( Trenmti - concha)-Prospondylus-Paleowaagia-Newaagia-Spondylus- may represent a phyletic series. Within this series, a number of Permian species display spines and/or scales, most notably Pseudomonotis ( Trematiconcha ) wandageensis Newell and Boyd, P. likharevi Newell and Boyd, Prospondylus acinetus Newell and Boyd and Paleowaagia cooperi Newell and Boyd (Newell and Boyd 1970, figs 14a, 16a, 23a-g, 24a-e, 28a, 29b, g). Waller (1978) has suggested that spondylids may have evolved 1064 PALAEONTOLOGY, VOLUME 41 from the superfamily Pectinoidea, rather than from the Pseudomonotidae. Despite the inherent difficulties associated with the assessment of phylogenetic affinities of many pteriomorph groups, ornament construction was nevertheless underway in the pectinoidan families Pseudomonotidae and Aviculopectiniae (for example, Girtypecten and Clavicosta; see Newell 1969), by the Late Palaeozoic. It is, therefore, unlikely that spinose ornamentation in pteriomorph bivalves arose initially as an adaptation to the muricid borers. It is also difficult to relate ornament to direct deterrence of other predatory methods, such as shell breakage by crustaceans. It is intuitively obvious that, to be effective against the latter group, shell spines and scales would need to be particularly well-developed, stout and strong, especially around the most vulnerable area, the commissure. Evidently, sculptural development in Permian Pectinoida, although present in some genera, was only weakly expressed. Shell ornament in late Mesozoic and particularly Tertiary spondylids is very different (Zavarei 1973; pers. obs.). It seems reasonable to suggest that the faunal changes resulting in a rise in predation pressure concomitant with the M.M.R. was responsible for the adaptive development of large, strong spines that characterize many Mesozoic and Tertiary taxa. Such ornament may be effective against a variety of predators, but it proposed here that muricid borers are potentially one of the most damaging, and potentially the most deterred by pronounced shell ornament, of all predatory groups. The family Chamidae, unrelated to the Spondylidae but again adopting the cemented habit, probably evolved as a branch of the byssate family Carditidae in the Late Cretaceous (Kennedy et al. 1970) and the pronounced shell ornament displayed by the former group may, conceivably, have arisen as a primary adaptation against shell boring predators, especially in view of the evolution of the cemented habit from a byssate precursor. Ornament in extant ribbed carditids, where present, tends to consist of nodes or pronounced scales, but further studies need to be undertaken on the ornament of Palaeozoic and Mesozoic carditids before the geological history of ornament in the Carditidae and Chamidae can be adequately assessed. The evidence, however, seems to suggest that ornamentation of the ‘upper’ free valve of epifaunal groups evolved initially as a response to some factor other than direct predatory pressure, and may have become coincidentally advantageous under the harsh predatory regimes characterizing the M.M.R. (Vermeij 1977, 1987), particularly with respect to shell boring by muricid gastropods. The original primary function may or may not have been lost and, if the latter is the case, then shell spines and lamellae in extant epifaunal bivalve taxa could be considered multifunctional. There are several possible hypotheses that may be put forward regarding the primary function of pronounced ‘upper’ valve shell ornament other than direct predatory inhibition which is the subject of the current study. Firstly, ornament may act to prevent the bivalve from becoming dislodged from cryptic microhabitats such as crevices. This does not apply to cementing taxa such as oysters, spondylids and chamids, although cementation to the sides of crevices may occur. However, some byssate taxa of the families Tridacnidae and Carditidae carry spine-like processes on both valves that may help to prevent dislodgement. The second hypothesis is that ornament may help to alleviate erosion. This applies potentially to both attached and unattached taxa. It has, however, never been satisfactorily demonstrated that the presence of ornament serves to increase shell strength in any way. Moreover, the spines of many widely distributed ornamented genera such as the semi-infaunal Pinna and Atrina, and certain members of the Spondylidae, such as Spondylus linguafelis Sowerby, are often elaborate and delicate and it is difficult to accept such an hypothesis. Erosion and endolithic boring of the shell by organisms such as clionid sponges can remove ornament from all or part of the shell surface and thus render the bivalve vulnerable to predation by muricids or crustaceans if the strength of the valve has been severely weakened, but if the ornament remains intact, it is likely the bivalve will be far more resistant to gastropod borers. Where chamids and spondylids are cemented to rocks in exposed areas, erosion of the shell may be extensive. Bernard (1976) stated that chamids living in such environments are large, thick-shelled with little ornamentation and very shallow in vertical distribution. In these areas, however, the intensity of predation is likely to be considerably less than at sheltered sites, especially for roving predators, such as muricids, that may be dislodged by wave action. In sheltered habitats, the STONE: PREDATOR DETERRENCE BY BIVALVES 1065 predation pressure is likely to be more intense and in these areas the ornament of chamids is often highly pronounced (Bernard 1976). The third hypothesis is that ornament may offer protection against the rasping activities of certain roving grazers, for example regular echinoids, feeding on algae growing on shells that may, over time, erode the shell and weaken it. The potential importance of grazing with respect to shell ornament is at present unclear. The fourth hypothesis considers spines and commarginal lamellae acting as stabilizing structures that prevent epifauna lying on soft substrates from being smothered. The spines of the extant chamid genus Arcinella may act in this way. This bivalve gains secondary freedom from an initial cemented phase early in ontogeny and rests on sandy or shelly substrates (Nicol 1952). The fifth hypothesis suggests that ornament may act to attract growth of epibionts on the shell, presumably by increasing the surface area of available substrate (Vance 1978 ; Feifarek 1987). Personal observations of the spondylid Spondylus americanus Hermann transferred to aquaria show that the colonizers are often algae, sponges, solitary corals, tube worms and hydroids. Epibiont growth may have a number of potentially beneficial effects. For example, biomineralizing colonizers may increase the effective strength of the shell and help to reduce erosion. Alternatively, epibionts may directly disguise the shell from visually hunting predators or may mask chemical cues released into the ambient environment by the bivalve which may serve as attractants for chemosensitive predators. There is good evidence that predatory gastropods are attracted by such cues (Carriker 1981). Some extant representatives of the families Spondylidae and Chamidae, for example Spondylus linguafelis and Chama lazarus Linnaeus, possess spines whose distal ends are highly intricate or enlarged into spatulate lamellae. It is conceivable that the consequent large surface area of such processes is particularly attractive to epibiont settlement. Such growth may provide serendipitous defence for sessile organisms such as cementing bivalves. The sixth hypothesis concerns shell ornament as structural supports for areas of sensory mantle tissue (Rudwick 1965; Kauffman 1969; Stenzel 1971). A mechanosensory or chemosensory capability may have several advantages. For example, it may provide the bivalve with an early warning system when under potential threat from an approaching predator, or it may warn against inclement environmental conditions, such as changes in turbidity or salinity. The intricate nature of the ornament of some members of the Spondylidae and Chamidae, referred to above, may indicate a corresponding intricacy of underlying mantle epithelium performing sensory functions. However, a sensory capability has only been suggested by analogy with the soft parts of brachiopods (Rudwick 1965), and comparative morphological evidence to substantiate this hypothesis in the Bivalvia has been lacking. Research is currently being undertaken by the author with regard to the hypothesized functions of pronounced shell ornament detailed above. The formation of shell ornament in epifaunal bivalves can only proceed within the intrinsic constraints of the bivalve Bauplan and extrinsic constraints such as water temperature, food availability, etc. Many epifaunal taxa have demonstrated a remarkable ability to radiate morphologically, particularly in warm waters. Nicol (1965, 1967) has stated that there are very few ornamented bivalves inhabiting cool waters of high latitudes and the deep sea. What ornament is evident, is always subdued. In addition, bivalves that adopt the cemented habit are absent from the Arctic and Antarctic (Nicol 1964). In warm temperate, subtropical and tropical waters many cemented taxa, such as members of the Spondylidae, Chamidae and Ostreidae, are particularly adept at ornament construction. It is suggested here that the sculptural radiation of highly ornamented epifaunal taxa may be a result of pressures imposed by the shell boring muricids from the Albian onwards, resulting in adaptive radiations in potentially vulnerable prey lineages. Whilst the nature and intensity of predation is evidently very different in cool water, the pre-adaptations that may have otherwise permitted the formation of sculpture, such as an ability to produce extensive growth in periodic rapid phases, may also be absent. Feifarek (1987) found that the rate of growth of spines in the spondylid Spondylus americanus is extremely rapid, about 1 mm per day. In addition, Paul (1981) has suggested the shell of the highly spinose warm water neogastropod Murex ( Murex ) pecten Lightfoot grows by rapid episodic incrementation. It is likely that such modes of growth can only be achieved in warm water because of high growth rates which may be associated with a combination of warm water temperatures as well as an adequate food supply. 1066 PALAEONTOLOGY, VOLUME 41 although the possibility that the effects of cold water enhance the dissolution of biogenic calcareous structures, retarding or preventing growth of pronounced ornament, must also be considered. SUMMARY Pronounced shell ornament in epifaunal bivalves has been shown to be effective at deterring shell boring muricid gastropods in aquarium experiments. No apparent inhibitory effect was observed for extraoral predation by asteroids. Gastropod deterrence may have become increasingly important with the adaptive radiation of the muricids in the Albian. Anti-boring defence may not be the sole function of pronounced shell ornament and it is likely that it arose initially for other reasons, in the family Spondylidae at least. Other hypothesized functions include, for example, attracting epibiont growth or outposts for sensory mantle tissue. These may have continued to play an important role for the epifaunal bivalve to the present day, despite the value that ornament has been shown to confer in deterring muricids. Acknowledgements. I am indebted to Dr Elizabeth Harper at the Department of Earth Sciences, University of Cambridge, for discussions concerning the manuscript and to Professor Brian Morton at the Swire Institute of Marine Science, Hong Kong, for use of aquarium facilities. This work forms part of a research studentship funded by the N.E.R.C., ref. GT4/94/136/G. Additional funding was provided by Gonville and Caius College, Cambridge. This is Cambridge Earth Science Publication no. 5263. REFERENCES ansell, A. d. 1969. Defensive adaptations to predation in the Mollusca. 487-512. In Proceedings of the Symposium on Mollusca , Cochin , 2. Marine Biological Association of India, Mandapam, India. — and morton, b. 1983. Aspects of naticid predation in Hong Kong with special reference to the defensive adaptations of Bassina ( Callanaitis ) calophylla (Bivalvia). 635-660. In morton, b. and dudgeon, d. (eds). Proceedings of the 2nd international workshop on the malacofauna of Hong Kong and Southern China , Hong Kong 1983. Hong Kong University Press, Hong Kong, 681 pp. - 1987. Alternative predation tactics of a tropical naticid gastropod. Journal of Experimental Marine Biology and Ecology, 111, 109-119. Bernard, F. R. 1976. Living Chamidae of the Eastern Pacific (Bivalvia: Heterodonta). Contributions in Science, Natural History Museum of Los Angeles County, 278, 1 43. carriker, M. R. i981. Shell penetration and feeding by naticacean and muricacean gastropods: a synthesis. Malacologia, 20, 40 3 — 422. carter, r. m. 1968. On the biology and palaeontology of some predators of bivalved Mollusca. Palaeogeography , Palaeoclimatology, Palaeoecology, 4, 29-65. cox, l. r. 1969. General features of Bivalvia. N2-N129. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part N. Mollusca 6 (7). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 1224 pp. croll, r. p. 1983. Gastropod chemoreception. Biological Reviews, 58, 293-319. day, r. w., barkai, A. and wickens, p. a. 1991. Trapping of three drilling whelks by two species of mussel. Journal of Experimental Marine Biology and Ecology, 149, 109-122. feder, h. m. 1955. On the methods used by the starfish Pisaster ochraceus in opening three types of bivalve molluscs. Ecology, 36, 764-767. feifarek, b. p. 1987. Spines and epibionts as antipredator defenses in the thorny oyster Spondylus americanus Hermann. Journal of Experimental Marine Biology and Ecology, 105, 39—56. galtsoff, p. s. and loosanoff, v. 1939. Natural history of Asterias forbesi. Bulletin of the U.S. Bureau of Fisheries, 49, 79-132. Hancock, d. a. 1955. The feeding behaviour of starfish on Essex oyster beds. Journal of the Marine Biological Association of the U.K., 34, 313-331. 1958. Notes on starfish on an Essex oyster bed. Journal of the Marine Biological Association of the U.K., 37, 565-589. harper, e. m. 1991. The role of predation in the evolution of cementation in bivalves. Palaeontology, 34, 455-460. STONE: PREDATOR DETERRENCE BY BIVALVES 1067 — 1994. Molluscivory by the asteroid Coscinasterias acutispina (Stimpson). 339-355. In morton, b. (ed. ). The malacofauna of Hong Kong and Southern China III. Proceedings of the 3rd international workshop on the malacofauna of Hong Kong and Southern China, Hong Kong 13 April-1 May 1992. Hong Kong University Press, Hong Kong, 504 pp. — 1997. The molluscan periostracum : an important constraint in bivalve evolution. Palaeontology, 40, 71-97. — and skelton, p. w. 1993a. The Mesozoic Marine Revolution and epifaunal bivalves. Scripta Geologica, Special Issue, 2, 127-153. 19936. A defensive value of the thickened periostracum in the Mytiloidea. The Veliger, 36, 36-42. jackson, j. b. c. 1977. Competition on marine hard substrata : the adaptive significance of solitary and colonial strategies. American Naturalist, 111, 743-767. kauffmann, E. G. 1969. Form, function and evolution. N129-N205. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part N. Mollusca 6 ( 1 ). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 1224 pp. Kennedy, w. J., morris, N. J. and taylor, J. d. 1970. The shell structure, mineralogy and relationships of the Chamacea (Bivalvia). Palaeontology, 13, 379—4 1 3. kohn, a. J. 1961. Chemoreception in gastropod molluscs. American Zoologist, 1, 291-308. lavoie, m. E. 1956. How seastars open bivalves. Biological Bulletin, 111, 1 14—122. logan, a. 1974. Morphology and life habits of the Recent cementing bivalve Spondylus americanus Hermann from the Bermuda Platform. Bulletin of Marine Science, 24, 568-594. morton, b. 1994. Prey preference and method of attack by Rapana bezoar (Gastropoda: Muricidae) from Hong Kong. 339-355. In morton, b. (ed.). The malacofauna of Hong Kong and Southern China III. Proceedings of the 3rd international workshop on the malacofauna of Hong Kong and Southern China, Hong Kong 13 April-1 May 1992. Hong Kong University Press, Hong Kong, 504 pp. Newell, n. d. 1 969. N335-N338. In moorf, r. c.(ed.). Treatise on invertebrate paleontology. Part N. Mollusca(l). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 1224 pp. — and boyd, d. w. 1970. Oyster-like Permian bivalvia. Bulletin of the American Museum of Natural History, 143, 219-281. nicol, D. 1952. A revision of the pelecypod genus Echinochama. Journal of Paleontology, 26, 803-817. — 1964. Lack of shell attached pelecypods in Arctic and Antarctic waters. The Nautilus, 11, 92-93. 1965. Ecologic implications of living pelecypods with calcareous spines. The Nautilus, 78, 109-116. — 1967. Some characteristics of cold-water marine pelecypods. Journal of Paleontology, 41, 1330-1340. PAUL, c. R. c. 1981. The function of the spines in Murex (Murex) pecten Lightfoot and related species (Prosobranchia: Muricidae). Journal of Conchology, 30, 285-294. petraitis, p. s. 1987. Immobilization of the predatory gastropod, Nucella lapillus, by its prey, My t i/us edidis. Biological Bulletin, 172, 307-314. pratt, d. M. 1974. Attraction to prey and stimulus to attack in the predatory gastropod Urosalpinx cinerea. Marine Biology, 27, 37—45. RUDWiCK, m. j. s. 1965. Sensory spines in the Jurassic brachiopod Acanthothiris. Palaeontology, 8, 604-617. Stanley, s. m. 1968. Post-Paleozoic adaptive radiation of infaunal bivalve molluscs a consequence of mantle fusion and siphon formation. Journal of Paleontology, 42, 214-229. — 1970. Relation of shell form to life habits of the Bivalvia (Mollusca). Memoir of the Geological Society of America, 125, 1-296. — 1981. Infaunal survival: alternative functions of shell ornamentation in the Bivalvia (Mollusca). Paleobiology , 7, 384—393. — 1988. Adaptive morphology of the shell in bivalves and gastropods. 105-141. In trueman, e. r. and Clarke, M. r. (eds). The Mollusca, Volume 2. Form and function. Academic Press. stenzel, h. b. 1971. Oysters. N953-N1224. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part N. Mollusca (i). Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 1224 pp. taylor, J. d. 1976. Habitats, abundance and diets of muricacean gastropods at Aldabra Atoll. Zoological Journal of the Linnean Society, 59, 155-193. — 1977. Diets and habitats of shallow water predatory gastropods around Tolo Channel, Hong Kong. 163-180. In morton, b. (ed.). Proceedings of the 1st international workshop on the malacofauna of Hong Kong and Southern China, 23 March-8th April, 1977. Hong Kong University Press, Hong Kong, 345 pp. 1068 PALAEONTOLOGY, VOLUME 41 - 1978. Habitats and diet of predatory gastropods at Addu Atoll, Maldives. Journal of Experimental Marine Biology and Ecology , 31, 83-103. — 1990. Field observations of prey selection by the muricid gastropods Thais clavigera and Morula musiva feeding upon the intertidal oyster Saccostrea cucullata. 837-855. In morton, b. (ed.). The marine flora and fauna of Hong Kong and Southern China. Proceedings of the 2nd international marine biological workshop , Hong Kong , 1986. Hong Kong University Press, Hong Kong, 1268 pp. — morris, N. J. and taylor, c. n. 1980. Food specialization and the evolution of predatory prosobranch gastropods. Palaeontology , 23, 375-409. - cleevely, r. j. and morris, n. j. 1983. Predatory gastropods and their activities in the Blackdown Greensand (Albian) of England. Palaeontology, 26, 521-533. vance, R. R. 1978. A mutualistic interaction between a sessile marine clam and its epibionts. Ecology, 59, 679-685. vermeij, G. J. 1977. The Mesozoic Marine Revolution: evidence from snails, predators and grazers. Paleobiology, 2, 245-258. — 1978. Biogeography and adaptation. Patterns of marine life. Harvard University Press, Cambridge, Mass., 332 pp. - 1987. Evolution and escalation. An ecological history of life. Princeton University Press, Princeton, New Jersey, 527 pp. waller, T. R. 1978. Morphology, morphoclines and a new classification of the Pteriomorphia (Mollusca: Bivalvia). Philosophical Transactions of the Royal Society of London, Series B, 284, 345-365. watters, G. T. 1993. Some aspects of the functional morphology of the shell of infaunal bivalves (Mollusca). Malacologia , 35, 315-342. wayne, T. a. 1987. Responses of a mussel to shell-boring snails: defensive behaviour in Mytilus edulisl The Veliger, 30, 138-147. williams, l. g., rittschof, d., brown, b. and carriker, m. r. 1983. Chemotaxis of oyster drills Urosalpinx cinerea to competing prey odors. Biological Bulletin, 164, 536-548. Wilson, j. G. 1979. What is the function of the shell ornamentation of Tellina fabula Gmelin? Malacologia , 18, 291-296. zavarei, a. 1973. Monographic des Spondylidae (lamellibranches) actuels et fossiles. Centre a> e s e s 73 73 73 73 a> C/5 C/5 C/5 C/5 g a G G o o o o D. £ a o- 00 G O g a g £ s o o g G 0) s 73 a G C O S g o u- *> g w D o-> a> a> o o o o =3 G O O o O 2 ■5b'5b'53j'5xj g ’5b "5b § GGGG^jGGq 13 ’y ”5 G G G G O O O O bij O G O G O G O G O X o o o o G o G o o ;>* 73 G G a a a a-g' a a g -O c c y ^ ■o -rf T3 "O T3 T3 X, •O 2? "2 -O -2 "2 ' G G G G --H d u u y J--* n> 77 ,_Ei ,P 5 77 ^ GO C/5 (/5 on C/5 on — i .— ■ i—i —< —< 1 — 1 X 13 13 13 13 73 s G O S-H > G a> >» 0D o "o bU jO o -G C/5 C/5 s s ° 2 g 5 0 >0 P*> >0 >C >5,^2? 03 03 03 03 03 03 [2 12 G G .. GGGGGGGG.^ DuuuD0022q2q2q2 2? 6 2 o X3 G G £ (D O N 4-J O G O -o £ 21 o ^ T3 3 3 O U 't'OO'too-HhNro-Hoo't^mio -Hrj-roTtrcimoooor'Xr'r-'ON autotrophic species; [H] = heterotrophic species. 1102 PALAEONTOLOGY, VOLUME 41 o U i4 fN — ' o u u, O' a g 3 3-s is d Oh C/3 £< S'T 5 •S £ •§ 3 .§ § .§ < ^ 3 ^ ^ ^ “ bo^S bo""^ bo a bo Yi <3 =3 a ^ a 2 3 ' ‘ C3 «0 ^ -3 3 uH. Co f.s <£ Q lP o 2 P\^r> ~ ^ o ^ £ fc ~9 <1> 3 cl5 T3 -^ s 3 'S .*> .5 hi C/3 .3 "c/3 .« >%~s ° o d £ d « 3-5, .2 §■ ■ S £ ' T3 tu 'C ^ , , ctf < "S e , 61; . § ■5 k , — , c a t £ £ £ ^ £ g _. ■« S -« u -c •5 5 c c ’C , g- 3 g d T3 _ S 3 o, S a d,— lS.g2SSci,gc-’2(i 6«K?pOftOX?0. f ^ 5^ -£ »-*■$ “> u ° s .3 •§ 4-J “3 'g-g §•'" X T3 -5 l ^ O C 3 = £ 'I : s .S -2 c 3^. oo O «-s 3 ~ 3 .3 3 .3 •S 3 -5 5 .5 C .§ 3 •§ ?3 S?3 2^! 333'CT3 K>.§ Sb^S ~ ~ ~ ~ " 3 S> §)■>( §3 B< cL S’ s 5 « ,5 ^ -5 CU W >3 -r a <$ S .g '£ 2 >s^' 5 -c> .R < O vr ^ 2* 3 3 > 3 ’3 O ’3 , -5 *J ■ 1- -3 ! "3 o ON 'sD O 1 ’ 0* Q £ £ 5 .5 § 5 •5 ^ 3 § ■f & ^ is 5.3 § -S bo d ^ a S .3 .3 S .3 s .5 ~ £ 3 £ S "3 § -<3 'c -« Hi i-sii W 60 § .R 'C il 1 o .2 a Cl, ° © a £ $ 3 Its • £ 3 © T) > ^3 •c © N3 1) X) '3 "O „ li CT3 08 o _i 0 6 0 4 0.2 0 0 Location Sites *♦' Log H A Ratio text-fig. 7. Graph of the heterotrophic ratio along the transect from the Weddell Sea (left) to Falkland Trough (right). Note the general increasing trend away from the Weddell Sea. the global circulation pattern and providing a site of upwelling adds to the significance of this the first study of dinoflagellate cysts in the area of remit of the British Antarctic Survey. However, it is pertinent to compare our results with those of Marret and de Vernal (1997) from the southern part of the Indian Ocean in the immediate vicinity of Antarctica. Using Principal Component Analysis on a data-set of 44 samples, Marret and de Vernal (1977) recognized two main groupings each of which contains three identified assemblages. Their most southerly group, a circum-Antarctic domain, contains identified dinoflagellate cyst assemblages to which the data presented herein can be compared. This circum-Antarctic domain is characterized by the high proportions of the cyst species Selenopemphix antarctica, Impagidinium pallidum , and round, brown Protoperidinium cysts referred to as Brigantedinium spp. It is immediately apparent from our data that our assemblages are also characterized by the same cyst taxa. Within the circum-Antarctic domain Marret and de Vernal (1997) defined three separate assemblages: a proximal Antarctic assemblage (Assemblage I), an Antarctic assemblage (Assemblage II), and a Subantarctic assemblage (Assemblage III). Assemblage I contains Selenopemphix antarctica (> 70 per cent.), with Impagidinium pallidum (1-35 per cent.) and Brigantedinium spp. (2-15 per cent.); Assemblage II consists of S'. antarctica (19-74 per cent.), with Brigantedinium spp. (8-56 per cent.) and /. pallidum (2-35 per cent.); Assemblage III is dominated by Brigantedinium spp. (16-72 per cent.), with Nematosphaeropsis labyrinthus (2 48 per cent.) and 5. antarctica (0-6-20 per cent.). We believe that our data are comparable to those of Marret and de Vernal (1997) and that our assemblages south of 60° S are similar to their Assemblage I, with high percentages of S. antarctica (0-68-1 per cent.), /. pallidum (0-66-7 per cent.) and round, brown Protoperidinium cysts (0-47-1 per cent.) and with significant proportions of Algidasphaeridiuml minutum (0-78-5 per cent.). To the north of 60° S we believe that our assemblages are similar to Assemblage II of Marret and de Vernal (1997), with . S’, antarctica (15-1-92-6 per cent.), round, brown cysts (3-1-51-3 percent.) and I. pallidum (0-4 0 per cent.). The possible subdivision of our data to the north of 60° S as intimated on the cluster analysis may be associated with Assemblage III of Marret and de Vernal (1997) especially on the increased presence of Nematosphaeropsis labrinthus , an important element of their HARLAND ET AL.: RECENT DINOFLAGELLATE CYSTS 1123 Assemblage III. However, we feel that there is insufficient evidence available at the moment to identify, with confidence, a subdivision of the dinoflagellate cyst assemblages north of 60° S. We are confident in identifying the two assemblages divided by the 60° S line of latitude and the clear similarity with Assemblages I and II of Marret and de Vernal (1997) which occur in the south Indian Ocean to the north and south of 65° S. In addition to describing various dinoflagellate cyst assemblages from the area as a result of the Principal Component Analysis, Marret and de Vernal (1997) developed a transfer function relating the assemblages to SST and SSS. The best analogue methodology was used and all the procedures (described in de Vernal et al. 1994) yielded comparable reconstructions. Testing against the known modern environmental parameters of temperature and salinity yielded excellent coefficients of correlation. Since that time Drs Marret and de Vernal have included our data and have found it to be fully compatible with theirs and have utilized it in further refinements of their transfer function. This makes us confident that our data are meaningful within the constraints that we identified earlier. DISCUSSION To date, few data are available on Recent dinoflagellate cyst distributions in the southern hemisphere except for around Australia and New Zealand, e.g. Baldwin (1987), Bint (1988), Bolch and Hallaegraff (1990), McMinn (1990, 1991, 1992), McMinn and Sun (1994), Sun and McMinn (1994) and from the southern Indian Ocean (Marret and de Vernal 1997). In most cases the dinoflagellate cysts recovered can be assigned to species already known from the northern hemisphere. There are very few species that can be said to be endemic to the southern hemisphere. This is perhaps not surprising given the planktonic nature of the thecate dinoflagellate life stage, but it may also reflect the lack of a detailed systematic study of the taxonomy of the southern forms. However, two species are sufficiently taxonomically unique to be identified confidently as endemic to the southern hemisphere: Dalella chathamense and Selenopemphix antarctica. The increased interest in the dinoflagellate cyst flora of the southern oceans, as revealed in bottom sediments, may well lead to the description of other endemic species and Impagidinium variaseptum Marret and de Vernal, 1997 may prove to be one of these. However, the presence of the two species mentioned above in any palynological assemblage is, for the moment, a clear indication of deposition in the southern hemisphere and in the case of Selenopemphix antarctica to deposition in a circum- Antarctic domain. In contrast with this rather limited evidence of endemism in dinoflagellate cyst distributions is the reinforcement that the major sea surface factor in controlling dinoflagellate cyst distributions is temperature. This is usually expressed in the latitudinal or climatic distribution trend and the biogeography of dinoflagellate cysts as first recognized by Wall et al. (1977) and later clearly demonstrated by Dale (1983). The results of the present study also show evidence of this latitudinal trend. As Dale (1996) pointed out, in coastal and neritic environments the distribution of dinoflagellate cysts follows a standard biogeographical zonation and appears to be bipolar on the global scale. The recognition of any of these biogeographical boundaries within Recent cyst distribution patterns provides a signal which can then be utilized in the sediment record to chart oceanographic and climatic changes. The latitudinal distribution of the core-top dinoflagellate assemblages and the clear differentiation of dinoflagellate cyst assemblages at 60° S appears to provide such a biogeographical boundary. It remains to be proven as to whether this boundary can be utilized in the Quaternary fossil record. This 60° S boundary appears to coincide approximately with the limit of maximum sea-ice, and is the approximate position of the 1 °C summer SST isotherm (Text-fig. 2). To the north of 60° S there are about eight months of open water from mid October to mid June (Sea Ice Climatic Atlas 1985). The limit of sea-ice is important in its role in the suppression of light energy and therefore as a limiting factor for photosynthesis and primary production from phytoplankton including the autotrophic dinoflagellates. Indeed, it is interesting to note that Hasle (1969) observed a major decline in dinoflagellate numbers south of 60° S along transects at 90° W and 150° W. However, to 1124 PALAEONTOLOGY, VOLUME 41 table 6. Log ratio of heterotrophic dinoflagellates to autotrophic dinoflagellates. Total cyst numbers are given as cysts per gramme of sediment. A minimum value of 1 (shown in italics) was assigned to those assemblages without autotrophic forms to facilitate the calculation. Core Log H: A Ratio Total Heterotroph Cysts Total Autotroph Cysts TC 004 0-4771 3 1 TC 006 0-301 2 1 TC 010 0 2 2 TC 044 0 1 1 PC 038 0-69897 5 1 TC 041 0-5441 7 2 GC 027 0-39794 5 2 TC 032 0-8239 20 3 KC 083 1-04663 334 30 KC 081 0-87091 208 28 PC 078 0-45426 74 26 KC 064 0-01773 25 24 KC 075 0-55091 192 54 KC 073 1-66276 230 5 KC 095 1-12571 187 14 the north of this line of latitude and beyond the winter sea-ice limit, numbers of dinoflagellates in the surface waters are controlled largely by nutrient availability. The dinoflagellate cyst floras that have been recovered in the present study reveal that north of 60° S the diversity and recovery of dinoflagellate cysts increases markedly. Most of this increase derives from congruentidiacean cysts belonging, by inference or from incubation experiments, to the modern dinoflagellate genus Protoperidinium. Indeed Holm-Hansen et al. (1977) noted that this genus was the most important dinoflagellate genus to be found south of the Antarctic Convergence or Polar Front. Protoperidinium is heterotrophic in its nutritional strategy, often feeding upon diatoms which are the most predominant constituent of the phytoplankton in Antarctic waters (Jacques et al. 1979). These heterotrophic dinoflagellates are r-strategists and take advantage of the high nutrient content which encourages the increased diatom populations upon which the heterotrophic dinoflagellates feed. The diatoms, whilst autotrophic, are also r-strategists. The area north of 60° S is a well-known area of upwelling from the effect of Ekman transport and katabatic winds off the Antarctic continent and it is to be expected that the effect of upwelling will be seen in the phytoplankton populations. We have already alluded to this in the increased numbers of Protoperidinium dinoflagellates and their cysts as seen in the core-top samples. An area of upwelling is also unstable and unpredictable and favours r-strategists. It is important, therefore, in this context to examine the numbers of autotrophic dinoflagellates and their cyst record in comparison with the heterotrophic dinoflagellates and their cysts. One of the ways of making this comparison is to look at the ratios between the two groups of dinoflagellate cysts. A simple intuitive approach was first described by Harland (1973) as the gonyaulacacean ratio although at the time it was not associated with nutritional strategies. Later it became more obvious that it was a measure of the numbers of peridiniacean cysts and hence of the importance of nutrient input into an environment encouraging heterotrophic nutrition. Recently, Powell et al. (1990) used the ratio of P-cysts to G-cysts as an indicator of nutrient enhancement and a guide to the history of coastal upwelling off the coast of Peru. They calculated a logarithmic ratio between peridiniacean and chorate cysts as a substitute for the peridiniacean/gonyaulacacean ratio. Dale (1996) argued the case to identify clearly the trophic categories, avoiding confusion with taxonomy or morphology. In this study we have identified autotrophs and heterotrophs within our assemblages and have indicated the same on Table 3. Table 6 shows the numbers of cysts per gramme of sediment HARLAND ET AL.: RECENT DINOFL AGELLATE CYSTS 1125 assignable to each category and the calculated log ratios of heterotrophs to autotrophs (the ‘ Id- cysts ’ and ‘A-cysts’ of Dale 1996). The results of plotting this ratio against the core-top samples along the transect (Text-fig. 6) reveals a general rising value from south to north with one of the major steps in the ratio occurring at about 60° S. This reinforces the other results presented in this paper and underscores the importance of the upwelling phenomenon in enhancing the numbers of the congruentidiacean dinoflagellate cysts and the position of the maximum limit of sea-ice. However, there is sufficient variation within the data for it to be treated with some caution. There is some potential, however, in using this methodology as a useful tool within the temporal record. Relationships to productivity are less easy to disentangle, not least because of the possible allochthonous nature of the record and the difficulties in identifying modern sedimentation, as the congruentidiacean dinoflagellate cysts are representatives of the second tier within the trophic web feeding upon the dominant primary producers, the diatoms. A general assumption is that increased numbers of diatoms will inevitably lead to increased numbers of congruentidiacean dinoflagellate cysts; this also assumes that there is a simple relationship between the numbers of thecate forms, feeding upon diatoms, and the number of cysts produced as a result of sexual reproduction. This is perhaps rather too many assumptions to make at the moment and certainly none of these relationships is as yet proven nor has any quantitative model been established. Indeed, we have little information, for instance, on the competition faced by heterotrophic dinoflagellates from foraminifera and other planktonic organisms for food. We believe this will prove to be an extremely fruitful and exciting area for dinoflagellate research in the future. The importance of these relationships is becoming more and more apparent such that the link between dinoflagellate productivity, cyst production and positions in the first or second tier of the trophic web becomes paramount if ever the dinoflagellate cyst record is to be used as a proxy for productivity. It is to be welcomed that new data are becoming available on the numbers of cysts falling through the water column and being sampled in sediment trap arrays (Dale and Dale 1992). These data will assist in our understanding of cyst production and their final incorporation as a thanatocoenosis in bottom sediments. In contrast to the discussion above on the role of the heterotrophic dinoflagellates, there is an increase in the absolute numbers of cysts derived from autotrophic dinoflagellates to the north along the studied transect (Text-fig. 6a). This trend increases toward the edge of the ACC and across the Antarctic Convergence as these .^-strategists gain prominence in the more stable, predictable and nutrient-poor environments of the South Atlantic. Nonetheless, these autotrophic dinoflagellate cysts remain a minor component of the assemblages dominated by the congruentidiacean heterotrophs described herein. Further confirmation of this trend across the Antarctic Convergence must await further study on a more extensive dataset. Finally, the relationship between the contained dinoflagellate cysts and the core-top sample lithologies also requires some comment. The poor recovery of dinoflagellate cysts south of 60° S also corresponds with an increase in the percentage of terrigenous sediment, low percentages of biogenic carbonate and silica and the presence of highly corrosive and oxygenated WSBW (Pudsey and Howe 1998). North of 60° S the increased numbers of dinoflagellate cysts is accompanied by increased percentages of biogenic material and a lower percentage of terrigenous sediment. This is further evidence to substantiate the effect of increased nutrient supply within the system leading to increased productivity and the importance of the sea-ice limit together with the nature of the bottom water masses. The relationships of biogenic material within core-top material is, however, also impossible to interpret without a clear understanding of the detailed sedimentation across the benthic boundary layer and the nature of the sedimentary record held in the uppermost millimetres of any sediment core and included in our core-top samples. Indeed, in Core 32 the Holocene in its entirety may be only 0T m thick. The different methods of core recovery may also be an important factor. However, the overall scale of this study and the comparisons with the work of Marret and de Vernal (1997) lead us to be confident in our identification of a biogeographical boundary for use within the palaeoceanographic history of the area. The biogeographical boundary identified herein is coincident with the maximum limit of sea-ice and, as a consequence, the length of time open sea 1126 PALAEONTOLOGY, VOLUME 41 conditions prevail, as well as the availability of nutrients from upwelling. The detailed ecology of dinoflagellates within this particular environment is complex and will need further research before an adequate paradigm becomes available. CONCLUSIONS Our study is the first to describe in detail an indigenous dinoflagellate cyst thanatocoenosis from Recent bottom sediments in the Antarctic region. It provides distribution data and taxonomy of 17 taxa along a transect from the Falkland Trough to the Weddell Sea and it demonstrates a latitudinal trend in the cyst distributions. In comparison with the regional oceanography of the area the maximum limit of sea-ice appears to be the defining parameter in the cyst distributions. Hence the analysis of temporal (downcore) dinoflagellate cyst data should provide a proxy to recognize fluctuations in the limit of winter sea-ice through the more recent geological record. The recognition of this biogeographical and oceanographic boundary has implications for the extent and volume of sea-ice around the continent of Antarctica, the production of cold, dense bottom water and the dynamics of the thermohaline circulation system. The recognition of this and other biogeographical boundaries in the geological record also allows the detailing of climatic change in the region. Finally, the potential of using the dinoflagellate cyst record of autotrophic and heterotrophic dinoflagellates for investigating the trophic web, productivity, nutrient availability and upwelling histories has been discussed and identified as an area of future fundamental research. Acknowledgements. The Centre for Palynology at the University of Sheffield and Mr S. Ellin and Mr B. Pigott, in particular, are thanked for their careful supervision of the qualitative palynological preparations of the core- top samples and for providing photographic facilities respectively. Particular thanks are due to Drs Anne de Vernal and Fabienne Marret who have been most generous with their help, advice and discussion as have the attendees of a recent workshop on ‘Late Quaternary to Recent dinoflagellate cysts and their (palaeo-) ecological significance' held in Bremen, Germany, hosted by Professor Helmut Willems and Dr Karin A. F. Zonneveld. Finally Dr J. B. Riding and an anonymous referee are thanked for their constructive criticisms of the manuscript. REFERENCES artzner, D. G. and dorhofer, G. 1978. Taxonomic note: Lejeunecysta nom. nov. pro. Lejeunia Gerlach 1961 emend. Lentin and Williams 1976 - dinoflagellate cyst genus. Canadian Journal of Botany , 56, 1381-1382. Baldwin, R. p. 1987. Dinoflagellate resting cysts isolated from sediments in Marlborough Sounds, New Zealand. New Zealand Journal of Marine and Freshwater Research, 21, 543-553. balech, E. 1973. Segunda contribucion al conocimiento del microplancton del mar de Bellingshausen. Contribucion del Instituto Antarctica Argentino, 107, 1-63. — 1974. El genero “ Protoperidinium ” Bergh, 1881 (“ Peridinium ” Ehrenberg, 1831, partim). Revista del Museo Argentino de Ciencias Naturales “ Bernardino Rivadavia ”, Hidrobiologia, 4(1), 1-79. 1976. Clave ilustrada de dinoflagelados antarticos. Publicacion Instituto Antarctica Argentino, 11, 1-99. — and el-sayed, s. z. 1965. Microplankton of the Weddell Sea. American Geophysical Union, Antarctic Research Series, 5, 107-124. barber, m. and crane, d. 1995. Current flow in the north-west Weddell Sea. Antarctic Science, 7, 39-50. baumann, k.-h. and matthiessen, J. 1992. Variations in surface water mass conditions in the Norwegian Sea: Evidence from Holocene coccolith and dinoflagellate cyst assemblages. Marine Micropaleontology, 20, 129-146. benedek, p. n. 1972. Phytoplanktonten aus dem Mittel- und Oberoligozan von Tonisberg (Niederrheingebeit). Palaeontographica, Abteilung B, 137, 1—71. bergh, r. s. 1881. Bidrag til Cilioflagellaternes Naturhistorie. Forelobige meddelelser. Dansk Natur- historiskforening i Kjobenhavn, Videnskabelige Meddelelser, Series 4, 3, 60-76. — 1882. Der Organismus der Cilioflagellaten. Eine phylogenetische Studie. Morphologisches Jahrbuch, 1, 177-288. bint, a. n. 1988. Recent dinoflagellate cysts from Mermaid Sound, northwestern Australia. Memoir of the Association of Australian Palaeontologists, 5, 329-341. HARLAND ET AL.\ RECENT DINOFLAGELLATE CYSTS 1127 bolch, c. J. and hallegraeff, g. m. 1990. Dinoflagellate cysts in Recent marine sediments from Tasmania, Australia. Botanica Marina, 33, 173-192. Bradford, M. r. 1975. New dinoflagellate cyst genera from the recent sediments of the Persian Gulf. Canadian Journal of Botany , 53, 3064—3074. broecker, w. s., bond, G., klas, M., bonani, G. and wolfli, w. 1990. A salt oscillator in the glacial Atlantic? 1. The concept. Paleoceanography, 5, 469^477. peteet, D. M. and rind, D. 1985. Does the ocean-atmosphere system have more than one stable mode of operation? Nature , 315, 21-26. bryden, h. l. and pillsbury, r. d. 1977. Variability of deep flow in the Drake Passage from year-long current measurements. Journal of Physical Oceanography , 7, 803-810. buck, K. R., bolt, p. A., bentham, w. N. and garrison, d. l. 1992. A dinoflagellate cyst from Antarctic sea ice. Journal of Phycology , 28, 15-18. bujak, J. p. 1980. Dinoflagellate cysts and acritarchs from the Eocene Barton Beds of southern England. 36-91 . In bujak, j. p., downie, c., Eaton, G. L. and williams, G. l. Dinoflagellate cysts and acritarchs from the Eocene of southern England. Special Papers in Palaeontology , 24, 1-100. 1984. Cenozoic dinoflagellate cysts and acritarchs from the Bering Sea and northern North Pacific, DSDP Leg 19. Micropaleontology , 30, 180-212. butschli, o. 1885. Erster Band. Protozoa. 865-1088. In Dr. H. G. Bronns Klassen und Ordnungen des Thier- Reichs, wissenschaftlich dargestellt in Wort und Bild. C. F. Winter’sche Verlagshandlung, Leipzig and Heidelberg. Carmack, e. c. and foster, t. d 1975. On the flow of water out of the Weddell Sea. Deep-Sea Research , 22, 711-724. chatton, e. 1933. Pheopolykrikos beauchampi nov. gen., nov. sp., dinoflagelle polydtnide autotrophe, dans l’Etang de Thau. Bulletin de la Societe zoologique de France, 58, 251-254. claparede, E. and lachmann, J. 1859. Etude sur les infusoires et les rhizopods. Memoires de Ilnstitut National Genevois, 6, 261-482. cookson, i. c. and eisenack, a. 1965. Microplankton from the Browns Creek Clays, SW. Victoria. Proceedings of the Royal Society of Victoria , 79, 1 19-131. dale, b. 1976. Cyst formation, sedimentation, and preservation: factors affecting dinoflagellate assemblages in Recent sediments from Trondheimsfjord, Norway. Review of Palaeobotany and Palynology, 22, 39-60. 1977. New observations on Peridinium faeroense Paulsen (1905), and classification of small orthoperidinioid dinoflagellates. British Phycology Journal, 12, 241-253. 1983. Dinoflagellate resting cysts: “benthic plankton”. 69-136. In fryxell, g. a. (ed.). Survival strategies of the algae. Cambridge University Press, Cambridge, 144 pp. 1996. Dinoflagellate cyst ecology: modeling and geological applications. 1249-1275. In jansonius, j. and McGregor, d. c. (eds). Palynology : principles and applications. 3. American Association of Stratigraphic Palynologists Foundation, Publishers Press, Salt Lake City, 911-1330. and dale, a. l. 1992. Dinoflagellate contributions to the deep sea. Woods Hole Oceanographic Institution, Ocean Biocenosis Series, 5, 1-76. davey, R. j. and williams, G. l. 1966. The genera Hystrichosphaera and Achomosphaera. 28-52. In davey, r. j., downie, c., sarjeant, w. a. s. and williams, g. l. Studies on Mesozoic and Cainozoic dinoflagellate cysts. Bulletin of the British Museum (Natural History), Geology, Supplement, 3, 1-248. deflandre, G. and cookson, i. c. 1955. Fossil microplankton from Australian Late Mesozoic and Tertiary sediments. Australian Journal of Marine and Freshwater Research, 6, 242-313. diesing, c. m. 1866. Revision der Prothelminthen, Abtheilung: Mastigophoren. Sitzungsberichte der Akademie der Wissenschaften zu Wien, Mathematisch-naturwissenschaftliche Klasse , 52, 287-401. dodge, j. d. 1989. Some revisions of the Family Gonyaulacaceae (Dinophyceae) based on a scanning electron microscope study. Botanica Marina, 32, 275-298. Edwards, L. E. and and RLE, v. a. s. 1992. Distribution of selected dinoflagellate cysts in modern marine sediments. 259-288. In head, m. j. and wrenn, j. h. (eds). Neogene and Quaternary dinoflagellate cysts and acritarchs. American Association of Stratigraphic Palynologists Foundation, Publishers Press, Salt Lake City, 438 pp. ehrenberg, c. g. 1830. Beitrage zur Kenntnis der Organisation der Infusorien und ihrer geographischen Verbreitung, besonders in Sibirien. Koniglich Akademie der Wissenschaften zu Berlin, Abhandlungen, Physikalische-Mathematische Klasse, 1-88. — 1831. Animalia evertebrata. Unpaginated. In hemprich, p. c. and ehrenberg, c. g. Symbolae physicae ... Pars zoologica. Akadamie der Wissenschaften zu Berlin, Abhandlungen, 10 pis. 1128 PALAEONTOLOGY, VOLUME 41 — 1838. Uber das Massenverhaltniss der jetz lebenden Kiesel-Infusorien und iiber ein neues Infusorien- Conglomerat als Polierschiefer von Jastraba in Ungarn. Koniglich Akademie der Wissenschaften zu Berlin, Abhandlungen , 1836, 1, 109-135. FENSOME, R. A., TAYLOR, F. J. R., NORRIS, G., SARJEANT, W. A. S., WHARTON, D. I. and WILLIAMS, G. L. 1993. A classification of living and fossil dinoflagellates. Micropaleontology, Special Publication, 7, 1-351. foldvik, a. and gammelsrod, t. 1988. Notes on Southern Ocean hydrography, sea-ice and bottom water formation. Palaeogeography, Palaeoclimatology, Palaeoecology, 67, 3-17. foster, t. d. and middleton, J. h. 1979. Variability in the bottom water of the Weddell Sea. Deep-Sea Research, 26, 743-762. grimm, E. c. 1987. CONISS: a FORTRAN 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares. Computers and Geosciences, 13, 13-35. grose, T. J., Johnson, J. a. and bigg, G. R. 1995. A comparison between the FRAM (Fine Resolution Antarctic Model) results and observations in the Drake Passage. Deep-Sea Research, 42, 365-388. GRUNIG, s. 1991. Quaternary sedimentation processes on the continental margin of the South Orkney Plateau, NW Weddell Sea (Antarctica). Berichte zur Polarforschung, 75, 1-196. Haeckel, E. 1894. Systematische Phylogenie. Entwurf eines natiirlichen Systems der Organismen auf Grutid ihrer Stammegeschichte, I. Systematische Phylogenie der Protisten und Pflanzen. Reimer, Berlin, 400 pp. harland, r. 1973. Dinoflagellate cysts and acritarchs from the Bearpaw Formation (Upper Campanian) of southern Alberta, Canada. Palaeontology, 16, 665-706. — 1977. Recent and Late Quaternary (Flandrian and Devensian) dinoflagellate cysts from marine continental shelf sediments around the British Isles. Palaeontographica, Abteilung B, 164, 87-126. — 1982. A review of Recent and Quaternary organic-walled dinoflagellate cysts of the genus Protoperidinium. Palaeontology, 25, 369-397. — 1989. A dinoflagellate cyst record for the last 0-7 Ma from the Rockall Plateau, northeast Atlantic Ocean. Journal of the Geological Society, London, 146, 949-951. — and howe, J. A. 1995. Dinoflagellate cysts and Holocene oceanography of the northeastern Atlantic Ocean. The Holocene, 5, 220-228. and long, D. 1996. A Holocene dinoflagellate cyst record from offshore north-east England. Proceedings of the Yorkshire Geological Society, 51, 65-74. — and reid, p. c. 1980. Systematics Part I. In harland, r., reid, p. c., dobell, p. and norris, g. Recent and sub-Recent dinoflagellate cysts from the Beaufort Sea, Canadian Arctic. Grana, 19, 211-225. hasle, G. R. 1969. Scientific results of marine biological research. Hvalradets Skriffter, 52, 1-168. head, M. j. 1993. Dinoflagellates, sporomorphs, and other palynomorphs from the Upper Pliocene St. Erth Beds of Cornwall, southwestern England. Memoir of the Paleontological Society , 31, 1-62. — 1994. Morphology and paleoenvironmental significance of the Cenozoic dinoflagellate genera Tectatodinium and Habibacysta. Micropaleontology, 40, 289-321. — 1996. Modern dinoflagellate cysts and their biological affinities. 1197-1248. In jansonius, j. and McGregor, D. c. (eds). Palynology : principles and applications. 3. American Association of Stratigraphic Palynologists Foundation, Publishers Press, Salt Lake City, 911-1330. — and wrenn, J. H. (eds). 1992. A forum on Neogene and Quaternary dinoflagellate cysts. 1-31. In head, m. J. and wrenn, J. h. (eds). Neogene and Quaternary dinoflagellate cysts and acritarchs. American Association of Stratigraphic Palynologists Foundation, Dallas, 438 pp. Hofmann, E. E. 1985. The large-scale horizontal structure of the Antarctic Circumpolar Current from FGGE drifters. Journal of Geophysical Research, 90, 7087-7097. holm-hansen, o., el-sayed, s. z., franceschini, G. a. and cuhel, R. L. 1977. Primary production and factors controlling phytoplankton growth in the Southern Ocean. Proceedings of the Third SCAR Symposium on Antarctic Biology, 1 1-49. howe, J. A., pudsey, c. J. and Cunningham, a. p. 1997. Pliocene-Holocene contourite deposition under the Antarctic Circumpolar Current, Western Falkland Trough, South Atlantic Ocean. Marine Geology, 138, 27-50. indelicato, s. r. and loeblich, a. r. in 1986. A revision of the marine peridinioid genera (Pyrrhophyta) utilizing hypothecal-cingular plate relationships as a taxonomic guideline. Japanese Journal of Phycology (Sorui), 34, 153-162. jacques, G., descolas-gros, c., grall, j.-R. and sournia, a. 1979. Distribution du phytoplankton dans la partie Antarctique de l’Ocean Indien en fin d’ete. Internationale Revue der gesamten Hydrobiologie, 64, 609-628. HARLAND ET AL.: RECENT DINOFLAGELLATE CYSTS 1129 Jordan, R. w. and pudsey, c. j. 1992. High-resolution diatom stratigraphy of Quaternary sediments from the Scotia Sea. Marine Micropaleontology , 19, 201-237. kofoid, c. A. 1911. Dinoflagellata of the San Diego region. IY. The genus Gonyaulax, with notes on its skeletal morphology and a discussion of its generic and specific characters. University of California Publications in Zoology , 8, 187-286. kroopnick, p. M. 1985. The distribution of 13C of SCO, in the world oceans. Deep-Sea Research , 32, 57-84. lemmermann, e. 1910. III. Klasse Peridiniales. In Kryptogamenflora der Mark Brandenburg und angrenzender Gebiete, III Algen I (Schizophyceen, Flagellaten, Peridineen ). Gebriider Borntraeger, Leipzig, 563-712, 869. lentin, J. K. and williams, G. l. 1981. Fossil dinoflagellates : index to genera and species, 1981 edition. Bedford Institute of Oceanography, Report Series, BI-R-81-12, 1-345. 1993. Fossil dinoflagellates: index to genera and species 1993 edition. American Association of Stratigraphic Palynologists Foundation, Contribution Series, 28, 1-856. lindemann, e. 1928. Abteilung Peridineae (Dinoflagellatae). 3-104. In engler, a. and prantl, k. (eds). Die Natiirlichen Pflanzenfamilien nebst ihren Gattungen und wichtigeren Arten insbesonderen den Nutzpflanzen. Zweite stark vermehrte und verbesserte Auflage herausgegeben von A. Engler. 2. Wilhelm Engelmann, Leipzig. locarnini, R. A., Whitworth, T., in and nowlin, w. d. Jr 1993. The importance of the Scotia Sea on the outflow of Weddell Sea Deep Water. Journal of Marine Research , 51, 135-153. loeblich, a. R. Jr and loeblich, a. r. in 1996. Index to the genera, subgenera, and sections of the Pyrrhophyta. Studies in Tropical Oceanography, University of Miami, 3, 1-94. mantell, G. A. 1850. A pictorial atlas of fossil remains consisting of coloured illustrations selected from Parkinson s “ Organic Remains of a former World", and Artis's “ Antediluvian phytology" . Henry G. Bohn, London, 207 pp. 1854. The medals of creation; or first lessons in geology and the study of organic remains. 2nd edition. 1. Henry G. Bohn, London, 91, 239-242. marret, f. and vernal, A. de 1997. Dinoflagellate cyst distribution in surface sediments of the southern Indian Ocean. Marine Micropaleontology, 29, 367-392. matsuoka, k. and bujak, J. p. 1988. Cenozoic dinoflagellate cysts from the Navarin Basin, Norton Sound and St. George Basin, Bering Sea. Bulletin of the Faculty of Liberal Arts, Nagasaki University, Natural Science, 29(1), 1-147. matthiessen , J. 1995. Distribution patterns of dinoflagellate cysts and other organic-walled microfossils in recent Norwegian-Greenland Sea sediments. Marine Micropaleontology, 24, 307-334. Mcminn, A. 1990. Recent dinoflagellate cyst distribution in eastern Australia. Review of Palaeobotany and Palynology, 65, 305-310. 1991. Recent dinoflagellate cysts from estuaries on the central coast of New South Wales, Australia. Micropaleontology, 37, 269-287. 1992. Recent and late Quaternary dinoflagellate cyst distribution on the continental shelf and slope of southeastern Australia. Palynology, 16, 13-24. — 1995. Why are there no post-Palaeogene dinoflagellate cysts in the Southern Ocean? Micropaleontology, 41, 383-386. and sun xuekun 1994. Recent dinoflagellate cysts from the Chatham Rise, Southern Ocean, east of New Zealand. Palynology , 18, 41-53. mosby, H. H. 1934. The waters of the Atlantic Antarctic Ocean. Scientific Results of the Norwegian Antarctic Expeditions 1927-1928, 11, 1-131. mud ie, p. j. 1986. Palynology and dinoflagellate biostratigraphy of Deep Sea Drilling Project Leg 94, Sites 607 and 611, North Atlantic Ocean. Initial Reports of the Deep Sea Drilling Project, 94, 785-812. nowlin, w. D. Jr and Clifford, M. 1982. The kinematic and thermohaline zonation of the Antarctic Circumpolar Current at Drake Passage. Journal of Marine Research, 40, 481-507. and klinck, J. M. 1986. Physics of the Antarctic Circumpolar Current. Reviews of Geophysics, 24, 469-491. o’brien, p. 1989. The magnetostratigraphy of marine sediments from Jane Basin, southeast of the South Orkney Microcontinent, Antarctica. Unpublished Ph.D. thesis, University of Southampton, England. OLBERS, D., GOURETSKI, V., SEISS, G. and schroter, J. 1992. Hydrographic atlas of the Southern Ocean. Alfred Wegener Institute, Bremerhaven, xvii + 82 pis. orsi, a. H., nowlin, w. d. Jr and whitworth, T. in 1993. On the circulation and stratification of the Weddell Gyre. Deep-Sea Research, 40, 169-203. 1130 PALAEONTOLOGY, VOLUME 41 — whitworth, t. hi and nowlin, w. D. Jr 1995. On the meridional extent and fronts of the Antarctic Circumpolar Current. Deep-Sea Research, 42, 641-673. ostenfeld, c. h. 1903. Phytoplankton from the sea around the Faeroes. Botany of the Faeroes based upon Danish investigations, 2. Det Nordiske Forlag, Copenhagen, 558-612. pascher, a. 1914. Uber Flagellaten und Algen. Berichte der Deutsche Botanische Gesellschaft, 32, 136-160. paulsen, o. 1905. On some Peridiniae and plankton diatoms. Meddelelser fra Kommissionen for Havundsogelser, Serie Plankton , 1(3), 1-21. powell, a. j., dodge, j. d. and lewis, j. 1990. Late Neogene to Pleistocene palynological facies of the Peruvian continental margin upwelling, Leg 112. Proceedings of the Ocean Drilling Program, Scientific Results, 112, 297-321. pudsey, c. J. 1992. Late Quaternary changes in Antarctic Bottom Water velocity inferred from sediment grain size in the northern Weddell Sea. Marine Geology, 107, 9-33. — barker, p. f. and Hamilton, n. 1988. Weddell Sea abyssal sediments : a record of Antarctic Bottom Water flow. Marine Geology, 81, 289-314. — and howe, j. a. 1998. Quaternary history of the Antarctic Circumpolar Current: evidence from the Scotia Sea. Marine Geology , 44, 1841-1876. — and king, p. in press. Particle fluxes, benthic processes and the palaeoenvironmental record in the northern Weddell Sea. Deep-Sea Research. reid, p. c. 1974. Gonyaulacacean dinoflagellate cysts from the British Isles. Nova Hedwigia, 25, 519-621. reinecke, p. 1967. Gonyaulax grindleyi sp. nov.: a dinoflagellate causing a red tide at Elands Bay, Cape Province, in December 1996. Journal of South African Botany, 33, 157-160. sarjeant, w. a. s. 1970. The genus Spiniferites Mantell, 1850 (Dinophyceae). Grana, 10, 74-78. Schiller, j. 1935. Dinoflagellatae (Peridineae) in monographischer Behandlung. 2. Teil, Leiferung 2. 161-320. In rabenhorst, L. Kryptogamen-Flora von Deutschland, Osterreich und der Schweiz. Akademische Verlagsgesellschaft, Leipzig. schutt, F. 1896. Peridiniales (Peridineae, Dinoflagellata, Cilioflagellata, arthrodele Flagellaten). 1-30. In engler, a. and prantl, k. (eds). Die natiirlichen Pflanzenfamilien nebst ihren Gattungen und wichtigeren Arten insbesondere den Nutzpflanzen unter Mitwirkung zahlreicher hervorragender Fachgelehrten, Teil. 1, Abteilung 1 b. Wilhelm Engelmann, Leipzig. sea ice climatic atlas. 1985. Volume 1, Antarctica. Naval Oceanography Command, Asheville, North Carolina. shimmield, G. b., derrick, s., mackensen, a., grobe, H. and pudsey, c. J. 1994. The history of biogenic silica, organic carbon and barium accumulation in the Weddell Sea and Antarctic Ocean over the last 150,000 years. 555-574. In zahn, r. et al. (eds). Carbon cycling in the Glacial Ocean: constraints on the ocean’s role in global change. NATO ASI Series , 117, 580 pp. stover, L. e. and evitt, w. R. 1978. Analyses of pre-Pleistocene organic-walled dinoflagellates. Stanford University Publications, Geological Sciences, 15, 1-298. sun xuekun and Mcminn, a. 1994. Recent dinoflagellate cyst distribution associated with the Subtropical Convergence on the Chatham Rise, east of New Zealand. Marine Micropaleontology, 23, 345-356. taylor, F. J. r. 1980. On dinoflagellate evolution. BioSystems, 13, 65-108. tectonic map of the scotia arc. 1985. 1:3000000, Sheet BAS ( Miscellaneous ), 3. British Antarctic Survey, Cambridge. vernal, A. de goyette, c. and rodrigues, c. G. 1989. Contribution palynostratigraphique (dinokystes, pollen et spores) a la connaissance de la mer de Champlain: coupe de Saint-Cesaire, Quebec. Canadian Journal of Earth Sciences, 26, 2450-2464. — turon, j.-l. and guiot, j. 1994. Dinoflagellate cyst distribution in high-latitude marine environments and quantitative reconstruction of sea-surface salinity, temperature, and seasonality. Canadian Journal of Earth Sciences, 31, 48-62. wall, d. 1965. Modern hystrichospheres and dinoflagellate cysts from the Woods Hole region. Grana Palynologica, 6, 297-314. 1967. Fossil microplankton in deep-sea cores from the Caribbean Sea. Palaeontology, 10, 95-123. — and dale, b. 1968. Modern dinoflagellate cysts and evolution of the Peridiniales. Micropaleontology, 14, 265-304. lohmann, G. P. and smith, w. k. 1977. The environmental and climatic distribution of dinoflagellate cysts in modern marine sediments from regions in the North and South Atlantic Oceans and adjacent seas. Marine Micropaleontology, 2, 121-200. HARLAND ET AL.\ RECENT DINOFLAGELLATE CYSTS 1131 whitworth, t. in, nowlin, w. d. Jr and worley, s. J. 1982. The net transport of the Antarctic Circumpolar Current through Drake Passage. Journal of Physical Oceanography , 12, 960-971. wood, G. d., Gabriel, A. M. and lawson, J. c. 1996. Palynological techniques - processing and microscopy. 29-50. In jansonius, J. and McGregor, d. c. (eds). Palynology : principles and applications. 1. American Association of Stratigraphic Palynologists Foundation, Publishers Press, Salt Lake City, 462 pp. wrenn, j. H. 1988. Differentiating species of the dinoflagellate cyst genus Nematosphaeropsis Deflandre & Cookson 1955. Palynology , 12, 129-150. zenk, w. o. 1981. Detection of overflow events in the Shag Rocks Passage, Scotia Ridge. Science, 213, 1113-1114. REX HARLAND DinoData Services 50 Long Acre, Bingham Nottingham NG13 8AH, UK and Centre for Palynology Department of Earth Sciences University of Sheffield Dainton Building Brookhill Sheffield S3 7HF, UK CAROL A. PUDSEY JOHN A. HOWE British Antarctic Survey High Cross, Madingley Road Cambridge CB3 OET, UK Typescript received 5 August 1997 Revised typescript received 14 January 1998 MERIEL E. J. FITZPATRICK Department of Geological Sciences University of Plymouth Drake Circus Plymouth PL4 8AA, UK CHAROPHYTES FROM THE LOWER CRETACEOUS OF THE IBERIAN RANGES (SPAIN) bv CARLES MARTI N-CLOSAS IlWCARMEN DIEGUEZ Abstract. In the Upper Barremian of the Iberian Ranges (Las Hoyas, Cuenca, Spain) an association of exceptionally well preserved charophyte thalli comprises four new form-species: Palaeonitella vermicularis sp. nov., Charaxis spicatus sp. nov., Clavatoraxis robustus gen. et sp. nov., and Clavatoraxis diaz-romerali sp. nov. This is the youngest fossil record of the genus Palaeonitella. The new form-genus Clavatoraxis is erected to include charophyte vegetative remains bearing spine-cell rosettes, a character attributed to the family Clavatoraceae. This is the first time an assemblage of charophyte vegetative remains has been described and related to assemblages of fructifications. This gives a good correlation at family level between the frequency of taxa found as vegetative remains and calcified fructifications. Two biocoenoses are represented: Clavatoraxis robustus displays adaptations found in extant charophytes living in permanent shallow water lakes whereas Clavatoraxis diaz-romerali was adapted to light-limited, probably deeper, environments. Palaeonitella vermicularis grew twisted round thalli of Clavatoraxis. Early Cretaceous freshwater communities appear to have been dominated by charophytes, and not by aquatic ferns as believed previously. Charophytes include complex green algae, which are considered to be part of the evolutionary lineage leading to vascular plants (Kenrick 1994). The fossil record of these fresh- to brackish-water plants is rich, extends from the upper Silurian to the present, and is composed mainly of calcified fructifications called gyrogonites and utricles. Fossil whole plant remains are extremely scarce. Lower Cretaceous whole plant remains of charophytes consist of silicified specimens from the British Purbeck (Harris 1939), the Morrison Formation of the United States (Peck 1957) and the Barremian of Argentina (Musacchio 1971). Although small fragments of calcified vegetative remains of charophytes are not uncommon in marls prepared for the study of fructifications or in thin sections of lacustrine limestones, this is the first time that a complete association of large vegetative charophyte remains has been found in the post-Palaeozoic fossil record. The study of these fossils is not only necessary to increase knowledge of the morphology of the plants producing the fructifications currently studied, but also enables us description of the structure of an Early Cretaceous freshwater community and underlines the significance of charophytes in subaquatic freshwater environments prior to the radiation of angiosperms. The material studied consists of exceptionally well preserved charophyte thalli from finely laminated lacustrine limestones in the La Huerguina Formation at Las Hoyas, near Cuenca, Spain (Text-fig. 1). The material was obtained in part by bed-by-bed sampling during the annual excavations of the site and also from the collection of Mr Armando Diaz-Romeral, from Cuenca, who discovered the charophyte remains. Charophyte thalli are calcified and preserved in three dimensions within laminites. In most specimens the surface is slightly corroded. This material was studied directly under light microscopy after immersion in an organic solvent. Limited etching with diluted acetic acid was necessary to prepare some specimens. Thin sections were also prepared to observe the internal anatomy of thalli. Since no three-dimensionally preserved fructifications were found attached to the thalli studied they have been named after already known or newly described form-taxa which are reserved for vegetative remains. A correlation between these taxa and the general systematics of fossil charophytes, which is based exclusively on fructifications, is currently only intended above the suprageneric level. For clavatoracean fructifications the systematics followed agrees with the phylogenetic system of Clavatoraceae proposed by Martin-Closas (1996). [Palaeontology, Vol. 41, Part 6, 1998, pp. 1133-1152, 5 pis] © The Palaeontological Association 1134 PALAEONTOLOGY, VOLUME 41 280 - 240- 200- meromictic lake (lithosome 2) meromictic lake (lithosomel) distal alluvial fan and swamp Upper Jurassic text-fig. 1. Geographical location and stratigraphical section of the palaeontological site of Las Hoyas; modified from Fregenal-Martinez and Melendez (1993). The marls and limestones of the La Huerguina Formation, which include the finely laminated limestones of Las Hoyas, were deposited in the south-eastern Iberian Basin, one of the Mesozoic intracontinental basins of the Iberian Plate. These basins contain up to 5000 m of Late Jurassic to Early Cretaceous sediments as a result of significant basement subsidence related to rifting of the Central Atlantic crust (Alvaro et al. 1979; Salas and Casas 1993). The south-eastern Iberian Basin was orientated following the general north-west to south-east trend of the Iberian Ranges. It opened towards the south-east to the Tethys sea, where the deposition of marine shallow water limestones dominated, whereas towards the opposite, north-western end of the basin, the sedimentation occurred mainly in freshwater facies, including the deposition of the La Huerguina Formation (Mas 1981 ; Melendez 1983; Fregenal-Martinez and Melendez 1994). This palustrine and lacustrine unit is up to 280 m thick in the area studied and includes two lithosomes, about 25 m thick, of finely laminated lacustrine limestones separated by palustrine and lacustrine stratified limestones and marls (Fregenal-Martinez and Melendez 1993). The laminated limestones correspond to the distal, anoxic facies of a meromictic lake (Gomez-Fernandez and Melendez 1991), which preserved floral and faunal remains exceptionally well and provided the charophyte specimens studied here. Biostratigraphical analysis of charophyte and ostracod associations (Dieguez et al. 1995 a) indicates that this unit is of Late Barremian age. The fossils found at Las Hoyas include representatives of almost all Lower Cretaceous continental groups. The site has provided significant new taxonomic and phylogenetic data about MARTIN-CLOSAS AND DIEGUEZ: CRETACEOUS CHAROPHYTES 1135 fish (Poyato-Ariza 1995), amphibians (McGowan and Evans 1995), dinosaurs (Perez-Moreno et al. 1994) and birds (Sanz et al. 1996) along with invertebrates, such as insects (Martinez-Delclos et al. 1995) and crustaceans (Rabada 1993). Plant fossils are also very abundant and well preserved and include charophytes, bryophytes, ferns, conifers, cycadales, bennettitales, gnetales and early angiosperms (Sanz et al. 1988; Dieguez et al. 1995 b). The enigmatic plant Montsecchia vidali (Zeiller 1902) Teixeira, 1954 is also present and constitutes one of the most abundant plant remains. Especially significant for this study is the absence of other freshwater macrophytes apart from charophytes. SYSTEMATIC PALAEONTOLOGY Division charophyta Migula, 1897 Class charophyceae Smith, 1938 Order charales Lindley, 1836 Form-genus palaeonitella Pia, 1927 emend. Type species. Palaeonitella cranii (Kidston and Lang 1921) Pia, 1927. Original diagnosis. Vegetative shoot clearly organized in nodes and internodes with whorls of short branches. Reproductive organs so far unknown. Many details of organization are reminiscent of Characeae (Pia 1927, p. 91). (‘Deutlich in Knoten und Internodien gegliederte vegetative Sprosse mit wirtelig gestellten Kurztrieben. Fortpflazungsorgane sind bis jetzt nicht bekannt. Viele Einzelheiten der Organisation erinnern an die Characeen’). Emended diagnosis. Vegetative shoot organized in nodes and internodes with whorls of short, non- branching branchlets. Thallus ecorticate. Reproductive organs so far unknown. Remarks. The ecorticate thallus of Paleonitella is the main character to separate this genus from other post-Palaeozoic form-genera such as Charaxis Harris, 1939. The absence of furcations in branchlets of Palaeonitella distinguishes this genus from the Devonian ecorticate genus Octochara Gess and Hiller, 1995. Thalli of another Devonian genus, Hexachara Gess and Hiller, 1995, are similar in organization to Palaeonitella but the laterals of the former should be termed bract-cells since they always bear large gyrogonites. Palaeonitella vermicularis sp. nov. Plate 1, figures 1^1 Derivation of name. The name refers to the flexible and filamentous aspect of the thalli, which resemble worms (Latin, vermis). Holotype. Specimen LH- 16100 from the collection of Mr Armando Diaz-Romeral, Museo de Cuenca (Cuenca, Spain), deposited in the Unidad de Paleontologia, Universidad Autonoma de Madrid. Paratypes. Specimens LH-1319 and LH-16102 (the latter from the collection of Mr Armando Diaz-Romeral), Museo de Cuenca (Cuenca, Spain), deposited in the Unidad de Paleontologia, Universidad Autonoma de Madrid. Type layer and locality. Second lithosome of finely laminated limestones of the La Huerguina Formation at Las Hoyas (Cuenca, Spain). 1136 PALAEONTOLOGY, VOLUME 41 Material. Specimen LH-16100 is an assemblage of abundant thalli, which are difficult to individualize. Specimens LH-1319, LH-16102, LH-16103 and LH-16104 contain individual portions of thalli. Diagnosis. Extremely fine filamentous and flexible ecorticate thalli with nodes bearing about twelve, 1- 2 mm long, single-celled branchlets and internodes separated at intervals of 1 -5-3-5 mm. Description. Charophyte vegetative remains flexible and filamentous in overall aspect, ecorticate and poorly calcified (PI. 1, figs 1-2). Nodes bear about ten (probably 12) small (1-2 mm long) single-celled branchlets and are separated by ecorticate internodes, which lack a cortex of cells (PI. 1 , figs 3^1). Internodes 1 -8-3-2 mm (mean 2- 3 mm) long and 0-2-0-8 mm wide. No fructifications have been found attached to these thalli. Comparisons. Palaeonitella cranii , from the Devonian Rhynie Chert, Scotland (Kidston and Lang 1921; Pia 1927), differs from P. vermicularis in its smaller size (relative difference 1:10) and possession of noded branchlets (single-celled in P. vermicularis). The other species described within the genus, Palaeonitella tarafiyensis Hill and El Khayal, 1983, from the Upper Permian of Saudi Arabia, is similar in general size and structure of branchlets to our material. However, the internodes of P. tarafiyensis are extremely long (up to 25 mm) and nodes are swollen to about twice the width of internodes, whereas in P. vermicularis internodes are short and nodes have about the same width as internodes. The Permian species appears much more rigid than the Cretaceous species, in spite of the similar preservation as lime-encrusted fossils. Remarks. Palaeonitella vermicularis is relatively common in the samples studied and is associated with Clavatoraxis robustus and Clavatoraxis diaz-romerali. On the basis of the ecortication of Palaeonitella it has traditionally been accepted that this form-genus is reminiscent of extant Characeae Nitelleae. However, extant representatives bear furcated branchlets, unlike Palaeonitella. On the other hand, some Characeae Chareae are also ecorticate, and this may have been the condition of many other taxa which are exclusively fossil, such as the Devonian genera Octochara and Hexachara. Therefore, at present, assignment of Paleonitella to Nitelleae is a hypothesis. Paleonitella vermicularis is the most modern species of the genus in the fossil record since other species were Palaeozoic. However, it is expected that the fossil record of the genus will increase and cover the gaps between fossil and present day charophytes bearing ecorticate thalli which are much like those described here. Form-genus charaxis Harris, 1939 Type species. Charaxis durlstonense Harris, 1939 from the Purbeck of Durlston (England), a lectotype proposed by Horn af Rantzien (1956). Diagnosis. ‘Vegetative charophyte organs agreeing in so far as they are known with Chara. Stem consisting of nodes and internodes, internode composed of a central cell surrounded by a ring of primary cortical cells which grow up and down from the nodes; and may cut off secondary cortical cells at their sides, primary cortical cells giving rise to spine cells. Leaves [branchlets] as in Chara , either corticated in the same way as the stem, or uncorticated’ (Harris 1939, p. 67). EXPLANATION OF PLATE 1 Figs 1-4. Palaeonitella vermicularis sp. nov. ; LH-16100 (holotype). 1, partial view of thallus showing flexibility and variation in size; x 5. 2, detail of thallus with ecorticate internodes; x 6. 3, detail of thallus with branchlets preserved; x 6. 4, detail of node showing about ten branchlet scars; x 30. Fig. 5. Charaxis spicatus sp. nov.; general view of LH-161 10 (holotype); x 4. PLATE 1 MARTIN-CLOSAS and BIEGUEZ, Palaeonitella, Charaxis 1138 PALAEONTOLOGY, VOLUME 41 Charaxis spicatus sp. nov. Plate 1, figure 5; Plate 2, figures 1-9 Derivation of name. Name refers to the spike-like form of the thallus. Holotype. Specimen LH-161 10 from the collection of Mr Armando Diaz-Romeral, Museo de Cuenca (Cuenca, Spain), deposited in the Unidad de Paleontologia, Universidad Autonoma de Madrid. Paratypes. Thin sections LH-16105-LH-16109, housed in the same museum. These samples were taken from a dark grey laminated mudstone. Type horizon and locality. Second lithosome of finely laminated limestones of the La Huerguina Formation at Las Hoyas (Cuenca, Spain). Material. LH-161 10, which is an apical fragment of thallus, is the only three-dimensionally preserved specimen found to date. LH- 16121 is a rock sample containing several horizons rich in charophyte remains which supplied the slices to prepare thin sections LH-16105-LH-16109. Diagnosis. Thallus of Charaxis with nodes formed by six nodal cells bearing up to 18 ecorticate branchlets which are longer than the internodes above, completely covering them. Internodes several millimetres long and c. 1 mm wide formed by an internodal cell coated by first six then 18 cortical cells. Gyrogonites ellipsoidal (400-530 pm x 260-290 pm), showing c. 16—18 circum- volutions and probably apical and basal necks. Description Description of three-dimensionally preserved specimen LH-161 10. Fragment of charophyte thallus bearing six large internodes (2-2— 5-3 mm long and c. 1 mm wide) which are covered by vertical, non-spiralized cortical cells (PI. 1, fig. 5). The preservation of this specimen does not allow a precise count of the number of cortical cells, which is, however, more than ten and less than 20. The five nodes preserved bear about 15 ecorticate, needle- like branchlets which are longer than the internode immediately above, covering it completely when not compressed. From this number of branchlets, which is similar to the most probable number of cortical cells, we deduce that the thallus was haplostichous, which means it had a cortex of primary cells only. Small scars at the base of nodes may correspond to bases of broken branchlets. No fructifications have been found attached to or in close association with this thallus. Description of specimens from thin sections LH-16105-LH- 16109. These specimens are about the half of the size of LH-161 10, but are identical to it in external morphology, number of branchlets and general cortication (PI. 2, fig. 1). Therefore we consider that samples LH- 16105 to LH-161 10 belong to the same species. Intemodes are formed by a large nodal cell (200-330 pm in diameter) coated by small cortical cells (diameter 40-80 ^m). Close to the nodes there are six large cortical cells (PI. 2, fig. 6). At variable but short distances from nodes, these cortical cells trifurcate : two small laterals and a larger central tube arise from each original cell (PI. 2, figs 3-4). This larger central cell becomes smaller towards the distal part of the internode. As a result, EXPLANATION OF PLATE 2 Figs 1-9. Charaxis spicatus sp. nov. 1, thin section LH-161 05; longitudinal section of thallus; x 20. 2, 7-8, thin section LH-16106. 2, longitudinal section through node showing insertion of three branchlets in each nodal cell (arrow); x 20. 7, oblique sections through distal part of internodes; x 30. 8, longitudinal section through fertile node showing three gyrogonites; x20. 3, thin section LH-16107; tangential section of node and subjacent internode showing cortical cells branching downwards (arrow); x 20. 4, thin section LH-16108; longitudinal section through node and internode with secondary cortical cells formed by upwards branching of primary cortical cells (arrow); x 20. 5-6, 9, thin section LH-16109. 5, transverse section through apical internodal cell and branchlets; x 30. 6, transverse section through proximal part of internode showing primary and secondary cortical cells of different size; x 30. 9, longitudinal section of gyrogonite; x 50. PLATE 2 MARTIN-CLOSAS and DIEGUEZ, Charaxis i ; 1140 PALAEONTOLOGY, VOLUME 41 close to the centre of an internode there are about 1 8 cortical cells of equal diameter, three per original cortical cell (PI. 2, fig. 7). In the centre of the internode, cortical cells coming from adjacent nodes do not interdigitate since cortications with 36 cells were not found. Nodes are formed by six large globular cells, from which three ecorticate branchlets emerge (PI. 2, fig. 2). As a result about 18 extremely long branchlets develop and appear to cover completely the next internode and node in an apical direction (PI. 2, fig. 5). Taking into account that the number of branchlets equals the final number of cortical cells per node, the cortex of this species could be termed haplostichous. Gyrogonites have been found attached to certain nodes and are ellipsoidal (400-530 /nn x 260-290 /mi) with an isopolarity index (height x 100/diameter) of 104-108 and about 16-18 circumvolutions (PI. 2, fig. 8). Several gyrogonites found in the same thin sections, which appear to belong to the same species, show apical and basal necks (PI. 2, fig. 9). Comparisons. The new species has approximately as many branchlets and cortical cells as the lectotype of the genus Charaxis durlstonense but differs in the absence of cortication in branchlets. According to Groves (1933), the Tertiary species Charaxis blassiana (Heer 1855) Harris, 1939 and Charaxis gypsorum (Saporta 1862) Harris, 1939 are also haplostichous, like Charaxis spicatus sp. nov. However, they differ from the new species by their reduced number of branchlets. Other Charaxis species considered by Harris (1939) are either different from the point of view of cortex organization or are insufficiently known. Charaxis striatus Peck, 1957 is known only from internodal fragments which appear to be coated by an extremely large number of cortical cells in comparison with the number of nodal cells or branchlets. Using this character it may be distinguished easily from the new species described here. Schudack (1989, 1993) suggested that Charaxis striatus should be included in the form genus Munieria Deecke, 1883, and may be synonymous with Munieria grambasti subsp. sarda Cherchi, Gusic, Schmidt and Schroeder, 1981, which he considers a Clavatoraceae. However, the genus Munieria , as originally defined by Deecke (1883), includes incompletely calcified vegetative remains which may correspond to several of the charophyte form-genera presently known. Remarks. Charaxis spicatus is strongly reminiscent of thalli of extant Chara, which may justify the inclusion of the new species in Characeae. However, the presence of gyrogonites bearing apical necks in the same horizon (but not in anatomical connection with thalli) makes this attribution unsure. In the Lower Cretaceous, gyrogonites with apical necks are typical of the extinct family Clavatoraceae. Form-genus clavatoraxis gen. nov. Derivation of name. From Clavatoraceae, the extinct charophyte family bearing this type of vegetative remains and axis (Latin). Type species. Clavatoraxis robustus sp. nov. Diagnosis. Verticillate thalli organized in nodes with branchlets separated by corticate internodes. Spine-cell rosettes are present at least in some parts of the thalli. Remarks. The presence of spine-cell rosettes was first noticed on thalli of Clavator reidii by Harris (1939). However, Harris found fructifications attached to the silicified vegetative remains that he studied, enabling him to ascribe such remains to a taxon using the fructification-based systematics of fossil charophytes. The new form-genus Clavatoraxis is created for sterile clavatoracean verticillated vegetative remains which cannot be attributed to any species of fructification. EXPLANATION OF PLATE 3 Clavatoraxis robustus gen. et sp. nov.; holotype, LH- 16111; young portion of thallus; x2-5. PLATE 3 MARTIN-CLOSAS and DIEGUEZ, Clavatoraxis 1142 PALAEONTOLOGY, VOLUME 41 Clavatoraxis robustus sp. nov. Plate 3; Plate 4, figures 1^4; Plate 5, figures 1-9 Derivation of name. From its overall robust appearance, making it one of the largest and strongest thalli known from a Recent or fossil charophyte. Holotvpe. Specimen LH-161 1 1 from the collection of Mr Armando Diaz-Romeral, Museo de Cuenca (Cuenca, Spain), deposited in the Unidad de Paleontologia, Universidad Autonoma de Madrid. Para types. Specimens LH-161 12-LH-l 61 14 and thin sections LH-161 16-LH-16 120 housed in the same museum. Type horizon and locality. Second lithosome of finely laminated limestones of the La Huerguina Formation at Las Hoyas (Cuenca, Spain). Material. LH-450 A/B (portion of thallus), LH-823 A/B (portion of thallus), LH-1020 (portion of thallus), LH-1924 (small portion of thallus), LH-1941-LH-1943 (three small portions of branchlets in the same rock sample), LH-7087 (portion of thallus), LH-7361 (apical portion of thallus), LH-8016 (impression of portion of thallus), LH-8017 (portion of branched thallus), LH-8047 (impression of portion of thallus), LH-81 14 (small portion of thallus), LH-8061 (portion of thallus), LH-13152 (portion of thallus with evidences of erosion), LH- 13153 (impression of clavatoracean utricle), LH- 13190 (portion of thallus with clavatoracean utricles scattered around), LH-13368 (portion of branchlet), LH-14114 (portion of thallus), LH-14180 (portion of thallus and impressions of utricles), LH-161 11 (apical part of young thallus, holotype), LH-161 12 (large mature thallus, paratype), LH-161 13 (portion of mature thallus, paratype), LH-161 14 (portion of mature thallus, paratype), LH-161 15 (large sample with abundant fragments of thalli which supplied rock slices for thin sections), five thin sections (paratypes) LH-161 16-LH-16120, LH-16122 (impression of two large portions of thalli and other smaller fragments). Diagnosis. Several hundreds of millimetres long and 2-3 mm wide thalli of the Clavatoraxis type with first-order opposite branching and spine-cell rosettes completely covering internodes which are not terminal. In terminal, last order branchlets, spine cell rosettes are organized in nodes, separated by corticate, rosette-free internodes. Description. Thalli supported by a main axis which is several hundreds of millimetres long and 2-3 mm wide, robust in overall appearance (PI. 3). Branches attached to main axis are opposite and may further branch dichotomously or trichotomously (PI. 3; PI. 4, fig. 1). First and second order internodes are 16-44 mm long, spirally corticate (PI. 5, fig. 1). Cortication formed by six cells near the nodes which interdigitate with the six cortical cells of an adjacent node giving the central part of an internode a 12-celled cortication (PI. 5, figs 2-3). Internodes covered with spirally arranged, spine-cell rosettes, c. 0 5 mm in basal diameter (PI. 4, fig. 2; PI. 5, fig. 5). Up to 25 spiral rows of spine-cell rosettes are observed in lateral views of internodes of the main axis (PI. 3). Spine-cell rosettes are hemispherical structures formed by a large number of club-shaped spine-cells, which are 200-250 //m long (PI. 5, fig. 6). The apical swelling of such clubs is spherical, c. 100 pm in diameter and filled with sparite. As a result, the surface of a spine-cell rosette is covered by crystalline spheres. Club- shaped spine-cells radiate from the base of a rosette which is directly open to a cortical cell, leaving a gap in the cortical cell wall at the insertion point. Last order branchlets organized in whorls of six are c. 3^4 mm long and formed by swollen nodes with short corticate internodes (PI. 4, fig. 3). Nodes are formed by spine-cell rosettes with wedge-shaped (not club-shaped) EXPLANATION OF PLATE 4 Figs 1-4. Clavatoraxis robustus gen. et sp. nov. 1, 3, paratype LH-161 12. 1, mature portion of thallus showing opposite branching; x 1. 3, detail, showing six branchlets per node; x 4. 2, paratype LH-161 13; young portion of thallus showing spiral arrangement of spine cell rosettes; x 2. 4, detail of mature branchlets in paratype LH-161 14; x4-5. PLATE 4 MARTIN-CLOSAS and DIEGUEZ, Clavatoraxis 1144 PALAEONTOLOGY, VOLUME 41 spine cells, whereas internodes are rosette-free (PI. 5, figs 7-8). Young, apical parts of thalli show closely packed last-order branchlets with nodes almost superimposed without leaving space for internodes. In the mature parts, last-order branchlets become loosely opened and their nodes are separated by short internodes. Since last order internodes are not covered by spine-cell rosettes, the cortication may be observed from the outside of thalli (PI. 5, fig. 4). However, only the wall of cortical cells adjacent to the axis is calcified and this results in the fossil internode having a striated surface (PI. 4, fig. 4). No fructifications have yet been found attached to the thalli studied. The vegetative characters of Clavator reidii as described by Harris (1939) are largely like the last-order branchlets of thalli of Clavator axis robustus described here. This may be evidence that Clavatoraxis robustus bore fructifications of the clavatoroid type or even that it represents the vegetative remains of genus Clavator. However, this conclusion is not supported by the material studied, which only contains atopocharoid utricles dispersed in the sediment around the thalli. Such utricles are globular, bottle-shaped and do not present calcified gyrogonites (PI. 5, fig. 9). Comparisons. Young and closely packed, terminal branchlets of the thallus of Clavatoraxis robustus are similar in external appearance to Munieria baconica Deecke, 1883 as figured by Conrad and Radoicic (1971, fig. 4) and Bystricky (1976, pi. 4, fig. 7). Also, the mature, loosely packed terminal branchlets have a superficial similarity to Munieria grambasti Bystricky, 1976. However, our material differs by having the nodes of such branchlets formed by spine-cell rosettes whereas in the genus Munieria such nodes are devoid of rosettes and bear cylindrical cells. Remarks. This is the most common charophyte found at Las Hoyas. Clavatoraxis diaz-romerali sp. nov. Text-figures 2a-c, 3a-d Derivation of name. After Mr Armando Diaz-Romeral from Cuenca (Spain), in acknowledgement for collecting and making available the material upon which this study is based. Holotype. Specimen LH-16123 from the collection of Mr Armando Diaz-Romeral, Museo de Cuenca (Cuenca, Spain), deposited in the Unidad de Paleontologia, Universidad Autonoma de Madrid. Paratype. Specimen LH-16101 and thin section LH-16124 prepared from the same sample (Museo de Cuenca). Type horizon and locality. Second lithosome of finely laminated limestones of the La Huerguina Formation at Las Hoyas (Cuenca, Spain). Material. Only the type material is known for this species. Diagnosis. Thallus of Clavatoraxis filiform and rigid in overall appearance. Nodes, separated by long internodes. Spine-cell rosettes small and formed by a reduced number of wedge-shaped calcite crystals are dispersed on the main axis leaving large bare areas. Description. Thalli filamentous and rigid in overall appearance. Internodes corticate and extremely long in comparison with the size of nodes and branchlets (Text-fig. 2a). Each internode is 6-9-20T mm long and EXPLANATION OF PLATE 5 Figs 1-9. Clavatoraxis robustus gen. et sp. nov.; thin sections of rock sample LH-16115. 1, thin section LH- 16116; tangential section through thallus showing cortical cells (c) covered by spine cell rosettes (r); x 15. 2, 7-8, thin section LH-161 17. 2, transverse section through proximal part of internode. 7, oblique section of young branchlet. 8, longitudinal section of mature branchlet. All x 15. 3-4, thin section LH-161 18. 3, transverse section through distal part of internode; x 15. 4, tangential section through branchlet; x 15. 5-6, thin section LH-161 19. 5, tangential section through surface of internode showing arrangement of spine-cell rosettes; x 15. 6, detail of spine-cell rosette; x 25. 9, thin section LH-16120; longitudinal section of atopocharoid utricle; x 30. PLATE 5 MARTIN-CLOSAS and DIEGUEZ, Clavatoraxis 1146 PALAEONTOLOGY, VOLUME 41 text-fig. 2. Clavatoraxis diaz-romerali sp. nov. a, LH-16123 (holotype); x2-5. B, detail of internode of holotype; x 5. c, LH- 16101 (paratype), detail of node bearing six branchlets (two cut off arrowed); x 15. 0-7-0-9 mm wide. Corticate cells, thin (80-100 /im in diameter) almost vertical or only slightly spiralized (Text- fig. 3b). Internodal cell 180-230 pm in diameter. Spine-cell rosettes hemispherical or hemiellipsoidal, 135-350 pm in maximum basal diameter, formed by wedge-shaped cells, inserted in the internodal cells or in last order branchlets leaving a gap in the wall at the insertion point (Text-fig. 3d). Spine-cell rosettes are dispersed on internodes, leaving large bare areas between them (Text-fig. 3c). Nodes bear six short, 1 -2—2-2 mm long, last order branchlets which are identical to last order branchlets of Clavatoraxis robustus (Text- figs 2c, 3a). They are covered by swollen spine-cell rosettes inserted at regular intervals (Text-fig. 2b). This gives a noded appearance to such last order branchlets. No fructifications have been found associated with the thalli of this species, but from the presence of spine- cell rosettes we infer that they are clavatoraceans. Comparisons. This species differs from C. robustus in its filamentous appearance and by bearing dispersed, rather than closely distributed, spine-cell rosettes. However, last order branchlets of the two species are identical. The cortication of this filamentous Clavatoraxis makes it easy to distinguish from other filamentous species of genus Palaeonitella which are ecorticate. Remarks. This species is rather uncommon in the lacustrine laminites studied. It has only been found in two samples which contain abundant thalli preserved together and associated with MARTIN-CLOSAS AND DIEGUEZ: CRETACEOUS CHAROPHYTES 1147 text-fig. 3. Clavatoraxis diaz-romerali sp. nov.; thin section LH-16124 of rock slice containing LH-16101. A, section through node; x 20. b, tangential section through internode; x 15. c, longitudinal section through internode; x 15. d, detail of spine-cell rosettes; x 55. Palaeonitella vermicularis. This may indicate peculiar ecological conditions for this species of Clavatoraxis. Owing to the identical structure of last order branchlets in both species of Clavatoraxis , the hypothesis that they are merely ecotypes of the same biospecies cannot be ruled out. COMPARISON WITH FRUCTIFICATIONS DISPERSED IN MARLY FACIES This fossil assemblage of vegetative remains may be compared with the charophyte fructifications found dispersed in palustrine marly layers of the same age and formation underlying the site and described by Martin-Closas in Dieguez et al. (19956). Marly layers provide the material to carry out current research on fossil charophytes, after sieving and sorting the calcified fructifications. In such palustrine facies, charophyte fructifications belong overwhelmingly to Clavatoraceae Atopo- charoidae, and particularly to Atopochara trivolvis var. triquetra (Grambast 1968) Martin-Closas, 1996. Less common are Clavatoraceae Clavatoroidae such as Clavator harrisii Peck, 1941 and Ascidiella crucial a (Grambast 1969) Schudack, 1993 and only certain samples are enriched with the atopocharoidean Globator maillardii var. trochiliscoides (Grambast 1966) Martin Closas, 1996 and the characean Mesochara harrisii Madler, 1952. These results correlate well with the relative abundance of Clavatoraxis robustus, vegetative remains probably related to Clavatoraceae Pia, 1927. Also, the rareness of Charaxis spicatus in the finely laminated limestones matches well the rarity of Mesochara harrisii in the laevigated marls if the affinity of Charaxis with Characeae Richard ex C. A. Agardh, 1824 is admitted. This is of special interest to palaeocharologists who, until now, have been unable to determine if the relative abundances of fossil fructifications reflect similar abundances of vegetative parts. However, there is also an inconsistency in the correlation of assemblages from different lithofacies. Nitellaceae Martin-Closas and Schudack, 1991 are not represented in assemblages of fructifications from marls whereas they may be represented by vegetative remains of Palaeonitella vermicularis if the affinity of the form-genus Palaeonitella with Nitellaceae Martin-Closas and Schudack, 1991 is accepted. This may be attributed to different ecological conditions during deposition of the lacustrine limestones and palustrine marls, but is most probably related to the slight calcification of reproductive remains of most Nitellaceae. The 1148 PALAEONTOLOGY, VOLUME 41 lack of calcification of their gyrogonites, which accounts for their poor fossil record, is well documented in the charophyte literature (Feist and Brouwers 1990). TAPHONOMY AND PALAEOECOLOGY All charophyte vegetative remains found at Las Hoyas are preserved in calcite, unlike other whole remains of Lower Cretaceous charophytes (Harris 1939; Peck 1957; Musacchio 1971), which are silicified. The preservation in calcite accounts for the slight epidiagenetic corrosion which has obliterated some details of the surface of the thalli and of fructifications dispersed on the same rock samples containing thalli. From the point of view of biostratinomy the charophyte assemblage is parautochthonous. The fragility of the preserved vegetative structures and their high degree of connection, along with the absence of rootlets, indicate that only slight translocation occurred before deposition. This enables reconstruction of the composition and structure of the ancient charophyte community. Extant charophyte communities display vertical zonation according to the adaptation of species to different light intensities, whereas horizontal zonations are strongly influenced by competition with angiosperms and by different tolerances to salinity and water level fluctuations (Corillion 1957). For the Early Cretaceous, horizontal zonations have been proposed based on the association of charophyte fructifications with the remains of other organisms, such as ostracodes or foraminifers, which are salinity markers (Schudack 1993). Vertical zonations have not yet been attempted since almost nothing was known about the adaptive morphology of fossil charophyte thalli and the relative abundances of the thalli that produce the fructifications usually found dispersed in the sediments. This is now possible with the material studied (Text-fig. 4). Clavatoraxis robustus is the most abundant taxon. It possessed a strong, self-supporting structure as deduced from the rigidity given by its large, branching and corticate thalli reinforced with a coat of spine- cell rosettes. In extant Chara, spine-cell rosettes have been interpreted as protective elements against herbivory. Amphipods, aquatic coleoptera and crayfish are significant charophyte grazers in permanent shallow water lakes and ponds (Proctor 1996). Thus, Clavatoraxis robustus probably lived in permanent shallow lacustrine facies, forming a dense framework of interconnected thalli, as do some extant associations dominated by robust Chara. Clavatoraxis diaz-romerali , although filamentous in overall appearance, was also rigid as demonstrated by its fracturation in long, straight fragments. These thalli probably stood upright at the bottom of the lake. The extremely long internodes of Clavatoraxis diaz-romerali are reminiscent of a similar adaptation we have observed in extant Chara when subjected to low light intensities. Also, the open distribution of spine-cell rosettes may indicate low grazing pressure, which is typical of deeper lacustrine facies (Proctor 1996). From these observations we deduce that Clavatoraxis diaz-romerali occurred in light-limited, probably deeper, habitats in comparison with Clavatoraxis robustus. This would also explain why the two taxa are not found in association. Palaeonitella vermicularis is a relatively common taxon in the charophyte associations studied. Its filamentous and flexible thalli were found twisted round thalli of both Clavatoraxis species. This does not indicate only a close ecological relationship between Palaeonitella and Clavatoraxis in the lake of Las Hoyas but also that Palaeonitella vermicularis was unable to stand upright in the bottom of the lake without the support of other plants. Thus, as in some recent associations containing Nitella , Palaeonitella vermicularis would have formed less-organized thickets that grew supported by the framework given by other charophytes, particularly Clavatoraxis. The palaeoecological role of Charaxis spicatus is not known, due to its scarcity in the fossil assemblage studied. FRESHWATER PLANT COMMUNITY EVOLUTION Charophytes are the only subaquatic macrophytes found in the palaeolake of Las Hoyas after ten years of systematic collection and study of large remains (both carbonaceous and calcitic) and palynomorphs. Plant remains other than charophytes, although abundant and well preserved, have been attributed to groups of land plants which occurred in subaerial habitats (Dieguez et al. 1995fi). MARTIN-CLOSAS AND DIEGUEZ: CRETACEOUS CHAROPHYTES 1149 text-fig. 4. Proposed zonation of charophytes in the palaeolake of Las Hoyas. Near-shore associations were dominated by Palaeonitella vermicularis (a) and Clavatoraxis robustus (b) whereas in deeper facies the latter species was replaced by Clavatoraxis diaz-romerali (c). The tree-fern represented at the shore is Weichselia reticulata , one of the most abundant plants in Las Hoyas. Crayfish represented are Delclosia martinelli, a nektonic species and Pseudastacus llopisi, a benthic species which probably lived in association with charophytes. The presence of a well structured and diverse charophyte community dominating the subaquatic environment in palustrine and lacustrine facies around the Barremian-Aptian boundary, along with the absence of other freshwater macrophytes, undermines current ideas on freshwater plant community evolution. To date it has been generally understood that ‘aquatic pteridophytes may have . . . occupied many of the aquatic niches prior to angiosperm diversification’, which occurred at the end of the Early Cretaceous (Collinson 1988, p. 326). This idea was based solely on the study of vascular plant remains. Although freshwater fern megaspores, particularly from Azolla, have been recorded in some Lower Cretaceous localities (Collinson 1980) they are far from being as abundant as charophyte fructifications. Besides this, freshwater ferns, which are small floating plants, have never been demonstrated to represent a serious competitor of charophytes. On the basis of evidence presented here we are now able to propose that before the radiation of freshwater angiosperms, continental subaquatic habitats were occupied extensively by well structured communities of charophytes in the absence of subaquatic embryophytes, whereas floating aquatic ferns were present only at certain localities. This hypothesis was suggested from the study of fossil charophyte fructifications by Martin-Closas and Serra-Kiel (1991) and is now supported by data from the fossil record of whole plant remains from Las Hoyas. The significance of subaquatic macrophytes for animal ecology is proposed to falsify this hypothesis. If charophytes were not the dominant macrophytes at Las Hoyas, benthic habitats would have remained devoid of any macrophytic vegetation, owing to the absence of subaquatic embryophytes during the whole Palaeozoic and early Mesozoic until the diversification of angiosperms. However, this is an untenable hypothesis, since huge populations of the crayfish Pseudastacus llopisi Via, 1971 were found at Las Hoyas, indicating that a significant macrophytic biomass existed to support them (Rabada 1993). Extant astacid crayfish feed not only on macrophytes but also need a dense 1150 PALAEONTOLOGY, VOLUME 41 macrophyte cover in which to shelter (Hobbs III 1991). The need for significant macrophytic cover formed by charophytes has also been given as an explanation for the development of particular groups of aquatic insects during the Early Cretaceous (Ponomarenko and Popov 1980). A consequence of this conclusion is that freshwater vegetation remained extremely conservative during plant evolution. Charophytes, which are considered to be among land plant ancestors (Kenrick 1994), appear to have dominated the subaquatic macrophytic vegetation from the late Silurian, when they are first recorded (Ishchenko and Ishchenko 1982), until the radiation of subaquatic angiosperms, which occurred by the end of the Early Cretaceous (Mai 1985). CONCLUSIONS Four charophyte species, preserved as large plant remains, are described from the palaeontological site of Las Hoyas (Lower Cretaceous of the Iberian Ranges, Spain). A new genus (C lav at or axis) has been created to include fossil vegetative remains related to Clavatoraceae. This genus includes two new species, Clavatoraxis robustus sp. nov. and C. diaz-romerali sp. nov. The other species are Charaxis spicatus sp. nov., probably related to fossil Characeae, and Palaeonitella vermicularis sp. nov., perhaps related to fossil Nitellaceae. The abundance of taxa represented in the vegetative remains of charophytes found at Las Hoyas correlates well, at the family level, with the corresponding abundance of calcified fructifications found dispersed in sediments in the same formation. The freshwater plant community of Las Hoyas was dominated by charophytes. Clavatoraxis robustus was the most abundant species and formed dense populations in the near-shore facies of the lake. Clavatoraxis diaz-romerali lived in deeper environments. Both species were associated with the filamentous Palaeonitella vermicularis , which was supported by other charophyte thalli. Charophytes were the only subaquatic macrophytes of Las Hoyas, in the absence of submerged embriophytic macrophytes. This appears to indicate that submerged plant communities were dominated by charophytes until the radiation of subaquatic angiosperms, i.e. during most of the Palaeozoic and Mesozoic. Acknowledgements. This paper is a contribution to the EU project ERBCHRX-CT93-0164 of the Human Capital and Mobility Program. It also contributes to the Spanish projects DGICYT PB93-0284 and DGICYT PB95-1 142-CO2-01. We acknowledge some key references provided by Mme N. Grambast-Fessard (Universite de Montpellier II). Constructive criticisms from M. Feist (Universite de Montpellier II) and an anonymous reviewer improved the text. REFERENCES agardh, c. a. 1824. Systema Algarum. Lundae Literis Berlingianis, Lund, 312 pp. alvaro, m., capote, r. and vegas, R. 1979. Un modelo de evolucion geotectonica para la Cadena Celtiberica. Acta Geologica Hispanica, 14, 172-181. bystricky, j. 1976. Munieria grambasti sp. nov. in Kalk-Gerolle der “Upohlav-Konglomerate” des Mittleren Vah-Gebietes (Klippenzone, Westkarpaten). Geologica Carpathica , 27, 45-64. cherchi, a., gusic, i., schmidt, m. and schroeder, r. 1981. Lacustrine Middle Cretaceous with Munieria grambasti sarda n. ssp. (Charophyta?) of Alghero (NW Sardinia). Revue de Micropaleontologie , 23, 138-150. collinson, m. e. 1980. A new multiple-floated Azolla from the Eocene of Britain with a brief review of the genus. Palaeontology , 23, 213-229. - 1988. Freshwater macrophytes in palaeolimnology. Palaeogeography, Palaeoclimatology, Palaeoecology, 62, 317-342. conrad, m. a. and RADOicic, R. 1971. On Munieria bakonica Deecke (Characeae) and Clypeinal solkani, n. sp. (Dasycladaceae). A case of homeomorphism in calcareous green Algae. Comptes Rendus des Seances de la Societe de Physique et Histoire Naturelle de Geneve, Nouvelle Serie, 6, 87-95. corillion, r. 1957. Les Charophycees de France et de l’Europe Occidentale. Travaux du Laboratoire de Botanique, Faculte des Sciences , Angers, 11-12, 1-499. deecke, w. 1883. Uber einige neue Siphoneen. Neues Jahrbuch fur Mineralogie, Geologie und Palaontologie, 1, 1-14. MARTIN-CLOSAS AND DIEGUEZ: CRETACEOUS CHAROPHYTES 1151 DIEGUEZ, C., MARTIN-CLOSAS, C., MELENDEZ, N., RODRIGUEZ-LAZARO, J. and TRIN^AO, P. 1995fl. Biostratigraphy. 77-79. In melendez, N. (ed.). Las Hoyas. A lacustrine Konservat-Lagerstatte, Cuenca, Spain. Universidad Complutense de Madrid, Madrid, 89 pp. — trin^ao, p., marti'n-closas, c. and lopez-moron, n. 1995 b. Palaeobotany. 29-32. In melendez, n. (ed.). Las Hoyas. A lacustrine Konservat-Lagerstatte, Cuenca, Spain. Universidad Complutense de Madrid, Madrid, 89 pp. feist, M. and brouwers, E. 1990. A new Tolypel/a from the Ocean Point dinosaur locality. North Slope, Alaska and the Late Cretaceous to Paleocene nitelloid Charophytes. Bulletin of the United States Geological Survey, 1990, F1-F4. fregenal-martinez, m. a. and melendez, N. 1993. Sedimentologia y evolution paleogeografica de la cubeta de Las Hoyas (Cretacico inferior, Serrania de Cuenca). Cuadernos de Geologia Iberica, 17, 231-256. - 1994. Sedimentological analysis of the Lower Cretaceous lithographic limestones of the “Las Hoyas” fossil site (Serrania de Cuenca, Iberian Range, Spain). Geobios, Memoire Speciale, 16, 185-193. gess, r. w. and hiller, n. 1995. Late Devonian charophytes from the Witteberg Group, South Africa. Review of Palaeobotany and Palynology, 89, 4 1 7—428. gomez-fernandez, j. c. and melendez, N. 1991. Rhythmically laminated lacustrine carbonates in the Lower Cretaceous of La Serrania de Cuenca Basin (Iberian Ranges, Spain). Special Publication of the International Association of Sedimentologists, 13, 245-256. grambast, L. 1966. Un nouveau type structural chez les Clavatoracees, son interet phylogenetique et stratigraphique. Comptes Rendus de I'Academie des Sciences, Paris, 262, 1929-1932. - 1968. Evolution of the utricle in the Charophyta genera Perimneste Harris and Atopochara Peck. Journal of the Linnean Society (Botany), 61 (384), 5-11. 1969. La symetrie de l’utricule chez les Clavatoracees et sa signification phylogenetique. Comptes Rendus de I'Academie des Sciences, Paris, 269, 878-881. groves, J. 1933. Fossilium Catalogus , II: Plantae, Pars 19: Charophyta. W. Junk, Berlin, 74 pp. Harris, t. m. 1939. British Purbeck Charophyta. British Museum (Natural History), London, 83 pp. heer, o. 1855. Flora Tertiaria Helvetiae I. J. Winterthur, Wuster, 118 pp. hill, c. R. and el-khayal, A. a. 1983. Late Permian plants including Charophytes from the Khuff Formation of Saudi Arabia. Bulletin of the British Museum ( Natural History ), Geology Series, 37, 105-112. hobbs in, H. H. 1991. Decapoda. 823-858. In thorp, j. h. and covich, a. p. (eds). Ecology and classification of North American freshwater invertebrates. Academic Press, San Diego, 911 pp. horn af rantzien, H. 1956. An annotated check-list of genera of fossil Charophyta. Micropaleontology , 2, 243-256. ishchenko, t. a. and ishchenko, A. A. 1982. [Finding of charophytes in the Upper Silurian of Podolia.] 21-32. In teslenko, Yu. v. (ed.). [Systematics and evolution of ancient plants.] Nauk Dumka, Kiev, 132 pp. [In Russian]. kenrick, p. 1994. Alternation of generations in land plants: new phylogenetic and palaeobotanical evidence. Biological Review, 69, 293-330. kidston, R. and lang, w. h. 1921. On Old Red Sandstone plants showing structure, from the Rhynie Chert Bed, Aberdeenshire. Part 5. The Thallophyta occurring in the Peat-Bed, the succession of the plants throughout a vertical section of the bed, and the conditions of accumulation and preservation of the deposit. Transactions of the Royal Society of Edinburgh, 52, 855-902. lindley, J. 1836. A natural system of botany. 2nd edition. Longman, London, 526 pp. madler, k. 1952. Charophyten aus dem Nordwestdeutschen Kimmeridge. Geologisches Jahrbuch, 67, 1-46. mai, d. h. 1985. Entwicklung der Wasser- und Sumpfpflanzen-Gesellschaften Europas vor der Kreide bis ins Quartar. Flora, 176, 449-511. marti'n-closas, c. 1996. A phylogenetic system of Clavatoraceae (fossil Charophyta). Review of Palaeobotany and Palynology, 94, 259-293. -and schudack, m. e. 1991. Phylogenetic analysis and systematization of post-Paleozoic Charophytes. Bulletin de la Societe Botanique de France, Actualites Botaniques , 1, 53-71. — and serra-kiel, j. 1991. Evolutionary patterns of Clavatoraceae (Charophyta) analysed according to environmental change during Malm and Lower Cretaceous. Historical Biology, 5, 291-307. martinez-delclos, x., nel, a. and popov, y. a. 1995. Systematic and functional morphology of Iberonepa romerali n. gen, n. sp. Belostomatidae Stygeonepinae from the Spanish Lower Cretaceous (Insecta, Heteroptera, Nepomorpha). Journal of Paleontology, 69, 496-508. mas, J. R. 1981. El Cretacico inferior de la region noroccidental de la Provincia de Valencia. Seminarios de Estratigrafia, Serie Monografias, 8, 1-408. 1152 PALAEONTOLOGY, VOLUME 41 McGowan, g. and EVANS, s. e. 1995. Albanerpetontid amphibians from the Cretaceous of Spain. Nature, 373, 143-145. melendez, n. 1983. El Cretacico inferior de la region de Canete-Rincon de Ademuz (Provincias de Cuenca y Valencia). Seminar ios de Estratigrafia, Serie Monografias, 9, 1-242. migula, w. 1897. Die Characeen Deutschlands, Osterreichs und der Schweiz. 1-765. In rabenhorst, x. (ed.). Kryptogamic Flora. Vol. 5. E. Kummer, Leipzig, 765 pp. musacchio, E. a. 1971. Charophytas de la Formacion La Amarga (Cretacico inferior), Provincia de Neuquen, Argentina, Revista del Museo de La Plata, 4 (37), 19-38. peck, r. E. 1941. Lower Cretaceous Rocky Mountain nonmarine microfossils. Journal of Paleontology, 15, 285-304. — 1957. North American Charophyta. Professional Paper of the United States Geological Survey, 294A, 1-44. PEREZ-MORENO, B. P., SANZ, J. L., BUSCALIONI, A. D., MORATALLA, J. J., ORTEGA, F. and RASSKIN-GUTMAN, D. 1994. A unique multitoothed ornithomimosaur from the Lower Cretaceous of Spain. Nature, 370, 363-367. pia, J. 1927. Handbuch der Palaeobotanik. Vol. I. R. Oldenbourg Verlag, Munich, 88 pp. Ponomarenko, A. G. and popov, Y. a. 1980. Palaeobiocenosis of Early Cretaceous Mongolia lakes. Palaeontological Journal, 3, 1-10. poyato-ariza, F. j. 1995. A revision of the ostariophysan fish family Chanidae, with special reference to the Mesozoic forms. Palaeo Ichtyologica, 6, 1-52. proctor, v. w. 1996. Charophytivory : 0-3 billion years of previously unexplored coevolution. Abstracts from the Second International Symposium on Extant and Fossil Charophytes, Madison ( Wisconsin ), 1, 29. rabada, d. 1993. Crustaceos decapodos lacustres de las calizas litograficas del Cretacico inferior de Espana: Las Hoyas (Cuenca) y el Montsec de Rubies (Lleida). Cuadernos de Geologia Iberica, 17, 345-370. salas, R. and casas, a. 1993. Mesozoic extensional tectonics, stratigraphy and crustal evolution during the Alpine Cycle of the Eastern Iberian Basin. Tectonophysics, 228, 33-55. SANZ, J. L., WENZ, S., YEBENES, A., ESTES, R., MARTINEZ-DELCLOS, X., JIMENEZ-FUENTES, E., BUSCALIONI, A. D., barbadillo, L. j. and via, l. 1988. An Early Cretaceous faunal and floral continental assemblage: Las Hoyas fossil-site (Cuenca, Spain). Geobios, 21, 611-613. — CHIAPPE, L. M., PEREZ-MORENO, B. B., BUSCALIONI, A. D., MORATALLA, J. J., ORTEGA, F. and POYATO-ARIZA, F. j. 1996. An Early Cretaceous bird from Spain and its implications for the evolution of avian flight. Nature, 382, 442-445. saporta, m. de 1862. Etudes sur la vegetation du Sud-est de la France a l’epoque tertiaire. Annales des Sciences Naturelles , Botanique, 4 (17), 191-311. schudack, m. e. 1989. Charophytenflora und fazielle Entwicklung der Grenzschichten mariner Jura/Wealden in den Nordwestlichen Iberischen Ketten (mit Vergleichen zu Asturien und Kantabrien). Berliner geowissenschaftliche Abhandlungen, Reihe A, 106, 409-443. — 1993. Die Charophyten im Oberjura und Unterkreide Westeuropas. Mit einer phylogenetischer Analyse der Gesamtgruppe. Berliner geowissenschaftliche Abhandlungen, Reihe E, 8, 1-209. smith, G. M. 1938. Botany. I. Algae and Fungi. Charophyceae. McGraw Hill, New York, 127 pp. teixeira, c. 1954. La flore fossile des calcaires lithographiques de Sta. Maria de Meya (Lerida, Espagne). Boletim da Sociedade Geologica de Portugal, 12, 1-14. via, l. 1971. Crustaceos decapodos del Jurasico superior del Montsec (Lleida). Cuadernos de Geologia Iberica, 2, 607-612. zeiller, r. 1902. Sobre las impresiones vegetales del Kimmeridgiense de Sta. Maria de Meya. Memorias de la Real Academia de Ciencias y Artes de Barcelona, 4 (26), 1-27. carles marti'n-closas Departament d’Estratigrafia i Paleontologia Facultat de Geologia 08071 Barcelona, Catalonia, Spain CARMEN DIEGUEZ Museo Nacional de Ciencias Naturales-C.S.I.C. Typescript received 11 July 1997 Calle Jose Gutierrez Abascal 2 Revised typescript received 12 January 1998 28006 Madrid, Spain PORIFERA AND CHANCELLOR! IDAE FROM THE MIDDLE CAMBRIAN OF THE GEORGINA BASIN, AUSTRALIA by DORTE MEHL Abstract. A rich assemblage of poriferan spicules and sclerites of the Chancelloriidae has been found in Mid Cambrian phosphatic sediments of the Georgina Basin. The hexactinellid spicules are especially diverse, and contain several new types. These include pulvinusactins (nom. nov.) and follipinules, strongly inflated triaxons which probably formed an armouring dermal layer in Thoracospongia, and cometiasters which may be the first Cambrian evidence of the Hexasterophora. Demosponge spicules, especially triaenes, are moderately diverse. Polyactine spicules with central canals are interpreted as proto-aster megascleres, which may have evolved into aster microscleres. Calcarean, heteractinellid spicules are also common. These features suggest an early Cambrian diversification of the Porifera. The systematic position of the Chancelloriidae is still controversial. Cambrian phosphatic sediments of the Georgina Basin are well known because of their well preserved fossils. Trilobites, brachiopods and molluscs were documented early (e.g. Opik 1961, 1970). The phosphates are also rich in microfossils showing soft-body preservation (Muller and Hinz 1992). Sponge spicules have been mentioned only in passing, although they are common, together with Chancelloria sclerites, in residues of washed sediments from the eastern Georgina Basin. In 1986, a joint venture project was initiated between K. J. Muller (Bonn) and J. H. Shergold (Canberra), new fossil collections in the Georgina Basin were made by Below, Laurie, Shergold and Walossek, and they also sampled from all the main formations. Material was collected in three main areas: most came from Rogers Ridge, some from Ardmore Outlier, and the remainder from the Thorntonia area. Sediments were processed in 1 5 per cent, acetic acid and the residues were screened and washed. The assemblage includes the rich fauna of sponge spicules and chancelloriid sclerites described herein. The purpose of this publication is the documentation of the Porifera and Chancelloriidae found in the Georgina Basin, and a discussion of phylogenetic and systematic aspects of these groups. STRATIGRAPHICAL AND PAL AEOECOLOGIC AL BACKGROUND The Georgina Basin covers about 325000 km2 of central northern Australia and contains exposures of Cambro-Ordovician rocks deposited in a broad, shallow epicontinental sea. During the Mid Cambrian, a series of organic-rich muds, associated with phosphorites, was deposited. The stratigraphical framework of the Middle Cambrian has been provided by Opik (1960, 1961), Smith (1972), Shergold and Druce (1980) and Shergold et al. (1985). Phosphogenesis persisted throughout most of the Mid Cambrian, and took place in the course of cyclic sedimentation related to upwelling during transgressions (Southgate and Shergold 1991). The stratigraphy and lithofacies have been described in detail by Shergold and Southgate (1986), so here only a short description of the formations is given, focusing on those in which sponge spicules and chancelloriian sclerites have been found. These are of early Mid Cambrian, Ordian to Undillan age. According to Southgate and Shergold (1991), the Mid Cambrian phosphorites of the Georgina Basin were deposited during two main transgressive episodes (Text-fig. 1). [Palaeontology, Vol. 41, Part 6, 1998, pp. 1153-1182, 7 pls| © The Palaeontological Association 1154 PALAEONTOLOGY, VOLUME 41 z § 5 X o LU CO QC LU > cc < THORNTONIA EAST z o D TREE h- z , LADY ANNIE cn 1— o 9?. YELVERTOFT 1 SEQUENCE 2 TRANSGRESSIVE SYSTEMS TRACT AND EARLY HIGHSTAND SYSTEMS TRACT Camooweal Dolomite Perkasie tr Lithofacies Association Member aj Cl li c — g LM 03 E New K o L L. O Haven | aj CO Formation 03 Graters Member — — Lithofacies ET Association 1 Warford Member 1 _l _ C Eastern New Mexico USA Eastern Greenland Germany Biochron < □C o c gj 'c _cd < a> "O T3 00 LU Himavatites Drepanites rutherfordi Juvavites magnus Malayites dawsoni Stikinoceras kerri Bull Canyon Formation (part) T rujillo Formation Orsted Dal Member Malmros Klint ! Member 1= O sz o g JD (O 5 03 CO £ Q) | = possible stratigraphic range of Aetosaurus occurrences text-fig. 6. Correlation of Aetosaurus-y\Q\&mg strata. Vertical bars indicate possible stratigraphical ranges of Aetosaurus occurrences. However, actual stratigraphical ranges are probably much shorter, but determination of this requires more precise stratigraphical data than are currently available for most Aetosaurus occurrences. As Jenkins et al. (1994) concluded, this assemblage shares many taxa with the German Stubensandstein and clearly is of Norian age. The closest similarity to the 0rsted Dal assemblage is the vertebrate assemblage of the Lower Stubensandstein, especially the co-occurrence in both units of Aetosaurus ferratus and Paratypothorax andressi, as well as Cyclotosaurus, Gerrothorax and Proganochelys. We thus regard the 0rsted Dal Member vertebrates as of early Norian age (Text- fig. 6). However, we note that Jenkins et al. (1994) presented no precise stratigraphical ordering of vertebrate fossil localities in the 1 50-200-m thick 0rsted Dal Member and that some taxa from the 0rsted Dal Member ( Gerrothorax , Plateosaurus) do not occur in the Lower Stubensandstein, but first appear in the Middle Stubensandstein (Benton 1993, table 1). Therefore, the possibility exists that the 0rsted Dal vertebrate assemblage includes temporal equivalents of both the Lower and Middle Stubensandstein. Germany In Germany, Aetosaurus is well documented from the Lower Stubensandstein {A. ferratus) and the Middle Stubensandstein (A. crassicauda) of the German Keuper (O. Fraas 1877; E. Fraas 1907; Wild 1989). Palynostratigraphy, vertebrate biostratigraphy, and sequence stratigraphy suggest that 1228 PALAEONTOLOGY, VOLUME 41 the Stubensandstein is of early to mid Norian age (Brenner 1973; Brenner and Villinger 1981; Visscher and Brugman 1981; Benton 1986, 1993; Wild 1989; Aigner and Bachman 1992; Kozur 1993; Lucas and Huber 1994). The most precise correlation available suggests that the Lower Stubensandstein is early Norian, whereas the Middle Stubensandstein is mid Norian (e.g. Benton 1993). This suggests that Aetosaurus in Germany has an early-mid Norian temporal range comparable to its temporal range in the Newark Supergroup (Text-fig. 6). Italy Wild (1989) documented Aetosaurus ferratus from the marine Calcare di Zorzino Formation ( = Zorzino Limestone) at Cene, near Bergamo in the Lombardian Alps of northern Italy. The Calcare di Zorzino Formation is a carbonate and turbidite facies that immediately overlies and is in part laterally equivalent to the Norian Dolomia Principale (= Hauptdolomit). After the regional progradation of platform carbonates (Dolomia Principale) during the early-mid Norian, extensional tectonism produced intraplatform depressions in which the Zorzino Limestone (Aralalta Group; Jadoul 1985) was deposited as patch reefs, turbiditic debris flows and lagoonal to freshwater facies (Jadoul et al. 1994). Palynostratigraphy and conodont biostratigraphy indicate that the fossil vertebrate locality in the Zorzino Limestone near Bergamo is very close to the Alaunian (mid Norian)-Sevatian (late Norian) boundary (Jadoul et al. 1994; Roghi et al. 1995; Tintori and Lombardo 1996). This indicates that the Aetosaurus occurrence documented by Wild (1989) is of late mid Norian age, and correlates with the younger part of the Himavatites columbianus Zone of the global Triassic ammonite biochronology (Tozer 1994). This provides a direct cross-correlation of an Aetosaurus occurrence to marine Triassic biochronology (Text-fig. 6). Discussion The direct cross-correlation of Aetosaurus from Italy to the middle Norian accords well with the inferred age of some Aetosaurus records in Germany and the United States. However, the German and American records suggest that Aetosaurus existed during both the early and mid Norian. Although this is consistent with cross-correlation to the Italian marine occurrence of Aetosaurus , the German and Newark Supergroup records obviously encompass a longer temporal range (Text-fig. 6) than the single Italian occurrence. The total known temporal range of Aetosaurus thus equals about four ammonite zones, which is about half of Norian time, approximately 5-7 million years on most numerical time scales (e.g. Kent et al. 1995). Aetosaurus thus emerges as a tetrapod taxon capable of providing a robust correlation across much of Late Triassic Pangaea and is an index fossil of early-mid Norian time. Acknowledgements. R. Barrick, P. Holroyd, D. Parris and M. A. Turner made it possible for us to study specimens of Stegomus. A. Tintori provided information on the age of the Zorzino Limestone, and A. Hunt provided useful discussion. The comments of two anonymous reviewers improved this article. The National Geographic Society (Grant 5412-95 to SGL) supported part of this research. REFERENCES aigner, t. and bachmann, G. H. 1992. Sequence-stratigraphic framework of the German Triassic. Sedimentary Geology , 80, 1 15-135. bajrd, D. 1986. Some Upper Triassic reptiles, footprints and an amphibian from New Jersey. The Mosasaur, 3, 125-135. ballew, K. l. 1989. A phylogenetic analysis of Phytosauria from the Late Triassic of the western United States. 309-339. In lucas, s. g. and hunt, a. p. (eds). Dawn of the age of dinosaurs in the American Southwest. New Mexico Museum of Natural History, Albuquerque, 414 pp. Benton, m. j. 1986. The Late Triassic tetrapod extinction events. 303-320. In padian, k. (ed.). The beginning of the age of dinosaurs. Cambridge University Press, Cambridge, 378 pp. 1993. Late Triassic terrestrial vertebrate extinctions: stratigraphic aspects and the record of the Germanic basin. Paleontologia Lombardia , Nuova Series, 2, 19-38. LUCAS ET AL.: TRIASSIC AETOSAURS 1229 and Clark, j. m. 1988. Archosaur phylogeny and the relationships of the Crocodylia. 295-338. In benton, m. j. (ed.). The phylogeny and classification of the tetrapods. Volume 1 : amphibians, reptiles, birds. Clarendon Press, Oxford, 377 pp. brenner, k. 1973. Stratigraphie und Palaogeographie des Oberen Mittelkeupers in Siidost- Deutschland. Arbeiten aus dem Institut fiir Geologie und Palaontologie der Universitdt Stuttgart, 68, 101-222. and villinger, E. 1981. Stratigraphie und Nomenklatur des siidwesdeutschen Sandsteinkeupers. Jahresheftte Geologische Landesamt von Baden- Wiirttemberg. 23, 45-86. camp, c. L. 1930. A study of the phytosaurs, with description of new material from western North America. Memoirs of the University of California, 10, 1—161. cornet, B. 1977. The palynostratigraphy and age of the Newark Supergroup. Unpublished Ph.D. dissertation, Pennsylvania State University. fraas, e. 1907. Aetosaurus crassicauda n. sp., nebst Beobachtungen iiber das Becken der Aretosaurier. Jahreshefte des Vereins fiir vaterlandische Naturkunde Wiirttemberg , 63, 101-109. fraas, o. 1877. Aetosaurus ferratus Fr. Die gepanzerte Vogel- Echse aus dem Stubensandstein bei Stuttgart. Wurttemberger Naturwissenschaften Jahrbuch, 33(3), 1-22. Gregory, J. t. 1957. Significance of fossil vertebrates for correlation of the Late Triassic continental deposits of North America. Report of the 20th Session, International Geological Congress, Section 2, 7-25. heckert, a. b., hunt, A. p. and lucas, s. G. 1996. Redescription of Redondasuchus reseri, a Late Triassic aetosaur (Reptilia: Archosauria) from New Mexico (U.S.A.), and the biochronology and phylogeny of aetosaurs. Geobios, 29, 619-632. hoffman, c. w. and Gallagher, p. e. 1989. Geology of the Southeast and Southwest Durham 7-5-minute quadrangles. North Carolina. Bulletin of the North Carolina Geological Survey, 92, 1-34. huber, p., lucas, s. G. and hunt, a. p. 1993a. Revised age and correlation of the Upper Triassic Chatham Group (Deep River Basin, Newark Supergroup), North Carolina. Southeastern Geology, 33, 171-193. — 19936. Vertebrate biochronology of the Newark Supergroup Triassic, eastern North America. Bulletin of the New Mexico Museum of Natural History and Science, 3, 179-186. huene, F. von 1908. On the age of the reptile faunas contained in the Magnesian Conglomerate at Bristol and in the Elgin Sandstone. Geological Magazine, 5, 99. 1920. Osteologie von Aetosaurus ferratus O. Fraas. Acta Zoologica, 1, 465-491. hunt, a. p. 1994. Vertebrate paleontology and biostratigraphy of the Bull Canyon Formation (Chinle Group, Upper Triassic), east-central New Mexico with revisions of the families Metoposauridae (Amphibia: Temnospondyli) and Parasuchidae (Reptilia: Archosauria). Unpublished Ph.D. dissertation. University of New Mexico. jadoul, F. 1985. Stratigrafia e paleogeografia del Norico nelle Prealpi Bergamasche occidentali. Rivista Italiana di Paleontologia a Stratigrafia, 91, 479-512. masetti, D., cirilli, s., berra, s., claps, m. and frisia, s. 1994. Norian-Rhaetian stratigraphy and paleogeographic evolution of the Lombardy basin (Bergamasc Alps). 15th International Association of Sedimentologists, Regional Meeting, April 1994, Ischia, Italy, Field Excursion Bl, Salermo, Italy, 5-38. JENKINS, F. A. Jr, SHUBIN, N. H., AMARAL, W. W., GATESY, S. M., SCHAFF, C. R., CLEMMENSEN, L. B., DOWNS, W. R., davidson, a. R., bonde, N. and osBjECK, f. 1994. Late Triassic continental vertebrates and depositional environments of the Fleming Fjord Formation, Jameson Land, East Greenland. Meddelelser om Gronland, Geoscience, 32, 1-25. jepsen, g. l. 1948. A Triassic armored reptile from New Jersey. State of New Jersey Department of Conservation Miscellaneous Geologic Paper, 1-20. kent, d. v., olsen, p. e. and witte, w. k. 1995. Late Triassic-earliest Jurassic geomagnetic polarity sequence and paleolatitudes from drill cores in Newark rift basins, eastern North America. Journal of Geophysical Research, 100, 14965-14998. kozur, h. 1993. Annotated correlation tables of the Germanic Buntsandstein and Keuper. Bulletin of the New Mexico Museum of Natural History and Science, 3, 243-248. lockley, M. and hunt, A. p. 1995. Dinosaur tracks and other fossil footprints of the western United States. Columbia University Press, New York. long, r. a. and ballew, k. l. 1985. Aetosaur dermal armor from the Late Triassic of southwestern North America, with special reference to material from the Chinle Formation of Petrified Forest National Park. Bulletin of the Museum of Northern Arizona, 54, 45-68. long, r. A. and murry, p. A. 1995. Late Triassic (Carnian and Norian) tetrapods from the southwestern United States. Bulletin of the New Mexico Museum of Natural History and Science, 4, 1-254. lucas, s. G. 1990. Toward a vertebrate biochronology of the Triassic. Albertiana, 8, 36-41. 1230 PALAEONTOLOGY, VOLUME 41 — 1993. The Chinle Group: revised stratigraphy and biochronology of Upper Triassic nonmarine strata in the western United States. Bulletin of the Museum of Northern Arizona, 59, 27-50. — and heckert, A. b. 1996. Late Triassic aetosaur biochronology. Albertiana, 17, 57-64. — and huber, p. 1993. Revised internal correlation of the Newark Supergroup Triassic, eastern United States and Canada. Bulletin of the New Mexico Museum of Natural History and Science, 3, 311-319. 1994. Sequence stratigraphic correlation of Upper Triassic marine and nonmarine strata, western United States and Europe. Memoir of the Canadian Society of Petroleum Geologists, 17, 241-254. — and hunt, a. p. 1993. Tetrapod biochronology of the Chinle Group (Upper Triassic), western United States. Bulletin of the New Mexico Museum of Natural History and Science, 3, 327-329. lull, r. s. 1915. Triassic life of the Connecticut Valley. Bulletin of the Connecticut Geologic and Natural History Survey, 24, 1-285. 1953. Triassic life of the Connecticut Valley revised. Bulletin of the Connecticut Geologic and Natural History Survey, 81, 1-336. lydekker, R. 1887. The fossil Vertebrata of India. Records of the Geological Survey of India, 20, 51-80. marsh, o. c. 1896. A new belodont reptile (Stegomus) from the Connecticut River Sandstone. American Journal of Science, 2, 59-62. ochev, v. g. and shishkin, m. a. 1989. On the principles of global correlation of the continental Triassic on the tetrapods. Acta Palaentologica Polonica , 34, 149-173. olsen, p. E., froelich, a. j., Daniels, D. l., smoot, J. p. and gore, p. J. w. 1990. Rift basins of early Mesozoic age. 142-170. In horton, j. w. and zullo, v. a. (eds). The geology of the Carolinas. University of Tennessee Press, Knoxville, 180 pp. — rent, D. v., cornet, b., witte, w. K. and schishe, R. w. 1996. High-resolution stratigraphy of the Newark rift basin (early Mesozoic, eastern North America). Bulletin of the Geological Society of America, 108, 40-77. Parker, J. M. hi 1966. Triassic reptilian fossil from Wake County, North Carolina. Journal of the Elisha Mitchell Society, 82, 92. ROGERS, R. R., SWISHER, C. C. HI, SERENO, P. C., MONETTA, A. M., FORSTER, C. A. and MARTINEZ, R. N. 1993. The Ischigualasto tetrapod assemblage (Late Triassic, Argentina) and 40Ar/39Ar dating of dinosaur origins. Science, 260, 794—797. roghi, G., mietto, P. and dalla vecchia, F. M. 1995. Contribution to the conodont biostratigraphy of the Dolomia di Forni (Upper Triassic, Carnian, NE Italy). Memoire Scienze Geologie, 47, 125-133. tintori, a. and lombardo, c. 1996. Gabonella agilis, gen. n. sp. n., (Actinopterygii, Perleidiformes) from the Calcare di Zorzino of Lombardy (North Italy). Rivista Italiana di Paleontologia e Stratigrafia, 102, 227-236. tozer, E. T. 1994. Canadian Triassic ammonoid faunas. Bulletin of the Geological Survey of Canada , 467, 1-663. visscher, h. and brugman, w. a. 1981. Ranges of selected palynomorphs in the Alpine Triassic of Europe. Review of Palaeobotany and Palynology, 34, 115-128. walker, a. D. 1961. Triassic reptiles from the Elgin area: Stagonolepis, Dasygnathus, and their allies. Philosophical Transactions of the Royal Society of London , Series B, 248, 103-204. westphal, f. 1976. Phytosauria. 99-120. In : Encyclopedia of paleoherpetology. Vol. 13. wild, r. 1989. Aetosaurus (Reptilia : Thecodontia) from the Upper Triassic (Norian) of Cene near Bergamo, Italy, with a revision of the genus. Revista del Museo Civico di Scienze Naturali ‘ Enrico Caffi', 14, 1-24. SPENCER G. LUCAS New Mexico Museum of Natural History and Science 1801 Mountain Road NW Albuquerque, NM 87104, USA ANDREW B. HECKERT Department of Earth and Planetary Sciences University of New Mexico Albuquerque, NM 87131-1116, USA PHILLIP HUBER Virginia Museum of Natural History Typescript received 6 March 1997 1001 Douglas Avenue Revised typescript received 3 November 1997 Martinsville, VA 24112, USA PALAEONTOLOGICAL ASSOCIATION ANNUAL ADDRESS ALL-TIME GIANTS: THE LARGEST ANIMALS AND THEIR PROBLEMS by R. MCNEILL ALEXANDER Abstract. The largest known swimming, walking and flying animals are all vertebrates. They include the blue whale (up to 190 tonnes), the largest sauropod dinosaurs (probably about 80 tonnes) and two flying animals estimated to have had masses of at least 75 kg, the pterosaur Quetzalcoatlus and the bird Argefitavis. Even larger sizes might be physically possible, but may not have been attained because problems associated with size may make excessively large animals competitively inferior. These problems are discussed with frequent reference to basic consequences of geometric similarity (areas are proportional to the squares of lengths and volumes to the cubes) and to the empirical rule that metabolic rates of similar animals tend to be proportional to (body mass)075. Excessively large animals would be liable to overheat, both in water and on land. Larger animals tend to have fewer individuals in each species, suggesting the possibility that the largest whales and dinosaurs approach the limits of size above which numbers would be unlikely to be large enough for long term viability. Even the largest dinosaurs seem to have been well able to support their weight on land. Flying animal size may have been limited more by the problem of taking off than by the power requirement for flight. The largest swimming animals are filter feeders and the largest land animals were herbivores, so neither are at the top of a long food chain. This paper reviews the largest animals known to have lived, at any time in the Earth’s history. I will consider the problems associated with their size, and ask why they did not evolve to be even larger. Aquatic, terrestrial and flying animals will be considered separately. Invertebrates cannot match the size of the largest vertebrates, so I will be concerned almost exclusively with vertebrates. Colonial animals such as corals are excluded from consideration. PRINCIPLES OF ALLOMETRY It may be helpful to start by noting some of the consequences of size differences, starting with a geometrical point. Bodies of identical shape, but different sizes (that is, geometrically similar bodies) have surface areas proportional to the squares of their lengths and volumes proportional to the cubes of their lengths: for example, a cube with sides twice as long as another has faces of four (= 22) times the area and has eight (= 23) times the volume. If the bodies are made of the same material, they have masses proportional to their volumes. Thus for different-sized animals of the same shape we expect to find mass proportional to (length)3 from which follows length proportional to (mass)033 and since area is proportional to length squared area proportional to (mass)067 [Palaeontology, Vol. 41, Part 6, 1998, pp. 1231-1245) © The Palaeontological Association 1232 PALAEONTOLOGY, VOLUME 41 Plainly, even closely related animals of different sizes are not precisely the same shape. Lions have relatively smaller brains and eyes than domestic cats (Davis 1962) but, in many other respects, groups of animals are remarkably close to geometric similarity. For example, the lengths of whales ranging from 30 kg dolphins to 100 tonne blue whales are proportional to (body mass)034 (Economos 1983). The lengths of the limb bones of mammals ranging from shrews to elephants are proportional to (body mass)035 (Alexander et al. 1979). However, in some cases we find marked deviations from geometric similarity. If Bovidae (antelopes, etc.) are considered separately from other mammals, their limb bone lengths are proportional to (mass)0 26 (Alexander et al. 1979). The wing spans of birds other than hummingbirds tend to be proportional to (mass)0 39 and those of hummingbirds to (mass)053 (Rayner 1988). Further to those geometrical points, we need to note that the pace of life is generally slower for larger animals. These generally make repetitive movements at lower frequencies than small animals : for example, sparrows in flight make about 20 wing beat cycles per second and swans about three cycles per second (see Rayner 1988). There is a tendency, in groups of related animals, for frequencies to be about proportional to (body mass)-025. For example, wing beat frequencies of birds (excluding hummingbirds) are proportional to (mass) 0 27 (Rayner 1988) and heart frequencies of mammals to (mass)-025 (Stahl 1967). However, not all frequencies scale so steeply. The stride frequencies of mammals using corresponding gaits are about proportional to (shoulder height)-0'5 and so to (mass)-017 (Pennycuick 1975). Another aspect of the slower pace of life for larger animals is that metabolic rates do not increase in proportion to body mass. The metabolic rate of a 2000 kg elephant is not 10000 times that of a 0-2 kg rat, but only about 1000 times. More generally, metabolic rates of similar animals of different sizes are found to be about proportional to (body mass)0 75. This applies not only to resting rates (Calder 1984), but also approximately to field metabolic rates and maximum aerobic rates, both of which are proportional, for mammals, to (mass)0 81 (Weibel and Taylor 1981 ; Nagy 1987). There are marked differences between groups, notably between ectotherms and endotherms; even with its body at a mammal-like temperature of 37 °C, a typical lizard uses oxygen only about one-quarter as fast as a mammal of equal mass (see Alexander 1981, fig. 1 1-4). However, within each group the 0-75 power law holds well. Most of the attempts that have been made to explain this law apply only to a limited range of organisms. However, a recent very genera4 theory (West et al. 1997) derives the law by considering the energy cost of distributing resources through a branching network of tubes (for example a blood system) in organisms of different sizes. The 0-75 power law of metabolic rate is related to the —0-25 law of frequencies. Consider two muscles that exert equal stresses while shortening by equal fractions of their lengths. The forces they exert are proportional to their cross sectional areas, and the distances they shorten are proportional to their lengths, so the amounts of work they do (force multiplied by distance) are proportional to their volumes, and so to their masses. If these muscles make up equal fractions of body mass and contract with frequencies proportional to (body mass)-0 25, their power outputs are proportional to (mass)0 75. There is another general rule (in this case, a very imprecise one) that is useful. Large species tend to have fewer members than small ones so that in most cases, for instance, the world population of a species of elephant will be fewer in number than the world population of a species of mouse. There is a general tendency for the population density of a species to be proportional to (body mass)-0 75 (Cotgreave 1993). This has been interpreted as implying an energetic equivalence rule: if species of different sizes have numbers proportional to (mass)-0 75 and metabolic rates proportional to (mass)0 75, total rate of food intake will be the same for species of all sizes. However, several points should be noted. First, there is a great deal of scatter about the regression line: in many cases, a common species is 1000 times as numerous as a rare one of similar size. Second, some studies of particular groups have found exponents markedly different from —0-75. And finally, carnivore species tend to have many fewer members than herbivore species of similar size (Peters and Raelson 1984). ALEXANDER: ALL-TIME GIANTS 1233 Subsequent sections of this paper apply the principles expounded in this one, in discussions of the consequences of large size for aquatic, terrestrial and flying animals. Crocodiles divide their time between land and water; they will be treated here as aquatic. SWIMMING ANIMALS (Text-fig. 1) First, I shall consider aquatic animals. Before discussing the consequences for them of large size, I shall review some of the largest of them. The blue whale ( Balaenoptera musculus ) is not only the heaviest modern animal, but also the heaviest known to have lived at any time. Adult females, which grow larger than males, reach lengths of 30 m and masses (determined by weighing the carcase in pieces) of over 120 tonnes (Lockyer 1976). The heaviest recorded weighed 190 tonnes. There are eight other species of baleen whales, with adult masses ranging from 7 tonnes (the minke whale, B. acutorostrata) upwards. All of them are filter feeders, using the fringes of their baleen to strain small crustaceans or fishes from the plankton. Antarctic krill (Euphausioidea, about 50 mm long) are particularly important for the blue whale, but other species, with differently spaced bristles on their baleen, take mainly copepods or fishes (Nowak 1991). There is a marked tendency for very large swimming animals to be filter feeders. The largest modern fishes are the whale shark, Rhincodon typus , which grows up to 13m long with an estimated mass of at least 1 5 tonnes ; and the basking shark, Cetorhinus maximus, which reaches 1 1 m and 8 tonnes (Matthews 1995). The basking shark feeds largely on copepods and crab larvae (Matthews and Parker 1950). The whale shark also feeds on zooplankton, but I know of no more precise description of its diet. Both these large sharks obtain their tiny prey by filtering. The largest known teleost fish, the Jurassic Leedsichthys (Martill 1988), was another filter feeder. It is known only from fragments, but these include the tail fin whose span of 2-7 m suggests a body length of 13 m. Some other very large aquatic animals are predators on larger prey, mainly fishes and squid. By far the largest is the sperm whale, Phvseter catodon , which attains 19 m and over 50 tonnes (Lockyer 1976). It feeds mainly on ammoniacal squid from substantial depths, prey whose bloated form suggests that they may not swim fast (Denton 1974). The killer whale ( Orcinus orca; males reach 10 m and 9 tonnes) eats seals as well as fishes and squid (Nowak 1991). Male elephant seals, Mirounga leonina, have masses up to 3-7 tonnes and again feed mainly on fishes and squid (Nowak 1991). The great white shark. Car char odon carcharis , preys on seals and dolphins as well as fishes (Wheeler 1975). The largest recorded modern specimen was 6-4 m long with a mass of 3-2 tonnes, but there are fossil Car char odon teeth whose size indicates a body length of 13 m (Randall 1973). The largest known aquatic reptiles include the Cretaceous pliosaur Kronosaurus (12 m long; Romer 1959); the Triassic ichthyosaur Shonisaurus (14 m; Kosch 1990); and the Cretaceous crocodilian Deinosuchus (15 m; Steel 1989). All of these were apparently predators, but we have no direct evidence of their diets. In comparison with these giants, the largest predatory teleosts are unimpressive. Current angling records are 0-71 tonnes for black marlin ( Makaira indica) and 0-68 tonnes for bluefin tuna ( Thunnus thymus; Matthews 1995). Both feed mainly on schooling fish (Wheeler 1975). The ocean sunfish Mo/a grows larger, apparently to more than one tonne, but eats smaller prey such as jellyfishes and young fish (Wheeler 1975). Giant squid ( Architeuthis harveyi) have immensely long tentacles, but the mantle is seldom more than 3 m long, and large specimens probably have masses of around 1 tonne (Clarke 1966). We m,ay ask why very large swimmers tend to be filter feeders. Consider first the rate (volume per unit time) at which water must be filtered. For animals taking the same food, this must be proportional to metabolic rate, and so to (body mass)0 75. Blue whales and other rorquals take mouthfuls of water, which are then squeezed out through the baleen to filter out the food. If mouth volume is proportional to body mass and mouth-filling frequency (like other physiological frequencies, see above) to (mass)-025, the rate of filtration will be proportional to (mass)0 75, as 1234 PALAEONTOLOGY, VOLUME 41 Metres text-fig. 1. Some of the largest known swimming animals, drawn to a uniform scale. The names of extinct animals are asterisked, a, Car char odon\ b, Rhincodon; c, Physeter ; d, Thunnus', E, Mala: f, Kronosaurus* ; G, Shonisaurus* ; H, Balaenoptera ; and I, Orcinus. required. Large size will present no problem. Right whales (Balaena) and filter-feeding sharks swim with their mouths open, straining out food with their baleen or gill rakers. Their rates of filtration will be mouth area multiplied by swimming speed. Geometric similarity would make mouth area proportional to (mass)0 67 so larger animals would have to swim a little faster to make filtration rates proportional to (mass)0 75 as required. Considerations of energy cost suggest problems for very large filter feeders. Animals taking the same prey can be expected to have filters of equal mesh size, even if their bodies are very different in size. To obtain volumetric flow rates proportional to (mass)0 75 through filters whose areas are proportional to (mass)0 67, linear flow rates and so pressure drops must be proportional to (mass)0 08. The power required for filtration is the volumetric flow rate multiplied by the pressure drop, and so will be proportional to (mass)0 75 x (mass)0 08 = (mass)0 83. Thus larger filter feeders may have to use a larger proportion of their food intake to drive the filtration process than smaller filter feeders. However, this conclusion could be avoided if fractal design made filter area increase with slight positive allometry (see Pennycuick 1992 on fractals). Also, at least some of the baleen whales appear to have fore stomachs which function as fermentation chambers, like the rumen of cattle (Herwig et al. 1984). If the chitin of crustacean exoskeletons is fermented, this may improve food utilization and so reduce the volume of water that must be filtered, alleviating the problem of energy cost for these very large filter feeders. In any case, the arguments in this paragraph fail to explain why the largest aquatic animals are filter feeders. Now consider predation on prey which are too large to be filtered and must be pursued ALEXANDER: ALL-TIME GIANTS 1235 individually. Slow prey may be able to escape from larger predators if they are better at swerving; the critical property is lateral acceleration (Howland 1974). The forces available for swerving can be expected to be proportional to muscle cross sectional area and so to (mass)0 67, and the accelerations they will provide will be proportional to (mass)-0'33. Thus predators can be expected to have trouble catching smaller prey. If the discrepancy of size between predators and prey is greater for larger predators, these may have most difficulty in catching prey. It may be significant that sperm whales feed largely on (probably) sluggish ammoniacal squid and killer whales hunt in groups, improving their chances of catching prey by making it harder for prey to escape by swerving (Howland 1974)." These arguments seem inconclusive; they fail to make it clear why so many of the largest swimmers are filter feeders. Another possible reason relates to the problem of maintaining a population of viable size, of very large animals. Filter feeders, taking food from relatively low in the food chain, have a more abundant energy supply than predators taking prey from higher in the food chain. If size were limited by the problem of obtaining enough energy to support a viable population, we would expect filter feeders to evolve to larger sizes than predators on large prey. Similarly, among terrestrial mammals herbivores have evolved to larger sizes than carnivores, and herbivore species have higher population densities than carnivore species of similar size. Similar reasoning might lead us to expect that because endotherms such as whales need more energy than ectotherms of similar size such as sharks, the largest animals should be ectotherms, which they are not. Similarly, terrestrial mammals need more energy than similar-sized (ectothermic) reptiles; thus we might expect reptiles to be more abundant than mammals of equal size, but they are not (Peters, 1983). These discrepancies show that we should be cautious in formulating arguments of this kind. It seems unlikely that the blue whale has reached the maximum size consistent with a viable population. Prior to human exploitation, it is estimated that the world population comprised 200000 individuals (Nowak 1991). A recent estimate that the minke whale population of the north- east Atlantic is now about 120000 has raised confidence in the viability of this species to such an extent that it has been suggested that some hunting could be permitted (Motluk 1996). In their guidelines for assessing threats of extinction, Mace and Lande (1991) associated their lowest level of threat (‘vulnerable’) with a population size of only 10000 or less. However, they were concerned with extinction in periods of the order of centuries, whereas our concern is with viability over periods of millions of years. If smaller populations were viable, larger animals would be possible. Another potential problem for very large animals is that excessively large ones would overheat. An animal may be thought of as a core, in which heat is liberated by metabolism and in which blood circulation maintains uniform temperature; enclosed by an insulating layer of skin with (in some cases) blubber, fur or feathers. The physics of heat conduction tells us that the temperature difference across the insulating layer is proportional to the metabolic rate divided by the thermal conductance of the insulation. Metabolic rate can be expected to be proportional to (body mass)0 75, as previously noted. Conductance should be proportional to surface area divided by insulation thickness, and so to (mass)067/(mass)033 = (mass)033. Then the temperature difference across the insulating layer will be proportional to (mass)0 75/(mass)033 = (mass)042, and excessively large animals would overheat. Ryg et al.'s (1993) calculations indicate that when a blue whale makes full use of the heat-insulating potential of its blubber, its basal metabolism is enough to heat it 40 K (centigrade degrees) above ambient, maintaining a typical mammalian body temperature of 38 °C in sea water at its freezing point of —2 °C. Field metabolic rates of large mammals are typically twice basal rates (Nagy 1987), so the whale’s problem is not to keep warm, but to avoid overheating. It does this by sending blood to the dermis, bypassing the blubber. Hokkanen (1990) calculated that with maximal blood flow to the dermis, a blue whale metabolizing at F5 times the estimated basal rate could just avoid overheating in water at 29 °C. Tropical surface water temperatures are about 27 °C. These data suggest that the largest whales may be near the maximum size set by the overheating problem. Even if this is the case, the largest fishes and aquatic reptiles are in no danger of overheating. They are much smaller than the blue whale, and their metabolic rates are presumably far below 1236 PALAEONTOLOGY, VOLUME 41 those of similar-sized mammals. Tunnies and other ‘warm blooded’ fishes owe their elevated body temperatures more to vascular heat exchangers than to their size (Carey 1982). Leatherback turtles (Dermochelys coriacea ) have metabolic rates intermediate between predictions for reptiles and mammals of their mass, enabling a 400 kg specimen to keep its body 18 K warmer than the water (Paladino et al. 1990). TERRESTRIAL ANIMALS (Text-fig. 2) Now I will review and discuss the largest terrestrial animals. Among these, the largest known are sauropod dinosaurs, all of them extinct. The linear dimensions of sauropods are known from skeletons but their masses can only be estimated. This has been attempted in two ways. First, scale models have been made of the animals as they are believed to have appeared in life and their volumes have been determined, preferably by a method that depends on Archimedes’ Principle. Then the volume of the living animal has been estimated by scaling up from the model, and the animal’s mass calculated by assuming a density in the range observed for related modern animals (see Alexander 1985). Alternatively, the circumferences of fossil leg bones have been used to estimate body mass, by extrapolating from empirical relationships established for modern mammals (Anderson et al. 1985). A relationship based on mammals seems appropriate because we know from fossil footprints that dinosaurs did not adopt the sprawling stance of modern reptiles, but walked more like mammals (Thulborn 1990; Lockley 1991). The largest dinosaur known from a reasonably complete (albeit composite) skeleton is Brachiosaurus brancai. It is about 25 m long, measured along the vertebral column (Paul 1988). Its mass has been estimated by both methods, yielding values ranging from 32 to 87 tonnes (Alexander 1989). Paul (1988) and Alexander (1989) both give values of 45-50 tonnes, and these are probably the best estimates. The Chicago skeleton of B. altithorax is a little smaller (Paul 1988). Other large sauropods known by more-or-less complete skeletons are Apatosaurus louisae (about 35 tonnes according to Alexander 1989, although Paul, who prefers very ‘skinny’ reconstructions, gives it only half that mass), and Diplodocus carnegiei (estimates range from 6 to 19 tonnes). A few bones are known of sauropods that may have been heavier than Brachiosaurus. The bones described as ‘ Ultrasaurus' seem to be from large specimens of B. altithorax of about 50 tonnes (Paul 1988). ‘ Supersaurus' may be a Diplodocus species (Paul 1988). Its scapulocoracoid is 2-7 m long, compared with 1-542 m for Diplodocus carnegiei. Hence if D. carnegiei had a mass of 15 tonnes (within the range of estimates given above) the mass of Supersaurus may have been 1 5 x (2700/ 1 542)3 = 80 tonnes. The huge femur of Antarctosaurus is 2-31 mm long, compared with 1-785 m for Apatosaurus louisae. If the latter had a body mass of 35 tonnes, geometric scaling suggests a mass for Antarctosaurus of 35 x (2310/1785)3 = 75 tonnes. However, the circumference of the femur is only 0-8 m, suggesting a more slender build and a lower mass (Paul 1988). It has been claimed that Seismosaurus may have had a mass of 100 tonnes (Gillette 1994), but the sparse remains (including no limb bones) seem inadequate to support the claim. These data suggest that the heaviest dinosaurs may have been between 50 and 80 tonnes. This is immensely heavier than the largest modern land animal, the African elephant ( Loxodonta africana: large males are around 5-5 tonnes; Laws 1966). Adult male masses for other very large land mammals include 2-2 tonnes for white rhinoceros (■ Ceratotherium simum ), 1-5 tonnes for hippopotamus ( Hippopotamus amphibius ) and 1-2 tonnes for giraffe ( Giraffa Camelopardalis ; Owen-Smith 1988). Although the large dinosaurs were sauropods, several other groups had members that were at least as heavy as any modern terrestrial animals. Mass estimates for herbivores include 5 tonnes for Iguanodon and 6 tonnes or more for Triceratops (Alexander 1989). The only terrestrial animals known to have approached the size of the large sauropods are a few gigantic mammals. The largest of these was probably Indricotherium , a hornless Oligocene rhinocerotoid which Economos (1981) estimated to have had a mass of 20 tonnes. Others, including myself (Alexander 1989) have suspected it of being even heavier, up to 34 tonnes. However, it now appears that the early restoration on which these mass estimates were based is misleading. A careful ALEXANDER: ALL-TIME GIANTS 1237 text-fig. 2. Some of the largest known terrestrial animals, drawn to scale. In the case of extinct animals (asterisked), only those known from reasonably complete skeletons are included. A, Giraffa\ B, Apatosaurus* ; c, Tyrannosaurus* ; D, Brachiosaurus* ; E, Indricotherium* ; and f, Loxodonta. analysis by Fortelius and Kappelman (1993) led to the conclusion that the bones that have been found come from specimens with an average mass of only 1 1 tonnes and that the largest specimens were probably little more than 15 tonnes. Two species of the related genus Paraceratherium were only a little smaller, and Fortelius and Kappelman argued that the largest complete mammoth ( Mammuthus ) skeleton may be from a 14 tonne animal. Other large extinct herbivores include pareiasaurs, dinocephalians and dicynodonts, but these were no larger than the largest modern mammals. The largest known terrestrial carnivores are much smaller than the sauropods. The best known is Tyrannosaurus rex , which was about 12 m long with a mass of about 7 tonnes (Alexander 1989; Farlow et al. 1995). Two other theropods, Giganotosaurus (Coria and Salgado 1995) and Car char odontosaurus (Sereno et al. 1996), may have been a little heavier. The largest rauisuchids (early archosaurs) attained lengths of 6 m (Benton 1997). Apart from these, and the theropods, there seem to have been no terrestrial carnivores of more than 1 tonne, at any time. The largest modern examples are polar bears ( Ursus maritimus; adult males are about 500 kg) and Siberian tigers ( Panthera tigris altaica, about 250 kg; Nowak 1991). The crocodilians have been discussed already, as aquatic carnivores. The question has often been asked, whether the largest dinosaurs could have supported their weight on land? The alternative would have been for them to have waded in water deep enough to have supported much of their weight by buoyancy. The question arises because for geometrically similar animals made of the same materials, weight increases as the cube of length, but bone and muscle cross sectional areas (and so strength) only in proportion to the square. Therefore, larger animals are expected to be less able to support their own weight. Evidence that the large sauropods could support their weight on land comes from several sources. First, morphological comparisons with terrestrial mammals such as rhinoceroses and elephants, and with the semiaquatic hippopotamus, favour terrestrial habits (Bakker 1971). Second, many sauropod footprints are more sharply defined than seems consistent with their having been formed under water (Thulborn 1990). Third, the dimensions of leg bones of large sauropods such as Apatosaurus indicate that they were amply strong enough to support the animals’ estimated weight. Alexander (1985) pointed out that bending moments due to components of force at right angles to 1238 PALAEONTOLOGY, VOLUME 41 the long axes of bones are more likely to set up dangerous stresses than are axial forces. With that in mind, I defined a ‘strength indicator’ which expressed the strength in bending of a leg bone (estimated from its dimensions) in relation to the load that the weight of the body would impose on it. If the bones of an extinct animal have strength indicators equal to those of homologous bones of a similarly proportioned modern one, they were strong enough to allow the extinct animal to move in dynamically similar fashion to the modern one. The legs of Apatosaurus are quite similar in the relative lengths of the bones to those of the African elephant Loxodonta , and homologous leg bones of the two species have very similar strength indicators. This implies that Apatosaurus had leg bones strong enough for it to have moved as athletically as elephants, which easily support their weight on land and indeed can run moderately fast, although they cannot jump. Hokkanen (1986) discussed how large a dinosaur could be, and concluded that even a sauropod of well over 100 tonnes could have legs strong enough to support itself on land. Thus sauropod size seems not to have been limited by problems of support. Another possibility we should consider is that dinosaur size was limited by the danger of overheating. Suppose first, as Bakker (1986) does, that the dinosaurs were endotherms with metabolic rates as estimated by extrapolation for mammals of their mass. We know that whales larger than any known dinosaur survive without overheating, even in the tropics where surface water temperatures may be as high as 27 °C. The effective temperatures of terrestrial habitats (averaged over day and night since we are considering very large animals which will heat and cool slowly) are probably seldom higher than this at the present day. In the Mesozoic, temperatures that we think of as tropical extended to higher latitudes than now, and equatorial temperatures seem to have been a few degrees higher (Hallam 1985). It seems necessary to explain what I mean by the effective temperature of a habitat. Different parts of the environment (air, ground, vegetation, sky) will be at different temperatures, and heat balance may also be affected by solar radiation. The ‘equivalent blackbody temperature’ (Campbell 1977) is the temperature at which a body that was not producing heat or evaporating water would reach equilibrium in the environment. By the effective environmental temperature I mean the equivalent blackbody temperature averaged over 24 h. The observation that whales can live in tropical seas suggests that the largest dinosaurs could have avoided overheating at similar effective environmental temperatures on land, even if their metabolic rates were as high as would be predicted for mammals of the same mass. In this argument I have not referred to the difference in heat loss rates in air and in water because, although small animals lose heat much faster in water, the difference is trivial for animals of more than 100 kg (Bell 1980). In another approach to the problem of overheating, Alexander (1989) considered the heat balance of a brachiosaur with mammal-like metabolism, estimating its rate of loss of heat by extrapolation from Bell’s (1980) data on cooling rates for smaller reptiles. I estimated that, unless it dissipated excess heat by evaporation of water, an endothermic brachiosaur would be at least 60 K warmer than its environment, which would be lethal except in extreme cold. Comparison with whales (as in the previous paragraph) suggests that this temperature difference has been overestimated, but even so we must doubt the viability of a brachiosaur with mammal-like metabolism, especially in warm Mesozoic climates, where the quantities of water that would have to evaporate to prevent overheating would be enormous. A more sophisticated analysis by Hokkanen (1989) led to a similar conclusion, that a Brachiosaurus with mammal-like metabolism would probably not be viable in a hot climate. Alexander (1989) also estimated body temperatures for ectothermic brachiosaurs, with metabolic rates as predicted for modern reptiles of equal mass. Unfortunately, my table 7.1 contained arithmetic inconsistency which has been pointed out to me by Dr Brian Bodenbender, to whom I am grateful. Also, my argument was simplistic: it should have taken account of the dependence of a reptile’s resting metabolic rate on body temperature. A corrected form of the argument follows. An animal with body temperature 7^,ody in an environment at temperature Tenv loses heat at a rate (7j,ody— Tenv)C/x, where C is the heat capacity of the body and r is the thermal time constant (the quantity given by Bell 1980, for many reptiles). This formula is explained by Alexander (1989). At ALEXANDER: ALL-TIME GIANTS 1239 equilibrium this heat loss is balanced by metabolic heat production at a rate R(m, ThoAy), that is at a rate that depends both on body mass and on body temperature. (^body- Teny) = R( ^body), Bennett and Dawson (1976) gave equations relating metabolic rate to body mass, for several groups of reptiles at several temperatures. I will use their equations for lizards, which cover the widest temperature range. These give metabolic rates for a 50 tonne brachiosaur of 770 W at a body temperature of 20 °C, 3270 W at 30 °C and 4840 W at 37 °C. These enable us to estimate the metabolic rate of a brachiosaur with reptile-like metabolism at any likely body temperature. The specific heat capacity of animal tissue is about 3500 J kg'1 K-1, so a 50 tonne brachiosaur would have a heat capacity C of 175 MJ K_1. We will assume a thermal time constant of 6 x 105 s (8 days). This is the shorter of the two estimates given by Alexander (1989; the other was 20 days), and is also shorter than an estimate of 12 days obtained by extrapolation from Loveridge’s (1984) data for crocodiles. The shortest estimate has been chosen as the least likely to predict overheating. Thus C/t will be taken to be 300 W/K and a temperature difference {ThoAy—Teny) of 10 K would be needed for equilibrium with a metabolic rate of 3000 W, the rate predicted for a body temperature of 29 °C. This tells us that with no evaporative cooling, a brachiosaur with a body temperature of 29 °C could be at equilibrium in an environment at 19 °C. Similarly, a brachiosaur with a body temperature of 38 °C could be at equilibrium in an environment at 23 °C. It seems unlikely that a brachiosaur with reptile-like metabolism could avoid overheating in hotter climates except by evaporative cooling. The latent heat of vaporization of water at 30-40 °C is 24 MJ kg-1, so the whole of the 4840 W produced by a brachiosaur at 37 °C could be dissipated by evaporation of 2 g of water per second, or 170 kg per day. This rate of loss seems entirely feasible; for example, a 3-7 tonne elephant lost 20 kg water per day by evaporation (Benedict 1936). Thus a brachiosaur with reptile-like metabolism could avoid overheating even in the hottest climates, provided it had an adequate water supply. Thus the size of large dinosaurs may have been limited by the danger of overheating if they had a mammal-like metabolism but not if they had a reptile-like metabolism. Dinosaur metabolic rates have been controversial since Bakker (1972) put the case for endothermy, but most of the points made have been inconclusive. Bakker’s most persuasive argument was that endothermic predators need bigger prey populations than ectothermic ones would do, to support their higher metabolic rates. He claimed to show that the ratio of predator to prey biomasses for dinosaur populations indicated endothermy, but Farlow (1976) showed that the evidence was equivocal. Weaver (1983) argued that Brachiosaurus could not have had mammal-like metabolism because, with a head of about the same size as that of a one tonne giraffe, it could not have eaten fast enough. If their metabolic rates are proportional to (body mass)0 75 (see above) a 50 tonne endothermic brachiosaur would need to eat 50° 75 = 19 times as much food as a 1 tonne giraffe with a similar-sized head. Barrick and Showers (1994) used the ratio of oxygen isotopes in Tyrannosaurus bone to argue that this dinosaur had a constant, uniform body temperature, like mammals (but see criticisms in Morell 1994 and Millard 1995). By contrast, Ruben et al. (1996) used computed axial tomography to show that the dinosaurs Nanotyrannus, Dromaeosaurus and Hypacrosaurus had no nasal turbinals. These structures are present in both birds and mammals, and serve as heat exchangers, cooling air as it is breathed out and condensing out much of its water vapour. Ruben et al. (1996) argued that, without nasal turbinals, endotherms with mammal-like metabolic rates would lose so much heat and water in their breath that endothermy was unlikely; dinosaurs were probably reptile-like in their metabolism. Whether the dinosaurs had mammal-like or reptile-like metabolic rates, Indricotherium was presumably mammal-like. For it, overheating may have been a serious problem. Another possibility is that dinosaur size was limited by the problem of maintaining a viable population (see Farlow 1993). Terrestrial habitats are more diverse and fragmented than the oceans, so world populations of terrestrial animals cannot be expected to comprise as many individuals as 1240 PALAEONTOLOGY, VOLUME 41 populations of ocean-living animals of equal body mass. Africa was supporting a population of 1-3 million elephants in 1979 (Nowak 1991). Population densities tend to be proportional to (body mass)-0 75 (Damuth 1981), with no clear difference between vertebrate ectotherms and endotherms (Peters 1983), so a continent capable of supporting 1-3 million 3 tonne elephants should be adequate to support 1 50000 50 tonne brachiosaurs, which would probably be enough for long-term (millions of years) viability. We have seen that the largest terrestrial carnivores were a great deal smaller than the largest herbivores. Similarly, modern carnivorous mammal species have lower population densities than similar-sized herbivores (Peters and Raelson 1984) and the largest carnivores are much smaller than the largest herbivores. FLYING ANIMALS (Text-fig. 3) Finally, I will review and discuss the largest flying animals. The Kori bustard ( Ardeotis kori ) seems to be the largest modern one, with masses of up to 16 kg (Maloiy et al. 1979). It takes off only with difficulty, and often runs instead of flying when approached. The largest albatrosses, vultures and swans all have masses around 10 kg and are much stronger fliers. The wandering albatross, Diomedea exulans , spends much of its time airborne, slope soaring over waves (Bevan et al. 1995). Vultures spend most of the day airborne, soaring either in thermals (Gyps species in Africa; Pennycuick 1972) or over the windward slopes of mountains ( Condor ; Pennycuick and Scholey 1984). By contrast, swans travel by flapping flight rather than soaring. The largest extinct birds, like the largest modern ones, were plainly flightless; their wings are rudimentary or even absent. The elephant bird Aepyornis stood 3 m tall, with an estimated mass of 450 kg (Amadon 1947). The largest known birds with well developed wing skeletons are the vulture- like teratorns (Campbell and Tonni 1983). The largest of these, Argentavis, is unfortunately known from only a few bones. Its mass has been estimated from the circumference of the tibiotarsus as 80 kg, five times the mass of the Kori bustard. This estimate is very imprecise, with 95 per cent confidence limits of 37 and 166 kg, but even the lower limit is far heavier than any modern flying bird. If the wing span was in the same proportion to humerus length as in condors, it was about 6 m, far greater than the 2-7 m span of the condor or the largest of all modern spans, the 3-4 m of the wandering albatross. All known pterosaurs had well developed wing skeletons and could presumably fly. Among them Pteranodon ingens is the largest known by a reasonably complete skeleton. Its wing span was 7 m, but it was remarkably lightly built, with an estimated mass of only 15 kg (Brower 1983). This mass was obtained by calculating the volume of the body and multiplying by 900 kg nY3, approximately the density of a plucked bird. A larger species, P. sternbergi, is estimated to have had a span of 9 m, a typical span for an ultralight aircraft (Frey and Martill 1996). Quetzalcoatlus northropi was even larger (Lawson 1975; Langston 1981). Only an incomplete wing skeleton has been found, but there is better material of smaller Quetzalcoatlus , either young specimens or a smaller species. The wing span of the large individual must have been about 12 m. If it were geometrically similar to Pteranodon (span 7 m) it would have been (12/7)3 times as heavy, about 75 kg. In fact, the wing skeleton was far from being geometrically similar to that of Pteranodon (the phalanges made up a smaller fraction of the span), so this estimate cannot be relied upon. Paul (1991) has estimated the mass of Quetzalcoatlus northropi as 250 kg. Arambourgiania (known only from a very few bones) may have had a slightly larger span than Quetzalcoatlus (Frey and Martill 1996). Now I will consider whether large animals can be expected to be able to generate the power needed for flight. A simple argument predicts that for geometrically similar aircraft, the power required for flight will be proportional to (mass)117 (see Rayner 1988), but the following argument predicts a lower exponent. Well-designed gliders of all sizes, from small gliders to large passenger- carrying craft, lose height at 0-5-FO m s 1 when gliding at optimum speed (Tucker and Parrott 1970). Thus they lose potential energy at rates proportional to their masses. This is the energy that keeps them airborne, so this observation suggests that the power required for flight is proportional ALEXANDER: ALL-TIME GIANTS 1241 Metres text-fig. 3. Some of the largest known flying animals, drawn to a uniform scale. Names of extinct animals are asterisked, a, Pteranodon ingens* ; B, Quetzalcoatlus northropi* ; c, Diomedea ; and d. Condor. to (mass)10. Whether power requirements increase in proportion to (mass)1 17 or to (mass)1 °, they increase faster than available metabolic power, which is expected to increase only in proportion to (mass)075. Thus large flying animals will have less power in reserve, and there must be an upper limit to the mass of flying animals. A glider sinking at 0-5 m s_1 is losing potential energy at a rate of 5 W kg-1 of body mass. To do work at this rate, muscles operating at the expected efficiency of about 25 per cent. (Astrand and Rodahl 1986) would have to use metabolic energy at a rate of 20 W kg 1 body mass. The maximum metabolic rates (calculated from oxygen consumption) of human endurance athletes are also about 20 W kg 1 (Astrand and Rodahl 1986), suggesting that a man-sized bird such as Argent avis might be just able to fly. Confirmation of this seems to be provided by the Gossamer Albatross , an ultra- light propeller-driven aircraft powered by a pedalling athlete which flew successfully across the English Channel in 1979 (MacCready 1995). Some animals are much better endurance athletes than humans; maximum metabolic rates of 40 W kg-1 have been recorded for 500 kg horses, and a remarkable 100 W kg 1 for the pronghorn antelope ( Antilocapra americana; mass about 32 kg; Lindstedt et al. 1991). Thus animals even larger than Argentavis and Quetzalcoatlus might well be able to produce enough power for flight. A flying bird (or pterosaur) probably needs some capacity for powered flight, but most very large birds (albatrosses, vultures, etc.) spend most of their airborne time soaring. The success of man- made gliders serves as evidence that craft much larger than Argentavis and Quetzalcoatlus can soar successfully, both in thermals and on the windward sides of slopes. There remains the question of whether such large animals could take off. Small birds can take off simply by jumping from the ground, hovering to keep themselves airborne, and then building up 1242 PALAEONTOLOGY, VOLUME 41 speed. Simple helicopter theory tells us that the power needed for hovering is much greater than for forward flight and (for geometrically similar craft) increases in proportion to (mass)1 17 (Alexander 1982). Therefore, large birds cannot hover, even to take off. They may take off by diving from a high perch, but to take off from level ground they often have to run like taxiing aircraft, as bustards and vultures do. Similarly, swans run over the surface of water to take off. The speed that a taxiing aircraft must reach, to take off, is the least speed at which the wings can provide enough lift to support it. It should correspond to the minimum gliding speed, which is between 5 and 10 m s-1 for various birds and a bat (Alexander 1982). Thus, animals that rely on running to take off may have to run moderately fast. The minimum speed is proportional to the square root of wing loading (that is, of body weight divided by wing area; Alexander 1982). It will generally be larger for larger animals because wing area is proportional only to (mass)0 67, in geometrically similar animals. Pteranodon is the largest flying animal for which wing loading, and so take-off speed, can be estimated with any confidence. Even in this case there is considerable uncertainty; the mass estimate may be inaccurate, and there has been controversy about the area of the wings. Estimates for a Pteranodon of 7 m span range from 2T to 4-6 m2 (Hazlehurst and Rayner 1992). Alexander (1994) argued on the basis of Unwin and Bakhurina’s (1994) interpretation of the shape of pterosaur wings that an intermediate value, perhaps 3-4 m2, was likely. If we accept this together with Brower’s (1983) mass of 15 kg, and assume a maximum lift coefficient of 1-5, Brower’s equation 2 gives a minimum speed of only 7 m s-1. It seems unlikely that Pteranodon could run as fast as this (it is about the speed of a men’s 1 500 m race), but if the wind were blowing at 7 m s-1 or faster (a moderate breeze) it could take off simply by facing into the wind and spreading its wings. This depends on its wings being remarkably large for its weight; its estimated wing loading of 43 N m-2 is much lower than those of the largest albatrosses and vultures (about 170 and 100 N m-2, respectively; Brower 1983). Quetzalcoatlus is estimated to have had 1-7 times the span of Pteranodon , so if it had the same aspect ratio its wing area was F72 times that of Pteranodon , and can be estimated as 10 m2. A 75 kg Quetzalcoatlus with this wing area would have had a wing loading of 74 N m“2, still a little lower than those of the largest vultures. That does not necessarily mean that it could have taken off as easily as a vulture; its enormous wings must have been difficult to manage, while it was still on the ground. If, however, it had the 250 kg mass estimated by Paul (1991), its wing loading would have been 245 N m-2, considerably higher than for albatrosses. Its minimum speed would then have been about 16 ms-1, in the speed range of galloping racehorses. Argent avis is estimated to have had double the span, four times the wing area and eight times the mass of a large vulture. This would give it twice the wing loading of a vulture and 2° 5 = 1-4 times the take-off speed. The problem of taking off may well have set the upper limit to the size of flying animals. CONCLUSIONS It is tempting to look for limits to the range of animal sizes and then to ask whether animals have ever reached them, and if not why not. That approach seems misguided for two reasons. First, all postulated limits depend on assumptions based one extant animals which may be false for extinct ones. For example, the metabolic rate of an unknown or extinct large animal may not be as predicted by allometric equations based on modern animals. Second, the evolution of larger animals will not necessarily occur whenever larger animals are possible; it will occur only when larger animals are favoured by natural selection. Very large animals may fail to evolve because their movements would be cumbersome, or because their activity would be constrained by the precautions they would have to take to avoid overheating, or for some other reason, even if they would be capable of life in the absence of competition. This paper does not show that larger animals than have evolved would be physically impossible, but it does suggest reasons why they might have been at a disadvantage. ALEXANDER: ALL-TIME GIANTS 1243 REFERENCES Alexander, R. McN. 1981. The chordcites. 2nd edition. Cambridge University Press, Cambridge, 510 pp. 1982. Locomotion of animals. Blackie, Glasgow, 163 pp. 1985. Mechanics of posture and gait of some large dinosaurs. Zoological Journal of the Linnean Society, 83, 1-25. 1989. Dynamics of dinosaurs and other extinct giants. Columbia University Press, New York, 167 pp. 1994. The flight of the pterosaur. Nature, 371, 12-13. jayes, a. s., maloiy, G. m. o. and wathuta, E. M. 1979. Allometry of the limb bones of mammals from shrews ( Sorex ) to elephant ( Loxodonta ). Journal of Zoology , 190, 155-192. amadon, D. 1947. An estimated weight of the largest known bird. Condor, 49, 159-164. anderson, J. F., hall-martin, a. and russell, d. A. 1985. Long bone circumference and weight in mammals, birds and dinosaurs. Journal of Zoology Series A, 207, 53-61. astrand, p.-o. and rodahl, k. 1986. Textbook of work physiology. 3rd edition. McGraw-Hill, New York, 756 pp. barker, R. T. 1971. Dinosaur physiology and the origin of mammals. Evolution, 25, 636-658. 1972. Anatomical and ecological evidence of endothermy in dinosaurs. Nature, 238, 81-85. 1986. The dinosaur heresies. Morrow, New York, 481 pp. barrick, r. E. and showers, w. J. 1994. Thermophysiology of Tyrannosaurus rex: evidence from oxygen isotopes. Science, 265, 222-224. bell, c. J. 1980. The scaling of the thermal inertia of lizards. Journal of Experimental Biology, 86, 79-85. benedict, F. G. 1936. The physiology of the elephant. Carnegie Institution, Washington, 302 pp. bennett, a. F. and dawson, w. r. 1976. Metabolism. 127-223. In gans, c. (ed. ). Biology of the Reptilia 5. Academic Press, London, 556 pp. benton, m. j. 1997. Vertebrate palaeontology. 2nd edition. Chapman and Hall, London, 452 pp. bevan, r. m., butler, p. j., woakes, a. j. and prince, p. a. 1995. The energy expenditure of free-ranging black- browed albatrosses. Philosophical Transactions of the Royal Society, Series B, 350, 119-131. brower, J. c. 1983. The aerodynamics of Pteranodon and Nyctosaurus, two large pterosaurs from the Upper Cretaceous of Kansas. Journal of Vertebrate Paleontology, 3, 84-124. calder, w. A. 1984. Size , function and life history. Harvard University Press, Cambridge, Ma., 431 pp. Campbell, G. s. 1977. An introduction to environmental biophysics. Springer, New York, 159 pp. Campbell, K. E. and tonni, E. p. 1983. Size and locomotion in teratoms (Aves : Teratornithidae). The Auk, 100, 390-403. Carey, F. G. 1982. Warm fish. 216-233. In taylor, c. r., johansen, k. and bolis, l. (eds). A companion to animal physiology. Cambridge University Press, Cambridge, 365 pp. Clarke, M. r. 1966. A review of the systematics and ecology of oceanic squids. Advances in Marine Biology, 4, 91-300. coria, r. a. and salgado, l. 1995. A new giant carnivorous dinosaur from the Cretaceous of Patagonia. Nature, 377, 224-226. cotgreave, p. 1993. The relationship between body size and population abundance in animals. Trends in Ecology and Evolution, 8, 244—248. damuth, J. 1981. Population density and body size in mammals. Nature, 290, 699-700. davis, D. d. 1962. Allometric relations in lions vs. domestic cats. Evolution, 16, 505-514. denton, E. j. 1974. On buoyancy and the lives of fossil and modern cephalopods. Proceedings of the Royal Society, Series B, 185, 273-299. economos, a. c. 1981. The largest land mammal. Journal of Theoretical Biology, 89, 211-215. 1983. Elastic and/or geometric similarity in mammalian design. Journal of Theoretical Biology, 103, 167-172. farlow, J. o. 1976. A consideration of the trophic dynamics of a late Cretaceous large-dinosaur community (Oldman Formation). Ecology, 57, 841-857. 1993. On the rareness of big, fierce animals: speculations about the body sizes, population densities, and geographic ranges of predatory mammals and large carnivorous dinosaurs. American Journal of Science, 293A, 167-199. smith, M. b. and robinson, J. M. 1995. Body mass, bone 'strength indicator’, and cursorial potential of Tyrannosaurus rex. Journal of Vertebrate Paleontology, 15, 713-725. fortelius, M. and kappelman, j. 1993. The largest land mammal ever imagined. Zoological Journal of the Linnean Society , 107, 85-101. 1244 PALAEONTOLOGY, VOLUME 41 frey, E. and martill, d. M. 1996. A reappraisal of Arambourgiania (Pterosauria, Pterodactyloidea) : one of the world’s largest flying animals. Neues Jahrbuch fur Geologie und Palaontologie, Abhandlungen, 199, 221-247. Gillette, D. D. 1994. Seismosaurus , the earth shaker. Columbia University Press, New York, 205 pp. hallam, a. 1985. A review of Mesozoic climates. Journal of the Geological Society, London, 142, 433 — 445. hazlehurst, G. A. and rayner, j. m. v. 1992. Flight characteristics of Triassic and Jurassic Pterosauria: an appraisal based on wing shape. Paleobiology, 18, 447—463. her wig, R. P., staley, j. T., nerini, m. K. and BRAHAM, h. w. 1984. Baleen whales: preliminary evidence for forestomach microbial fermentation. Applied and Environmental Microbiology, 47, 421-423. hokkanen, j. e. i. 1986. The size of the biggest land animal. Journal of Theoretical Biology, 118, 491^199. — 1989. Heat exchange in large animals. Unpublished Ph.D. thesis. University of Leeds. — 1990. Temperature regulation of marine mammals. Journal of Theoretical Biology, 145, 465 — 485. howland, H. c. 1974. Optimal strategies for predator avoidance: the relative importance of speed and manoeuvrability. Journal of Theoretical Biology, 47, 333-350. kosch, b. F. 1990. A revision of the skeletal reconstruction of Shonisaurus popularis (Reptilia: Ichthyosauria). Journal of Vertebrate Paleontology, 10, 512-514. Langston, w. 1981. Pterosaurs. Scientific American, 244 (2), 92-102. laws, R. M. 1966. Age criteria for the African elephant, Loxodonta africana. East African Wildlife Journal, 4, 1-37. lawson, d. a. 1975. Pterosaur from the latest Cretaceous of West Texas: discovery of the largest flying creature. Science, 187, 947-948. LINDSTEDT, S. L., HOKANSON, J. F., WELLS, D. J., SWAIN, S. D., HOPPELER, H. and NAVARRO, V. 1991. Running energetics in the pronghorn antelope. Nature, 353, 748-750. lockley, M. 1991 . Tracking dinosaurs: a new look at an ancient world. Cambridge University Press, Cambridge, 238 pp. lockyer, c. 1976. Body weights of some species of large whales. Journal du Conseil pour /’ Exploration de la Mer , 36, 259-273. loveridge, j. p. 1984. Thermoregulation in the Nile crocodile, Crocodylus niloticus. Symposia of the Zoological Society of London, 52, 443-467. MacCReady, p. b. 1995. Gossamer aircraft and where they lead. 239-245. In abbott, a. v. and wilson, d. g. (eds). Human-powered vehicles. Human Kinetics, Champaign, II., 279 pp. mace, G. m. and lande, r. 1991. Assessing extinction threats: toward a reevaluation of IUCN threatened species categories. Conservation Biology, 5, 148-156. maloiy, G. M. o., Alexander, r. McN., njau, R. and jayes, A. s. 1979. Allometry of the legs of running birds. Journal of Zoology, 187, 161-167. martill, d. m. 1988. Leedsichthys problematicus, a giant filter-feeding teleost from the Jurassic of England and France. Neues Jahrbuch fur Geologie und Palaontologie, Monatshefte, 1988, 670-680. Matthews, l. h. and Parker, H. w. 1950. Notes on the anatomy and biology of the basking shark ( Cetorhinus maximus (Gunner)). Proceedings of the Zoological Society of London, 120, 535-576. Matthews, p. 1995. The new Guinness book of records 1996. Guinness Publishing, London, 320 pp. millard, a. r. 1995. The body temperature of Tyrannosaurus rex. Science 267, 1666. morell, v. 1994. Warm-blooded dino debate blows hot and cold. Science, 265, 188. motluk, a. 1996. Crisis looms over whaling ban. New Scientist, 150 (2036), 6. nagy, K. A. 1987. Field metabolic rate and food requirement scaling in birds and mammals. Ecological Monographs, 57, 112-128. nowak, R. m. 1991. Walker s mammals of the world. 5th edition. Johns Hopkins University Press, Baltimore, 1629 pp. owen-smith, r. n. 1988. Megaherbivores : the influence of very large body size on ecology. Cambridge University Press, Cambridge, 369 pp. paladino, f. v., o’connor, m. p. and spotila, j. r. 1990. Metabolism of leatherback turtles, gigantothermy, and thermoregulation of dinosaurs. Nature, 344, 858-860. paul, G. s. 1988. The brachiosaur giants of the Morrison and Tendaguru with a description of a new subgenus, Giraffititan, and a comparison of the world’s largest dinosaurs. Hunteria, 2, 1-13. 1991. The many myths, some old, some new, of dinosaurology. Modern Geology, 16, 69-99. pennycuick, c. j. 1972. Soaring behaviour and performance of some East African soaring birds observed from a motor-glider. Ibis, 114, 178-218. 1975. On the running of the gnu ( Connochaetes taurinus) and other animals. Journal of Experimental Biology, 63, 775-799. ALEXANDER: ALL-TIME GIANTS 1245 1992. Newton rules biology. Oxford University Press, Oxford, 111 pp. and scholey, k. d. 1984. Flight behaviour of Andean condors Vultur gryphus and turkey vultures Cathartes aura around the Paracas Peninsula, Peru. Ibis, 126, 253-256. peters, R. H. 1983. The ecological implications of body size. Cambridge University Press, Cambridge, 329 pp. and raelson, j. v. 1984. Relations between individual size and mammalian population density. American Naturalist , 124, 498-517. randall, J. E. 1973. Size of the great white shark ( Carcharodon ). Science , 181, 169-170. rayner, j. m. v. 1988. Form and function in avian flight. Current Ornithology , 5, 1-66. romer, A. s. 1959. The vertebrate story. 4th edition. University of Chicago Press, Chicago, 437 pp. RUBEN, J., H1LLENIUS, W. J., GEIST, N. R., LEITCH, A., JONES, T. D., CURRIE, P. J., HORNER, J. R. and ESPE, G. 1996. The metabolic status of some late Cretaceous dinosaurs. Science, 273, 1204-1207. ryg, m., lydersen, c., knutsen, l. o., bjorge, a., smith, t. g. and oritsland, N. a. 1993. Scaling of insulation in whales and seals. Journal of Zoology, 230, 193-206. SERENO, P. C., DUTHEIL, D. B., IAROCHENE, M., LARSSON, H. C. E., LYON, G. H., MAGWENE, P. M., SIDOR, C. A., varicchio, d. j. and wilson, j. a. 1996. Predatory dinosaurs from the Sahara and Late Cretaceous faunal differentiation. Science, 272, 986-991. stahl, w. R. 1967. Scaling of respiratory variables in mammals. Journal of Applied Physiology, 22, 453^460. steel, R. 1989. Crocodiles. Helm, London, 198 pp. thulborn, t. 1990. Dinosaur tracks. Chapman and Hall, London, 410 pp. tucker, v. a. and Parrott, G. c. 1970. Aerodynamics of gliding flight in a falcon and other birds. Journal of Experimental Biology, 52, 345-368. unwin, d. m. and bakhurina, n. n. 1994. Sordes pilosus and the nature of the pterosaur flight apparatus. Nature, 371, 62-64. weaver, J. c. 1983. The improbable endotherm: the energetics of the sauropod dinosaur Brachiosaurus. Paleobiology , 9, 173-182. weibel, E. r. and taylor, c. R. 1981. Design of the mammalian respiratory system. Respiration Physiology, 44, 1-164. west, G. b., brown, j. h. and enquist, B. J. 1997. A general model for the origin of allometric scaling laws in biology. Science , 276, 122-126. wheeler, a. 1975. Fishes of the world: an illustrated dictionary. Ferndale, London, 366 pp. R. MCNEILL ALEXANDER School of Biology University of Leeds Leeds LS2 9JT, UK Typescript received 19 August 1996 Revised typescript received 1 October 1997 A PHYLOGENETIC TEST OF ACCELERATED TURNOVER IN NEOGENE CARIBBEAN BRAIN CORALS (SCLERACTINIA: FAVIIDAE) by KENNETH G. JOHNSON Abstract. Documenting patterns of long-term faunal change is an important application of palaeontological data, but questionable results may be obtained if the potential effects of sampling bias are not considered. Analysis of fossil Caribbean reef coral occurrences indicates significant species turnover during the late Neogene. The goal of this study is to test this pattern for a subset of the entire fauna by using phylogenetic information to identify problematical taxa and periods of poor sampling. A phylogeny for 40 species from the faviid genera Caulastraea, Colpophyllia, Diploria, Favia , Hadrophyllia, Manicina and Thysanus was inferred using 23 multistate characters. Although the relationships are homoplasious, some stable groups emerged. One group includes the Colpophyllia species, another includes Manicina , Hadrophyllia and Thysanus species. As currently defined, both Favia and Diploria are paraphyletic stem groups. The inferred evolutionary tree was used to estimate species richness and proportional origination and extinction rates. When ghost lineages are considered, the magnitude of species richness estimates increases resulting in lower estimates of proportional origination and extinction. However, the pattern of faunal change within the group remains largely unchanged, with increased origination during the Late Miocene followed by extinction during the Late Pliocene and early Pleistocene. Palaeontological data are used to document patterns of long-term faunal change. As such they provide the primary data to assess the role of potential causes and to develop predictions of the potential consequences of periods of rapid turnover in the history of life. However, bias in sampling can lead to inconclusive or misleading results. Two strategies have been applied to overcome artefacts due to incomplete or uneven sampling (Smith 1994). One approach is to compile occurrences of high level taxa such as genera or families, and use these data to estimate taxonomic ranges. The distribution of ranges through time can then be analysed using evolutionary metrics. By including large numbers of taxa in their analysis, proponents of this approach are able to detect large-scale patterns free of sample bias. Supraspecific level taxa are likely to be better sampled (Gilinsky 1991), but suffer from problems of definition and comparability (Eldredge and Cracraft 1980) because they usually are defined as arbitrary subdivisions of a particular lineage rather than as holophyletic groups. An alternative approach is to compare stratigraphical occurrence patterns at low taxonomic levels with phylogenetic information (Novacek and Norell 1982). Stratigraphical evidence has been widely used to test hypotheses of phylogeny. For example, the stratocladistic method proposed by Fisher (1991) explicitly includes stratigraphical information into the procedure for comparing competing hypotheses of relationship. Conversely, phylogeny can be used to assess the role of biased sampling in generating apparent patterns of faunal change (Benton 1994). Combining stratigraphical and phylogenetic data in an evolutionary tree will usually require interpretations of ‘ghost lineages’ and hypothetical range extensions (Norell 1992). Therefore, the estimates of rates of taxonomic evolution can change when phylogenetic information is included in an analysis of diversity patterns. Recent work has provided evidence for rapid change in the Caribbean reef coral fauna during the late Neogene (Budd et al. 1996). Accelerated species turnover has been documented using a [Palaeontology, Vol. 41, Part 6, 1998, pp. 1247-1268] © The Palaeontological Association 1248 PALAEONTOLOGY, VOLUME 41 comprehensive compilation of all known reef coral occurrences from over 70 Miocene to Recent localities (Budd et al. 1992). Study of the patterns of species first and last occurrence using this database indicates a significant turnover event in the Late Pliocene to Pleistocene, with extinction rates as high as 30 per cent, of the fauna per million years during the late Pliocene. Furthermore, study of extinction selectivity among ecological groups indicates that species with small, short-lived colonies which reproduce sexually have overall higher rates of origination and extinction throughout the Neogene (Johnson et al. 1995). The origination and extinction of these types of taxa is the main mode of faunal change in late Neogene Caribbean reef corals. The goal of this study is to test the hypothesis of accelerated Pilo-Pleistocene faunal change by including phylogenetic information in the analysis. There are over 1 70 species in the complete database, so developing a baseline phylogeny of the entire fauna is a prohibitively large task. Instead, a phylogeny has been inferred for the subset of taxa which experience the highest turnover. Most of these taxa come from the Faviidae, and have been classified into seven genera in which colonies form through primarily intratentacular division (Vaughan and Wells 1943). Three of these genera are restricted to the Caribbean region ( Manicina , Hadrophyllia and Thysanus ) whilst the others ( Caulastraea , Colpophyllia, Diploria and Favia ) have a broad geographical distribution including Mediterranean and Pacific occurrences. Several of the genera ( Manicina , Hadrophyllia , Favia and Thysanus) include species which tend to live as small, free-living colonies in off-reef and reef marginal environments. Previous attempts to reconstruct coral phylogeny using explicit cladistic techniques and skeletal characteristics have been hampered by the presence of excessive homoplasy and therefore poor resolution of species relationships (e.g. Budd and Coates 1992). But even homoplasious characters can be used to define groups (Pandolfi 1989), and analysis of skeletal characteristics on living material has been shown to be substantially in agreement with analyses of molecular and soft tissue characteristics (Potts et al. 1993; Budd et al. 1994). Furthermore, some workers (Veron 1995) have suggested that hybridization among disparate coral taxa is possible and has occurred repeatedly in the evolution of the Pacific coral fauna. If this is the case, then inferring phylogeny using parsimony is clearly an inappropriate approach for this group. Regardless, the cladistic approach applied to corals in this work resulted in useful interpretable hypotheses of relationships among the study taxa. A cladistic analysis is used to reconstruct the phylogeny of meandroid faviid corals. The inferred phylogeny includes a high degree of homoplasy but groups are relatively stable. By combining the inferred phylogeny with the stratigraphical distribution of the study taxa, an evolutionary tree is constructed for the group with implications for evolution and biogeography. The tree suggests that the Caribbean fauna was largely isolated from the Mediterranean by the Miocene, and subsequent evolution was primarily moulded by extinction of Mediterranean lineages and radiation of Caribbean lineages. Adding ghost lineages and extinctions resulting from cladistic branching events to analysis of faunal turnover does not alter the general pattern of increased richness and turnover in the late Neogene. Obviously, including ghost lineages will increase richness estimates, and, in general, decrease estimates of per taxon rates of origination and extinction. Proportional rates of species origination remain constant throughout much of the late Paleogene and Neogene, but decrease during the Pilo-Pleistocene. This pattern is similar whether or not ghost lineages and range extensions are included. Similarly, the pattern of variation in species extinction is not changed by considering phylogenetic information. Although estimates of the rate of both background origination and extinction are lower when range extensions and ghost lineages are included, estimates of the magnitude of the late Neogene period of accelerated extinction are comparable. PHYLOGENY Taxa A total of 40 species from the Faviidae has been included in the present analysis (Appendix). Taxa with Neogene and Recent distributions were taken from a comprehensive compilation of Caribbean JOHNSON: CARIBBEAN BRAIN CORALS 1249 table 1. List of characters included in the analysis including the type of character (discrete or continuous), a priori ordering of character states (ordered or unordered), and the number of character states. Measures of character fit (consistency index (Cl) retention index (RI) and rescaled consistency index (RC) on the consensus cladogram are also listed. These are in general minimum estimates across the range of all 72 equally parsimonious trees. Amb. = ambiguous. Character Type Order States Amb. Cl RI RC 1. Attachment of skeleton D U 2 Y 0-25 0-70 017 2. Meandroid series sinuosity D u 3 Y 0-40 0-77 0-31 3. Frequency of wall development C o 4 N 0-50 0-92 0-46 4. Symmetry of bud geometry D u 3 Y 0-40 0-50 0-20 5. Calicular platform shape D u 2 Y 013 0-50 0-06 6. Calice relief C 0 4 N 0-21 0-54 012 7. Calice or valley width C o 4 N 0-25 0-74 019 8. Epitheca D 0 3 N 0-20 0-67 013 9. Relative costa thickness D u 2 Y 0-20 0-43 009 10. Coenosteum C u 5 Y 0-31 0-59 018 11. Exothecal dissepiments D o 2 N 0-33 0-60 0-20 12. Costa continuity D u 2 N 013 0-50 006 13. Complete septal cycles C o 4 N 0-17 0-35 006 14. Septal spacing C o 3 N 013 0-28 004 15. Septal thickness D u 2 Y 010 0-40 004 16. Columella width C o 3 Y 0-20 0-69 014 17. Columella continuity D u 2 N 100 100 100 18. Septal lobes D u 2 N 0-50 0-83 0-42 19. Paliform lobes D u 2 N Oil 0-56 006 20. Endothecal dissepiments C o 3 Y 0-15 0-58 0-09 21. Wall structure D u 2 N 0-50 0-94 0-47 22. Double or single paratheca D u 2 N 0-33 0-75 0-25 23. Maximum colony size C u 3 N 0-25 0-70 018 Neogene coral occurrences (Budd et al. 1994). Several new taxa from the upper Neogene of the Dominican Republic (Budd and Johnson 1998) have also been included. Paleogene taxa were obtained from lists included in a review of the Eocene Caribbean faunas (Budd et al. 1992). Eocene species from Jamaica (Wells 1935; Zans et al. 1962), and Oligocene material from Antigua and Puerto Rico (Vaughan 1919; Frost and Weiss 1979; Frost et al. 1983) have been taken from published lists and new collections. The Paleogene and early Miocene fauna from Chiapas, Mexico (Frost and Langenheim 1974) was also considered. All extant Caribbean species from the seven genera as well as endemic species from the distinctive Brazilian fauna (Verrill 1901) were included. However, congeners from the Pacific and Mediterranean faunas have not been included. Although the exact biogeographical relations between the Mediterranean, Caribbean and Pacific biotas are not well understood, none of the included taxa has been described from outside the Caribbean Basin. However, Frost (1977) briefly compared the Mediterranean and Caribbean Oligocene faunas and suggested that some taxa might be synonymous, but he did not complete a full revision of the faunas. Species classified into Favia, Diploria , and possibly Colpophyllia , have been described from the Mediterranean. Characters Skeletal morphology was characterized using 23 characters with a total of 64 discrete character states (Table 1) scored from type material when possible. When type material was not available, characters were scored from published descriptions and examination of material in museum 1 Septal Spacing (per 5 mm) 4 8 12 16 PALAEONTOLOGY, VOLUME 41 1250 PALAEONTOLOGY, VOLUME 41 text-fig. 1. Distribution of septal spacing among the study taxa. Cut-offs to create discrete character states from this semi-quantitative measure were selected by visual examination of the distribution. Gaps are suggested at six and 12 septa per 5 mm resulting in three groups of species. collections. Many of the characters are summaries of continuous variation in colony or corallite form which may not fall into non-overlapping (discrete) character states. Although some workers have criticized the use of characters with overlapping variation in phylogeny reconstruction (Chappill 1989), coral morphology is notoriously poor in features which are expressed as a few clearly non-overlapping states, so relationships among the study taxa are unlikely to be resolved if these attributes are not included. However, previous studies which used both overlapping and non- overlapping characters suggest that overlapping characters are more likely to be homoplasious (Stevens 1991). A modified version of simple gap coding (Archie 1985) was used to subdivide continuous character distributions into discrete character states. The data for character coding are derived from a combination of measurement of type and accessory material and published species descriptions. Approximate ranges for each measured character were ranked by their midpoints, and plots examined for gaps in the character frequency distribution (Text-fig. 1). Where gaps were not evident, character state boundaries were defined at all levels where few taxa possessed ranges of variation which crossed the boundary between character states. Choosing the number of character states is a compromise between maximizing information content and maintaining consistency between characters (Archie 1985). Increasing the number of character states in a particular character increases that character’s potential to resolve more groups but simultaneously increases the likelihood of homoplasy as random error in character scoring may obscure any phylogenetic signal. The division of a continuous character into discrete states remains an arbitrary act, but greater division of a character will not result in conflicting hypotheses of relationships (Thiele 1993). JOHNSON: CARIBBEAN BRAIN CORALS 1251 Coding a larger range of character states will increase tree resolution, but this resolution is likely to be unstable. Each character was assigned equal weight in the analysis regardless of its range. Determining the range of a character was an important aspect of character selection and scoring, and therefore already involved many a priori assumptions regarding character weighting. Adding additional assumptions to the analysis will only decrease the parsimony of the resulting hypothesis, and cannot add any additional information into the analysis (Farris 1990). Characters have been ordered where there is a clear set of steps between the states, but other characters were left unordered. Characters were not explicitly polarized prior to the selection of the most parsimonious tree. Instead, an outgroup was included in the analysis and the shortest unrooted trees including both an ingroup and an outgroup were subsequently rooted at an internal node with a basal polytomy (Maddison et al. 1984). Character states Analysis of character states is arguably the most important component of any phylogenetic analysis, so considerable space is devoted here to discussion of how coral morphology was reduced to a set of characters with discrete character states. A complete list of character states scored for each taxon is included in the Appendix. 1. Attachment of skeleton. Almost all reef-corals live permanently attached to a hard substrate, a few taxa are free-living during most of their lives. As in all scleractinian corals, a pelagic (or motile benthic) larval stage settles on a hard substrate prior to skeleton development. However, two strategies exist which allow free-living species to avoid permanent attachment. In some taxa, the original attachment points are small pieces of rubble (especially skeletal plates from the calcareous green algae Halimeda sp.), and as the coral grows, the lower surface becomes larger than its attachment substrate and so becomes effectively free on the sea floor. In other cases, colony attachment points are not well developed, and the colony is broken loose either by physical or biological agents. In either case, the strategy allows populations to live in habitats with high sedimentation (Gill and Coates 1977). States : 0 = free-living; 1 = attached. 2. Meandroid series sinuosity. Meandroid series result from intramural budding which is not followed by the construction of walls between daughter polyps. In meandroid colonies, the orientation of budding and subsequent extension of polyps is expressed in the meander form of the colony. Sinuosity of the meander valley ranges from straight to sinuous, but no intrinsic order is obvious from the geometry of colony formation. Therefore, this character has been left unordered in the analysis. If the meandroid series is branching, this character refers to the nature of the valley between branching points. This character has not been scored for phaceloid, plocoid, or cerioid taxa because the budding history cannot be clearly assessed from the arrangement of corallites on the surface of a colony. For these taxa, this character has been scored as missing. States: 0 = mostly straight ; 1 = greatly curved ; 2 = sometimes sinuous. 3. Frequency of wall development. All taxa considered in this analysis utilize intramural fission to some degree during colony development. However, various colony forms may be constructed depending on whether walls are erected between sister polyps. Phaceloid, plocoid and cerioid forms result when walls are constructed after the formation of a new bud, whilst flabellate and meandroid colonies result when walls develop only occasionally. However, strictly meandroid or cerioid/ plocoid forms occupy the ends of a continuum of colony forms. The character is best coded by counting the number of continuous valley sections in a colony relative to the number of growth centres (stomodaea). Because soft tissue is not preserved in extinct taxa, this character was scored 1252 PALAEONTOLOGY, VOLUME 41 by estimating the relative lengths of continuous sections of meander valley. This is a more-or-less continuous character, so there is some scope for variability with taxa, especially considering that mechanical damage to colonies during life can divide continuous series and trigger the formation of new calical walls during recovery and overgrowth of the damaged region. However, the terminal states (no new walls compared with inevitable wall development) is generally invariable within species. States : 0 = walls always develop; 1 = walls develop in most (approximately two-thirds) new buds; 2 = walls develop in few (approximately one-third) new buds; 3 = walls never develop between new buds. 4. Symmetry of bud geometry. Scleractinian polyps are characterized by hexagonal symmetry, so new centres may develop in any one of six directions. However, in many meandroid forms, the direction of budding is geometrically constrained to one, two or three directions (Text-fig. 2). Meander valleys in taxa that are constrained to uni- or bi-directional growth are straight or sinuous single series resulting in a flabellate colony form. If tri-directional growth is possible, branching meander series can develop. Morphometric analysis of colonies of Manicina areolata suggests that polyps may be polymorphic with respect to budding direction. In M. areolata , stomodaea are invariably located over branching points in the meander series, but can also be positioned between branching points, and new centres may originate on the margins or interior to meandroid series. However, as new centres which develop internally are limited to bi-directional grow, new branch points are invariably added to at the ends of the meander series. This character was illustrated by Matthai (1926) who distinguished between stomodaea which form in linear series by repeated intra- tentacular budding on the distomodaeal mode and stomodaea formed by dichotomous branching or terminal forking. States : 0 = only uni-directional; 1 = only bi-directional; 2 = sometimes multi- directional. 5. Calicular platform shape. In the taxa considered here, septa are elevated above the columella and provide support for tentacle attachment. When the polyp is completely retracted into the ‘valley’, the soft tissue is protected by the septal plates. The margins of the septal may be gently inclined to nearly vertical. Variable preservation of the study material results in uncertainty regarding this character in some of the less abundant species. States: 0 = sloping or V-shaped; 1 — steep-sided or U-shaped. 6. Calice relief. This character describes the difference in elevation (relative to the upward growth direction) of the columella and the upper surface of the septa. It is a semi-quantitative character with numerical ranges defined by dividing the range along approximate discontinuities in the distributions measure material. However, in some cases, material was excessively worn or damaged, so measurements might be considered as minima, and the character state assignment might be questionable. Because of its inherent order, this character has been left ordered in the analysis. States: 0 = low (< 2 mm); 1 = medium (2-4 mm); 2 - high (4-10 mm); 3 = very high (> 10 mm). 7. Calice or valley width. This semi-quantitative character describes the wall-to-wall distance across a meandroid valley or corallite diameter in non-meandroid colonies. It is roughly equivalent to two times the major septal length plus the width of the columella, and is closely conserved within the meandroid species. Previous work on cerioid faviids ( Montastraea ) suggests that this character is perhaps the most useful diagnostic for recognizing reef-coral morphospecies within generic groups (Budd 1993). Assuming that the character states really fall along a continuum, this is included as an ordered character. States: 0 = small (< 5 mm); 1 = medium (5-10 mm); 2 = large (10-15 mm); 3 — very large (> 15 mm). 8. Epitheca. In the meandroid faviids, the epitheca is a distinctive non-trabecular thecal tissue deposited in a modified cavity on the perimeter of the skeletal secreting layer (Sorauf 1972; Stolarski JOHNSON: CARIBBEAN BRAIN CORALS 1253 text-fig. 2. Illustration of three modes of bud geometry in meandroid faviid corals. 1995) which is thought to provide a protective cover for exposed skeleton. Such a function would be crucial for free-living corals in reef-marginal environments to deter infestation of boring organisms. Environmental variation may be significant in this character, but the states are general enough to include intraspecific variation. In several cases, outer surfaces of a colony were not preserved, so this character is coded as missing. States : 0 = absent or very reduced; 1 = reduced; 2 = well-developed. 9. Relative costae thickness. The relative thickness of major and minor costae may be equal, but in some forms, minor (third and fourth order) septa and costae can be less than half as thick as major septa. This character was used by Duncan (1863, 1864) to distinguish various forms of Hadrophyllia , Thysanus and Manicina. States : 0 = equal; 1 = unequal. 10. Coenosteum. Coenosteum is skeleton deposited by coenosarc tissues. In meandroid forms, coenosteum develops between adjacent series and reflects the complex packing of the meander network. In colonies restricted to only uni-directional or bi-directional budding (character 4), adjacent corallites are always sister polyps, so the development of coenosteum is geometrically forbidden. Therefore, taxa with restricted budding geometry are scored as ‘absent’. Similarly, by definition, coenosteum is undeveloped in phaceloid colonies, and these taxa have been scored as ‘absent’. No attempt was made to distinguish between these two character states to avoid overweighting the distinction between colony forms included as other characters. Transitions among the character states are not restricted to a linear sequence (e.g. it is possible to proceed from an absent coenosteum to a wide coenosteum without intermediate steps), so this character was left unordered. Coenosteum is invariably present or absent, but its width can be related to the stage of formation of a new bud. Therefore, maximum coenosteum widths were used when coding the character. States: 0 = absent; 1 = present with adjacent walls; 2 = present and narrow (less than meandroid valley width); 3 = present with medium width (equal to meandroid valley width); 4 = present and wide (greater than valley width). 1 1 . Exothecal dissepiments. This character indicates the presence and relative abundance of tabular or vesicular horizontal structures extending between costal plates. Several taxa have been scored as ‘missing’ because of a shortage of well-preserved material in the current collection. States: 0 = absent; 1 = present. 12. Continuity of costae. A score for this character was determined by whether or not costae (or septa) are continuous between adjacent meander series. It is meaningless for flabellate or phaceloid colony forms and has been coded as ‘missing’ for several taxa. In meandroid forms, this character can in part reflect the relative sinuosity and proximity of the meander series to neighbouring series, 1254 PALAEONTOLOGY, VOLUME 41 and possession of confluent septa suggests greater colony integration among adjacent meander series (Coates and Oliver 1973). States'. 0 = discontinuous; 1 = continuous. 13. Number of septal cycles. The number of septal cycles has long been recognized as a significant character in corals. In forms with corallites formed by extramural budding, it can more easily scored by counting septa, but in meandroid forms the relative lengths and widths of septa must be examined. This character is related to septal spacing (character 14). States: 0 = three complete; 1 = more than three complete ; 2 = nearly four complete ; 3 = four or more complete. 14. Septal spacing. This character is scored using the number of septa per 5 mm along a meander series. It is related to septal width and the number of septal cycles. Because it is a relatively continuous character, discrete levels were assigned through visual inspection on a range of septal spacing measured on type material or taken from species descriptions (Text-fig. 1). Divisions were made between six and 12 septa per 5 mm reflecting discontinuities in the distribution at those points. Some taxa had overlap between adjacent character states. Manicina puntagordensis , Favia leptophylla and F. macdonaldi were scored as having fewer than six septal per 5 mm and Colpophyllia elegans was scored as having more than 12 septa per 5 mm. An alternative analysis with these four taxa scored between six and 12 septa per 5 mm did not change the hypothesized relationships among the taxa. States: septa per 5 mm: 0 = less than six; 1 = between six and 12; 2 = more than 12. 15. Equality of septal thickness. Major septa are generally longer than minor septa, and they may be thicker. In taxa with septothecal wall structures, this character should be structurally related to costal thickness (character 9), but often costae are equal and septa are unequal. States: 0 = equal; 1 = unequal. 16. Columella width. This character describes the width of the columella relative to the overall valley width. Absolute columella width is likely to be structurally correlated with overall valley width (character 7), so relative width was scored to avoid implicitly over- weighting corallite size. States: 0 = less than or equal to one-quarter valley width; 1 = one-third valley width; 2 = one-half width or wider. 17. Columella continuity. In meandroid colonies, the columella can be an continuous structure, with no easily recognizable corallite centres. But, in some taxa, corallite centres are clearly evident from the degree of septal inflection and by breaks in the columella. These breaks are often more clearly demonstrated in taxa with reduced or poorly developed columella and may be related to the process of budding. States: 0 = continuous; 1 = discontinuous. 18. Septal lobes. There is a great deal of confusion regarding septal and paliform lobes (character 19). As used here, septal lobes can only be found on lamellar septa composed primarily of a single fan system of simple trabeculae. Septal lobes are internal lobes formed by a second fan system. In contrast, paliform lobes are vertical extensions of septa formed by one or more trabecular bundles, and not generally composed of a second fan system. Some workers (Chevalier 1975; Veron et al. 1977) considered the presence of well-developed septal lobes to be significant, and based the definition of a new scleractinian family (Trachyphyllidae) largely on this character. However, this characteristic is widespread within the taxa considered here even though they are classified as members of the Faviidae. States: 0 = absent; 1 = present. 19. Paliform lobes. Paliform lobes are vertical extensions of the medial margins of septa formed by one or more trabeculae. They are not true pali because they do not form through the process of septal substitution (Wells 1956), in which the medial margin of a exoseptum bifurcates as it grows upwards and new septa are inserted into the calice. Paliform lobes are distinguished from septal JOHNSON: CARIBBEAN BRAIN CORALS 1255 lobes by forming from a single or multiple trabeculae which are not arranged as a fan system. In general, paliform lobes are vertical extensions of skeleton with margins that are free from the parent septum, and are usually associated with a thickening of the inner ends of septa. Although these structures are considered distinct from septal lobes (character 18), they may represent proto- septal lobes. However, they have been coded as distinct because no specimens examined in this study have both well-developed septal and paliform lobes. States'. 0 = absent; 1 = present. 20. Endothecal dissepiments. These structures are similar to exothecal dissepiments, but develop internally. States: 0 = absent or very few; 1 = intermediate; 2 = abundant. 21. Wall structure. Two distinct wall structures have been recognized in the study taxa. In all cases, septo-costal plates extend across the theca, but the margins of calices are defined by different structures. Septotheca is formed by septal skeleton and develops as a thickening and fusing of adjacent septo-costal plates. In contrast, thecal skeleton may not be not genetically related to the septa, in which case the theca is constructed by abundant and closely spaced dissepiments. This style of wall is termed paratheca (Wells 1956). States: 0 = septothecal; 1 = parathecal. 22. Double or single wall In some parathecate colonies, wall development between adjacent meander series appears to be co-ordinated resulting in a distinctive double wall structure. In these forms, the walls appear as clearly defined thin plates separated by a constant distance. Although this character applies only to parathecal forms, it was scored for all taxa. This will increase the relative weight of wall structures in the phylogenetic inference. States: 0 = single wall; 1 = double wall. 23. Size of colony. Although colony growth is indeterminate, some species of corals tend to have smaller maximum colony sizes than others. This is no doubt a reflection of the life history or environmental tolerences of taxa, with some forms that utilize clonal reproduction through fragmentation possessing large (although not necessarily connected) colonies. Other taxa are rarely found as large colonies, especially free living species which depend on some degree of mobility to survive in sediment-rich environments (Johnson 1992). Although maximum colony size is clearly a continuous character, no effort was made here to define size categories statistically. Categories were defined by roughly dividing the total range of size variation of all known Neogene Caribbean coral taxa into three groups (Budd et al. 1994). States: 0 = small (< 0T m); 1 = intermediate (0T-0-3 m); 2 = large (> 0-3 m). Phylogenetic inference Paup version 3.1.1 (Swofford 1993) was used to find the most parsimonious trees which describe the relationships among the taxa. Because of the relatively large number of taxa, a heuristic search was performed followed by total branch swapping of the set of all shortest unrooted trees. The maximum number of trees held in memory was 2000. The initial trees were found using random addition sequence with 100 iterations to help assure that the identified trees were close to global minima. Once the set of minimum trees was found, the tree was rooted and character state reconstructions were calculated relative to an outgroup consisting of Caulastraea portoricensis. Both Matthai (1928) and Wells (1956) present hypotheses of ‘morphogenetic trends’ in colonial corals with colonies formed by primarily extratentacular budding plesiomorphic to meandroid and flabellate colonies formed by exclusively intratentacular division. Caulastraea is the only genus in the Faviidae characterized by phaceloid colonies (Veron et al. 1977) and therefore most probably is part of a more pleisiomorphic lineage of the family than the other lineages included in this study. Extant species of Caulastraea are widely distributed across the Indo-Pacific region (Veron 1993) and have occurred since the Oligocene in the Caribbean, Indo-Pacific and Mediterranean regions (Chevalier 1961; Frost and Weiss 1979; Pfister 1980; Budd et at. 1994). Therefore, as a group. 1256 PALAEONTOLOGY, VOLUME 41 text-fig. 3. Strict consensus tree calculated from 78 trees with length 175. The relationships among species are ambiguous at three nodes. Apomorphies and support indices associated with branches are included as Table 2. JOHNSON: CARIBBEAN BRAIN CORALS 1257 Caulastraea species are among the most widespread of all faviid corals. Strict consensus trees were formed from multiple equally parsimonious trees, and characters were optimized on the consensus cladogram assuming accelerated character transformation. A cladistic permutation tail probability test (PTP) was used to assess the phylogenetic signal in the character matrix (Archie 1989; Faith and Cranston 1991). This test is a comparison of the length of the observed shortest tree with tree lengths for a set of 100 character matrices obtained by randomly reassigning character states for each taxon. The PTP test is designed to examine the amount of cladistic covariation in the data. The analysis was performed using code written by me for the automatic scripting of PAUP commands in NEXUS format. For each randomized data set, a heuristic search was used so the estimate of minimum tree length for each iteration is conservative. The PTP test suggests that significant phylogenetic signal exists in the character matrix, with the observed most parsimonious tree shorter than 99 replicate trees (P — 0-01). The initial heuristic search identified 78 equally parsimonious trees each with 175 steps. Five ambiguous nodes exist on a strict consensus of these trees (Text-fig 3); the relationships among some Favia species and the outgroup are not well resolved. Similarly, the relationships among Colpophyllia elegans, C. duncani and a group containing the other Colpophyllia species are not resolved. The relationships among Manicina mayori, M. puntagordensis , and M. species B and the relationships among the three Thysanus species are also not resolved. Some polytomies might be expected if several new lineages originate from another lineage which is not evolving new apomorphies, so further manipulation of the characters (e.g. reweighting) to increase resolution was not attempted. Two randomization tests were also applied to assess the support for hypotheses of group monophyly. The ‘evolutionary bootstrap’ works by finding maximally parsimonious trees for a series of pseudo-random replicates of a character matrix constructed by randomly resampling (with replacement) the vector of character states for each taxon (Felsenstein 1985). The proportion of these trees which include a particular monophyletic group is used as a measure of support for that group. There may be serious objections to this test (reviewed by Sanderson 1995), but it is widely used in molecular systematics where large numbers of characters are available. Bootstrap support estimates were obtained using the Random Cladistics program with 100 pseudoreplicates (Siddall 1995). Clade stability was also assessed using a modified jackknife procedure in which a series of cladograms were constructed for subsets of taxa with each taxon removed (Lanyon 1985). This is a way of examining the effects of individual taxa in the analysis. Although a different version of this test can be performed using the Random Cladistics package, this analysis was performed using a scripting program written by the author. Jackknife support values were calculated by determining group frequency from a total of 39 replicate trees constructed using heuristic searches. In each replicate, Caulastraea portoricensis was left in the analysis as the outgroup. If multiple shortest trees were found, a strict consensus tree was calculated for that replicate. The frequency distribution of all possible groups of taxa defined in the all-taxon consensus tree was then derived from the 39 replicate consensus trees by counting the number of trees in which each group was defined. The total number of iterations in which a group could possibly be found is equal to the number of iterations minus the number of taxa in the group, because if a taxon is not included in the analysis, it will not occur in the resulting tree. Jackknife percentages for each group were calculated by dividing the frequency of each group by the number of trees in which the group could possibly have occurred, the higher the percentage, the more stable the group to the effects of missing or ‘ problematical ’ taxa. Character state reconstruction on the consensus tree is ambiguous for nine characters (Table 1). The tree as a whole has low consistency (rescaled consistency index = 0T4; retention index = 0-78), but high homoplasy levels are in part related to the large number of taxa included in the analyses (Sanderson and Donoghue 1989). Homoplasious characters include septal and costal architecture, but characters associated with budding and the corallite wall and columellae provide more support for the consensus tree. Contrary to expectation, results of a Kruskal-Wallace rank sum test suggest that discrete characters are not more consistent with the consensus cladogram than continuous characters (x 2 = 0-53; 1 d.f. P = 0-47). 1258 PALAEONTOLOGY, VOLUME 41 table 2. Branch stability measures and apomorphies for branches. Branch numbers refer to Text-figure 4. Jackknife frequencies and are shown with the maximum number of replicates for each group indicated in parentheses. Character states were optimized on the cladogram assuming accelerated change, and ambiguous apomorphies indicated by asterisk. Branch Jackknife Apomorphies 1 — 6 (1-0), 8 (1-0), 10 (3-0), 11 (1-0), 13 (1-2), 16 (1-0), 19 (1-0), 23 (1-2) 2 0-69 (36) 10 (3-1)*, 12 (0-1), 16 (1-2)*, 20 (0-2) 3 0-76 (37) 19 (1-0), 21 (0-1) 4 — 6 (1-0), 10 (1-2)* 5 — 23 (1-0) 6 — 7 (1-2), 10 (1-4)*, 14 (1-0) 7 — 8 (1-0), 13 (1-0), 14 (1-0), 19 (1-0), 20 (0-2) 8 — 6 (1-0), 7 (1-0), 8 (1-2), 14 (1-0), 22 (0-1) 9 0-89 (33) 14 (1-2), 15 (1-0)*, 16 (1-2)* 10 — 13 (1-0) 11 0-97 (34) 7 (1-0), 10 (3-2), 23 (1-0) 12 — 4 (2-1) 13 0-97 (35) 13 (1-3), 15 (0-1)*, 19 (1-0) 14 0-53 (36) 6 (1-0), 9 (0-1) 15 0-51 (37) 8 (1-2) 16 — 13 (3-2), 14 (2-1) 17 — 3 (0-1), 16 (2-1) 18 — 1 (1-0), 10 (2-4), 20 (0-1) 19 — 10 (2-1), 12 (0-1) 20 0-64 (11) 3 (0-1), 10 (3-4), 15 (1-0), 16 (1-2)*, 20 (0-1)* 21 0-59 (37) 13 (1-0) 22 — 7 (1-0), 12 (0-1), 14 (1-0), 16 (2-1)* 23 — 19 (1-0) 24 0-53 (13) 3 (1-2) 25 — — 26 0-50 (14) 5 (1-0), 10 (4-1) 27 0-97 (37) 11 (1-0), 15 (0-1), 23 (1-2) 28 — 7 (1-0), 8 (1-0), 13 (1-2), 14 (1-2) 29 — 12 (0-1) 30 0-44 (16) 16 (2-1)*, 20 (1-2)* 31 0-97 (37) 6 (1-0), 7 (1-0), 23 (1-0) 32 — 8 (1-2), 2 (1-3), 14 (1-2), 15 (0-1) 33 — 10 (1-3), 12 (0-1) 34 — — 35 0-53 (19) 6 (1-2), 8 (1-0), 12 (0-1) 36 — 13 (1-0), 20 (2-0) 37 0-80 (20) 7 (1-2), 16 (1-0), 19 (1-0), 21 (0-1) 38 0-97 (32) 2 (1-2), 13 (1-2), 15 (0-1), 17 (0-1) 39 0-82 (34) 5 (0-1)*, 10 (1-2), 22 (0-1) 40 0-83 (35) 3 (2-1), 7 (2-3) 41 100 (36) 5 (1-0)*, 12 (1-0), 13 (2-1), 14 (1-0), 15 (1-0), 19 (0-1) 42 — 6 (2-3) 43 0-84 (37) 23 (1-2) 44 — — 45 — 3 d-2) 46 — 13 (2-3), 14 (1-2) 47 — — 48 — — 49 — 7 (2-1), 14 (1-2) JOHNSON: CARIBBEAN BRAIN CORALS 1259 TABLE 2. ( cont .) Branch Jackknife Apomorphies 50 0-96 (27) 1 (1-0), 3 (2-3), 4 (2-1), 10 (1-0), 23 (1-0) 51 — 7 (2-3), 14 (1-0) 52 0-50 (28) 5 (0-1), 6 (2-1) 53 0-58 (36) 2 (1-0)*, 4 (1-0)*, 6 (1-0)*, 11 (1-0), 19 (0-1), 20 (2-0) 54 — 9 (0-1), 13 (1-3), 14 (1-2), 15 (0-1) 55 — 2 (0-1)*, 5 (1-0), 6 (0-1)* 56 — 4 (0-1)*, 20 (0-1 )* 57 0-48 (31) 8 (0-2), 18 (0-1) 58 — — 59 0-44 (32) 9 (0-1)*, 15 (0-1) 60 0-54 (37) 13 (1-2) 61 — 2 (1-2), 5 (1-0), 6 (1-2), 7 (2-3), 8 (2-1) 62 — 13 (2-3), 14 (1-2), 18 (1-0), 19 (0-1) 63 0-85 (34) 2 (1-0), 4 (1-2), 10 (0-2), 16 (0-1) 64 — 12 (1-0), 20 (2-1) 65 0-54 (35) 6 (1-3), 9 (1-0)* 66 — — 67 0-92 (36) 1 (0-1)*, 5 (1-0)*, 7 (2-3), 22 (0-1), 23 (0-1) 68 — 1 (1-0)*, 6 (3-2), 14 (1-0) 69 — 5 (0-1)*, 9 (0-1) 70 — 15 (1-0), 16 (1-0) Groups The consensus tree suggests a distinct Favia subgroup, including Favia maodentrensis, F. favioides, F. fragum, F. gravida, and F. vokesae. This group is supported by three unambiguous apomorphies and can be found in a high proportion of the jackknife trees (Table 2). The shortest tree which does not include this group is two steps longer than the current hypothesis, and the three characters which support this group all have greater than median rescaled consistency indices. The group is characterized by decrease in both corallite and colony size and a narrowing of the coenosteum. A second group, including Colpophyllia, Hadrophyllia, Thysanus and Manicina species, is also well supported. Apomorphies include deeper calices, a reduced epitheca, and the development of confluent costae. However, jackknife support for this group is not as strong as for some of the other groups, and only one additional step is required for an hypothesis which does not include this clade. Within this large group, two main subgroups are defined, one including the Colpophyllia species and the other including Hadrophyllia, Thysanus and Manicina species. Jackknife frequency for the Colpophyllia clade is very high (0-97) and the shortest tree which does not include the group is two steps longer than the current hypothesis. The Colpophyllia clade is supported by four unambiguous character state changes including the development of sinuous meandroid series, the insertion of minor septa which are thinner than the major septa. Most importantly, the Colpophyllia clade is characterized by the development of a discontinuous columella. A smaller subgroup is stable within the Colpophyllia clade. This group includes the three Neogene species Colpophyllia natans, C. amaranthus and C. breviserialis, and is defined by a total of six character state changes, one of which is ambiguous. This is the most stable group in the current hypothesis, with support from all possible jackknife trees. This group is supported by a loss of fourth order septa, which results in a decrease in septal number accompanied by an increase in septal spacing, as well as the loss of septa with unequal thickness. Paliform lobes can be identified in all three Neogene Colpophyllia species. All of these characters are highly homoplasious with 1260 PALAEONTOLOGY, VOLUME 41 rescaled consistency indices less than 0-10, demonstrating that homoplasious characters can be used to substantiate stable groups. A group including species of Manic ina , Hadrophyllia and Thysanus (MHT) is supported by five apomorphies, including the development of a free-living mode of growth and generally smaller colonies and the cessation of wall development between newly budded polyps. New buds are constrained to a linear series with no branching or forking resulting in a flabellate growth form and loss of coenosteum. These characters are related to loss of attachment to the substrate. Previous work has shown that colony size in the extant free-living coral Manicina areolata is constrained by the ratio of tissue surface area to colony mass (Johnson 1992) so that large colonies experience high mortality rates due to reduced colony mobility. Adopting a free-living mode allowed these taxa to occupy sediment-rich reef marginal environments equivalent to shallow or deep seagrass beds and mangrove fringe systems. Stable subgroupings can be recognized within the MHT group. One node defines a group including the five meandroid Manicina species. This node is supported by four unambiguous apomorphies, and is characterized by the reacquisition of branched meandroid series with straight meander valleys between branch points accompanied by coenosteum development between adjacent branches. These characters are all more consistent with the hypothesis than average. However this group has relatively low jackknife support (0-85) and only one step is required for a tree which does not include the group. A second subgroup has high jackknife support (0-92) and includes three meandroid Manicina species with attached colonies. These taxa are also characterized by intermediate colony size and deeper calices with steeply sloping septal margins. EVOLUTIONARY TREE An evolutionary tree was constructed by superimposing cladistic relationships on to the stratigraphical range of species whilst minimizing hypothetical range extensions (Text-fig. 4). A single tree was selected from the set of most parsimonious trees by resolving ambiguous nodes using stratigraphical information (Smith 1994). The three ambiguous nodes in the more apomorphic part of the tree were treated first. For the ambiguous node in the Colpophyllia group, C. duncani was selected as the sister taxon to a group including C. elegans and the other Colpophyllia species. The alternative hypothesis places C. elegans as the pleisomorphic sister group to C. duncani and the other Colpophyllia species, and requires a range extension for C. elegans through the Mid Eocene. Similar reasoning was used to resolve the Thysanus and Manicina ambiguous nodes. A group including Thysanus corbicula and Thysanus excentricus is the hypothetical sister group of Thysanus navicula , and Manicina puntagordensis is identified as the pleisomorphic sister taxon to a group including M. mayori and M. aff. mayori. An ambiguous node involving the three Oligocene Diploria species was resolved by inferring a hypothetical monophyletic group including all three taxa. The alternative relationship identified two distinct stem groups, one including D. antiguensis and D. dumblei and the other including only D. portoricensis. The addition of stratigraphical information was not able to resolve the remaining ambiguous node which appears along the Favia stem groups. A single tree was selected from two remaining hypotheses which suggests that a group including Favia dominicensis, F. aff. dominicensis , and F. macdonaldi is a sister group to F. gregoryi. The tree includes several hypotheses of ancestry when no apomorphies occur along cladogram branches. For example, Colpophyllia breviserialis is identified as the ancestor of C. natans because no autapomorphies are hypothesized for C. breviserialis. Branching events are drawn at or below boundaries for convenience; they are assumed to have occurred sometime within the time interval after the boundary. Last occurrences of taxa which are drawn at boundaries also reflect imprecise age assignment, and actual extinctions are assumed to have occurred in the time interval prior to the boundary. Major range extensions are required for the more pleisomorphic taxa, and multiple origins of Favia and Diploria lineages are suggested. Favia is widely dispersed in both time and space. The genus has been described from the Cretaceous of Europe and the Caribbean (Vaughan and Wells JOHNSON: CARIBBEAN BRAIN CORALS 1261 Ma 10 30 50 Manicina mayori Manicina aff. mayori Manicina puntagordensis Manicina aff. areolata Manicina areolata Manicina geisteri Manicina jungi Manicina grandis Thysanus navicula Thysanus corbicula Thysanus excentricus Hadrophyllia saundersi Colpophyllia natans Colpophyllia breviserialis Colpophyllia amaranthus Colpophyllia willoughbiensis Colpophyllia mexicanum Colpophyllia elegans Colpophyllia duncani Diploria sarasotana Diploria portoricensis Diploria dumblei Diploria antiguensis Diploria strigosa Diploria clivosa Diploria labyrinthiformis Diploria bowersi Diploria zambensis Fa via gravida Fa via fragum Favia maoadentrensis Favia vokesae Favia favioides Favia weisbordi Favia leptophylla Favia aff. dominicensis Favia dominicensis Favia macdonaldi Favia gregoryi Caulastraea portoricensis text-fig. 4. Evolutionary tree created by superimposing the selected cladogram onto stratigraphical ranges of the study taxa whilst minimizing hypothesized range extensions. Asterisks indicate extant species. 1262 PALAEONTOLOGY, VOLUME 41 1943), and has developed a pan-tropical distribution since that time (Pfister 1980; Budd et al. 1992; Budd et al. 1994). The biogeographical origins of the distinctive coral fauna of north-eastern Brazil has never been demonstrated conclusively (Laborel 1967). However, this phylogeny suggests that the two endemic Favia species, F. leptophylla and F. gravida , are not closely related. Their most common ancestor is likely to be an unknown Paleogene species. Therefore, the biogeographical origins of this fauna is complex, including repeated migration of coral species from into and out of Caribbean and Brazilian reef communities. As currently recognized, Diploria is also a paraphyletic group. Frost (1977) suggested close similarities among several Mediterranean and Caribbean species of Diploria , and species described from the Oligocene of Europe (Vaughan and Wells 1943; Pfister 1980) have been placed in the genus. However, the European forms have discontinuous columellar structures, and may represent a different lineage than the Caribbean forms (Chevalier 1961; Budd and Johnson 1998). Neogene Diploria species are restricted to the Caribbean region, but it is unlikely that the Miocene and later Caribbean Diploria species were directly derived from the Oligocene species. In contrast, there is strong evidence of monophyly for both Colpophyllia and the MHT group. Colpophyllia was abundant in Europe during the Oligocene (Pfister 1980) and Miocene (Chevalier 1961), but became restricted to the Caribbean in the Neogene. The tree suggests that the Neogene Caribbean lineage originated from within the Paleogene lineage and has remained isolated from the Mediterranean fauna. No species of Manicina, Hadrophyllia or Thysanus has been described from outside the Caribbean, so a hypothesis of monophyly for this group is supported by the distribution data. LATE NEOGENE TURNOVER The evolutionary tree was used to compare evolutionary rates both with and without phylogenetic information. Phylogeny has two main effects on the distribution of taxa through time. First, the timing of species origination may be extended below the first occurrence of the species in the record, resulting in an hypothetical range extension. These range extensions will only alter estimates of origination rates and total species richness in earlier time intervals; they can have no effect on estimates of extinction rates. A phylogeny can also suggest the presence of undiscovered ancestral taxa termed ‘ghost lineages’ (Norell 1992). These are lineages predicted by tree topology. Since the age of both first and last occurrence of ghost lineages can be estimated on the evolutionary tree, including ghost lineages can alter estimates of species richness, origination and extinction through time. Taxonomic turnover was analysed using standard techniques (Gilinsky 1991). The study interval has been divided into nine time periods of roughly equal duration, and an estimate of total species richness was obtained by counting the number of lineages which occur within or both before and after each interval. Some conventions were adopted to estimate the number of first and last occurrences in each time interval, so that first occurrences which correspond to boundaries are attributed to the interval after the boundary, but last occurrences mapped on to a boundary are attributed to the interval prior to the boundary (Text-fig. 5). All ghost lineages are counted if they were supported by at least one apomorphy. If both sister lineages associated with a branching event are supported by an apomorphy, then the branching event is assumed to be associated with the extinction of the parent lineage. If one of the sister lineages is not supported by an apomorphy, then it is assumed to be the parent lineage. Therefore, most branching events result in two first occurrences and one last occurrence. Proportional rates are used to estimate the magnitude and timing of taxonomic turnover. These are calculated as the number of first and last occurrences divided by the taxonomic richness within each time interval. Under a wide range of extinction models, these estimates of true branching and extinction rates are likely to be biased by differences in interval duration, but no individual metric has been proposed that provides unbiased estimates under a range of typical extinction models (Foote 1994). As expected, the addition of ghost lineages increased estimated richness (Text-fig. 6a), with the JOHNSON: CARIBBEAN BRAIN CORALS 1263 text-fig. 5. Rules for counting species richness and the number of first and last occurrences with and without hypothetical ancestors. In each case, vertical bars represent species ranges and horizontal lines are boundaries of stratigraphical intervals. Parts b and c include hypotheses of relationships and ghost lineages, a, when first occurrences are associated with interval boundaries, the species is assumed to have originated after the boundary, but last occurrences at boundaries are attributed to the prior time interval, b, hypothetical ancestors are assumed to become extinction during cladistic branching, c, all hypothetical ancestors are considered when multiple taxa appear to arise simultaneously. long branches in the Favia and Diploria taxa increasing apparent richness during the Paleogene and Early Miocene. The general pattern remains unchanged with richness reaching a peak in the Late Miocene and Plio-Pleistocene faunas, but the highest total richness estimated including ghost lineages occurred during the Late Miocene with a decrease in richness during the Pliocene and Pleistocene. The number of first appearances (Text-fig. 6b) is also increased when phylogenetic predictions are included, but the overall pattern of high numbers of first appearances in the late Miocene is similar with and without ghost lineages. This increase might be caused by the increase in species richness during the Late Miocene. Proportional origination rates estimated including phylogeny suggest high origination throughout the Miocene. Patterns of species extinction through time are not changed by including phylogenetic information in the analysis. The estimates of pre- Pliocene background extinction rates are lower due to increased richness estimates when ghost lineages are counted, but neither the timing nor the magnitude of the Plio-Pleistocene extinction event is significantly altered. Including the ghost lineages has increased the relative difference between the Plio-Pleistocene time of species extinction relative to background extinction. For this group of reef-corals, there was radiation throughout the Miocene resulting in an increased number of species until the late Miocene. However, during the Plio-Pleistocene, most of the taxa suffered extinction. Although the addition of phylogeny did not cause substantial change to the results of the analysis of taxonomic turnover, it does allow identification of periods with poor sampling, especially for the Favia and Diploria species during the Oligocene and early Miocene. Examination of the evolutionary tree also facilitates identification of potentially problematical occurrences of particular taxa. For example, the first appearance of Thysanus excentricus in the lower Miocene results in considerable range extension for the MHT lineage. As the database is refined, this occurrence will be examined in detail to insure that both the age assignment and identification are correct. Therefore, although adding phylogenetic information to stratigraphical ranges can aid the detection of periods of poor sampling, the approach suffers from several potential sources of error. Most serious is the requirement for a stable phylogenetic hypotheses. This may not be possible for some problems, especially for large datasets or for groups without clearly defined discrete morphological characters. Alternate methods for detecting uneven sampling exist for such cases. For example, a sample completeness index can be calculated as the ratio of the number of taxa found in a particular interval to the number which occur both before and after the interval. A 1264 PALAEONTOLOGY, VOLUME 41 u o'