M \ Z CO ± CD ± CD X j SMITHSONIAN ~ INSTITUTION NOlinillSNI NVINOSHIIWS S3IHVH8n LIBRARIES SMITHSONIAN CD 2 ,v, CD z , ^ ^ >2SS»L^\ = — ./ * /* -r Aw#]teftv -i . , / ' z -h /Sjgmfft x 01 - '_ W- ? > S X > ^ ^ ^ ^ ^ Jg ^ ^ 1 NVINOSHilWS S3iavaan LIBRARIES SMITHSONIAN_ INSTITUTION NOlinillSNI NVINQSH1IWS — CD ^ CO -2 , 4/1 _ “ 2 V liJ ^ UJ / n ^ y*^\V'‘sV/Vi> q ^ l-j SMITHSOmAN^INSTITUTION_ NOlinillSNI JvIVINOSHlIINS ^Saiavaail^LIBRARI ES^SMlTHSONIAh o m - 2- 'v^jvasv^ m CD x CD — CD i“nvinoshiiins saiavaan libraries Smithsonian institution NoiimusNi nvinoshihm Z CD Z \ W Z iD Z 2 '\^F > 5 'KyV > S^SMITHSONIAN ^INSTITUTION W NO!inillSNI_ NVINOSHHINS^S 3 iaVaan\lBRARI£S SMITHSONIAr — CD z \ CD — CO _ Z ll) cd 1x1 cd ^ stfmw/. A — i X v'W , j| NVINOSHilWS^SS I ava a ll^ LIBRAR I ES^SMiTHSONIAN” INSTITUTION NOlinillSNI NVIN0SH1IIAI * S H ~ ■<*<£■*" /SSKSs. = /5Sfni^ 2 O -x “ 5 , _ L-' „ “* -tp —v • — > Q X^Y DL/^ N^ilivy Q S^ SMITHSONIAN^ INSTITUTION “ NOliniliSNI^NVINOSHillAIS^S 3 I aVH 3 IT ^Ll B RAR I ES^SMITHSONIA £ % ro i /#^\ I I /#^\ 1 § » ltp» . loP- -^ol > v jx (§^ > pf 33 CD ,07 33 cd m ^ X^va^z m '' Ajvasi^z m z CD — CD — CD — sii“nvinoshiiws S3 lava an libraries Smithsonian institution NoiimusNi nvinoshhiai 2 CD Z , CD Z CD Z 2 < Nj. 2 < A 2 .jagfo, < Xt£5q> \ _j Z V^_ v\ -c Z A. — ( Z m 2 w- - z ^ X £ — 1§ c HI .« £ life. . . v^l/ h 0 g ^ e 5 ON^NOlinillSNII^NVINOSHillAIS^Sa IHVHen^LIBRARI ES^SMITHSONIAN^INSTITUTION ^ NOIlflili! O “ \v ° X3E?^ “ ^ovS'v o A. “ ^Too^Sx 0 m '' \w "' v'VASi^x ro X'oos>*i>>' ro .*_ in V> £ ^ ^~-* £ CO — n”LIBRARIES SMITHSONIAN INSTITUTION NOlinillSNl NVINOSHillNS S3 I HVH 8 IT LIB RAR I — "• Z. * CO 2 *o “ 2 )^v >'* 5 w DN^NoiiniiisNs^NviNosHiiiAis^sa i ava a n2u brar i es^smithsonian institution Noiinm 2 ' x' _j 2 %'^T os N^feu^/ £ tr ^ 15s/ X WX/alVC/ “ \-lW! co 5 ^ST - 5 5?' ' 5 _j J - _ 2 -• n libraries Smithsonian jnstitution NoiinmsNi NViiMosHiiNS ssiavaan ^librar O “ XcSsooTX O ~ O Z m ns ov^ ^ m 'xcle^ « n»/ m '‘X, ^ m ^'---~^ Z " jjf £ to X £ co ON NOlinillSNrNVINOSHlIINS SBIHVHan LIBRARIES SMITHSONIAN INSTITUTION NOlinilJ ^ 2^ (/) 2 »*■• ■ ^ -'••*•* ^ (/> § z ^ ^ ^ 3» 2 X > 2 '’S, > ^ ^ ^ ■*• 21 2 1 Hi 2 LI B RAR I ES^SMITHSONIAN INSTITUTION NOIlflillSNI NVIN0SH1IINS S3 I H VB B 11 _ L * B R AR 1,1 yy — CO “ CO CO o XQq pC7 “ o — - — — — 2 ON ^ NOlinillSNI^NVINOSHlIINS SBIHVaBIl LIBRARIES SMITHSONIAN INSTITUTION NOIlfUIJ :r r~ .. Z r~ z “ m ■" \w ~ X!vasa>- m iy) ~ .a ^ ^ C/) — IT ~LI 8 RAR I ES SMITHSONIAN INSTITUTION NOIlflillSNI NVIN0SH1IINS S3 I HVa B IT LI B R AR 2 CO 2 CO 2 gi £ 5 \%v ^ =? 5 =5 z /mwms co ^ 0 A 25 cm Conglomerate with finer, tabular and subequant clasts Parted limestone Irregularly bedded limestone Ribbon limestone Nodular limestone Quartzose dolostone Shale: black (bk), brown (bn) green (gn), grey (gy). red (rd) ▼ Chert / Graptolitic interval text-fig. 2. Lithological logs through Cow Head and St Paul’s Inlet. ST PAUL'S INLET the Cow Head Group for the Tremadoc-Arenig boundary interval at indicating graptolitic horizons (after James and Stevens, 1986). Geology of the Cow Head Group The following account is based largely on James and Stevens (1986), from whom additional details may be obtained. The Cow Head Group consists of up to 500 m of shales, hemipelagic limestones. 4 PALAEONTOLOGY, VOLUME 34 carbonate grainstones and limestone conglomerates. The detrital carbonates were deposited through the action of lime turbidites and limestone debris flows derived in part from lithified or semi-lithified sediments of the shelf edge and upper slope to the north-west. Most, but not all, clasts from any particular conglomerate yield fossils from a limited stratigraphic range and are approximately coeval with fossils from the overlying shales and limestones. This permits an unusually high degree of correlation between typical shelf faunas and those inhabiting deeper, open ocean environments, although care needs to be exercised in recognizing reworked faunal assemblages. The most common fossils in the limestone clasts are trilobites (see Kindle and Whittington, 1958), brachiopods (see Ross and James 1987) and conodonts (see Pohler et at. 1987) of the North American conodont province. The interbedded shales and limestones yield graptolites, conodonts of the North Atlantic province, inarticulate brachiopods and occasional trilobites, together with rare examples of other invertebrates and possible fish remains. The graptolite record extends from the middle Cambrian to the early middle Ordovician (late Arenig); details of the post- Tremadoc, early Ordovician graptolites are given by Williams and Stevens (1987, 1988a). A fuller account of previous investigations into graptolites from the Cow Head Group is also provided in the latter work. Other recent publications on the invertebrate faunas include accounts of the conodonts straddling the Cambrian-Ordovician boundary (Barnes 1988) and the Tremadoc-Arenig boundary (Stouge and Bagnoli 1988), and notes on the Radiolaria (lams and Stevens, 1988; Stevens and lams 1988). A monographic study of the Cambrian trilobite fauna has recently been completed by Ludvigsen et ah (1989). Biostratigraphic correlation between the isolated sections through the Cow Head Group, based mainly on graptolites and trilobites, shows that several conglomerate horizons can be traced throughout the entire Cow Head area. Those conglomerates in the most north-easterly exposures are thickest and coarsest, and interpreted to have been deposited closest to source. The most proximal sections of Stearing Island and Lower Head are composed almost entirely of conglomerate with only narrow interbeds; one boulder at Lower Head is 200 m across (Kindle and Whittington, 1958). The distal sections such as that at Green Point are mainly shale and thin-bedded limestone; here the conglomerates are thin with only small clasts up to 20 cm diameter. The proximal sections in the Cow Head Group were originally upslope from the more distal, and this may have controlled the distribution of graptolites to some extent. A noticeable feature of the distal section is the development of red, bioturbated shales in the late Tremadoc and Arenig, although at Cow Head itself the only heavily oxidized sediments present are found in a 3-3 m thick greenish dolostone interval at the Tremadoc-Arenig boundary. The overall colour change in Arenig shale interbeds, from dominantly green and black proximally to almost entirely red in the more distal sections, suggests an oxygen minimum upslope, with increased ventilation in deeper waters as found at the present time (see James and Stevens 1986). The ubiquitous red strata across the Tremadoc-Arenig boundary must, however, reflect important changes in the structure of at least the western reaches of the Iapetus Ocean, and may be of global significance (Stevens in prep.). It is possible that a stratified Cambrian ocean with anoxic bottom waters changed into a mixed ocean with oxygenated bottom waters during the Tremadoc, perhaps as a result of the glacial event postulated by Fortey and Morris (1982). Such a change, particularly if it occurred during the Tremadoc-Arenig boundary interval, may well have influenced the course of graptolite evolution in a similar fashion to the extinction and radiation event during the late Ordovician (Barnes and Williams 1990). Stratigraphic nomenclature of the Cow Head Group The earliest workers who studied the rocks of western Newfoundland, namely Richards, Billings and Logan (in Logan 1863), correlated the Cow Head strata with similar rocks in Quebec, particularly with those at Levis. Logan (1863) placed the Cow Head Group in his Division P. Schuchert and Dunbar (1934) concluded that the Cow Head was in part a tectonic breccia of middle Ordovician age. Kindle and Whittington (1958), following the lead given by Johnson (1941), recognized that the strata represented a sequence of sediments with an orderly stratigraphy and that WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 5 the breccias were part of the package, rather than being later, tectonically-derived material. They used a sequence of numbered ‘Beds’ to describe the Cow Head section. These are not beds in the usual sedimentological sense, nor do they conform to formal lithostratigraphic convention as recognized internationally. Rather they are hybrid biostratigraphic/lithostratigraphic units defined on their fossil content and gross overall lithology. They are, however, convenient to use and were extended to all parts of the Cow Head Group by James and Stevens (1986). Strata yielding a late Tremadoc graptolite fauna occur in the upper part of Bed 8 (Text-figs 2 and 3), and are here assigned to the Aorograptus victoriae Zone. The base of this zone is as yet undefined, but the top is marked by the appearance of graptolites indicative of the early Arenig T. approximate Zone at the base of Bed 9 (Williams and Stevens 1988a). The more formal lithostratigraphic nomenclature proposed by James and Stevens (1986) shows correlation with faunal changes at the Tremadoc-Arenig boundary only in the more proximal sections (e.g. James and Stevens 1986, fig. 43), where it lies at the boundary between the underlying Stearing Island Member and overlying Factory Cove Member of the Shallow Bay Formation. WESTERN BROOK POND (NORTH) MARTIN POINT (NORTH) text-fig. 3. Lithological logs through the Cow Head Group for the Tremadoc-Arenig boundary interval at Western Brook Pond, Martin Point and Green Point, indicating graptolitic horizons (after James and Stevens, 1986). Correlation with other sequences Full discussion of correlation between graptolitic zonal schemes of the early Ordovician was given by Cooper (19796). Since that time, additional sections have been studied; we therefore include an 6 PALAEONTOLOGY, VOLUME 34 COW HEAD W. NFL D P. fruticosus T . akzharcnsis T. approximatus A. vlctoriae zones not yet defined VICTORIA AUSTRALIA g Be4 T. fruti. (3 stipe > Be3 T. fruti. (3 & 4) Be2 T. fruticosus (4) Bel T. fruticosus & T. approximatus La 3 T. approximatus La2 A. victoriae Lai. 5 Pslgraptus <& Cionograptus Lai D. scltulum & Anisograptus NEW ZEALAND T. fruticosus T. approximatus CANADIAN CORDILLERA J T. fruticosus T . approximatus no fauna recorded T. fruticosus T. approximatus C. flexills - A. vlctoriae Anlsograptus - Staurograptus CENTRAL GREAT BRITAIN AlAIAkvx S. pusllla D. f. flabelli- forme OSLO NORWAY Ddymograptus Beds Ceratopyge Beds Dictyonema Beds HUNNEBERG SWEDEN _ no fauna recorded T. approximatus T. phyllo- graptoides no fauna recorded hunj;ang CHINA no fauna recorded Adeiograptus - Cionograptus text-fig. 4. Correlation of the late Tremadoc-early Arenig graptolite zones of the Cow Head Group with other sequences (data based largely on: 1- Williams and Stevens 1988a and this paper; 2-VandenBerg 1981; 3 - Cooper 1979a; 4 - Lenz and Jackson 1986; 5 - Berry 1960; 6 - Stubblefield and Bulman 1929; 7 - Monsen 1925 ; 8 - Maletz and Erdtmann 1987; 9 -Wang and Erdtmann 1986). updated correlation chart to include a selection of these (Text-fig. 4), although some (e.g. Hunneberg and Oslo) are currently under investigation and detailed biostratigraphic discussion is not yet possible. For the purpose of the present paper, we merely reiterate the precise correlation possible in continuously graptolitic successions across the Tremadoc-Arenig boundary interval, where a rapidly evolving graptoloid fauna, as found in the A. victoriae Zone in the Cow Head Group, becomes extinct and is then replaced by the somewhat low-diversity but distinctive dichograptid fauna of the T. approximatus Zone (see Williams and Stevens 1988a). Several anisograptid genera, including Rhabdinopora , surprisingly seem to have been unaffected by this evolutionary event, maintaining their relative abundance from the late Tremadoc through to the T. akzharensis Zone. These taxa are, however, rare in the succeeding P. fruticosus and later Arenig zones, where the fauna is dominated by dichograptids and sigmagraptines. Justification for employing the chronostratigraphic series ‘Tremadoc’ and ‘Arenig’ is, however, more difficult owing to the incomplete nature of original British sections and the current state of flux regarding their definition (see Fortey 1988). Ongoing biostratigraphic studies within the Cow Head Group, particularly of the conodonts and trilobites (see Barnes et al. 1988; Stouge and Bagnoli 1988; Williams and Stevens 1988/6), are permitting precise correlation between the various schemes at this level. These support the assumption made by most previous graptolite workers (e.g. Bulman 1970; Cooper 19796) that the major faunal turnover documented at a level equivalent to the boundary between the A. victoriae and T. approximatus zones lie close to the traditionally accepted position of the Tremadoc-Arenig boundary. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 7 TAXONOMIC PROBLEMS ASSOCIATED WITH LATE TREMADOC GRAPTOLITES The earliest graptolites from the Cambrian were all benthic (see Rickards 1977); the first planktic graptolites evolved during the Cambrian-Ordovician boundary interval, including the ubiquitous and familiar group of Rhabdinopora flabelliformis . These Tremadoc anisograptids have traditionally been assigned to the Dendroidea, characterized by numerous, commonly irregular dichotomies, presence of bithecae, and a sclerotized stolon. Fortey and Cooper (1986) produced a revised high level, phylogenetic classification, in which they assigned all nematophorous (i.e. planktic and epiplanktic) graptolites to the Graptoloidea, restricting the Dendroidea to benthic genera. For the purpose of the present work, we accept this revised notion ; consequently all taxa described herein are considered to be Graptoloidea. Bulman (1970) referred all Tremadoc nematophorous graptolites to the Anisograptidea ; these were subsequently split into four subfamilies, the Adelograptinae, Anisograptinae, Staurograptinae, and Rhabdinoporinae, although Fortey and Cooper (1986, p. 683) doubted that these groupings served any useful, phylogenetically- related purpose. Further studies utilizing isolated material such as the present one are required before any additional revision of high-level classification is possible; until that time we follow Fortey and Cooper (1986) in using an undivided family Anisograptidae. Previous publications describing late Tremadoc graptolites have assigned them to both anisograptid and dichograptid taxa, for while some elements of the fauna (e.g. Rhabdinopora ) are clearly identical to earlier Tremadoc taxa, others appear more similar in overall rhabdosome form to ‘typically’ Arenig dichograptids, commonly having only two to four stipes and relatively simple thecal style. These species have been variably assigned to anisograptid genera such as Adelograptus and Kiaerograptus , and to dichograptid genera including Tetragraptus and Didvnwgraptus (e.g. Jackson 1974; Cooper and Stewart 1979). None of these previous studies had the opportunity of utilizing isolated, three-dimensional material to substantiate deductions made from flattened, non- isolated species. Most material described here from the Cow Flead Group is flattened, but three nodular limestone horizons (at Martin Point South, Green Point and St Paul’s Inlet) have yielded three-dimensional graptolites that can be isolated from the rock using acetic acid. These generally lack fine detail of periderm structure owing to the rather coarse, granular nature of preservation, probably related to partial breakdown of the organic material during subsequent burial and tectonic deformation. A few specimens do, however, reveal some ultrastructure, including cortical bandages overlying the fusellar increments (Text-fig. 5). Most studies of periderm ultrastructure have been made on graptolites of middle Ordovician and Silurian age, Rickards et al. (1982) recording no studies at all on material from the Tremadoc. Our material is thus important in providing a comparison of structures described from later taxa, although the main value of isolated specimens from the Cow Head Group is in permitting observation of proximal development. This has allowed us to make the following conclusions: 1. Whereas presence of bithecae associated with autothecae is variable, all taxa from this interval possess a sicular bitheca. A sicular bitheca has never been recorded from any Arenig dichograptid or sigmagraptine species, although it is apparently present in all earlier Tremadoc anisograptids (see Rickards 1975, 1977). 2. With the exception of the sicular bitheca, proximal development of several late Tremadoc species is almost indistinguishable from that of certain early Arenig dichograptid and sigmagraptine taxa described by Williams and Stevens (1988c/). 3. Although overall rhabdosome form of a few taxa are similar to Arenig dichograptids and sigmagraptines, irregular occurrence of delayed dichotomies commonly results in extra stipes of variable number. This contrasts with the regular, fixed nature of branching in the Arenig taxa. 4. Genera such as Rhabdinopora and Clonograptus which are found earlier in the Tremadoc and continue through into the Arenig have distinctive proximal developments unlike those of the majority of the fauna, and bithecae throughout the rhabdosome. We consider that the presence of a sicular bitheca and irregular dichotomous branching does not permit the assignment of any late Tremadoc graptolites to the Dichograptidae or Sigmagraptinae, PALAEONTOLOGY, VOLUME 34 text-fig. 5. SEM micrographs showing details of ultrastructure on isolated metasicula of Kiaerograptus bulmani (Thomas, 1973), GSC 87446, SPI43 (complete specimen figured PI. 3, fig. 12). a, cortical bandages, x 110. b, fusellar increments, x 110. unless the traditionally accepted views of these high-level classifications is significantly modified. All taxa described herein are therefore assigned to existing or new anisograptid genera. It does, however, seem likely that several late Tremadoc genera give rise to Arenig forms through loss of bithecae and the fixing of dichotomous branching, perhaps suggesting a polyphyletic origin for the dichograptids and sigmagraptines. This will be the subject of a future study incorporating both Tremadoc and Arenig material, and is outside the scope of the present paper. SYSTEMATIC PALAEONTOLOGY Descriptive nomenclature employed conforms to that of Bulman ( 1970), Cooper and Fortey (1982) and Williams and Stevens (1988a). Particular note should be made of the term Tutellum\ introduced by Williams and Stevens (1988a, p. 20) to describe the ‘lip’ or ‘spoon-shaped’ process found at the sicular aperture of many Ordovician graptolites on the side of th 1 1 (cf the virgella, which is a spine). The acetate overlay technique described by Williams and Stevens (1988a), p. 23) was employed to assist in distinguishing species and in comparing isolated specimens with flattened material. Line drawings were made whilst using a Wild M5A microscope with ‘camera lucida' attachment. Light photographs of isolated and non-isolated material were taken with a Wild M400 photomicroscope, using fibre-optic fight source and with slabs immersed in 95% ethanol. Scanning electron micrographs were taken using a Hitachi S570 with a 120 film back. All figured specimens are housed in the collections of the Geological Survey of Canada, Ottawa (GSC). Specimen localities and horizons in the systematic section are referred to in abbreviated form; collected sections (see Text-fig. I and Williams and Stevens 1988a) are the ‘Ledge’ on the Cow Head Peninsula (CHN), St Paul's Inlet, North Tickle (SPI), Western Brook Pond, north section (WBN), Martin Point, north and south sections (MPN and MPS), and Green Point (GP). Numbered intervals refer to those used in Text-figures 2 and 3. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 9 Order graptoloidea Family anisograptidae Bulman, I960 Genus kiaerograptus Spjeldnaes, 1963 Type species. Didymograptus kiaeri Monsen, 1925, pp. 172-175, pi. 2, figs 9, 10, 12-14, 16, pi. 4, figs 6-8. By original designation. Diagnosis (revised). Rhabdosome horizontal or declined, with two primary stipes, one or both of which may dichotomise at the second thecal pair to produce three or four stipes. Autothecae simple or with sigmoidal curvature, prothecal folds occasionally present. Sicula with bitheca; other bithecae present in early forms, apparently absent in later taxa. Remarks. The definition of Kiaerograptus was revised by Bulman (1970) and by Cooper and Stewart (1979); the description of K. quasimodo by Rushton (1981) and of taxa in the present study necessitate a broadening of the understanding to include rhabdosomes with more than two stipes as originally defined. Further revision might permit restriction of the genus to include only taxa with sigmoidally curved thecae and prothecal folds, as found in K. kiaeri . K. quasimodo. and the two new species erected in the present study (K. undulatus and K. magnus). This morphological feature is, however, often difficult to recognize in flattened material, and apparent folding of the dorsal margin is sometimes a post-mortem effect related to diagenetic flattening. All species included within Kiaerograptus from western Newfoundland have similar proximal development, but show some variation in thecal style; none, however, possesses any bithecae other than that of the sicula, and possibly at the dichotomies of first order stipes. Species from earlier in the Tremadoc, such as K. kiaeri and K. quasimodo have bithecae associated with autothecae throughout much of the rhabdosome, and a gradual reduction in bithecae would probably be documented if a continuous stratigraphic succession of taxa could be found. Kiaerograptus pritchardi (T. S. Hall, 1899) Plate 1, fig. 1; Text-fig. 6a-l 1899 Didymograptus pritchardi. n. sp.; T. S. Hall, p. 167, pi. 17, figs 7 and 9; pi. 19, figs 8 and 10. 1938 b Didymograptus pritchardi T. S. Hall; Harris and Thomas, pi. 1, fig. 13. I960 Didymograptus pritchardi T. S. Hall; Thomas, pi 1, fig. 14. 71962 Didymograptus tenuiramis sp. nov. ; Obut and Sobolevskaya, pp. 84—85, pi. 5, fig. 3. 1966 Didymograptus pritchardi T. S. Hall; Berry, pp. 429^430, pi. 45, fig. 1 ; pi. 46, fig. 1 ; pi. 47, figs 1 and 2. 1974 Didymograptus (?) stelcki n. sp. ; Jackson, pp. 52-53, pi. 5, figs 5 and 7; text-fig. la, b. non 1974 Kiaerograptus pritchardi (T. S. Hall); Jackson, p. 51, pi. 5, fig. 3; text-fig. 2a. c. d (= A.? fHiformis sp. nov.). 19796 Kiaerograptus cf. pritchardi (T. S. Hall); Cooper, fig. 5a. non 1982 Kiaerograptus pritchardi (T. S. Hall); Gutierrez-Marco, fig. 2/(= K. taylori). Type specimen (designated Berry 1966, p. 429). The lectotype is Nat. Mus. Victoria No. P14238, figured by T. S. Hall (1899, pi. 17, fig. 7), from La2 near Lancefield, Victoria, Australia. Diagnosis (revised, incorporating Berry’s redescription of type material). Rhabdosome with two (or occasionally more) long, slender, gently declined stipes, straight or dorsally convex, widening rapidly from 04-0-6 mm proximally to a maximum of 0 5-0-9 mm. Sicula inclined, 10-1-5 mm long, 0-25-0 3 mm wide at aperture. Autothecae number 9-9-5 in 10 mm, overlapping two-fifths to one half of their total length, inclined at about 10° to dorsal margin. Bithecae apparently absent except for sicular bitheca. 10 PALAEONTOLOGY. VOLUME 34 text-fig. 6. Kiaerograptus pritchardi (T. S. Hall. 1899), GP38, all x 5 except a ( = x 10). a and b, GSC 87381. c, GSC 87393. d, GSC 87390. e, GSC 87412. f, GSC 87383 (also figured PI. 1, fig. I). g, GSC 87368. H, GSC 87378. i, GSC 87411. j, GSC 87399. k, GSC 87406. l, GSC 87407. EXPLANATION OF PLATE I Fig. 1. Kiaerograptus pritchardi (T. S. Hall, 1899). GSC 87383, GP38, x 10 (also figured Text-fig. 6f). Figs 2-4. Kiaerograptus undulatus sp. nov. MPS42C, x 10. 2, GSC 87286. 3, GSC 87330 (also figured Text-fig. 8o). 4, GSC 87327. Figs 5-7. Kiaerograptus magnus sp. nov. GP38, x 10. 5, GSC 87388. 6, GSC 87331. 7, GSC 87361 (also figured Text-fig. 8n). Figs 8 and 9. Kiaerograptus bulmani (Thomas, 1963). MPS42C, x 5. 8, GSC 87329. 9, GSC 87317. Figs 10-16. Paratemnograptus isolatus gen. et sp. nov. 10, GSC 87375, GP38, x 10. 1 1, GSC 87315, MPS42C, x 10. 12, GSC 87287, CHN8.30, x 5. 13, GSC 873846, detail of distal branching, GP38, x 5. 14, GSC 87288, CHN8.30, x 2.5. 1 5, GSC 87385, GP38, x2-5. 16, GSC 87289, CHN8. 30, x 5. PLATE 1 WILLIAMS and STEVENS, Kiaerograptus , Paratemnograptus 12 PALAEONTOLOGY, VOLUME 34 Material and localities. Many flattened, non-isolated specimens from GP38; others from MPN17A, 17B. One possible poor isolated specimen from MPS42C. Description. The rhabdosome consists of two slender stipes occasionally reaching over 35 mm long; second order branching has not been observed in our material. The stipes are 0- 3-0-5 mm (commonly 0-4 mm) wide at t h 1 1 . increasing only slightly to a maximum 0-5 mm (cf. type material). Narrow widths are probably due to preservation in oblique orientation, the larger measurements probably being more representative of the true widths. The sicula is TO-1 T 5 mm long (cf. I -4 mm for type material), is usually inclined rather then perpendicular to the stipes, and has a gentle convex curvature with respect to the rutellar margin. The aperture shows a pronounced rutellum and is typically 0-25 mm wide. Thl 1 presumably buds from the prosicula, growing down along the rutellar margin for 0-75-0-85 mm before turning sharply out and growing slightly downwards for the remainder of its 0-6-0-8 mm length. The base of the rutellar margin of the sicula is left free for 0-15-0-4 mm (commonly 0-2 mm), whereas the ventral wall of thl1 subtends an angle of 60-80° with the sicular axis. The sicular bitheca has not been observed unequivocally, but by comparison with other taxa is almost certainly present. One specimen appearing to show a sicular bitheca reveals it to extend only slightly beyond the point where the ventral wall of thl1 diverges from the sicula, probably explaining its cryptic form. Thl2 buds from thl1 high on the reverse side, growing initially across the sicula in an almost horizontal direction before turning down to run along the antirutellar margin. It remains in contact with this margin until the sicular aperture is reached, at which point thl2 bends abruptly out, subtending an angle of 50-60° with the sicular axis. This angle is maintained for 0-6-0-9 mm until the aperture is reached. Remaining autothecae are almost straight, inclined at 10-15° with the dorsal stipe margin, but with a slightly concave ventral wall and gently flared aperture in most specimens. Occasionally, however, the ventral wall is straight; it is possible that the flaring is a preservational artefact related to differential lateral spread on flattening. Apertures are simple but deep, occupying one half to two-thirds of total stipe width. Thecal overlap represents a little under one half total thecal length, while thecal density is an almost constant 8-10 in 10 mm throughout the rhabdosome. Autothecal length appears to be related to size of rhabdosome, but it is unclear whether thecal growth is continuous throughout astogeny as demonstrated for several Arenig dichograptids by Williams and Stevens (1988a). Bithecae have not been observed apart from that of the sicula, and it is unlikely that they existed. Remarks. Erdtmann et al. (1987) referred K. pritchardi to their new genus Paradelograptus ; this genus is, however, characterized by slender thecae similar to Adelograptus and Kinnegraptus and lacks a sicular bitheca. K. pritchardi appears to be a well-defined, consistent species with little variation in rhabdosome form and dimensions, in contrast to most other coeval taxa. It is easily separated from these by its distinctive proximal region. The Newfoundland representatives of K. pritchardi have rather smaller dimensions than those recorded by Berry (1966) for the type specimens. Berry recorded proximal widths of 07-08 mm widening to a maximum 0-8-0-9 mm in his text descriptions, but measurements from his illustrations and Cooper’s (1979a, fig. 17/c) figure of the lectotype demonstrate proximal widths of 0-6 mm. As noted above, thecal length (and consequently stipe width) appears to have increased during growth of the rhabdosome. As the type specimens have much longer stipes than our specimens, stipe widths between the two populations are considered to be compatible, while thecal densities are identical. The length of the sicula does, however, appear to be consistently longer in the type material (1-4 mm) than in the Newfoundland specimens (1-0-1-15 mm). The single specimen of Didymograptus tenuiramis figured by Obut and Sobolevskaya (1962) is poorly preserved and seems to have suffered tectonic deformation. It does, however, appear very similar to K. pritchardi and is here tentatively referred to this species. Although Obut and Sobolevska (1962, fig. 4) refer the interval yielding D. tenuiramis to earliest Arenig, the associated assemblage could equally well be placed in the late Tremadoc as it contains ‘ Temnograptus ’ species and predates the first occurrence of T. approximatus. Specimens from the Yukon, northern Canada which Jackson (1974) referred to a new species Didymograptus (?) stelki , agree even more closely with the Australian types of K. pritchardi than those from Newfoundland, and we have no hesitation in assigning them to this species. Most of our specimens of K. pritchardi originate from Green Point, where the late Tremadoc WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 13 interval is represented by a succession of fine-grained, fissile shales deposited in a rather deeper, more distal environment than elsewhere in the Cow Head Group. This distribution may be related to original environmental restraints, or may be due to poor preservation in rather coarser lithologies elsewhere. A similar problem exists for slender Arenig graptoloids in the Cow Head group, Kinnegraptus and Adelograptus being largely restricted to the more distal, fine-grained facies, deposited in deeper water. text-fig. 7. Kiaerograptus cf. K. taylori (T. S. Hall, 1899), x 5. a d, GP38; a, GSC 87369; b, GSC 87386; c, GSC 87392; d, GSC 87410. e-g, MPS42C ; e, GSC 87356; F, GSC 87357; g, GSC 87358. Kiaerograptus cf. K. taylori (T. S. Hall, 1899) Text-fig. 7a-g cf. 1899 Didymograptus taylori , n. sp.; T. S. Hall, pp. 167-168, pi. 17, figs 1 1 and 12. cf. 1960 Didymograptus taylori T. S. Hall; Thomas, pi. I, fig. 15. Material and localities. Seven flattened, non-isolated specimens from GP38 and MPS42C. Description. The rhabdosome is composed of two stipes up to 25 mm long, with a deflexed or declined form and separated by an angle of 90-120°. They measure 0-6-0-8 mm wide at the first thecal aperture; the larger width is found in specimens with longer thecal lengths and higher inclinations to the dorsal stipe wall, but may be related to lateral spread in some instances. The stipes soon attain their maximum width of 10 mm, which is then maintained throughout the rhabdosome. The sicula is a consistent L5-L6 mm long, with an apertural width of 0-3-0-45 mm. It is initially straight, but has a convex curvature with respect to the rutellar margin over the distal 0-5 mm. A short nema is occasionally present; the sicular bitheca has not been seen, but is almost certainly present. Proximal 14 PALAEONTOLOGY, VOLUME 34 development has not been observed, but is probably similar to other late Tremadoc graptolites with a prosicular origin for th 1 1 . With the exception of the sicula, bithecae appear to be absent. Autothecae have a low initial inclination of about 10° to the dorsal margin, but this increases throughout their length to reach a maximum of 3(MK)° near the aperture. Thecal length is somewhat variable; overlap is about one half of total length in early thecae, reducing to about one third distally. Free ventral thecal margins are markedly concave, particularly towards the apertures which occupy one third to one half of total stipe width, giving a markedly denticulate appearance to the ventral stipe margin. Thecal density is somewhat variable 8-10 in 10 mm proximally, but reduces to a constant 8 in 10 mm distally. Remarks. Although the generic assessment of K. taylori has been discussed relatively recently (e.g. by Cooper and Stewart 1979, p. 790), no additional specimens appear to have been described since the original description by T. S. Hall in 1899. From his remarks (1899, p. 168) it seems that Hall possessed more than the one specimen illustrated; unfortunately there are several discrepancies between his written description, figure at natural size and the illustration recorded as x 3 magnification. Thomas (1960, fig. 15) has since provided a rather clearer figure of the specimen at natural size. Because of Hall's poor original description and lack of revisions using the type material, assignment of our Newfoundland specimens cannot be certain and we therefore refer them to K. cf. taylori. This species is unlike any other taxa from the late Tremadoc of the Cow Head Group, with the exception of K. pritchardi , from which it differs by its longer sicula, more robust form, steeply inclined stipes and narrower thecal apertures. Bulman (1950) compared his new species Didymograptus primigenius with D. taylori ; the overall dimensions and rhabdosome form of this taxon from the middle Tremadoc of Quebec are, however, closer to those of K. pritchardi. It is distinguished from this species by its more steeply inclined thecae and higher thecal density of 1 1 in 10 mm. Kiaerograptus undulatus sp. nov. Plate 1, figs 2-4; Plate 3, figs 1 and 2; Text-fig. 8a-h cf. 1937 Didymograptus norvegicus , n. sp.; Monsen, pp. 176-177, pi. 2, figs 7 and 8; pi. 4, figs 4 and 5; fig. 6. 1983 ? Kiaerograptus sp. cf. K. quasimodo Rushton; Henderson, p. 155, fig. 5 g-j. Derivation of name. From undulatus (Latin) meaning 'wavy', referring to the folded dorsal stipe margin. Type specimen. The holotype is GSC 87413, from Green Point (GP40). Figured Text-figure 8a. Diagnosis. Small rhabdosome composed of four (occasionally two or three) slightly declined, radiating stipes, measuring 0-7-0-8 mm wide proximally with rapid increase to the maximum TO mm. Sicula T5-T8 mm long, almost straight, with apertural width of 0-25 mm. Prominent sicular bitheca filling much of ‘notch' of basal rutellar margin. Autothecae strongly curved, with strong prothecal folds, wide apertures occupying one half of total stipe width and numbering 9-10 in 10 mm. Many or all autothecae with bithecae opening into large apertures. Material and localities. Fifteen flattened, non-isolated specimens from GP38, 40; MPS42C; CH8-34. One isolated, three-dimensional specimen from MPS42C. Description. The species is known only from small proximal fragments with stipes up to 8 mm long. The rhabdosome typically consists of four, gently declined, radiating stipes, formed by the dichotomous division of th21 and 22. Occasional specimens with two horizonal stipes considered to belong to this species have, however, been found. Stipes are generally 0-7-0-8 mm wide proximally, with rapid increase to F0 mm, although a few specimens are 10 mm wide proximally. The dorsal stipe margin is characterized by pronounced prothecal folds, although these are less conspicuous in more poorly preserved, flattened material. The sicula is F5-F8 mm long and almost straight throughout its entire length, with an apertural width ot 0-25 mm. Proximal development has not been observed clearly, but evidently agrees with that of other late WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GR APTOLITES 15 text-fig. 8. a-h, Kiaerograptus undulatus sp. nov., a-g x 5, it x 2 5. a, GSC 87413, Holotype, GP40. b, GSC 87296, CHN8.34. c, GSC 87318. MPS42C. d, GSC 87328, MPS42C. E, GSC 87359, MPS42C. f, GSC 87371, GP38. G, GSC 87414, GP40. h, GSC 87322, MPS42C. i p, Kiaerograptus magnus sp. nov. i-n GP38, x 5; i, GSC 87365; j, GSC 87404; k, GSC 87363; l, GSC 87372; m, GSC 87408; n, GSC 87361, Holotype (also figured PI. 1, fig. 7). o, GSC 87330, MPS42C (also figured PI. 1, fig. 3). p, GSC 87325, MPS42C. Tremadoc taxa. Thl1 diverges from the sicula relatively high, leaving the basal rutellar margin free for 05- 0-6 mm. Much of this ‘notch’ is, however, commonly filled by the sicular bitheca, giving a more robust and ‘filled-in’ appearance to the proximal region. Thl 2 also leaves the sicula above the level of the sicular aperture, leaving the antirutellar margin free for 01-0-2 mm. The free ventral margins of thl1 and l2 measure TO mm and 0-8 mm respectively; both show pronounced downward curvature throughout their free portions and splay out towards the apertures, which are 0-35-0-4 mm in diameter (half total stipe width). Th2L and 22 are commonly dichotomous, giving rise to the typically ‘tetragraptid’ form; in these specimens thl1 and l2 possess bithecae which open into large apertures alongside those of the autothecae and directly below the point of branching. It is unclear whether these bithecae are present in the two-stiped forms, or whether they occur throughout the rhabdosome. End-on views of the single isolated specimen suggest that they are indeed present in at least the succeeding few thecae, unless this specimen 16 PALAEONTOLOGY. VOLUME 34 is showing a third-order dichotomy. All autothecae throughout the rhabdosome show the same characteristic strong curvature, prothecal folds and wide apertures occupying one half of total stipe width. Thecal overlap is greater than one half, while thecal density is a constant 9-10 in 10 mm. Remarks. K. undulatus is a very distinctive form when well preserved owing to the sinuous nature of the thecae, which gives an appearance reminiscent of the Arenig sinograptids. The outline of the dorsal wall is somewhat variable, from specimens with strong prothecal folds to others with an almost straight dorsal margin. Although this may be partly an original morphological feature, the folds may well have been reduced by differential lateral spread on compaction, as described for the Upper Ordovician Dicellograptus complanatus Lapworth by Briggs and Williams (1981) and Williams et al. (1982). Henderson’s (1983) specimens of K. ? cf. quasimodo agree well with K. undulatus , although they are all two-stiped forms. The types of A'.? quasimodo described by Rushton (1981) from the middle or upper Tremadoc subsurface of central England differ, however, by their longer sicula and more steeply inclined thecae, resulting in a slightly higher thecal count. Most specimens of A.? quasimodo were two-stiped forms, although one possible three-stiped specimen with a higher thecal density was recorded by Rushton (1981, fig. 3c). K. ? quasimodo is clearly similar to A. undulatus and may well represent an ancestral taxon. A. undulatus is also comparable with Didymograptus norvegicus Monsen: this has a folded dorsal margin and equivalent thecal densities, but a rather smaller, inclined sicula 14 mm long and two reclined stipes. The only other associated species with which A. undulatus may be confused is K. bulmani sp. nov. The latter species has a much more slender and open rhabdosome, rather more gently inclined thecae with a marginally lower thecal density of 8-9 in 10 mm, and seems to lack the prominent folded dorsal margin (although one two-stiped specimen possibly referable to this species does have prothecal folds). Kiaerograptus magnus sp. nov. Plate 1, figs 5-7; Plate 3, figs 4 and 7; Text-fig. 7i-p Derivation of name. From magnus (Latin) meaning ‘large’, in reference to the large and robust sicula and proximal region. Type specimen. The holotype is GSC 87361, from Green Point (GP38). Figured Plate 1, fig. 7 and Text-figure 8n. Diagnosis. Robust rhabdosome with four, three or two stipes 1-2 mm wide proximally. Sicula up to 2-3 mm long, almost straight, with pronounced rutellum and aperture 0 5 mm diameter. Autothecae simple, inclined at 3CMf0° to dorsal margin, numbering 9-10 in 10 mm. Bithecae apparently lacking with exception of large sicular bitheca. Materials and localities. Ten flattened specimens from MPS42C and GP38. Four isolated, three-dimensional specimens from SP143 and MPS42C. Description. The rhabdosome is robust with two, three or four stipes, F2 mm wide proximally and increasing rapidly to over F5 mm within 5 mm. Only proximal fragments have been positively identified, with stipes up to 6 mm long. The sicula is long and wide, reaching up to 2-3 mm long measured along the gently convex rutellar margin, with a conspicuous rutellum projecting (F2 mm and long nema which is occasionally thickened or lorked (Text- fig. 8m, n). The sicula is 0-5 mm diameter at its aperture. Thl1 buds from the prosicula, growing down in contact with the metasicula for about 1 mm before diverging gently out at 20-30°. A large sicular bitheca fills most of the notch left between the rutellar margin of the sicula and ventral wall of thl 1 . The arrangement ol the sicula and first theca is highly symmetrical in young growth stages, giving an appearance approaching that found in the Arenig genus Isograptus. This symmetry is, however, lost during astogeny, as thl1 continues to grow with a free ventral wall up to FI mm long. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 17 Thl2 buds from high up th 1 1 , as does Th2L It grows down and across the sicula, its ventral wall intersecting the antirutellar apertural margin of the sicula. Its original angle of 30° subtended with the sicular axis decreases slightly throughout its distal portion, resulting in a concave free ventral margin up to 10 mm long. Remaining development is similar to other Kiaerograpius taxa, th2‘ and 22 normally being dichotomous, although one or both dichotomies may be suppressed. Remaining autothecae are straight and inclined at 30-40° to the dorsal margin. They overlap about one half their length, have simple apertures occupying about one half of total stipe width, and number 9-10 in 10 mm. No bithecae have been observed apart from that of the sicula. Remarks. The robust proximal region and large sicula separate K. magnus from all other coeval Kiaerograpius species and give an appearance reminiscent of Clonograptus. However, K. magnus lacks the common bithecae characteristic , of this genus during the late Tremadoc, and appears to only have a maximum of four stipes, although more complete specimens might potentially possess further delayed, dichotomous branching. The forked and thickened nemata present on some specimens are unusual for graptolites from this stratigraphical interval and may have some taxonomic significance. text-fig. 9. a-h, Kiaerograptus bulmani (Thomas, 1973). a, GSC 87400, GP38, x 10, b, GSC 87420, GP40, x 5. c-h, x 2-5 ; c, GSC 87324, MPS42C ; d, GSC 87339, MPS42C ; e, GSC 87295, CHN8.34 ; f, GSC 87364, GP38 ; G, GSC 87366, GP38; H, GSC 87319, MPS42C. i, K. bulmani (Thomas, 1973)7, GSC 87323, MPS42C, x2-5. Kiaerograptus bulmani (Thomas, 1973) Plate 1, figs 8 and 9; Plate, 3 figs 5, 6, 8-14; Text-fig. 9a-i 1971 Tetragraptus otagoensis (Benson and Keble); Erdtmann, pp. 259-260, pi. 33, figs 1-3. 1973 Tetragraptus bulmani sp. nov.; Thomas, pp, 530-531, pi. 2, figs b and c. 1979 Tetragraptus bulmani Thomas; Cooper and Stewart, p. 795, text-fig. 8/?, k. Type specimen. The holotype is specimen No. 64419 in the Mines Department Museum, Melbourne. From the middle Lancefieldian (La2), loc. 68, Staurograptus Gully, Parish of Springfield, Victoria. 18 PALAEONTOLOGY, VOLUME 34 Diagnosis (revised, incorporating descriptions by Thomas (1973) and Cooper and Stewart (1979)). Small, slender rhabdosome with four or three, radiating, gently declined stipes increasing from 0-4-0- 5 mm wide proximally to a maximum 0-8 mm (0-5-0-6 mm in scalariform or oblique preservation). Thecae simple, straight, gently inclined at about 20° and numbering a constant 8-10 in 10 mm. Sicula with bitheca, other bithecae apparently lacking except at dichotomies. Material and localities. Twenty flattened specimens from CH8-34; MPS42C; GP38, 40. Fifteen isolated, three- dimensional specimens from SPI43, MPS42C and GP38. Description. The rhabdosome is composed of four radiating, slender stipes up to 20 mm long. Proximally they have a dorso-ventral width of 0-4-0-5 mm, increasing to 0-5-0-7 mm in 5 mm and reaching a maximum of 0-8 mm. Stipes are commonly preserved in oblique or scalariform view, resulting in rather narrower widths of 0-5-0-6 mm. Rare preservation of the rhabdosome in lateral view reveals the stipes to be gently declined. Occasionally one dichotomy is suppressed, resulting in a three-stiped rhabdosome, although the majority of specimens from the Cow Head Group possess four stipes. One specimen possibly referable to K. bulmani (Text-fig. 9i) has only two, horizontal stipes, suggesting suppression of both dichotomies; this example however has a strongly folded dorsal margin and may not belong to this species. The sicula is 1-4 mm long (but apparently only 1-2 mm in non-isolated specimens), measured along the rutellar margin, with an apertural diameter of 0-2-0-25 mm. It is straight or almost straight throughout its length, with a small but conspicuous rutellum projecting 0-08 mm beyond the antirutellar, apertural margin. Thl1 buds from the prosicula on the rutellar side and grows down in contact with the sicula for 0-8-0-85 mm before turning out, after which it subtends an angle of 40° with the sicular axis for the remaining 0-7-0-9 mm of its length. Th 1 1 has an almost constant diameter of 0-2-0-23 mm during the second portion of its development and opens into a simple aperture. A sicular bitheca is invariably present, opening at a level varying from 0- 1 5 mm above the point of divergence of th 1 1 to just below the level of divergence. Development may be sinistral or dextral. Th 1 2 buds from th 1 1 about 0-5 mm below the apex of the sicula; it grows down and across at 30° to the sicular axis, maintaining this direction of growth throughout its length. Subsequent development and branching patterns appear to be typically ‘dichtograptid ’, with thl2, thd1 and th22 dichotomous (one or both branchings may be suppressed). Bithecae appear to be absent apart from that of the sicula, and possibly at dichotomies (Plate 3, fig. 13). Remaining thecae are simple, straight, gently inclined at about 20° to the dorsal margin and have apertures occupying one third to one half of total stipe width. Thecal overlap is about one-third, while thecal density is a uniform 8-10 in 10 mm throughout the rhabdosome. Rare flattened specimens appear to exhibit prothecal folds, but these are not present in isolated material. Critical observation suggests that they may be due to lateral spread of the apertural regions during flattening in oblique or scalariform orientation, apertural walls becoming visible on both sides of the stipe margin (e.g. Text-fig. 9b, g). Remarks. K. bulmani is distinct from other taxa at this stratigraphical level due to its narrow stipes and widely spaced thecae. Our specimens appear to agree with the Australian types in all respects, except in lacking flared thecal apertures. Such flaring is, however, common in many graptolites with straight, simple thecae, due to post-mortem, differential lateral spread during flattening of the rhabdosome and is, therefore, of no taxonomic importance. K. bulmani differs from Kiaerograptus otagoensis (Benson and Keble, 1936) by its rather narrower stipes and lower thecal densities; Erdtmann’s ( 1971 ) specimens referred to K. otagoensis are from Martin Point and clearly belong to K. bulmani. K. bulmani may be distinguished from K. undulatus sp. nov. by that species’ rather different proximal development, wider stipes and prominent prothecal folds. Occasionally, however, specimens are found preserved in oblique or scalariform orientation which could be assigned to either one of the species. The similarity in thecal style between K. bulmani and Kiaerograptus pritchardi (T. S. Hall) was noted both by Thomas (1973) and by Cooper and Stewart (1979). Our isolated and flattened material reveals that K. bulmani has a similar proximal development to that shown by both K. pritchardi and K. taylori , which is why we refer that species to Kiaerograptus rather than retaining within the dichograptid genus Tetragraptus. K. otagoensis is also similar and should be referred to this genus. Cooper and Stewart (1979) remarked that K. bulmani was rather similar to the WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 19 Bendigonian (lower Arenig) species Tetragraptus harti T. S. Hall. Williams and Stevens ( 1988a) recently redescribed this taxon from the D. bifidus Zone of the Cow Head Group and transferred it to the genus Etagraptus. Although similar in overall rhabdosome form, E. harti is a true dichograptid without a sicular bitheca, and any similarity to K. otagoensis is entirely hoinoeomorphic. Genus paratemnograptus nov. Type species. Paratemnograptus isolatus sp. nov. By monotypy. Diagnosis. Pauciramous, radiate rhabdosome with up to sixteen stipes arising from two primary stipes by three orders of widely spaced, delayed, irregular, dichotomous branching. Sicula with bitheca. Autothecae gently curved with moderate inclination, simple apertures and apparently lacking bithecae. Remarks. Proximal branching conforms to a standard tetragraptid plan, with two primary stipes and th2x and th22 dichotomous. Subsequent dichotomies are delayed and irregular, many large rhabdosomes possessing only four stipes. Overall form may, therefore, be similar to either Tetragraptus or Temnograptus , although both these genera are Arenig in age and lack bithecae. Paratemnograptus further differs from the diagnosis of Temnograptus given by Bulman (1970, p. VI 13) in having irregular dichotomous stipe division and non-denticulate thecae. The type species of Temnograptus , namely T. multiplex (Nicholson, 1868), is poorly known and based on inadequate, flattened and deformed material from an uncertain stratigraphic level. Further work may ultimately prove Paratemnograptus to be synonymous with Nicholson's genus. Paratemnograptus isolatus sp. nov. Plate 1, figs 10-16; Plate 2, fig. 4; Plate 4, figs 1-8; Text-fig. 10a-o 71899 Tetragraptus decipiens, n. sp.; T. S. Hall, pp. 168-169, pi. 17, figs 13-15; pi. 18, figs 16-19. 71904 Temnograptus noveboracensis sp. nov.; Ruedemann, pp. 619-620, pi. 5, figs 15-20, 35, 36. 71920 Tetragraptus decipiens , T. S. Hall; Keble, pp. 199-200, pi. 34, fig. 1 a-e. 71947 Temnograptus noveboracensis Ruedemann; Ruedemann, p. 284, pi. 44, figs 14—16; pi. 45, figs 1-4. 71962 Temnograptus aff. noveboracensis Ruedemann; Obut and Sobolevskaya, p. 79, pi. 3, fig. 3. 71966 Tetragraptus decipiens T. S. Hall; Berry, pp. 423-424, pi. 44, figs 5, 10, 11. 71969 Tetragraptus decipiens T. S. Hall; Bulman and Cooper, pp. 215-216, pi. 1, figs 1-4; fig. 3 a-c. 71974 Clonograptus sp. A; Jackson, pp. 46^47, text-fig. 4. 1974 Clonograptus sp. B; Jackson, p. 47, text-fig. 1 m. n. 71974 Tetragraptus decipiens T. S. Hall; Jackson, pp. 53-54, pi. 5, fig. 4. 1979a Temnograptus aff. regularis (Tornquist, 1904); Cooper, p. 58, pi. 1/; fig. 24. 7 19796 Tetragraptus decipiens T. S. Hall; Cooper, fig. 5 f. 1979 Temnograptus sp.; Cooper and Stewart, pp. 793-795, text-fig. 8c. 71979 Tetragraptus decipiens T. S. Hall; Cooper and Stewart, pp. 795-796, text-fig. 8a, b. 7 1982 Temnograptus sp. ; Gutierrez-Marco, fig. 2k. Derivation of name. From isolatus (Latin) meaning detached or separate, in reference to the widely spaced, irregular dichotomous branching. Type specimen. The holotype is GSC 87284, from the Ledge, Cow Head Peninsula (CHN8.30). Figured Text- fig. 10l. Diagnosis. Large rhabdosome with four to sixteen slightly flexuous, radiating stipes increasing rapidly from 0-8-F2 mm wide proximally to F4 mm maximum. Slender sicula with sicular bitheca. 20 PALAEONTOLOGY, VOLUME 34 funicle composed of thl 1 and thl2 2-5— 3-0 mm wide. Thecae simple, overlap one half, thecal density 9-10 in 10 mm. Material and localities. Numerous flattened specimens from CHN8.30; MPN17B; MPS42C; GP38, 40. Over twenty isolated, three-dimensional specimens from SP143, MPS42C. Description. The rhabdosome consists of four to sixteen long, slightly flexuous, radiating stipes reaching over 70 mm long and widening rapidly from 0-8-1 -2 mm proximally to a maximum I -4 mm which is then maintained. The sicula is 1 -6-1-8 mm long; it is straight throughout its length and relatively slender, reaching 0-2 mm diameter at the aperture. It has a small but conspicuous rutellum extending 0-1-0-15 mm beyond the antirutellar margin. Thl1 generally buds from the prosicula on the rutellar margin, although in one well- preserved specimen it buds from the antirutellar side, then swings immediately across to the rutellar margin. Thl1 grows down along the rutellar margin for 0-75 mm before turning outwards, subtending an angle of 40° with the sicular axis which is maintained throughout the remainder of its length. The distal rutellar margin of the sicula is left free for 0-25-0-3 mm. A sicular bitheca buds from the sicula below the point of origin of thl1, opening into an aperture a little above the point of divergence of the ventral wall of thl1 from the sicula. Development may be either right- or left-handed; thl2 buds from thl1 above its point of deflection, growing down and across the sicula and the ventral wall of thl2 intersects the base of the antirutellar sicula margin. Thl2 is dichotomous, giving rise to tfG1 and th2'\ as are each of these subsequent thecae to give the typical ‘ tetragraptid ’ proximal plan. The funicle, consisting of the first two thecae, is 2- 5-3-0 mm wide. Subsequent autothecae have a typically dichtograptid appearance; their angle of inclination with the dorsal margin increases from 30° initially to 50° towards the aperture, which is simple. Thecal overlap is one half of total length, while apertures occupy one half to two-thirds of total stipe width. Bitheca appear to be lacking with the exception of the sicular bitheca. Thecal density is a uniform 9-10 in 10 mm throughout the rhabdosome. Remarks. Although the overall form is distinctive, details of thecal morphology or proximal development are rarely seen in flattened specimens owing to common preservation in scalariform orientation. Most rhabdosomes have only four stipes, but sufficient specimens have been found with additional distal dichotomies to determine the variability of this morphological feature. There are no other associated species which might be confused with P. isolatus ; as can be seen from the list of synonymies, both the generic and specific identity of this taxon have, however, been problematic. Temnograptus regularis (Tornquist) as described by Tornquist (1904) and Monsen (1937) certainly appears similar, but both our material and that described by Cooper (1979a) has more widely spaced dichotomies, more slender stipes and is much earlier (late Tremadoc as opposed to middle Arenig). Temnograptus noveboracensis Ruedemann was based entirely on distal stipe fragments and is, therefore, not a strictly valid taxon. Ruedemann (1947, pi. 44, figs 14-16) did, however, figure three fragments from the Cow Head Group and these are likely to belong to P. isolatus. Tetragraptus decipiens T. S. Hall has been recorded previously from the late Tremadoc and early EXPLANATION OF PLATE 2 Fig. 1. Aorograptus victoriae (T. S. Hall, 1899). GSC 87309, MPS42C, x2-5 (also figured Text-fig. 1 1 L). Figs 2 and 3. Adelograptus cf. A. tenellus (Linnarsson, 1871). 2, GSC 87376, GP38, x 5. 3, GSC 87307, MPN17B, x 10. Fig. 4. Paratemnograptus isolatus gen. et sp. nov. GSC 87362, GP38, x 10 (also figured Text-fig. 10b). Fig. 5. Clonograptus sp. B. GSC 87314, MPS42C, x 2-5 (also figured Text-fig. I5j). Fie. 6. Clonograptus sp. A. GSC 87354, MPS42C, x 2-5. Figs 7-11. Rhabdinopora sp. 7, GSC 87308, MPNI7B, x 10. 8, GSC 87290, CHN8.30, x 5. 9, GSC 87291. CHN8.30. x 2-5. 10, GSC 87396, GP38, x 5. 11. GSC 87292, CHN8.30, x2-5. 13, GSC 87293, CHN8.30, x 2-5. Fig. 12. Dendroid indet., distal fragment. GSC 87316, MPS42C, x 5. PLATE 2 WILLIAMS and STEVENS, Late Tremadoc graptolites 22 PALAEONTOLOGY, VOLUME 34 WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GR APTOLITES 23 Arenig of Australasia and North America; the type specimens are, however, poor and are only juveniles (see Berry 1966). The only description of T. decipiens including anything more than juveniles was by Keble (1920). The taxonomic affinities of T. decipiens were discussed by Williams and Stevens (1988#), who concluded that many lower Arenig specimens were probably juvenile representatives of T. approximatus approximatus Nicholson. We furthermore believe that the Tremadoc examples of T. decipiens are probably juvenile representatives of our new species P. isolatus , in which case T. decipiens would be a senior synonym. The dimensions of the sicula given by Berry (1966) for the types of T. decipiens are, however, greater than those for P. isolatus ; he recorded that the sicula was 1 -9-2-4 mm long and 0-4-0- 5 mm wide at the aperture (cf. 1 -6— 1 -8 mm long and 0-2 mm wide for P. isolatus). Until a better population of T. decipiens is collected from the type locality, including large, mature rhabdosomes and specimens in which the presence or absence of a sicular bitheca can be determined, the synonymy with P. isolatus must remain questionable. Genus aorograptus nov. Derivation of name. From aoros (Greek), meaning pendulous, hanging or waving, in reference to the pendent nature of the rhabdosome. Type species. Bryograptus victoriae T. S. Hall, 1899, p. 165, pi. 17, figs 1 and 2. Diagnosis. Pendent or declined rhabdosome with regular, commonly delayed, dichotomous branching from two primary stipes. Sicula and most autothecae with bithecae; autothecae composed of simple, dichgraptid-like tubes, commonly curved with fairly high distal inclinations, stipes relatively robust. Remarks. Until more extensive taxonomic revision of Middle and Upper Tremadoc graptolites is accomplished, this genus is essentially monotypic. A. victoriae , the type species, has been previously assigned to both Bryograptus and Adelograptus. It differs from the former genus by having two, rather than three, primary stipes, and from the latter in having a relatively robust, large rhabdosome with regular branching. When preserved in radiate, rather than pendent, orientation, the rhabdosome gives an appearance which would normally have been referred to Clonograptus. As discussed elsewhere, we consider this genus to be a typically dichograptid, Arenig genus, lacking bithecae or any other ‘dendroid’ features (in the traditional sense). It is therefore likely that many specimens referred previously to Clonograptus are actually representatives of our new genus Aorograptus preserved in radiate (horizontal) orientation. Several previous authors have referred to the possibility that Bryograptus evolved to give the lower Arenig dichograptid genus Pendeograptus and/or the pendent didymograptids (see Fortey and Cooper 1986 for discussion). In our opinion, it is likely that Aorograptus evolved from Bryograptus in the late Tremadoc through loss of one primary stipe and the stolon system, then subsequently gave rise to Pendeograptus through loss of bithecae and further stipe reduction. It is not, however, the ancestor of Didymograptus ( Didymograptellus ) Cooper and Fortey, 1982, which almost certainly evolved from the Didymograptus ( Expansograptus ) nitidus group of extensiform didymograptids (see Williams and Stevens 1988#). text-fig. 10. Paratemnograptus isolatus gen. et sp. nov., a-c x 5, d-j x 2-5, k-o x 1. a, GSC 87301, MPN17B. b, GSC 87362, GP38 (also figured PI. 2, fig. 4). c, GSC 87370, GP38. d, GSC 87333, MPS42C. e, GSC 87402, GP38. F, GSC 87280, CHN8.30. G, GSC 8728 1 . CHN8.30. H, GSC 87382, GP38. i, GSC 87282, CHN8.30. j, GSC 87283, CHN8.30. k, GSC 87355, MPS42C. l, GSC 87284, Holotype, CHN8.30. m, GSC 87334, MPS42C. n, GSC 87415, GP40. o, GSC 87311, MPS42C. 24 PALAEONTOLOGY, VOLUME 34 Aorograptus victoriae (T. S. Hall, 1899) Plate 2, fig. 1; Plate 3, fig. 15?; Plate 4, figs 9-14; Plate 5, figs 1-8; Text-fig. 11a-q 1899 a Bryograptus victoriae , n. sp.; T. S. Hall, p. 165, pi. 17, figs 1 and 2. 1899 a Bryograptus clarki, n. sp.; T. S. Hall, pp. 165-166, pi. 17, figs 3 and 4. 18996 Bryograptus victoriae', T. S. Hall, p. 450, pi. 22, figs 11 and 12. 1914 Bryograptus sp. ; T. S. Hall, pi. 8, figs 5 and 6. 1932 Bryograptus victoriae T. S. Hall; Harris and Keble, pi. 4, fig. 2. 1933 Bryograptus pauxillus sp. nov. ; Benson, p. 403 ( nom . mid.). 1936 Bryograptus hwmebergensis Moberg; Benson and Keble (pars), pp. 269-270, pi. 30, figs 1-11 (non pi. 30, figs 14 and 15 = A. cf. tenellus (Linnarsson)?). 71936 Bryograptus simplex Tornquist; Benson and Keble, p. 270, pi. 30, figs 12 and 13. 1938 b Bryograptus victoriae T. S. Hall; Harris and Thomas, pi. 1, fig. 7. 1938 Bryograptus clarki T. S. Hall; Harris and Thomas, pi. 1, fig. 8. 1941 Adelograptus victoriae (T. S. Hall); Bulman, p. 115 (no description or figures, but refers to Adelograptus and synonymises A. clarki). 1955 Adelograptus asiaticus', Mu, p. 30, pi. 10, figs 4-7. 71955 Adelograptus sinicus'. Mu, p. 30, pi. 10, fig. 8. 1960 Bryograptus victoriae T. S. Hall; Thomas, pi. 1, fig. 6. 1960 Bryograptus clarki T. S. Hall; Thomas, pi. 1, fig. 7. 71960 Adelograptus victoriae (T. S. Hall); Berry, pp. 46-47 (remarks only, no descriptions or figures). 1966 Adelograptus clarki (T. S. Hall); Berry, pp. 419-421, pi. 44, figs 2 and 4. 1966 Adelograptus victoriae (T. S. Hall); Berry, pp. 421^422, pi. 44, fig. I. 1968 Adelograptus kazakhstanensis Tzaj, n. sp. ; Tzaj, pp. 493-494, pi. 5, fig. 2. 1968 Bryograptus ulutanensis Tzaj, n. sp.; Tzaj, p. 495, pi. 5, fig. 3. 1969 Bryograptus 7 sp. of T. S. Hall; Bulman and Cooper, fig. Aa, b. 1974 Adelograptus victoriae (T. S. Hall); Jackson, p. 45, pi. 5, fig. 2; text-fig. 2a. 1974 Adelograptus kazakhstanensis Tzaj; Tzaj, pi. 37, pi. 1, figs 6 and 7. 1974 Bryograptus ulutanensis Tzaj; Tzaj, pp. 38-39, pi. 2, figs 1-3; fig. 4. 1974 Bryograptus sp.; Tzaj, p. 39, pi. 2, fig. 4. 1979 a Adelograptus clarki (T. S. Hall); Cooper, pp. 54-55, pi. 2a, 6; fig. 19a-c. 19796 Adelograptus victoriae (T. S. Hall); Cooper, fig. 5 g. 1979 Adelograptus victoriae (T. S. Hall); Cooper and Stewart, pp. 784-785, text-fig. 8 g,j, I. 1979 Adelograptus asiaticus Mu; Wang et al., pp. 499-500, pi. I, figs 6 and 7; fig. 8 a-e. 1979 Adelograptus simplex (Tornquist); Wang et al., p. 501, fig. 9a. 1979 Adelograptus victoriae (T. S. Hall); Wang et al., p. 501, fig. 96. Type specimen. Nat. Mus. Victoria No. PI 4240 (figured by Hall 1899, pi. 44, fig. 1) was designated lectotype by Berry (1966, p. 421). From the middle Lancefieldian (La2) near Lancefield, Victoria, Australia. Diagnosis. Pendent or declined rhabdosome with many stipes increasing from 06-08 mm wide proximally to a maximum T2 mm. Autothecae with concave ventral margin, flared aperture and with bithecae, thecal density increasing from 8 in 10 mm proximally to 10 in 10 mm distally. EXPLANATION OF PLATE 3 Figs I and 2. Kiaerograptus undulatus sp. nov. GSC 87436, MPS42C. 1, x 20, 2, x 40. Figs 3, 4, 7. Kiaerograptus magnus sp. nov. 3, GSC 87433, SPI43. 4, GSC 87474, MPS42C. 7, GSC 87473, MPS42C. All x 40. Figs 5, 6, 8-14. Kiaerograptus bulmani (Thomas, 1963). 5, GSC 87462, GP38. 6, GSC 87434, SPI43. 8 and 9, GSC 87459, GP38. 10, GSC 87442, SPI43. 1 1, GSC 87443, SPI43. 12, GSC 87446, SPI43 (also figured Text- fig. 5a, b). 13, GSC 87488, MPS42C. 14, GSC 87444, SPI43. All x40. Fig. 15. Aorograptus victoriae (T. S. Hall, 1899)7, GSC 87465, MPS42C, x40. Scanning electron micrographs of isolated specimens. PLATE 3 WILLIAMS and STEVENS, Kiaerograptus , Aorograptus 26 PALAEONTOLOGY, VOLUME 34 Material and localities. Many isolated, three-dimensional and flattened, non-isolated specimens from all localities in the late Tremadoc of the Cow Head Group described in this paper. Description. The rhabdosome has a pendent form with up to sixteen branches formed by four delayed dichotomies and sometimes exceeds 60 mm in diameter. Occasionally specimens are preserved flattened in horizontal orientation; in this instance the rhabdosome has a radiate, ‘clonograptid ’ appearance. Stipe widths vary depending on astogeny, but are commonly 0-6-0-8 mm proximally, increasing distally to a maximum T2 mm. The sicula is large, measuring 1 -4-2-0 mm long; although such variation is not found in most other associated taxa, detailed observation has revealed continuous variation between the extremes and taxonomic division based solely on this criterion is therefore not warranted. The sicula is more or less straight, increasing gradually in diameter to 0-25-0-3 mm at the aperture. The nema is commonly preserved, reaching up to 4 mm long, and the rutellum is pronounced, extending 0-15 mm beyond the antirutellar margin. Thl 1 buds from the prosicula, growing down in contact with the rutellar margin for 0-75 mm before bending out at an angle of 70° to the sicular axis. It subsequently curves down throughout its length, ending subparallel to the sicular axis after 0-8-1 -2 mm. The aperture has a short selvage and is 0 3-0-4 mm wide (one half to two- thirds total stipe width). A sicular bitheca buds from the sicula 0-5 mm below the point of origin of thl1. It varies tremendously in length, from little more than a concealed foramen to a theca with an aperture just above the point of divergence of the ventral wall of thl1. The distal notch between the rutellar margin of the sicula and thl1 is also rather variable in size, from 0-4-0-6 mm long. Thl ’ buds from thl1 not far below its point of origin, growing down and across the sicula at an angle of 20-30°. Development may be either dextral or sinistral. The ventral wall of thl2 intersects the antirutellar margin of the sicular aperture, after which the theca arches gently down towards the thecal aperture, the free portion of ventral wall measuring 0-8-1 -0 mm. Thl2 is dicalycal, giving rise to both th2* and th22. T1121 and th22 are also normally dicalycal, although one dichotomy is occasionally suppressed to give an asymmetrical branching pattern. Delayed but fairly regular dichotomous branching occurs throughout the rhabdosome, resulting in third or fourth order stipes in mature specimens. Each autotheca possesses a bitheca, whose apertures open on alternating sides of the stipe; these are also clearly visible at each dichotomy in isolated material. Such a pattern of thecae is strongly reminiscent of typical anisograptids, but careful examination has failed to reveal any hint of a stolon system embedded in the dorsal margin. Thecal style is consistent throughout the rhabdosome, autothecae possessing concave ventral margins with flared apertures which occupy about one half total stipe width. Interthecal septae have an initial inclination of 10° to the dorsal margin, increasing ventrally to 20-30°. Thecal overlap is approximately one half total thecal length. Thecal density is unusual in that it increases from 8 in 10 mm proximally to 10 in 10 mm distally. It is unclear whether this is due to more steeply inclined thecae, shorter thecae, or greater thecal overlap, but is opposite to the situation found in most graptolites where thecal density decreases distally. Remarks. All previously described specimens of A. victoriae , including the types, have been small rhabdosomes with only second order dichotomies. However, the lectotype has identical proximal dimensions and form and we have no hesitation in assigning our material to this species. Bulman (1941) was the first to recognize that A. clarki was synonymous with A. victoriae. Berry (1966) subsequently considered the two to be distinct taxa, A. clarki being distinguished by lateral rather than dichotomous branching and less strongly declined stipes. Cooper (1979a) remarked that the two would probably prove conspecific; Cooper and Stewart (1979) formally synonymized them. EXPLANATION OF PLATE 4 Figs 1-8. Paratemnograptus isolatus gen. et sp. nov. 1, GSC 87449, SPI43. 2, GSC 87450, SPI43. 3, GSC 87439, MPS42C. 4, GSC 87482, MPS42C. 5, GSC 87424, MPS42C. 6, GSC 87435, MPS42C. 7, GSC 87486, MPS42C. 8, GSC 87440, MPS42C. All x 40. Figs 9-14. Aorograptus victoriae (T. S. Hall, 1899). 9 and 1 1, GSC 87437, MPS42C, x 20. 10, GSC 87418, note bitheca, MPS42C, x 20. 12, GSC 87422, MPS42C, x 20. 13, GSC 87477, MPS42C, x40. 14, GSC 87425, MPS42C, x 40. Scanning electron micrographs of isolated specimens. PLATE 4 WILLIAMS and STEVENS, Paratemnograptus , Aorograptus 28 PALAEONTOLOGY, VOLUME 34 text-fig. 1 1 . For legend see opposite. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 29 We see no justification in retaining two separate species in the light of our own work and that of previous authors, and therefore follow Bulman (1941) and Cooper and Stewart (1979) in regarding A. clarki as a junior synonym of A. victoriae. The species of ‘ Bryograptus ’ figured by Hall (1914) and Bulman and Cooper (1969) is identical to our mature specimens of A. victoriae. Specimens assigned to ‘ B. pauxillus, sp. nov.’ (Benson 1933) and ‘ B. hunnebergensis Moberg’ (Benson and Keble 1936) were recognized by Bulman (1941, p. 115) as belonging to A. victoriae. The proximal ends of ‘ Bryograptus simplex Tdrnquist’ figured by Benson and Keble (1936) appear similar in branching pattern and overall form to A. victoriae , but the sicula is much longer (3 mm and 4-5 mm if their magnifications are correct). Tornquist’s original specimens (1904, pp. 3^1, pi. 1, figs 1-4) have a similarly long sicula, but are recorded from the T. phyllograptoid.es Zone of southern Sweden. Williams and Stevens (1988a) considered this interval to be equivalent to the lower Arenig T. akzharensis Zone of the Cow Head Group. It is therefore most likely that B. simplex is synonymous with Pendeograptus fruticosus (J. Hall) or P. cf. P. pendens (Elies) as described by Williams and Stevens. The similarity of A. victoriae to the lower Arenig P. fruticosus is remarkable: proximal development, rhabdosome branching and thecal style (autothecae in A. victoriae) are all very similar, although the two may be distinguished by the longer sicula of P. fruticosus and bithecae and more numerous branching in A. victoriae. The various Chinese and Russian species described by Mu (1955), Wang et al. (1979) and Tzaj (1968, 1974) all appear to be synonymous with A. victoriae , as do the specimens figured by Wang et al. (1979, fig. 9a) as A. simplex (Tornquist). As noted in the discussion of Aorograptus , the original assignation of A. victoriae to Bryograptus is invalid following the definition given by Bulman (1970, p. V39), who stated that Bryograptus is an anisograptid which develops ‘from three primary stipes by irregular and apparently lateral branching’. Obut's (1957) inclusion of both Bryograptus and the dichograptid genus Pendeograptus within a family Bryograptidae is therefore clearly unacceptable. Genus adelograptus Bulman, 1941 Type species (by original designation). Bryograptus ? Hunnebergensis Moberg, 1892, p. 92, pi. 2, figs 5-7 (?8 and 9). Diagnosis, (revised using Bulman 1941, p. 114). Rhabdosome declined or horizontal, often somewhat lax and flexuous, formed from two primary branches by regular or irregular, commonly delayed, dichotomous branching. Sicular bitheca always present, additional bithecae and stolothecae present in some species, absent in others, autothecae straight, with simple apertures and low inclination, stipes consequently slender. Remarks. The revision of Adelograptus permits incorporation of many slender, regularly branching taxa previously accommodated within the rather unsatisfactory genus Clonograptus. The type species of Clonograptus (C. rigidus ) is now recognized as having a Lower Arenig age and probably belongs within the dichograptids (see previous discussion in text). Other more robust, pendent species originally assigned to Bryograptus (e.g. ' B. ’ victoriae) but since transferred to Adelograptus (Bulman 1941) because of their two primary stipes are here assigned to a new genus Aorograptus (see generic remarks). Although such a classification still has its limitations, it is closer to a true phylogenetic grouping text-fig. 11. Aorograptus victoriae (T. S. Hall, 1899), a-g x 5, h-q x 2 5. a, GSC 87374, GP38, b, GSC 87401, GP38. c, GSC 87405, GP38. d, GSC 87297, CHN8.32. e, GSC 87295, CHN8.34. f, GSC 87367, GP38. G, GSC 87397, GP38. h, GSC 87326, MPS42C. i, GSC 87379, GP38. J, GSC 87332, MPS42C. k, GSC 87310, MPS42C. l, GSC 87309, MPS42C (also figured PI. 2, fig. 1). m, GSC 87355, MPS42C. n, GSC 87321, MPS42C. o, GSC 87320, MPS42C. p, GSC 87336, MPS42C. Q, GSC 87313, MPS42C. 30 PALAEONTOLOGY, VOLUME 34 than that used previously, all members having similar proximal developments and thecal styles. It permits the transfer of Clonograptus tenellus Linnarsson to Adelograptus as suggested by Maletz and Erdtmann (1987), making sense of Hutt’s (1974) observation that C. tenellus and A. hunnebergensis have identical proximal development patterns and may only be distinguished following subsequent branching. In his original diagnosis, Bulman (1941) stated that branching in Adelograptus was apparently lateral rather than dichotomous. All studies using isolated material of the genus since that time, including the present study and that of Hutt (1974), have found branching to be dichotomous; the diagnosis is therefore consequently emended. The genus Par adelograptus was erected recently by Erdtmann et al. (1987) for non-bithecate forms which would previously have been assigned to Adelograptus or Clonograptus. The genus is characterized by slender thecae with simple or modified apertures, and considered to be ancestral to Kinnegraptus Skoglund, 1961 and other kinnegraptid genera. All their described species are from the lower Arenig and lack a sicular bi theca ; none of our taxa may therefore be accommodated within this genus. Adelograptus altus sp. nov. Plate 5, figs 9-13; Plate 5, figs 14? and 15?; Text-fig. 12a-g 1979 Adelograptus sp. ; Cooper and Stewart, text-fig. Id-f, h (no description). Derivation of name. From altus (Latin) meaning ‘high’, in reference to the relatively high level of divergence of the first two thecae from the sicula. Type specimen. The holotype is GSC 87430, an isolated specimen mounted on an SEM stub, from MPS42C. Figured Plate 5, figure 12. Diagnosis. Sicula T5-F8 mm long with distal convex curvature, with both rutellar and antirutellar margins free distally. Sicular bitheca opens at same level where ventral wall of th 1 1 diverges from sicula. Thl1 and l2 are gently declined with concave free ventral margins and gently flared apertures, increasing from 0-2 mm diameter to 0-4— 0-5 mm at the aperture. Material and localities. Nine isolated, three-dimensional proximal fragments, eight flattened, non-isolated proximal fragments. Several possible mature, non-isolated rhabdosomes. From CHN8.30, SPI43, MPN17B. MPS42C. Description. The species is defined primarily on its distinctive pattern of proximal development. Overall form is apparently similar to that of A. cf. tenellus (Moberg) with the exception of a slightly narrower funicle, but the two are clearly separated by the proximal form seen both in flattened and isolated material. The sicula is F5-F8 mm long measured along the rutellar margin, with a distal convex curvature. The sicular aperture is 0-2-0-25 mm wide, with a pronounced rutellum extending 0 05—0- 1 mm beyond the antirutellar margin. Thl1 buds from the prosicula on the rutellar side, growing down along this margin for 0-6-0-75 mm EXPLANATION OF PLATE 5 Figs 1-8. Aorograptus victoriae (T. S. Hall, 1899). 1, GSC 87421, MPS42C, x 40. 2, GSC 87457, MPS42C, x 40. 3, GSC 87426, MPS42C, x 40. 4, GSC 87475, x40. 5, GSC 87472, MPS42C, x20. 6, GSC 87453, MPS42C, x 40. 7, GSC 87478, MPS42C, x20. 8, GSC 87487, MPS42C, x 20. Figs 9- 13. Adelograptus altus sp.nov. MPS42C, x 40 except Fig. 1 1 (= x 20). 9, GSC 87429. 10 and 1 1, GSC 87441. 12, GSC 87430, Holotype. 13, GSC 87455. Figs 14 and 15. Adelograptus altus sp. nov.? Juvenile growth stages, x40. 14, GSC 87460, GP38. 15, GSC 87427, MPS42C. Scanning electron micrographs of isolated specimens. PLATE 5 WILLIAMS and STEVENS, Aorogrciptus , Adelograptus 32 PALAEONTOLOGY, VOLUME 34 text-fig. 12. a-g, Adelograptus altus sp. nov., a-e x 5, F and G, x 2-5. a, GSC 87302, MPN17B. b, GSC 87377, GP38. c, GSC 87360, MPS42C. d, GSC 87337, MPS42C. e, GSC 87341. MPS42C. f, GSC 87285, CHN8.30. G, GSC 87300, SPI43. h and i, Adelograptus antiquus (T. S. Hall, 1899)7, x 5. h, GSC 87340, MPS42C. i, GSC 87416, GP40. before turning sharply out, subtending an angle of 70-80° with the distal sicular axis. The rutellar wall of the sicula is free for 0-4-0-6 mm distally, while the sicular bitheca opens cryptically at the same level at which the ventral wall of th 1 1 diverges from the sicula. The point of origin of the bitheca is unclear and appears to be concealed by the early dorsal wall of th 1 1 . The stipe is 01 5-0-2 mm wide where th 1 1 leaves the sicula; the free ventral wall of th 1 1 has as strong concave curvature, leading to a splayed-out aperture and an undeformed apertural stipe width of 0-4-0-5 mm. The ventral wall of th 1 1 is free for L0-L2mm before the aperture is reached, which is 0-25-0-3 mm wide (i.e. two-thirds of total stipe width). Thl2 buds from th 1 1 0-5 mm below its origin, growing immediately across and down the sicula at 45-50° from the sicular axis. Development may be either right- or left-handed; it is therefore meaningless to discuss reverse and obverse aspects of the sicula, as these vary from one specimen to the next. Once thl2 has reached the antirutellar margin of the sicula it turns up slightly, subtending an angle of 3(U40° with the distal sicular axis in most cases. It has a similar concave ventral margin and splayed-out aperture to thl1, the stipe width measuring 0-2 mm initially, but increasing to 0-4-0-5 mm (undeformed) by the aperture. The budding of th21 and 22 appears to be typically isograptid, with thl2 dicalycal. There is no evidence for further dichotomies in isolated material, but one non-isolated specimen assigned to this species (Text-fig. 12e) and those of Cooper and Stewart (1979) show dichotomous branching of th2x and 22. Measurement of several slender ‘clonograptid ’ rhabdosomes demonstrates several with funicles of equivalent width to that which would be expected from isolated specimens. We conclude that although branching is variable, mature rhabdosomes have overall appearances of those specimens illustrated in Text-figure 12f, g. Remarks. The description of A. altus is based primarily on isolated, three-dimensional material, making comparison with other similar Adelograptus species difficult if known only from flattened, non-isolated specimens. The proximal budding pattern is similar to those shown by A. cf. A . tenellus and Adelograptus sp. A, but the sicula is longer than that of A. cf. A. tenellus and shorter than that of Adelograptus sp. A, while most specimens have a prominent ‘notch' between the free ventral wall of thl2 and the distal antirutellar margin of the sicula. Comparison using overlays clearly shows the incompatibility of the three species in terms of exact budding patterns and angles of thecal inclination. The specimens figured by Cooper and Stewart (1979) as Adelograptus sp. are identical to those described here, with the exception of thl1 which is marginally shorter. Unfortunately they did not describe their material or make any reference to it in the text. No other comparable specimens have been described or figured previously. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GR APTOLITES 33 text-fig. 13. Adelograptus cf. A. tenellus (Linnarsson, 1871), a-f x 5, g x 10, h-j x 2 5. A, GSC 87403, GP38. b, GSC 87389, GP38. c, GSC 87409, GP38. d, GSC 87303, MPN17B. E, GSC 87347, MPS42C. f, GSC 87348, MPS42C G, GSC 87349, MPS42C. h, GSC 87398, GP38. i, GSC 87394, GP38. J, GSC 87395, GP38. Adelograptus cf. A. tenellus (Linnarsson, 1871) Plate 2, figs 2 and 3; Text-fig. 1 3a— j cf. 1871 Dichograptus tenellus ; Linnarsson, p. 795, pi. 16, figs 13-15. cf. 1909 Clonograptus tenellus Linnarsson (and vars.); Westergard, pp. 68-72, pi. 4, figs 17-29. cf. 1929 Clonograptus tenellus (Linnarsson); Stubblefield, pp. 278-262, text-figs 1, 8-11. 1936 Bryograptus hunnebergensis Moberg; Benson and Keble (pars), pp. 269-270, pi. 30, figs 14 and 15 (non pi. 30, figs 1-11 = A. victoriae (T. S. Hall)), cf. 1987 Adelograptus tenellus (Linnarsson); Maletz and Erdtmann, pp. 180-182, pi. 1, figs a-c, pi. 2, figs a-m. Material and localities. About ten flattened specimens from MPN17B, MPS42C and GP38 and five isolated, three-dimensional specimens from MPS42C. Description. Rhabdosome with several slender stipes formed by delayed dichotomous branching from two primary stipes. The largest rhabdosome seen has a diameter of about 30 mm, with four dichotomies on the most complete portion, suggesting a total of thirty-two stipes. Dichotomous branching is apparently irregular, with a normal spacing of 3-5 mm (i.e. every two or three thecae). Thecal outline is rarely seen owing to 34 PALAEONTOLOGY, VOLUME 34 preservation in scalariform view, but when present stipe width is seen to measure 0-35-0-4 mm proximally, increasing distally to a maximum 0-6 mm. The sicula is 1-1-1 -4 mm long, and has a gentle convex curvature with respect to the rutellar margin in the distal one-third to one halt' its length. It is 0-2-0-25 mm wide at the aperture, with a slight rutellum. Thl1 buds from the prosicula on the rutellar margin. It grows down in contact with this margin for 0-6-0-7 mm, before deflecting sharply out, subtending an angle of about 100° with the sicular axis. This leaves the distal rutellar margin of the sicula free for 0-4-0-5 mm. Thl1 then curves gently downwards until its aperture is reached, leaving a free ventral wall 1 -0—1 - 1 mm long. The thecal aperture is 0-3 mm wide with a prominent flaring at the tip in some specimens. A sicular bitheca is present, originating a little below the point of origin of thl1 on the obverse side. The level of its aperture lies a little above the point of deflection of thl1, and is therefore not seen except in isolated specimens. Thl2 buds high up from thl1, growing down and across the sicula on the reverse side, then curving out so that its ventral wall cuts the base of the antirutellar sicular margin. It subtends an angle of 60-70° with the sicular axis at its point of divergence; this angle is maintained for the remaining TO- 1-1 mm of growth, although the theca sometimes curves down very slightly before the aperture. The funicle formed by the sicula and first two thecae measure 2-5— 2-8 mm long when preserved horizontally to bedding. Th2‘a buds from thl2 on the rutellar margin of the reverse side some 0-7 mm above the base of the rutellum. It follows the dorsal wall of thl1 until just before the aperture is reached; at this point th21b buds from th21a, the two growing in contact for about 0-15 mm before the aperture of thl1 is reached. They then diverge to give the first dichotomous branch. Although bithecae are apparently lacking on most thecae, branching fragments belonging to this or a related species show a bithecal aperture above the aperture of the autotheca when dichotomous branching occurs. Bithecae also occur at the dichotomies of several other unrelated late Tremadoc taxa, and are thought to represent an intermediate stage towards total loss of bithecae. Th22a buds from thl2 near its point of divergence from the sicula, developing and branching in a similar fashion to th21a. Each stipe then divides dichotomously every two to three thecae. Thecal density is a low, 6-7 in 10 mm where visible, although this is difficult to determine owing to frequent branching and usual preservation in scalariform view. Remarks. Adelograptus tenellus was revised recently by Maletz and Erdtmann (1987), who selected a neotype and thoroughly discussed the morphological variation found within the species. They conclude the nominate species to be a lower Tremadoc form occurring definitely only in Scandinavia, the Baltic region and Britain. Records of the species from late Tremadoc strata are, therefore, likely to be erroneous. Our material differs from the type material in having a shorter funicle and noticeably lower thecal density. Variation is so great that definition of a new taxon is withheld pending further, more detailed quantitative studies of late Tremadoc material both from western Newfoundland and elsewhere. A few previously published descriptions include material comparable to ours; some of Westergard’s (1909) specimens of C. tenellus and varieties are very similar, but there is a great deal of variation in his figured specimens and probably more than one species represented. C. tenellus kingi Benson and Keble, 1936 is similar in overall form, but thecal density is extremely high (17-21 in 10 mm). Their specimen of 'C. tenellus' (1936, pi. 32, fig. 4) also has a high thecal count. Benson and Keble (1936, pi. 30, figs 14 and 15) figured two proximal fragments more-or-less identical to our material; these are referred to Bryograptus simplex Tornquist in the plate description, as are figs 12 and 1 3. The latter two specimens have a very different appearance, are referred to ‘ B. hunnebergensis ' in the text, and probably belong to A. victoriae (see discussion of A. victoriae elsewhere in this paper). The Newfoundland specimens of A. cf. tenellus have a wider funicle than ‘C. tenellus sensu lato' of Cooper (19796, fig. 5c) and Cooper and Stewart (1979, fig. 8m); both these appear to be a different species. WILLIAMS AND STEVENS: NEWFOUNDLAND TREMADOC GRAPTOLITES 35 Adelograptus antiquus (T. S. Hall, 1899)? Text-fig. 12h, i ? 1 899 Leptograptus antiquus , n. sp.; T. S. Hall, p. 166, pi. 17, figs 5 and 6. ? 1979a Adelograptus! antiquus (T. S. Hall); Cooper, pp. 51-54, pi. 2, fig. c-e\ fig. \la-k. 18. 71979 Kiaerograptus antiquus (T. S. Hall); Cooper and Stewart, pp. 791-792, text-fig. 8 d, e. (summary only) Material and localities. Two flattened specimens from MPS42C and GP40. Remarks. Careful comparison of these two specimens with Adelograptus species from western Newfoundland reveals them to differ in terms of their small, slender sicula, wide funicle and low thecal density. These however appear to agree with those given for A. antiquus by Cooper (1979 flj 3 C C CD £ 'w a? CD t - O 0) CD w, O a ■o E CD o c CD X o CO B H, I H - S A,E K - S L P - S Q -S M - S N - S X v,w S.T-S U -s Y -S ( 1 Outcrops designated A- N.) i2 Outcrops designated S-Y.) text-fig. 1. For legend see opposite. 152 PALAEONTOLOGY, VOLUME 34 LOCALITY AND GEOLOGY The major localities yielding Spongiophyton are listed below and in Text-figure 1 ; these include some of Dawson’s localities as well as other outcrops in New Brunswick and Gaspe, all of late Lower Devonian (Emsian) age. 1 . Atholville, New Brunswick: Route 17 roadcut into Atholville (Locality P, Text-fig. 1) and outcrop along Beauvista Drive (Locality Q, Text-fig. 1). Specimens occur scattered on bedding surfaces of channel fills at the former and as stacks of individuals in grey bands of poorly-bedded sandy claystone at the latter. 2. Localities El, K, M, and N (Gensel and Andrews 1984; Text-fig. 1) along the Restigouche River, near Dalhousie Junction, New Brunswick, from which several vascular plants have been described (Gensel 1982). The thalli are sporadic and usually are intermixed with the more abundant vascular plant remains. 3. Cross Point, Quebec: at the Bordeaux Quarry and just to the north along Route 132. These rocks were included in the Gaspe Sandstone Group (McGerrigle 1950; Alcock 1935) and may fall within the LaGarde Formation of Dineley and Williams (1968). The Bordeaux Quarry, one of Dawson’s collecting sites, consists of a sequence of red-brown sandstones alternating with conglomerate bands. The thalli occur in the sandstone along with slender ribbed axis (? Psilophyton) and Prototaxites ‘logs’. These same sediments are exposed along a new roadcut on Route 132, where bedding surfaces show alternating conglomerate and sandstone layers, with the latter being a thin ‘wash’ of pebbles. Thalli densely cover the bedding surfaces of the sandy layers; a specimen of Prototaxites about 5 m long and 0-8 m wide and numerous smaller ones also were found in or immediately adjacent to these layers. 4. At several horizons of the Battery Point Formation near Cap-aux-Os, Gaspe and along the Laurencelle road which leads from Cap-aux-Os to Cap-des-Rosiers Est, Gaspe Peninsula, Quebec (localities S, T, U, Y of Gensel and Andrews 1984; Text-fig. 1). Thalli are rare to fairly abundant and appear to have been transported. 5. South shore of Gaspe Bay at several localities. (See McGregor 1977 for stratigraphic correlations.) TAPHONOMY, MATERIAL AND METHODS Considerable variation in mode of burial exists among thalli obtained from these outcrops. The majority are spread along bedding planes of fluvial, often channel-deposited sediments, with no preferred orientation (PI. 1, figs 1 and 5), suggesting they were transported moderate distances prior to deposition. At locality Q, thalli occur in stacks several entities deep and are so tightly held together that it is difficult to determine if they represent one plant or several (PI. 1, figs 2-4). Thick coverage of some bedding surfaces by many such stacks suggest that these thalli were growing in considerable abundance in some areas and buried rapidly near (or at?) their site of growth. We tend to discount the possibility that the stacks of thalli resulted entirely from the way in which they grew because they are not all orientated the same way up within a stack. The thalloid plants are preserved as compressed cuticles which occasionally show impressions of internal cells, thus mostly demonstrating general morphology. We use the term cuticle to refer to the inert outer covering of these fossils. It is resistant to oxidative maceration and shows a pattern of cell outlines on the inner surface, features found in the cuticle of higher plants. Its chemical composition is unknown, but its morphology compares closely with the lipid-derived cuticle known to occur only in embryophytes (higher plants). Although some algae apparently possess a thin, protein-rich outer covering that withstands some acid hydrolysis. (Hanic and Cragie 1969), these have not been shown to be resistant to coalification during fossilization. EXPLANATION OF PLATE 1 Figs 1-7. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. 1-5, various ways the thalli occur in the sediment; 1, specimen from Route 132, Quebec locality with numerous scattered thalli on the bedding plane; this specimen was etched in HF, GSC 93004, x0-8. 2, surface view of mass of thalli from a poorly bedded siltstone, Atholville locality. New Brunswick, GSC 93003, x L2. 3, lateral view of mass of stacked thalli, Atholville locality. New Brunswick, GSC 93005, x 1-4. 4, mass of stacked thalli from Atholville; N.B. locality, isolated by maceration in HF, GSC 93008, x 3. 5, numerous thalli after etching a single bedding plane, locality N, New Brunswick, GSC 93006, x L2. 6, SEM of thallus isolated by maceration, distal region to left; basalmost divisions produce upper and lower lobes (a), next 1-3 divisions produce side-by-side ones (b); note constrictions in lobes, pores (some indicated by arrows); roughness and cracks probably produced by deterioration under SEM vacuum, x 10. 7, thallus fragment with four dichotomies mostly in same plane, some with apices preserved, poral surface up, lying on etched surface of sediment, GSC 9301 1, x 4. PLATE 1 GENSEL, CHALONER and FORBES, Spongiophyton 154 PALAEONTOLOGY, VOLUME 34 The cuticles range from a shiny to dull black or brown colour to ones which are naturally weathered to a red-brown colour. Some lighter coloured portions, especially distal tips, appear waxy. A single specimen may exhibit some portions that are red-brown and others, black; possibly it was specimens of this kind which caught Dawson's eye. Thallus surfaces range from smooth to finely pockmarked to rough in texture ; the latter may have been partially corroded during diagenesis. At the Rte. 132, Quebec locality, the sandstone matrix apparently has imprinted outlines of grains on the thalli, causing a distinctive deeply pockmarked surface pattern. Individual thalli, or regions of a given thallus, vary in the extent to which the cuticle is eroded. Many splits and cracks are present, and more may be induced during SEM treatment. Features we interpret as resulting from erosion include differences in pore outline, the frequent absence of the thinner, here designated lower, part of the thallus, and the depressions located at branch tips. Bulk maceration of specimens in HF provided best results in elucidating thallus morphology. Thalli cleared only after long oxidation in Schulze’s solution. Individual fragments reacted slightly differently to oxidation, suggesting that alteration of the original substance prior to or during fossilization was quite variable. Selected specimens were cleared in Shulze's solution at timed intervals, being examined and photographed with a light microscope at 10, 30, and 60 minute intervals. At the end of 48 hours, no evident destruction of surface features, or changes in pore outline or size, were observed in well-preserved specimens. In poorly preserved thalli (partially oxidized, possibly more strongly compressed), pores became progressively more irregular in outline with prolonged maceration. Extensively cleared thalli may appear ‘cellular’ as a result of differential breakdown of the thick cuticle (PI. 3, figs 5 and 6). Differential erosion, particularly of the inner cuticle surface, is evident in thin sections (PI. 3, fig. 1). After long maceration the cuticles appear spongy in construction (PI. 3, fig. 7). This probably was the basis for Krausel’s original suggestion of a spongy structure (Schwammstructur) of the thallus (see also Chaloner el al. 1974, p. 934). To determine regularity of pore spacing, camera lucida drawings were made of specified areas of selected specimens. Using a customized software program called MEASUR (S. Case, pers. comm.) pore location was digitized, spacing calculated, and mean nearest neighbour determined. No regular pattern was evident. Rock fragments containing numerous thalli and individual thalli isolated by maceration in HF were embedded in plastic, sectioned, and ground thin until transparent. Isolated thalli also were embedded in araldite or glycol methacrylate, sectioned with a microtome, and examined with LM and TEM for ultrastructural detail. Both oxidized and unoxidized specimens were mounted on slides in CMC non-resinous mounting medium, or were glued to glass or cardboard slides with gum tragacanth. Isolated thalli were also mounted on stubs, coated with gold-palladium and examined with an ETEC SEM. Specimens were photographed using a Leitz Aristophot, a Wild Photomacroscope, or a Zeiss photomicroscope. Type and figured specimens are stored in the collections of the Geological Survey of Canada at Ottawa, Ontario and bear the numbers GSC no. 93003-93020. SYSTEMATIC PALAEONTOLOGY Family spongiophytaceae Krausel, 1954 Genus spongiophyton Krausel, 1954 Type species. Spongiophyton lenticulare (Barbosa) Krausel, 1954, p. 206, figs 5-7 of Barbosa, 1949; from the upper Punta Grossa beds, Parana, Brazil. S. minutissimum Krausel, 1954 Plates 1-3; Text-figs 2-6 Type specimens. P. 264/9, Krausel (1954), PI. 28, figs 72 and 73. Original diagnosis. Thallus klein, meist nur wenige mm messend, gabelig gelappt, nrit stark verdickten Radern. Innenbau wie bei Sp. nanum, die zahlreichen Locher aber klein, nadelstichartig, ihre Durchmeser meist 60 bis 100, selten bis 150 /mi, oft quer verbreitert, Oberflachenzellen wabig- vieleckig. GENSEL ET AL. \ CANADIAN EMSIAN SPONGIOPH YTON 155 text-fig. 2. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. a, form 3 thallus, poral side up, with smooth to finely ridged surface and few pores (one at arrow); branching is mostly in one plane; wrinkling and constrictions near lobe apices are interpreted as a result of preservational factors, GSC 93012, x 7. B, form 2 thallus, poral surface up, with many pores (arrows), each pore located in raised area of surface producing bumpy appearance. GSC 93013, x 12. c, form 3 thallus fragment isolated by maceration, showing two dichotomies in two different planes, but with lobes parallel (arrows indicate pores), GSC 93010, x 14. d, form 1 thallus with numerous constrictions, dichotomies in same plane, numerous pores, GSC 93015, x 12. e, form 1 thallus bearing numerous short branches (b) as well as exhibiting major dichotomies, GSC 93016, x 9. 156 PALAEONTOLOGY, VOLUME 34 text-fig. 3. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. a, an isolated thallus with several lobe apices visible, after short HF etch; from sequence near Sawdonia acanthotheca locality (locality M), New Brunswick, GSC 93007, x 6. B, SEM of horizontal thallus lobes; note breakdown of apex to right and absence of spores or other cell masses in depression, x 46. c, LM of thallus with two pairs of lobes, intermediate between short vertical branches and ‘normal’ lobes, departing from poral surface; lobe apices collapsed, GSC 93017, x 10. d and F, SEM of thalli with short vertical branches departing from poral surface; apical depressions are probably a result of collapse of thinner cuticle in that area; d, x 15; f, x 11. e, SEM of thallus with very pronounced constrictions and possible short vertical branches near left, pores, x 12. Emended diagnosis. Thallus cylindrical, originally circular or elliptical in cross section and at least 2 cm long. Width of thalli 02-5-5 mm. Thalli may exhibit constrictions along their length. Thalli branch dichotomously several times, with most lobes 3-10 mm long and with rounded apices. Short erect branches (1-2 mm long) occur on poral surface of some thalli. Pores extend through cuticle mostly on one surface, this being 2-4 times thicker than aporal surface (75-250 //m vs 30-60 //m), the thicker cuticle extending around the margins onto the edge of the aporal surface. Poral and aporal surfaces smooth, aporal surface often longitudinally folded. Inner surfaces of cuticle may GENSEL ETAL . CANADIAN EMSIAN SPONGIOPH YTON 157 text-fig. 4 a-q. Camera lucida drawings of various Spongiophyton minutissimum thalli, showing differences in extent and angle of branching, x 6. 158 PALAEONTOLOGY, VOLUME 34 retain rectangular cell outlines, 20-43 //m long and 9-12 p m wide, although often degraded and vermiform in appearance. Pores circular to oval, randomly spaced, 22-5 x 9 pm to 99 x 90 pm in diameter, with vertical, fissured or (occasionally) bevelled edges. Description. The plant fossils consist of the very thick cuticles of dorsiventral, apparently elliptical thalli which dichotomize at various intervals and on the surface of which pores occur (PI. 1, figs 6 and 7; Text-figs 2—4). Thalli vary mainly in size, pore density and location, extent of branching, and surface features. Three categories (forms 1-3) of thalli are recognized: (1) the majority are smooth-surfaced, with a number of pores located on the upper surface, many branches, and ‘constrictions’ (PI. 1, figs 6 and 7; Text-fig. 2d, e); (2) some are smooth- surfaced except that each pore occurs in a small projection resulting in an overall bumpy appearance (Text- fig. 2b); and (3) some thalli are smooth but longitudinally ridged, have very few pores located along lateral margins, and bear short vertical branches (Text-figs 2a, c and 3c). We presently regard this variability in surface topography, pore distribution, and branching type to be intra-specific, perhaps resulting from different parts of a given plant being represented, populational differences, and/or preservational differences. We thus refer all of the specimens to a single species. The dorsiventral thalli are usually incompletely preserved, being up to 10 mm long and ranging from 0-3— 2-5 mm wide (PI. 1 ; Text-figs 2 and 3). The thick amorphous cuticle is smooth externally (Text-figs 2 and 3) and rough internally (Text-fig. 5c, d). One surface of the cuticle is thicker than the other (PI. 3, figs 1 and 3); we interpret the thicker surface, on which pores and branches occur, to be the dorsal surface of a more or less flattened horizontal thallus. Pores may also occur near the margins on both upper and lower surfaces of some (especially form 3) thalli. The thalli branch dichotomously at least six times at 0-2-3 mm intervals, being dense on some specimens and very sparse on others. Branching results in formation of lobes extending in the same plane as the original thallus (PI. 1, fig. 7; Text-figs 2a, b and 3c) or in lobes lying one on top of the other but still with their long axes parallel (PI. 1, fig. 6; Text-fig. 2c). These may curve upwards, downwards or laterally (in relation to the presumed horizontal position of the main thallus). It is not unusual to observe three to four levels of thallus lobes belonging to one specimen in the stacks of fossils preserved at the Atholville locality. Additionally, branching may result in one horizontal and one short vertical lobe at right angles to each other (Text-figs 2e and 3d, f). Constrictions occur within the lobes of some thalli resulting in a sausage-string type of appearance (PI. 1, fig. 6; Text-figs 2d, e and 3e. The region between some constrictions almost resembles very short upright branches. The upright branches are 1-2 mm tall and occur singly or in 2 rows (Text-fig. 3d, f). While upright branches occur on all forms of thalli, they are most abundant on form 3, being located along a centraJ ridge area. Most exhibit pores (PI. 2, fig. 2). The branches are narrower towards their base and swell or flare distally (Text-fig. 3a, c, d, f). Some of these branches terminate in rounded apices, often with a slight depression (PI. 2, fig. 1 ; Text-fig. 3a). A few exhibit small protrusions extending from the apex (PI. 2, figs 3 and 4) while others have at their tips a deep cup-like depression which usually is irregular in outline (PI. 2, fig. 2). The interior of the cup revealed no organization such as spores or vegetative propagules. Their structure suggests the depressions formed as a result of collapse and breakdown of cuticle at the branch apex. Horizontal branch tips exhibit similar depressions (Text-fig. 3b). Circular to oval shaped pores occur on the presumed upper (and thicker) surface of the first two forms of thalli and along the margins of the third form of thallus (PI. 1, fig. 6; PI. 3, figs 5 and 6; Text-fig. 2b-d). They are variably spaced, from 3-5 //m to 350 pm apart. Average distance between pores on selected specimens are 30, 53, 75, 178, 222 pm. The pores range in size from 22-5 pm long by 9 pm wide to 99 pm long by 90 pm wide. SEM study shows the pore margin often to consist of indented fissures or rounded outlines (PI. 2, figs 4 and 6; Text-fig. 5b). Only a few exhibit a bevelled edge (PI. 2, fig. 5) as has been described in Spongiophyton nanum by Chaloner et al. (1974). EXPLANATION OF PLATE 2 Figs 1-5. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. 1 and 2, SEM of apical region of vertical branches showing various degrees of collapse; 1, x 40; 2, x 31. 3 and 4, general view and detail of short vertical branches with protrusion at apex - also appearing somewhat degraded ; note occurrence of pores on branches (arrows); 3, x38; 4, x 120. 5, SEM of pore with bevelled edges, some evidence of internal surface of cuticle, x 1100. 6, SEM of pore with irregular margin, x 120. PLATE 2 GENSEL, CEIALONER and FORBES, Spongiophyton 160 PALAEONTOLOGY, VOLUME 34 text-fig. 5. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. a, interior contents of a thallus, GSC 93014, x 14. b, LM of pore on thallus cleared with Schulze’s solution until nearly translucent; region around pore is darker than rest, GSC 93020, x 65. c, SEM of inner cuticle surface, poral side, showing outlines of rectangular cells; pore in centre, x 1 10. d, SEM of inner cuticle surface, poral surface, appearing vermiform, probably a result of degradation or borings; pore in centre, x 220. E, ventral surface of extensively cleared thallus with characteristic longitudinal folds, GSC 93018, x 3 1 . f, SEM of inner cuticle surface showing rectangular cell outlines; this probably reflects type of cell construction immediately below the cuticle, x 550. GENSEL ET AL.: CANADIAN EMSIAN SPONGIOPH YTON 161 Thickness of the poral, presumed upper surface is 75-250 pm, and that of the aporal one is about 30-60 pm. However, the form 3 thalli exhibit poral and aporal surfaces of more equal thickness. The thinner lower thallus surface often is partly broken down or may be entirely absent (Text-fig. 5a). Intact lower surfaces have been observed mostly near the tips of some lobes and rarely on more completely preserved specimens. Where present, the lower surface exhibits longitudinal ridges and appears fragile and rather wrinkled except at the margins where it is transitional to the thicker, upper surface (Text-fig. 5e). Inside the thallus occurs a thin granular layer of material which is usually hght-brown in colour (Text-fig. 5a). We do not believe this is rock matrix (left after HF treatment) but is a remnant of the internal contents of the thalli. SEM of a cut transverse section of the thallus cuticle end-on shows it to be amorphous (PI. 3, fig. 3). Thin sections examined with LM and TEM show an absence of internal structure in the cuticle, except for minute structures perpendicular to the outer surface interpreted as borings or cracks (PI. 3, figs I, 2, 4). Particularly interesting is the absence on all but a few specimens of regular ridges or pegs corresponding to depressions between epidermal cells as usually occurs on vascular plant cuticles or in other species of Spongiophyton. However irregularities of the inner surface of the cuticle may represent the position of anticlinal walls in life (PI. 3, fig. 1). Possibly many such ridges were lost or obscured through erosion of the inner cuticle surface. Examination of the interior cuticle surface has revealed few with cellular patterns (Text-fig. 5c, f). More frequently they exhibit a vermiform pattern (Text-fig. 5d) which compares well with the ‘borings’ described by Chaloner et al. (1974) for S. nanum from Ghana. Cleared thalli may exhibit an apparent cell-like pattern (PI. 3, figs 5 and 6), especially in photographs. Close examination suggests these result from cracks caused by differential breakdown of the cuticle after prolonged oxidation. We term this a pseudocellular pattern and regard it as different from the cell outlines preserved on some inner cuticle surfaces. Elemental analysis of two different specimens show element ratios similar to the Ghana Spongiophyton specimens (Table 2). table 2. Elemental percentage composition of Spongiophyton from Canada and Ghana. The difference of the sum from 100 is probably accounted for by oxygen. N C H S S. minutissimum , Canada 118 74-88 8-08 0 S. minutissimum, Canada 103 6901 7-55 0 S. nanum, Ghana 2-70 78-40 8-40 — COMPARISONS AND DISCUSSION The thalli are clearly referable to the genus Spongiophyton Krausel as emended by Chaloner et al. (1974). Characters considered diagnostic of the genus by the latter authors, and exhibited by the Canadian material, are: a tubular thallus with cuticular covering, dichotomous or sub-dichotomous branching and rounded apices; cuticle with internal cellular reticulum and circular-fusiform pores largely confined to one surface of the thallus. The Canadian specimens are older and morphologically more diverse than other undoubted Spongiophyton specimens. The thalli branch much more frequently and in more than one plane, and also bear more short vertical branches than previously known. Our data also confirm the interpretation of Chaloner et al. (1974) that some features considered diagnostic by Krausel (dark bodies on surface, the ‘spongy’ or hyphal pattern) are in fact the result of degradation, either during preservation or the clearing process, of the thick cuticles. A pseudocellular pattern may result from cuticular breakdown in the Canadian specimens. The constrictions common in the Canadian thalli probably are a result of preservational factors or may reflect environmental fluctuations. Species of Spongiophyton The Canadian specimens are most similar to Krausel’s species S. minutissimum , based on consideration of his few illustrations and brief description and on study of his figured specimens. Many extensively cleared thallus fragments from Canada are identical to 5. minutissimum in 162 PALAEONTOLOGY, VOLUME 34 exhibiting thickened margins, a character considered by Krausel as distinctive for that species. Thallus appearance, cuticle thickness, and pore shape, size, and density of the Canadian material corresponds very closely to S. minutissimum. Two possible differences exist - maximum thallus width in the type material (up to 5 mm) exceeds that of the Canadian fossils and the dark bodies described by Krausel for S. minutissimum are not found on the Canadian specimens. The more extensive preservation and greater abundance of specimens from Canada provides some characters not available from the type material, limiting further comparison. In the absence of major characters separating them, and indeed with strong evidence supporting their identity, it seems reasonable to expand the concept of a known species rather than create a new one. Differences between the Canadian specimens and other species of Spongiophyton include pore morphology and size, thallus size, and cuticle thickness. S. nanum and S. lenticulare are the best known species. The Canadian specimens differ from S. nanum in their smaller pores that mostly lack a bevelled margin. Branching is more profuse in the Canadian specimens than in S. nanum where only a few dichotomies or vertical branches have been recorded. Thallus diameter is half that of S. nanum. Cuticle thickness of the Canadian specimens (up to 250 pm) is much greater than that of S. nanum (60-80 /mi) and both exceed the thickness of most vascular plant cuticles. Similarly, the Canadian specimens differ from S. lenticulare in pore morphology, those of the latter species being elongate, slit-like structures with folded edges of cuticle extending to the outside, and their thicker cuticle. The internal cellular pattern of S. lenticulare consists of more elongate cells than occur in either S. nanum or the Canadian material. Krausel and Venkatachala (1966) placed S. hirsutum in Aculeophyton, because of its hairlike papillae. Krausel’s species S. articulation is based on broken cuticle fragments which exhibit a very pronounced longitudinal striped pattern and transverse corrugations. These remains bear some resemblance to very over-macerated thalli from Canada, but are too poorly known for further comparison. Notably absent in all of these species is any evidence of reproductive structures. Chaloner et al. (1974) suggested that the short vertical branches present on S. nanum thalli were perhaps sites of reproductive organs, but no conclusive evidence was obtained. Many short vertical branches of the Canadian specimens which exhibited depressions were examined for evidence of spores or other possible reproductive structures. Occasionally a mesh-work of material was present in the depressions, but more commonly only fissures were observed along the margins. Both types of structure are interpreted to have resulted from degradation of cuticle in apical regions. Other identifications of thalli as Spongiophyton are less certain. Boureau and Pons (1973) assigned thalli from Bolivia to S. lenticulare. Pore shape agrees with that of Krausel’s S. lenticulare but some other features are problematical, recalling protuberances termed ‘capsules’ (see below) in Orestovia and other genera by Istchenko and Istchenko (1981). The Canadian specimens differ, not only in pore outline, but also in lacking dark round bodies and any evidence of internal ‘hyphal ramifications' as described for the Bolivian specimens. The latter should be compared more closely EXPLANATION OF PLATE 3 Figs 1-7. Spongiophyton minutissimum Krausel from New Brunswick and Gaspe. 1, ground thin section of thallus in rock matrix, showing much thicker poral and thinner aporal surfaces ; dark contents in middle may be remains of inner cells and may correspond to the lighter material seen in Text-fig. 5a, x 54. 2, TEM of cuticle showing absence of internal structure, except for possible borings, x 700. 3, SEM of transverse cut surface end-on showing differential thickness of poral and aporal surfaces and absence of structure other than borings, x 280. 4, LM of section of cuticle showing borings, x 300. 5-7, cleared thalli showing breakdown of cuticle producing a pseudocellular pattern; 5, cleared thallus fragment with thicker margins as illustrated by Krausel as typical of this species; some evidence of pseudocellular pattern at arrow, shown enlarged in fig. 6, GSC 93019, x 31 ; 6, detail of pseudocellular pattern of specimen in fig. 5, can be emphasized by manipulating lighting of microscope; several pores visible, apparently ‘ringed’ by pseudocellular pattern, GSC 93019, x 65. 7, over-cleared cuticle with spongy appearance similar to several illustrated by Krausel, x 39. PLATE 3 GENSEL, CHALONER and FORBES, Spongiophyton 164 PALAEONTOLOGY, VOLUME 34 to the several thalloid types described by Istchenko and Istchenko (1981) from the Voronezh anticline, USSR. As noted earlier, the specimens described as Spongiophyton by Zdebska (1978) may in fact represent other taxa. Although Zdebska’s species 1 bears a superficial resemblance to Spongiophyton thalli, no details of pore type or cellular construction are evident. Species 2 consists of fragmentary cuticles with isodiametric cell outlines and pores, suggesting a filamentous rather than parenchymatous organization. Neither type shows any indication of being part of a tubular thallus like Spongiophyton. Edwards (1982) suggested that these fragments resemble some cuticles of Nemato thallus. Other putative spongiophytes Other genera are allied with Spongiophyton in the family Spongiophytaceae (Table 1) because they exhibit a thalloid construction with thick, resistant, sometimes flexible cuticles and lack definitive evidence of reproductive structures. Comparison of these taxa is hampered somewhat because interpretation of particular morphological structures varies, depending on the worker involved or the time of publication and corresponding knowledge of Devonian plant diversity. It also is extremely difficult to interpret structural detail on opaque cuticles of comparatively undifferentiated organisms from photographs and descriptions. The cuticles of these other taxa may also have been strongly affected by taphonomic factors and preparation techniques. Thus the family may not be as coherent as it appears. The broad, ribbon-like cuticularized axes of Orestovia are generally similar to Spongiophyton but longer, wider, less frequently branched, radially symmetrical, and with a thinner cuticle. The outer surface is smooth or covered with some form of tiny emergence (depending on author). Circular pores with slightly raised margins, often bordered by several concentric layers of mostly isodiametric ‘cells’, occur randomly. Extraporal regions bear the outline of elongate-rectangular cells. The pores and associated structures are interpreted by some workers as stomata (Ergolskaya 1934, 1936; Krassilov 1981), and by others as reproductive structures (Krausel and Venkatachala 1966, Istchenko and Istchenko 1981). Krassilov (1981) further reported the presence of conducting cells with thickened wall patterns in Orestovia , suggesting it may be a vascular plant. Our preparations of thalli, conforming to Ergolskaya’s O. petzii from the Barzas coal, support some of his conclusions concerning stomata. Krassilov’s specimen appears papillate (= O. devonica of Ergolskaya), whereas the specimens available to us are smooth. Obviously, further documentation is needed to resolve several attributes of the genus. Despite this, Spongiophyton, including the Canadian material, can be distinguished from Orestovia in gross thallus organization, symmetry, and details of pore construction. The genus Aculeophyton was established by Krausel and Venkatachala (1966) for cuticular fragments of thalli from western Siberia, originally placed by Ergolskaya (1934, 1936) in Orestovia devonica. The genus differs from Orestovia mainly in the presence of papillae, conical in A. sibirica and hair-like in A. hirsutum. Krassilov considered that other characters outweighed the presence of papillae and that Aculeophyton and Orestovia are synonymous. Istchenko and Istchenko (1981) described several new genera and species of thalloid plants from the Lower Devonian of the Voronezh region, USSR, placing some in the Spongiophytaceae and some in a second family, the Bitelariaceae. Bitelarian cuticles reflect distinct ‘cell’ patterns interpreted by the Istchenkos as a meristoderm (without a cuticle) and by Johnson and Gensel (1989) as a cuticular epithelium. A number of other characters such as branching pattern and presence of vascular tissue in Bitelaria, further distinguish bitelarians from all thalli placed in the Spongiophytaceae, as summarized in Johnson (1989) and Johnson and Gensel (1987, 1989). Istchenko and Istchenko (1981) assigned the Voronezh fossils to several genera ( Orestovia , Orestovites, Voronejipliyton, Rhytidophyton , Bitelaria and Donotela ) relating them to the algae. They interpreted the protruding round pores found on thalli of the first four genera as reproductive structures (termed capsules) reminiscent of conceptacles or nemathecia, as found in brown and red algae. When mature, each structure supposedly opened and released its contents, leaving behind a pore. The same structures in Orestovia appear to us very like sunken stomata or in some cases like GENSEL ET AL.: CANADIAN EMSIAN SPONGIOPH YTON 165 the dark bodies or ‘grossorgane’ of Krausel and Venkatachala (1966). No evidence of ‘capsules' or stomata exists for the Canadian Spongiophyton. Rhytidophyton superficially seems most similar to Spongiophyton but apparently consists of radially symmetrical tubular thalli with pronounced vertical folding. The thalli are up to 0-6 cm wide and dichotomize, with the resultant branches forming a U-shaped pattern. One or more hemispherical protuberances 1 -5—3-5 mm in diameter occur just below the dichotomy, or singly elsewhere on the thallus, which when lost form large pores. Capsules (or the oval apertures that remain after their disintegration) are irregularly distributed on the thallus surface. The thallus is interpreted as consisting of an outer cuticle, a middle fibrous zone, and an inner region of coal. Voronejiphyton Istchenko and Istchenko (1981), based on a few specimens, is very similar to Rhytidophyton , apparently differing in exhibiting occasional longitudinal ridges and internal wall thickening. In addition to lacking ‘capsules’, Spongiophyton lacks obvious folds and exhibits a single-layered thick cuticle. Its short vertical branches usually are more extensively developed or larger than those of Rhytidophyton and are not located at points of branching. Orestovites is similar to Orestovia , differing only in rather minor features, such as the presence of several cuticle layers, hemispherical structures, major cracks in the cuticle and an irregular pattern of cells on the inner cuticle surface. Other thalloid Devonian plants Chaloner et al. (1974) compared Spongiophyton with several other Devonian plants of thalloid construction but with one or more features attributed to land plants, e.g. Prototaxites , Parka , Protosalvinia and Nematothallus. They all differ in apparently lacking the type of tubular cuticularized thallus plus pores seen in Spongiophyton , Orestovia and Auculeophyton and detailed comparison of most of them is unnecessary. It is intriguing to note, however, that the vertical branches of the Canadian Spongiophyton resemble the proposed reconstruction of Protosalvinia by Niklas and Phillips (1976) even though many differences between the two taxa exist, including their postulated mode of growth (Niklas and Chaloner 1976). Reproductive structures are known for Protosalvinia. The several types of isolated cuticles attributed to Nematothallus (sensu Edwards 1982) or Cosmochlaena (Edwards 1986) have been compared at times with Spongiophyton or other spongiophytes. These latter plants differ from Nematothallus , as stated by Edwards (1982), in their tubular construction and apparently parenchymatous cell structure. Further, the isolated nematophyte cuticles are not as thick as those of Spongiophyton and some related taxa. The original concept of Nematothallus was of a system of tubes covered on the upper, and perhaps lower, surface by a cuticle and possibly bearing spores among the tubes. Edwards suggested the nearly isodiametric cell outlines of the associated cuticles represented outlines of filament tips. This taxon, and Prototaxites , served as the basis for Lang’s Nematophytales (Lang 1937). While he suggested other taxa may be included in that group, later research has shown several of them to be differently constructed. We agree with Edwards (1982) that Spongiophyton probably had a parenchymatous organization, which would contrast strongly with the above taxa. We also believe the term nematophyte should be restricted to plants of tubular (filamentous) construction as originally proposed by Lang, thus excluding Protosalvinia , the Spongiophytaceae, and probably several other enigmatic early land plants (Strother 1988). Proposed growth habit The extensively preserved Canadian Spongiophyton provides a basis for modifying concepts of its growth habit. The dorsiventral, probably cross-sectionally elliptical, tubular nature of thallus lobes is confirmed. Profuse branching, both in the same plane and at right angles, produces a growth form recalling that of some thallose liverworts such as Conocephalum or Marchantia. This extensive branching is not consistent with or feasible to the growth model proposed by Niklas and Chaloner (1976) based on studies of S’, nanum. Whether S. minutissimum actually grew differently from 5. nanum is unclear; certainly data concerning multidimensional branching were sparse at the time the model was proposed. 166 PALAEONTOLOGY, VOLUME 34 These specimens are found in fluvial sediments, some appearing more extensively transported than others. Their habitat may have been similar to that of many extant thallose liverworts - stream or pond margins on a flood plain. They probably formed mats or stands several centimetres to tens of centimetres broad. The stacks of thalli found at the Athoville locality document that several levels of branches may occur on a single organism, as if the older portions were partially buried and newly produced ones grew upwards (towards light?). Where thalli have been sectioned in situ in the matrix, the thicker (presumably upper) surface is commonly uppermost in the rock, but not consistently enough to support the possibility of the plants growing within the environment of deposition. The short vertical branches appear different from other thallus lobes, but perhaps they elongated and became more parallel to the main thallus when older, as suggested by the specimen illustrated in Text-figures 2e and 4k, n. Variation in branching and thallus orientations are depicted in the reconstruction in Text-figure 6. text-fig. 6. Proposed reconstruction of Spongiophyton minutissumum plants. The question of affinities Despite the abundance and variety of specimens of S. minutissimum, many questions remain particularly on the nature of its reproductive structures and whether or not it possessed conducting tissues. This has important bearing on its affinities - particularly in relation to whether it represents an ‘algal’ grade of organization or one more comparable with embryophytes. It appears to be a non-vascular plant with a resistant cuticle and pores. A resistant cuticle is generally regarded as an adaptation to a terrestrial habitat, since only terrestrial higher plants ( = embryophytes) are known to possess one. Mishler and Churchill (1984, GENSEL ET AL.: CANADIAN EMSIAN SPONGIOPH YTON 167 1985) and others have postulated that a cutin-containing cuticle is a synapomorphy of the embryophyte clade. This might be further tested by chemical analysis of the cuticles of several plant types, including representatives of the charophyte-embryophyte clade, representatives of other algal clades that possess an outer covering, and of the enigmatic types discussed above. If lipid-rich (cutin-containing) cuticles occur only in the embryophytes, and if Spongiophyton cuticles have a similar composition, then one could place it in that lineage. The same might be true of Orestovia or other taxa of enigmatic affinity. Thus, although the thalloid, presumed parenchymatous construction of Spongiophyton has led workers to suggest it is related to algae, one could also envision it representing an algal-derived form that had not yet attained the grade of complexity of bryophytes or vascular plants. Its affinity to the charophyte-embryophyte clade sensu Mishler and Churchill (1984, 1985) remains uncertain as it is possible that several extinct lineages, derived from any of several algal clades, may have become adapted for terrestrial existence and possessed a resistant cuticle. More fossils of these enigmatic types, and careful analysis of all aspects of morphology and chemistry, might address these questions. Documenting all combinations of adaptations to a terrestrial existence among Silurian- Devonian plants promises to reveal more fully the intricate story of invasion of the land by plants and of diversity of lineages at that time. Acknowledgements. The authors thank Susan Whitfield, Staff Arist, Biology Department, University of North Carolina at Chapel Hill, for making the line drawings and reconstruction and Mr Graham Lawes, Biology Department, Royal Holloway and Bedford New College for help in preparing sections and TEM photos. Appreciation is extended to Dr Robert Carroll and Ms. Delice Allison for facilitating the loan of specimens from the Dawson collection at the Redpath Museum, Montreal, Canada. This research was supported by NSF grants DEB 80-1 1705, BSR 83-15670, and BSR 8800432 to Patricia G. Gensel. REFERENCES alcock, f. J. 1935. Geology of Chaleur Bay region. Canada Department of Mines , Geological Survey Memoires. 183 pp. barbosa, o. 1949. Vegetais fosseis do Devoniano do Brasil e da Bolivia. Mineracao e Metalurgia , 14, 81-84. boucot, a. j. 1985. The relevance of biogeography to palaeogeographical reconstructions. 79-80. In chaloner, w. G. and Lawson, j. d. (eds). Evolution and environment in the Late Silurian and Early Devonian. Philosophical Transactions of the Royal Society of London. Series B , 309, 1-342. boureau, m. E. and pons, d. 1973. Sur des empreintes vegetales devoniennes du Sud de la Bolivie. Compte Rendus de T Academic Sciences de Paris , Sene D , 276, 2151-2153. chaloner, w. G., mensah, m. k. and crane, m. d. 1974. Non-vascular land plants from the Devonian of Ghana. Palaeontology , 17, 925-947. dineley, d. l. and williams, b. p. j. 1968. The Devonian continental rocks of the lower Restigouche River, Quebec. Canadian Journal of Earth Sciences , 5, 945-953. Edwards, d. 1982. Fragmentary non-vascular plant microfossils from the late Silurian of Wales. Botanical Journal of the Linnean Society, 84, 223-256. — 1986. Dispersed cuticles of putative non-vascular plants from the Lower Devonian of Britain. Botanical Journal of the Linnean Society, 93, 259-275. ergolskaya, z. v. 1934. New data on the origin of coal. Chemistry of hard fuel , 5, 33-39. [In Russian], — 1936. Petrographical examination of the Barzas coals, Kuznetsk Basin. Transactions of the Central Geological and Prospecting Institute, 70, 3-54. [In Russian], gensel, p. G. 1982. On the contributions of Sir J. W. Dawson to the study of early land plants (Devonian) and current ideas concerning their nature, diversity, and evolutionary relationships. Proceedings, 3rd North American Paleontological Convention, 1, 199-204. — and Andrews, h. n. 1984. Plant life in the Devonian. Praeger Publishers, New York, 380 pp. hanic, l. a. and craigie, j. s. 1969. Studies on the algal cuticle. Journal of Phycology, 5, 89-102. ISTCHENK.O, t. a. and istchenko, a. a. 1981. Middle Devonian flora from the Voronezh Anticline. Akademia Nauk Ukraniskoi SSR, Kiev, 112 pp. [In Russian], Johnson, n. G. 1989. Bitelaria, an early Devonian vascular plant and its significance relative to the origin and radiation of early land plants. Unpublished PhD thesis. University of North Carolina. 145 pp. 168 PALAEONTOLOGY. VOLUME 34 and gensel, p. G. 1987. A new occurrence of Bitelaria from the Emsian of New Brunswick, Canada, and its significance relative to early land plant evolution. Abstracts '. Botanical Society of America, 74, 685. - 1989. The early Devonian (Emsian) land plant Bitelaria , previously allied with thallophytes, is a tracheophyte with a specialized cuticular structure. Abstracts : Botanical Society of America, 76, 167. krassilov, v. 1981. Orestovia and the origin of vascular plants. Lethaia, 14, 234-250. krausel. r. 1954. Spongiopbyton nov. gen. (Tallophita) und Haplostigma Seward (Pteridophyta) im Unter- Devon von Parana. 195-210. In Paleontologia do Parana. Vol. Comemorativo do 1 Centenario do Estado do Parana, Publicado pela Comissao de Comemoracoes do Centenario do Parana. Curitiba, Brasil. — and venkatachala, b. s. 1966. Devonische Spongiophytaceen aus Ost- und West-Asien. Senckenbergiana Lethaea , 47. 215-251. lang, w. h. 1937. On the plant remains from the Downtonian of England and Wales. Philosophical Transactions of the Royal Society of London, Series B , 227, 245-291. livermore, r. a., smith, a. G. and BRiDEN, J. c. 1985. Palaeomagnetic constraints on the distribution of continents in the late Silurian and early Devonian. 29-56. In chaloner, w. g. and lawson, j. d. (eds). Evolution and Environment in the Late Silurian and Early Devonian. Philosophical Transactions of the Royal Society of London, Series B, 309, 1-342. mcgerrigle, H. w. 1950. The geology of eastern Gaspe. Department of Mines, Geological Report , 35, 1-168. mcgregor, d. c. 1973. Lower and Middle Devonian spores of eastern Gaspe. L Systematics. Palaeontographica, Abteilung B , 142, 1-77. 1977. Lower and Middle Devonian spores of eastern Gaspe, Canada. II. Biostratigraphy. Palaeontographica, Abteilung B , 163, 111-142. mishler, b. D. and churchill. s. P. 1984. A cladistic approach to the phylogeny of the ‘bryophytes’. Brittonia , 36, 406-424. 1985. Transition to a land flora: phylogenetic relationships of the green algae and bryophytes. Cladistics, 1. 305-328. niklas, k. j. and chaloner, w. G. 1976. Simulations of the ontogeny of Spongiopbyton, a Devonian plant. Annals of Botany, 40, 1-1 1. and Phillips. T. L. 1976. Morphology of Protosalvinia from the Upper Devonian of Ohio and Kentucky. American Journal of Botany, 63, 9-29. penh allow, d. p. 1889. (Introductory notes by Sir J. W. Dawson). On Nematophyton and allied forms from the Devonian (Erian) of Gaspe and Bay des Chaleurs. Transactions of the Royal Society of Canada, 6, 27-47. scotese, c. r., van der voo, r. and barratt, s. F. 1985. Silurian and Devonian base maps. 57-77. In chaloner, w. G. and lawson, j. d. (eds). Evolution and environment in the Late Silurian and Early Devonian. Philosophical Transactions of the Royal Society of London, Series B, 309, 1-342. Strother, p. k.. 1988. New species of Nematothallus from the Silurian Bloomsburg Formation of Pennsylvania. Journal of Paleontology, 62, 967-982. sommer. F. w. 1959. Spongiophylates, una nova ordem de talofitos, de classe fossile dos algomycetes. Ana is de la Academia brasileira de Ciencias, 31, 43. zdebska, d. 1978. On Spongiopbyton from the Lower Devonian of Poland. Acta Palaeobotanica, 109, 13-20. P. G. GENSEL Department of Biology CB no. 3280 University of North Carolina Chapel Hill N.C. 27599-3280 W. G. CHALONER Department of Biology School of Life Sciences Royal Holloway and Bedford New College Egham. Surrey TW20 0EX W. H. FORBES Department of Geology University of Maine at Presque Isle Presque Isle Maine 04769 Typescript received 20 October 1988 Revised typescript received 12 February 1990 TEUTHID CEPHALOPODS FROM THE UPPER JURASSIC OF ANTARCTICA by PETER DOYLE Abstract. Two teuthid cephalopods, Trachyteuthis cf. hastiformis (Riippell) and muensterellid gen. et sp. nov., are described from the Nordenskjold Formation (Upper Jurassic) of the northeastern Antarctic Peninsula. These specimens, the only recorded teuthids from Gondwana, are closely related to European species and suggest a more widespread distribution in the Late Jurassic than was previously known. Dur ing the Antarctic summer of 1987-1988 two fossil teuthid specimens were collected by the author from the Nordenskjold Formation, a late Jurassic-early Cretaceous black shale sequence exposed in the northeastern Antarctic Peninsula (Text-fig. 1 ). These specimens represent the only known teuthids from any of the Gondwana continents, and as such are of importance to our understanding of teuthid distribution. Teuthids ( = Vampyromorpha of Bandel and Leich 1986 and Engeser 19886) are rare fossils given the relative abundance of other fossil cephalopods (ammonites and belemnites). A survey of the teuthid fossil record shows that these cephalopods are most commonly preserved in fine-grained sediments deposited under anoxic or otherwise restricted conditions, and the present specimens are no exception. Lower Jurassic specimens are commonest, especially from the widespread Toarcian black shales of Europe (Posidonienschiefer, Jet Rock, e.g. see Riegraf et al. 1984; Engeser 19886; and Doyle 1990 for summaries), and North America (Fernie Formation, e.g. Hall 1985; Hall and Neuman 1989). Exceptionally well preserved specimens are also known from the Middle Jurassic (Callovian) Oxford Clay of England (e.g. Donovan 1983), and similar-aged anoxic sediments in the Ardeche, France (Fischer and Riou 1982). Upper Jurassic teuthids are well represented in the Solenhofen Limestone of southern West Germany (Crick 1896; Bandel and Leich 1986; Engeser 1986) and in the Kimmeridge Clay of England (Owen 1855; Hewitt and Wignall 1988). Cretaceous restricted facies have also yielded teuthids: from the Lower Aptian ‘Tock’ of northern West Germany (Engeser and Refiner 1985), the Santonian Fish Bed of the Lebanon (Woodward 1883; Roger 1946; Engeser and Reitner 1986), and the Upper Cretaceous Niobrara Formation (Kansas) and Pierre Shale (Manitoba) of North America (e.g. Miller and Walker 1968; Nicholls and Isaak 1987) . The relative paucity of teuthid specimens has lead to an anomalous distribution pattern. Thus, apart from the specimens found in anoxic sediments in the United States, Cuba (Schevill 1950), the Lebanon, and the Cape Verde Islands, West Africa (Reitner and Engeser 1982), the majority of specimens are from Europe (see Engeser 19886). Prior to the present study, teuthids were unknown from Gondwana, as the only record from Queensland, Australia (Moore 1870) has been found to be an indeterminate bivalve fragment (Engeser and Phillips 1986). The purpose of this paper is to document the new record and discuss its implications for palaeobiogeography. GEOLOGICAL SETTING From late Jurassic to early Tertiary times the northern Antarctic Peninsula was an active volcanic arc formed by the southeastward subduction of the proto-Pacific plate. During subduction, a 5-6 km thick sedimentary sequence was deposited in a retro-arc basin (the Larsen Basin) to the east of the arc. The Nordenskjold Formation is a distinctive sequence of air-fall ashes and black | Palaeontology, Vol. 34, Part 1, 1991, pp. 169—178-1 © The Palaeontological Association 170 PALAEONTOLOGY, VOLUME 34 text-fig. 1. Locality map for the northeastern Antarctic Peninsula, showing the distribution of known Nordenskjold Formation exposures. The specimens described below were collected from the type locality at Longing Gap. mudstones of late Jurassic to early Cretaceous age which is thought to form the base of the Larsen Basin succession (Macdonald et al. 1988). The Nordenskjold Formation is exposed at five localities along the northeastern coast of the Antarctic Peninsula (Farquharson 1983; Text-fig. 1), and its stratigraphy has recently been revised by Whitham and Doyle (1989). Two members are recognized within the Nordenskjold Formation at Longing Gap, the type locality for the formation (Text-fig. 1). The Longing Member ranges in age from Kimmeridgian to Tithonian and is dominated by parallel-laminated black mudstones with subordinate, thin ash layers. The Ameghino Member ranges in age from Tithonian to Berriasian and is characterized by structureless mudstones and thicker ash layers (Whitham and Doyle 1989). A detailed sedimentological study of the Nordenskjold Formation is currently being carried out by Dr A. G. Whitham (British Antarctic Survey). Both teuthid specimens were obtained from near the top of the Longing Member at Longing Gap. They were associated with a fauna consisting of the ammonites Virgatosphinctes spp. and Lithacoceras sp., and the bivalves Retroceramus spp. and Arctotis sp., of Tithonian age (Whitham and Doyle 1989). Sedimentological (parallel lamination, etc.) and palaeoecological (low faunal diversity, a lack of true benthos and trace fauna) indices show that the Longing Member was deposited under low oxygen, anaerobic to episodically dysaerobic conditions (Doyle and Whitham in press). Although the Longing Member fauna had a primarily pelagic or pseudoplanktonic mode of life, the teuthids were collected from an interval with some benthic colonization, though lacking DOYLE: ANTARCTIC TEUTHID CEPHALOPODS 171 bioturbation, suggesting dysaerobic rather than the anaerobic zone conditions characteristic of the lower part of the member (Doyle and Whitham in press). SYSTEMATIC PALAEONTOLOGY The terminology used below is discussed in detail in Jeletzky (1966), and the classification largely follows that of Engeser ( 19886) (see discussion below). Both specimens are housed in the collections of the British Antarctic Survey (BAS) in Cambridge. Comparative material was examined in the British Museum (Natural History) (BMNH), London. Annotation of synonymy lists follows the convention of Matthews (1973). Subclas coleoidea Bather, 1888 Order teuthida Naef, 1916 Remarks. Jeletzky (1966) employed the order Teuthida Naef, 1916 for all known fossil squid. However, Bandel and Leich (1986) studied in detail specimens of the Solenhofen teuthids Leptoteuthis, Plesioteuthis and Trachyteuthis , and concluded that they possessed only eight arms, linked by basal webs, and were therefore most closely related to the Recent cephalopod V ampyroteuthis . This lead to the adoption of the order Vampyromorpha Robson, 1929 for all fossil ‘teuthids’ by Berthold and Engeser (1987), Engeser and Bandel (1988) and Engeser (19886), and the contention that the fossil ‘teuthids’ were not directly ancestral to the Recent Teuthida. The more conservative usage of the order Teuthida Naef, 1916 is maintained below, however, as the specimens discussed below shed no further light on this discussion. Suborder mesoteuthina Naef, 1921 Family trachyteuthididae Naef, 1921 Genus trachyteuthis Meyer, 1846 (= Coccoteuthis Owen, 1855, Voltzia Schevill, 1950; junior subjective synonyms) Type genus. Sepia hastiformis Riippell, 1829, by subsequent designation (Bulow-Trummer 1920, p. 248). Diagnosis. See Naef (1922, p. 137). Remarks. The form of Trachyteuthis , and its relative similarity to the present-day cuttlebone of Sepia officianalis , has led some authors to consider that this genus is actually representative of the Sepiida rather than the Teuthida. Schevill (1950) described a new genus from the Oxfordian of Cuba, Voltzia , which he considered distinct from Trachyteuthis , as it apparently possessed ‘phragmocone deposits’ similar to those of Sepia. Donovan (1977) questioned the distinction of the nominal genera Voltzia and Trachyteuthis , but went further in suggesting that Trachyteuthyis was a true sepiid, attributing the lack of phragmocone to solution of the delicate aragonite plates after burial. In one specimen of Trachyteuthis from Solenhofen (BMNH 83730), Donovan reported fragments attributable to phragmocone debris as a ‘lag’ beneath the shell After sectioning, I found that this specimen revealed no further debris beneath the dorsal shield, and no indication of an extensive phragmocone development, and as such this evidence is not unequivocal. No traces of phragmocone were found in the Antarctic specimen, which is preserved as a thin shield < 1 mm thick, built of successive lamellae. An ink sac is present in this specimen, and though slightly displaced, it is found directly beneath the thin gladius, without any trace of intervening phragmocone plates. Hewitt and Wignall (1988) have studied the mineralogy of Trachyteuthis specimens from the English Kimmeridge Clay, and have determined that its original mineralogy was francolite, rather than aragonite. These authors used this as additional evidence against sepiid affinities of Trachyteuthis , arguing that one would expect a sepiid ‘cuttlebone’ to be aragonitic, rather than 172 PALAEONTOLOGY, VOLUME 34 phosphatic. The Antarctic specimen described below is also phosphatic, but the possibility of diagenetic replacement of original aragonite cannot be ruled out, especially since specimens of the thin-shelled ammonite Haploceras and belemnoid ? Belemnoteuthis (both originally aragonitic) are found as crushed, phosphatic films in the Nordenskjold Formation. In summary, it seems probable that despite the close morphological similarities between the shells of Trachyteuthis (= Voltzia) and Sepia, the absence of a proven phragmocone and the possible original phosphate shell mineralogy of the former suggest that assignment to the Teuthida rather than to the Sepiida is more correct. Range. Definite records from the Lower Oxfordian to Tithonian of southern West Germany (Bavaria), England (Dorset, North Yorkshire), USSR (Volga region), Cuba (Vinales region) and Antarctica (Graham Land). A single doubtful record from the Lower Aptian of northern West Germany (Heligoland). cf. *1829 cf. v. 1855 cf. v . 1 896 cf. 1 922 cf. v. 1977 cf. v. 1988a cf. 19886 v. 1988 Trachyteuthis cf. hastifonnis ( Ruppell, 1829) Text-figs 2a, b and 4 Sepia hastifonnis Ruppell, p. 9, pi. 3, fig. 2. Coccoteuthis latipinnis Owen, p. 124, pi. 7. Coccoteuthis hastifonnis Ruppell; Crick, p. 439, pi. 14. Trachyteuthis hastifonnis (Ruppell); Naef, p. 137, text-fig. 51. Trachyteuthis sp. Donovan, p. 32, text-figs 8 and 9. Trachyteuthis hastifonnis (Ruppell); Engeser, p. 82, text-fig. lc. Trachyteuthis hastifonnis ( Ruppell); Engeser, p. 59. [Full and extensive synonymy given], lossil teuthid; Anonymous, p. 15, text-fig. 6. [Colour photograph of specimen described below]. Type specimen (of Trachyteuthis hastifonnis). Holotype, Senckenberg Museum, Frankfurt-am-Main, register number XI 1328. Lower Tithonian, Solenhofen Limestone, Miihlheim, Bavaria, West Germany. Material. One specimen, BAS D. 9007. 33, uppermost Longing Member, Nordenskjold Formation (Whitham and Doyle 1989, p. 6). Longing Gap, Graham Land, Antarctic Peninsula. Preserved intact in a carbonate concretionary horizon yielding the ammonite Virgatosphinctes rotundidoma Uhlig of Tithonian age (Whitham and Doyle, 1989, text-fig. 6/). Description. The single specimen collected comprises the majority of the median field of a small (total preserved length 90-5 mm) Trachyteuthis gladius. It is preserved in a carbonate concretion allowing three-dimensional preservation, in a formation that otherwise yields compressed fossils. The specimen consists of two parts, naturally split by freeze-thaw action. These represent ventral and dorsal surfaces of the gladius divided cleanly along shell lamellae, the two parts united are less than 1 mm thick. The dorsal fragment (Text-fig. 2b) is the most recognizable of the two as Trachyteuthis. It exhibits (in negative, as it is the undersurface of the topmost part of the gladius) a median field with a narrow (width 23-4 mm) central region composed of a series of closely spaced pustules arranged in arcuate arrays which correspond to growth lines. A central ridge or median keel is present. The median field is completed by relatively smooth lateral areas (‘ Seiteplatte' of Naef 1922, text-fig. 51). These are incomplete, but display some longitudinal striation. Finally, there are displaced fragments of a probable wing present at the left posterior of the shell. The ventral fragment (Text-fig. 2a) is less easily recognizable as representative of Trachyteuthis , as there are no pustules or definable field areas present. The fragment consists of an almost smooth shield with some traces of arcuate growth lines in the central area. Part of the lateral area of the median field is definable in the right anterior of the specimen, and fragments of a wing in the left posterior. This portion of the gladius very clearly shows the lamellar construction of the gladius and, where successive lamellae have exfoliated, neither it nor the underlying matrix displays any evidence of phragmocone deposits. The presence of an ink sac is indicated by a dull black mass up to 5 mm thick beneath the ventral portion of the gladius in the right posterior of the fragment (Text-fig. 2a). DOYLE: ANTARCTIC TEUTHID CEPHALOPODS 173 text-fig. 2. Trachyteuthis cf. hastiformis (Riippell). Specimen D. 9007. 33, Tithonian, Longing Member, Nordenskjold Formation, Longing Gap. Ventral and dorsal fragments of naturally split gladius, x 1. a, ventral fragment showing the ink sac beneath the thin lamellae of the shell, b, dorsal fragment showing the characteristic pustules of the median field. Abbreviations: MA, median asymptote; MF, median field; MK, median keel; Sp, 1 Seiteplatte' ; W, wing. A reconstruction of the gladius is given in Text-figure 4. Remarks. This specimen is clearly representative of the genus Trachyteuthis , and is very close to specimens from the Kimmeridge Clay of England and the Solenhofen Limestone of West Germany. However, its small size, which may indicate that it is a juvenile, and its incomplete preservation, allow only tentative assignment to the species Trachyteuthis hastiformis (Riippel). Trachyteuthis palmeri (Schevill) (Lower Oxfordian, Cuba) and T. zhuravlevi Hecker and Hecker (Lower Volgian, Volga region, USSR) are poorly known, and differ only in their greater width and elongate form, respectively. Suborder kelaenina Starobogatov, 1983 Family muensterellidae Roger, 1952 munsterellid gen. et sp. nov. Text-figs 3a, b and 4 Material. A single specimen, BAS D. 9008. 3, found loose in the uppermost Longing Member (approximately equivalent horizon to BAS D. 9007. 33), Nordenskjold Formation. Longing Gap, Graham Land, Antarctic Peninsula. Associated Virgatosphinctes and Retroceramus specimens indicate a Tithonian age. 174 PALAEONTOLOGY, VOLUME 34 Description. The specimen consists of a gladius with a preserved length of 82.5 mm, comprising a broad spoon- shaped conus with a rhachis extending anteriorly from it (Text-fig. 3a). The spoon-shaped conus is incomplete, but has an approximate maximum width of 37 mm. It is preserved flattened with no indications of concentric or other growth lines upon its dorsal surface. The median field of the gladius is developed as a rhachis, commencing as a median ridge or raised area in the posterior of the gladius, then extending anterior of the conus. The median field is completed by smooth lateral outgrowths (' Seiteplatte') which accompany the rhachis for half of its length and up to 60 mm of the total length of the gladius, and indistinct surface features on the conus indicate the possible position of the median asymptotes which border the median field (Text-fig. 4). The rhachis diverges anteriorly at an angle of approximately 5°, and expands to a maximum width of 4 mm. As the rhachis expands, it divides anteriorly from its original raised area on the conus to produce two laterally placed ridges with an intervening, smoother area. There is some indication of a weak median keel in the centre of this region, but preservation is too poor for this to be unequivocal. text-fig. 3. Muensterellid gen. et sp. nov. Specimen D. 9008. 3, Tithonian, Longing Member, Nordenskjold Formation, Longing Gap. a, dorsal view of gladius x 1. b, sketch representation of same view, xl. Abbreviations: C, conus; FR, free rhachis; Pr, preparation marks; Sp, ‘ Seiteplatte A reconstruction of the gladius is given in Text-figure 4. Remarks. The unusual divided form of the rhachis, and apparent absence of growth lines in this specimen, initially gave rise to doubts about its actual cephalopod affinities. However, despite this, the regular form of the conus and its relationship with the rhachis confirm that this specimen undoubtedly represents a (new) teuthid taxon, and it is certainly not representative of any known non-cephalopod mollusc, plant (cf. Engeser and Phillips 1986) or even fish (P. Forey, pers. comm. 1987). DOYLE: ANTARCTIC TEUTHID CEPHALOPODS 175 The form of the specimen discussed most closely resembles taxa of the Muensterellidae (Kelaenina). Specifically, the presence of a ‘free rhachis’ distinguishes it from otherwise similar specimens of Palaeololigo Naef (Palaeologinidae, Mesoteuthina), which have a broader median field. Of the Muensterellidae, the Tithonian genera Listroteuthis Naef and Muensterella Schevill are closest, especially the former which has a similar conus shape. The only other muensterellid with a divided rhachis is the Campanian form Tusoteuthis Logan (= Kansasteuthis Miller and Walker; see Nicholls and Isaak 1987, p. 734). The gladius of Tusoteuthis has a leaf-shaped conus with a robust ‘free rhachis' starting immediately from its anterior. The rhachis does not diverge significantly to the anterior, and is much more robust than that of the present specimen. Difference of rhachis design in otherwise similar spoon-shaped gladii of Recent squid was noted by Toll (1988), illustrating the potential for variability in this feature. The rhachis of the Recent family Bathyteuthidae would seem to be analogous to the Antarctic specimen, having lateral rods joined by a central U-shaped area. The only Prototeuthina which approaches the present specimen is the genus Maioteuthis Reitner and Engeser (Plesioteuthididae). Maioteuthis has a much reduced conus and an extremely long and Trachyteuthis Muensterellid text-fig. 4. Suggested reconstructions of the Antarctic teuthids, not to scale. Trachyteuthis redrawn after Naef (1922, fig. 51). Abbreviations: C, conus; LA, lateral asymptote; MA, median asymptote; MF, median field; MK, median keel; Sp, ‘ Seiteplatte' ; W, wing. 176 PALAEONTOLOGY, VOLUME 34 narrow median field which divides anteriorly to present a weak median keel (Reitner and Engeser 1982, text-fig. 2). The Antarctic specimen resembles Maioteuthis only in having a divided median field with a faintly developed median keel, but differs greatly in possessing a spoon-like conus with an anteriorly extensive ‘free rhachis’, demonstrating its muensterellid affinities. In summary, the overall form of the gladius (conus and rhachis) of this specimen would support the erection of a new genus within the Muensterellidae. However, the single specimen available does not permit the formal designation of a new taxon. PALAEOBIOGEOGRAPHICAL CONSIDERATIONS There are too few records to provide any definite conclusions about the palaeobiogeography of Mesozoic teuthids. However, the discovery of fossil teuthid gladii from Gondwana is significant in illustrating that the present observed European bias is artificial, induced to some extent by the fragility of the remains and a greater intensity of study in western Europe. Therefore, some primary observations are presented here. In addition to its European (England, West Germany) occurrences (see Engeser 1 988 A and references therein), the genus Trachyteuthis is recorded from the Lower Volgian of the USSR (Volga region) (Hecker and Hecker 1955), the Lower Oxfordian of western Cuba (as Voltzia) (Schevill 1950) and now the Tithonian of Antarctica. The majority of these specimens are remarkably similar to the western European representatives, especially the Antarctic example, suggestive of an almost worldwide distribution for Trachyteuthis in the Late Jurassic, transgressing boreal and Tethyan realm boundaries observed in other marine groups. Cretaceous trachyteuthids are represented only by a possible Trachyteuthis from the Lower Aptian of Heligoland (northern Germany) (Engeser and Reitner 1985) and Upper Cretaceous records from the Lebanon (e.g. Lihanonteuthis Kretzoi) and North America (e.g. Actinosepia Whiteaves) (Engeser 1988/5). The Muensterellidae have similarly disparate geographical records. Muensterella and associated genera (i.e. Listroteuthis Naef, Calaenoteuthis Naef) are presently known only from the Lower Tithonian of West Germany (see Engeser 1988/5). The Antarctic muensterellid, described from sediments of similar age, has many points in common with these European genera, and like Trachyteuthis, is a possible indicator of a formerly more widespread distribution. Cretaceous muensterellids are relatively rare, but there is some indication of less centred distribution pattern than is presently observed. Thus, while Tusoteuthis (? = Kansasteuthis, Niobrarateuthis and Enchoteuthis) is only recorded from the Upper Cretaceous of North America (Nicholls and Isaak 1987), two undescribed Australian teuthid specimens with affinity to Tusoteuthis are preserved in the BMNH collections. These specimens, from the Lower Cretaceous (Albian) of Queensland, Australia, (BMNH C. 59211, C. 59276) resemble Tusoteuthis, but are larger, possessing a ribbed conus and multiple grooved free rhachis, and undoubtedly represent a new taxon. Acknowledgements . The specimens described above were collected while I was employed by the British Antarctic Survey, to which organization I extend my thanks for the opportunity to describe them. I gratefully acknowledge assistance in the field from Dave O'Dowd and Donny Stewart. I thank my co-worker Andy Whithanr for his useful comments and discussion. REFERENCES anonymous. 1988. The Natural Environment Research Council report for 1987/1988. NERC, Swindon, 51 pp. bandel. K. and leich, H. 1986. Jurassic Vampyromorpha (dibranchiate cephalopods). Neues Jahrbuch fiir Geologie und Palaontologie, Monatshefte , (1986), 129-148. bather, F. a. 1888. Professor Blake and shell-growth in Cephalopoda. Annals and Magazine of Natural History, (1888), 421-427. berthold, t. and engeser, t. 1987. Phylogenetic analysis and systemization of Cephalopoda. Verhandlungen des Naturwissenschaftlichen Vereins in Hamburg, N.S., 29. 187-220. DOYLE: ANTARCTIC TEUTHID CEPHALOPODS 177 bulow-trummer, e. von. 1920. Fossilium Catalogus 1: Animalia. Pars 11; Cephalopoda Dibranchiata. Junk, Berlin, 313 pp. crick, G. c. 1896. On a specimen of Coccoteuthis hasitformis Riippell, sp., from the Lithographic Stone (Lower Kimmeridgian) of Eichstadt, Bavaria. Geological Magazine, Decade 4, 3, 439 — 443. donovan, d. r. 1977. Evolution of the dibranchiate Cephalopoda. Symposia of the Zoological Society of London , 38, 15-48. - 1983. Mastigophora Owen, 1856: a little-known genus of Jurassic coleoids. Neues Jahrbuch fiir Geologie and Paldontologie , Abhandlungen , 165, 484-495. doyle, p. 1990. Teuthid cephalopods from the Lower Jurassic of Yorkshire. Palaeontology , 33, 193-207. — and whitham, a. g. in press. Palaeoenvironments of the Nordenskjold Formation, an Antarctic Late Jurassic-Early Cretaceous black shale-tuff sequence. In tyson, r. v. and pearson, t. h. (eds). Modern and ancient shelf anoxia. Geological Society Special Publications, London. engeser, t. 1986. Beschreibung einer wenig bekannten und einer neuen Coleoiden-Art (Vampyromorphoidea, Cephalopoda) aus den Untertithonium von Solnhofen und Eichstatt (Bayern). Archaeopteryx, (1986), 27-35. 1988r/. Fossil ‘octopods’-a critical review. 81-87. In clarke, m. r. and trueman, e. r. (eds). The Mollusca , vol. 12. Palaeontology and neontology of cephalopods. Academic Press, San Diego, 355 pp. 1988(6. Fossilium Catalogus 1 Animalia. Pars 130 Vampyromorpha (‘ Fossile Teuthiden'). Kugler, Amsterdam, 167 pp. — and bandel, k. 1988. Phylogenetic classification of coleoid cephalopods. 105-115. In wiedmann, j. and kullmann, j. (eds). Cephalopods - present and past. Schweizerbart, Stuttgart, 763 pp. — and Phillips d. 1986. Redescription of two specimens previously recorded as fossil teuthids (Coleoidea, Cephalopoda). Bulletin of the British Museum (Natural History ), Geology, 40, 249-264. - and reitner, j. 1985. Teuthiden aus dem Unterapt (‘Tock') von Helgoland (Schleswig-Holstein, Norddeutschland). Paldontologisches Zeitschrift, 59, 245-260. - 1986. Coleoidenreste aus der Oberkreide des Libation in Staatlichen Museum fiir Naturkunde in Stuttgart. Stuttgarter Beitrdge zur Naturkunde, Serie B , 124. 1 15. farquharson, G. w. 1983. The Nordenskjold Formation of the northern Antarctic Peninsula: an Upper Jurassic radiolarian mudstone and tuff sequence. British Antarctic Survey Bulletin , 60, 1-22. fischer, j. c. and riou, b. 1982. Les teuthoi'des (Cephalopoda, Dibranchiata) du Callovien Inferieur de la Voulte-Sur-Rhone (Ardeche, France). Annales de Paleontologie, 68, 295-325. hall, r. 1985. Paraplesioteuthis hastata (Munster), the first teuthid squid recorded from the Jurassic of North America. Journal of Paleontology, 59, 870-874. — and neuman, a. G. 1989. Teudopsis cadominensis , a new teuthid squid from the Toarcian (Lower Jurassic) of Alberta. Journal of Paleontology, 63, 324—327. hecker, e. l. and hecker, r. f. 1955. Remains of Teuthoidea from the Upper Jurassic and Lower Cretaceous of the Middle Volga area. Voprosy Paleontologii, 2, 36-44. [In Russian]. iiewitt, r. a. and wignall, p. b. 1988. Structure and phylogenetic significance of Trachyteuthis (Coleoidea) from the Kimmeridge Clay of England. Proceedings of the Yorkshire Geological Society, 47, 149-153. jeletzky, j. a. 1966. Comparative morphology, phylogeny and classification of fossil Coleoidea. University of Kansas Paleontological Contributions. Mollusca, Art. 7, 162 pp. MACDONALD, D. I. M., BARKER, P. F., GARRETT, S. W., INESON, J. R., PIRRIE, D., STOREY, B. C., WHITHAM, A. G., kinghorn, r. r. f. and marshall, j. e. a. 1988. A preliminary assessment of the hydrocarbon potential of the Larsen Basin, Antarctica. Marine and Petroleum Geology, 5, 34-54. Matthews, S. c. 1973. Notes on open nomenclature and on synonymy lists. Palaeontology, 16, 713-719. meyer, H. von. 1846. Mitteilungen an Professor Bronn gerichtet. Neues Jahrbuch fiir Mineralogie, Geognosie, Geologie und Petrefaktenkunde, (1846), 595-599. miller, h. w. and walker, m. w. 1968. Enchoteuthis melanae and Kansasteuthis lindneri, new genera and species of teuthids, and a sepiid from the Niobrara Formation of Kansas. Transactions of the Kansas Academy of Science, 71. 176-183. MOORE, c. 1870. Australian Mesozoic geology and palaeontology. Quarterly Journal of the Geological Society of London , 26, 226-261. naef, a. 1916. Systematische Ubersicht der mediterranean Cephalopoden. Pubblicazioni della Stazione zoologica di Napoli, 1, 11-19. - 1921. Das System der dibranchiaten Cephalopoden und die mediterranen Arten derselben. Mitteilungen aus der Zoologischen Station zu Naepel , 22, 527-542. - 1922. Die fossilen Tintenfische - Eine palaozoologische Monographic. Fischer, Jena, 322 pp. 178 PALAEONTOLOGY, VOLUME 34 nicholls, e. and isaak, h. 1987. Stratigraphic and taxonomic significance of Tusoteuthis longa Logan (Coleoidea, Teuthida) from the Pembina Member, Pierre Shale (Campanian), of Manitoba. Journal of Paleontology, 61, 727-737. owen, R. 1855. Notice of a new species of an extinct genus of dibranchiate cephalopod (Coccoteuthis latipinnis ) from the Upper Oolitic shales at Kimmeridge. Proceedings of the Geological Society of London, 11, 124—125. reitner, j. and engeser, T. 1982. Teuthiden aus dem Barreme der Insel Maio (Kapverdische Inseln). Paldontologisches Zeitschrift , 56, 209-219. riegraf, w., werner, G. and lorcher, F. 1984. Der Posidonienschiefer . Biostratigraphie, Fauna und Fazies des siidwestdeutschen Untertoarciums ( Lias e). Enke, Stuttgart, 195 pp. ROGER, j. 1946. Les Invertebres des couches a poissons du Cretace Superieur du Liban. Memoires de la Societe Geologique de France, N.S., 51, 1-92. 1952. Sous-classe des Dibranchiata Owen, 1836. 689-755. In piveteau, j. (ed.). Traite de paleontologie, vol. 2. Masson, Paris, 755 pp. ruppell, E. 1829. Abbildung und Beschreibung einiger neuen oder wenig bekannten Versteinerungen aus der Kalkschieferformation von Solnhofen. Bronner, Frankfurt am Main, 12 pp. schevill, w. E. 1950. An Upper Jurassic sepioid from Cuba. Journal of Paleontology, 24, 99-101. storabogatov, y. i. 1983. The system of the Cephalopoda. 4-7. In storabogatov, y. i. and nesis, k. n. (eds). Taxonomy and ecology of cephalopods. Zoological Institute, USSR Academy of Sciences, Leningrad, 77 pp. [In Russian], toll, r. b. 1988. Functional morphology and adaptive patterns of the teuthoid gladius. 167-182. In trueman, e. R. and clarke, M. R. (eds). The Mollusca, vol. 11. Form and Function. Academic Press, San Diego, 504 pp. whitham, a. G. and doyle, p. 1989. Stratigraphy of the Upper Jurassic-Lower Cretaceous Nordenskjold Formation of eastern Graham Land, Antarctica. Journal of South American Earth Sciences. 2, 371-384. woodward, s. p. 1883. On a new genus of fossil ‘ Calmary ’ from the Cretaceous formation of Sahel-Alma, near Beirut, Libanon, Syria. Geological Magazine, Decade 2, 10, 1-5. PETER DOYLE Nature Conservancy Council Typescript received 25 October 1989 Northminster House Revised typescript received 5 March 1990 Peterborough PEI 1UA, UK A NEW SCLE R ACTINI AN-LI KE CORAL FROM THE ORDOVICIAN OF THE SOUTHERN UPLANDS, SCOTLAND by COLIN T. SCRUTTON and EUAN N. K. CLARKSON Abstract. New, discoidal fossils preserved as moulds from the middle Ordovician (Caradoc) of the Southern Uplands are shown to possess characteristic coralline microarchitecture. They are solitary, zoantharian corals with cyclic, hexameral septal insertion. Successive cycles are arranged in a system of nested triads similar to patterns associated with septal substitution in scleractinian corals. The corallum lacks tabulae or dissepiments but is epithecate with the point of origin a basal disc as in Scleractinia rather than a cone as in Rugosa. The new coral is named Kilbuchophyllum discoidea gen. et sp. nov., and is placed in the new family Kilbuchophyllidae and the new order Kilbuchophyllida. It is interpreted as an early example of skeletal acquisition by the group of anemones that ultimately gave rise to the Scleractinia in the Middle Triassic. The phylogeny of the Zoantharia is briefly discussed in the light of this new material. A striking feature of the geological distribution of fossil corals is the sequential ranges of the two major and crudely homoeomorphic groups possessing well developed septa. The Rugosa appear in the mid-Ordovician and become extinct at the end of the Permian (Scrutton 1979, 1988; Hill 1981), whilst the Scleractinia first occur in the middle Triassic and persist to the present day (Wells 1956; Oliver 1980). No early Triassic corals are known. The fundamental distinction between these two groups of corals lies principally in their modes of septal insertion, serial in four quadrants in the Rugosa, and cyclic, hexameral in the Scleractinia (Oliver 1980). Other zoantharian corals occur in the Palaeozoic but are less comparable. The small, enigmatic Devono-Carboniferous order Hexacorallia is also strongly septate but distinct in septal pattern from both Rugosa and Scleractinia (Hill 1981). A third major group of exclusively Palaeozoic and colonial corals, the Tabulata, have variably and generally weakly developed septa for which no definite pattern of insertion has yet been established. Claimed rugosan insertion in Agetolites (Kim 1974) requires restudy before its significance can be assessed. In addition, pre-Ordovician beds have yielded a small number of coralline organisms, one of which, Cothonion , has well developed septa and may derive from the same stock as the Rugosa (Scrutton 1979; Jell 1984). Although direct descent of the Scleractinia from the late Palaeozoic Rugosa has been claimed by Schindewolf (1942) and others, the alternative view that the Scleractinia evolved independently from anemone precursors in the middle Triassic has been strongly argued by Oliver (1980). Over the years, a number of Palaeozoic corals had been described as exhibiting scleractinian characters. However, Hill (1960) noted that ‘all Palaeozoic corals claimed... to be Scleractinia have subsequently been proved to be Rugosa...’. The sole uncertainty she allowed was the record of apparent Permian age of species of the genus Omphalophyllia by Minato (1955). This Japanese material is too poorly preserved to be reliably interpreted and its restudy is required; Minato (1955, p. 1 80) considered the possibility that it was related to Lophocarinophyllum. On the other hand, the host rock is now interpreted as an olistostrome in a Triassic accretionary complex containing Carboniferous, Permian and Triassic olistoliths; thus there are no positive data to support Permian Omphalophyllia (Makoto Kato pers. comm.). The type material of Omphalophyllia is a Triassic scleractinian coral, considered a synonym of Conophyllia by Wells (1956). Since then, the possibility of a scleractinian presence in the Palaeozoic has continued to be raised. Krasnov (1970) considered Scleractinia of fungiid type to have separated from the Rugosa in the early to mid Palaeozoic with (Palaeontology, Vol. 34, Part 1, 1991, pp. 179-194, 1 pl.| © The Palaeontological Association 180 PALAEONTOLOGY, VOLUME 34 the Calostylidae as the most likely ancestral group. However, Smith (1930) had already shown Calostylis to have rugosan septal insertion and this was confirmed by Weyer (1973). More recently, Erina and Kim (1980) considered the Ordovician Sumsarophyllum and their new genus Tjanshanophyllia to both show fungiid characteristics on the basis of many cycles of perforate septa and the reported lack of an epitheca. They did not demonstrate cyclic, hexameral septal insertion, however, which must be considered the critical evidence for rejecting classification with the Rugosa and supporting comparison with the Scleractinia. Oliver (1980) concluded at that time that no known Palaeozoic coral demonstrated cyclic septal insertion. Thus we regard our description here of a new Ordovician solitary coral with scleractinian characteristics as the first well documented case of a Palaeozoic scleractiniamorph. Our claim is based on a full assessment of the structure and development of the coral, including and principally, the clear expression of hexameral cyclic septal insertion. We presuppose our conclusions concerning the nature of this material and scleractinian coral terminology is used throughout for the morphological descriptions (Wells and Hill 1956). Abbreviations. All material we have collected is housed in the Royal Museum of Scotland, Edinburgh (RMS). Additional material referred to is housed in the British Geological Survey, Edinburgh (BGS) and The Natural History Museum, Department of Zoology, London (BM(NH)Z). FIELD OCCURRENCE In the Southern Uplands of Scotland, Ordovician rocks extend as a continuous belt from the North Sea to the Irish Sea (Text-fig. 1). They are largely confined to the Northern Belt and consist in the main of Arenig volcanics overlain by a suite of younger sediments. The oldest sediments are Llanvirn-Llandeilo red and grey cherts; these are succeeded by black shales of Glenkiln age ( gracilis and peltifer Zones), overlain in turn by Caradoc greywackes, grits and shales. Whereas all these sediments are typically of deep-water origin, there are a number of localities, referred to by Peach and Horne (1899), which have yielded shelly fossils. The best exposures are at Kilbucho (National Grid Reference NT 060338) and at Wallace’s Cast in the Wandel Burn (NT 967263) (Text-fig. 1). Other occurrences were noted by Ritchie and Eckford (1935) westwards to Duntercleuch and Snar, north-west of Leadhills. At Kilbucho and Wallace’s Cast, imperfectly exposed though they are, basal greywackes are overlain by a coarse conglomerate with clasts of igneous rock, limestone and mudstone with undistorted fossils. This conglomerate fines upwards into siltstone turbidites and mudstones, cleaving subparallel with the bedding, which yield a rich assemblage of fossils, usually found in a rather distorted state. The total thickness of the conglomerates and associated siltstones and mudstones is no more than 5 m. These sediments are interpreted as debris-flow deposits, probably triggered seismically. They were originally laid down in shallow waters, following which, large unstable masses of partly and unlithified sediments slumped rapidly into deep-water, burying their faunas in the process. The sediments at the two localities of Kilbucho and the Wandel Burn are very similar, though they lie 12 km apart along the strike. It is quite possible that they record the events of a single debris flow of vast size although this cannot be confirmed. They could, on the other hand, have been smaller, separate, but near contemporaneous debris-flows from the same source. All the fossils are well-preserved as moulds (Text-fig. 2a-d), though distorted and not infrequently cracked, possibly during transportation or through diagenetic effects. The fossiliferous mudstones and siltstones are a classic obrution deposit, the fossils often being preserved at an angle to the bedding. A full description of the localities and other faunas is in preparation (Clarkson, Harper, Owen and Taylor in prep.). As well as a rich variety of brachiopods and trilobites, there are also bryozoans (especially common at Wallace's Cast), ostracodes, bivalves, gastropods, nautiloids and crinoids in addition to the material described here. Scattered solitary rugose corals are present at Wallace’s Cast and rarely in mudstone clasts in the conglomerate at Kilbucho. The new genus described here SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 181 text-fig. I Map of the mid-north Southern Uplands of Scotland indicating collecting sites at Kilbucho and Wallace’s Cast. Area of detailed map located in inset. is known only from two poor fragments from Wallace’s Cast but is common in a wide range of ontogenetic stages in a coarse-silt grade turbidite at Kilbucho, where it appears not to be associated with rugose corals. The graywacke group in which the Kilbucho-Wandel Burn sequence occurs lies within Tract 2 of the Southern Uplands (Leggett et al. 1979) and belongs to the Kirkcolm Formation (J. Floyd pers. comm.). The trilobite fauna is fairly diverse, there being twelve species (A. Owen pers. comm.), and there are up to twenty-four species of brachiopods (D. A. T. Harper pers. comm.). Amongst the trilobites, the most common faunal elements are the mid-Caradoc (Balclatchie and Ardwell) Calyptaulax brongniartii (Portlock) (see Clarkson and Tripp 1982) and Illaenus convergens, with subsidiary Stenopareia, Cybeloides, Paraharpes and Remopleurides. The numerous brachiopods are very similar to those of the Bardahessiagh Formation, Pomeroy, Northern Ireland (Mitchell 1977), of high Ardwell age, but some of the elements are also found in the Balclatchie and Ardwell Beds at Girvan. The total age range of the Kilbucho-Wandel faunas cannot for the moment be dated more accurately than mid-Caradoc (Soudleyan - Actonian). 182 PALAEONTOLOGY, VOLUME 34 EXPLANATION OF PLATE 1 Figs 1-8. Kilbuchophyllia discoidea gen. et sp. nov.; all scanning electron micrographs of gold coated latex replicas; Ordovician, nhd-Caradoc; Kilbucho, near Biggar, southern Scotland. 1 and 2, RMS 1989.36.4, mature septal blade, corallum axis to left; note pattern of diverging columnar units in fractured face in fig. 2; 1, x 15; 2, x 60. 3, RMS 1989.36.5, oblique view of individual trabecular spine set on internal surface of epitheca at peripheral margin of corallum, x 150. 4, RMS 1989.36.2, plan view of pair of trabecular spines in mid septum, x 150. 5 and 6, RMS 1989.36.6, external surface of epitheca, periphery of corallum bottom right; 5, general view showing prominent septal grooves and growth ridges, x 15; 6, detail of growth ridge crossing interseptal ridge, x45. 7, RMS 1989.36.4, pattern of nested triads developed about third order septum in centre of figure; first order septum to left of group, second order septum to right of group, x 15. 8, RMS 1989.36.5, papillose axial structure of merged trabecular spines, x 30. text-fig. 2. Kilbuchophyllia discoidea gen. et sp. nov. Ordovician, mid-Caradoc, Kilbucho, near Biggar, southern Scotland, a-d, original moulds; a, RMS 1989.36.2, calical surface of immature specimen in which septa are weakly linked spines (compare with Text-figs 3e, f); axis of symmetry vertical, x 5; b, RMS 1989.36.1 (holotype), calical surface of mature specimen in which septa are solid blades (compare Text-fig. 3g, h); axis of symmetry vertical, x2-5; c and d, RMS 1989.36.7, calical surface; undersurface of epitheca, part and counterpart, x 3. E, RMS 1989.36.8, latex replica of external surface of epitheca showing well developed interseptal grooves and growth ridges, x 4. PLATE 1 SCRUTTON and CLARKSON, Kilbuchophyllia 184 PALAEONTOLOGY, VOLUME 34 PREPARATION The mouldic material, preserved in a coarse-silt grade, quartz-rich turbidite with a substantial mica and clay mineral matrix, was cleaned in a weak solution of Calgol in an ultrasonic bath and latex replicas were made using standard techniques. Although these replicas demonstrate overall three-dimensional appearance of the coral, distortion of the septal blades, either taphonomic or tectonic, tends to obscure the detailed interseptal relationships in most specimens. Therefore, septal patterns were traced directly from the moulds using a camera lucida attachment on the microscope. These reflect the growth of the septal blades on the upper surface of the epitheca and are thus likely to reflect the interseptal relationships most accurately. For the purposes of illustration in Text-figure 3, these patterns have been reversed to show the standard calical view of septal arrangement in corals. The latex replicas were used for SEM study of septal microarchitecture. Selected replicas were coated with gold under vacuum to a thickness of 12-15 nnr and examined at a range of magnifications using a Cambridge Instruments Stereoscan 240 in the Biomedical E.M. Unit at the University of Newcastle upon Tyne. Problems of creep in the latex were solved by working at low energy levels, between 0-5 and 4 kV. Very low magnification pictures taken at settings for maximum depth of field, such as that in Text-figure 3h, suffer from slight spherical distortion but have been preferred for their clarity over light microscope photographs. In coral studies, mouldic preservation is usually regarded as of limited value. It is less of a disadvantage in the present material because of the discoidal growth form and lack of horizontal elements between the septa : no macrostructural detail is lost. However, the SEM results obtained here, suggest that all mouldic material may repay closer examination. MORPHOLOGICAL CHARACTERISTICS A full description of this new species is given below. The present discussion concentrates on the two most important features bearing on the anthozoan cnidarian nature of the material and its phylogenetic relationships within the class: microarchitecture of the skeletal elements and septal pattern. Microarchitecture SEM study of gold-coated latex replicas reveals the preservation of elements of about 20 //m and above in the better preserved material. Individual septal spines of up to 200 /mi diameter, in specimens RMS 1989.36.2 and 5, are constructed of upward and outwardly diverging columnar units of indeterminate length and subrectangular to rhomboidal to irregular (?oblique) section, c. 20 /mi across (PI. 1, figs 3 and 4). Viewed from above, terminations give the appearance of overlapping roof tiles, possibly helically arranged in a conical stack. Where spines are first linked to form continuous but beaded septal plates, the intervening ridges are composed of units of similar size and shape. In larger coralla, in which individual spines have been subsumed into dentate, flat faced septal blades, the lateral faces of the blades have the appearance of a uniform fabric of overlapping scales (PI. 1, figs 1 and 2). On the fractured surface of a septal tooth, internal upward fanning of columnar units is visible; the effect of overlapping scales is produced by oblique terminations of these units at the surface. No substructure is visible within these units. The axial structure in some smaller coralla is composed of a cluster of discrete spines (PI. 1, fig. 8). These have the same structure as the septal spines and are clearly septal in origin. The basal surface of the epitheca shows circumferential ridges, demarcating growth increments, and sometimes, except in the axial area, radiating septal grooves (PI. 1, fig. 5; Text-fig. 2d,e). In between ridges, the surface is smoother and may be very smooth to almost featureless. At the ridges, a cluster of overlapping triangular to arcuate elements averaging 40 //m across forms a low scarp slope directed towards the axis (PI. 1, figs 5 and 6). Individual elements are oriented radially and offlap towards the periphery, the ultimate series in each ridge subparallel to the inter-ridge surface of the epitheca. The calicular surface of the epitheca is rather smooth and undulating in places but elsewhere shows sub-vertically orientated elements with low pyrimidal terminations approximately 20-40 //m across. These define a fabric which appears to have a crudely radial orientation in places (PI. 1, fig. 3). SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 185 Septal pattern Septa are arranged radially, reaching up to 0-8 of the corallum radius in length, on a flat, circular (see below) basal disc. Pattern is most readily detected in the smaller coralla with about 30-40 septa. In larger, mature coralla, with up to 120 septa, not only are most specimens incomplete but septal arrangement becomes increasingly irregular. Two features reveal the septal pattern : the relative length of the septa, and curvature of the inner ends of septa of higher order to face, or rest against the flanks of septa of lower order (PI. 1, fig. 7). In RMS 1989.36.2 (Text-figs 2a and 3e, f), 12 septa of approximately equal length extend 0 8 radius to the axis. Of these, alternate septa are each flanked by two shorter septa, between 0-5 and 0-75 radius in length, whose axial ends turn, more or less strongly, towards each other and the opposite faces of the dividing, longer septum. There are thus 12 of these shorter septa. Eight of these are again each flanked by a pair of even shorter septa, 01 to 0 5 radius in length and converging on opposite faces towards the axis. This repeated pattern of septal convergence leads to the appearance of nested triads of septa of which there are six, separated by six of the longest septa which have no divergent septal groups. These latter are interpreted as six first cycle septa, alternating with six second cycle septa that form the axis of each set of nested triads. The successive groups of diverging septa represent, respectively, 12 third cycle septa and 16 fourth cycle septa, amounting to 40 septa in all. The fourth cycle here is regarded as incomplete, numbering 24 septa when complete. The two triads lacking fourth cycle septa are adjacent and flank an axis of bilateral symmetry defined by a short septal blade in the axial area of the coral. All of the available material, except the smallest specimen, clearly shows this septal pattern of nested triads and often some weak indication of an overall bilateral aspect. Two further specimens unequivocally, and several others less certainly, demonstrate the hexameral symmetry of the pattern of triads. One, RMS 1989.36.4 (PI. 1, fig. 7), is only slightly larger than RMS 1989.36.2 and shows complete first to fourth cycles of septa and an incomplete fifth cycle containing 10 septa. The pattern of nested triads is uncertain and probably anomalous in one sextant; the specimen is damaged at this point. A bilateral symmetry is suggested by the 5th cycle occurring almost exclusively in two opposite sextants. RMS 1989.36.1 (Text-fig. 3g, h) is close to the maximum diameter known so far for this coral. It also shows complete first to fourth cycles of septa, whilst the fifth cycle is better developed but still incomplete with 31 septa and the sixth cycle rarely developed and represented by 6 septa. Bilateral symmetry is again suggested by septal arrangement in and around the axial area and by slightly higher septal numbers in two opposite sextants. However, the numerical difference is small and peripheral preservation incomplete so that septal number may be higher than apparent. Only two relatively immature specimens are available (Text-fig. 3a-d). No pattern of convergence is seen in the smallest specimen, RMS 1989.36.3 (Text-fig. 3a, b), and septal identity is uncertain: the interpretation shown draws on comparison with the pattern developed in larger coralla. The larger specimen shows weakly developed triads. We have been conservative in our interpretive sketch (Text-fig. 3c) and faint traces on the specimen suggest the possibility of greater axial extension of the third order septa towards the second order septa (Text-fig. 3d). The appearance of third order septa in more mature specimens suggests that some extension or strengthening of their axial ends takes place as growth proceeds. The details of septal insertion cannot be substantiated by a study of septal grooves on the underside of the epitheca. These are only rarely well-developed and tend to fail almost completely in an axial area c. 4 mm across. Often the whole epitheca appears to lack septal grooves (Text-fig. 2d). Supporting evidence is limited to faint indications of peripheral triads, as at top-right in Text- figure 2e. Many specimens show varying degrees of irregularity in insertion. However, an overall pattern emerges of a pair of adjacent sextants relatively retarded and a further pair of opposite sextants relatively accelerated. The pattern is symmetrical about the plane of bilateral symmetry where this is clear from features in the axial area of the corallum. In the smallest coralla available (Text-fig. 3a-d), retardation is already apparent in the adjacent pair of sextants (orientated towards the 186 PALAEONTOLOGY, VOLUME 34 text-fig. 3. For legend see opposite. SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 187 bottom of each figure). Acceleration in lateral sextants does not become marked until the insertion of the 5th cycle begins. It may be so extreme in some larger corallites that an initial impression is given of eight rather than six sets of nested triads. An idealized representation of the septal pattern in these corals is given in Text-figure 4a. AFFINITIES AND RELATIONSHIPS Anthozoan affinities The gross morphological features of these specimens immediately suggest coralline affinity. The only other reasonable possibility seems to be a relationship to the Porifera, based on a crude homeomorphy with forms like Haplistion (Rigby 1987). No other phylum is known to produce a structure of this size range and form. The microarchitecture of the septa clearly rules out poriferan affinity and strongly supports assignment to the Anthozoa Cnidaria. The characteristic pattern of elements in the septal spines can be matched very closely among the Scleractinia (for example, Sorauf 1972). In particular, the appearance of granulations on the lateral faces of septa in Fungia , representing one spherulitic cluster of crystallites (Sorauf 1972, pi. 14, fig. 5), is indistinguishable in appearance from the tips of the septal spines in the present material, although smaller in size (PI. 1, figs 3 and 4). Granulations on the septal faces in Cladocora (Sorauf 1972, pi. 13, fig. 4) are also similar. The fabric on the lateral faces of septal blades (PI. 1, figs 1 and 2) compares with that in Fungia scutarea (Sorauf 1972, pi. 1 1, fig. 2) but is much coarser. The elongate units defined in the present material are assumed to be composed of bundles of fine acicular crystals, not resolvable here either because of the limitations of the moulding medium, or recrystallization, or both. However, there seems to be sufficient evidence to establish the septal spines as trabeculate. Such microstructure appears to be characteristic of the anthozoan Cnidaria. This evidence, together with the discoidal epithecate form and the radial distribution of the spinose or bladelike septa, clearly identifies this material as a zoantharian coral. Affinities within the Anthozoa In detail, this material is unlike any other known Palaeozoic coral, either from the established Rugosa or Heterocorallia, or the more scattered and problematic Cambrian material. It is grossly most similar to some solitary, discoidal Rugosa (for example, Hill 1981, fig. 39) but is fundamentally distinguished from them by its septal arrangement. These new specimens unequivocally show six- fold cyclic insertion in contrast to the serial insertion in four quadrants of rugose corals.. The septal development in Hexacorallia, based on four primary septa (Hill 1981), is even more distinct. On the other hand, this pattern of cyclic insertion is indistinguishable from that in scleractinian corals (Vaughan and Wells 1943; Wells 1956; Jell 1980; Oliver 1980). The tendency for cycles, particularly above the third, to be incomplete when higher cycles are initiated is common in Scleractinia. The evidence of bilateral symmetry is also seen in septal development in many scleractinians and, as pointed out by Oliver (1980), is a reflection of the fundamental radiobilateral symmetry of all known anthozoans. A dorso-ventral polarity in the insertion of septal cycles is a feature of some scleractinians (Vaughan and Wells 1943; Wells 1956; Oliver 1980) and we interpret the relative retardation of a pair of sextants in the present material to indicate the equivalent of the ventral pole text-fig. 3. Kilbuchophyllia discoidea gen. et sp. nov. Ontogenetic series, Ordovician, mid-Caradoc, Kilbucho, near Biggar, southern Scotland. Photographs (b and d) and scanning electron micrographs (f and h) of latex replicas of calical surfaces are matched with interpretive sketches based on information from original moulds and corresponding replicas. Plane of bilateral symmetry vertical, supposed dorsal pole at top. Septal cycle indicated as follows: protosepta, long heavy lines; 2nd, 3rd and 4th cycles, successively shorter light lines; 5th cycle, spots; 6th cycle, unornamented, a and b, RMS 1989.36.3, x 8. c and d, BGS 9936, x 8. e and f, RMS 1989.36.3, x 6. G and h, RMS 1989.36.1 (holotype), x2-5. PALAEONTOLOGY, VOLUME 34 in scleractinians. In conformity with scleractinian usage, we have orientated the presumed dorsal pole uppermost in the material described here. However, we are not aware of relative acceleration in the pair of opposite sextants in scleractinians. thxt-fig. 4. a, Kilbuchophyllia discoidea gen. et sp. nov. Idealized reconstruction of septal pattern. Plane of bilateral symmetry vertical with supposed dorsal pole uppermost, b, Fungiacyathus symmetricus (Pourtales). BM(NH)Z 1880.1 1.25.123, Recent specimen, collection station details uncertain, x4. The distinctive pattern of nested triads of second and higher orders of septa is similar to a version of Portales Plan which is developed in some scleractinian corals. Pourtales Plan is regarded as a reflection of septal substitution during ontogeny, when the peripheral ends of exosepta split to accommodate subsequent entosepta. Vaughan and Wells (1943, p. 34) stated that it may be assumed that substitution has occurred when septa of a higher cycle unite with those of a lower cycle. A range of patterns of uniting septal ends is possible in detail, but the arrangement in the present material is remarkably similar to that exhibited by such Scleractinia as Fungiacyathus symmetricus (Text-fig. 4b; Vaughan and Wells 1943, pi. 34, figs la and 4) and Balanophyllia ( Eupsammia ) zelandiae (Squires 1958, p. 73, fig. 28). However, this is not identical to the classic pattern in dendrophyllid corals illustrated by Vaughan and Wells (1943, fig. 13) and Wells (1956, fig. 239) in which the entosepta are less well-developed than the exosepta. Also the pattern in the present material is equivocal. The axial septum of a triad is usually more or less structurally continuous and the flanking septa bend towards but do not always touch or merge with the axial septum. This does not immediately suggest the process of substitution. In some cases the peripheral ends of existing septa are deflected around the tips of newly inserted septa, at this stage a string of septal spines, in a manner suggesting substitution. However, we cannot be certain that these instances are not irregularities in insertion rather than clues to its character. If septal splitting did occur, the appearance of septa in the smaller specimens suggests that it was unlikely to have affected either the first or second orders. It seems also that further work is needed on the origin of some of the patterns attributed to Pourtales Plan in living corals. Thus it is premature to claim septal substitution as occurring in this Ordovician material. This very close similarity to the Scleractinia is reflected in other features. The origination of septa as discrete spines, subsequently linked by thin blades of material to give a beaded appearance to the septa, is very reminiscent of the early stages of skeletal development in some Recent corals (see, for example, Jell 1980). The rather confused appearance and irregularities in metasepta insertion in the early ontogenetic stages mentioned and illustrated by Jell are very similar to those seen here. Furthermore, the coarsely denticulate upper margins of the septal blades in mature coralla are also SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 189 closely comparable to those seen in many scleractinian corals but are not a characteristic of the Rugosa. An epitheca or holotheca is almost universal among rugose corals but, as a well developed feature, is confined largely to some ahermatypic (mainly caryophyllid and dendrophyllid) forms among the Scleractinia. In the Rugosa and Tabulata, it appears always to develop from an initial conical structure secreted by the polyp on settlement and metamorphosis, whereas in the Scleractinia it develops on the edge of the basal disc (Jell 1980). In the present material the central area of the epitheca is featureless and flat; there is no sign of a conical stage in development (Text- fig. 2d,e). The microarchitecture of the epitheca shows similarities with that described for Manicina by Sorauf (1972), although the structures preserved here are much coarser in scale. Also, the character of the upper surface conforms closely in appearance to the secondary layer on the surface of the basal disc of Porites lutea illustrated by Jell (1980). These specimens occur with a rich invertebrate fauna, preserved almost exclusively as moulds, the vast majority of which originally had calcium carbonate shells or skeletons. The skeletal material is thus assumed to have been calcium carbonate. Whether the skeleton was originally calcitic or aragonitic is much more speculative. Little is known of microarchitecture in the rugose corals, widely regarded to have been originally calcitic (Sandberg 1984), although internal ultrastructure appears to be identical to that in the Scleractinia. Sorauf (1980, p. 335) considered that biomineralization in the Rugosa closely resembled that in the Scleractinia, differing only in original mineralogy. In any case, in the material described here the finest detail of the microarchitecture is not preserved. The only evidence is indirect; similarity to the Scleractinia is so close that the original mineralogy may well have been the same, that is to say aragonitic. Phylogenetic relationships The evidence suggests very close affinity between this Ordovician material and the Scleractinia among the Zoantharia Anthozoa. It seems highly improbable that intermediates over a period of 220 Ma could all have escaped preservation and/or detection even if it is assumed that these corals remained ecologically confined to oceanic environments. In fact the associated organisms clearly indicate a shelf and/or upper slope fauna. It seems more probable that the Palaeozoic specimens represent an earlier, ultimately unsuccessful attempt at skeletonization by the same group of anemones that later gave rise, probably polyphyletically, to the Scleractinia. Such a conclusion requires the existence of anemones with a cyclic hexameral pattern of mesenterial insertion at least as early as the mid-Ordovician. Thus it strongly supports the rejection of the Rugosa as ancestral to the Scleractinia (Oliver 1980). The ancestral anemone group is usually considered to be the Corallimorpharia, identical to scleractinian polyps but skeletonless (Wells and Hill 1956; Hill 1981; Oliver and Coates 1987). although Hand (1966) has suggested the possibility of the reverse relationship on functional grounds, with the Corallimorpharia and Actiniaria evolved from the Scleractinia by loss of the skeleton. The new Ordovician material, however, appears to favour the former scenario. Furthermore, if its septal pattern can be confirmed to be identical to one known to result from septal substitution in living corals, this isolated skeletonized species would itself presumably require an anemone precursor already possessing paired mesenteries. Thus the range of the Corallimorpharia, and/or the closely related Actiniaria, must be extended back at least that far to provide the same ancestral anemone stock for this and the Scleractinia. Anemones have an almost non-existent fossil record (Scrutton 1979) but it now seems possible that all the various groups of anemones may have differentiated during the initial cnidarian radiation in the late Precambrian. A possible phylogeny for the Palaeozoic Zoantharia Anthozoa, modified after that of Oliver and Coates (1987), is given in Text-figure 5. The present material has the same relationship to the Scleractinia as the Middle Cambrian Cothoniida probably, but perhaps less certainly, has to the Rugosa (Jell and Jell 1976; Scrutton 1979; Oliver and Coates 1987). The latter are regarded as having evolved from the Zoanthiniaria, in which later mesenterial couples are inserted serially in only one pair of sextants (Wells and Hill 1956; Hill 1981). The relationships of the other major group of Palaeozoic corals, the Tabulata (taken to include the Heliolitida) is equivocal. Some or all 190 PALAEONTOLOGY, VOLUME 34 Free. Palaeozoic post-Palaeozoic V € 0 S D C p post T / / I 1 9 1 Cothon r 1 ? iida He r 9 / ‘terocoralli R a ugosa / l Zoa Tabulata nthiniaria Ac itiniaria 1 ? \_Kilbu chophy Cora 'llida limorpharic i 1 1 Scleractinia — ► text-fig. 5. Phylogeny of the Anthozoa Zoantharia (modified after Oliver and Coates 1987). Rugosa have been claimed to have direct tabulate ancestry (Flower 1961) but, although the earliest skeletal ontogenetic stage appears always to be conical as in the Rugosa, we regard this as most unlikely (Scrutton 1979; Neuman 1984). Septa are absent or weakly developed in the Tabulata and no general pattern of insertion has been demonstrated. Tabulate corallites rather seldom show bilateral symmetry; septal development is usually radially uniform and 12 septa are sufficiently common, together with a rare instance of preservation of twelve tentacled favositid polyps, for a fundamental dodecal symmetry to have been claimed for the group (Copper 1985; Mistiaen 1989). These factors suggest a corallimorpharian or actiniarian ancestor to be as, if not more, likely for this group than a zoanthiniarian ancestor among known orders of anemones, although it seems equally possible that the tabulates evolved from a separate group of anemones now extinct. SYSTEMATIC PALAEONTOLOGY Phylum cnidaria Hatschek, 1888 Class anthozoa Ehrenberg, 1834 Subclass zoantharia de Blainville, 1830 Order kilbuchophyllida nov. SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 191 Diagnosis. As for genus. Discussion. The Kilbuchophyllida is homoeomorphic to a high degree with the Scleractinia. However, the combination of a solitary discoidal, epithecate form lacking dissepiments, with solid, bladed septa nested in triads and more or less strongly accelerated in the lateral sextants, does not appear to occur among the Mesozoic to Cenozoic scleractinian corals. Although the combination of characters in the only known species is unique, all, with the possible exception of the pattern of septal acceleration, are individually or severally found in various scleractinians. Ultimately, the classification of this species in a new order is based on its stratigraphic separation and our presumption of lack of direct descent to the Scleractinia. Family kilbuchophyllidae nov. Diagnosis. As for genus. Genus kilbuchophyllia gen. nov. Derivation of name. After the type locality, Kilbucho (pron. -bukko), near Biggar, southern Scotland. Diagnosis. Solitary, discoidal, epithecate radiobilateral corals showing hexameral, cyclic septal insertion. Septa, spinose to solid blades, arranged in a pattern with the internal ends of higher order septa turned towards or resting against the flanks of lower order septa. Adjacent sextants (?ventral pole) with retarded septal insertion in early ontogeny, lateral sextants accelerated in later ontogeny. No dissepiments. Kilbuchophyllia discoidea sp. nov. Plate 1, figs 1-8; Text-figs 2, 3, 4a, 6 Diagnosis. Circular, solitary, discoidal corals with diameter up to 28 mm and estimated maximum 120 septa. Up to six cycles of septa of which the fourth sometimes and the fifth and sixth cycles always are incomplete. Second and higher orders involved in pattern of nested triads. Insertion retarded in adjacent (?ventral) sextants in early ontogeny, accelerated in lateral sextants in later ontogeny. Axial area with discrete trabeculae (?pali) merging to form papillose or contorted axial structure. Weak bilateral symmetry usually apparent. Epitheca a flat disc with concentric growth ridges and occasionally septal grooves. No dissepiments. Holotype. RMS 1989.36.1. Ordovician, middle Caradoc; Kilbucho, near Biggar, southern Scotland. Paratvpes. RMS 1989.36.2-12; BGS 9936. Same horizon and locality as holotype. Description. Solitary, circular, discoidal corals ranging from 2-6 mm diameter with 15 septa to 27-5 mm with estimated 120 septa (Text-fig. 6). In small coralla, septa either discrete trabecular spines or in lower order septa, spines linked by a low thin ridge giving septa a beaded appearance. In larger coralla, spines subsumed in smooth faced blades, c. 0-3 mm thick, with coarsely toothed upper margin in all but highest order septa, although less completely fused peripherally and particularly adaxially. Individual spines c. 0-2 mm diameter with axes c. 0-3 mm apart; septa! teech spaced c. 0- 7-0-9 mm. Height of spines or septal blades 0-75—1 -0 mm in smaller coralla, rising to c. 2-3 mm high in the largest coralla. Septal height greatest at mid length of smaller coralla, migrating to axial end of septa in larger coralla. Six first (protosepta) and six second cycle septa of more or less equal length, 0-8 radius in all but smallest coralla. Higher cycles, up to sixth, successively shorter in length. Second and higher cycles of septa involved in a pattern of nested triads, with higher cycles at their inner ends turned towards or resting against the flanks of lower cycles. The first two cycles complete in smallest available corallum, third cycle complete between 34 mm diameter, fourth cycle complete by about 10 mm, fifth cycle absent from smaller coralla, ?complete in largest coralla, sixth cycle variably present only in largest coralla and never complete. Septal insertion retarded in adjacent sextants (at ?ventral pole of polyp) in early ontogeny, accelerated in lateral sextants about plane of bilateral symmetry in later ontogeny. Axial area with 192 PALAEONTOLOGY, VOLUME 34 discrete trabecular spines, ?equivalent to pali, in smallest coralla. With size increase, spine bases variably embedded to form flat or slightly arched papillose area. In one case, spines linked as extensions of septa to form dome of twisted, interlocked plates. Bilateral symmetry may be weakly defined by a more or less well-developed bladed element in axis but sometimes not obvious. Epitheca a flat disc with peripheral depth 0-3-0- 5 mm high. Central area featureless and may be almost smooth throughout but concentric growth ridges usually and radiating septal grooves sometimes clearly developed around central area. There are no dissepiments. Discussion. Except for the smallest specimens, all the material is elliptical in plan. When apparent, the plane of bilateral symmetry is not coincident with the long axis of the ellipse and the shape is due to tectonic distortion in the rock. Strain analysis yields a value of Ri of T06. Allowing for the difficulty of measuring axes accurately in some of the material, this suggests that the coral was originally effectively circular. 100 - * no. of septa so - t 0 0 10 -J 20 Diameter (mm) 30 text-fig. 6. Kilbuchophyllia discoidea gen. et sp. nov. Plot of septal number against diameter for better preserved material. Both parameters estimated in many cases because of damage to margins of specimens. Holotype indicated by asterisk. Variation in most features in the material available, allowing for ontogenetic stage, is relatively limited. The axial structure is the most variable aspect of mature specimens. One coral, BGS 9936, representing an early ontogenetic stage, is unique in possessing a distinct low rim linking the peripheral ends of septa. Whether or not this is aberrant, or the rim is obscured by thickening of the upper surface of the epitheca in larger coralla, is unknown. The specimens often appear to have suffered some damage before final burial, consistent with their presence in a debris flow. In particular, the septal blades in the larger specimens are often damaged and their upper margins incomplete. Because of incomplete preservation there is an SCRUTTON AND CLARKSON: ORDOVICIAN CORAL 193 element of estimate in all the data on Text-figure 6, although the error is considered unlikely to exceed 10%. Range. This species is known so far from some 20 specimens and fragments from the type locality. Two fragments have been recovered from similar beds of the same age at Wallace's Cast, Wandel Burn, 12 km west- south-west along strike, southern Scotland. Acknowledgements . We are grateful to all those with whom we have discussed various aspects of this study, particularly Stephen Cairns (Smithsonian Institution, Washington, D.C.), Bill Oliver (US Geological Survey, Washington, D.C.), Makoto Kato (Hokkaido University, Sapporo), Keith Rigby (Brigham Young University, Utah), Martin Le Tissier and Graham Young (University of Newcastle upon Tyne). Susan Bruce (University College, Galway) kindly contributed the smallest specimen, which she collected. Brian Turner commented on the matrix to the specimens, scanning electron micrographs were taken by Trevor Booth and Text-figures 1, 3, 5, 6 were drafted by Christine Jeans; Peter Lewis and Brian Tuffs helped with preparation (all University of Newcastle upon Tyne). Simon Moore (Natural History Museum, London) and Peter Brand (British Geological Survey, Edinburgh) kindly arranged the loan of material in their care. REFERENCES blainville, h. m. d. de 1830. Zoophytes. Dictionaire de Sciences Naturelle de Paris , 60, 1-546. clarkson, e. N. k. and tripp, r. p. 1982. The Ordovician trilobite Calyptaulax brongniartii (Portlock). Transactions of the Royal Society of Edinburgh , 72, 287-294. copper, p. 1985. Fossilized polyps in 430-Myr-old Favosites corals. Nature , 316, 142-144. ehrenberg, c. G. 1834. Beitrage zur physiologischen Kenntnis der Corallenthiere im allgemeinen, und besonders des rothen Meeres, nebst einem Versuche zur physiologischen Systematik derselben. Koniglischen Akademie der Wissenschaft , Physiologische-Mathematischen Abhandlung , 1832, 225-380. erina, m. v. and kim, a. i. 1980. On some Ordovician Scleractima-like corals from the south Tien-Shan. Acta palaeontologica polonica , 25, 375-379, pis 18-21. flower, R. H. 1961. Montoya and related colonial corals. Memoirs , New Mexico Institute of Mining and Technology , 7, 1-97, 52 pis. hand, c. 1966. On the evolution of the Actiniaria. Symposium , Zoological Society of London , 16, 135-146. hatschek, b. 1888-91. Lehrbuch der Zoologie, eine morphologische Ubersiclit des Thierreiches zur Einfurung in das Studium dieser Wissenschaft , Lief 1-3. Gustav Fischer, Jena, iv + 432 pp., 407 figs. hill, d. 1960. Possible intermediates between Alcyonaria Tabulata, Tabulata and Rugosa, and Rugosa and Hexacoralla. International Geological Congress , 21, (sect. 22), 51-58. - 1981 Rugosa and Tabulata. xl + Fl-F762, figs 1-462. In teichert, c. (ed. ). Treatise on invertebrate paleontology , Part F, Coelenterata , suppl. 1. Geological Society of America and University of Kansas Press, Boulder, Colorado and Lawrence, Kansas. jell, j. s. 1980. Skeletogenesis of newly settled planulae of the hermatypic coral Porites lutea. Acta palaeontologica polonica , 25, 311-320, pis 5-8. - 1984. Cambrian cnidarians with mineralised skeletons. Palaeontographica americana , 54, 105-109. jell, p. a. and jell, j. s. 1976. Early Middle Cambrian corals from western New South Wales. Alcheringa, 1, 181-195, 12 figs. kim, a. i. 1974. On the phylogeny and systematical position of some tabulatomorpha. In sokolov, b. s. (ed.). Drevnie Cnidaria, 1, 118-122. Nauka, Novosibirsk. krasnov, e. v. 1970. Filogenez i problema tselostnosti gruppy Scleractinia. 15^10, 8 figs. In grigor’eva, a. d. (ed.). Trudy II Vsesoyuznogo Simpoziuma po izuchennyu iskopaemykh korallov SSSR , 4. Mezozoiskie korally SSSR. Nauka, Moscow, 112 pp. [In Russian], leggett, j. k., mckerrow, w. s. and eales, M. H. 1979. The Southern Uplands of Scotland: a Lower Palaeozoic accretionary prism. Journal of the Geological Society , 136, 755-770. minato, M. 1955. Japanese Carboniferous and Permian corals. Journal of the Faculty of Science , Hokkaido University , 7, (4), I -202, pis 1-43. mistiaen, b. 1989. Importance de la symetrie d’ordre douze chez les Tabulata. Comptes rendues de T Academic des Sciences de Paris , 308, ser. II, 451-456, 3 figs. 194 PALAEONTOLOGY, VOLUME 34 mitchell, i. h. 1977. The Ordovician Brachiopoda from Pomeroy, Co. Tyrone. Palaeontographical Society Monograph , 1-138. neuman, b. E. E. 1984. Origin and early evolution of rugose corals. Palaeontographica americana , 54, 119-126, 2 figs. Oliver, w. a., jr. 1980. The relationship of the scleractinian corals to the rugose corals. Paleobiology , 6, 146-160, 8 figs. - and coates, a. G. 1987. Phylum Cnidaria. 140-193, 44 figs. In boardman, r. s., cheetham, a. h. and rowell, a. j. (eds). Fossil Invertebrates. Blackwell Scientific Publishers, Palo Alto, Oxford, etc., x + 515 pp. peach, b. N. and horne, J. 1899. The Silurian rocks of Britain, vol. 1 : Scotland. Memoir of the Geological Survey of Great Britain , xviii + 749 pp. rigby, j. k. 1987. Phylum Porifera. 116-139, 21 figs. In boardman, r. s., cheetham, a. h. and rowell, a. j. (eds). Fossil Invertebrates. Blackwell Scientific Publishers, Palo Alto, Oxford, etc., x + 515 pp. ritchie, m. and eckford, r. j. a. 1935. The Haggis Rock of the Southern Uplands. Transactions of the Geological Society of Edinburgh, 13, 371-377. sandberg, p. a. 1984. Recognition criteria for calcitised skeletal and non-skeletal aragonites. Palaeontographica americana , 54, 272-281, I pi. schindewolf, o. h. 1942. Zur Kenntnis der Polycoelien und Plerophyllen. Reichsamt fur Bodenforschung, Abhandlung, n.s., 204, 1-324, pis 1-36. scrutton, c. t. 1979. Early fossil cnidarians. 161-207. In house, m. r. (ed.). The origin of major invertebrate groups. Academic Press, London and New York, x + 515 pp. - 1988. Patterns of extinction and survival in Palaeozoic corals. 65-88. In larwood, g. p. (ed.). Extinction and survival in the fossil record. Clarendon Press, Oxford, x + 365 pp. sorauf, J. e. 1972. Skeletal microstruture and microarchitecture in Scleractinia (Coelenterata). Palaeontology, 15, 88-107, pis 11-23. 1980. Biomineralisation, structure and diagenesis of the coelenterate skeleton. Acta palaeontologica polonica, 25, 327-343, pis 13-17. smith, s. 1930. The Calostylidae, Roemer: a family of rugose corals with perforate septa. Annals and Magazine of Natural History, (10), 5, 257-278, pis 10-12. squires, d. f. 1958. The Cretaceous and Tertiary corals of New Zealand. New Zealand Geological Survey, Palaeontological Bulletin, 29, 1-107, 16 pis. vaughan, t. w. and wells, J. w. 1943. Revision of the suborders, families, and genera of the Scleractinia. Special Papers , Geological Society of America, 44, xvi+ 1-363, 51 pis. wells, j. w. 1956. Scleractinia. F328^144, figs 222-339. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part F. Geological Society of America and the University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 498 pp. - and hill, d. 1956. Anthozoa - general features. F 1 6 1 — 1 65, fig. 132. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part F. Geological Society of America and the University of Kansas Press, Boulder, Colorado and Lawrence, Kansas, 498 pp. weyer, d. 1973. Uber den Ursprung der Calostylidae Zittel 1879 (Anthozoa Rugosa, Ordoviz-Silur). Freiberger Forschungshefte, C282, 23-87, 15 pis. COLIN T. SCRUTTON Department of Geology The University Newcastle upon Tyne NE1 7RU, UK Present address: Department of Geological Sciences University of Durham South Road, Durham DH1 3LE, UK EUAN N. K. CLARKSON Typescript received 8 November 1989 Revised typescript received 7 March 1990 Grant Institute of Geology West Mains Road Edinburgh EH9 3JW, UK THE TAXONOMY AND SHELL CHARACTERISTICS OF A NEW ELK AN 1 1 D BRACHIOPOD FROM THE ASHGILL OF SWEDEN by LARS E. HOLMER Abstract. A new elkaniid brachiopod genus and species, Tilasia rugosa , is described from the Ashgill (Harju Series) Boda Limestone in the Siljan district (province of Dalarna), Sweden. It is the first record of the lingulacean family Elkaniidae from the Upper Ordovician. The material of T. rugosa , which is one of the largest described member of the family, is well preserved and allows an account of the micro-ornamentation and shell structure. The strongly rugose exterior has a divaricate ornamentation with minute rhomboid pits, previously not known among the elkaniids. Elkaniid brachiopods are common and widely distributed mainly in the Upper Cambrian and Lower Ordovician (Tremadoc-lower Llanvirn); the family has not previously been recorded from beds younger than the Middle Ordovician. Here a new genus and species, Tilasia rugosa from the Upper Ordovician (Ashgill) Boda Limestone in the Siljan district, province of Dalarna, Sweden (Text-fig. 1), is described. The rare but well preserved material of this large elkaniid also permits an account of the shell structure and micro- ornamentation, not previously known from this group. MATERIALS AND METHODS In the Siljan district. Lower Palaeozoic (Upper Cambrian? to Silurian) rocks crop out within a tectonically complex ring-structure, which probably represents a hypervelocity impact crater (Text- fig. 1; see Jaanusson 1982 for a review). The Boda Limestone (within the Amorphognathus ordovicicus Biozone: Bergstrom 1971) is a large (maximum diameter, 1000m; thickness, 140m), lens-shaped, stromatactis-bearing unit, with a high carbonate content. Although reef-like, it lacks an organic frame; it represents a carbonate mound, possibly comparable with modern lithoherms (Jaanusson 1979, 1982). The phosphatic inarticulates (discinaceans) from these beds have previously been described by Lindstrom (in Angelin and Lindstrom 1880) and Holmer (1987). The stratigraphy and fauna of the Boda Limestone were summarized by Jaanusson (1958, 1982). The material was prepared from the limestone by etching with 10% buffered acetic acid (see Jeppsson et al. 1985 for details); the outer or inner surfaces of the valves were covered with a layer of epoxy resin to avoid fragmentation during the etching process. To study shell structure, specimens embedded in epoxy resin were sectioned, polished and subsequently etched with 4% hydrochloric acid for 4 seconds; the counterparts of the sectioned valves were used to make thin sections for examination in transmitted light. The type material is housed in the Department of Palaeozoology, Swedish Museum of Natural History (SMNH), and in the Department of Geology, University of Lund (LO). Detailed descriptions of the localities (Ostbjorka, Boda, Jutjarn, and Skalberget; Text-fig. 1) in the Siljan district are given by Thorslund (1936; see also Jaanusson 1982). (Palaeontology, Vol. 34, Part 1, 1991, pp. 195— 204.| © The Palaeontological Association 196 PALAEONTOLOGY, VOLUME 34 text-fig. 1. Location map of the Siljan district, province of Dalarna, Sweden, showing the ring-structure with Lower Palaeozoic rocks (shaded) and the localities investigated (filled circles). 1, Jutjarn; 2, Ostbjorka; 3, Skalberget. SYSTEMATIC PALAEONTOLOGY Class lingulata Goryansky and Popov, 1985 Order lingulida Waagen, 1885 Superfamily lingulacea Menke, 1828 Family elkaniidae Walcott and Schuchert, 1908 Diagnosis. See Rowefl (1965, p. H270). Genera assigned. Monobolina Salter, 1866; Elkania Ford, 1886; Broeggeria Walcott, 1902; Lamanskya Moberg and Segerberg, 1906 [= IDictyobolus Williams and Curry. 1985]; Elkanisca Havhcek, 1982; Tilasia gen. nov. Discussion. The detailed morphology of many of the elkaniid genera listed above remains poorly known, perhaps partly because they have usually been described from material from argillaceous sequences (e.g. Broeggeria, Monobolina , Elkanisca)', well preserved complete specimens from carbonates have generally not been available. The elkaniid affinity of Monobolina was recently questioned by Havlicek (1982, p. 50). However, the new data on the morphology of M. plumbea (Salter) presented by Lockley and Williams (1981, p. 15, figs 31-34) indicates that it belongs within the Elkaniidae. Lockley and Williams (1981) also HOLMER: SWEDISH ELKANIID BRACHIOPOD 197 described the new species M. crassa, which extended the range of the family into the Middle Ordovician (Llandeilo). The poorly known Lower Ordovician (Tremadoc) Lamanskya Moberg and Segerberg, 1906, from Oland, Sweden, was previously placed questionably among the Strophomenidina (Williams 1965, p. H863), but is now considered to be an elkaniid (Holmer 1989); the type (and only) species, L. splendens Moberg and Segerberg, is widely distributed in the Lower Ordovician of Sweden, and is currently being redescribed. The Irish Lower Ordovician genus Dictyobolus Williams and Curry, 1985 (type species D. transversus Williams and Curry), which is here referred to the elkaniids, appears to be a junior synonym of Lamanskya (Holmer unpublished). The likewise poorly known Aulonotreta kuraganica Andreeva, 1972 from the Lower Ordovician of the Ural Mountains probably also represents a new genus of the elkaniid brachiopods (L. E. Popov, personal communication 1989). Genus tilasia gen. nov. Type species. Tilasia rugosa sp. nov. Etymology. In honour of Daniel Tilas (1712-1772), who published the first detailed account of the Lower Palaeozoic strata of Dalarna (Tilas 1740). Diagnosis. Large, transversely suboval, moderately and subequally biconvex, rugose shell; exterior pitted with rhomboid pits. Ventral pseudointerarea with wide propareas and deep, triangular pedicle groove; ventral umbonal muscle scar divided by anteriorly directed extension of the pedicle groove. Dorsal pseudointerarea with wide median groove and narrow propareas. Species assigned. Tilasia rugosa sp. nov.;? Obolus ? sp. 3 Cooper, 1956. Tilasia rugosa sp. nov. Text-figs 2-5 Holotype. SMNH Bi 1 33686, almost complete shell (width 26-6 mm, length 22-4 mm, thickness 10 0 mm) from the Boda Limestone, Jutjarn quarry, Siljan district, Dalarna (coll. M. Frye). Paratypes. All material from the Boda Limestone, Siljan district, Dalarna; SMNH Br 133691, incomplete dorsal valve, Skalberget quarry (coll. E. Jarvik; flank facies; locality 8 in Jaanusson 1982, fig. 3; SMNH Brl02556o, incomplete dorsal valve (previously identified as fragmentary dorsal valve of Orbiculoidea ? gibba in Holmer 1987, p. 320), Skalberget quarry (flank facies; coll. J. Martna); LO 5956, incomplete ventral valve, Ostbjorka (coll. S. L. Tornquist); LO 5957 (not figured), incomplete ventral valve, Boda (coll. S. L. Tornquist). Total of two dorsal and two ventral valves. Etymology. Latin rugosus , wrinkled; alluding to the rugose ornamentation. Diagnosis. As for genus. Description. Shell large (up to 26 6 mm wide and 22-4 mm long in one specimen), and moderately, subequally biconvex, 38% as thick as wide (Text-figs 2a and 3e); transversely suboval in outline. Ornamentation strongly rugose (see also below) with regularly disposed, up to 0-5 mm high rugae, on average 05 mm apart (Text-figs 3a, d, f, i and 4a-f). Ventral valve (of holotype) 1 mm longer than dorsal valve, 84% as long as wide, but less convex (about I mm difference), 15% as high as wide (Text-figs 2a and 3a, e). Interior of ventral valve not known in detail; ventral pseudointerarea 14-16 mm wide (in two specimens), occupying 50% of valve width, with well developed propareas, F3 mm wide; deep, triangular pedicle groove, 4-3 mm wide and 1-6 mm long; ventral umbonal muscle scar divided by anteriorly directed extension of pedicle groove (Text-figs 2b and 4g— i). An 198 PALAEONTOLOGY, VOLUME 34 text-fig. 2. Tilasia rugosa sp. nov. a, lateral profile of complete shell, based on SMNH Br 133686. b, ventral interior, based on LO 5956. c, dorsal interior, based on SMNH Brl 33686. All x 6. U, umbonal muscle scar; VL, vascula lateralia\ C, central muscle scar; PL, platform; A, anterior lateral muscle scar; VM, vascula media. unfigured, poorly preserved fragment of a ventral valve (No. LO 5957) shows a section through an elevated platform, directly anterior to the umbonal muscle scars, but the detailed morphology of the platform is not known. Dorsal valve (of holotype) 80% as long as wide, and 23% as high as wide (Text-figs 2a and 3b, e). Dorsal pseudointerarea with median groove, 5 6 mm wide and 0 6 mm long; exact dimensions of propareas unknown, but they appear to be narrower than the ventral ones (Text-figs 2c and 3g-h). Dorsal umbonal muscle scar situated directly anterior to median groove; central and anterior lateral muscle scars situated on an elevated, subtriangular platform, 7 mm wide and 9 mm long, with low median septum; well-developed vascular markings with vascula lateralia diverging anterolaterally from umbonal muscle scar, and vascula media diverging anterior to anterior lateral muscle scars (Text-figs 2c and 3g-h). Remarks on ontogeny. All the examined specimens represent adults. The early ontogeny of T. rugosa is not known; the apical region of the valves is fragmentary. The regularly shaped, biconvex shells do not show any major interruptions or changes in the growth pattern during the juvenile and adult stages; the major concentric rugae are formed at regular intervals. In an early part of the juvenile stage (when the shell is up to 3 mm wide and 2 mm long) the rugae are densely spaced, about 016 mm apart; during later growth stages they become gradually more widely spaced, up to 0-8 mm apart; a fully grown shell appears to have up to about fifty major rugae (Text-fig. 3a, d, f). In some specimens there are minor, more irregular rugae between the major ones (Text-fig. 4a-c). Discussion. Tilasia rugosa differs from mot other elkaniids (such as species of Broeggeria, Elkanisca , and Monobolina) mainly in being more biconvex and strongly rugose. It is most similar to species of Elkania. However, T. rugosa differs in being less biconvex and more rugose; the thickness of the type species E. desiderata (Billings) (Rowell 1965, p. H270, fig. 164: 1 a-c) is about two-thirds of its width and most species of Elkania , like E. hamburgensis (Walcott), are smooth, having only weakly developed growth lines (Rowell 1965, fig. 164: 1 d-f). Lamanskya splendens Moberg and Segerberg, 1906 (p. 71, pi. 3:17) and ‘ Aulonotreta' kuraganica Andreeva, 1972 (p. 46, pi. 7: 1-3) differ in being more strongly biconvex; the thickness of the latter is up to three-quarters of its width; moreover, the dorsal platforms of these two species are much higher (Holmer unpublished; Andreeva 1972, pi. 7: 3). HOLMER: SWEDISH ELKANIID BRACHIOPOD 199 text-fig. 3. Tilasia rugosa sp. nov., Boda Limestone (Ashgill), Siljan district, Dalarna. a-h, holotype, complete shell, Jutjarn, SMNH Brl33686; A, ventral exterior, x2; b, internal mould of dorsal valve, x2; c, posterior profile, x2; d, oblique posterior view of ventral valve, x2-5; E, lateral profile, x2; f, oblique lateral view of ventral valve, x 2-5; G, detail of B, x 2-8; H, detail of latex cast of B, x 3-4. i, exterior of incomplete dorsal valve, Skalberget, SMNH Brl02556a, x2-5. T. rugosa is comparatively large for the family, the maximum width being almost 27 mm. Most other elkaniids (such as Broeggeria , Elkania , and Elkanisca) are generally up to 10 mm wide; only Monobolina crassa (maximum width 23 mm) and ' Aulonotreta' kuraganica (maximum width 25 mm) are more than 20 mm wide. Cooper (1956, p. 193, pi. 9f: 16, 11a: 1) described a large, unnamed obolid, Obolusl sp. 3, from the Middle Ordovician Pratt Ferry beds of Alabama, USA. The interior of this species is unknown, but the strongly rugose exterior, and the general shape of the shell indicate that it might possibly be related to Tilasia. As noted above, T. rugosa is the youngest described elkaniid and the first record of the family from the Upper Ordovician (Harju Series). Remarks on autecology. The type of environment in which T. rugosa lived is uncertain. The holotype is a complete, articulated shell, which has probably not been transported for any great distance after death, but its exact location within the carbonate mound is not known. The mound core of the Boda Limestone is generally poor in sedentary macro-organisms, and has dominantly a vagile fauna of trilobites, gastropods, cephalopods, and pelecypods (Jaanusson 1982, p. 28). Two dorsal valves, which were collected from the flank facies of the mound, are fragmentary and may have been transported. Although many fossil lingulaceans appear to have been infaunal burrowers comparable with their Recent representatives, this is an unlikely mode of life for T. rugosa. The following characters makes it comparatively poorly adapted for burrowing (see Bassett 1984 and Savazzi 1986 for a detailed 200 PALAEONTOLOGY, VOLUME 34 discussion of this life strategy): (1) The moderately biconvex shell is transversely suboval and wider than long (rather than elongate and ‘streamlined’ as in Lingula). (2) The visceral area and the sites of muscular attachments are more posteriorly placed as compared with other lingulaceans (the dorsal anterior lateral muscle scars are placed at about 40% the valve length from the posterior margin in T. rugosa, whereas in Lingula , for example, the ratio is about 60-70%). (3) The ornamentation is strongly rugose (rather than smooth, or with burrowing sculptures). Thus, T. rugosa was probably better adapted to some kind of epifaunal mode of life. The pedicle foramen appears to have remained open throughout ontogeny. Occurrence. T. rugosa is restricted to the Ashgill Boda Limestone of Dalarna. MICRO-ORNAMENTATION Under the SEM, the etched rugose exterior of two dorsal valves revealed a regular pattern of pits covering the post-larval surface (Text-fig. 4a-f). The apical region of the valves is fragmentary, and the ornamentation is not known from this part of the shell. The pits are evenly distributed and closely packed, less than 10 /mi deep, subequal in size and shape, elongate rhomboid, up to 100 /mi long and 30 jum wide, with the largest dimension arranged text-fig. 4. Tilasia rugosa sp. nov., Boda Limestone (Ashgill), Siljan district, Dalarna. a, exterior of incomplete dorsal valve, the location of B indicated, Skalberget, SMNH Brl 33691, x 5. b, c, d, details of a, x 19, x 60, x 150, respectively, e, exterior of partly exfoliated, incomplete dorsal valve (see also Text-fig. 3i), Skalberget, SMNH Brl02556a, x 5. f, detail of E, x 196. G, interior of incomplete ventral valve, Ostbjorka, LO 5956, x 10. h, detail of G, x 20. i, oblique lateral view of G, x 15. HOLMER: SWEDISH ELK AN 1 1 D BRACHIOPOD 201 perpendicular to the direction of growth (Text-fig. 4c). The geometry of the ornamentation could not be investigated in detail, owing to the considerable degree of fragmentation and exfoliation in the two available valves. However, the pits appear to be arranged in offset radiating rows (sensu Wright 1981, p. 446). Each rhomboid pit is defined by two pairs of parallel ridges (each up to 5 pm wide), which are disposed obliquely across the valve surface and intersect at about 30-40°. This type of sculpture is very suggestive of the so-called divaricate pattern of ornamentation, which is responsible for a wide range of sculptures (including burrowing terraces) in molluscs and arthropods, but it has also been reported from some lingulacean brachiopods (see Seilacher 1972 and Savazzi 1986 for reviews). A divaricate ornamentation of pits has not previously been reported from the elkaniids, but the Lower Ordovician species Dictyobolus [= ILamanskya] transversus Williams and Curry (1985, p. 189, figs 2-7) and Lamanskya splendens Moberg and Segerberg have an essentially identical type of ornamentation; a similar type of sculpture also appears to be developed in ‘ Aulonotreta' kuraganica Andreeva. As noted above, these taxa are here considered to belong within the family (Holmer, unpublished). Ornamentation comparable to that of the elkaniids is also known from three other brachiopod groups: (1) The problematic articulate brachiopod Dictyonella has rhomboid pits, very similar to those of Tilasia and arranged in a strict divaricate geometry (see Wright 1981 for a detailed discussion); however, this brachiopod is not otherwise comparable with the elkaniids. (2) Rhomboid, post-larval pits, only some 6 pm across, and arranged in divaricate rows have been described by Popov et al. (1982, fig. 1: 2) and Holmer (1986, fig. 40) from the thin-shelled Ordovician lingulacean Paterula. In the paterulids, the larval shell is also pitted, with minute, circular, cross-cutting pits, about 2-4 pm across, which are closely comparable with the larval pits of most acrotretaceans (see Biernat and Williams 1970). Popov et al. (1982, p. 103) suggested that both the larval and post-larval pits of Paterula represent moulds of a vesicular periostracum, as in the ‘bubble raft’ model originally proposed for the acrotretacean larval shell (Biernat and Williams 1970). It is entirely possible that the post-larval pits of Tilasia represent a cast of similar structures in the periostracum (see also Williams 1990). (3) Most paterinids (like Dictyonina and Micromitra) appear to have divaricate types of post-larval pitted ornamentation, whereas the larval shell is smooth (e.g. Rowell 1965); in Dictyonites and Lacunites , there are rounded, open perforations, 20-200 pm across, which penetrate the valves (Cooper 1956; Wright 1981 ; Holmer 1986, 1989), and the problematic phosphatic brachiopod Volborthia (sometimes doubtfully referred to the paterinids) possesses some kind of pitted, divaricate ornamentation, which has not been studied closely (Ushatinskaya et al. 1988, pi. 6: 6a). Other types of pitted post-larval ornamentation have been reported and discussed by Wreight (1981), Savazzi (1986), and Holmer (1986, 1987, 1989). SHELL STRUCTURE Because of the limited material available, only a single fragment of the postero-lateral portion of a dorsal valve was sectioned (Text-fig. 5). The rugose exterior of this fragment is still covered by the calcareous matrix of the Boda Limestone (Text-fig. 5a). The pitted ornamentation, described above, is developed in the outermost primary layer, which is only about 10 pm thick (Text-fig. 5e). In etched sections examined under the SEM, it has a densely granular appearance, but the size of individual apatite granulae could not be determined, and the layer appears to lack birefringence. The boundary to the secondary layer is not well defined (Text- fig. 5e), and the primary layer is not easily ‘peeled off’ as in some discinaceans (Holmer 1987). The secondary layer is primarily built up of laminae, up to 0-3 mm thick, which are roughly wedge-shaped in section, and inclined at a low angle to the outer valve surface. The laminae have a porous appearance both under the SEM and the light microscope, and possess a well-developed baculate structure (sensu Holmer 1989), with criss-crossing slender apatite baculae, about 1-2 //m across (Text-fig. 5b, c). The detailed internal structure of the baculae could not be determined; they 202 PALAEONTOLOGY, VOLUME 34 text-fig. 5. a. Polished and etched section through a fragment of a dorsal valve of Tilasia mgosa sp. nov., the location of b and e indicated, Boda Limestone (Ashgill), Skalberget, Siljan district, Dalarna, SMNH Br 102556c, x 27. b, detail of a, the location of c is indicated, x 180. c, detail of B, x 750. D, detail of c, x 2250. e, detail of a, x 1660. F, detail of E, x 5900. are covered by numerous minute apatite granulae, which sometimes are cube-shaped, up to 0-5 /mi across (Text-fig. 5d); these structures are possibly related to secondary cyrstal growth during diagenesis. In the inner part of each lamina the interbacular spaces are empty, which causes the baculae to stand out in relief in etched sections; on the outer zone, directly beneath the primary layer, these spaces appear to be filled by a granular apatite matrix (Text-fig. 5b-d, f). The thick baculate laminae are separated by thin, homogenous lamellae, consisting of minutely granular apatite (Text-fig. 5b). The apatite of the secondary layer is strongly birefringent, and the main preferred orientation of the c-axes appears to be roughly normal, or at a high angle to the laminae; only in some of the thin granular lamellae are there indications of a different preferred oaxis orientation, parallel relative to the lamellae. The shell structure of Tilasia is nearly indentical to that of other Lower Palaeozoic lingulaceans discussed by Holmer (1989). The shell structure of most Lower Palaeozoic lingulaceans can be interpreted in the light of what is now known about Recent Glottidia , which has a well-defined primary layer and a baculate structure penetrating the organic laminae of the secondary layer (see Iwata 1982; Watabe and Pan 1984; Pan and Watabe 1988 for details). Holmer (1987) and Ushatinskaya et al. (1988) noted that the shell structure of fossil discinaceans is comparable with that of the lingulaceans, and that they can also be compared with their Recent representatives (see Iwata 1982). Ushatinskaya et al. (1988, p. 49; see also Hewitt 1980; Popov and Ushatinskaya 1986; HOLMER: SWEDISH ELKANIID BRACHIOPOD 203 Ushatinskaya and Zezina 1988) suggested that the shell structures present in both the fossil and Recent phosphatic brachiopods could have been formed by a complete post-mortem redistribution of phosphate, and phosphatization of the organic matter in the shell. One of the main reasons for this proposal seems to be that phosphatic, rod-like strutures, somewhat similar to the brachiopod baculae, have been described by Hewitt and Stait (1985) from the phosphatized connecting rings of some Ordovician cephalopods. There are two kinds of rod-like structures present in the cephalopod connecting rings; the first type apparently represent secondarily phosphatized spicules, originally consisting of aragonite (Hewitt and Stait 1985, figs 5 and 7), whereas the second type is formed by 'dendritic granular crystals on the interior of the connecting ring’ (Hewitt and Stait 1985, fig. 2). For obvious reasons, the phosphatized aragonite spicules are most unlikely to be comparable with the baculae described from lingulacean and discinacean brachiopods. The second irregular, dendritic pattern of granular apatite ‘rods’ appears to have grown in contact with the surface represented by the connecting ring, rather than representing isolated criss-crossing rods as in the lingulacean baculae. Moreover, the sections of Recent Glottidia , examined by Iwata (1982), Watabe and Pan (1984), and Pan and Watabe (1988) were prepared using freshly killed specimens; it is highly unlikely that a complete redistribution of phosphate could have occurred in these specimens as was suggested by Ushatinskaya et al. (1988). Acknowledgements . This study was carried out at the Department of Palaeozoology, Swedish Museum of Natural History, Stockholm. I am grateful to Lennart Andersson (Stockholm), who prepared the art work and to Uno Samuelsson (Stockholm), who did the dark room work. Kristina Lindholm (Lund) and Louis Liljedal (Lund) kindly arranged the loans from the Tornquist collection (Department of Historical Geology and Palaeontology, University of Lund). I am also grateful to Valdar Jaanusson (Stockholm), Stefan Bengtson (Uppsala), and Sir Alwyn Williams (Glasgow) who offered comments on the manuscript. The work was supported by a grant from the Swedish Natural Science Research Council. REFERENCES andreeva, o. n. 1972. Brachiopods from the Ordovician Kuragan Suite in the southern Urals. Paleon- tologicheskij Zhurnal , 1972, 45-56. [In Russian], angelin, N. p. and lindstrom, G. 1880. Fragmenta Silurica e dono Caroli Henrici Wegelin. Samson and Wallm, Holmiae [Stockholm], 60 pp. bassett, M. G. 1984. Life strategies of Silurian brachiopods. Special Papers in Palaeontology, 32, 237-263. bergstrom, s. m. 1971. Conodont biostratigraphy of the Middle and Upper Ordovician of Europe and eastern North America. Geological Society of America Memoir, 127, 83-157. biernat, g. and williams, a. 1970. Ultrastructure of the protegulum of some acrotretide brachiopods. Palaeontology, 13, 491-502. cooper, G. a. 1956. Chazyan and related brachiopods. Smithsonian Miscellaneous Collection, 127, 1-1245. havlicek, v. 1982. Lingulacea, Paterinacea and Siphonotretacea (Brachiopoda) in the Lower Ordovician sequence of Bohemia. Sbornik geologickych ved, Paleontologie, 25, 9-82. hewitt, r. a. 1980. Microstructural contrasts between some sedimentary francolites. Journal of the Geological Society of London, 137, 661-667. — and stait, b. 1985. Phosphatic connecting rings and ecology of an Ordovician ellesmerocerid nautiloid. Alcheringa, 9, 229-243. holmer, l. e. 1986. Inarticulate brachiopods around the Middle-Upper Ordovician boundary in Vastergotland. Geologiska Foreningens i Stockholm Forhandlingar , 108, 97-126. 1987. Discinacean brachiopods from the Ordovician Kullsberg and Boda limestones of Dalarna, Sweden. Geologiska Foreningens i Stockholm Forhandlingar , 109, 317-326. — 1989. Middle Ordovician phosphatic inarticulate brachiopods from Vastergotland and Dalarna, Sweden. Fossils and Strata , 26, 1-172. iwata, k. 1982. Ultrastructure and calcification of the shells in inarticulate brachiopods. Part 2. Ultrastructure of the shells of Glottidia and Discinisca. Journal of the Geological Society of Japan, 88, 957-966. [In Japanese], 204 PALAEONTOLOGY, VOLUME 34 jaanusson, v. 1958. Leptaena limestone. 189-191. In magnusson, n. h. (ed.). Lexique stratigraphique international , l(2c), 1-498. 1979. Carbonate mounds in the Ordovician of Sweden. Izvestya Akademii Nauk Kazakhckoj SSR , Seriya Geologicheskaya , 4, 92-99. [In Russian]. 1982. Ordovician in Dalarna. 15-42. In bruton, d. l. (ed.). Field excursion guide. IV International Symposium on the Ordovician System. Paleontological Contributions from the University of Oslo, 279, 1-217. jeppsson, l., fredholm, d. and mattiasson, b. 1985. Acetic acid and phosphatic fossils - a warning. Journal of Paleontology , 59. 952-956. lockley, m. G. and williams, a. 1981. Lower Ordovician Brachiopoda from mid and southwest Wales. Bulletin of the British Museum of Natural History, (Geology), 35. 1-78. moberg, J. c. and segerberg, c. o. 1906. Bidrag till kannedomen of ceratopygeregionen med sarskild hansyn till dess utvecklmg i Fogelsangstrakten. Lunds Universitets Arsskrift, 2. 1-116. pan, c.-m. and watabe, n. 1988. Shell growth of Glottidia pyramidata Stimpson (Brachiopoda: Inarticulata. Journal of Experimented Marine and Biological Ecology, 119, 45-53. popov, l. e. and ushatinskaya, g. t. 1986. On secondary changes in the microstructure of calcium-phosphatic shells of inarticulate brachiopods. Izvestiya Akademii Nauk SSSR, Seriya Geologicheskaya, 10. 135-137. [In Russian]. — zezina, o. n. and nolvak, j. 1982. Microstructure of the apical parts of inarticulates and its ecological importance. Byuleten Moskovskogo obshchestva ispytatelej priody, otdel biologicheskij, 87. 94—104. [In Russian], rowell, a. J. 1965. Inarticulata. H260-H269. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part H. Brachiopoda 1{2). Geological Society of America and University of Kansas Press, Lawrence, Kansas, 927 pp. savazzi, E. 1986. Burrowing sculptures and life habits in Paleozoic lingulacean brachiopods. Paleobiology , 12, 46-63. seilacher, a. 1972. Divaricate patterns in pelecypod shells. Lethaia, 5. 325-343. thorslund, p. h. 1936. Siljansomradets brannkalkstenar och kalkindustri. Sveriges Geologiska Undersokning , Series C, 398. 1-64. tilas, d. 1740. Mineral-Historia ofwer Osmunds-berget uti Rattwiks Sochn och oster-Dalarne af Daniel Tilas. Swenska Wetenskaps Academiens Handlingar for Manaderna Januar. Februar. Martius , 1, 202-209. ushatinskaya, g. t. and zezina, o. n. 1988. On the probable post-mortem redistribution of phosphatic matter in older inarticulate brachiopods. Doklady AN SSSR, 300, 700-703. [In Russian], - zezina, o. n., popov, l. e. and putivtseva, n. v. 1988. Microstructure and mineral composition of brachiopods with calcium phosphate shells. Pcdeontologicheskij Zhurnal, 1988, 45-55. [In Russian], watabe, n. and pan, c.-m. 1984. Phosphatic shell formation in atremate brachiopods. American Zoologist, 24. 977-985. williams, a. 1965. Lamanskya Moberg and Segerberg, 1906. H863 . In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part H. Brachiopoda 2(2). Geological Society of America and University of Kansas Press, Lawrence, Kansas, 927 pp. - 1990. Biomineralization in the lophophorates. 67-82. In carter, j. g. (ed.). Skeletal biomineralization : patterns , processes and evolutionary trends. Van Nostrand Reinhold, New York, 399 pp. — and curry, G. b. 1985. Lower Ordovician Brachiopoda from the Tourmakeady Limestone, Co. Mayo, Ireland. Bulletin of the British Museum of Natural History, (Geology), 38. 183-269. wright, a. d 1981. The external surface of Dictvonella and of other pitted brachiopods. Palaeontology , 24. 443-481. L. E. HOLMER Institute of Palaeontology Box 558 S-751 22 Uppsala, Sweden Typescript received 8 December 1989 Revised typescript received 28 February 1990 CUTICULAR ULTRASTRUCTURE OF THE TRILOBITE ELLIPSOCEPHALUS POLYTOMUS FROM THE MIDDLE CAMBRIAN OF GLAND, SWEDEN by J. E. DALINGWATER, S. J. HUTCHINSON, H. MUTVEI and D. J. SIVETER Abstract. Hand specimens and polished sections of the cuticle of the trilobite Ellipsocephalus polytomus Linnarsson from the Middle Cambrian of Oland, Sweden have been examined in incident light and, after etching, with the scanning electron microscope. A thin (25-50 //m) outer layer comprises about twenty lamina units; the structure of these units is interpreted as representing the original inorganic material of the cuticle, and therefore also reflecting the structure of the original organic template. X-ray microanalysis strongly suggests that this outer layer is now composed of calcium phosphate. Cavities, 15 pm in diameter, in the outer layer connect to 3 pm diameter canals which extend across the principal layer of the cuticle: these resemble the gland ducts of a Recent millipede. Pore canal pathways may be represented by elongate openings on the undersurface of the outer layer, and structures resembling the interprismatic septa of Recent decapod crustaceans are seen in angled slices. Other primary microstructures identified are relict organic material and fibres which may have bound together the major layers of the cuticle. Horizontal tubules on the undersurface of the outer layer are possibly infilled borings of cyanobacteria. Major subdivisions of Ellipsocephalus cuticle in life are proposed as: a very thin outermost epicuticle, an outer laminated layer, and a principal layer, the original structure of which is represented only by disc-like extensions on the perpendicular canals which pass across it. Trilobite cuticular microstructure has been extensively investigated over the past twenty years (see Dalingwater 1973; Teigler and Towe 1975; Dalingwater and Miller 1977, Stormer 1980; Wilmot and Fallick 1989; Wilmot 1990u), yet our knowledge of the overall structure of the cuticle is far from complete, and the only detailed information on ultrastructure of lamina units has been provided by Mutvei (1981) from a Flexicalymene species from the upper Ordovician of Iowa. In this paper we describe ultrastructural detail from an outer cuticular layer of the Middle Cambrian trilobite Ellipsocephalus polytomus Linnarsson, 1877, from Enerum, Oland, Sweden, which is superior to anything previously reported from any trilobite cuticle. We analyze our observations in relation to Recent arthropod material, assess the implications for views on the overall structure of the trilobite cuticle and outline areas for further investigation. MATERIALS AND METHODS The collections of the Naturhistoriska Riksmuseet in Stockholm contain specimens of the Middle Cambrian trilobite genus Ellipsocephalus preserved in different lithologies: shales, limestones and even conglomerates. However, the best-preserved material seems to be in the Middle Cambrian glauconitic limestones from Enerum and Borgholm on the Baltic island of Oland. Specimens of Ellipsocephalus polytomus from these localities are almost exclusively cranidia, although the collections also include a few complete dorsal exoskeletons. A series of pieces of glauconitic limestone containing cranidia of Ellipsocephalus polytomus and Pciradoxides sp. fragments collected by Westergard from ‘a boulder at Enerum, Oland in 1930’ were selected for study. The surface of the cuticle of Ellipsocephalus was examined and photographed in incident light. IPalaeontology, Vol 34, Part 1, 1991, pp. 205-217, 3 pls.| © The Palaeontological Association 206 PALAEONTOLOGY, VOLUME 34 Slices of the limestone, approximately 2 mm thick, were cut away from blocks of material with a thin high-speed diamond wheel, after one face had been flattened on rotating wheels covered with carborundum-impregnated papers and lubricated with water, and polished to a mirror finish using ultra-fine diamond pastes on felt buffing wheels. Material prepared in this way was examined under a stereo binocular microscope in incident light. Further slices prepared in a similar fashion were etched in a supersaturated aqueous solution of ethylenediaminotetracetic acid (disodium salt) for up to three hours, with the etching process observed from time to time under a stereo binocular microscope in incident light. Etched slices were carefully washed in de-ionised water, air dried, gold sputter-coated and examined with a Cambridge S360 scanning electron microscope (SEM) under optimum conditions for high resolution (short working distance, high accelerating voltage, small aperture size, small spot size). In all, eleven etched preparations were made for SEM examination; each preparation contained at least two and as many as five sections of Ellipsocephalus cuticle, as well as those of Paradoxides sp. Most of the slices were deliberately cut in such a way that the cuticle was sectioned more or less perpendicular to the cuticle surface, though a few angled slices were accidentally produced and a few deliberately achieved. In addition, an accidental but fortuitous break of cuticle along a low angle from the horizontal was made; part and counterpart of this break were examined unetched with the SEM. Two further preparations were carbon coated and analysed with the LINK system of X-ray microanalysis attached to a Cambridge S360 SEM. All preparations are stored with their parent specimens (Ar 46218a-/) in the Sektionen for Paleozoologi, Naturhistoriska Riksmuseet, Stockholm (RM). DESCRIPTION OF THE CUTICLE Terminology We follow Dennell's (1973) terminology for horizontal laminations of the cuticle: each lamina unit is considered to consist of a narrower lamina and a wider inter-lamina. This contrasts with the view of Bouligand (1965) who considered the lamination of arthropod cuticles as an artefact resulting from the sectioning of horizontal sheets of fibres with fibre orientation changing from one sheet to the next. (For a more detailed discussion see Dalingwater and Mutvei 1990). Hand specimens A consistent feature of Ellipsocephalus specimens from Oland is a thin outer layer which has a faint pinkish tinge: this was commented on by Teigler and Towe (1975, pp. 138-139) who also established that a similar thin outer layer in a Silurian calymenid from Poland was composed of calcium phosphate, probably in the form of apatite. Specimens of Paradoxides sp., in the same beds on Oland do not have a thin layer of this nature. The layer does not completely cover all parts of every Ellipsocephalus specimen: it is often worn away from the prominence of the glabella (Text-fig 1a) and, in a few examples, seems to be absent, possibly removed on the counterpart. In the latter situation, the brown exposed ‘surface’ of the cuticle has the shiny appearance characteristic of other well-preserved trilobites. It is possible to find two Ellipsocephalus cranidia side by side on the same bedding plane, one with a pinkish outer layer, the other apparently without. However, when examined under a microscope, at least traces of the outer layer can be found on all specimens. In the very rare ‘complete’ specimens of Ellipsocephalus , all parts of the dorsal exoskeleton are seen to be covered by the outer layer. Hand specimens and polished slices viewed in incident light When the surface of the outer layer is viewed in incident light at low magnifications, almost its entire area appears to be patterned with small circular punctations, about 15 pm in diameter (Text-fig. 1b). The spacing of these punctations is somewhat irregular: in places they are almost contiguous, contrasting with small clear patches, but on average they are 15 //m apart. In areas where the outer DALINGWATER ET AL.\ TRILOBITE CUTICULAR ULTRASTRUCTURE 207 text-fig. 1. Ellipsocephalus polytomus Linnarsson from Enerum, Oland, Sweden. Specimen RM Ar 46218c. a, cranidium, x 5. b, detail of surface punctations, x 100. layer has been worn away or removed on the counterpart, the punctations can still clearly be seen, and also some light circular areas about 40 /nn in diameter each perforated by a minute ( c . 1 /mi) opening. These light circles are about 200 pm apart: a similar spacing to that of the patches devoid of punctations on the outer layer surface. In polished slices, sections of Ellipsocephalus cuticle can easily be identified by their shape and by the possession of a thin whitish outer layer, which at higher magnifications is seen to contain darker spherulitic structures around 15 //m in diameter. The region of cuticle below the outer layer is dark brown and penetrated by numerous fine perpendicular canals which stand out as they are paler than the ground material of the inner layer. SEM preparations General structure of the cuticle. The great predominance of cranidia on the surface of the hand specimens led us to assume that the great majority, if not all, of the Ellipsocephalus material was of sections of that part of the cephalon. As in material examined with the light microscope, the shape of many of the sections reinforced the validity of this conclusion. The etching process left a thin outer layer, 25-50 pm thick, standing clear and unaltered from the rest of the cuticle, up to 200 pm thick, which was etched inwards. Not only could the outer layer be viewed in perpendicular section, but its inner undersurface could also be examined, for example in preparation RM Ar 46218fi-E4 (Text-fig. 2). In that particular preparation, the perpendicular face clearly shows that the outer layer comprises about twenty lamina units, each just over 1 pm thick. The majority of preparations show a similar aspect to that of E4, but a few are different, possibly the result of : (i) slight differences in preparation technique, including direction of sectioning and quality of polishing; (ii) original differences in the cuticles, possibly including those related to the size of the animal; (iii) localized diagenetic differences. In preparation RM Ar 46218e-E9 (PI. 1, fig. 1), the outer layer is somewhat thicker (nearly 50 pm thick) than in most other preparations and the lamination is very clearly defined. There are about thirty-five lamina units, each nearly E5 /an thick except for the outer five units which are thinner. In areas of a few preparations, for 208 PALAEONTOLOGY, VOLUME 34 text-fig. 2. Scanning electron micro- graph of etched perpendicular section of Ellipsoceplialus polytomus Linnarsson cuticle outer layer, also with a view of undersurface of that layer. Preparation RM Ar 462186-E4, x 800. example RM Ar 462186-El (PI. 1, fig. 2), the lamination is less clear and transforms laterally into a zone of semi-prismatic calcite crystallites, and in one preparation, RM Ar 462186-Ei (PI. 1, fig. 3), the lamination is penetrated by calcite crystallites. Preparations RM Ar 462186-El and -Ei are both perpendicular sections (deduced from the perpendicular pathways of their canals) and so this transformation or penetration is a real phenomenon and not an artefact produced by angled sectioning. In many sections round or elliptical cavities, up to 1 5/mi in diameter and up to 20 //m high, extend from the lower edge of the outer layer to near the surface of the cuticle. However, they never reach beyond the uppermost fine lamina units, nor was any connection between these cavities and the cuticle surface observed in any of the sections examined. Another feature in many sections is an outermost non-laminate region of cuticle, up to 2 /mr thick and with a dense homogenous appearance. This can be seen most clearly in Plate 1, figure 1 and Plate 3, figure 4. The main region of the cuticle (for convenience termed the principal layer) consists of fine crystallites, presumably of calcite, sometimes with their long axes arranged roughly perpendicular to the cuticle surface. This region shows little detail apart from this feature, but is penetrated by perpendicular canals, approximately 3 /mi in diameter (PI. 1, fig. 4). These canals have disc-like lateral extensions about 0-5 //m thick and on average the same distance apart (PI. 1, fig. 5). Preparations in which the principal layer is etched deeply inwards illustrate how numerous and ubiquitous these canals are (PI. 1, fig. 6). Lamina unit ultrastructure. At higher magnifications, considerable ultrastructural detail can be resolved. At first, a bewildering array of apparently different structures was observed. But eventually, by always taking micrographs at a standard series of screen magnifications, it became clear that at least some of the apparent variation was the result of viewing essentially similar EXPLANATION OF PLATE 1 Figs 1-6. Ellipsoceplialus polytomus Linnarsson, Middle Cambrian, Enerum, Oland, Sweden. Scanning electron micrographs of etched sections of cranidial cuticle. 1-3, outer laminated layer in preparations RM Ar 46218e-E9, 6-El, 6-Ei, respectively, all x 650. 4-6, perpendicular canals in the principal layer in preparations RM Ar 462186-E5, x 300, 6-E5, x 8000, e-E9 x 250, respectively. PLATE 1 DALINGWATER et al., Ellipsocephalus polytomus cuticle 210 PALAEONTOLOGY, VOLUME 34 structures at arbitrary magnifications, with slight differences in preparation method, angle of slicing and angle of viewing also contributing to variability. Plate 2, figure 1 shows lamina units with their sectional edges flattened by the polishing procedure, whereas those in Plate 2, figure 2 show a more broken appearance. The interface between the perpendicular face and the horizontal undersurface of the outer layer was also examined (PI. 2, fig. 3). In all three micrographs the laminae appear to be composed of arrays of rods, with more or less circular cross-sections, linked together in sheets; some sheets seem to arc across the inter-laminae. A detailed view of the undersurface of the outer layer (PI. 2, fig. 4) shows that the fingerprint-like patterns seen in Text-figure 2 are produced by arced sheets of fibrous material. A near-horizontal view of a lamina unit in an unetched break (PI. 2, fig. 5) reveals a herringbone-like pattern of rods. In contrast, a near-horizontal slice, despite being subjected to flattening and polishing (or perhaps because of this) shows a mosaic of fibrous and rod- like material from different levels of the cuticle (PI. 2, fig. 6, which is a detail of PI. 3, fig. 5). Polygonal patterns. In sub-surface areas of the unetched preparation viewed from above, polygonal areas about 40 pm across and delimited by slightly raised ridges can be detected (PI. 3, fig. 1). Cavities. Round or elliptical cavities have already been mentioned as a consistent feature of the outer layer of cuticle. At low magnifications, arrays of these cavities can be seen, with the broken upper portions of perpendicular canals below them (PI. 3, fig. 4). In preparations sliced at an angle of a few degrees from the horizontal, the outer layer is perhaps somewhat disrupted by the effect of the etching process on the principal layer; the latter can be seen through the cavities (PI. 3, fig. 5). One preparation in which the principal layer has been etched inwards to a considerable extent, leaving the outer layer roofing a miniature cave (PI. 3, fig. 6), shows the stumps of canals as stalactic projections from the cave roof, clearly connecting to the cavities in the outer layer which are ‘illuminated' by the electron beam striking the top surface of the cuticle and ‘shining through it’. On the right of the micrograph, a rather stouter perpendicular canal is the only one left extending from the inner matrix to the outer layer. Other structures. The undersurface of the outer layer in some preparations seems to be covered by a thin coating skin through which some details of that undersurface can still be seen. This skin often peels back or breaks open to reveal clearer details. This phenomenon can just be seen on the bottom right of Plate 1, figure 1. Roughly star-shaped arrays of fibrous or crystalline material (PI. 3, fig. 2) stand out below the general level of the undersurface of the outer layer in some preparations. In some areas of nearly all preparations, horizontal tubular structures 1-2 pm in diameter criss-cross the undersurface of the outer layer, sometimes forming node-like structures where they intersect (PI. 3, fig. 3). COMPOSITION Semi-quantitative elemental analysis, using the LINK system of X-ray microanalysis attached to the SEM gave peaks for calcium and phosphorus in the outer layer, whereas the principal layer showed a strong peak only for calcium with lesser peaks for silicon and iron and only a trace of phosphorus. EXPLANATION OF PLATE 2 Figs 1—6. Ellipsocephalus polytomus Linnarsson, Middle Cambrian, Enerum, Oland, Sweden. Scanning electron micrographs of etched sections (except 5) of cranidial cuticle showing details of outer layer, all x 9000. 1 and 2, lamina units in preparations RM Ar 46218e-E9, b- E4. 3, interface between vertical section and undersurface, preparation RM Ar 462186-E4. 4, undersurface, preparation RM Ar 462186-E4. 5, unetched low angle break, preparation RM Ar 46218e-E8. 6, low angle slice, preparation RM Ar 462 1 8A- E6. PLATE 2 DALINGWATER et a!., Ellipsocephalus polytomus cuticle 212 PALAEONTOLOGY, VOLUME 34 text-fig. 3. Etched perpendicular section of Ellipsocephalus polytomus Linnarsson cuticle outer layer. Left, scanning electron micrograph; right, spot X-ray microanalysis for phosphorus. Preparation RM Ar 4621 8- A2, x 450. A spot analysis for phosphorus showed an exact co-incidence of the concentration of phosphorus with the outer layer (Text-fig. 3) and also suggested that the 3 pm perpendicular canals contain high concentrations of phosphorus. DISCUSSION Subdivisions of trilobite cuticle Stormer (1980) discussed the broad divisions of the trilobite cuticle and generally supported Dalingwater and Miller's (1977) view that it consisted of an outer prismatic layer and a principal layer with three distinct laminate zones - an outer zone with narrow lamina units, a middle zone with a few relatively wide units and an inner zone with a few narrow units. Stormer also recognized that rarely are all regions of the cuticle equally well represented or well preserved in any one example. Teigler and Towe (1975) have argued for two basic layers of cuticle, suggesting that the thin outer layer may be prismatic or pigmented or apatitic. Our interpretation of Ellipsocephalus cuticle is that in life the outer laminated layer had only the thin apparently structureless outermost layer above it, the latter possibly representing an epicuticle. Furthermore, the lateral transition between laminated cuticle and prismatic cuticle in one preparation and the invasion of the laminated layer by calcite crystallites in another suggests that the prismatic layer observed in the cuticle of many trilobites may not be an original layer. However, much more evidence is needed before we can firmly draw this conclusion. Ultrastructural detail of lamina units Calcified cuticles of Recent arthropods, for example those of decapod crustaceans, have organic EXPLANATION OF PLATE 3 Figs 1-6. Ellipsocephalus polytomus Linnarsson, Middle Cambrian, Enerum, Oland, Sweden. Scanning electron micrographs of etched sections (except 1) ofcranidial cuticle. 1, low angle break, showing prismatic structures, preparation RM Ar 46218e-E8, x 500. 2, arrays of fibrous material on outer layer undersurface, preparation RM Ar 462186-E4, x 9000. 3, tubular structures on undersurface of outer layer, preparation RM Ar 46218e-E9, x 1200. 4-6, cavities in the outer layer; 4, perpendicular sectional view, preparation RM Ar 462186-Ei, x 300; 5, from above, preparation RMAr 462186-E4, x 1000; 6, from slightly below, ‘illuminated’ by beam striking top surface, preparation RM Ar 46218e-E9, x 300. PLATE 3 DALINGWATER et ai, Ellipsocephalus polytomus cuticle 214 PALAEONTOLOGY, VOLUME 34 templates upon or within which inorganic salts are deposited. On analysis, these templates show at least three levels of structural organization (Giraud-Guille 1984u). Near-molecular associations of chitin and proteins to form microfibrils represent the first level; associations of microfibrils in reticulate, macrofibrillar or homogenous arrays form the second level; and spatial arrangements of level two associations (e.g. macrofibrils in helicoidal arrays) give the third level. Minerals are probably deposited within the reticulate arrangement of microfibrils in the decapod crustacean exocuticle and around the macrofibres of the calcified zone; homogenous arrays of microfibrils effectively fill all available space in the uncalcified endocuticle. It is quite possible to envisage the three-dimensional arrangement both of organic template and deposited minerals in the decapod crustacean calcified zone if one accepts the Bouligand-Neville interpretation of laminated cuticles. In fact, the model was originally proposed after examination of Carcinus calcified zone macrofibrils (Bouligand 1965), but later shown to be more widely applicable to microfibrillar arrangements, for example in insect cuticles (Neville 1975). It is, however, at the third level of cuticular architectural organization that the Dennell- Mutvei-Dalingwater view (see Dalingwater and Mutvei 1990 for a more detailed discussion) does not accord with the Bouligand-Neville model: the former suggest that laminae are real structures and that sheets of fibres arc across the inter-laminae. In this context it is interesting to note that it is difficult to produce a satisfactory three-dimensional helicoidal arrangement for the reticulate associations of exocuticular microfibrils: Giraud-Guille (1984«, p. 81, fig. 6) has illustrated a semi- helicoidal pattern with fibre direction changing in blocks, but even that is not easily reconciled with the reality of her excellent micrographs. In Recent decapod crustacean material examined with the SEM, it is difficult to distinguish between organic template and deposited minerals, even with the help of transmission electron micrographs of the same material in which all inorganic material has been removed by decalcification prior to sectioning. So interpretation of lamina unit ultrastructure of the trilobite material described here is extremely difficult, because in addition to great structural complexity we also have to consider the effects of replacement and diagenesis. We tentatively suggest that in EUipsocephalus the laminae are composed of horizontal sheets of rods with further sheets of material arcing at low angles across the inter-laminae and connecting adjacent laminae. The rods probably represent the original inorganic material of the cuticle, and possibly also reflect the original organic template. Significance of a finely laminated outer cuticular layer Many extant arthropods from all the major groups (crustaceans, insects, chelicerates) have an outer cuticular layer with fine lamina units - invariably much finer than those in central regions of their cuticles. This outer layer is very likely to have been formed pre-ecdysially, i.e. under the old cuticle prior to moulting, whereas central and inner regions of the cuticle are usually formed after ecdysis. Possibly slower pre-ecdysial formation in some ways results in the formation of narrow lamina units, but perhaps a functional explanation is more likely. A region of narrow lamina units on the outside of a cuticle will have considerable crack-stopping ability. This holds good with lamination interpreted either according to the Bouligand-Neville model or the Dennell-Mutvei-Dalingwater explanation (for further discussion see Dalingwater 1985, p. 360). Cavities and caned s The cavities in the outer layer appear circular or egg-shaped in many perpendicular sections, but a few are pear-shaped with the narrower end pointing upwards. Only in sections close to their mid- line do pear-shaped structures reveal their true shape; glancing slices will appear round or oblong. We therefore suggest that the most complex aspect seen reflects the true shape - resembling that of an upwardly pointing pear. The 3 pm canals which characterize the principal layer connect the cavities to the inner surface of the cuticle and therefore originally to the epidermis. On the other hand, the cavities do not quite extend to the surface of the cuticle, nor do they appear to be connected to the surface. However, they extend so close to the surface, that in hand specimens DALINGWATER ET AL.\ TRILOBITE CUTICULAR ULTRASTRUCTURE 215 illuminated from above they can be seen through the thin (less than 10 //m) overlying layer of cuticle. Furthermore, slight abrasion will easily remove this overlying layer and expose the tops of cavities. These cavities and canals are similar in position and dimensions to the Osmolska cavities described and discussed in great detail by Stormer (1980). Stormer considered that this type of cavity occurred below the prismatic layer, but Wilmot (1990/?) has clearly shown that they usually occur within the prismatic layer. Stormer (1980) suggested a chemosensory function for the Osmolska cavities, whereas Wilmot (19906) preferred to interpret the cavities and canals as some type of modified pore canal. However, the most closely analogous structures to the cavities and canals that we have encountered in an extensive search through the literature are the gland ducts of the millipede Glomeris convexa which have dilated tips within a finely-laminated outer region of cuticle (Richards 1951, p. 55, fig. 32c). Gland ducts may be concerned with the secretion and maintenance of the epicuticle. The dilated tips of the gland ducts in Glomeris are shown to connect to the surface of the cuticle by minute canals. As mentioned above, we have not detected such openings in Ellipsocephalus , but connections to the surface by minute canals would show up only very rarely in sectional slices. The canals in the endocuticle of Flexicalymene which Mutvei (1981, p. 230, fig. 5) termed pore canals have a diameter of about 0-3 //m, similar to that of the pore canals in Recent arthropod cuticles, and do not connect to cavities. They do, however, show a feature of similarity with the canals in Ellipsocephalus'. disc-like lateral extensions which Mutvei called horizontal lamellae or laminae. They almost certainly reflect ultrastructural elements of the principal layer, but whether they represent the laminae themselves or structures within lamina units is uncertain. Mutvei (1981, p. 229, fig. 4) described wider ducts, 3-7 //m in diameter, in Flexicalymene cuticle. There may also be two types of canal in Ellipsocephalus cuticle : the great majority are the 3 pm diameter canals which connect to cavities, but slightly wider canals which do not connect to cavities (e.g. to the right in PI. 3, fig. 6) may account for the irregularities in the spacing of punctations as seen in surface views of hand specimens and the presence of light circular areas on worn surfaces of hand specimens. Absence from the outer layer of any structures that can definitely be regarded as pore canals is puzzling. In an outer (and presumably pre-ecdysially formed) layer of cuticle a supply-line for minerals and for other materials required for mineralization would be needed after ecdysis. In Recent decapod crustacean cuticles pore canals almost certainly carry out this function (Roer and Dillaman 1984). However, pore canals are essentially organic structures, so they may not necessarily be preserved as canals. Elliptical openings are present on the undersurface of the outer layer (PI. 2, fig. 4) reminiscent of pore canal pathways: thus pore canals may indeed originally have passed upwards through the Ellipsocephalus cuticle outer layer. Significance of other primary microstructures Polygonal structures observed at a sub-surface level in the outer laminated layer (PI. 3, fig. 1) may be equivalent to the interprismatic septa of calcified cuticles of Recent decapod crustaceans. The walls of the septa in these Recent cuticles represent cell margins transformed into cuticular material and show concentrations of cation-binding glycoproteins and maximum carbonic anhydrase activity (Giraud-Guille 19846). Thus the walls represent sites of calcification initiation. It is important to note that they do not extend to the surface of the cuticle and are therefore distinct from polygonal surface ornament whose shapes and sizes are not necessarily related to epidermal cell shapes. Giraud-Guille (19846) has clearly shown that interprismatic septa coincide precisely with underlying epidermal cells. The thin coating skin (PI. 1 , fig. 1 ) on the undersurface of the outer layer may represent a deposit of relict organic material from the dissolution of the principal layer. Relict organic material has been identified in other trilobite cuticles by Dalingwater (1973) and Teigler and Towe (1975). The roughly star-shaped arrays also on the undersurface of the outer layer (PI. 3, fig. 2) may be the remains of fibrous structures binding together this and the underlying principal layer. Dennell 216 PALAEONTOLOGY, VOLUME 34 (1973) identified horizontal arrays of fibres in decapod crustacean cuticles which he suggested might bind together adjacent lamina units. Secondary microstructures The 1-2 /mi horizontal tubules on the undersurface of the outer layer (PI. 3, fig. 3) are interpreted as secondary structures because they are irregular in appearance and inconsistent with other cuticular structures in their arrangement. They are remarkably similar in dimensions and appearance to borings described by Runnegar (1985) from shells of the gastropod Yuwenia bentleyi from the Lower Cambrian Pavara Limestone of South Australia. Runnegar concluded that these borings were made by cyanobacteria rather than by fungi. Although the nodal structures in the tubules in Ellipsocephalus could be interpreted as fungal reproductive bodies, in other aspects the resemblance to the borings described by Runnegar is so close that it seems reasonable to consider the tubules in Ellipsocephalus also as infilled borings of cyanobacteria. Composition of the cuticle in Ellipsocephalus Although the outer layer is now almost certainly composed of calcium phosphate in the form of apatite and the principal layer (except for the 3 pm canals) of calcium carbonate in the form of calcite, it is uncertain if this reflects the original composition. Teigler and Towe (1975) did, however, demonstrate a high concentration of phosphorus in an outer layer of a Recent crab cuticle. One suggestion that we can make at this stage is that the outer layer may originally have had a different composition from the principal layer, since detailed microstructures are preserved in the former but not in the latter except as discs around perpendicular canals. Alternatively, the outer layer could have had a different structure from the principal layer which was more predisposed to replacement ; preferential replacement could have resulted in better preservation of microstructural detail. We intend to make further studies of the composition of the cuticle employing a range of techniques including cathodoluminescence. Concluding remarks Ultrastructural details described here from Ellipsocephalus cuticle are the finest so far from any trilobite cuticle and it is ironic that they are possibly also the oldest such details described from any arthropod cuticle. But, before any firm conclusions can be drawn about the general structure of trilobite cuticle, more work is needed on a range of cuticles using careful preparation techniques and taking advantage of the increased resolution of the current generation of SEMs. Parallel studies of Recent crustacean cuticles are also needed to elucidate the precise positional relationships of organic template and inorganic impregnating minerals. There are signs that arthropod cuticle workers are at last breaking out of the straightjacket imposed by the Bouligand-Neville model of cuticular architecture (Neville 1975). Compere and Goffinet (1987a, b ), for example, have described new and exciting structural details, from decapod crustacean cuticles, which do not fit the model. We need to know about and to be able to explain the reasons for differences between the cuticles of different species and of higher taxa, as well as attempting to identify features of similarity. The significance of the work described and discussed here is not only in the discovery of such exceptionally fine details in a trilobite cuticle, but also in heralding a new phase of fossil arthropod cuticle research made possible by new techniques and new instruments. Acknowledgements. We thank the staff of the School of Biological Sciences Electron Microscope Unit for their help, advice and technical expertise. We are most grateful to Dr Paul Selden for his constructive comments on a preliminary version of the manuscript, Mr Les Lockey for photographic work and Miss Lisa Monks for typing the final copy. This study was financially supported by Grant 287-1 18 of the Swedish Natural Science Research Council. DALINGWATER ET AL.\ TRILOBITE CUTICULAR ULTRASTRUCTURE 217 REFERENCES bouligand, Y. 1965. Sur une architecture torsadee repandue dans de nombreuses cuticles d’arthropodes. Compte Rendu Hebdomadaire des Seances de I'Academie des Sciences, Paris, 261, 3665-3668. compere, p. and goffinet, g. 1987a. Ultrastructural shape and three-dimensional organization of the intracuticular canal systems in the mineralized cuticle of the green crab Carcinus maenas. Tissue and Cell , 19, 839-857. - 19876. Elaboration and ultrastructural changes in the pore canal system of the mineralized cuticle of Carcinus maenas during the moulting cycle. Tissue and Cell , 19, 859-875. dalingwater, j. E. 1973. Trilobite cuticle microstructure and composition. Palaeontology , 16, 827-839. - 1985. Biomechanical approaches to eurypterid cuticles and chelicerate exoskeletons. Transactions of the Royal Society of Edinburgh ( Earth Sciences ), 76, 359-364. - and miller, j. 1977. The laminae and cuticular organisation of the trilobite Asaphus raniceps. Palaeontology, 20, 21-32. - and mutvei, h. 1990. Arthropod exoskeletons. 83-96. In carter, j. g. (ed.). Skeletal biomineralization : patterns , processes and evolutionary trends. Van Nostrand Reinhold, New York, 399 pp. dennell, r. 1 973. The structure of the cuticle of the shore-crab Carcinus maenas (L. ). Zoological Journal of the Linnean Society , 52, 159-163, 5 pis. giraud-guille, M.-M. 1984a. Fine structure of the chitin-protein system in the crab cuticle. Tissue and Cell, 16, 75-92. - 19846. Calcification initiation sites in the crab cuticle: The interprismatic septa. An ultrastructural cytochemical study. Cell and Tissue Research, 236, 413-420. linnarsson, j. G. o. 1877. Om faunan i lagran med Paradoxides olandicus. Sveriges Geologiska Undersokning. Afhandlingar och Uppsatser, Stockholm, Series C, No. 22, 1-24, pis 1, 2. mutvei, h. 1981. Exoskeletal structure in the Ordovician trilobite Fie xicaly metre. Lethaia, 14, 225-234. Neville, a. c. 1975. Biology of the arthropod cuticle. Springer-Verlag, Berlin, Eleidelberg, New York, xvi + 448 pp. richards, a. G. 1951. The integument of arthropods. University of Minnesota Press, Minneapolis, xvi + 41 1 pp. roer, r. and dillaman, r. 1984. The structure and calcification of the crustacean cuticle. American Zoologist, 24, 893-909. runnegar, b. 1985. Early Cambrian endolithic algae. Alcheringa, 9, 179-182. stgrmer, l. 1980. Sculpture and microstructure of the exoskeleton in chasmopinid and phacopid trilobites. Palaeontology, 23, 237-271. teigler, d. j. and towe, k. m. 1975. Microstructure and composition of the trilobite exoskeleton. Fossils and Strata, 4, 137-149, 9 pis. wilmot, n. v. 1990a. Cuticular structure of the agnostine trilobite Homagnostus obesus. Lethaia, 23, 87-92. - 19906. Primary and diagenetic microstructures in trilobite exoskeletons. Historical Biology, 4, 51-65. - and fallick, a. e. 1989. Original mineralogy of trilobite exoskeletons. Palaeontology, 32, 297-304. j. e. dalingwater and s. j. hutchinson Department of Environmental Biology The University, Manchester M13 9PL, UK H. MUTVEI Sektionen for Paleozoologi Naturhistoriska Riksmuseet 104 05 Stockholm, Sweden D. J. SI VETER University Museum Parks Road, Oxford OX1 3PW, UK Typescript received 27 January 1990 Revised typescript received 23 February 1990 CONTRASTING FEEDING STRATEGIES IN BIVALVES FROM THE SILURIAN OF GOTLAND by LOUIS LILJEDAHL Abstract. Two examples of contrasting feeding strategies in bivalves from the Silurian of Gotland are presented. The first shows a deposit-feeding community of protobranchs in which non-siphonate species greatly dominate siphonate ones. This is probably the result of extensive bioturbation by the non-siphonate species causing agitation of the fine-grained sediment and consequent disturbance of the feeding of siphonate species. Tiering of this community is also suggested, based on observations on abundant, silicified material. The second example depicts shallow subtidal life associations of Ilionia prisca in preferred orientation. This species shows special characteristics typical of extant deeply burrowing suspension-feeders of the superfamily Lucinacea. It is suggested that Ilionia prisca had a unique feeding strategy of anterior inhalation through a mucus tube, and also that it oriented itself obliquely to the direction of wave action, both for optimal intake of suspended food particles and for the avoidance of inhaling its own waste products. Possibly Ilionia prisca also lived in symbiosis with sulphur-oxidizing bacteria. The beds discussed are intercalated with shales and it is assumed that the whole bivalve population was instantaneously killed off when smothered by mud. Bivalves are perhaps the most thoroughly investigated of all marine invertebrates and many studies have been devoted to the feeding habits of this group. Throughout their evolutionary history, bivalves have occupied a large spectrum of aquatic habitats and are thus well suited for palaeoecological reconstructions. The feeding habits and trophic relations of benthic invertebrates have been classified by various workers in different ways (e.g. Stanley 1968; Walker and Bambach 1971). Most bivalves are generally considered to be suspension feeders or deposit feeders or carnivores. The classification of organisms as true suspension feeders or true deposit feeders, however, is made difficult by the presence of ‘opportunistic feeders’, i.e. those capable of using more than one feeding method (Cadee 1984). Deposit feeders ingest organic matter trapped in the substrate in which they live and therefore must actively move about in search of food. Their gills are simply built and mainly used for respiration, while the collection of food particles is provided by palp proboscides. Siphons, when present, are used for respiration (Cox 1969). The protobranchs discussed in this paper include one opportunistic deposit feeder, the solemyoid Janeia silurica (believed to have been symbiotic with chemoautotrophic bacteria: see Liljedahl 1984a, 19846, 1984c). In suspension feeders, the gills are more complex than those of the deposit feeders and are mainly used for food collection. Suspension feeders normally remain fixed in one position and passively feed on particles which come to them through the water. When present, siphons are, in contrast with the deposit feeders, used for feeding. Also within the suspension feeders ‘opportunistic’ feeders are present. The Silurian Ilionia prisca is assumed to have lived in symbiosis with chemoautrophic bacteria and is thus considered an ‘opportunistic’ filter feeder (Liljedahl in prep.). Bivalves play an important role in the tiering relationships (relative vertical (ecological) positions of organisms within a community) in many Recent biotas, where different trophic categories or feeding groups may be recognized (Ausich and Bottjer 1984). The chemical stratification and related environmental changes within a sediment may be (Palaeontology, Vol 34, Part 1, 1991, pp. 21 9-235. | © The Palaeontological Association 220 PALAEONTOLOGY, VOLUME 34 considerable. Accordingly, the ecological relationships of organisms downwards from the surface can be more extreme below the sediment/water interface than above it. The first example considered in the present paper, is represented by deposit feeders. In this, community tiering may be established (for detailed analysis see Liljedahl 1985). Indirect competitive interactions may also have been present here: the feeding habits of one trophic group (the non-siphonate deposit feeders) is suggested to have made the substrate unsuitable for representatives of another trophic group, the siphonate deposit feeders (see Rhoads and Young 1970; Levinton and Bambach 1975). As filterers, suspension feeders are sensitive to sudden environmental changes (in contrast to deposit feeders). Above the sediment/water interface, ecological stratification may also occur, depending on different susceptibility to fouling among the suspension feeders. The second example is provided by the deeply buried ‘opportunistic’ suspension feeder Ilionia prisca (Hisinger). It inhabited a substrate of low species diversity in a shallow subtidal environment of low oxygen and high sulphur content, unsuitable for most other bivalves. It is suggested that Ilionia prisca oriented itself with its anterior-posterior axis obliquely to wave movement, i.e. with its anterior inhalant mucus tube against the flow of suspended food particles. By analogy with its living relatives (Reid and Brand 1986) it is also assumed that Ilionia prisca housed chemoautotrophic bacteria in the gills, the bacteria being important nutritional providers for the bivalve (Liljedahl in prep.). INTERACTIONS BETWEEN THE DEPOSIT-FEEDING BIVALVES OF MOLLBOS The material from Mollbos 1 consists of 2743 silicified valves, of which 684 are articulated. Only one specimen (Nuculodonta gotlandica) has been observed in life-position (Text-fig. 1g). They were all isolated by acid etching and the preferred life-positions of each of the Mollbos species are thus mainly based on morphological and statistical grounds by analogy with modern counterparts. One of the advantages of the acid etching method is that the whole preserved shelly fauna is recovered, i.e. all sizes are represented (Liljedahl 1984a). Above all it is possible to obtain enough material for fairly reliable statistical processing (Liljedahl 1985). The bivalves form an important constituent of the Mollbos fauna. Although it is a typical soft- bottom community, this fauna contains a conspicuous amount of sessile benthos such as stromatoporoids, tabulate corals, rugose corals, crinoids, etc. probably due to close vicinity to a reef. It abounds in infaunal burrowers, e.g. protobranch bivalves, gastropods, and annelid worms (Liljedahl 1983). f The Wenlockian Halla Beds at Mollbos consists of a compact, strongly argillaceous calcilutite which is fairly hard due to silicification (Liljedahl 1983, p. 8). The high percentage of deposit feeding text-fig. 1. a, Nuculoidea lens. External dorsal view of articulated specimen, anterior to the left, SGU TYPES 894, 895, sample G77-28LJ, x4-3. b, Nuculodonta gotlandica. External dorsal view of articulated specimen, anterior to the left, SGU TYPES 1202, 1203, sample G79-82LJ, x 3-9. c, Nuculoidea lens. External lateral view of a left valve, SGU TYPE 901, sample G77-28LJ, x 3 9. d, Nuculodonta gotlandica. External lateral view of a left valve, SGU TYPE 1036, sample G78-2LL, x4-4. e, Nuculoidea lens. Internal postero-ventral view of holotype (right valve) showing from left to right, anterior adductor muscle scar, anterior pedal protractor muscle scar (first arrow from the left), visceral attachment muscle scar (second arrow), anterior pedal retractor muscle scar (third arrow), and pedal elevator muscle scar (fourth arrow), SGU TYPE 842, sample G77-28LJ, x 3-5. F, Nuculodonta gotlandica. Internal posteroventral view of a right valve showing from left to right, anterior adductor muscle scar, anterior pedal protractor muscle scar (first arrow from the left), anterior pedal retractor muscle scar (second arrow), and visceral attachment muscle scar (third arrow), SGU TYPE 1200, sample G79-82LJ, x 4. g, Nuculodonta gotlandica. Only specimen of the bivalve fauna of Mollbos 1 found in life-position, just below original sediment surface, LO 6084t, loose boulder, x 1-3. All specimens are silicified and all samples are from Mollbos 1. LILJEDAHL: BIVALVE FEEDING STRATEGIES 221 text-fig. 1 . For legend see opposite. 222 PALAEONTOLOGY, VOLUME 34 animals suggests that this sediment was rich in bacteria, as is often the case in fine grained substrates (Zobell 1938: Newell 1970). It is concluded that the Mollbos infaunal bivalve fauna is autochthonous and undisturbed except for post mortem phenomena, such as disturbances by scavengers and burrowing deposit feeders (Liljedahl 1985; also see Johnson 1960 for criteria for life associations). The bivalve fauna is numerically dominated by deposit feeding species (90% of the bivalve population: Liljedahl 1985). It comprises four nuculoid species, Nuculodonta gotlandica Liljedahl, 1983 (44% of Mollbos bivalves), Nuculoidea lens Liljedahl, 1984 (27%), Palaeostraba baltica Liljedahl, 1984 (0-7%), Caesariella lindensis (Soot-Ryen, 1964) (0-4%), and one solemyoid, Janeia silurica Liljedahl, 1984 (18%). Shell morphology of the different deposit feeding bivalves shows a common theme with minor variations (Liljedahl 1 984 oscillating water movement text-fig. 8. Orientation of anteroposterior axis (broken lines) of llionia prisca in relation to water movement. Anterior inhalant tube (open circle) and posterior exhalant siphon (filled circle), thin arrows showing direction of waste products, a, hypothetical in line-orientation, b, observed oblique orientation. A — * ♦ — -o B fig. 8a). In the latter case, with an oscillating wave movement, the chances of inhaling its own waste products are greater than in case of oblique orientation. Rhythmic trapping A number of the limestone beds at Grogarnshuvud 1 abound in llionia prisca while others contain few or no specimens. Each of the beds discussed is intercalated with calcareous shales. The thickness of the limestone beds ranges from 70 to 130 mm. The shales are usually 10 mm thick but can in places reach 60 mm (see also Sundquist 1982). It is suggested that each limestone bed represents one life association of llionia prisca , although there is a conspicuous lack of juvenile specimens (Liljedahl, in prep.). Sundquist (1982, pp. 87-89) assumes that the calcareous shale beds represents the final stage of a previous ephemeral incident, such as a storm, etc. The shales were deposited rhythmically and possibly some of them represent volcanic ash-falls rich in silica, indicated by the presence of silicified fossils. On such occasions a large number of floating nautiloid shells were stranded and oriented parallel to the shore due to storm-wave action. The water was heavily loaded with suspended particles, which eventually came to rest, resulting in a deposit considerably thicker than the present thin shale beds. The fouling of the water and/or sedimentation of the fine grained material most probably was catastrophic for the bivalves and the infaunal species were forced to escape. However, when overburden stress reaches a critically high value, burrowing infaunal organisms can not escape burial. Experiments on living polychaete/bivalve communities show that this value (40 Kpa) corresponds to a burial depth of c. 28 cm (Nicols et al. 1978). Specimens of llionia prisca are found down to a depth of 10 cm or more in the sediment and with an overburden of a thick layer of clay (now considerably compacted). This limit could have been reached in the present community at Grogarnshuvud 1 and the infaunal bivalves fatally trapped. It seems as if no reworkers, including protobranch bivalves, gastropods, annelid worms etc. survived, since llionia prisca was preserved undisturbed in ‘life’ position. As stated, specimens of llionia prisca have been found at different depths in the beds and even at the sediment surface (all orientation-measured specimens). A number of individuals are inclined, with their antero-posterior axis dipping at an angle of 10-15° to the bedding plane (Text-fig. 9). This suggests that these individuals were killed during the rocking movement of burrowing, perhaps while attempting to escape. According to Stanley (1972) and Nicols et al. (1978), in a series of experiments, individual burrowing ability of each bivalve species resulted in differences in escape efficiency. Although no escape structures have been found in the different beds at Grogarnshuvud, the slurry-like nature of the sediment may account for their absence. LILJEDAHL: BIVALVE FEEDING STRATEGIES 233 text-fig. 9. In situ specimens of Ilionia prisca on eroded bedding surfaces, Grogarnshuvud 1, Ludlovian Hemse Beds. A, articulated specimen preserved as internal mould of surrounding sediment, x 0.6. b, articulated specimen preserved as druse filled internal mould, x 0-7. The specimens found on the bedding planes either: (1) succeeded in escaping burial; their gills were, however, eventually clogged due to the large amount of fine grained suspension in the very turbid water; or (2) they were killed in ‘life’ position, or rather ‘death’ position after having burrowed themselves downwards in the sediment and later isolated by erosion caused by the ephemeral violent event. The presence of some specimens, preserved as drusy filled cavities (Text- fig. 9b), supports this latter assumption. They indicate extremely rapid burial and enough compaction forces to prevent the ligament to open the valves after soft part corruption. The absence of Ilionia prisca from some of the beds may result from the sedimentation of the fine grained material not being rapid enough for catastrophic burial. Alternatively the high turbidity event may have been too short for the bivalves to be suffocated. In either case the bivalves might have been able to escape and survive. It is also possible that the bivalves had not yet colonized the area after the previous catastrophic event. It seems as if only one or a few age classes colonized the area after each previous mud sedimentation event. Presumably either these individuals were killed during the following catastrophic incident before they were able to reproduce, or the environment was simply unfavourable for their young offspring (see Rhoads and Young 1970). Repository. Specimens with their numbers prefixed RMMO are deposited in the type collection of the Swedish Museum of Natural History, Box 50007, S-104 05 Stockholm, Sweden, those prefixed SGU TYPE are deposited in the type collection of the Geological Survey of Sweden, Box 670, S-751 28 Uppsala, Sweden, and those prefixed LO in the type collection of the Geological Institute, Lund University, Solvegatan 13, S-223 62 Lund, Sweden. Acknowledgements. The present paper is a longer version of a talk given at the Murchison Symposium on 2 April 1989 at Keele University. Sincere thanks are due to Sven Laufeld, Jin' Kri'z, Anita Lofgren and Euan Clarkson for valuable comments and improvements of the manuscript and also to Euan Clarkson for linguistic help. Lennart Jeppsson kindly gave access to huge samples from Mollbos 1. A travel grant from Naturvetenskapliga Forskningsradet is gratefully acknowledged. 234 PALAEONTOLOGY, VOLUME 34 REFERENCES allen, j. a. 1958. On the basic form and adaptions to habitat in the Lucinacea (Eulamellibranchia). Philosophical Transactions of the Royal Society of London, Series B, 241, 421 — 484. ausich, w. i. and bottjer, d. j. 1985. Phanerozoic tiering in suspension-feeding communities on soft substrata: Implications for diversity. 255-274. In vallentine, j. w. (ed.). Phanerozoic diversity patterns', profiles in macroevolution. Princeton University Press, 441 pp. bambach, R. K. 1971. Adaptation in Grammysia obliqua. Lethaia , 4, 169-183. berg, c. J. and alatolo, p. 1984. Potential of chemosynthesis in molluscan mariculture. Aquaculture, 39, 165-179. bergman, c. F. 1979. Ripple marks in the Silurian of Gotland, Sweden. Geologiska Foreningens i Stockholm Forhandlingar , 101, 217-22. buchanan, j. b. 1958. The bottom fauna communities across the continental shelf off Accra, Ghana (Gold Coast). Proceeding of the Zoological Society of London , 130, 1-56. cadee, g. c. 1984. ‘Opportunitistic feeding’, a serious pitfall in trophic structure analysis of (palaeo)faunas. Lethaia , 17, 289-292. CAVANAUGH, C. M., GARDINER, S. L., JONES, M. L., JANNASCH, H. W. and WATERBURY, J. B. 1981 Procaryotic Cells in the hydrothermal vent tube worm Riftia pachyptilia Jones : possible chemoautotrophic symbionts. Science, 213, 340-342. cox, L. r. 1969. General features of the Bivalvia. N2-128. In moore, r. c. (ed.). Treatise on invertebrate paleontology. Part N. Mollusca 6, Bivalvia 1 . Geological Society of America and University of Kansas Press, Boulder, Colorado, 489 pp. DANDO, P. R, SOUTHWARD, A. J., SOUTHWARD, E. C., TERW1LLINGER, N. B. and TERWILLINGER, R. C. 1985. Sulphur- oxidizing bacteria and haemoglobin in gills of the bivalve mollusc Myrtea spinifera. Marine Ecology - Progress Series, 23, 85-98. eriksson, c.-o. and laufeld, s. 1978. Philip structures in the submarine Silurian of northwest Gotland. Sveriges Geologiska Undersokning, Series C, 736, 1-30. hadding, a. 1958. The pre-Quaternary sedimentary rocks of Sweden: 7. Cambrian and Ordovician limestones. Lunds Universitets Arsskrift, 2, 54, 1-262. Also in Kungliga Fysiografiska Sdllskapets Handlingar, 69, 1-262. hickman, c. s. 1984. Composition, structure, ecology and evolution of six Cenozoic deep-water mollusk communities. Journal of Paleontology, 58, 1215-1234. hisinger, w. 1837. Lethaea Svecica seu Petrificata Sveciae, iconibus et characteribus illustrata. Stockholm, 124 pp. Johnson, r. G. 1960. Models and methods for analysis of the mode of formation of fossil assemblages. Geological Society of America , Bulletin, 71, 1075-1086. laufeld, s. 1974. Preferred orientation of orthoconic nautiloids in the Ludlovian Hemse Beds of Gotland. Geologiska Foreningens i Stockholm Forhandlingar , 96, 157-162. levinton, j. s. 1977. Ecology of shallow water deposit-feeding communities Quisset Harbor, Massachusetts. Ecology of Marine Benthos, 2, 191-228. levinton, j. s. and bambach, r. c. 1975. A comparative study of Silurian and recent deposit-feeding bivalve communities. Paleobiology, 1, 97-124. liljedahl, l. 1983. Two silicified Silurian bivalves from Gotland. Sveriges Geologiska Undersokning, Series C, 799, 1-51. - 1984a. Silurian silicified bivalves from Gotland. Sveriges Geologiska Undersokning, Series C, 804, 1-82. - 19846. Janeia silurica, a link between nuculoids and solemyoids (Bivalvia). Palaeontology, 27, 693-698. - 1985. Ecological aspects of a silicified bivalve fauna from the Silurian of Gotland. Lethaia, 18, 53-66. manten, a. a. 1971. Silurian reefs on Gotland. Elsevier, Amsterdam, 539 pp. marsh, l. f. 1984. Mode of life and autecology of Silurian-Devonian Grammysiidae (Bivalvia). Palaeontology, 27, 679-691. mcalester, a. l. 1965. Systematics, affinities, and life habits of Babinka, a transitional Ordovician lucinoid bivalve. Palaeontology, 8, 231-246. morton, J. e. 1960. The responses and orientation of the bivalve Lasaea rubra Montagu. Journal of the Marine Biological Association of the United Kingdom, 39, 5-26. morton, j. e. 1962. Habit and orientation in the small commensal bivalve mollusc, Montacuta ferruginosa. Animal Behaviour, 10, 126-133. Newell, r. c. 1970. Biology of intertidal animals. American Elsevier, New York, 555 pp. LILJEDAHL: BIVALVE FEEDING STRATEGIES 235 nicols, j. a., rowe, G. T., Clifford, c. h. and young, r. a. 1978. In situ experiments on the burial of marine invertebrates. Journal of Sedimentary Petrology , 48, 419^125. o’gower, a. k. and nicol, p. i. 1971. Orientation of the bivalve Anadara trapezia (Deshayes) relative to water currents. The Veliger , 13, 275-278. reid, r. G. and brand, d. G. 1986. Sulfide-oxidizing symbiosis in lucinaceans: Implications for bivalve evolution. The Veliger , 29, 3-24. rhoads, d. c. 1970. Mass properties, stability and ecology of marine muds related to burrowing activity. 392-406. In crimes, t. p. and harper, j. c. (eds). Trace fossils. Seel House Press, Liverpool, 574 pp. - and young, d. k. 1970. The influence of deposit-feeding organisms on sediment stability and community trophic structure. Journal of Marine Research , 28, 150-178. soot-ryen, h. 1964. Nuculoid pelecypods from the Silurian of Gotland. Arkiv for Mineralogi och Geologi, Kungliga Vetenskapsakademien , 3, 489-519. spiro, b., greenwood, p. b., southward, a. j. and dando, p. R. 1986. 13C/12C ratios in marine invertebrates from reducing sediments: confirmation of nutritional importance of chemoautotrophic endosymbiotic bacteria. Marine Ecology - Progress Series, 28, 233-240. 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). Geological Society of America, Memoirs , 125, 1-296. - 1972. Functional morphology and evolution of byssally attached bivalve mollusks. Journal of Paleontology, 46, 165-212. southward, E. c. 1986. Gill symbionts in Thyasirids and other bivalve molluscs Journal of Marine Biological Association of the United Kingdom , 66, 889-914. sundquist, b. 1982. Wackestone petrography and bipolar orientation of cephalopods as indicators of littoral sedimentation in the Ludlovian of Gotland. Geologiska Foreningens i Stockholm Forhandlingar, 104, 81-90. walker, k. r. and bambach, r. c. 1974. Feeding by benthic invertebrates: classification and terminology for palaeoecological analysis. Lethaia, 7, 67-78. zobel, c. e. 1938. Studies on the bacterial flora of marine bottom sediments. Journal of Sedimentary Petrology , 8, 10-18. Typescript received 14 November 1989 Revised typescript received 19 March 1990 LOUIS LILJEDAHL Department of Historical Geology and Palaeontology Solvegatan 13 S-233 62 Lund, Sweden LIZARD EGG SHELLS FROM THE LOWER CRETACEOUS OF CUENCA PROVINCE, SPAIN by ROLF KOHRING Abstract. The Lower Cretaceous vertebrate-bearing coaly marls and limestones of Una (Province of Cuenca, Spain) have yielded fragmentary reptilian egg shells. The shell is of gekkonid microstructure type, and thus they can be confidently assigned to the lizards. These fragments represent the oldest known gekko-like egg shells. Fossil egg shells have been reported nearly worldwide, especially from Upper Cretaceous and Tertiary deposits, and they have been assigned, according to their microstructure and biomineralization, to turtles, crocodiles, dinosaurs, and birds (reviewed by Hirsch and Packard 1987) . Fossilized egg shells of snakes and lizards, however, are only rarely described, owing to their largely non-mineralized composition ; nearly all squamates produce eggs with soft shells consisting of interlacing protein fibrils and some calcareous matter, probably homologous to the membrana testacea of avian eggs (Schleich and Kastle 1988). Only the recent gekkonids (Lacertilia) develop calcified (and thus fossilizable) rigid egg shells, characterized by a continuous layer, composed of tightly abutted jagged columns, and surface nodes. The thickness of both fossil and recent gekkonid egg shells ranges from 35 to 280 /mi (Schleich and Kastle 1988). Fossil gekko-like egg shells are reported from the Lower Miocene of Kenya (Hirsch and Harris 1989), the Oligocene of the Mainz Basin (Schleich and Kastle 1988), the Lower Eocene of Wyoming (Hirsch and Packard 1987), the Cretaceous/Tertiary-boundary of Peru (Hirsch in Mourier et aL 1988) , the Upper Cretaceous of both Montana (Hirsch and Quinn, in press) and India (Sahni et al. 1984), and the Lower Cretaceous of Mongolia (Alifanov 1989). One genus (Ilerdaesaurus sp.) of this material is under study by A. Richter (Berlin). According to Hoffstetter (1964) gekkonids are known since the Upper Jurassic. The lizards of Una are not yet described (Krebs, pers. comm.) LOCALITY AND STRATIGRAPHY The coal-bearing marls and limestones of Una (province of Cuenca, Spain) have yielded tetrapods, especially frogs (Fey 1988), turtles and lizards (Krebs, pers. comm.), crocodiles (Brinkmann 1989), and early mammals (Henkel and Krebs 1969). They have been dated as Upper Barremian on the basis of palynomorphs (Mohr 1989), ostracodes, and charophytes (Schudack 1989). The palaeoenvironment of Una is postulated to have been lacustrine, with marshy, deltaic deposits (Gierlowski-Kordesch and Janofske 1990). The egg shells described here are well preserved; only their margins are partially pyritized, as is typical also for the gastropods, ostracodes, and charophytes of Una. THE MATERIAL Description In all, eight tiny, dark brown coloured shell fragments (5 x 10 mm), embedded in the coaly sediment, have been studied in thin sections and by scanning electron microscopy (SEM, Cambridge Stereoscan 360). In thin section, the egg shells display a continuous layer with hardly visible fine, closely spaced growth-stage lines and a light coloured secondary layer in the outer part, which is obviously a diagenetic structure (Hirsch, pers. IPalaeontology, Vol 34, Part 1, 1991, pp. 237-240, 1 pl.| © The Palaeontological Association 238 PALAEONTOLOGY, VOLUME 34 comm.) (PL 1, fig. 5). It is never pyritized in any of the specimens. In XPL an extinction pattern with cone- shaped wedges in the upper part of the primary layer and partially in the secondary layer is visible. In SEM studies, some further morphological features can be observed. The shell consists of a nearly complete homogenous calcitic layer without recognizable shell units (PL 1, figs 2 and 6), and therefore is very similar to the recent gekko Ptyodactylus (Schleich and Kastle 1988). In its upper part, the 20 /rm thick secondary layer with horizontal crystallites is visible (PI. 1, fig. 4). This characteristic structure has been mentioned also from Upper Cretaceous gekko-like egg shells of Montana (Hirsch and Quinn, in press) and from hadrosaurian egg shells (Hirsch and Packard 1987). The surface is covered with a thin 3 /an mineralized layer, as is typical for nearly all recent gekkos (Schleich & Kastle 1988). However, this layer has never been reported from fossil lizard eggs. The shell thickness, including surface nodes, is 170-180 //m. The diameter of these nodes is about 100 /nn (PI. 1, fig. 1). Discussion Other distinctive reptilian egg shell microstructures, such as pores, pore openings, and basal aragonitic mammallae could not be found. Probably the rigid gekkonid egg shell is not homologous to those of other reptiles or birds. The remarkable structural similarities to modern gekkonid egg shells allow the assignment of the Una material to the lizards. These late Barremian fragments are the oldest known certain lizard egg shells. Due to the poor knowledge of the problematic diagenetic pattern of egg shells an identification of the very thin outer layer as a mineralized organic cover seems hitherto impossible. The fragmentation of the shells suggests substantial transport. The primary shapes and sizes of the eggs cannot be reconstructed. A single thin-shelled (about 50 //m) fragment is known from the Upper Jurassic (Kimmeridgian) coaly marls and limestones of Guimarota (Central Portugal), where turtle egg shells have been described (Kohring 1990). It is similar in microstructure to recent gekkonid egg shells, but its real taxonomic position is uncertain (PI. 1, fig. 8). Acknowledgements. I thank Professor B. Krebs (Berlin) for material and information on Una, and Dr K. F. Hirsch (Denver) for useful remarks. The field work was supported from the Deutsche Forschungsgemeinschaft (DFG). My sincere thanks go to Miss H. Bosbach (Berlin) for reading the typescript critically, to Dr J. Reitner (Berlin) for providing me with recent lizard egg shells, and to Dr D. Martill (Milton Keynes) and Dr M. J. Benton (Bristol) for useful comments. REFERENCES alifanov, v. r. 1989. The oldest gecko (Lacertilia, Gekkonidae) from the Lower Cretaceous of Mongolia. Paleontological Journal 23. 128-131. brinkmann, w. 1989. Vorlaufige Mitteilung fiber die Krokodilier-Faunen aus dem Ober-Jura (Kimmeridgium) der Kohlegrube Guimarota, bei Leiria (Portugal) und der Unter-Kreide (Barremium) von Una (Provinz Cuenca, Spanien). Documenta naturae 56, 1-26. EXPLANATION OF PLATE 1 Figs 1-6. Lizard egg shell fragments from the Lower Cretaceous of Una. 1, outer surface with nodes, x 50. 2, lateral view with homogenous calcitic layer, secondary layer, outside is up, x 100. 3, lateral view, note secondary layer, x 200. 4, Secondary layer, x 400. 5, lateral view in thin section in ordinary light, outside with a secondary layer is up, note pyritized margins, x 50. 6, lateral view, x 80. Fig. 7. Recent gekko egg shell, Tarentola sp., lateral view with nodose outer surface, x 300. Fig. 8. Uncertain gekko-like egg shell from the Upper Jurassic of Guimarota, lateral view, x 300. Specimens are housed in the Institut ffir Palaontologie, Freie Universitat Berlin under the registered numbers Un Bar ES 1-8. PLATE 1 KOHRING, lizard egg shells 240 PALAEONTOLOGY, VOLUME 34 fey, b. 1988. Die Anurenfauna aus der Unterkreide von Una (Ostspanien). Berliner geowissenschaftliche Abhandlungen A 103, 1-125. gierlowski-kordesch, E. and janofske, d. 1990. Paleoenvironmental reconstruction of the Weald around Una (Serrania de Cuenca, Cuenca Province, Spain). In wiedmann, j. (ed.). Cretaceous of the Western Tethys. Proceedings 3rd International Cretaceous Symposium, Tubingen 1987, Schweizerbart, Stuttgart. henkel, s. and krebs, b. 1969. Zwei Saugetier-Unterkiefer aus der Unteren Kreide von Una (Provinz Cuenca, Spanien). Neues Jahrbuch fur Geologic und Palaontologie , Monatshefte, 1969, 449 463. hirsch, k. f. and Harris, j. 1989. Fossil eggs from the Lower Miocene Legetet Formation of Koru, Kenya: snail or lizard? Historical Biology , 3, 61-78. — and Packard, m. j. 1987. Review of fossil eggs and their shell structure. Scanning Microscopy, 1, 383-400. — and quinn, b. In press. Eggs and eggshell fragments from the Upper Cretaceous Two Medicine Formation of Montana. Journal of Vertebrate Paleontology. hoffstetter, r. 1964. Les Sauria du Jurassique superieur et specialements les Gekkota de Baviere et de Mandchourie. Senckenbergiana Biologica, 45, 281-324. kohring, R. 1990. Upper Jurassic chelonian eggshell fragments from the Guimarota Coalmine (Central Portugal). Journal of Vertebrate Paleontology, 10, 128-130. mohr, b. 1989. New palynological information on the age and environment of Late Jurassic and Early Cretaceous vertebrate localities of the Iberian Peninsula (eastern Spain and Portugal). Berliner geowissenschaftliche Abhandlungen A, 106, 291-301. MOURIER, TH., BENGTSON, P., BONHOMME, M., BUGE, E., CAPPETTA, H., CROCHET, J.-Y., FEIST, M., HIRSCH, K. F., JAILLARD, E., LAUBACHER, G., LEFRANC, J. P., MOULLADE, M., NOBLET, C., PONS, D., REY, J., SIGE, B., TAMBAREAU, y. and taquet, p. 1988. The Upper Cretaceous-Lower Tertiary marine to continental transition in the Bagua basin, northern Peru. Paleontology, biostratigraphy, radiometry, correlations. Newsletters on Stratigraphy, 19, 143-177. sahni, a., rana, r. s. and prasad, G. v. R. 1984. SEM studies of thin egg shell fragments from the Intertrappeans (Cretaceous-Tertiary Transition) of Nagpur and Asifabad, peninsular India. Journal of the Paleontological Society of India , 29, 26-33. schleich, h. h. and kastle, w. 1988. Reptile egg-shells. Gustav Fischer Verlag, Stuttgart, 123 pp. schudack, m. 1989. Charophytenfloren aus den unterkretazischen Vertebraten-Fundschichten bei Galve und Una (Ostspanien). Berliner geowissenschaftliche Abhandlungen A, 106, 409-443. Manuscript received 18 January 1990. Revised manuscript received 25 April 1990 R. KOHRING Institut fur Palaontologie Freie Universitat Berlin Schwendenerstrasse 8, 1000 Berlin 33, Germany NOTES FOR AUTHORS The journal Palaeontology is devoted to the publication of papers on all aspects of palaeontology. Review articles are particularly welcome, and short papers can often be published rapidly. A high standard of illustration is a feature of the journal. Four parts are published each year and are sent free to all members of the Association. Typescripts should conform in style to those already published in this journal, and should be sent to Dr Dianne Edwards, Department of Geology, University of Wales College of Cardiff CF1 3YE, who will supply detailed instructions for authors on request (these are published in Palaeontology 1990, 33, pp. 993-1000). Special Papers in Palaeontology is a series of substantial separate works conforming to the style of Palaeontology. SPECIAL PAPERS IN PALAEONTOLOGY In addition to publishing Palaeontology the Association also publishes Special Papers in Palaeontology. Members may subscribe to this by writing to the Membership Treasurer: the subscription rate for 1991 is £45-00 (U.S. $80) for Institutional Members, and £20-00 (U.S. $36) for Ordinary and Student Members. A single copy of each Special Paper is available on a non-subscription basis to Ordinary and Student Members only , for their personal use, at a discount of 25 % below the listed prices: contact the Marketing Manager. Non-members may obtain Nos 35^43 (at cover price) from Basil Blackwell Ltd., Journal Subscription Department, Marston Book Services, P.O. Box 87, Oxford OX2 0DT, England, and older issues from the Marketing Manager. For all orders of Special Papers through the Marketing Manager, please add £1-50 (U.S. $3) per item for postage and packing. PALAEONTOLOGICAL ASSOCIATION PUBLICATIONS Special Papers in Palaeontology For full catalogue and price list, send a self-addressed, stamped A4 envelope to the Marketing Manager. Numbers 2 1 1 and 13-30 are still in print and are available together with those listed below: 31. (for 1984): Systematic palaeontology and stratigraphic distribution of ammonite faunas of the French Coniacian, by w. j. Kennedy. 160 pp ., 42 text-figs., 33 plates. Price £25 (U.S. $50). 32. (for 1984) : Autecology of Silurian organisms. Edited by m. g. bassett and j. d. lawson. 295 pp.. 75 text-figs., 13 plates. Price £40 (U.S. $80). 33. (for 1985): Evolutionary Case Histories from the Fossil Record. Edited by j. c. w. cope and p. w. skelton. 202 pp., 80 text-figs., 4 plates. Price £30 (U.S. $60). 34. (for 1985): Review of the upper Silurian and lower Devonian articulate brachiopods of Podolia, by o. i. Nikiforova, t. l. modzalevskaya and m. g. bassett. 66 pp., 6 text-figs., 16 plates. Price £10 (U.S. $20). 35. (for 1986): Studies in palaeobotany and palynology in honour of N. F. Hughes. Edited by d. j. batten and d. e. g. briggs. 178 pp., 29 plates. Price £30 (U.S. $60). 36. (for 1986): Campanian and Maastrichtian ammonites from northern Aquitaine, France, by w. j. Kennedy. 145 pp., 43 text-figs., 23 plates. Price £20 (U.S. $40). 37. (for 1987): Biology and revised systematics of some late Mesozoic stromatoporoids, by rachel wood. 89 pp., 31 text- figs., 7 plates. Price £20 (U.S. $40). 38. (for 1987): Taxonomy, evolution, and biostratigraphy of late Triassic-early Jurassic calcareous nannofossils, by p. r. bown. 118 pp., 19 text-figs., 15 plates. Price £30 (U.S. $60). 39. (for 1988): Late Cenomanian and Turonian ammonite faunas from north-east and central Texas, by w. i. Kennedy, 131 pp., 39 text-figs., 24 plates. Price £30 (U.S. $60). 40. (for 1988): The use and conservation of palaeontological sites. Edited by p. r. crowther and w. a. Wimbledon. 200 pp., 31 text-figs. Price £30 (U.S. $60). 41. (for 1989): Late Jurassic-early Cretaceous cephalopods of eastern Alexander Island, Antarctica, by p. j. howlett. 72 pp., 9 text-figs., 10 plates. Price £20 (U.S. $40). 42. (for 1989): The Palaeocene flora of the Isle of Mull, by m. c. boulter and z. kvacek 149 pp., 23 text-figs., 23 plates. Price £40 (U.S. $80). Field Guides to Fossils and Other Publications These are available only from the Marketing Manager. Please add £100 (U.S. $2) per book for postage and packing plus £1-50 (U.S. $3) for airmail. Payments should be in Sterling or in U.S. dollars, with all exchange charges prepaid. Cheques should be made payable to the Palaeontological Association. 1. (1983): Fossil Plants of the London Clay, by m. e. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $16) (Members £6 or U.S. $12). 2. (1987): Fossils of the Chalk, compiled by e. owen; edited by a. b. smith. 306 pp., 59 plates. Price £11-50 (U.S. $23) (Members £9-90 or U.S. $20). 3. (1988): Zechstein Reef fossils and their palaeoecology, by n. hollingworth and T. Pettigrew, iv + 75 pp. Price £4-95 (U.S. $10) (Members £3.75 or U.S. $7-50). 1982. Atlas of the Burgess Shale. Edited by s. conway morris. 31 pp., 24 plates. Price £20 (U.S. $40). 1985. Atlas of Invertebrate Macrofossils. Edited by j. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii + 241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24-95. © The Palaeontological Association, 1991 Palaeontology VOLUME 34 • PART 1 CONTENTS Late Tremadoc graptolites from western Newfoundland S. H. WILLIAMS and R. K. STEVENS 1 Middle Triassic holothurians from northern Spain A. B. SMITH and J. GALLEMI 49 A new upper Ordovician bryozoan fauna from the Slade and Redhill Beds, South Wales C. J. BUTTLER 77 Middle Ordovician bivalves from Spain and their phyletic and palaeogeographic significance C. BABIN CUld J.-C. GUTIERRBZ-MARCO 109 Spongiophyton from the late Lower Devonian of New Brunswick and Quebec, Canada P. G. GENSEL, W. G. CHALONER and W. H. FORBES 149 Teuthid cephalopods from the Upper Jurassic of Antarctica p. DOYLE 169 A new scleractinian-like coral from the Ordovician of the Southern Uplands, Scotland C. T. SCRUTTON and E. N. K. CLARKSON 179 The taxonomy and shell characteristics of a new elkaniid brachiopod from the Ashgill of Sweden L. E. HOLMER 195 Cuticular ultrastructure of the trilobite Ellipsocephalus polytomus from the Middle Cambrian of Oland, Sweden J. E. DALINGWATER, S. J. HUTCHINSON, H. MUTVEI and D. J. siveter 205 Contrasting feeding strategies in bivalves from the Silurian of Gotland L. LILJEDAHL 219 Lizard egg shells from the Lower Cretaceous of Cuenca Province, Spain R. KOHRING 237 Primed in Green Britain at ilie University Press , Cambridge ISSN 0031-0239 Till- PALAEONTOLOGICAL ASSOCIATION The Association was founded in 1957 to promote research in palaeontology and its allied sciences. COUNCIL 1991-1992 President '. Professor J. W. Murray, Department of Geology, The University, Southampton S09 5NH Vice-Presidents'. Dr P. R. Crowther, City of Bristol Museum and Art Gallery, Queen’s Road, Bristol BS8 1RL Dr P. A. Selden, Department of Extra-Mural Studies, University of Manchester, Manchester M13 9PL Treasurer : Dr M. E. Collinson, Department of Biology, King’s College, London W8 7AH Membership Treasurer '. Dr H. A. Armstrong, Department of Geological Sciences, The University, South Road, Durham DH1 3LE Institutional Membership Treasurer: Dr A. R. I. Cruikshank, 72 Thirlmere Road, Hinckley, Leicestershire LE10 OPF Secretary : Dr J. A. Crame, British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 OET Circular Reporter : Dr D. Palmer, Department of Geology, National Museum of Wales, Cardiff CF1 3NP Marketing Manager (Sales): Dr C. R. Hill, Department of Palaeontology, British Museum (Natural History), London SW7 5BD Marketing Manager ( Publicity ): Dr P. B. Wignall, Department of Earth Sciences, The University, Leeds LS2 9JJ Public Relations Officer : Dr D. M. Martill, Department of Earth Sciences, The Open University, Milton Keynes MK7 6AA Editors Dr J. E. Dalingwater, Department of Environmental Biology, University of Manchester, Manchester M13 9PL Dr P. Doyle, Department of Earth Sciences, Thames Polytechnic, London El 2NG Dr D. Edwards, Department of Geology, University of Wales College of Cardiff, Cardiff CF1 3YE Dr P. D. Lane, Department of Geology, University of Keele, Keele, Staffordshire ST5 5BG (co-opted) Dr A. R. Milner, Department of Biology, Birkbeck College, Malet Street, London WC1E 7HX Dr P. D. Taylor, Department of Palaeontology, British Museum (Natural History), London SW7 5BD Other Members Dr E. A. Jarzembowski, Brighton Dr W. J. Kennedy, Oxford Dr R. B. Rickards, Cambridge 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: Professor S. H. Williams, Department of Earth Sciences, Memorial University, St John’s, Newfoundland A1B 3X5. China: Dr Chang Mee-mann, Institute of Vertebrate Palaeontology and Paleoanthropology, 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 : Prof. 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. U.S.A. : Prof. A. J. Rowell, Department of Geology, University of Kansas, Lawrence, Kansas 66044. Prof. N. M. Savage, Department of Geology, University of Oregon, Eugene, Oregon 97403. Prof. M. A. Wilson, Department of Geology, College of Wooster, Wooster, Ohio 44961. Germany: Prof. F. T. Fursich, Institut fur Palaontologie, Universitat, D8700 Wurzburg, Pliecherwall 1 MEMBERSHIP Membership is open to individuals and institutions on payment of the appropriate annual subscription. Rates for 1991 are: Institutional membership . . . £60 00 (U.S. $108) Student membership . . £1 1-50 (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 A. R. I. Cruikshank, 72 Thirlmere Road, Hinckley, Leicestershire LE10 OPF. Student members are persons receiving full-time instruction at educational institutions recognized by the Council. On first applying for membership, an application form should be obtained from the Membership Treasurer: Dr H. A. Armstrong, Department of Geological Sciences, The University, South Road, Durham DH1 3LE. Subscriptions cover one calendar year and are due each January; they should be sent to the Membership Treasurer. All members who join for 1991 will receive Palaeontology , Volume 34, Parts 1M. Enquiries concerning back numbers should be directed to the Marketing Manager. Non-members may subscribe, and also obtain back issues up to 3 years old, at cover price through Basil Blackwell Ltd, Journal Subscription Department, Marston Book Services, P.O. Box 87, Oxford OX2 0DT, England. For older issues contact the Marketing Manager. Cover: Bolboforma intermedia Daniels and Spiegler (Incertae Sedis, possibly a calcified algal cyst) from Site 552A, southwest margin of Rockall Plateau, late Miocene NN9-10. x 800. Bolboforma was planktonic and cysts are found in epicontinental shelf sea deposits, thus providing a useful biostratigraphic link with oceanic sequences. A SPIDER AND OTHER ARACHNIDS FROM THE DEVONIAN OF NEW YORK, AND REINTERPRETATIONS OF DEVONIAN ARANEAE by PAUL A. SELDEN, WILLIAM A. SHEAR and PATRICIA M. BONAMO Abstract. The oldest known spider, from the Devonian (Givetian) of Gilboa, New York, is Attercopus fimbriunguis (Shear, Selden and Rolfe), parts of which were originally described as a trigonotarbid, possibly of the genus Gelasinotarbus. Previous reports of Devonian spider fossils, from the Lower Emsian of Alken- an-der-Mosel, Germany, and the Pragian of Rhynie, Scotland, are shown to be erroneous identifications. Attercopus is placed as sister-taxon to all living spiders, on the basis of characters of the spinneret and the arrangement of the patella-tibia joint of the walking legs. A cladogram of the relationships of all pulmonate arachnids is presented. A pulmonate arachnid from Gilboa, related to Araneae and Amblypygi, is described as Ecchosis pulchribothrium Selden and Shear, gen. et sp. nov., and additional arachnid material is described. A devonian age for the oldest known fossil spider was set by Hirst when he described Palaeocteniza crassipes Hirst, 1923, from the Pragian Rhynie Chert of Aberdeenshire, Scotland. The description of another fossil assigned to the Araneae, Archaeometal devonica Stormer, 1976, from the Emsian of Alken-an-der-Mosel, Germany, added more evidence for the antiquity of the order. The find of a spider spinneret (Shear, Palmer et al. 1989) from the Givetian of Gilboa, New York, provided conclusive evidence for the validity of the Devonian as the earliest period in which spider fossils are known to occur. In this paper, results of a re-examination of the Rhynie and Aiken spider fossils are presented: the fossils are not spiders, and are reinterpreted as a probable juvenile trigonotarbid and an indeterminate fossil, respectively. The Gilboa spider is placed in a new genus, Attercopus , described here. The new genus includes only the animal previously called Gelasino- tarbus? fimbriunguis (Shear et al. 1987), which we now regard as the only known Devonian spider, and the oldest known fossil of the Araneae. In addition, podomeres originally placed in Arachnida incertae sedis by Shear et al. (1987) are redescribed here, with the addition of new material, as Ecchosis pulchribothrium gen. et sp. nov., and placed in Pulmonata incertae sedis (it may be an amblypygid), and other arachnid remains from Gilboa are described. RHYNIE PALAEOCTENIZA In 1923, Hirst described Palaeocteniza crassipes as a spider from the Pragian Rhynie Chert of Scotland. James Locke and W.A.S. carried out a detailed photographic study of the specimen (British Museum (Natural History) (BM(NH)) In 24670) in 1987 and 1988. The fossil is in a small chip of chert mounted on a microscope slide. Even if the fossil were to be removed from the slide, no additional views could be obtained, owing to the opacity of the chert behind the specimen. The specimen itself is highly three-dimensional, as are many of the arthropod remains from Rhynie, and thus difficult to photograph. Adding to the problems are the cloudiness of the matrix, opaque inclusions, and the very small size of the specimen, about 0-85 mm long. In addition to photographs of the whole specimen at low magnifications (Text-fig. 1), a series of about thirty-five optical sections was made at higher magnification, using the very shallow depth-of- field characteristic of Nomarski Differential Interference Contrast (NDIC - see below. Methods). These photographs were printed at a large size and each was carefully examined for evidence of IPalaeontologv, Vol. 34, Part 2, 1991, pp. 241-281, 7 pls.| © The Palaeontological Association 242 PALAEONTOLOGY, VOLUME 34 text-fig. 1. Palaeocteniza crassipes Hirst. 1923. a, b, two views, at different planes of focus, of the holotype (and only known) specimen (BM(NH) In 24670), seen from the left side, anterior to the left, x 130. spider autapomorphies. In addition, each photograph was traced seriatim on a graphics pad andnJie resultant digitized images were stacked and reconstituted as a rotatable virtual solid using the Jandel computer program PC3D™ (see below. Methods). We had hoped that James Locke’s efforts to reconstruct the specimen using this program would allow us to examine further details, but this was not to be. The level of resolution attainable was too low, and there were considerable difficulties in digitizing the images, since shallow as the depth-of-field was, at the necessary magnifications subjective judgement was still required as to what was in the plane of focus and what was not, resulting in further blurring of the lines. A careful examination of the specimen itself and of the serial photographs proved to give the most information. The general condition of the specimen, much crumpled and folded, suggests that it may be a moult. Hirst (1923) noticed a small, thin, scarcely visible object dorsal to the abdomen, which he supposed to be the detached carapace. Since the carapace detaches when arachnids moult, if this identification is correct, its presence and position are further evidence for the specimen being a cast exoskeleton. The prosoma is almost entirely concealed behind the dorsally flexed legs and palps. While the palps appear to be complete, all of the legs on the left side of the specimen (facing the viewer) lack their distal portions. The abdomen is complexly crushed and folded. Hirst (1923) provided a detailed drawing, which, however, incorporates some errors. The proportions of the right palp are not correct in comparison with the left, to which a segment has been added. In ‘restoring’ the loose piece of cuticle to its supposed position as carapace, the SELDEN ET AL.. DEVONIAN ARACHNIDS 243 mass of wrinkles and folds above the leg coxae (perhaps the true carapace) has been omitted, and some of the folds in this structure appear to have been confused with parts of the palps. The second or third left leg has the tibia omitted. In the region of the supposed abdomen. Hirst noted that what had been made in the drawing to resemble spinnerets might be folds of cuticle. This is definitely so; the apparent internal structures of the abdomen are also cuticular folds on the right side of the specimen, seen through the left side. In attempting to determine the affinity of this fossil, a process of elimination was followed. The general appearance and structure of the body (a prosoma with five pairs of leg-like appendages, and an abdomen attached by a narrowed portion) establishes that it is an arachnid, and that it may belong to the known orders Araneae, Amblypygi, Uropygi, Schizomida, or Trigonotarbida. The presence of leg-like (not raptorial) palps rules out Amblypygi, Uropygi, and Schizomida, at least as they are presently known. Devonian trigonotarbids differ from potentially contemporaneous spiders in a number of ways. While both groups may have segmented abdomens, trigonotarbids have three tergal plates per segment and lack spinnerets. The eyes of any contemporaneous spiders were likely to have been grouped on a centrally located tubercle, as in the modern mesothele spiders, while those of Devonian palaeocharinid trigonotarbids are dispersed in three groups: a median group of two, and two lateral groups which may consist of several minor and major lenses each (Shear et al. 1987). All the Devonian trigonotarbids we have examined have a simple bicondylar hinge joint between the patella and tibia, and spiders have a monocondylar rocking joint in this position. Close examination of the abdomen of the specimen failed to reveal any evidence for or against segmentation (despite the clear segmental lines in his illustration. Hirst (1923, p. 460) wrote: ‘...it is impossible to be quite certain whether this [the abdomen] is segmented or not.’). Thus the number of tergites that might be present for each segment cannot be ascertained. The ‘spinnerets' have already been alluded to; as Hirst inferred, this is in fact a fold of the abdominal cuticle that can be traced continuously until it merges with other folds of the structure. The entire abdomen was also carefully examined for spinnerets, because we suspected that it might have been twisted through 180°, and because in living mesothele spiders the spinnerets are located about in the middle of the ventral surface of the abdomen, which is supposedly their primitive position. We found no indication whatsoever of spinnerets. Careful focusing revealed that among the crushed mass of the prosoma was an object that resembles an eye tubercle and seems to bear at least two hemispherical lens-like protrusions. Unfortunately this evidence is inconclusive, because at least two eye lenses would be present on a median tubercle both in trigonotarbids and spiders. The complicated folding and distortion of the carapace and its concealment behind the legs made it impossible for us to find any indication of lateral eye groups. The patella-tibia articulation can be seen on just one of the legs, probably the left third leg. It may be possible to make out two dorsally situated articular condyles on the distal end of the patella, but at the level of magnification required to see them, the optical properties of the chert interfere significantly. In summary, the fossil carries none of the autapomorphies of spiders that could be seen on a specimen of this size and level of preservation, but its identity as a trigonotarbid is only suggested (by the possible pattern of patella-tibia articulation). It should be pointed out, however, that scores of trigonotarbids have been seen in the Rhynie chert, and that this specimen is the only one for which a spider identity has been suggested. Our hypothesis is that Palaeocteniza crassipes Hirst is a moulted exoskeleton from an early instar trigonotarbid. ALKEN ARCHAEOMETA One of only four fossil sites with Devonian terrestrial animals, Alken-an-der-Mosel, Germany, has yielded impression fossils of lower Emsian age, including trigonotarbids, scorpions, eurypterids, and arthropleurids (Stormer 1976; Brauckmann 1987). One fossil from this deposit, Archaeometa? 244 PALAEONTOLOGY, VOLUME 34 devonica Stormer, 1976, was identified as a spider (Stormer 1976). A policy against type-specimen loans at the Senckenberg Museum, which houses this specimen, meant that we were unable to examine the original. However, we were able to study a plaster cast, and the photograph and drawing published by Stormer. The specimen consists of an elongate blob with a few transverse lines at one end and a vaguely indicated region at the other which may be part of some plant remains (Stormer 1976, figs 48 and 49; pi. 5, fig. 2 a,b). Stormer indicated that he had before him Petrunkevitch’s drawing of Archaeometa nephilina Pocock, 1911, from the Upper Carboniferous of Britain. This drawing (Petrunkevitch 1949, fig. 159) shows a featureless carapace with seven legs radiating from it, and an elongate abdomen with two longitudinal lines and four or five terminal segments. There are two similar specimens of A. nephilina in the British Museum (Natural History) which were examined in 1986 by W.A.S., and subsequently by P.A.S. Specimen In 15863 is the more complete and was the specimen figured by Petrunkevitch. It is relatively poorly preserved and little can be added to the diagrammatic illustration and brief description. Specimen In 31259, the holotype, does not show the transverse ‘segmental' lines seen in In 15863. The cuticle is tuberculate and the abdomen bears longitudinal folds; neither of these features are found in contemporaneous spider fossils (e.g. Eocteniza silivicola , figured on Pocock’s pi. II, fig. 4), but are more reminiscent of other Carboniferous arachnid groups. There are other details visible on this specimen which would reward a detailed restudy. Nevertheless, there are no features which would distinguish either of these specimens as a spider rather than any other arachnid. In any case, the resemblance of Archaeometa? devonica to these two specimens is vague and probably coincidental. There seems to be no reason to consider Archaeometa? devonica as a spider or a fossil arachnid of any sort. THE GILBOA ARACHNIDS Early reports on the Gilboa fauna (Shear et al. 1984) raised the possibility of spiders being among the animals present. The tip of an arachnid walking leg tarsus was illustrated, and diagnosed as being from a spider largely on the basis of serrate ventral setae similar to the silk-handling accessory claws found in some living araneoid spiders. However, in later studies, the possibility of spiders being present receded as it became clear that another related group of arachnids, the Trigonotarbida, dominated the fauna. We were also unable to demonstrate conclusively in the fossils any autapomorphies of spiders. Shear et al. (1987), in a detailed study of the trigonotarbids, assigned all pulmonate arachnid fossils from Gilboa to this extinct order, which was placed as the plesiomorphic sister group to the other pulmonate orders. One animal represented only by legs was assigned with some doubt to the trigonotarbid genus Gelasinotarbus, and given the species epithet fimbriunguis. This name referred to the characteristic claws, set with ventral cuticular fimbriae, not found in any other trigonotarbids. Other characters in these legs, present but undetected in 1987, we now recognize as conclusive evidence of a spider. A single femur with a patch of acute spinules near its base was called Arachnida Incertae sedis B; its cuticle is similar to that of fimbriunguis, and other similar femora have now been found in direct connection with pieces of undoubted fimbriunguis. A third group of specimens, consisting of podomeres and cuticular fragments, was referred to Arachnida Incertae sedis A. Re-examination of these specimens and of new material with the same distinctive cuticle has produced evidence that they belong to a pulmonate arachnid, close to Amblypygi and Araneae. To complicate matters further, the tarsus illustrated as a possible spider in Shear et al. (1984, fig. 1 b) is undoubtedly trigonotarbid ; it has smooth claws and lacks a tarsal organ. Late in 1988, conclusive evidence for spiders finally turned up in the Gilboa material: a spinneret (Shear, Palmer et al. 1989). This discovery triggered a search for other possible spider parts, and it was soon realized that the spinneret belonged with the legs described in 1987 as Gelasinotarbus? fimbriunguis. In addition, some previously unassigned chelicerae and some pieces of carapace belong to this animal. SELDEN ET A L. \ DEVONIAN ARACHNIDS 245 The ‘clasp-knife’ form of the chilecera, places it in the Pulmonata ( = Arachnidea sensu van der Hammen 1977; made up of the orders Trigonotarbida, Uropygi, Schizomida, Amblypygi, and Araneae). Illustrated here for comparison are chelicerae of the uropygid Mastigoproctus giganteus (PI. 7, fig. 5), and the amblypygid Heterophrynus elaphus (PI. 7, fig. 6), and see Shear et al. (1987, figs 7, 67, 68) for photographs of trigonotarbid chelicerae. A number of characters unequivocally place the chelicera in Araneae (see discussion under phylogenetic relationships). A cheliceral gland, found only in spiders, is present. The cheliceral fang of A.fimbriunguis lacks setae, which are present in all other pulmonates. In all other orders of Pulmonata, the largest cheliceral teeth are at the end of the tooth row opposing the tip of the fang (subchelate condition), while in A. fimbriunguis, as in the vast majority of spiders, the largest teeth occur part-way along the row and nearer to the fang articulation than to the fang tip (the subchelate condition occurs in a small number of spiders, but the described arrangement is found only in spiders, among the pulmonates). On the basis of outgroup comparison with, for example, scorpions, the subchelate state is primitive. Thus there are three definite spider synapomorphies present in the chelicera. A significant apomorphy of spiders is the presence of cheliceral venom glands. Whilst the evidence is not entirely certain, in at least two specimens of A. fimbriunguis chelicerae there may be a subterminal venom pore near the fang tip (PI. 1, fig. 7). In addition, as discussed in the detailed descriptions, the articulations present make it clear that the A. fimbriunguis chelicera must have been orthognath. The legs of A. fimbriunguis bear numerous lyriform organs; only in spiders are lyriform organs found on podomeres other than the metatarsi. The pieces of carapace are referred to A. fimbriunguis on the basis of their similarity of cuticular patterning. The evidence that the spinneret, chelicera, legs, and carapace fragments all come from the same morphospecies is overwhelming. All the chelicerae are identical, except for some size differences, and all of the podomere types (trochanter, femur, etc.) are identical within each type. All specimens, including the spinneret and carapace fragments, have the same distinctive cuticular ornamentation, a pattern which appears in no other Gilboa specimens except those that can be unequivocally assigned to the spider on the grounds given above. Finally, the chelicerae and basal leg podomores occur in organic connection on a number of slides. Therefore these Gilboa specimens are considered to belong to the same species, Attercopus fimbriunguis. There are numerous fragments of cuticle among the Gilboa slides which resemble the cuticle of A. fimbriunguis at first sight, and which we at first thought could belong to the body of the spider. Some of these were figured by Shear et al. (1987) and referred to as Arachnida Incertae sedis A. This animal is characterized by: generally large size; scale-like ornament rather than reticulation; setal sockets which range from small to very large; striated macrosetae and thick, striated, bifid spines (PI. 7, figs 4 and 8); groups of slit sensilla and lyriform organs; ornamented trichobothrial base on the patella. Minute, c. 0 005 mm, circular organs occur on the cuticle surface and appear, at low magnification, similar to the characteristic little slit sensilla of Attercopus , but examination at higher magnifications reveals a circular hole rather than a central slit, so they are not the same organ. None of these minute pores bears a seta, and their function is unknown; nevertheless, the difference in morphology from the little slit organs of Attercopus gives a useful criterion for distinguishing the two cuticle types. New information on Arachnida Incertae sedis A has been discovered during the present study, and the animal is named Ecchosis pulchribothrium gen. et sp. nov., below. The presence of lyriform organs suggests that E. pulchribothrium could be a spider, but the distinctive ornamented trichobothrial socket on the patella is puzzling. Virtually identical trichobothrial sockets are found on the living amblypygid Heterophrynus elaphus (PI. 7, fig. 2), but this animal has a qi ite dilferent leg articulation pattern to that in E. pulchribothrium , and a lyriform organ only on the metatarsus. The identity of E. pulchribothrium thus remains unclear, but we suggest that it is either an aberrant amblypygid or a member of an extinct, undiagnosed arachnid order. 246 PALAEONTOLOGY, VOLUME 34 GEOLOGICAL SETTING Stratigraphy The fossils occur in a grey shale in the upper part of the Panther Mountain Formation at a locality on Brown Mountain, Gilboa, Schoharie Co., New York (7§' quadrangle sheet 6168 IV NW 1945, approx. 271272 m N by 142951 m E; Banks et al. 1985). Further locality details can be found in Banks et al. ( 1 972). The original site has now been destroyed to make way for a pump-storage power plant associated with Schoharie Reservoir, but much of the fossil-bearing shale was removed to the Department of Biology, State University of New York at Binghamton, for later processing. The Panther Mountain Formation is part of the Hamilton Group, upper Middle Devonian Erian Series, and is equivalent to the middle Givetian of Europe. Palaeoecologv Detailed discussion of the taphonomy and palaeoecology of the biota is given in Shear (1986), Shear et al. (1987) and Shear and Bonamo (1988). The Gilboa lithology is a dark grey mudstone. The fauna occurs in close association with mats of interlocking spiny stems of the lycopod Leclercqia. Consideration of the manner of preservation of the plants suggested to Banks et al. (1985) that they were buried in situ by low-energy flood waters. Shear et al. (1984) suggested that the animals, which were living at the site or may have been carried in by the flow, came to rest by the localized reduction of velocity created by the mesh of Leclercqia. the ‘natural sieve’ effect would exclude large pieces of arthropod cuticle, while the most minute particles could have passed through. Almost all the arthropods recovered from the Gilboa site were undoubtedly terrestrial. The only exception to this is the occurrence of eurypterid fragments. In the Devonian, these animals lived in both marine and freshwater aquatic habitats, and some were amphibious (Selden 1984, 1985), so their presence in the Gilboa mudstones is not problematical. In addition to the external evidence of sedimentology and associated land flora for the habitat of the arthropods, palaeophysiology provides further proof of their terrestriality (Selden and Jeram 1989). Trichobothria are fine hairs sensitive to high-frequency vibrations, and could only function in air. They occur on the Gilboa pulmonates Gelasinotarbus bonamoae , G. bifidus (Shear et al. 1987, figs 105-120), and Ecchosis pulchribothrium (see below), and the pseudoscorpion (Shear, Schawaller and Bonamo 1989). Book- lungs for air breathing occur in the trigonotarbids of Gilboa (Shear et al. 1987). While we have no evidence of trichobothria or book-lungs in the Gilboa spider Attercopus , all living spiders are terrestrial apart from the secondarily aquatic Argyroneta aquatica , found in fresh waters of Europe, and the littoral, southern hemisphere Desidae. The phylogenetic discussion (below) indicates that if Attercopus were aquatic, it would also have been secondarily so, since all other Pulmonata are primarily terrestrial. MATERIAL AND METHODS Preservation The animal fossils are preserved as minute, undistinguished, brown to black flakes, which are unrecognizable as animals when in the rock and under incident light microscopy, but transmitted light reveals their zoological nature. The cuticle appears brown in transmitted light, and the depth of colouration is directly correlated with the thickness of the cuticle (or the number of layers of cuticle superimposed in the specimen). The chemical composition of the cuticle is not known; the brown colouration suggests it is organic, but the reduction of much of the plant material in the same beds to carbon indicates the likelihood that the arthropod cuticle has also been altered, probably by repolymerization of the organic molecules, during diagenesis. The arthropods are strongly compressed, necessitating the use of special techniques, such as NDIC, to separate overlapping layers of cuticle. For the same reason, scanning electron microscopy (SEM) is virtually useless for the study of these fossils, revealing only surface features: both original structures and diagenetic effects. SELDEN ET AL.: DEVONIAN ARACHNIDS 247 The fossils are fragmentary; only rarely are podomeres and other parts found in organic connection with others. However, the occurrence of such specimens is vital for the correct identification of loose podomeres and reconstruction of the animals. The dearth of pieces of carapace and abdomen of the arachnids can be explained by the fact that podomeres have two surfaces, so that when compressed together they remain coherent and are less likely to fragment than the body parts which consist of a single sheet of cuticle. The carapace and abdomen cuticle is represented by the many ‘scraps’ which occur on the slides. The nearly complete trigonotarbid carapaces and abdomens described by Shear et al. (1987) are rare, and mostly consist of both left and right (or dorsal and ventral) surfaces compressed together. Further discussion of the preservation of the Gilboa fauna is given in Shear et al. (1987). Methods The specimens were recovered from the rock matrix by digestion in concentrated hydrofluoric and hydrochloric acids (see Shear et al. 1987; Shear and Bonamo 1988, for details). After washing in distilled water, the animal fossils were separated from the abundant plant fragments, as far as possible, and mounted in CMC or Clearcol on plain microscope slides. The preparation was done in the laboratory of P.M.B. in Binghamton, and the prepared slides were then sent to Hampden- Sydney for study by P.A.S. and W.A.S. The slides were studied using an Olympus Vanox II biological microscope with a Nomarski Differential Interference Contrast (NDIC) facility. This illumination is particularly useful at high magnification and for the optical separation of closely adpressed layers of cuticle. Use was made of an Olympus SZH stereomicroscope for low magnification work, particularly on comparative extant material; for photography, this was cleared of muscles by soaking overnight in a solution of potassium hydroxide. Camera lucida attachments to both microscopes facilitated accurate drawing of the specimens, and photographs were taken on 35 mm Kodak Technical Pan film at ASA 50 with Olympus PM 10 cameras mounted on these instruments. On plates and text-figures, unless stated otherwise, all photographs were taken in transmitted light with NDIC on the Vanox. The computer program Jandel PC3D™ (available from Jandel Scientific, 2526 Bridgeway, Sausalito, California 94965, USA) was used for the three-dimensional reconstruction of Palaeocteniza crassipes , and the program MacClade 2.1 (Maddison and Maddison 1987) was extremely useful in the phylogenetic analysis. Abbreviations and conventions used in text-figures are as follows; a, anterior, antero-; ar, articulation; ch, chelicera(l); cl, claw; co cx, costa coxalis; cu. cuticle; Cx, coxa; d, dorsal; di, distal; e, edge; f, fold; Fe, femur; gl, gland; i, inferior, infero-; m, arthrodial membrane; ma, marginal; me, median; ms, macroseta; Mt, metatarsus; p, posterior, postero-; pa sp, palpal spinules; Pa, patella; pd, paired; po, poison duct opening; pr, proximal; ps, prosoma; r, ridge; s, superior, supero-; sc, sclerite; si, slit sensilla; sr, serrated; st, sternum, su, surface; t b, trichobothrial base; Ta, tarsus; ta or, tarsal organ; Ti, tibia; Tr, trochanter; tv, transverse; v, ventral; X, artefact. Unless stated otherwise in the legend to camera lucida drawings: dashed lines show linear features showing through cuticle from behind; finely dotted areas are internal surfaces; coarse dots show arthrodial membrane; setal sockets and slit sensilla (where shown) are infilled in black when on surfaces showing through from behind; prominent spores (where shown) are in black. Repository ami authorship Type and figured material is deposited in the Department of Invertebrates, American Museum of Natural History, New York (numbers prefixed AMNH). but are referred to in the text by their slide numbers. Most slide numbers consist of a series number (the first two numbers, e.g. 411.7, or the first only if only two numbers are present, e.g. 329), followed by the number of the slide within the series. The last, slide, number is prefixed with the letters AR (or Ar) on the slide itself, and quoted thus in earlier publications; these letters are omitted here for brevity. The slide may include more than one specimen, commonly of a different arthropod, but quoting the slide number makes retrieval of specimens for future study easier, facilitates references to earlier papers on the Gilboa 248 PALAEONTOLOGY. VOLUME 34 table 1. List of specimens mentioned in text. Slide No. AMNH No. Illustration Brief description A iter copus fimbriunguis 329. 1 43162 PI. 3, fig. 4; Text-fig. 6d palpal femur + patella 329.3 43163 PI. 3, fig. 2; Text-fig. 6 b femur 329.3 43163 PI. 4, fig. 1 ; Text-fig. 7a distal tibia 329.3 43163 PI. 4, fig. 10; Text-fig. 7f metatarsus 329.38 43168 PI. 4. fig. 8 metatarsus 329.39 43098 Text-fig. 12 b patella 329.53 43099 PI. 4, fig. 9 tibia 329.57 43100 Text-fig. 12 f metatarsus 329.58 43101 Shear et al. 1987, fig. 134 holotype, metatarsus, tarsus 329.59 43102 PI. 3, fig. 3; Text-fig. 6c distal femur + patella 329.59 43102 Text-fig. 12c trochanter 329.69 43106 PI. 2, fig. 5; Text-fig. 5e various; femur, patella, tibia 329.69 43106 PI. 6, fig. 5; Text-fig. 9d palpal tarsus 329 . 70 43107 Text-fig. 12 a paratype, femur + patella 329.70 43107 Text-fig. 1 2 d, e 2 metatarsi, proximal tarsus 329.16.34 43164 PI. 5, fig. 2 tarsus 329.22.9 43165 PI. 1, fig. 7; Text-fig. 4e chelicera 329. 31a. Ml 43166 PI. 3, fig. 7; Text-fig. 6e various; femur + patella 329. 31a. M2 43047 PI. 6, fig. 4 legs 334. la. 4 43170 PI. 5, figs 1 and 3; Text-figs 8a-c 2 legs, patella to tarsus 334. la. 6 43171 PI. 2, fig. 4; Text-fig. 5d femur 334. la. 7 43172 PI. 1, figs 6 and 8; Text-fig. 4c chelicera 334. la. 8 43173 PI. 4, figs 6; Text-fig. 7e tibia 334. la. 9 43174 PI. 2, fig. 1 ; Text-fig. 5 a femur 334.16. 12 43175 PI. 3, fig. 5; Text-fig. 6g distal femur + patella 334.16.34 43176 Text-figs 10. and 1 1 a, b, c spinneret 334.16.38 43177 PI. 5, fig. 5; Text-fig. 8d tarsus 334.16.86 43178 PI. 3, fig. 6; Text-fig. 6f femur + patella 411.02. 12M.6 43179 PI. 6, figs 1 and 2; Text-fig. 9 a metatarsus + tarsus 411.7.19 43052 paratype, femur 411.7.33 43180 PI. 1, figs 4 and 5; Text-fig. 4d chelicera 411.7.45 43181 PI. 4, fig. 3; Text-fig. 7c distal tibia 411 . 19.83 43182 PI. 2, fig. 2; Text-fig. 5 b coxa 411.19.98 43183 PI. 4, fig. 7; Text-fig. 7g distal tibia 411.19.102 43184 PI. 2, fig. 7; Text-fig. 5h 3 coxae, 1 trochanter 411.19.243 43185 PI. 3, fig. 8 proximal femur 411 . 19.248 43186 PI. 4, fig. 5; Text-fig. 7d patella 411 . 19.250 43187 PI. 2, fig. 8; Text-fig. 5g coxa 411 . 19.251 43188 PI. 4, fig. 1 1 metatarsus 411.20.25 43189 PI. 4, fig. 2; Text-fig. 7 b patella 2002.12.49 43190 PI. 4, fig. 4 tibia 2002.12.79 43191 PI. 3, fig. 1 ; Text-fig. 6a femur 2002.12.90 43192 PI. 1, figs 2 and 3; Text-fig. 4 b cheliceral teeth 2002.12.102 43193 PI. 1, fig. 1 ; Text-fig. 4a anterior carapace Ecchosis pulchribothrium 411.1.33 43194 PI. 7, fig. I paratype, distal femur 411.7.37 43195 PI. 6, fig. 6; Text-fig. 9 b holotype, patella -Fprox. tibia 411.7.86 43111 Shear et al. 1987, figs 149 and 150 paratype, distal patella 411.19.96 43198 PI. 6, fig. 3; Text-fig. 9c patella 411 . 19.137 43169 PI. 7, fig. 4 large, bifid spine 411 . 19.184 43195 PI. 7, fig. 3 lyriform organ 411 . 19.188 43196 PI. 7, fig. 8 paratype, probable tibia 411.19.206 43197 PI. 7, fig. 7 sheet of cuticle 2002.9.13 43097 PI. 2, fig. 3; Text-fig. 5 c coxa Arachnida incertae sedis 334. la. 4 43198 PI. 5, Fig. 3 flagelliform appendage 2002.9.20 43199 PI. 5, Fig. 4 flagelliform appendage SELDEN ET AL.: DEVONIAN ARACHNIDS 249 fauna in which slide numbers are used, and locates the specimen to the original rock sample. Thus it will be possible in the future to collate data on the whole Gilboa biota to a fine degree of accuracy. Table 1 lists the described specimens both by their AMNH accession number and the slide number. A complete list of the microscope slides which bear fragments of Attercopus fimbriunguis, Ecchosis pulchribothrium , and Arachnida incertae sedis is deposited as Supplementary Publication No. SUP 14040, 5 pp., at the British Library, Boston Spa, Wetherby, Yorkshire LS23 7BQ, England. Copies of this can be obtained by writing to the British Library at the above address, enclosing prepaid coupons available from most libraries throughout the world. In addition to the fossils, the following material (both males and females, and from the W. A. Shear Collection, unless otherwise stated) of extant arachnids was studied for comparative purposes: Araneae: Liphistius sumatranus Thorell, Sumatra, American Museum of Natural History collection; Amblypygi : Heterophrynus elaphus Pocock, Ecuador; Uropygi : Mastigoproctus giganteus (Lucas), Florida; Schizomida: species indet., Mexico. Following previous practice (Shear el al. 1987), authorship of new taxa is attributed to Selden and Shear. Bonamo discovered and supervised the preparation of the Gilboa material ; Selden and Shear are responsible for other information and ideas in this paper. RECONSTRUCTION OF THE GENERALIZED LEG OF ATTERCOPUS The reconstruction (Text-fig. 2) reflects a combination of the known morphology of various legs, some of which are suspected to be leg 1 by their close relationship with palpal femora and chelicerae, but for most specimens the leg -to which they belong is not known. The reconstruction is to be used as a key to interpretation of the fossils, and for comparative purposes in a general sense. However, it must be remembered that no one leg of Attercopus fimbriunguis looked exactly like this reconstruction, and in particular, the relative proportions of the podomeres would have varied between legs. There are a number of ways in which the orientation of podomores can be inferred. Inferior and superior are fairly straightforward : comparison of the articulation points with those of living spiders, together with a consideration of the way the leg has to work as a functional unit, is normally sufficient. Assessing which is anterior and which posterior is less easy. The trochanter can be oriented by observing its relationship to the coxa, the orientation of which is known because of the asymmetry in the joint and comparison with extant arachnids. However, there are no trochanters connected to femora which are sufficiently well preserved to enable the following of the orientation down the leg. Since most joints beyond the coxa are symmetrical, their morphology is of little use in orientation, but there is an asymmetrical distribution of slit sensilla and lyriform organs around the distal joints of podomeres. The palpal femur bears a patch of spinules in an inferior position, to one side of its sagittal plane. The function of these spinules is not known, but we are assuming that, whatever their function (see below), they are most likely to occur on the anterior side of the podomere. Therefore, the palpal femur can be oriented, and since this podomere is attached to a patella, this podomere can also, and so on down the leg. A further logical step is required in the assumption that the apparent similar distribution of slit sensilla on palpal podomeres and on the podomeres of other legs reflects a real serial homology. These assumptions have only been made in order to provide an orientation for the reconstructed generalized leg, and not for any other purpose. Should the orientation prove to be incorrect, then the references to anterior and posterior would simply require reversal. PHYLOGENETIC RELATIONSHIPS OF ATTERCOPUS F1M BRIUNGUS Cladistic analysis Characters and character states used in the analysis are listed in Table 2, the data matrix is given in Table 3, and the cladogram in Text-figure 3. The tree was rooted by arbitrarily including an ancestor plesiomorphic for all characters. 250 PALAEONTOLOGY, VOLUME 34 a si text-fig. 2. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). A, reconstruction of a typical walking leg, posterior aspect, b, diagrammatic representation of walking leg joints, distalmost to the left; each joint is viewed from the distal direction with the anterior to the left, the inner circle representing the distal podomere, the outer the proximal podomere; solid circles are articulation points and straight lines are articulation axes, short lines represent slit sensilla. The body-coxa joint is highly diagrammatic; the lower articulation representing the coxosternal attachment, the upper triangle representing the attachment of the coxa to the prosomal marginal cuticle. The upper coxa-trochanter articulation is a movable sclerite set in the arthrodial membrane, which allows rocking. Slit sensilla omitted from coxal distal joint. The trochanter-femur joint is a horizontal pivot. The femur-patella joint is a superior bicondylar hinge, and there is a sclerite embedded in the inferior arthrodial membrane. The patella-tibia joint has a superior articulation, but a close connection of the podomeres inferiorly allows the joint to work as a loose vertical pivot. The tibia-metatarsus joint is a superior bicondylar pivot. The metatarsus-tarsus joint bears antero- and posterosuperior articulations forming a superior bicondylar hinge, but the joint may be uncoupled on relaxation of the muscles, allowing rocking. Shear et al. (1987) presented a cladistic analysis based on 23 of the same characters as used here. The additional characters accommodate the division of the Araneae into Attercopus , Mesothelae, Mygalomorphae, and Araneomorphae. If a character is not discussed below, the discussion will be found in the 1987 paper. Some of the previously used 23 characters have been re-evaluated; in the following discussion, the character number given is from Table 2, and the character number from Shear et al. (1987) is in brackets. Original characters. Character 8 [5] has been recoded. Further investigation of the patella-tibia articulation demonstrated that the joint in living spiders has an additional specialization, compression zone Y (CZY, see later), not present in Attercopus. Further, while the joint is immobilized (fixed) in Amblypygi, considerable movement is possible at that articulation in legs 2-4 of Uropygi and Schizomida (in leg 1 the patella and tibia are entirely fused without trace of a suture). We do not know if the condition on the more posterior legs of Uropygi and Schizomida represents a reversal or the retention of a primitive condition, but we decided to code it as a primitive retention on the grounds of parsimony. Character 9 [16] has also been recoded, because an SELDEN ET AL. DEVONIAN ARACHNIDS 251 table 2. Characters and character states used in the phylogenetic analysis. Characters Plesiomorphic state Apomorphic state 1. cheliceral segmentation 3-segmented 2-segmented 2. plagula ventralis absent present 3. book-lungs absent present 4. sperm flagellum 9 + 2 9 + 3 5. segment 7 broad narrowed 6. eggs not protected protected by secretions 7. lateral eyes minor lenses present minor lenses absent 8. Pa-Ti joint bicondylar hinge 1, rocking, no CZY 2, rocking with CZY 3, immovable 9. labium absent present 10. grouped slits/lyriforms absent present 11. tarsal organ absent present 12. cheliceral poison gland absent present 13. silk glands absent present 14. tibial lyriforms absent present 15. cheliceral fang setose naked 16. cheliceral gland absent present 17. male palp unmodified modified 18. abdominal segments visible hidden 19. tartipores absent present 20. sternum broad, unitary reduced, divided 21. palps leg-like raptorial 22. leg 1 leg-like antenniform 23. posterior sucking stomach present absent 24. abdominal flagellum absent present 25. palp coxae free fused 26. postabdomen 2-segmented 3-segmented 27. abdominal tergites entire divided 28. fimbriae on claws absent present 29. spinules on palpal Fe absent present 30. Ti-Mt organ absent present 31. clavate trichobothria absent present 32. anterior media spinnerets absent 1, present 2, lost 33. chelicerae orthognath labidognath 34. cleaning brush on palp absent present 35. anal glands absent present 36. male flagellum unmodified modified 37. central nervous system partly in abdomen wholly in prosoma 38. trichobothria present absent examination of specimens has convinced us that a labium (sternite of the palpal segment modified as a lower lip) does not in fact occur in Amblypygi, Uropygi, and Schizomida. In amblypygids, a long projection goes forward from the sternite of the first leg, but could not function as a labium. In uropygids and schizomids, the palpal sternum is an immovable pentagonal sclerite and the ventral wall of the preoral cavity (camerostome) is formed by the fused palpal coxae. Character 5, the narrowing of segment 7, has replaced [18]: presence or absence of a pedicel. We think that the key feature here is the reduction in width of that segment, which occurs to a greater (Araneae, Amblypygi) or lesser (Trionotarbida, Uropygi, Schizomida) degree in all of the taxa involved. 252 PALAEONTOLOGY, VOLUM E 34 table 3. Data matrix used in the phylogenetic analysis. 0 = plesiomorphic state, 1 = apomorphic state, 2, 3 = alternative apomorphic states, ? = character state uncertain. See text for details. Characters 12345 1 67890 12345 2 67890 12345 3 67890 12345 678 Trigonotarbida 11171 ?0000 00000 00000 00?00 01000 00000 0?1 Attercopus 111?? ??1?1 urn l??0? 00 ??0 ??1 10 0?000 0?1 Mesothelae mu 1 121 1 urn 1 1000 00000 00001 11000 010 Mygalomorphae mu 1 121 1 mu 11110 00000 00000 02000 010 Araneomorphae mu 11211 m 1 1 11110 00000 00000 01100 010 Amblypygi mu 1 1301 10000 00001 11000 00000 00010 010 Uropygi mu 11101 10000 00001 mu 10000 00001 000 Schizomida mu 11101 10000 00001 urn 10000 00000 100 New characters. Characters 10 and 14: slit sensilla are unique to chelicerates. We have assumed . that the primitive arrangement was scattered, single slits on most or all body surfaces, and these still occur in all arachnids. However, the slits, which function as cuticular strain gauges, are found in greater numbers near articulations or points where the cuticle is likely to be stressed (Barth 1978, 1985). This has led in turn to the formation of loosely organized groups of slits, and thence to tightly coupled, parallel slits, commonly surrounded by a cuticular border, known as lyrifornr organs. In true lyriform organs the slit sensilla are neurally integrated to act as a single organ; this integration is recognized morphologically where the slits are as close together as their individual widths, and are parallel to each other. They may change in length gradually across the organ, giving the appearance of the arrangement of strings in a lyre or harp. A multiplicity of lyriforms is clearly apomorphic, and in character 14, the presence of lyriforms on the leg tibiae stands in for this increase in their number. In trigonotarbids, we have not detected grouped slits or lyiforms, though large slits occur in greater numbers near the distal ends of podomeres (see Shear et al. 1987, figs 1 1. 46, 79-81). Lyriforms occur in amblypygids and uropygids only on the distal ends of the metatarsi of legs 2-4, and are oriented parallel to the long axis of the leg; spiders have this metatarsal lyriform, which is oriented perpendicular to the long axis of the leg, as well as many additional lyriforms on other podomeres which are oriented parallel to the long axis (Barth 1985; Barth and Stagl 1976; Moro and Bali 1986). Character 1 I : typical tarsal organs (Blumenthal 1935; Forster 1980) occur on the walking leg tarsi of all living Pulmonata (Amblypygi and spiders, Forster 1980. and pers. obs. ; antenniform legs of Amblypygi, Foelix et al. 1975 (‘pit organ'); walking legs of Uropygi, pers. obs. and R. Forster, pers. comm.; walking legs of Schizomida, pers. obs. and R. Forster, pers. comm.). We have not detected this organ on the tarsi of trigonotarbids, but it is present in Attercopus. While similar structures are found on the tarsi of scorpions and ticks (Foelix and Axtell 1972; Foelix and Schabronath 1983), they appear ultrastructurally different and their homology has not been established. Thus the presence of the tarsal organ is treated here as a synapomorphy for the orders of Pulmonata excepting Trigonotarbida, though it may later be shown to be more widespread in Arachnida. Character 15: a naked cheliceral fang is apomorphic by comparison with the setose condition of the palp and walking legs, with which the chelicera is serially homologous. Among the Pulmonata, a naked cheliceral fang is found only in spiders, al! other pulmonate orders have a brush of setae on the fang (see, for example, PI. 7. figs 5 and 6; Shear et al. (1987) figs 7, 67, 68). Character 16: the cheliceral gland described by Forster and Platnick (1984) has been reported only in spiders; it has been found in all species so far examined from a wide selection of families (R. Forster, pers. comm.). Raymond Forster (pers. comm.) stated that he has found a series of scattered pores near the midpoint of the ventral surface of the chelicera in amblypygids, which he considers a cheliceral gland. Using light microscopy (including oil immersion examination of cleared cuticle) we were not able to confirm these observations, but a purposeful search for the gland may reveal it in orders other than Araneae. In pseudoscorpions, glands also open on the chelicera (Vachon 1966), but they are very distinct in appearance and probably not homologous. We propose the presence of this distinctive gland is yet another autapomorphy for the order Araneae. Character 18; in opisthothele spiders, the segmentation of the abdomen is suppressed and is either entirely SELDEN ET AL.. DEVONIAN ARACHNIDS 253 concealed from external view, or revealed only on the maturity of males of a few species of mygaloniorphs, and even then only in the anterior part. This is a synapomorphy for Mygalomorphae and Araneomorphae. Character 19: tartipores - these peculiar structures, like small, collapsed pastries (hence the name), evidently mark the position of spigots on the spinnerets in previous mstars (Kovoor 1986; Coddington 1989). They do not occur in Attercopus nor in mesotheles (pers. obs. on Liphistius sumatranus and L. malayanus). The number of spigots on spider spinnerets increases with each instar; in mesotheles the increase is accomplished by adding more pseudosegments to the spinneret. We consider this mechanism primitive, and the presence of tartipores synapomorphic for mygalomorph and araneomorph spiders. Character 26: a two-segmented postabdomen is present in trigonotarbids, spiders, and amblypygids. Counting segments shows that uropygids and schizomids have added a third, basal segment (probably by the narrowing of the segment just in front of the primitive two-segmented postabdomen), which we consider a synapomorphy for that group, correlated with the postanal abdominal flagellum. Characters 28 and 29: fimbriate claws and palpal femoral spinules are autapomorphies of Attercopus, by outgroup comparison and the criterion of ‘special structures’. Characters 30 and 31 : a highly specialized organ for detecting deflection of the metarsus with respect to the tibia is present among spiders only in living mesotheles (Platnick and Goloboff 1985). Likewise, special club- shaped trichobothria (Foelix 1985) are unique to this group (Platnick and Goloboff 1985). Character 32: by outgroup comparison, the loss of the anterior median spinnerets is autapomorphic for mygalomorph spiders. We might add here that there are other spinneret and spigot characters that may prove useful for phylogenetic analysis among spiders; some of these have already been described by Coddington (1989) and others are under study by J. M. Palmer and J. A. Coddington. Character 33 : labidognath chelicerae are found only in araneomorph spiders and are autapomorphic for that group. Character 38: the distribution of trichobothria in the Arachnida has been discussed by Kaestner (1968), and Reissland and Gorner (1985). They are found in spiders, amblypygids, uropygids, schizomids, palpigrades, scorpions, pseudoscorpions, and mites, but not in solifuges, ricinuleids, or opihonids. Their occurrence in scorpions and palpigrades, both considered primitive arachnids, and their general appearance elsewhere argues for considering their absence in any arachnid a loss. We have not found trichobothria in trigonotarbids, nor in Attercopus. Shear et al. (1987) described trichobothria in the supposed trigonotarbid Gelasinotarbus bonamoae , but new studies of this animal have convinced us that it is not, after all, a trigonotarbid, nor does it seem to be a spider. The loss of trichobothria is thus proposed as another autapomorphy for Trigonotarbida. We are more concerned about the complete lack of trichobothria encountered during our high-magnification studies of well-preserved podomeres of Attercopus. We have found no mention in the literature of spiders without trichobothria, and R. Forster and N. Platnick, who have surveyed hundreds of species using SEM, reported that they have found no spiders which lack these sense organs (R. Forster, pers. comm.). Had we not found tarsal organs and longitudinally oriented lyriforms on Attercopus podomores, as well as having been able to match their cuticle to that of the isolated spinneret, we would question our assignment of these fossils to Araneae. We must regard the loss of trichobothria in Attercopus as an autapomorphy independent of their loss in trigonotarbids. Cladogram. Using these 38 characters, we have produced a 36-step cladogram (Text-fig. 3) with a consistency index of 0 97. In an earlier, preliminary report on the spinneret of Attercopus fimbriunguis, Shear, Palmer et al. (1989), were able to narrow down the number of possible cladograms for spider sub- and infra- orders to three, arguing as follows. Recent views of spider evolution divide the Order Araneae into two suborders. Suborder Mesothelae includes a small number of species today restricted to southeast Asia, Indonesia, and Japan; they are united by a number of synapomorphies, including a peculiar sense organ between the tibiae and metatarsi of the legs (see above). Mesotheles are better known to arachnologists for their primitive characters, including an externally segmented abdomen and the possession of eight (rarely seven) spinnerets, which are located not at the end of the abdomen, but near the middle of its ventral surface. Suborder Opisthothelae includes all other spiders, in which the number of spinnerets has been reduced to six, four, or two and moved to the posterior end of the abdomen, which is not externally segmented. Within this group, Mygalomorphae (‘tarantulas’ in the North American sense) have lost all vestiges of the anterior median spinnerets, while Araneomorphae carry a cribellum (repeatedly lost in many lines) 254 (0 -a IS PALAEONTOLOGY, VOLUME 34 0 ) rd X O- o; C3 JX 00 IX o D 03 T3 £ o N U cn text-fig. 3. Cladogram of relationships between Attercopus gen. nov., infraorders of Araneae, and orders of Pulmonata, as inferred by the cladistic analysis (see text for details). The cladogram has a length of 36 and a consistency index of 0-97. homologous to the anterior median spinnerets of mesotheles, and have chelicerae rotated to the labidognath position, so that the fangs point toward one another. The spinneret is described in detail below. Using information from the description. Shear, Palmer et al. (1989) were sure the spinneret could not have come from the living clade of mesotheles, because in mesotheles the large lateral spinnerets of each pair are pseudosegmented, with spigots in ranks of 2, 3, or 4 on the mesal surface of a pseudosegmental ring, and the smaller, single-articled median ones bear only a single spigot. Because the Devonian spinneret is not pseudosegmented, yet bears more than one spigot, it could not have come from a mesothele spider similar to those living today. Araneomorph spiders are ruled out because the spigots of their spinnerets are strongly differentiated from one another and from those of mygalomorph spiders in characteristic ways, and all spigots on the fossil specimen are of the same size and shape. Mygalomorph spiders have single-articled posterior median spinnerets with numerous spigots arranged as they are in the fossil. The presence of undifferentiated, or only weakly differentiated, spigots that are more densely clustered near the tip of the spinneret is consistent with mygalomorph spider posterior median spinneret anatomy. However, both mygalomorph and araneomorph spinnerets have peculiar nipple-shaped structures called tartipores (see above), which represent the positions of spigots in previous instars. Tartipores are not present on the Devonian spinneret. In addition, mygalomorph spinnerets usually have two types of spigots present. Finally, the form of the spigots themselves does not, in detail, agree with that of mygalomorph spigots. Mygalomorph spigots usually have an articulated shaft, which joins the base by means of a well-defined, sleeve-like fold. At least the distal third of the shaft has sculpture. However, the rastelloid clade of mygalomorphs have non-articulated shafts and extremely fine sculpture, visible only when viewed with the SEM. Diagenetic changes in the fossil spinneret may have made it impossible to resolve such fine detail as the distal shaft sculpture. Mesothele spigots, on the other hand, are uniform in morphology, with a broad, conical base and a long, gradually tapering, unsculptured distal shaft that merges smoothly into the base. The spigots of the fossil are of this type. Considering the absence of tartipores, of a sleeve-like fold at the base SELDEN ET AL. DEVONIAN ARACHNIDS 255 of the spigot shaft, and the likelihood that distal sculpture is absent, the spigots are more like mesothele spigots than mygalomorph ones. Therefore, the combinations of apomorphies found in the three living clades would seem to exclude the fossil from all of them. The problem then becomes placement of the fossil as a sister group of one, two or all of these clades. The presently accepted 3-taxon statement for the groups of spiders so far discussed is: Mesothelae (Mygalomorphae (Araneomorphae)). The fossil spinneret is probably not from a spider belonging to the sister group of either Araneomorphae or Mygalomorphae, because to place it in either of those positions would require the ad hoc secondary loss of tartipores in the fossil clade. Thus, either Attercopus fimbriunguis would prove to be the sister group of all other spiders, of only mesotheles, or of opisthotheles, leaving a basal trichotomy in the cladogram of spider suborders. Shear, Palmer et al. (1989) ended their argument at this point, because additional Attercopus fragments had not yet been identified, and no characters were available to resolve the trichotomy. Careful examination of the legs of A. fimbriunguis has provided evidence that the trichotomy can be resolved in favour of this Devonian clade as the sister group of all other spiders. This evidence comes from the structure of the patella-tibia joint, which, as we (Shear et al. 1987) and others (Manton 1977; van der Hammen 1977, 1985, 1986; Shultz 1989) have shown, is of great phylogenetic significance. In trigonotarbids, this joint is a simple bicondylar hinge, probably the plesiomorphic form at least for Pulmonata (Shear et al. 1987). In the other pulmonate orders, it becomes a specialized rocking joint, with a single dorsal condyle and held together with strong muscles. In spiders, three lyriform organs are found on the posterior surface and two on the anterior, and this rich array of proprioceptors is associated with the complex movement of this joint in more than one plane (Manton 1977). The additional complex mobility of the patella-tibia joint is conferred at least in part by a posterior emargination, occupied by lightly sclerotized cuticle and extending proximally from the distal edge, which Manton called ‘compression zone Y’ (CZY). The presence of CZY pushes the middle lyriform of the three posterior ones almost to the proximal edge of the podomere. However, in amblypygids, this joint, while retaining vestiges of the rocking articulation, is nearly immobile. In uropygids and schizomids the first leg patellae and tibiae are entirely fused and no separate patella appears. On the walking legs (2-4) the joint is movable, but, as discussed above, we are not certain if this mobility is primary or secondary. The condition of this joint in A . fimbriunguis is of great interest ; the rocking articulation is present but CZY is absent. Functionally, this suggests substantially less mobility at this joint than in other spiders, but more than in trigonotarbids. It is suggested that the common ancestor of Araneae and the ‘pedipalp’ orders (Uropygi, Ainblypygi, Schizomida) had the type of joint found in A. fimbriunguis, which is still present in Uropygi and ‘locked’ in the legs of Amblypygi; the presence of CZY in Mesothelae and Opisthothelae is a synapomorphy for them alone. The meaning of this is that A. fimbriunguis represents a clade of spiders forming the sister group to Mesothelae + Opisthothelae, and could justifiably be made the single member of a new suborder. There are several interesting autapomorphies for the Devonian spider. Most obvious are the fimbriate claws, described above. These do not occur on any other spider known to us and differ strongly from the smooth claws of trigonotarbids. Secondly, the patches of acute spinules at the inner base of the palpal femora would appear to be unique among spiders. Somewhat worrisome, but a potential third autapomorphy, is the absence of trichobothria. It may be that they are present and we have not found them, but given our close examination of the material, this is extremely unlikely. These additional observations have an effect on the cladogram published by Shear et al. (1987). One result has been to affirm the basal position in the cladogram of Trigonotarbida as the plesiomorphic sister group of all the other included orders of Pulmonata. The evidence lies in the lack of tarsal organs and lyriforms in trigonotarbids, and the presence of these features can be considered synapomorphic for the other orders. (However, if the ‘tarsal organ’ of scorpions and the 256 PALAEONTOLOGY, VOLUME 34 Haller’s Organ in ticks are homologous to the tarsal organ of spiders, amblypygids and uropygids, then the loss of it may be an autapomorphy of trigonotarbids.) The basal, plesiomorphic position of the trigonotarbids, which in general resemble ‘spiders without spinnerets’, emphasizes the strongly derived nature of Amblypygi, Uropygi, and Schizomida. Secondly, the earlier conclusion that the Amblypygi are the sister group of Uropygi + Schizomida, and not of Araneae, is reinforced. It can be further suggested that the key adaptations of the ancestor of the ‘pedipalp’ clade were the development of raptorial palps, probably articulating in the horizontal plane, antenniform first legs used as a ranging device for palpal strikes, and finally, as Manton (1977) wrote, partial or total immobilization of the patella-tibia joint to strengthen the knee, which must undergo extreme flexure in connection with the other modifications of legs to allow the animals to slip sideways into narrow crevices. In uropygids, the joints are far more mobile on legs 2-4 than in amblypygids, but the patella -tibia joint has been entirely lost in the first legs. Schizomids may be seen as a derived clade of uropygids; the movement of their palps in the vertical plane and the subdivision of the carapace are secondary changes designed to increase the flexibility of the whole body to allow for movement in the small spaces between soil particles. But the fused patellotibia of the first leg remains as a vestige of their common ancestry with uropygids. It should also be recognized that naked cheliceral fangs, cheliceral glands, transversely oriented metatarsal lyriforms, and the presence of lyriforms on podomeres other than metatarsi, are probable autapomorphies of Araneae, joining the better known features of cheliceral poison glands, opisthosomal silk glands and spinnerets, and the palpal intromittent organ in mature males. SYSTEMATIC PALAEONTOLOGY Order araneae Clerck, 1757 Emended diagnosis. Pulmonata with paired abdominal appendages modified as silk-spinning organs; chelicera with cheliceral gland; cheliceral fang with poison gland opening, and without setae; adult male palps modified for sperm transfer; numerous longitudinally oriented lyriform organs present on walking legs in addition to transverse one on distal metatarsus. Genus attercopus gen. nov. Derivation of name. English dialect (from Old English) attercop, a spider. Tvpe and only known species. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). Diagnosis. Spider with patch of minute cuticular spinules on proximal infero-?anterior surface of palpal femur; minute cuticular fimbriae on inferior surface of all tarsal claws; without longitudinal emargination on posterior side of distal edge of patella of walking legs. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987) Plate 1; Plate 2, figs 1, 2, 4-8; Plate 3; Plate 4; Plate 5, figs 1-3, 5; Plate 6, figs 1, 2, 4, 5; Text-figs 4; 5a, b, d-h; 6; 7; 8; 9a, c; 10; 12. 1987 Gelasinotarbus ? fimbriunguis . Shear, Selden and Rolfe; Shear et al., pp. 60-65, 71, figs 128-140. 1987 Arachnida Incertae sedis B. Shear, Selden and Rolfe; Shear et al., pp. 70, 71, figs 151-157. Type specimens. Listed in Shear et al. (1987), p. 60. Additional material. A complete list of the specimens referred to this species is deposited in the British Library, Boston Spa, Yorkshire, England, as Supplementary Publication No. SUP 14040, 5 pp.; see Repository above for availability of this publication. SELDEN ET AL.: DEVONIAN ARACHNIDS 257 Diagnosis. As for the genus. Description Cuticle. The cuticle pattern of Attercopus fimbriunguis is characteristic, and readily identifiable. The surface sculpture was described in Shear et al. (1987, p. 64) as being reticulate, with one side (distal, normally) of each polygonal cell being thicker than the other sides; the sculpture of Incertae sedis B was described (Shear et al. 1987, p. 70) as being similar. This sculpture pattern can be confirmed here, but with added detail: first, the distal side of each polygon of the reticulum actually forms the proximal side of the distally adjacent cells, and second, the sculpture dissolves into smooth cuticle in places, such as over most of the distal parts of the tarsus and the chelicera. Two distinct sizes of setal socket and the presence of long, fine setae without bifid tips were mentioned by Shear et al. ( 1 987) ; the cuticle of Incertae sedis B was described as lacking this bimodality of setal sockets. The present study confirms that two sizes of setal sockets may be present, for example, on most of the leg segments there are small sockets with long, fine setae, and larger sockets bearing larger, long setae. This bimodality can, in fact, be seen on the published figures of Incertae sedis B (Shear et al. 1987, figs 151-154), but it is somewhat variable, and is not, alone, diagnostic for the genus. Many of the setae can be seen to be finely serrate, and the macrosetae bear serrae on their convex surface. Most characteristic of Attercopus fimbriunguis is the presence of very small cuticular organs scattered across the cuticle surface (PI. 1, fig. 1). Their distribution may be quite dense, for example on the spinneret (Text-figs 10 and 1 1 a, b). At low magnification (up to about x 100), these appear very much like small setal sockets: a circle or oval of dark cuticle, about 0-006 mm in diameter. At higher magnification, however, the central pore is revealed as a slit, and thus these organs are true slit sense organs. In addition, larger slit sensilla are found at the joints. They may occur singly, at the distal end of the tarsus for example, in groups, such as those adjacent to the distal articulations of the femur, or in lyriform organs, examples of which can be seen at the distal ends of the patella, the tibia and the metatarsus. The distribution of the larger slits and lyriforms on the generalized leg is shown in the reconstruction (Text-fig. 2). A major surprise in the present study was to find no evidence of trichobothria on any of the leg segments. The report of one on specimen 411.7.19 (Shear et al. 1987, p. 70) is incorrect; study of many more specimens of femora has shown that these podomeres are susceptible to the occurrence of circular dark patches, the origin of which is unknown, but which may be pre- or post-mortem fungal or parasitic attacks. That the dark patches occur only rarely, and then in different places on the same podomere (e.g. on palpal femora), is evidence that they are not a feature of A. fimbriunguis. Carapace and abdomen. Three pieces of cuticle may represent parts of the carapace. 2002. 12. 102 is a sheet of typical reticulate A. fimbriunguis cuticle, with small slit organs scattered over the surface, which lacks setal sockets except at one end where large sockets occur, adjacent to two large, oval holes; nearby are what appear to be the edges of two further holes (PI. 1, fig. 1 ). On one side of the specimen is an edge with a narrow doublure, and that part of the specimen which is folded over also has an edge to it. The holes are interpreted as possible eyes, and the edges as the carapace margin. The margin is not scalloped, as it is in trigonotarbids. A similar edge, with a narrow doublure, occurs on specimen 329.31. It is noteworthy that the carapace of Liphistius is almost devoid of setae except around the margins, and adjacent to the group of eyes (which are situated in the midline at the anterior edge of the carapace) some large setae are present. Specimen 41 1 . 1 1 . 3 is a chelicera of A. fimbriunguis which is superimposed on a large sheet of A. fimbriunguis cuticle. The cuticle sheet is torn down the centre and displaced so that it is overlapping; short lengths of edge can be seen on the sheet, but no eyes are present. Three characteristics suggest that this specimen belongs to the carapace : first, the size of the sheet in comparison to the size of the chelicera, second, the lack of podomere structures, and third, the features of the presumed carapace fragment 2002 . 12 . 102 mentioned above (lack of setal sockets except near the presumed anterior edge) also occur in this specimen. Sternum. The sternum, which consisted of a cushion-like surface in life, occurs in the fossil as a rectangular strip of cuticle, about five times as long as wide (not all of it may be preserved), on specimen 41 1 . 19.83 (PI. 2, fig. 2). Articulations are present at the points where the coxae meet the sternum. There are three pairs of these visible in the specimen, one side of each pair adjacent to each of the two coxae preserved. The anterior end does not preserve this feature, and the posterior end is missing. If the well-preserved coxa on this specimen belongs to leg 4 (see below), then the sternum is probably produced backward between coxae 4. Chelicera. The chelicera (PI. 1, figs 2-8) is equant in shape. Specimen 334. la. 7 is nearly complete and shows proximal articulations along a joint plane which is nearly at right-angles to the tooth row. The articulations 258 PALAEONTOLOGY. VOLUME 34 eyes? ar text-fig. 4. Atter copus fimbriunguis (Shear, Selden and Rolfe, 1987), explanatory drawings for specimens illustrated on Plate 1. a, 2002. 12.102, anterior part of carapace, small slit sensilla shown on internal surface only, b, 2002 .12.90, distal end of chelicera. c, 445 . 1 a . 7, whole chelicera with fang, proximal joint edges shown at left (near side is partly detached), foreign cuticle fragment (X) lying behind specimen, d, 411 .7.33, nearly complete chelicera lacking fang, showing tooth row and cheliceral gland (both on far side). E, distal end of chelicera with fang, tooth row (distal end partly obscured by artefact). Scale bar represents 0.5 mm for all specimens; see materials and methods for abbreviations and conventions. EXPLANATION OF PLATE 1 Figs 1-8. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1, anterior part of carapace showing possible eyes and large setal sockets at anterior, also typical cuticle sculpture and small slit sensilla elsewhere, explanatory drawing in Text-figure 4 a, 2002 . 12 . 102, x 70. 2, distal end of chelicera showing tooth row, fang articulations, and position of cheliceral gland, explanatory drawing in Text-figure 4b, 2002. 12.90, x 107. 3, distal end of tooth row of specimen shown in fig. 2, showing cheliceral gland, 2002.12.90, x215. 4, chelicera, lacking fang, showing general shape, tooth row, and position of cheliceral gland, explanatory drawing in Text-figure 4 D, 41 1 .7.33, x 95. 5, distal end of tooth row of specimen shown in fig. 3, showing cheliceral gland at end of tooth row, 411.7.33, x 235. 6, whole chelicera, showing general shape, articulation of fang, and poison gland opening, foreign cuticle fragment lying across part of tooth row, explanatory drawing in Text-figure 4c, 334. \a.l x 55. 7, distal part of chelicera showing tooth row, fang articulation, poison duct opening, and serrated ridge on fang, artefact lying across distal end of tooth row, explanatory drawing in Text-figure 4e, 329.22.9, x 132. 8, distal part of specimen shown in figure 6, showing details of fang articulation, poison gland opening, serrate ridge, and tooth row, 334. la. 7 x 105. PLATE I SELDEN et al Attercopus 260 PALAEONTOLOGY, VOLUME 34 text-fig 5. For legend see p. 262. EXPLANATION OF PLATE 2 Figs 1, 2, 4-8. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1, femur in posterior aspect, distal to right, explanatory drawing in Text-figure 5a, 334. la. 9, x 64. 2, left coxa (probably of leg 4), sternum (top to right), fragment of coxa ?3, and piece of marginal cuticle of prosoma, posterior aspect, explanatory drawing in Text-figure 5b, 411 . 19.83, x 62. 4, femur, posterior aspect, distal to left, explanatory drawing in Text-figure 5d, 334. la. 6, x93. 5, complex grouping of podomeres, including chelicera (dark mass on right), palpal femur, leg 72 femur, patella, tibia, and tarsus (all on left), and plant cuticle and spores, explanatory drawing on Text-figure 5e, 329.69, x 80. 6, trochanter, distal aspect, inferior to top, fragment of coxa attached at bottom left, explanatory drawing in Text-figure 5f, 411.19.1 02, x93. 7, three coxae (two at top, one at bottom left) and trochanter (bottom right), explanatory drawing in Text-figure 5h, 334. la. 9, x 66. 8, coxa, posterior aspect, explanatory drawing in Text-figure 5g, 411 . 19.250, x 117. Fig. 3. Ecchosis pulchribothrium gen. et. sp. nov. Ventral part of coxa, posterior aspect, explanatory drawing in Text-figure 5c, 2002.9. 13, x 1 10. PLATE 2 SELDEN et al., Attercopus , Ecchosis 262 PALAEONTOLOGY, VOLUME 34 are arranged in such a way that it is difficult to envisage this chelicera being anything other than orthognath. The teeth are in a single row of about 8-1 1 teeth (8 in small, 1 1 in large specimens). The smallest teeth occur near the fang tip, the larger occur closer to the basal articulation of the fang, and largest of all is third or fourth from the end of the row nearest the fang articulation. There are no subsidiary teeth, and the teeth are not greatly different in size, the smallest is about half the size of the largest. The fang curves gently to a point adjacent to end of tooth row. A possible orifice for the poison gland may be seen subterminal to the fang tip on specimens 334. la. 7 and 329.22.9 (PI. 1, figs 7 and 8); other specimens do not show the fang tip. The inner surface of the fang bears a ridge of fine serrations extending the length of the tooth row. Most of the cuticle surface bears only a sparse scattering of setal sockets; setae are numerous near the teeth, but do not occur in a comb or brush. The setae are finely serrate. There are no setae on the fang. The cheliceral gland openings can be seen on specimens 2002 .12.90, 329 . 3 1 a . M I , and 4 1 1 . 7 . 33 at the end of the tooth row near the fang tip (PI. 1, figs 3 and 5). A few slit sensilla occur adjacent to the fang articulations. Coxa. Coxae are present on a number of specimens, but commonly these bear numerous other podomeres compressed together (on PI. 2, fig. 7 three coxae and a trochanter occur together), so the coxal morphology is better interpreted from the few isolated examples (e.g. PI. 2, figs 2 and 8). Understanding the coxal morphology is aided by study of the coxa of Liphistius in conjunction with the fossils. The coxa on specimen 411.19.83 probably belongs to leg 4, since it occurs at the rear of the sternum (see below) which appears to have attachment points for at least two, and probably three, coxae in front. If this coxa is not leg 4 then it would be leg 3. Adjacent, and anterior to, the main example on this specimen, is a small portion of the medial side of the next coxa anterior, also attached to the sternum, with some membrane between the two. The coxa is of the boat-like form typical of most arachnids, although on this specimen the ventral surface is mainly missing. The anterior dorsal edge runs with a thickened line from an attachment point with the sternum towards the distal margin, but about two-thirds of the way along towards the distal margin, it dips ventrally; the next part up to the distal edge is missing. The posterior dorsal edge is also thickened in a line, which runs horizontally for about one-third of the way to the distal edge then dips towards the ventral, for a distance of about half the length from the sternum to the dip, and then runs to the distal edge at this lower elevation. Specimen 41 1 . 19.250 (PI. 2, fig. 8) is most useful for reconstructing the shape of the podomere. The anterior articulation at the distal joint lies at the end of a long ridge of thickened cuticle (the costa coxalis) which extends in a proximodorsal direction towards, and closely approaching, the anterior dorsal edge. The posterior articulation consists of a sclerite which originates at the posterior edge of the joint in an anterior position, and runs dorsally, separated from the joint edge by membrane (see PI. 2, fig. 8). The morphology of the distal joint is very similar to that of the Recent Liphistius. The strip of cuticle running along the dorsal side of the coxae, the lateral marginal plate, and also seen in Liphistius , can be seen on 411 . 19.83. On this specimen the posterior sclerite is folded onto the anterior side of the distal joint. Trochanter. Trochanter morphology is difficult to interpret because so many of the few specimens are folded together with coxae or femora. The best specimens are 334. la. 9 (PI. 2, fig. 7), which is attached to coxae, but relatively easy to make out, and 411.19. 102 (PI. 2, fig. 6), a separate trochanter. The trochanter is a short podomere, the inferior surface is nearly twice as long as the superior and was bulbous in life. The interior surface bears numerous large setal sockets. Proximal articulations consist of a prominent, thick triangular projection which marks the anterior articulation, slightly inferior in position; the posterior articulation shows text-fig. 5. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987) (a, b, d-h) and Ecchosis pulchribothrium gen. et sp. nov. (c), explanatory drawings for specimens illustrated on Plate 2. a, 334. la. 9, femur of walking leg, posterior aspect, b, 41 1 . 19.83, left coxa (of leg 4?), sternum, fragment of next anterior coxa, and piece of marginal prosomal cuticle (folded and twisted), ventral surface of coxa ?4 absent, c, 2002.9. 13, ventral part of coxa of Ecchosis , ventral surface to lower right, distal joint to left (superior side absent), spore in black. D, 334. la. 6, femur of walking leg, posterior aspect, e, 329.69, complex group of podomeres (setae omitted), plant cuticle, spores (in black). F, 411.19.102, trochanter, distal aspect, inferior to top, posterior to left, including fragment of coxa (shaded) with posterior articulation. G, 41 1 . 19.250, ventral half of coxa, torn and folded, posterior aspect, inferior to lower right, distal joint to left (superior side absent), setae omitted. h, 334. la. 9, three coxae and trochanter, two coxae at top, one at lower left, trochanter at lower right, setae and interior surfaces of upper two coxae omitted. Scale bar represents 0.5 mm for all specimens; see material and methods for abbreviations and conventions. SELDEN ET AL.: DEVONIAN ARACHNIDS 263 only as a darkened edge, the major part of this being on the coxal side of the joint. However, 411.19.102 (PI. 2, fig. 6) shows a portion of the coxa attached at this point, and reveals the detail of the articulation well. The distal joint bears anterior and posterior articulations; these are not well expressed, being only dark but discrete edges to the podomere. They are connected by a fold of cuticle on the inferior edge which marks the distal termination of the bulbous part of the inferior surface. Superoanteriorly on the distal edge there is a group of slit sensilla; a group of slit sensilla occurs in exactly the same position in Liphistius, and is useful for identifying the orientation of loose podomeres. Specimen 329.59, which was figured by Shear et al. (1987, fig. 140) as a possible median organ of some kind, is now reinterpreted as half of a trochanter. The cuculliform shape described by Shear et al. (1987, p. 64), is incorrect, since there is only a single layer of cuticle present, the inferior surface of the trochanter, and basally, the two proximal articulations can be seen. Femur. The femur is an easily recognizable podomere, and occurs on many slides. The characteristic palpal femur, with a patch of spinules, is described below. The femur is a long podomere, with a bicondylar horizontal pivot joint proximally (PI. 3, fig. 8) and a greatly inferiorly emarginated distal joint with a dorsal bicondylar hinge. Specimens 334. In. 6, 334. In. 9, 2002.12.79, and 329.3 (PI. 2, figs 1 and 4; PI. 3, figs 1 and 2) show typical podomeres. Longitudinal rows of setal sockets occur on the superior surface, and similar rows are found on the inferior surface. The anterior and posterior sides are devoid of setae. The articulations on the proximal joint occur on pronounced promontories. The distal joint bears curved rows of slit sensilla adjacent to the articulations, which are situated superoposteriorly and superoanteriorly. Fewer slits occur in the anterior group than in the posterior. Some variation in the femora is noticeable, in greater or lesser amounts of emargination at the inferior side of the distal joint. This can be accounted for by differences between the legs. In Liphistius , the emargination is greatest on legs 2 and 3, whereas on leg 4 and on the palp there is less emargination; the least emargination of all occurs on leg 1. The amount of emargination is correlated with the degree of flexure required during stepping of the legs, and the activities of the palp. Specimen 329. 3 la. Ml (PI. 3, fig. 7) shows a femur with little emargination in connection with a chelicera and palpal femora; this presumably belongs to leg 1 . The palpal femur is not very large (the largest is specimen 329.63, figured in Shear et al. 1987, fig. 155), and bears a patch of cuticular spinules on its proximal infero-?anterior surface (PI. 3, fig. 4). The spinules are not setae, but cuticular projections, and were figured by Shear et al. (1987, figs 156 and 1 57). By assuming that these were used towards the mouth or towards the anterior/mesal, then they would be on the inner, proximal prolateral side. The bases for the supposition that this podomere is palpal are, first, that modifications to the prosomal limbs in spiders are more likely to affect the palp than any other leg, and second, that when this podomere is found connected together with other organs, it is found adjacent to the chelicera in all cases. Apart from the patch of spinules, the palpal femur is similar to the other femora. There is a bicondylar pivot joint with a horizontal axis at the proximal end of the podomere, and a superior bicondylar hinge distally, with a greatly emarginated inferior surface. Rows of slit sensilla occur adjacent to the distal articulations. Setae on the podomere occur in rows; principally two rows superiorly, two inferiorly, and one retrolaterally. Specimens 329.3 (PI. 3, fig. 2) and 329.63 show right femora, and 329. 1 (PI. 3, fig. 4) shows the left femur in connection with the patella. Two palpal femora are present on 329. 31a. Ml, together with the chelicera, and other podomeres. Patella. The patella is a short podomere, with the curved superior surface more than twice the length of the inferior surface. Specimens are shown on Plate 3, figures 3-7, and Plate 4, figures 2 and 5. The proximal joint bears supcroanterior and superoposterior articulations corresponding to those distally on the femur. The inferior part of this joint, however, is emarginated, more so posteriorly than anteriorly, and two dark, recurved areas are present in inferoposterior and inferoanterior positions. By comparison with living spiders, amblypygids, and uropygids, these areas mark the sites of suspension of the arcuate sclerite: a distally procurved sclerite lying in the membrane of the greatly emarginated Fe-Pa joint, and facilitating flexion from the extreme extension possible at this joint. The sclerite itself seems unlikely to be preserved, but nevertheless, one appears to be present on specimen 329. 31a. Ml, on leg ?1 (PI. 3, fig. 7). Distally, there is a strong superior articulation; the distal joint is not a bicondylar pivot, as stated by Shear et al. (1987, p. 63), but is monocondylar. Three lyriform organs are situated in an inferoposterior position, and two occur inferoanteriorly, on the distal joint. Of especial interest here, is the lack of a pronounced emargination (CZY) on the posterior side of the distal joint, seen in Liphistius and all other spiders. In this respect, the A. fimbriunguis patella more closely resembles that of the ambulatory legs of uropygids. The superior surface bears about four large setae in addition to the smaller ones. Smaller setae occur elsewhere, especially superoproximally and inferiorly. 264 PALAEONTOLOGY, VOLUME 34 text-fig. 6. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987), explanatory drawings for specimens illustrated on Plate 3. a, 2002. 12.79, posterior aspect of right walking leg femur, spores omitted, b, 329.3, anterior aspect of right palpal femur, spore omitted, c, 329.59, distal end of left femur and attached patella, posterior aspect, d, 329.1, posterior aspect of left palpal femur and attached patella, spores omitted, e, 329 . 3 1 a . M 1 , detail of joints of left femur and patella, including, sclerite, posterior aspect, f, 334 . 1 6 . 86, femur and patella, foreign cuticle omitted. G, 334. 1 6. 12, distal femur and patella. Scale bar represents 0.5 mm for all specimens; see material and methods for abbreviations and conventions. EXPLANATION OF PLATE 3 Figs 1-8. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1, femur, posterior aspect, distal to right, circular spores attached, explanatory drawing in Text-figure 6a, 2002. 12.79, x 84. 2, right palpal femur, anterior aspect, distal to right, patch of spinules on near (anterior) surface, black spore attached, explanatory drawing in Text-figure 6b, 329.3, x 133. 3, distal end of femur, patella, posterior aspect, explanatory drawing in Text-figure 6 c, 329. 59, x 73. 4, left palpal femur and patella, patch of spinules on far (anterior) surface, dark spores attached, explanatory drawing in Text-figure 6d, 329 . 1 , x 74. 5, distal end of femur and patella, explanatory drawing in Text-figure 6g, 334.16. 12, x 115. 6, femur and patella, foreign cuticle fragment overlying proximal part of femur, explanatory drawing in Text-figure 6f, 334. 16.86, x 71. 7, part of complex grouping of podomeres showing distal femur and patella, posterior aspect, distal to left, details including sclerite at proximal joint of patella, distal patella with attached fragment of tibia, explanatory drawing in Text-figure 6e, 329. 31 a. Ml, x 68. 8, proximal end of femur showing large setal sockets, 41 1 . 19.243, x 60. ^ PLATE 3 SELDEN et al., Attercopus 266 PALAEONTOLOGY, VOLUME 34 text-fig. 7. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987), explanatory drawings for specimens illustrated on Plate 4. a. 329.3, distal end of tibia, detritus shown at X, setal sockets not differentiated according to surface and setae omitted. B, 41 1 . 7.45, distal end of tibia, inferior aspect, spore shown in black, c, 41 1 .20.25, patella, inferior aspect, spore shown in black, detritus by X. d, 411 . 19.248, distal aspect of patella, focused to show details of distal joint, superior to left, e, 334. la. 8, tibia, setae omitted, f, 4 1 1 . 19.98, distal end of tibia and proximal piece of metatarsus, superior aspect, setae and sockets omitted. G, 329.3, metatarsus, proximal end to left, distal to right, superolateral aspect, setal sockets omitted. Scale bar represents 0.5 mm for all specimens; see material and methods for abbreviations and conventions. EXPLANATION OF PLATE 4 Figs I -1 1 . Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1 , distal end of tibia, dark mass of detritus, explanatory drawing in Text-figure 7 a, 329 . 3, x 70. 2, patella, inferior aspect, distal to left, pieces of detritus on right, and spore, explanatory drawing in Text-figure 7 b, 41 1 .20.25, x 132. 3, distal end of tibia, showing sensilla and articulations, spore in top left, explanatory drawing in Text-figure 7c, 41 1 .7.45, x 125. 4, distal end of tibia showing sensilla, 2002. 12.49, x 65. 5, patella, details of distal joint, explanatory drawing in Text-figure 7d. 41 1 . 19.248, x 190. 6, tibia, distal to right, explanatory drawing in Text-figure 7e, 334. la. 8, x 66. 7, distal end of tibia attached to proximal part of metatarsus, superior aspect, explanatory drawing in Text-figure 7 g, 411.19. 98, x 124. 8, metatarsus, distal to left, attached spore at top, 329.38, x 58. 9, metatarsus, distal to left, 329.53, x 53. 10, metarsus, broken into two parts, distal to right, superolateral aspect, explanatory drawing in Text-figure 7f, 329.3, x46. 11. metatarsus, superolateral aspect, distal to left, 41 I . 19.251, x 92. PLATE 4 SELDEN et ah. At ter copus 268 PALAEONTOLOGY, VOLUME 34 text-fig. 8. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987) (a-d) and Arachnida incertae sedis (b), explanatory drawings for Plate 5. a-c, 334. la. 4, two walking legs, patella to tarsus; a, detail of distal end of tarsus, setae, sockets, and slit sensilla omitted ; b, complete specimen with adjacent Arachnida incertae sedis flagelliform appendage, setae and sockets omitted for clarity; c, detail of metatarsus-tarsus joint, distal to top, setae and sockets omitted, arthrodial membrane shown in coarse stipple. D, 334. 16.38, tarsus, showing tarsal organ and slit sensilla, setae omitted, spore at proximal end. Scale bar represents 0.25 mm for a and c, 1.5 mm for b, and 0.5 mm for d; see material and methods for abbreviations and conventions. Tibia. This podomere is about three times as long as wide (PI. 4, fig. 6). When flattened in the fossils, it appears rectangular, lacking the distal emargination and the proximal promontories of the femur. It can be distinguished from the metatarsus by the superodistal lyriform organ of the latter, which has the slit sensilla arranged transversely. The proximal joint of the tibia bears a strong superior articulation. The distal joint is a superior bicondylar hinge. Adjacent to one side of the distal articulations is a row of slit sensilla, and there are lyriforms situated close to the inferior on this side, and on the opposite side of the joint in an anterior/posterior position. Features of the distal joint are shown on Plate 4, figs 1, 3, 4, 9. It is not possible to orient the tibia since the only specimens which are in direct connection with the patella and also preserve the distal joint are obscured by other podomeres. Metatarsus. The metatarsus is the longest podomere on the leg, the longest being nearly four times as long as wide, in the flattened state. The proximal joint is a superior bicondylar hinge (see tibia, above). The distal joint explanation of plate 5 Figs 1-3, 5. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1, detail of distal end of tarsus shown in upper part of fig. 3, showing setation, serrate macroseta inferiorly, tarsal organ superiorly, and arrangement of paired and median fimbriate claws, explanatory drawing in Text-figure 8a, 334. la. 4, x 165. 2, tarsus, distal to top, 329.16.34, x92. 3, complex grouping of two walking legs of Attercopus with adjacent flagelliform appendate of Arachnida incertae sedis, explanatory drawing in Text-figure 8b, 334. la. 4, x 94. 5, tarsus, distal to left, showing tarsal organ, claws, spore at proximal end, explanatory drawing in Text- figure 8 d, 334. 16.38, x 76. Figs 3 and 4. Arachnida incertae sedis. 3 flagelliform appendage with 12 segments (including distal?), showing setae and slit sensilla, adjacent to legs of Attercopus , explanatory drawing in Text-figure 8b, 334. la. 4, x 94. 4, 8-segmented flagelliform appendage (including distal?), showing setae and slit sensilla, 2002.9.20, x 80. PLATE 5 SELDEN et a/., Attereopus , Arachnids incertae sedis 270 PALAEONTOLOGY, VOLUME 34 text-fig. 9. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987) (a, d) and Ecchosis pulchribothrium gen. el. sp. nov. (b. c), explanatory drawings for Plate 6. a, 41 1 .02. I2M.6, metarsus and overlying tarsus, setal sockets not differentiated according to surface and setae ommited. b, 411.7.37, patella and proximal end of tibia, inferior aspect, distal to top. c, 41 1 . 19 . 96, patella, superior aspect, distal to left, d, 329 . 69, palpal tarsus, setal sockets not differentiated according to surface and setae omitted. Scale bar represents 0.5 mm for all specimens; see material and methods for abbreviations and conventions. is readily recognized by the large lyriform organ situated in a superior position, which characteristically has the slits arranged at right angles to the long axis of the leg. The lyriform is situated at the base of a cuticular projection which bears articulations at either side (PI. 4, figs 8-1 1 ; PI. 5, fig. 3 ; PI. 6, fig. 1 ). Though resembling a bicondylar hinge, the arrangement here is actually a rocking joint. As in spiders, the two ‘condyles’ are projections which articulate with the tarsus only loosely, the joint being held by muscles, and the joint allows rocking in an antero-posterior direction as well as flexure, as necessary (see Manton 1977 ; Clarke 1984, 1986). The metatarsus is well clothed with setae, some of which are long and thin, and macrosetae are present EXPLANATION OF PLATE 6 Figs 1, 2, 4, 5. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). 1, metatarsus and overlying tarsus, distal to left, explanatory drawing in Text-figure 9 a, 411 . 02.12M.6, x 60. 2, detail of distal end of tarsus shown in fig. 1, showing claw fimbriae and tarsal organ, explanatory drawing in Text-figure 9a, 411.02. 12M .6, x 154. 4, complex grouping of walking leg podomeres, including tibiae, metatarsi, and tarsi, showing setae, claws, and tarsal organs, 329. 3 In. M2, x 98. 5, palpal tarsus, showing attachment to metatarsus fragment, setae, and single fimbriate claw, explanatory drawing in Text-figure 9d, 329.69, x 80. Figs 3 and 6. Ecchosis pulchribothrium gen. et sp. nov. 3, patella, superior aspect, distal to left, explanatory drawing in Text-figure 9c, 41 1.19.96, x 65. 6, patella and proximal end of tibia, inferior aspect, distal to top, explanatory drawing in Text-figure 9b, 411 .7.37, x90. PLATE 6 SELDEN et a /., Attercopus , Ecchosis 272 PALAEONTOLOGY, VOLUM E 34 text-fig. 10. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987), explanatory drawing for Text-figure 1 1, posterior median spinneret, distal to right, specimen is folded into three layers at thickest, small slit sensilla shown on near surfaces, setal sockets on far surfaces shown in dotted lines for clarity. Scale bar represents 0.5 mm; see material and methods for abbreviations and conventions. interiorly and inferodistally. The macrosetae are very prominent, see, for example, Plate 5, figure 3, and Shear et al. (1987, figs 132 and 133). A few slits are present inferoanteriorly and inferoposteriorly around the joint margin, adjacent to the macrosetae. No trichobothria have been seen on this podomere. Tarsus. The tarsus (PI. 5, figs 1-3, 5; PI. 6, figs 1, 2, 4, 5) is about five-sixths the length of the metatarsus, and is similarly profusely clothed with two sizes of setae, and macrosetae occur inferiorly. Except proximally in larger specimens, the reticulate pattern characteristic of this genus is absent. The proximal joint bears two articulations which correspond to the articulations on the metatarsus. The leg tarsi are parallel-sided, and the distal joint bears three emarginations, in inferior, anterior, and posterior positions. Pairs of slit sensilla occur adjacent to the anterior and posterior embayments. There are three claws on the leg tarsi : the lateral claws are long and curved, the median claw is also quite long, and is thicker in mid-section and shorter than the lateral claws. All claws bear rows of fimbriae along their inferior edges (PI. 5, fig. 1 ; Shear et al. 1987, figs 137-139). Trichobothria cannot be seen on any of the fossil specimens. A tarsal organ is present in a superodistal position (PI. 5, fig. 1 ; PI. 6, figs 2 and 4). The palpal tarsus (PI. 6, fig. 5; Shear et al. 1987, fig. 135) is not parallel-sided, but tapers distally and is also distinguished by the presence of only a single fimbriate claw. Spinneret. The single spinneret (specimen 334. lb. 34; Text-figs 10 and 11a-c), believed for reasons already discussed (Shear, Palmer et al. 1989) to be a posterior median spinneret, is about 0-94 mm long and represents a nearly complete single article of typical semifusiform shape. The specimen appears to have been torn along the median surface, and subsequently the torn edge (now nearest the observer as the spinneret is mounted on a microscope slide) was folded under itself. The cuticle is typical of A. fimbriunguis, ornamented with distinct polygonal cells, setal sockets and slit sense organs, the latter two structures densely but evenly scattered over the entire surface. Some of the sockets bear large, smooth setae, and a single serrate seta is present (Text-fig. 10). Spigots are scattered along the median surface only, and more densely clustered distally. Though folding and consequent superposition of structures makes an exact count difficult, at least 24 distinct individual spigots can be seen. There appears to be no significant variation in spigot size and form. Each spigot consists of a conical SELDEN ET AL. DEVONIAN ARACHNIDS 273 text-fig. 11. Attercopus fimbriunguis (Shear, Selden and Rolfe, 1987). a-c, 334 . 1 ^ . 34. a, posterior median spinneret, distal to right, explanatory drawing in Text-figure 10, x 70. b, detail of base of lowermost, distally- directed, terminal spigot shown in a and c, with ?setal socket, distal to top, x 1200. c, distal end of spinneret showing detail of cuticle and spigots, x 130. base approximately twice as long as wide, which narrows abruptly to a slender shaft no more than three times as long as the base. What appears to be a large setal socket is found on some of the spigot bases; the consistent position of this structure and careful focussing confirms that it is on the spigot base and is not a feature of overlying or underlying spinneret cuticle. The articulation of the base with the shaft is smooth, lacking a collar. No sculpturing is detectable on the distal part of the shaft, but extraordinarily fine sculpture, as is found on the shafts of some rastelloid mygalomorph spiders (J. Palmer, pers. comm.), may have been obliterated during diagenesis. Subclass pulmonata ( sensu Firstman 1973) incertae sedis Genus ecchosis gen. nov. Derivation of name. Greek, ec -, out of, from, and chosis , a heaping-up of earth; referring to the earth-dam for the pump-storage power station which now covers the Gilboa locality. Type and only known species. Ecchosis pulchribothrium sp. nov. Diagnosis. Pulmonate with patellar trichobothrium, the basal collar of which is ornamented with reticulate pattern of oval and lunate reticulate thickenings; thick, striated spines with bifid tips on some other podomeres. 274 PALAEONTOLOGY, VOLUME 34 text-fig. 12. Alter copus fimbriunguis (Shear, Selden and Rolfe, 1987), explanatory drawings for specimens illustrated in Shear et al. (1987). a, 329. 70, left walking leg femur and patella, anterior aspect, see Shear et al. (1987, tig. 129). b, 329.39, patella, inferior aspect, distal to right, see Shear et al. (1987, fig. 128). c, 329.59, trochanter, inferior aspect, distal to top, superior surface absent, specimen previously described and illustrated in Shear et al. (1987, fig. 140) as 'undetermined median structure’, d, 329.70, distal end of metatarsus and proximal tarsus, distal to right, setal sockets not differentiated according to surface and setae omitted from tarsus, see Shear et al. (1987, fig. 132). e, 329.70, distal end of metatarsus showing slit sensilla and serrate macrosetae, distal to left, setal sockets not differentiated according to surface, spores in black, see Shear et al. (1987, fig. 133). f, 329.57, distal part of metatarsus, distal to left, setae (except macrosetae) omitted, see Shear et al. (1987, fig. 131). Scale bar represents 0-5 mm for all specimens; see material and methods for abbreviations and conventions. SELDEN ET A L. : DEVONIAN ARACHNIDS 275 Ecchosis pulchribothrium sp. nov. Plate 2, fig. 3; Plate 6, figs 3 and 6; Plate 7, figs 1, 3, 4, 7, 8; Text-figs 5 c and 9 b, c. 1987 Arachnida incertae sedis A, Shear, Selden and Rolfe; Shear et al. , pp. 70, 71, figs 146-150. Derivation of name. Latin, pulcher , beauty, and bothrium , a cup. Type specimens. Holotype : patella and base of tibia, on slide 411.7.37. Paratypes : patella? on slide 411.7.86; distal end of femur, on slide 411.1.33; two parts of unknown podomere with large sockets and striated spines, on slide 411 . 19.188. Additional material. A complete list of the specimens referred to this species is deposited in the British Library, Boston Spa, Yorkshire, England, as Supplementary Publication No. SUP 14040, 5 pp.; see Repository above for availability of this publication. Diagnosis. As for the genus. Description Cuticle. Large sheets of cuticle (PI. 7, figs 1 and 7) of this animal occur in the Gilboa material, and are characterized by an ornament of small scales, resembling a reticulate ornament, thickened at one side, in which the connections between the thickenings have been lost. The scales are arranged in straight or arcuate parallel rows; the arcuate patterned cuticle is presumed to represent podomeres which have become opened out. The cuticle bears setal sockets whose diameters range in size from small (0015 mm) to extremely large (045 mm), and additionally there are small (0-005 mm), circular pores scattered across the cuticle. The largest sockets only occur on one type of podomere. On what is presumed to be body cuticle, the setal sockets range up to 0 075 mm in diameter, and these larger ones commonly have a raised rim or broad spine on one side of the socket. This pattern is particularly emphasized on what are presumed to be edges of tergites, where a large thorn has a small spine articulated at its base; such an arrangement appears to be common on the cuticle of amblypygids. Large slit sensilla are also present on these pieces. The macrosetae are conspicuously striated, and the very large, thick spines are not only striated but also have bifid tips, a feature lacking on smaller setae (PI. 7, figs 4 and 8). The cuticle of Ecchosis resembles that of Liphistius in the following features: scale-like sculpture, minute pores on cuticle surface, raised rim to larger setal sockets, and striations on macrosetae. Coxa. The inferior surface and distal joint of the coxa is preserved on slide 2002.9. 13 (PI. 2, fig. 3). The costa coxalis can be seen to run as a thickened ridge towards the anterior dorsal edge of the podomere (which is not preserved). Close to the preserved proximal termination of the costa coxalis, and running at an angle from it towards the distal edge, is a folded piece of cuticle which is believed to represent the stiffened cuticle by which the coxa articulates dorsodistally with the body wall, in comparison with the coxa of Liphistius. The posterior and superior margins of the distal joint are folded across the anterior surface and the costa coxalis. The ventral surface is covered with setal sockets and richly supplied with pores; the inferoanterior surface bears fine setae. There is a fragment of the dorsal edge preserved at the proximal end of the podomere. No other specimen of this podomere is known. Femur. A large femur is present on slide 41 1 . 1 . 33 (PI. 7. fig. 1 ). Only the distal half is preserved, including parts of the distal joint: one of the articulations, a small group of slit sensilla adjacent and just superior to the articulation, and the emarginated inferior border. A number of small setal sockets are present, and two longitudinal rows of three or four larger sockets run along the inferior side of the podomere. Patella. One definite patella is present, on slide 411.7.37 (PI. 6, fig. 6), attached to the proximal end of a tibia. The patella is easily recognized by its emarginated inferior proximal edge, which bears inferoanterior and inferoposterior crescentic articulation points, for attachment of the arcuate sclerite (not preserved). The superior edge of the proximal joint is not preserved. The superior surface of the patella is twice the length of the inferior surface; it bears four or five setal sockets, some with setae, and a short distance proximal to the superior articulation of the distal edge lies an ornamented trichobothrial base. Three small slit sensilla occur between this bothrium and the articulation point, which is present at the extremity of the distally produced superior side of the distal joint. The inferior side of the distal joint is fairly straight, and is characterized by 276 PALAEONTOLOGY, VOLUME 34 two groups of large slit sensilla, the slits at an angle distally diverging from the midline, and an interiorly positioned single large slit which runs at an angle of about 80° from the others (this slit may be part of another group, but dark material obscures the podomere at this point). The presence of an inferior articulation at the distal joint is suspected, but not clearly seen because of the obscuring dark matter, because there is an articulation on the corresponding inferior side of the piece of tibia which is inserted into the patella (PI. 6, fig. 6). Two other podomeres bear an ornamented trichobothrial base. The best preserved specimen is 411.7.86 (Shear et al. 1987, figs 149 and 150). The bothrium consists of a ring of thickened cuticle surrounding a hole. Outside this ring is a collar of patterned cuticle which is more than three times the diameter of the hole. The pattern consists of a reticu'um of thickened cuticle defining elliptical and lunate shapes. In the other specimens (41 1 . 7 . 37, 41 1 . 19.96) the morphology appears to be identical, as far as can be made out in these less well preserved examples. In no case is a hair seen emerging from the hole. In the original description, the podomere bearing the well preserved example (41 1 .7.86) was described as a possible femur because its distal end appears to have an inferior emargination (Shear et al. 1987, fig. 149). Now that the femur of Ecchosis pulchribothrium is known, it is certain that the earlier described podomere is not a femur. Specimen 411 7.86 could, however, be another patella. The bothrium occurs adjacent to the superior distal articulation; slit sensilla may be present on the emarginated inferior side of the distal joint, but could not be seen because of the folding (Shear et al. 1987, fig. 149). Both the inferior side and the proximal joint are not well preserved in 41 1 .7.86; it is uncertain whether this specimen represents a different podomere with the same kind of trichobothrial base, or another patella. The third specimen bearing an ornamented trichobothrial base (411.19.96, PI. 6, fig. 3) resembles a trochanter at first sight; closer inspection, however, reveals that it has been proximodistally compressed to some degree, and the proximal joint is incompletely preserved. It, too, could be a patella. The preserved inferior surface is short, and bears three groups of slit sensilla. Two of these are situated close to the inferior articulation (which is not strongly developed), and they diverge distally at an angle from the midline. The other group diverges at an 80° angle from the first two groups, and is situated on the other side of the midline from them. The trichobothrial organ is obscured by folding; it is situated, like those on the other two podomeres, a short distance behind and a little to one side of the superior distal articulation. Tibia. Only the fragment attached to the patella in 41 1 .7.37, described above, is known with certainty. This piece has superior and inferior proximal articulations. It is interesting in that its lateral sides appear to diverge distally; possibly it was a tumid podomere in this leg in life. In addition, a number of examples of a long podomere with extremely large setal sockets occurs among the specimens; 411.19. 188 (PI. 7, fig. 8) and 329.46 are good examples. The proximal end of the podomere does not occur on any of these specimens, but these podomeres are at least three times as long as wide, and have two rows of large sockets, each row with at least 8 sockets, along their length. In addition to the rows of major sockets, there are about 10 rows of smaller setal sockets running along the length of the podomere. There is commonly a smaller seta adjacent to each major socket. The large sockets bear thick, spindle-shaped movable spines, each about four times as long as maximally wide in the compressed state. The spines have straight striations running along their length, are EXPLANATION OF PLATE 7 Figs 1. 3, 4. 7. 8. Ecchosis pulchribothrium gen. et sp. nov. 1, inferodistal part of femur, inferior to top, distal to right, showing cuticle sculpture, 411.1.33, x 53. 3, part of distal joint of unknown podomere showing slit sensilla grouped into lyriform organ, 411.19.184, x 72. 4, thick, striated, bifid spine on unknown podomere, 411.19.137, x 53. 7, patch of cuticle (part of body not known) showing cuticle sculpture, 41 1 . 19.206, x 89. 8, superodistal part of unknown podomere showing cuticle sculpture, setae, spine and their sockets, and lyriform organ, 411.19.188, x 1 18. Figs 2 and 6. Extant amblypygid Heterophrynus elaphus , specimens cleared in potassium hydroxide. 2, trichobothrial base adjacent to superior articulation at distal joint of tibia 4, x 135. 6, left chelicera, ectal aspect, dense setation around teeth removed for clarity, transmitted light under ethanol on Olympus SZH stereomicroscope, x 7-5. Fig. 5. Extant uropygid Mastigoproctus giganteus, left chelicera, mesal aspect, specimen cleared in potassium hydroxide, dense setation around teeth removed for clarity, transmitted light under ethanol on Olympus SZH stereomicroscope, x 7-5. PLATE 7 SELDEN et a/., Ecchosis , Mastigoproctus , Heterophrynus 278 PALAEONTOLOGY. VOLUME 34 broad at the base, and have a bifid tip (PI. 7, fig. 4). The normally shaped macrosetae present on the podomeres are also striated, and do not have bifid tips. The smallest setae are relatively short. The distal end of one podomere is preserved (PI. 7, fig. 8), and shows a longitudinal lyriform organ. There is no conclusive evidence of the identity of these large podomeres. The short trochanter and patella, and the terminal tarsus can all be ruled out. Of the long podomeres, all pulmonate metatarsi have a lyriform or group of slit sensilla at the distal end, in which the slits are aligned transversely. Pulmonate femora bear rows of slit sensilla rather than lyriforms, characteristic articulation points, and are normally distinctly emarginated. It is therefore most likely that the long podomeres represent tibiae. Well developed lyriform organs occur on the distal ends of the tibiae of spiders, but not of amblypygids or uropygids (Barth 1978, fig. 3). Discussion. Is the ornamented sense organ, which is one of the characteristics of Ecchosis , a true trichobothrium? Among living arachnids, the trichobothrium is fairly widespread, occurring in all groups except Ricinulei, Solifugae, and Opiliones. Ornamented trichobothrial bases are known from living spiders, although none has the same type of pattern see in Ecchosis. It is also rare to find a trichobothrium on the patella of an arachnid; they occur more commonly on the more distal podomeres of the legs. A literature search for spiders with patellar trichobothria revealed none, and R. Forster (pers. comm.) is aware of no spider with patellar trichobothria. However, a study of specimens of other Pulmonata revealed that whereas uropygid patellae bear no trichobothria, they are present on the patellae of legs 2, 3, and 4 of Amblypygi. Weygoldt (1972) described two trichobothria on each walking-leg (2, 3, 4) patella of all species of Charinus , and we observed this same pattern on Heterophrynus elaphus. Quintero (1980) described these organs on the patella of Acanthophrynus coronatus , and called them ‘campaniform sensilla’, but they do not seem to differ in morphology from the tibial trichobothria. They bear a fine hair emerging from the central hole, as drawn by Quintero (1980, fig. 6) and so are not campaniform sensilla. Of especial interest is the ornamentation of the collar (PI. 7, fig. 2); it is remarkably similar to that observed in E. pulchribothrium , and quite different from that on the trichobothria found on uropygids and spiders. The patella of amblypygids is different in shape from that of E. pulchribothrium , being specialized for immobility and twisted to enable the crevice locomotion of these bizarre animals (Manton 1977), so that whilst their patellar trichobothria lie adjacent to the superior distal articulation, this articulation is situated in a triangular notch in the distal edge of the podomere. It is therefore possible that Ecchosis is an amblypygid, but without additional evidence, the genus cannot be assigned to that group. It is probable that in the Devonian there were Pulmonata with a mosaic of characters which today are found iu separate taxa. Class arachnida Lamarck, 1801 incertae sedis Plate 5, figs 3 and 4 Five specimens (329 . 60, 329 . 62, 334. In. 4, 41 1 .2.4, 2002 .9.20: PI. 5, figs 3 and 4) of lengths of short segments are present in the Gilboa material. The segments are about one and a half times as long as wide and all are virtually identical, apart from the terminal one in some specimens. No more than 12 occur together. Each has a distal collar into which the next succeeding segment is inserted, and this collar bears setal sockets all round. The cuticle is patterned with transversely elongate reticulate sculpture, and scattered across the surface are some small pores which resemble the little slit sensilla of Attercopus (they differ slightly, however, in that these always appear eliptical or lunate even at low magnification). The setae are very long and thin, and do not have bifid tips (there arc many specimens of another type of flagellar appendage in which the segments are about three times as wide as long, in the compressed state, in which the setae have bifid tips with branches of different lengths). There is no evidence to link these flagellar appendages with any arachnids, except that the little pores, if they are slit sensilla, would confirm an arachnid rather than any other arthropod group. These organs might be the caudal flagellum of a uropygid (and evidence is amounting that Gelasinotarbus bonamoae may prove to be one of these animals) or could be the flagelliform first leg of an amblypygid. Similar antenniform appendages with slit sensilla have also been found in Stephanian deposits from Kansas (A. J. Jeram, pers. comm.). SELDEN ET A L. : DEVONIAN ARACHNIDS 279 Acknowledgements. We thank Ray Forster and Norman Platnick for sharing their observations on the morphology of a wide range of pulmonate arachnids with us, Andy Jeram for information on the many new fossil Pulmonata he is turning up, and Jonathan Coddington and Jacqueline Palmer for discussion on the identity of the spinneret. We thank Sam Morris and Norman Platnick for the loan of material in the care of The British Museum (Natural History) and The American Museum of Natural History respectively, and W. Struve (Senckenberg Museum) for the preparation and loan of a plaster cast of Archaeometa? devonica. P.A.S. is extremely grateful to the faculty and staff of Hampden-Sydney College for their hospitality during an extended study visit in 1989. This work was supported by a grant from the US National Science Foundation (BSR 88-180-27) to W.A.S. and P.M.B., and by travel funds for P.A.S. from The University of Manchester and The Royal Society of London. REFERENCES banks, H. p., bonamo, P. M and Grierson, j. D. 1972. Leclercqia complexa gen. et sp. nov., a new lycopod from the late Middle Devonian of eastern New York. Review of Paleobotany and Palynolog r, 14. 19^10. - 1985. The flora of the Catskill clastic wedge. Geological Society of America Special Papers , 201, 1-22. barth, F. G. 1978. Slit sense organs: ‘Strain gauges’ in the arachnid exoskeleton. Symposia of the Zoological Society of London, 42, 439-448. — 1985. Slit sensilla and the measurement of cuticular strains. 162-188. In barth, f. g. (ed.). Neurobiology of arachnids. Springer-Verlag, Berlin, Heidelberg, New York and Tokyo, x + 385 pp. — and stagl, j. 1976. The slit sense organs of arachnids. A comparative study of their topography on the walking legs (Chehcerata, Arachnida). Zoomorphologie , 86, I -23. blumenthal, h. 1935. Untersuchungen fiber das ‘Tarsal-organ’ der Spinnen. Zeitschrift fur Morphologic und Okologie der Tiere , 29, 667-719. brauckmann, c. 1987. Neue Arachmden-Funde (Scorpionida, Trigonotarbida) aus dem westdeutschen Unter- Devon. Geologica et Palaeontologica , 21, 73-85. Clarke, j. 1984. On the relationship between structure and function in the leg joints of Heteropoda venatoria (L.) (Araneae: Eusparassidae). Bulletin of the British Arachnological Society , 6, 181-192. — 1986. The comparative functional morphology of the leg joints and muscles of five spiders. Bulletin of the British Arachnological Society, 7, 37—47. coddington, j. a. 1989. Spinneret silk spigot morphology: evidence for the monophyly of orbweaving spiders, Cyrtophorinae (Araneidae), and the group Theridiidae plus Nesticidae. Journal of Arachnology, 17, 71-95. firstman, b. 1973. The relationship of the chelicerate arterial system to the evolution of the endosternite. Journal of Arachnology 1, 1-54. foelix, r. f. 1985. Mechano- and chemoreceptive sensilla. 118-137. In barth, f. g. (ed.). Neurobiology of arachnids. Springer-Verlag, Berlin, Heidelberg, New York and Tokyo, x + 385 pp. — and axtell, R. c. 1972. Ultrastructure of Haller’s organ in the tick Amblyomma americum L. Zeitschrift fur Zellforschung und Mikroskopische Anatomie, 124, 275-292. — chu-wang, i. w. and beck, l. 1975. Fine structure of tarsal sensory organs in the whip spider Admetus pumilio (Amblypygi, Arachnida). Tissue and Cell, 7, 331 -346. — and sharbronath, j. 1983. The fine structure of scorpion sensory organs. I. Tarsal sensilla. Bulletin of the British Arachnological Society, 6, 53-67. forster, r. r. 1980. Evolution of the tarsal organ, the respiratory system and the female genitalia in spiders. 269-284. In gruber, j. (ed.). Verhandlungen der 8 Internationaler Arachnologen-Kongress, Wien. H. Engermann, Vienna, 505 pp. — and platnick, n. i. 1984. A review of the archaeid spiders and their relatives, with notes on the limits of the superfamily Palpimanoidea (Arachnida, Araneae). Bulletin of the American Museum of Natural History, 181, 1-230. hammen, l. van der. 1977. A new classification of Chelicerata. Zoologische Mededelingen, Leiden, 51, 307-319. 1985. Functional morphology and affinities of extant Chelicerata in evolutionary perspective. Transactions of the Royal Society of Edinburgh ( Earth Sciences ), 76, 137-146. — 1986. Comparative studies in Chelicerata IV. Apatellata, Arachnida, Scorpionida, Xiphosura. Zoologische Verhandelingen, Leiden, 226, I -52. 280 PALAEONTOLOGY, VOLUME 34 hirst, s. 1923. On some arachnid remains from the Old Red Sandstone (Rhynie Chert Bed, Aberdeenshire). Annals and Magazine of Natural History (9), 12, 455^474. kaestner, a. 1968. Invertebrate zoology II. (Translated and adapted from the German by h. w. and L. R. levi.). Wiley, New York, London and Sydney, ix + 472 pp. kovoor, J. 1986. Comparative structure and histochemistry of silk-producing organs in arachnids. 160-186. In nentwig, w. (ed.). Ecophysiology of arachnids. Springer-Verlag, Berlin, Heidelberg, New York, London, Paris and Tokyo, ix + 448 pp. maddison, w. p. and maddison, d. r. 1987. MacClade . version 2.1. (A phylogenetics program for the Apple Macintosh™ computer, distributed by the authors at: Museum of Comparative Zoology, Harvard University, Cambridge, MA 02138, USA.) manton, s. M. 1977. The Arthropoda. Habits, functional morphology, and evolution. Clarendon Press, Oxford, xxii + 527 pp. moro, s. d. and bali, g. 1986. The topography of slit sense organs in the whip scorpion, Thelyphonus indicus (Arachnida, Uropygida). Verhandlung der Naturwissenschaftlichen Verein im Hamburg ( NF ), 28, 91-105. petrunkevitch, A. 1949. A study of Paleozoic Arachnida. Transactions of the Connecticut Academy of Arts and Sciences. 37. 69-315. platnick, N. i. and goloboff, p. a. 1985. On the monophyly of the spider suborder Mesothelae (Arachnida: Araneae). Journal of the New York Entomological Society. 93, 1265-1270. pocock, r. i. 191 I. A monograph of the terrestrial Carboniferous Arachnida of Great Britain. Monograph of the Palaeontographical Society. 64 ( ), 1-84, 3 pis. Quintero, D. 1980. Systematics and evolution of Acanthophrynus Kraepelin (Amblypygi, Phrynidae). 341-347. In gruber, J. (ed.). Verhandlungen der 8 Internationale r Arachnologen-Kongress, Wien. H. Engermann, Vienna, 505 pp. REISSLAND, a. and GORNER, p. 1985. Trichobothria. 138-161. In barth, f. g. (ed.). Neurobiology of arachnids. Springer-Verlag, Berlin, Heidelberg, New York and Tokyo, x + 385 pp. shultz, j. w. 1989. Morphology of locomotor appendages in Arachnida : evolutionary trends and phylogenetic implications. Zoological J owned of the Linnean Society 97, 1-56. selden, p. a. 1984. Autecology of Silurian eurypterids. 39-54. In bassett, m. g. and lawson, j. d. (eds). Autecology of Silurian organisms. Special Papers in Palaeontology. 32. 1 -295. 1985. Eurypterid respiration. Philosophical Transactions of the Roval Society of London. Series B. 309. 219-226. - and jeram, a. J. 1989. Palaeophysiology of terrestrialisation in the Chelicerata. Transactions of the Royal Society of Edinburgh (Earth Sciences). 80, 303-310. shear, w. a. 1986. A fossil fauna of early terrestrial arthropods from the Givetian (upper Middle Devonian) of Gilboa, New York, USA. Actas X Congreso Internacional de Aracnologia. Jaca. Espaha. 1. 387-392. - and bonamo, p. m. 1988. Devonobiomorpha, a new order of centipeds (Chilopoda) from the Middle Devonian of Gilboa, New York State, USA. and the phytogeny of centipcd orders. American Museum Novitates. 2927, 1-30. — Grierson, j. d., rolfe, w. d. i., smith, e. l. and Norton, r. a. 1984. Early land animals in North America : evidence from Devonian age arthropods. Science. 224. 492^194. palmer, j. m., coddington, j. a. and bonamo, p. m. 1989. A Devonian spinneret : early evidence of spiders and silk use. Science. 246, 479-481. - schawaller, w. and bonamo. p. m. 1989. Record of Palaeozoic pseudoscorpions. Nature. 341. 527-529. selden, p. a., rolfe, w. d. L, bonamo, p. m. and grierson, j. d. 1987. New terrestrial arachnids from the Devonian of New York (Arachnida, Trigonotarbida). American Museum Novitates. 2901, 1-74. stormer, l. 1976. Arthropods from the Lower Devonian (Lower Emsian) of Alken-an-der-Mosel, Germany. Part 5: Myriapoda and additional forms, with general remarks on fauna and problems regarding invasion of land by arthropods. Senckenbergiana Lethaea. 57, 87-183. vachon, m. 1966. Les conduits evacuateurs des glandes cheliceriennes chez les pseudoscorpions (Arachn.). Senckenbergiana Biologica. 47, 29-33. weygoldt, p. 1972. Charontidae (Amblypygi) aus Brasilien. Beschreibung von zwei neuen Charinus- Arten, mit Anmerkungen zur Entwicklung, Morphologie und Tiergeographie und mit einem Bestimmungschlussel fur die Gattung Charinus. Zoologisches Jahrbucher. Abteilung fur Systematik. Geographie und Biologie der Tiere, 99, 107-132. SELDEN ET AL.: DEVONIAN ARACHNIDS 281 PAUL A. SELDEN Department of Extra-Mural Studies University of Manchester Manchester M13 9PL, UK WILLIAM A. SHEAR Department of Biology Hampden-Sydncy College Hampden-Sydney, Virginia 23943, USA Typescript received 2 February 1990 Revised typescript received 31 March 1990 PATRICIA M. BONAMO Center for Evolution and the Paleoenvironment State University of New York Binghamton, New York 13901, USA ORDOVICIAN G RAPTOLITES FROM THE EARLY HUNNEBERG OF SOUTHERN SCANDINAVIA by KRISTINA LINDHOLM Abstract. A graptolite fauna of Early Hunneberg age is described from southern Scandinavia (Scania, Vastergotland, Oslo region). Correlation and boundaries within the interval are discussed and it is suggested that the Hunneberg Stage be elevated to series rank, interposed between the Tremadoc and the Arenig. One new dichograptid genus, Hunnegraptus , and three Scandinavian representatives of it, H. copiosus , H. tjernviki , and H. robustus , are erected. The genus is multiramous, with long first-order stipes, and shows presumed rejuvenation of gerontic specimens. It is likely to be most closely related to Clonograptus. Six additional species are formally named: Kiaerograptus supremus (Anisograptidae), Clonograptus (C.) magnus, Tetragraptus longus, T. krapperupensis (Dichograptidae), Paradelograptus elongatus, and P. tenuis (Sinograptidae). The sequence containing these taxa is divided into three zones: the K. supremus Zone which probably starts in the Late Tremadoc, the A. murrayi Zone, and above that the H. copiosus Zone, which underlies the Late Hunneberg Tetragraptus phyllograptoid.es Zone. The fauna covers part of the interval when anisograptids gave way to graptolites of the dichograptid development stage, and the observed steps in this evolution (loss of bithecae) are described. Graptolites from the Upper Tremadoc and Lower Arenig of southern Scandinavia have been known for over a hundred years, e.g. Tullberg (1880), Holm (1881), Brogger (1882), Herrmann (1883, 1885), Tornquist (1901, 1904), Strandmark (1902), Monsen (1925, 1937), Spjeldnaes (1963), and Erdtmann (1965a). Yet, the fauna described in this paper, which comes from a ‘post-Tremadoc, pre-Arenig' level, remained unrecognized until Tjernvik (1956) made his overview of the Lower Ordovician of Sweden. From a darker band in a grey shale unit at Storeklev, at Mt Hunneberg (Text-fig. 1 A), he mentioned a few peculiar graptolites, which he referred to as ‘undescribed dichograptids’ in his correlation table. No description of this fauna has been given to this day. The fauna, together with several accompanying species, was later found in lithologically similar beds in the Oslo region, mainly by N. Spjeldnaes in the Slemmestad area and by B.-D. Erdtmann in central Oslo. More recently, I re-collected both the Storeklev and the Slemmestad localities. Finally, I identified the fauna, and also older post-Tremadoc graptolites, in a grey to nearly black shale sequence in the Krapperup drillcore in NW Scania. The rarity of identifiable graptolites in the basal beds of the core and absence of the otherwise ubiquitous Ceratopyge Limestone make lithostratigraphic and chronostratigraphic correlation of these basal beds difficult. Judging by circumstantial evidence, however, all of the basal beds probably belong to the Hunneberg. From my own observations in the Lower Ordovician of southern Scandinavia (Lindholm 1991), a closely similar sequence of facies and faunas is developed in the Oslo region, Mt Hunneberg in Vastergotland, and SE Scania. All areas can be regarded as lying within a single confacies belt, equivalent to Jaanusson’s (1976, 1982) Scanian and Oslo confacies belts, at least until the end of Arenig time. They undoubtedly represent a single, original depositional basin which included, as a thicker and further offshore facies, the NW Scanian sequence of the Krapperup bore core. All three areas, Oslo, Mt Hunneberg, and Krapperup, have been cut by various forms of late Carboniferous to early Permian intrusives. In other respects, the geological settings of the three areas, as seen today, differ due to their later geological history. The Oslo region is a large, more or less continuous area of well exposed Cambrian to Silurian sediments close to the Scandinavian fold | Palaeontology, Vol. 34, Part 2, 1991, pp. 283-327. | © The Palaeontological Association 284 PALAEONTOLOGY. VOLUME 34 text-fig. 1. Location of the study area, a. Outline map showing location of investigated localities, b. Detail map of the Slemmestad area, showing the structural complexity; redrawn from Bockelie (1982). The main sampling localities of Lower Hunneberg rocks were the coastal sections at Grundvik and Hagastrand, a road- cut just south of Slemmestad crossroads (eastern side of the road), the new standard section (1 ; western side of the road), and a long continuous roadside exposure (western side) marked ‘2’. This is the ‘ Rortunet ’ section (Bodalen or Buss-stop Nybygget of previous collectors), which is most complete stratigraphically at its southern end. It is cut by a couple of minor thrust faults and Permian dykes. South of area ‘2’ is another roadside exposure (western side), partly hidden behind trees. This section is the Kiaerograptus locality of Spjeldnaes (1963) which, however, also contains Hunneberg age beds. In addition, a few samples are labelled with street names, and one sample derives from the islet Gjeitungholmen. LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 285 belt. The beds are gently, to somewhat more complexly, folded, with minor associated thrust faults particularly in the Slemmestad area where structure can be easily observed (Text-tig. I B). Mt Hunneberg, on the other hand, constitutes a very small, isolated but well exposed area of Cambrian and Lower Ordovician sediments on a Precambrian basement, protected by a thick dolerite cap. The beds are flat-lying. Finally, the Krapperup area (NW Scania) is a small (1x7 km) fault-bounded block, surrounded by Mesozoic rocks, and lying within a zone of intense block faulting, the individual blocks of which are sometimes less than 1 km in width. The NW Scania area forms part of a broad tract of discontinuous outcrops, extending from NW to SE Scania, of Lower Palaeozoic rocks. The beds, and graptolite faunas described here, however, are known only from the western part of Scania. STRATIGRAPHY Zonation of the sequence. In Scandinavia, the interval from the base of the Ceratopyge Limestone to the base of the T. phyllograptoid.es Zone (roughly corresponding to the faunas described herein) has not formally been divided into graptolite zones; a ‘ Didymograptus ? stoenneri Zone’ was indicated by Erdtmann (1965b, text-fig. 5; not defined in text) for the time interval spanning the Ceratopyge Limestone, but based on finds from the very top of the unit only (Erdtmann 1965a, p. 108). The same graptolite fauna is also present in W. Scania, in the Fagelsang core (Hede 1951 ; "D. balticus' Zone, followed by a major hiatus). Here, D. ? stoenneri is found well above the Ceratopyge Limestone. The species was not found in the Krapperup core, despite examination of every cm of the lowermost metres of the core, nor was anything else as primitive-looking. I propose a subdivision, based on the Krapperup core sequence, into (Text-fig. 2): a Kiaerograptus supremus Zone (Krapperup core, 155-06 m (base)-148-22 m); an Araneograptus murrayi Zone (148-22 m-132-73 m); and a Hunnegraptus copiosus Zone (132-73 m-1 12-80 m). The base of the K. supremus Zone corresponds to an undefined level within the Ceratopyge Limestone. The bases of the A. murrayi and H. copiosus Zones are defined by the incoming of their zone fossils. The base NW SCANIA HUNNEBERG OSLO WALES D. balticus T. phyilo- graptoides H. copiosus D. balticus T. phyllo- graptoides H. copiosus D. balticus T. phyllo- graptoides H. copiosus Aremig Hiatus A. murrayi K. supremus M. (E.)\ armata X"' ' i i i * i 1 i i A, SBrratus £ Hiatus? ^4.' serratus l ~T T T AT T T Y T ~t ? A. sedgwicki ? Hiatus S. pusilla S. pusilia text-fig. 2. Attempted correlation of part of the Lower Ordovician between Scandinavia and Wales. The lithologies are clastic, except where indicated; * shows stratigraphic position of beds described by Molyneux and Doming ( 1989). 286 PALAEONTOLOGY, VOLUME 34 of the overlying T. phyllograptoides Zone is defined by the incoming of T. phyllograptoides or a considerable increase in horizontal and reclined tetragraptids, whichever comes first, and roughly corresponds to the base of the T. approximatus Zone elsewhere. The approximate distribution of species is shown in Text-figure 3. text-fig. 3. Summary of observed ranges of taxa present in topmost Tremadoc to Lower Hunneberg beds in southern Scandinavia. (1) indicates estimated relative position of the graptolite-rich horizon at Mt Hunneberg and in the Oslo region. The case for a Hunneberg Series. It has long been known (e.g. Skevington 1966) that there is a sizeable hiatus between the Tremadoc and the Arenig in their respective type areas in Wales. The fauna described herein, of Hunneberg age, fits into this hiatus. Also, there is no general agreement yet as to where to put the boundary between the Tremadoc and the Arenig. This uncertainty concerns mainly beds of an age corresponding to La 2-La 3 in the Australasian stratigraphic scheme, and sometimes also beds of Be 1 -Be 2 age (e.g. Rushton 1985). As things stand, the Hunneberg interval can thus be regarded in four different ways: 1. as a series filling the gap between the Tremadoc and the Arenig; 2. as the basal stage of the Arenig; 3. as the topmost stage of the Tremadoc; 4. as part Tremadoc, part Arenig. The trend is nowadays towards a reduction of the number of series, e.g. the suggested amalgamation of the Llanvirn and the Llandeilo. Still, I am in favour of the introduction of a new Hunneberg Series, interposed between the Tremadoc and the Arenig, as previously suggested by Erdtmann (1988). In my opinion, this is the easiest way round a difficult problem. Even from the British point of view, it would be an advantage: what is now Tremadoc and Arenig in their respective type areas would remain so, whereas the beds of ‘ uncertain ' age in the Lake District and LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 287 South Wales described by Rushton (1985), Molyneux and Rushton (1988), and Molyneux and Doming (1989) would belong to the Hunneberg. The beds described in the above papers are all of La 2 age. My examination of the graptolites described by Molyneux and Rushton (1988) and comparison with Scandinavian and Spanish material proves them to be considerably older than the T. approximatus Zone (La 3). I would recommend a Hunneberg Series of the extent originally suggested by Tjernvik ( 1956), not the revised concept of Tjernvik and Johansson (1980) who referred the topmost zone, that of D. balticus / M . (V.) aff. estonica ('Transition beds’), to the overlying Billingen Stage. It seems that the most practical definition of the Hunneberg Series would be in terms of conodonts, as comprising the conodont zones of P. proteus and P. elegans. This would closely fill the Tremadoc/Arenig hiatus in the type areas. The base of the Hunneberg would correspond to the Scandinavian top of the Tremadoc, however, which is equivalent to a level higher than the top of the Tremadoc in its type area in Wales (Skevington 1966; Henningsmoen 1973). From elsewhere in Wales and adjacent areas, Rushton (in Whittington et al. 1984) mentions younger beds which he refers to the Tremadoc. He includes a trilobite fauna correlated with the Shumardia pusilla Zone of Scandinavia which (Regnell 1960) lies within the Ceratopyge Shale, and a younger Angelina sedgwicki Zone fauna which cannot be correlated with Scandinavia. Therefore, beds equivalent to the Scandinavian topmost Tremadoc Apatokephalus serratus Zone (Ceratopyge Limestone) are not definitely known in Wales. An approximate correlation between the Scandinavian and Welsh faunas is given in Text- figure 2. The top of the graptolite sequence here described is considerably older than the oldest graptolite fauna in the type area of the Arenig. That fauna was described by Zalasiewicz (1986) and corresponds to a level no lower than the upper part of the D. balticus Zone or more probably the P. densus Zone of Scandinavia (Lindholm 1991). According to Fortey and Owens (1987, p. 99) no strata of Tetragraptus approximatus Zone age have been proven to exist in Wales, although they suspect rocks equivalent in age to the upper part of the zone to be present. All of the graptolite fauna described herein appears to be older than the T. approximatus Zone. Cooper and Lindholm (1991) have made an attempt at estimating the relative duration of the different intervals in the Early Ordovician. That study indicates that the duration of the Hunneberg Series, as proposed here, is longer than the Tremadoc, taken as the Rhabdinopora flabelliformis desmograptoides Zone - Apatokephalus serratus Zone (the Scandinavian concept). It is only slightly shorter than the ‘remaining’ Arenig and of approximately equal length to the combined Llanvirn-Llandeilo. A further argument for a Hunneberg Series is the disagreement between workers on different fossil groups if the beds in question are ‘Tremadoc’ or ‘Arenig’ in age. Graptolite workers have generally considered the ‘La 2’ beds as ‘Tremadoc’, whereas conodont workers, working in different facies, call coeval beds ‘Arenig’. It should be noted here that the base of the conodont zone of P. proteus lies considerably lower than the base of the T. approximatus Zone, contrary to the views of Barnes et al. (1988) (Lofgren in prep.). Different graptolitic facies have been treated equally ambiguously (Lindholm 1984): typical La 2 beds have been referred to the Tremadoc, whereas the coeval A. murrayi beds have been considered to be of Arenig age (e.g. Thoral 1935; Destombes et al. 1969). The works of Williams and Stevens (1991), Stouge and Bagnoli (1988) and Lofgren (in prep.) have added to the precision in correlation between the graptolite and conodont zonation. According to conodont evidence, the lower part of the La 2 graptolite fauna is of Tremadoc age (that described by Williams and Stevens (1991) from Newfoundland) whereas higher parts (this work) belong to the P. proteus conodont zone, generally regarded as of Arenig age. LOCALITIES In Scandinavia, the Lower Hunneberg beds outcrop only in the Oslo region (east-central Oslo and Slemmestad) and at Mt Hunneberg. In Scania they are known only from the Krapperup core, the basal beds also from the Fagelsang core (D. balticus Zone of Hede (1951)). Based on lithological similarity, they appear to be present 288 PALAEONTOLOGY, VOLUME 34 both further to the south (SE Scania) and to the north (Hanrar at Lake Mjosa). These beds are, however, unfossiliferous. The only graptolite-bearing outcrop of Lower Hunneberg beds at Mt Hunneberg is at Storeklev, in the south-west wall of the mountain. Here, the Lower Hunneberg is represented by shale, and is thicker than elsewhere. The sequence gradually thins and shale gives way to limestone towards the eastern wall of the mountain. At Storeklev, graptolites are found scattered through the lower part of the shale, within which there is one rich band, 2- 15-2-32 m above the hiatus separating the Cambrian from Ordovician beds (Tjernvik 1956). The collections investigated from Storeklev consist of T. Tjernvik’s original material (PU Vg 124-127), B.-D. Erdtmann’s collections from the early 1960s (TUB HUN-S/2. 18-2.3/001-058) and my own collections from 1979-1986, belonging to Lund University. The Oslo region, c. 200 km north-west of Mt Hunneberg, contains several outcrops of Lower Ordovician graptolite shale, but the Lower Hunneberg beds are found only in the central part, at Galgeberg and Toyen (both in east-central Oslo) and in the Slemmestad area (c. 20 km south-west of the Oslo localities). The Galgeberg and Toyen localities were temporary construction sites, and are now inaccessible, whereas the Slemmestad area contains several well-exposed localities (road sections and beach sections; Text-fig. I b). My own collecting at Slemmestad has shown the graptolites to be less rare than at Storeklev, but at both localities there are unusually rich horizons. The collections investigated from the Oslo region consist of material from Galgeberg collected in the 1930s by T. Strand and A. Heintz (PMO 58.965-58.970); B.-D. Erdtmann’s collection from the Toyen underground station (GPIT1-T30; PMO 73.652); collections from various localities in the Slemmestad area, mainly by N. Spieldnaes, to a minor extent by G. Henningsmoen and D. Bruton (PMO 137, 73.187-73.192, 73.200, 73.204, 97.702, 97.705-97.706, 97.708, 108.557-108.574, 108.598-108.599, 112.966-1 12.970, 113.031-113.033, 120.751); and finally, my own collections from various localities in the Slemmestad area - the most productive one being Grundvik between Slemmestad and Naersnes to the south. My collections are all measured in sections, and belong to Lund University. The investigated part of the Krapperup core (situated c. 230 km S of Mt Hunneberg) consists of the lowermost c. 42 m (155 06-112-80 m), comprising the Lower Hunneberg beds. 193 samples, not all of which contained identifiable graptolites, have been taken out of this part of the core. The core was drilled in the 1940s and belongs to Lund University. Its diameter is 62 mm. All the material examined consists of medium grey to almost black, non-calcareous, shale/mudstone. The preservation of the graptolites varies from flattened to full relief, infilled with pyrite or, commonly in the lowermost part of the Krapperup core, with calcite. In the latter case, the periderm is usually very brittle and partly flakes off during splitting of the slab or preparation. Also, some of these graptolites were partly compressed and deformed before infilling with calcite. They are, consequently, often hard to identify. GRAPTOLITE TERMINOLOGY The terminology in general follows that of Bulman (1970) and Cooper and Lortey (1982, 1983; isograptid development type, dextral and sinistral mode, consecutive and delayed dichotomies etc.). Didymograptid and tetragraptid proximal part refers to the length of first-order stipes (several vs. one theca each). Profile stipe width refers to measurements made from the dorsal edge of the specimen to the ventral wall of a theca, at its aperture - the aspect of the stipe is referred to as "profile view'. Lateral stipe width refers to specimens in dorsal or ventral view ("dorsoventral view' ; horizontal preservation of multiramous specimens), that is, measurements are made from side to side of the stipe. The number of thecae in 10 mm has usually been measured over the available number of thecae, and then recalculated. Stipe divergence angles are measured as the angle resulting from the tangents of the dorsal side of the stipes across a specified thecal aperture. Secondary cortex cover in general refers to what appears to be an 'envelope' around the stipe, compressed to a film in the bedding plane in an arbitrary preservational aspect of the specimen; only exceptionally does the cortex cover appear to have thickened the stipe into a robust "rod’. The terms sicular bitheca and plaited thecal structure are explained in the section on evolution. Dichograptid stipe indicates a stipe without triad budding or plaited thecal structure, i.e. "fully graptoloid’. Graptoloid thecal notation is used throughout. In the systematic section, the suprageneric classification of Lortey and Cooper ( 1986) has, in general, been followed (see discussion on the Sigmagraptinae, however). In the synonymy lists the signs recommended by Matthews (1973) have been used. Under the heading of "Associated species’ are listed only the species found on the same bedding plane as the species under discussion. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 289 EVOLUTION AND PHYLOGENY The fauna discussed in this paper represents a stage in graptolite evolution when (along various lineages) biradiality had generally been attained and bithecae were in the process of being lost. The coexistence of sinistral and dextral forms of a species is a common feature in early Elunneberg time. The fauna does not verify any general trend towards reduction in the number of stipes present in a rhabdosome. Loss of bithecae. Bithecae were lost along different lineages in a rather restricted time period. Late Tremadoc and Early Hunneberg. Late Tremadoc graptolites showing various degrees of bithecal reduction were described by Williams and Stevens (1991). I have not been able to verify if the various steps in this transformation follow in the same order (homotaxially) in different groups, nor if bithecae were lost progressively along a stipe or instantaneously. I have seen a limited number of combinations of primitive and advanced traits (Text-fig. 4a) that can be listed as five steps in a morphological series: 1. The typical anisograptid : a fully bithecate rhabdosome, with normal triad budding, i.e. successive groups of one autotheca, one bitheca, and a stolotheca, produced at stolonal nodes. These groups alternate regularly (Bulman 1970, fig. 8), so that, from one side of the rhabdosome, only every second bitheca can be seen (Text-fig. 4a: 1, B). In profile view, the thecae are seen to bud laterally (Text-figs 4c, d, 5a, b). The alternation is seen as a zig-zag or sinuous pattern in dorsal view. 2. A fully bithecate rhabdosome with irregularities in the triad budding, that is, two or more successive bithecae may be present on one side of the rhabdosome (Text-fig. 4a: 2, c, d), e.g. Kiaerograptus supremus. 3. Only a sicular bitheca is present, i.e. the bitheca associated with th l1 and present between the sicula and th l1 on the obverse side. The stipes have traces of triad budding, here termed plaited thecal structure. The name has been chosen to illustrate the zig-zag or sinuous path of the common canal, as seen in dorsal view, caused by the fact that the thecae still alternate, even though the hi thecae have been lost (Text-fig. 4a: 3, e), and their proximal parts produce a 'herringbone' or plaited structure (in dorsal view). The budding is closer to the dorsal side of the rhabdosome than in the bithecate species examined, suggesting the possibility that the transition from lateral to dorsal budding was a gradual one. 4. The sicular bitheca remains, but the stipes are of normal dichograptid appearance (Text-fig. 4a: 4), as in Hunnegraptus copiosus. This change in the stipes is apparently coupled with a reduction in total thecal length. A growth stage preserved in relief (Text-fig. 4 F). might give a clue as to the disappearance of the sicular bitheca. It appears to have the proximal part of the bitheca, which has stopped growing. The 'aperture' is covered by periderm. Since only one specimen has been found, this interpretation is uncertain. The specimen could be pathological or deformed by compression. 5. The last primitive character, the sicular bitheca, is lost, and the 'dichograptid' development stage is reached (Text-fig. 4a: 5). In addition to what I have observed, Williams and Stevens (1991), using isolated specimens, found that residual bithecae may be found associated with dichotomies after disappearance of bithecae from the rest of the stipes. Phytogeny. The phylogeny of the fauna is difficult to trace, mainly because of the rarity of well- preserved graptolites of Late Tremadoc age. What is evident is that the Paradelograptus group nourished in Early Hunneberg time, with at least six species present in Scandinavia. The genus belongs in the family Sinograptidae (see further discussion with systematic descriptions) which, judging from proximal and thecal characters, derives its origin from Adelograptus tenellus , and thus not via an unspecified dichograptid, as suggested by Lortey and Cooper (1986, text-fig. 1 1 ). The Sinograptidae constitutes one of the independent lineages with bithecal reduction. Another is the Clonograptus s.s. lineage. The earliest representatives of the lineage known from relief material n+1 n+2 n+3 290 PALAEONTOLOGY, VOLUME 34 3 ■=* -a * ■ 50 & g, 5 73 73 C_> i/~) ^ 0 CU r- • 2 5 ! ^ 72 ; ^ c : o =s u, c 73 - * c ^ 5 03 o o . Cl, X O . „ n ^ X c O (73 X . i X C3 03 3 ■« (50 C MV X > CP a> C3 D 03 > 73 O o a,'*- * .c aj a3 " x> .5 E c/o O CP P> ^ -a C/2

"O S D (1) > c/2 (j }rj x 73 X 00 C c3 73 X CP H (73 00 X H 60 si ‘I ^ ^ 73 O ^ c 3 7 7 ^ o 5 ^ c x 73 X 73 „ O H CPX cX Q LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 291 (Lindholm and Maletz 1989) have a sicular bitheca and plaited thecal structure (C. aft. multiplex) or a sicular bitheca and normal dichograptid stipes (C. rigidus). Lindholm and Maletz (1989) restricted Clonograptus s.s. to species without bithecae along the stipes. Hunnegraptus is a probable descendant of an early Clonograptus species or of one of its ancestors. A certain variation in first- order stipe length is known in Clonograptus , and the length was accentuated in Hunnegraptus. It is possible that one group of didymograptids (D. undulatus , D. protobalticus-balticus, D. geometricus etc.) derives its origin from Hunnegraptus , through suppression of higher-order dichotomies. Likewise, a number of horizontal and reclined tetragraptids (and Dichograptus species?) may derive their origin from a species of Clonograptus s.s. related to C. magnificus-multiplex . Broad stipe fragments with thecae of reclined tetragraptid type (long, somewhat curved thecae, with high inclination and high thecal overlap) are sometimes met with in the Lower Hunneberg fauna. Other taxa are more problematic. For instance, did Kiaerograptus give rise to another group of didymograptids (another separate lineage with bithecal reduction) and/or the earliest isograptids (see p. 320)? What is the origin of the early ‘corymbograptids ’ found in Scandinavia, Britain and Spain - probably the very earliest didymograptids - and the 3- and 5-stiped forms? SYSTEMATIC PALAEONTOLOGY Repositories of specimens. Abbreviations used are as follows: GPI, Institute of Geology and Palaeontology, Gottingen, Germany; GSC, Geological Survey of Canada, Ottawa, Canada; LO and LR. Department of Historical Geology and Palaeontology, Lund, Sweden; PMO, Palaeontological Museum, Oslo, Norway; PU, Palaeontological Institute, Uppsala, Sweden; RM, National Museum of Natural History, Stockholm, Sweden; SGU, Geological Survey of Sweden, Uppsala, Sweden; TUB, Technical University, Berlin, Germany. Order graptoloidea Lapworth, 1875 Diagnosis (from Fortey and Cooper 1986). Graptolites in w'hich the nema is retained in the adult stage. Incerti subordinis Family anisograptidae Bulman, 1950 Diagnosis (from Fortey and Cooper 1986). Paraphyletic group, sicula retains nema in adult stage, bithecae present, rhabdosome more or less bilaterally symmetrical, and quadriradiate, triradiate or biradiate. Remarks. The Anisograptidae is a very heterogenous group, with many of its biradiate taxa closely similar to various taxa within the Dichograptina, the only difference being the presence of bithecae along the stipes in anisograptids. In my opinion, to obtain a phylogenetically based classification, the inclusion of taxa with bithecae along the stipes will eventually have to be accepted in the Dichograptina, thus necessitating a redefinition of that group. What would be left in the Anisograptidae, in that case, would be its tri- and quadriradiate taxa, which are probably rather closely genetically related, since they appear in a relatively short interval of time just after the origin of planktonic forms. Additionally included would be those biradiate taxa that cannot be linked with a dichograptinid form. For practical purposes, this change would make classification (above the genus level) easier, since most forms are not well enough preserved to reveal bithecae. Genus kiaerograptus Spjeldnaes, 1963 Type species. Kiaerograptus kiaeri (Monsen, 1925). Diagnosis (based on Spjeldnaes 1963; Rushton 1981; and author’s observations). Rhabdosome biradiate, composed of two reclined to declined stipes, one of which may be aborted after the first 292 PALAEONTOLOGY, VOLUME 34 theca; one stipe may branch near, or at some distance from, the sicula. In some rare cases, an extra proximal theca may represent an aborted third stipe. Autothecae are of dichograptid type but may have isolated distal parts. Bithecae present at sicula and along stipes. Triad budding not always regular. Kiaerograptus supremus sp. nov. Text-fig. 5 v 1965a Kiaerograptus kiaeri (Monsen); Erdtmann, pp. 106-107, pi. 2, figs 1 and 2; pi. 3, fig. 4. Name. Latin supremus , uppermost, indicating its position as the last of the fully bithecate species in the Krapperup core. Material. 46 specimens, of which 44 come from the 1 5 1 -96—144-57 m level of the Krapperup core and 2 from the Toyen section, Oslo (both found on PMO 73.652; illustrated by Erdtmann 1965a). Holotypc LO 5970T (Text-fig. 5a), paratype LO 597 1 1. Associated species. ?Trograptus sp., P. onubensis , A. murrayi. Stratigraphic range. K. supremus and A. murrayi Zones. Diagnosis. Rhabdosome composed of two undivided declined stipes. Proximal development comparable to isograptid type. Bithecae present throughout stipes but sometimes not regularly alternating. Length of sicula l-7-2-0mm, stipe width 0-8-1 T mm, 12-13 thecae in 10 mm, divergence of stipes 115-140°. Description. The species is a typical anisograptid, with bithecae developed at most or all available nodes (with possible reductions in the stratigraphically youngest specimens - no pyritized specimens are available above 147-66 m). The proximal development type resembles the isograptid development, i.e. th 1 2 is dicalycal. As seen in Text-figure 5 a, though, theca 22 emerges from the sicula-facing side of th l2, indicating that the triad (alternate) budding mechanism operates already in this position. Both sinistral and dextral forms are found (compare Text-fig. 5a with 5b). A sclerotized stolon system has not been observed: a relief specimen (now unfortunately lost) from the Krapperup core, filled with clear calcite, appeared 'empty' inside. The sicula is tube-like, I -7—2-0 mm long and 0-3-0-45 mm wide at the aperture, depending on the degree of compression. The first bud emerges approximately 0-25 mm from the apex of the sicula. The stipes show typical triad budding, i.e. the autothecae are seen to emerge alternately from opposite sides of the stipe. The first bi theca of each stipe (as well as the sicular bitheca) seems to occur on the obverse side. Thecal length, including stolothecae, can be estimated at 2 mm. Thecal width at the aperture is 0-5 mm, sometimes slightly more in flattened specimens. The free ventral part of the thecae is somewhat curved, especially if the proximal part of the theca is more completely pyritized than the distal part (see Text-fig. 5 a). The inclination of the distal parts of thecae varies from 30° to 45° depending on the degree of compression. There are 12-13 thecae in 10 mm. The thecal overlap is difficult to estimate due to the triad budding: the thecae do not bud dorsally, but laterally. In regular triad budding, the budding point of every second theca is on the unexposed side of the specimen. Such a theca will be seen only as a wedge between the preceding and the following theca (see Text-figs 4 and 5). The bithecae are about 0-4 mm long and 0-1-0-15 mm wide. They do not reach the aperture of the previous autotheca. Text- figure 5b shows a stipe with irregular triad budding: the bithecae associated with th 22 and 42 are on the obverse side, whereas that associated with th 62 is on the reverse side. The profile stipe width is 0-8-0-9 mm in relief specimens, and 0-9—1 - 1 mm in flattened ones. The stipe divergence angle is 115-140°. Remarks. Within the studied area, the species was found only in the Krapperup core and the Toyen section in Oslo (Erdtmann 1965a). From the latter area only two specimens from a shale band at the very top of the Ceratopyge Limestone unit were found. This limestone is considered as the top of the Tremadoc in Scandinavia. Most or all of it is younger than the youngest Tremadoc beds present in the type area. Because the Krapperup core lacks the limestone, it is a little difficult to correlate the two occurrences of the species. On circumstantial evidence (Fagelsang core), the Krapperup specimens are somewhat younger. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 293 text-fig. 5. Kiaerograptus supremus sp. nov. from the Krapperup core, a, holotype, LO 5970T, a dextral specimen in full relief, 151 -46—15 1-50 m. b, LO 597 1 1. a sinistral specimen in low relief, with irregular triad budding, 150-13- 150- 17 m. Both specimens drawn under vertical light. The specimens found in the Krapperup core are mostly rather small, the longest stipe seen consisting of 16 thecae (Text-fig. 5 b), whereas an average stipe consists of only 3-6 thecae. On the whole, the beds with K. supremus are fairly poor in graptolites, at least those well enough preserved for identification. A. murrayi appears in the higher part of the range of K. supremus. K. supremus seems most closely related to K. klotsehichini (Obut, 1961 ). This species was referred to Didymograptus, but the original illustration (Obut 1961, pi. 1, fig. 7, 7 a) shows indications of triad budding along the stipes. The measurements of that species are close to those of K. supremus , except for a shorter sicula (but the illustration gives the impression of a longer sicula than mentioned in the description) and a slightly narrower final stipe width. K. klotsehichini was found in clay shales in the southern part of the Ural Mountains. Apparently it was not associated with any other species and its precise age is unknown. ? Didymograptus sp. Bulman, 1954, is probably the oldest Kiaerograptus species so far known, found at a rather low level of the Dictyonema Shale in the Oslo region ; it has an outline fairly close to that of K. supremus. It differs mainly in having a longer sicula, a slightly narrower final width, and stipes that distally become nearly horizontal. Bulman (1954, p. 36) noted that there was no trace of bithecae or stolothecae, but the material is totally flattened, thus making it impossible to see such details. Both K. klotsehichini and ? Didymograptus sp. are known only from a few specimens, so it can be supposed that only a part of the full range of variation has been revealed. Two other species, K. kicieri (Monsen, 1925) and K. quasimodo Rushton, 1981, show a great inherent variation. I have studied the material of K. kiaeri , 470 specimens, that formed the basis of the publication by Monsen (1925), and among these specimens the variation in, for example, stipe attitude, number of thecae in 10 mm, and the number of thecae with isolated distal parts is such that the end members of the variation would hardly have been recognized as belonging to the same species, were it not for all the intermediate specimens. The excellently preserved material of K. kiaeri described by Spjeldnaes (1963) shows, in addition to this variation, at least three successive bithecae on the same side of the stipe (see Text-fig. 4d) - a type of irregularity found also in K. supremus. K. quasimodo resembles K. kiaeri in the variation of, for example, stipe attitude and 294 PALAEONTOLOGY, VOLUME 34 distal isolation of thecae, but also has a variable number of stipes (which is comparatively rare in K. kiaeri - less than 2%); sometimes stipes are 'aborted’ after their first theca (see Rushton 1981, figs 2 and 3). K. quasimodo also has occasional second-order branching close to the sicula, giving three-stiped specimens. Compared with these two species, K. supremus has differently shaped bithecae, a more constant and lower stipe divergence angle, more rigid stipes, and apparently no thecae with isolated distal parts. The stipes also diverge from the sicula closer to its aperture. I interpret the latter three characters as more advanced, probably indicating that K. supremus comes from a higher stratigraphical level. Genus araneograptus Erdtmann and VandenBerg, 1985 Type species. Dictyonema macgillivrayi nom. nov. T. S. Hall, 1897 (= Dictyonema grande T. S. Hall, 1891 ; non D. grandis Nicholson, 1873). Diagnosis (taken from Erdtmann and VandenBerg 1985). ' Rhabdosome siculate, biradial, produced by dichotomous division (similar to Clonograptus ), generally at steadily increasing intervals, to eighth or ninth order [or possibly more] (usually fourth to sixth); adjacent branches connected by more or less regularly spaced dissepiments; autothecae in proximal portions denticulate with concave ventral margins and of moderate inclination; bithecae not observed. Juvenile specimens, up to the third-order dichotomy, cannot be assigned to a particular species, because of their identical morphology and structural development.’ Remarks. The absence of bithecae in the type species cannot be considered proven on the basis of the Australian material used by Erdtmann and VandenBerg, since this material is completely flattened and cannot possibly reveal such characters. For this reason I leave Araneograptus with the Anisograptidae. Also the biradiality of the rhabdosome is not proven beyond doubt. All the illustrated details of proximal ends (Erdtmann and VandenBerg 1985, fig. 6a-c) show an asymmetry which could be interpreted, instead, as three primary stipes. If this is the case, the genus is a junior synonym of Rhabdinopora Eichwald. Araneograptus murrayi (J. Hall, 1865) Text-figs 6, 7, ? 1 8 c 1865 Dictyonema Murrayi J. Hall, pp. 138-139, pi. 20, figs 6 and 7 [photographs seen], 1865 Dictyonema quadrangularis J. Hall, p. 138, pi. 20, fig. 5. 1873 Dictyonema grandis Nicholson, pp. 134-136, fig. 1. v 1937 Dictyonema cf. murrayi J. Hall; Monsen, pp. 89-92, pi. 11, fig. 2. 1982 Dictyonema murrayi J. Hall; Mu et al., p. 295, pi. 73, fig. 1. 1982 Dictyonema quadr angular e J. Hall; Mu et al., p. 295, pi. 73, figs 2-4. 1982 Dictyonema maximum Xu sp. nov.; Mu et al., pp. 295-296, text-fig. 101, pi. 73, figs 5-7. 1982 Dictyonema ziyangense Xu sp. nov.; Mu et al., p. 296, pi. 74, fig. 3. 1985 Dictyonema pulchellum T. S. Hall; Rushton, p. 332, figs 1 and 2. 1985 Dictyonema sp. Rushton, p. 332, figs 3 and 4. 1987 Araneograptus murrayi (J. Hall); Gutierrez Marco and Acenolaza, pp. 325-330, pi. I v 1988 ‘ Dictyonema ’ cf. yaconense Turner; Molyneux and Rushton, pp. 65-66, fig. 8. Lectotype. GSC 962 a, J. Hall’s ( 1865) pi. 20, fig. 7; Text-fig. 6 herein; designated lectotype by Gutierrez Marco and Acenolaza (1987). Material. From the Krapperup core ( 148 . 22-109 . 86 m), c. 30 surfaces (each 30 cm2) with 1- > 5 specimens of different sizes, ranging from juveniles to fragments of giants. The species is most common in the lower part of LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 295 its range, up to 136-69 mm. At 141 m the species is very common, forming layers each a couple of millimetres thick. The species is absent at Mt Hunneberg, and only a few specimens have been found in the Oslo region, all except one slab from the Slemmestad area: PMO 58.967 from the T. phyllograptoides Zone (more probably lower) at Galgeberg (east-central Oslo), a large specimen figured by Monsen (1937), associated with a juvenile; PMO 137 from Gjeitungholmen, Slemmestad, allegedly from the upper Tremadoc Ceratopyge Shale; PMO 108.568, 108.569, 108 . 572 - together containing 6 specimens (of which 3 are juveniles) from 0-5-10 m above the Ceratopyge Limestone at Slemmestad crossroads; PMO 1 12.967+ 1 12.969, PMO 1 12.969+ 1 12.966, two relatively large specimens from Prestengveien, Slemmestad; and PMO 120.751, three specimens from 21-70 m at the new standard section (=4-1 m above the Ceratopyge Limestone), central Slemmestad. Associated species. K. supremus , H. copiosus, T. krapperupensis , horizontal tetragraptids Cquadribrachiatus’- type), three-stiped extensiform tetragraptids, Didymograptus sp. 1, P. antiquus , P. elongatus , P. tenuis. Stratigraphic range. A. nmrrayi to T. phyllograptoides Zones, possibly also lower and higher. Maximum abundance in the A. nmrrayi Zone. Diagnosis (based on the Scandinavian material and Gutierrez Marco and Acenolaza (1987)). Rhabdosome conical, mostly subtending an angle of 60-75° when flattened, angle decreasing distally in big specimens. Meshwork normally has 3-4 stipes in 10 mm, and 2-3 dissepiments in 10 mm, but the total variation ranges well outside these limits. The shape of the meshes is variable, from rectangular to oval. The lateral stipe width is over 1 mm, the thickness of the dissepiments is variable. The maximum length of the rhabdosome is unknown, but at least 30 cm. Description. Not much is known about the details of proximal growth pattern of the species. A few immature specimens have been found at different levels in the Krapperup core. A couple of these could possibly support a biradiate origin of the rhabdosome, while others seem asymmetrical enough to indicate a triradiate origin. The sicula is L8- 1-9 mm long where seen in full but presumably somewhat longer, perhaps up to 2-5 mm, in some more mature specimens. A very short nema of normal thickness is seen in a couple of the immature specimens. No specimen is well enough preserved to show beyond doubt a biradiate origin or any details of proximal development. At two levels (those of Text-figs 17 and 18), pyritized immature specimens of various species occur. Some of these are pendent and may belong to A. nmrrayi but, due to the lack of dissepiments, this cannot be proved unequivocally. All pendent forms seen in obverse view have a sicular bitheca. One of the specimens seen in reverse view (Text-fig 18c), shows a dicalycal theca l2 and a two-stiped origin. It apparently lacks plaited thecal structure. In some of the slightly larger specimens (Text-fig. 7f, h) the sicula, and sometimes also more of the proximal region, is covered with cortical tissue, extending on to the nema, which is then up to more than 1 mm thick. A couple of thecae are seen in partial profile view in one of the immature specimens, giving an estimate of 11-5 thecae in 10 mm. The thecae seem to be straight tubes of normal dichograptid appearance. On the other hand, a Moroccan specimen (Text-fig. 7b) shows a few thecae in relief which are very denticulate, the distal part of the ventral side being almost at right angles to the dorsal margin of the stipe. This high angle could, however, be due to distortion. The thecae number about 12-5 in 10 mm in this specimen. The difference in thecal shape between the two specimens can be explained in different ways: either the thecal shape changes along the rhabdosome, or the slightly oblique position of the thecae in the Scandinavian specimens obscures their true shape. Another possibility is, of course, that there is more than one species which cannot be distinguished on the basis of cone shape and mesh pattern alone. Ruedemann (1947, p. 171 ) commented on the thecae thus: ‘Thecae numbering 9-10 in 10 mm; apparently with acute extensions of apertural margins.’ Rushton (1985, fig. 2c) showed elongate thecae with high overlap and high distal inclination. Normally, only the dorsal side of the stipes is seen, since this is the outward-facing side of the cone and also represents the surface most easily exposed by splitting. The lateral stipe width is mostly 1-2—1 -5 mm in flattened specimens. Specimens with some relief often have thinner stipes, down to 10 mm. The stipe width of immature specimens is sometimes as low as 0-6-0-7 mm. The dissepiments are rather regularly spaced within a specimen (closer in the proximal part, though), but the number of dissepiments per length unit varies markedly from one specimen to another, from about 4 in 10 mm down to 1-5. The average density is about 2-3 dissepiments in 10 mm. Also the stipe density varies between specimens. This variation is due to the frequency of dichotomies 296 PALAEONTOLOGY, VOLUME 34 text-fig. 6. GSC 962a, lectotype slab of A. murrayi (J. Hall) containing three specimens, x 1. a, specimen of quadrangularis type with short meshes and relatively broad dissepiments, b, lectotype. c, specimen with short meshes and relatively thin dissepiments. The specimens are associated with numerous rhabdosomes of Clonograptus rigidus. (see Text-fig. 7 c) and the angle of the cone. There are normally about 3-5-4 stipes in 10 mm, but the total variation ranges from 3 to 5. The variation in stipe and dissepiment density gives a marked variation in size and shape of the meshes. Another factor influencing this is the thickness and shape of the dissepiments. The meshes can thus be square, rectangular, nearly circular, or oval. The thickness of the dissepiments varies from considerably thicker to noticeably thinner than the stipes, but in most specimens they are of about the same thickness as the stipes. The dissepiments are sometimes uniformly thick, in others thinner in their middle part. Secondary cortical additions to the stipes and dissepiments can, under special circumstances, almost fill out the meshes (Text-fig. 7e). The formation of dissepiments seems to have been very regular, these being inserted in every second or third position at the same time (or rather, the sariie distance from the sicula), so that the meshes form diagonal rows across the specimen (see Text-fig. 7a, c). This pattern is disturbed where dichotomies occur. These are relatively frequent in proximal parts (see Text-fig. 7f) but rare in the distal part of large specimens. Text-figure 7c shows two zones of stipe division, both apparently induced by irregularities in the mesh pattern. The left zone compensates for the loss of a stipe (t in the figure), the right-hand one seems to compensate for a deflection to the right of one stipe, as seen by the very small mesh to the right of this stipe slightly more proximally than the point of dichotomy (this interpretation seems more probable than that a stipe division was planned for). In both cases the dichotomies compensating a disturbance are paired, followed by an additional dichotomy slightly later. Paired dichotomies were illustrated also by Rushton (1985, fig. 4). A couple of relief specimens (see Text-fig. 7 d) have what appears to be later additions attached on the outside of the rhabdosome, pouch-like ‘balconies’ that join the stipes on their dorsal side. They do not seem to form part of the normal dissepiments. Their function is likely to have been to direct water currents through the rhabdosome meshwork. A distal end fragment (Text-fig. 7a) shows that dissepiments are present at normal frequency to the very distal end of the stipes, i.e. they are produced as the stipe grows. Further, the distalmost dissepiments have full width, but the 2-3 last produced of them seem to be less dense. This may explain the apparent lack of strength, leading to the disruption shown in Text-figure 7 a. The angle of the cone is normally 60-75°, but in a couple of cases angles as low as 40-50° have been observed. As seen from some very large Spanish specimens, the angle of the cone decreases distally. The angles of the larger Scanian specimens were impossible to measure, since the drillcore surfaces contain only small fragments LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 297 osliStes #o§Ws»w MM; text-fig. 7. Araneograptus murrayi (J. Hall). Black dots in rhabdosomes indicate the positions of stipe divisions, a, LO 5972t, Krapperup core, 137-70 m, the distal end of a large specimen; stippling indicates not fully corticized dissepiments, dashed outlines represent superposed phyllocarids. b. PMO 120.752, detail of a Moroccan specimen showing thecae in some relief, c, LO 5973t, Krapperup core, 137-60 m, part of a large specimen with relatively elongate meshes and two zones of branching; + represents the termination of a stipe. d, PMO 112.966, Slemmestad, detail of a specimen with pouch-like ‘balconies', e, PMO 58.967 (= Monsen 1937, pi. 11, fig. 2), Galgeberg, Oslo, detail showing cortical overgrowth of a mesh, f, PMO 137, upper Tremadoc (?), Gjeitungholmen, Slemmestad, a specimen with the proximal part covered by cortical tissue (stippling). G, LO 5974t, Krapperup core 147-66-147-72 m, a specimen without cortical overgrowth of the proximal part, and possibly indicating a triradiate origin, h, LO 5975t, Krapperup core 140-30 m, a specimen with cortical overgrowth of sicula and nema. of rhabdosomes. The maximum number of stipe dichotomies is unknown but. judging from the very low number of branchings on the drillcore surfaces, the rhabdosomes must have been very big, with a cone length of perhaps more than 20 cm. Remarks. The large Scanian specimens are closely similar to the specimen figured by J. Hall (1865, pi. 20, fig. 7), except that their dissepiments are on the average somewhat thicker. Hall, according to the figured material of Dictyonema murrayi , allowed a certain variation in the number of dissepiments per length unit, the specimen in plate 20, figure 6 having a dissepiment spacing equal to that of D. quadrangularis figured on the same plate. Hall seems to have found the thickness of the dissepiments more important than their spacing. But, on the type slab GSC 962 n, containing plate 20, figure 7 and two additional specimens (Text-fig. 6), all three shapes co-occur on one surface, indicating their probable conspecificity. Gutierrez Marco and Acenolaza (1987) also synonymized D. yaconense from South America and Nyssenia zemmourensis from northern Africa, as well as some variously named European finds, with A. murrayi. Interestingly, they noted that the descriptions of different species fitted different parts of one rhabdosome. They hesitated to synonymize A. pulchellus, although they noted that in some 298 PALAEONTOLOGY, VOLUME 34 characters the two species were partly overlapping. Their main argument was that no specimen of A. pulchellus of the size of the larger A. murrayi specimens has ever been found. Although they did not regard D. grandis Nicholson as a junior synonym, I suggest that it is. According to Nicholson (1873), D. grandis differs from D. murrayi by having conical form, more frequent bifurcation, and meshes wider than long. The first two differences are easily explained by Nicholson’s specimen being the proximal part of a rhabdosome and Hall’s specimens more distal fragments. The third difference, the shape of the meshes, can be explained by tectonic distortion, coupled with more closely-spaced dissepiments in the proximal part of the rhabdosome. Both J. Hall’s and Nicholson’s types derive from Levis, Quebec. A. murrayi has a world-wide distribution, including Europe (Scandinavia, Great Britain, Germany, France, Spain), northern Africa, eastern North America, South America (as D. yaconense; Argentina, Bolivia), and NW China. The species was listed, but not illustrated, from the Taimyr area of the Soviet Union by Obut and Sobolevskaya (1962). A more detailed account of the distribution is given by Gutierrez Marco and Acenolaza (1987). In addition, the possibly conspecific A. pulchellus is found in Australasia and western Canada. Both species seem to be restricted to a relatively narrow stratigraphical interval, corresponding to the Australasian stage La 2, and in some cases the basal part of La 3. Suborder dichograptina Lapworth, 1873 Diagnosis (from Fortey and Cooper 1986; emended by Lindholm and Maletz 1989). Graptoloids lacking bithecae along the stipes, and without virgella. Remarks. The diagnosis by Fortey and Cooper has been emended to incorporate in the Dichograptina the anisograptid/dichograptid intermediary forms with a sicular bitheca, but without bithecae along the stipes. Since the loss of bithecae apparently occurred in different lineages during a relatively short period of time, the Dichograptina, like the Anisograptidae, will be a paraphyletic group, no matter where the boundary between the two groups is drawn (however, see remarks on the Anisograptidae, p. 291). Superfamily dichograptacea Lapworth, 1873 Diagnosis (from Fortey and Cooper 1986; slightly emended by Lindholm and Maletz 1989). Dichograptinids lacking isograptid symmetry, number of orders of dichotomy in rhabdosome not limited. Remarks. The diagnosis by Fortey and Cooper has been emended to include forms which apparently have unlimited capacity for dichotomy, e.g. the genus Clonograptus. Family dichograptidae Lapworth, 1873 Diagnosis (from Fortey and Cooper 1986). Dichograptaceans lacking prothecal folds and sigmagraptine proximal end. Genus hunnegraptus gen. nov. Name. From Ml Hunneberg. Type species. Hunnegraptus copiosus gen. et sp. nov. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 299 Diagnosis. Dichograptids with a didymograptid proximal part and two or more orders of stipes produced by dichotomous division; first-order stipes extensiform or declined, normally consisting of approximately 5-25 thecae. Sicula tube-like; a sicular bitheca present at least in the type species. Proximal development isograptid, sinistral and dextral forms co-occurring. Thecae straight tubes of dichograptid type, no bithecae observed along stipes. Secondary cortical thickening may be present; possible stipe regeneration in gerontic specimens. Species included. H. copiosus sp. nov., H. tjernviki sp. nov., H. robustus sp. nov., and provisionally H. novus (Berry, 1960), H. gulinensis (Wang, 1981), H. minor (Wang, 1981 ), H. sichuanensis (Wang, 1981 ), H. sp. (Wang, 1981). Wang’s taxa are probably synonyms. Remarks. The Scandinavian specimens of the genus are here divided into three contemporaneously occurring species. However, it is possible that they all belong to one species, representing different growth stages. Unfortunately, though, the ’mature’ and ‘gerontic’ specimens are very rare compared to the ‘adolescents’ and a continuous spectrum of variation cannot be proven based on the available material. Therefore the three species H. copiosus (abundant, and the only one of the three found as immature specimens; ‘young stage’), H. tjernviki (less common; ‘mature stage’), and H. robustus (rare; ‘gerontic stage’) are here described as separate entities. If their synonymy can be proved, the name H. copiosus takes precedence. Synonymy would imply that all thecae kept growing through the entire life of the colony (cf. Williams and Stevens 1988), or until the apertures were covered by cortex. The ‘gerontic’ rhabdosomes are too flattened to prove that such ‘choking’ with cortex took place, but nearly all specimens of H. robustus have irregularly placed thin lateral stipes, as thin as those of the other two species. These stipes are connected to the main body of the rhabdosome by the cortex, and thus cannot be superimposed stipes belonging to other specimens. They do not influence the direction or thickness of the main stipes. They appear to represent a rejuvenation of the colony, extra stipes being inserted later than the surrounding branches. This could be to compensate for zooids no longer active in that part of the rhabdosome. These stipes are probably not metacladia, and I have seen no report of comparable stipe formation in any other graptoloid. The Scandinavian occurrences are restricted to the H. copiosus Zone, the zone directly underlying the Tetragraptus phyUograptoides Zone. Provisionally included in Hunnegraptus is H. novus (Berry, 1960), which has a sicular bitheca but normally no bithecae along the stipes. This taxon is probably older than the Scandinavian occurrences, however, since it is reported to co-occur with Anisograptus (Berry 1960). Provisionally included are also Kiaerograptusl gulinensis, K.l sp., Adelograptus minor , and A. sichuanensis , all described by Wang (1981 ) from probable Late Xinchangian beds of Sichuan, central China. All of these are most likely conspecific, the amount of variation among them being smaller than that within the type species of Hunnegraptus. Multiramous species with dichograptid thecae and prolonged first-order stipes have also been described from Spain (Gutierrez Marco 1982, 1986, pp. 290-304) and Czechoslovakia (Kraft and Mergl 1979). The material from Spain is of Early Hunneberg age, and the associated fauna contains i.a. A. murrayi. The age of the Czech material is uncertain. The relationship of these species to Hunnegraptus is not clear. The genus is presumably most closely related to Clonograptus , in which first-order stipes may be prolonged, e.g. in the type species, C. rigidus. The sicular bitheca seen in H. copiosus is also found in Clonograptus milesi , and probably also in the type species (Lindholm and Maletz 1989). Regenerated stipes of Hunnegraptus type are unknown in Clonograptus. The latter fact has convinced me that the distinction between the two taxa should be on a generic, rather than subgeneric, level. Hunnegraptus copiosus gen. et sp. nov. Text-figs 8a-f and 18f, ?h, j 1987 Dichograptid sp. 1 Maletz, p. 136, text-fig. 44:9; 10, pi. 5, figs 1 and 2. p 1987 Dichograptid sp. 2 Maletz, pp. 136-137, text-fig. 44:3?, 5-8. 300 PALAEONTOLOGY, VOLUME 34 LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 301 Name. Latin copiosus , abundant, refers to the abundance of the species in all sections. Material. 736 specimens, of which 1 1 1 come from east-central Oslo (Toyen, 3-55-3-68 m above the Ceratopyge Limestone, and Galgeberg), 213 from various localities in the Slemmestad area, 165 from Storeklev, Mt Hunneberg, and 247 from the Krapperup core (132-73-117-00 m). Except for Krapperup, the stratigraphic range is rather limited, with specimens collected from a single horizon in a section or, as in Storeklev and the new standard section in central Slemmestad, from a relatively thin band (1-4- 1-8 m and 4-6-5-6 m above the Ceratopyge Limestone, respectively). The holotype is found on PMO 58.969 from Galgeberg, central Oslo (Text-fig. 8e, f); the paratypes are LO 5976t-5979t, LO 6090t, LO 6094t, and PMO 108.599 (localities given in figure captions). The species is very common and is found on most slabs within its range, mostly as two- stiped, didymograptid-like, specimens. Associated species. H. tjernviki, H. robustus , A. murrayi , Clonograptus s.s. sp. indet., T. longus , T. cf. krapperupensis, narrow- and broad-stiped horizontal tetragraptids Cquadribrachiatus’-type), three-stiped extensiform tetragraptids, gen. et sp. indet. 1, Didymograptus sp. 2, Isograptus sp., P. antiquus , P. pritchardi , P. elongatus , P. tenuis , P. cf. rams. Stratigraphic range. H. copiosus Zone. Diagnosis. Didymograptid proximal part and one, two, or possibly more orders of stipes. First- order stipes normally declined, consisting of approx. 7-22 thecae each. Sicula mostly 11—1-2 mm long, sicular bitheca present. Thecae straight tubes with a thecal inclination of 15-20°. Stipe width 0-5-0-6 mm, approx. 12 thecae in 10 mm. No bithecae along stipes. Description. The sicula is tube-shaped, 1 -0—1 3 mm long, 0-25 mm wide at the aperture; and its distal part is inclined towards the stipe2 side. The proximal part of the nema is somewhat thickened, like the cauda (Hutt 1974). A slightly curved bitheca is present on the obverse side of the sicula, budding from th 1 1 at mid-length of the sicula (Text-figs 8 b, f and 1 8 f). The aperture of the bitheca is positioned where th l1 bends away from the sicula. Theca l1 buds from the sicula approximately 01-0-2 mm from its apex. In early growth stages consisting of the sicula and th 1 1 , the sicula and theca make a more or less symmetrical pair, with the bitheca in a central position. The proximal development type is isograptid, theca 21 budding from th l2 in its most proximal part. There are both sinistral and dextral specimens (compare Text-fig. 8 a with 8 d) but too few well- preserved specimens have been found for any statistical evaluation of predominance. The proximal part of the protheca is relatively narrow, resulting in a rather low thecal inclination (15-20°). A slight prothecal folding can be seen in pyritized relief specimens (Text-fig. 8 a, b, d). The metathecae are simple tubes, straight or nearly so; the total thecal length is about 1-2 mm, the dorsoventral thecal width at the aperture about 0-25-0-3 mm. The thecal overlap is about 40-50%, and there are normally 1 1-5-13 thecae in 10 mm (total range 10-5-14). The profile stipe width is 0-5-0-6 mm, bithecae are absent along stipes, both in their proximal and distal parts. The first-order stipes consist of 7-22 thecae (observed range) and make an angle of about 120-180° (normally 130-160°) if seen in profile view. A pyritized specimen possibly belonging to the species (Text-fig. 1 8 H ) shows a first-order stipe consisting of only 3 thecae. Stipe division is dichotomous; the longest second-order stipes text-fig. 8. a-h Hunnegraptus copiosus sp. nov. a, b, LO 5976t and LO 5976 + , Storeklev 2-15-2-32 m, counterparts of a dextral relief specimen, c, LO 5977t, Storeklev 2-32 m, an almost flattened specimen, showing the most typical appearance of the species, d, PMO 108.599, Slemmestad, a full relief sinistral specimen; note the difference in thecal spacing between the stipes: associated specimens show no tectonic distortion, e, f, holotype, PMO 58.969, Galgeberg, east-central Oslo, a sinistral specimen (proximal part is a mould) with the fourth second-order stipe presumably primarily missing; part of one stipe is pyritized, showing absence of bithecae; f is drawn from a latex cast and shows the sicular bitheca in low relief. G, LO 5978t, 22-2 m in the standard section, Slemmestad, a dextral specimen (the sicular part is a mould), one of the bigger specimens, in a preservation showing the low degree of rigidity of the stipes. H, LO 5979t, associated with specimen shown as a and b; the specimen has highly unequal length of first-order stipes, a, b, d, f are drawn from latex casts under vertical light, i-k, H. tjernviki sp. nov. I, PU Vg 125, Storeklev 227-230 cm, a stipe fragment in profile view, j, LO 5980t, Grundvik, Slemmestad, the longest stipe fragment, showing four orders of stipes; the drawing is a combination of counterparts. K, holotype, PU Vg 124, Storeklev 227 cm. 302 PALAEONTOLOGY, VOLUME 34 seen are more than 40 mm long. A branching stipe fragment probably belonging to the species (Text-fig. 1 8 j) shows isograptid type branching. No complete specimen shows more than two orders of stipes, but it cannot be excluded that the branching continues. In one specimen (LR 6. from Storeklev) an 'aborted' stipe division can be seen : part of a theca projects from the dorsal side of a stipe, but the stipe continues in its previous direction. Also the holotype may have lost a stipe in the same way (Text-fig. 8 e) : no trace of a fourth second- order stipe is seen, but in this case the stipe is bent as if branching had occurred. The majority of the specimens found are too small for branching to have occurred. These specimens look like declined didymograptids, but relief specimens in obverse view show the characteristic sicular bitheca. No marked secondary cortical thickening has been observed in this species. Specimens are often somewhat flexuous. No gerontic specimens were found, however, and the possibility of cortical thickening at a later growth stage cannot be ruled out. Remarks. The low thecal inclination and the faint prothecal folding observed in relief specimens of H. copiosus might suggest a relationship with the Sigmagraptidae. However, the proximal part differs markedly from that of the type species of Sigmagraptus (cf. Cooper and Fortey 1982, fig. 61): the length/width ratio of the sicula is smaller, the prothecal part of theca l1 is shorter, th 21 buds off th l2 slightly later, so that it crosses th l1 instead of following its dorsal side, and also neither th l1 nor th l2 has the characteristic sharp bend seen in sigmagraptids. Except for the sicular bitheca, the proximal structure in Hunnegraptus is normal dichograptid. The presence of both sinistral and dextral forms must be assumed to be a primitive character (common among the Anisograptidae). The Chinese species described by Wang (1981 ) are all very similar to H. copiosus. The illustrations show a proximal part virtually identical to that of H. copiosus , and sicular length and figures given for thecal characters differ insignificantly from those of that species. The only visible difference lies in the position of second-order dichotomy, which is closer to the sicula in the Chinese species. All the specimens illustrated by Wang (1981) have one or two stipe orders, like H. copiosus. H. novus (Berry, 1960) is known only as a two-stiped form. Also this taxon has a thecal shape reminiscent of H. copiosus. Hunnegraptus tjernviki gen. et sp. nov. Text-fig. 8i-k p 1987 Dichograptid sp. 2 Maletz, pp. 136-137, fig. 44: 1, 2, 4; pi. 5, fig. 3. Name. In honour of Torsten Tjernvik, the discoverer of the Early Hunneberg graptolite fauna. Material. 32 more or less fragmentary specimens, of which 1 1 are from Oslo, 10 from the Slemmestad area, 9 from Storeklev, and 2 from Krapperup. The range coincides with that of H. copiosus. Holotype PU Vg 124 (Text-fig. 8k.) and paratype PLf Vg 125 from Storeklev; paratype LO 5980t from Grundvik, Slemmestad. Associated species. H. copiosus , T. longus , P. antiquus. Stratigraphic range. H. copiosus Zone. Diagnosis. Didymograptid proximal part and up to four or possibly more orders of stipes. First- order stipes horizontal or slightly declined, consisting of several thecae. Sicula approximately 1-5 mm long, sicular bitheca suspected. Thecae straight tubes with an inclination of about 30°. Stipe width 0-8-1 -2 mm, about 1 T5 thecae in 10 mm. No bithecae along stipes. Description. The species is known from fewer specimens, and also in less detail, than H. copiosus. Most specimens consist of stipe fragments only. The sicula is about 15 mm long; no specimen is well enough preserved to reveal a possible sicular bitheca or the proximal development. The thecae are simple straight tubes, approximately I -4-1 -7 mm long and 0-4-0-5 mm wide at the aperture. The thecal inclination is about 30° in full profile view, less in obliquely preserved specimens. Profile stipe width is O8-b0mm in proximal parts, 0 9 12 mm in more distal parts; lateral stipe width (dorsoventral view) is about 05-0-6 mm. There are 10-5-12 thecae in 10 mm, and the thecal overlap is about 50%. Bithecae are not present along the stipes. The observed LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 303 length of first-order stipes falls within the range of variation seen in H. copiosus', the branching is dichotomous. The first-order stipes diverge at ?150-180°. A stipe fragment from Grundvik, Slemmestad (Text-fig. 8 j ) shows four orders of stipes, the most proximal stipe seen being of second or higher order, judging from the angles of dichotomy. Cortical thickening of stipes is variable, but mostly not readily observed. Remarks. The species differs from H. copiosus , and the species described by Wang ( 1981 ), primarily in having broader and somewhat more rigid stipes. The more rigid appearance might be explained in part by longer thecae and higher thecal overlap and in part by slight cortical thickening. In contrast to H. copiosus , no immature specimens of H. tjernviki have been observed. Hunnegraptus robustus gen. et sp. nov. Text-fig. 9 ? 1987 Dichograptid sp. 3 Maletz, p. 138, pi. 5, fig. 4. Name. Denoting the robust character of the species, as compared to H. copiosus and H. tjernviki. Material. Six specimens, all illustrated in Text-figure 9. Holotype LO 5981T, from 2 60 m at Storeklev, Mt Hunneberg. Paratypes PUVgl26, LO 5982t, LO 5983t, from the graptolite-rich band 215-2-32 m at Storeklev; LO 5985t, Storeklev, 2-60 m; LO 5984t from Grundvik. Associated species. H. copiosus. Stratigraphic range. H. copiosus Zone. Diagnosis. Didymograptid proximal part and up to four or more orders of stipes. First-order stipes consist of several thecae. Profile stipe width 1 -8 2-5 mm; lateral stipe width 1 0-2-5 mm, depending on cortical cover. Dichotomous and irregular lateral branching. Lateral stipes usually narrower than the rest of the rhabdosome. Description. The sicula and proximal development are unknown, since only distal parts are seen in profile view. The observed combined length of first-order stipes is 24-33 mm; the observed range of second-order stipes is from 24 to more than 50 mm. The thecae are long straight tubes, their length c. 3 mm, width 0-4 mm, and overlap about 75%. There are about 12 thecae in 10 mm. The profile stipe width is 1-8-2-5 mm (Text-fig. 9c). The lateral stipe width varies considerably, depending on the amount of secondary cortex cover: normally 1 -5—2-5 mm, but in some cases as thin as 10 mm. Secondary cortical cover is less marked in a distal direction, but this is in no way regular (see Text-fig. 9a). Thin (0-7-L0 mm wide) lateral stipes occur irregularly in five of the six specimens (see Text-fig. 9a-b, d f). These stipes are connected to the rest of the rhabdosome by the cortical thickening, and thus cannot represent superimposed fragments of other specimens. As discussed for the genus, I consider these stipes to have been formed secondarily and thus ignore them when counting the stipe order: the maximum found is four stipe orders, in the holotype. Text-figure 9e shows a specimen and its counterpart, with three broken lateral stipes seen on the counterpart only (after some preparation), indicating that they did not grow in the plane represented by the four main stipes. The lateral stipes shown in Text-figure 9f seem to have been originally directed slightly upwards, and later bent down to the bedding plane by compaction. Remarks. The species differs from the other two described Hunnegraptus species, and the species of Wang (1981), in its longer thecae, more robust stipes and thick cortex cover, as well as the occasional lateral stipes. No immature specimens have been identified. The lateral stipes appear to have been formed later than the surrounding parts of the rhabdosome (see the remarks on the genus). Genus clonograptus Nicholson, 1873 Type species. Graptolithus rigidus J. Hall, 1858. 304 PALAEONTOLOGY. VOLUME 34 text-fig. 9. Hunnegraptus robustus sp. nov. a, holotype, LO 598 IT, Storeklev 2-60 m, showing numerous thin lateral stipes, possibly a sign of regeneration; the drawing is a combination of counterparts, b, LO 5982t, Storeklev 2-32 m, arrows point to lateral stipes; note the two closely arranged stipes on the first-order stipe, connected proximally by cortical tissue, c, LO 5983t, Storeklev 2- 15-2-32 m, a stipe fragment showing thecae in profile view, d, PU Vg 126, Storeklev 227-230 cm, the specimen commented on by Tjernvik (1956, LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 305 Diagnosis (from Lindholm and Maletz 1989). Dichograptid with bilateral rhabdosome produced by dichotomous division occurring at irregular intervals; second dichotomy in most species consecutive, forming a tetragraptid proximal part, but can be delayed for a couple of thecae; branches diverging proximally, while distally diverging, subparallel, or flexuous; thecal shape variable, unknown in many species assigned to the genus; central disc unknown, secondary development of cortical overgrowth in many species, particularly in proximal parts. Remarks. The genus Clonograptus was treated by Lindholm and Maletz (1989). It was reinterpreted as a form genus ( Clonograptus sensu lato ) consisting of the phylogenetically based subgenus Clonograptus ( Clonograptus ) (= Clonograptus sensu stricto ) and additionally a number of species not known in enough detail for inclusion in any phylogenetically based group. In the same paper Clonograptus was transferred to the Dichograptidae, since there are no bithecae along the stipes in the type species. Clonograptus cf. norvegicus Monsen, 1937 Text-fig. 10f cf. 1937 Clonograptus norvegicus Monsen, pp. 198-200, pi. 20; non pi. 5, fig. 22. cf. 1987 Clonograptus norvegicus Monsen; Maletz, p. 58, fig. 41:1, 2. Material. One incomplete, nearly flattened, specimen with proximal part, associated with scattered stipe fragments of the species, found on PMO 73 . 200 and 73 . 204 (counterparts), in grey shale from 0-5 m above the Ceratopyge Limestone at Bodalen, Slemmestad. Associated species. None. Stratigraphic range. H. copiosus Zone, possibly also T. phyllograptoides Zone. Diagnosis (of C. norvegicus , based on Monsen 1937). A clonograptid, irregularly branching to at least 13 orders of stipes. Second- to fourth-order stipes progressively longer, but within 3-8 mm in length; higher orders on average 10 mm long or more. A marked cortex cover (peridermal film?) gives a lateral width of 3 mm proximally and down to less than 2 mm distally. The cortex obscures all thecal details. Possibly 8-9 thecae in 10 mm. Description. In my specimen, no details of proximal development or thecae are visible due to the cortex cover and the horizontal orientation of the rhabdosome. The position of the sicula can be seen, and the outline of the stipes and branching points within the cortex film can be partly discerned, indicating that the primary stipes are of unequal length. There is probably one theca in one first-order stipe and 2-3 in the other. The thecal spacing is unknown, however, and if it is much less than 10 in 10 mm it could indicate that the specimen is triradiate rather than biradiate. The lateral width of the stipes excluding the cortex cover is 03-06 mm, the total lateral width varies from 15 to 2 0 mm in the proximal part down to less than 1 mm in distal parts. The branching pattern seems to be somewhat irregular, but in general the distances between branchings increase in a distal direction. Seven orders of stipes are seen, and the length of second- to fifth-order stipes are (assuming two primary stipes) 2-6 mm, 3-7 mm, 4-9 mm, and 6-10 mm. The specimen may not have been fully planar; it appears that at least one stipe crosses the others at a lower level in the slab. The branching angles are variable, c. 60-120° in proximal branchings and 45-70° in the more distal parts. Higher-order stipes sometimes curve to adopt a more parallel orientation. pp. 1 17-1 18); a broken lateral stipe (arrow) is seen on the lower left stipe, e, LO 5984t, Grundvik, Slemmestad, piece and mirror image of counterpart shown to illustrate lateral stipes at an angle to the plane formed by the main stipes; the lateral stipes on the upper left are connected to the main stipe by thick cortex proximally. f, LO 5985t, Storeklev 2-60 m, a stipe fragment with paired lateral stipes. Stippling indicates flexure - the lateral stipes were originally directed upwards. 306 PALAEONTOLOGY, VOLUME 34 text-fig. 10. a, c, d, Clonograptus magnus sp. nov; a, holotype PMO 108.564-108.565, Slemmestad; the drawing is a combination of counterparts; c, d. PMO 108.561-108.562, same locality as a; d is an enlargement of the obliquely positioned fragment in c, showing the thecae in profile view; c is combined from piece and counterpart, b, Clonograptus sp. 1, PMO 97.708, Slemmestad; drawing from latex cast. E, Clonograptus sp. 2, LO 5986t, Krapperup core 1 42-46— 1 42-56 m. F, Clonograptus cf. norvegicus , PMO 73-200, 73-204; the drawing is a combination of counterparts. LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 307 Remarks. The fragmentary nature and lack of information on proximal development and thecal shape of the specimen makes identification very difficult. C. norvegicus is a rare species and the original description by Monsen (1937) was based on only one specimen with most details obscured by the cortex cover. That specimen has at least 13 orders of stipes and is more robust than the present specimen, which is unlikely to have been of that size. The robustness may be caused entirely by a thicker cortex cover, associated with its greater size. Monsen’s specimen was said to be associated with ‘ Didymograptus minutus var. pygmaeus \ hence Monsen assigned it to the middle part of the Arenig. However, small pendent specimens very much resembling D. minutus occur also at a level slightly below the base of the T. phyllograptoides Zone (see Text-fig. 14c, F). Monsen’s specimen comes from Grundvik, Slemmestad, one of the localities where the beds below the T. phyllograptoides Zone are easily accessible. Erdtmann (19656, p.496) reported having found fragments of the species at a very low level in the Toyen section. However, his specimens have since been lost and the statement cannot be verified. Maletz (1987) reported the species from Mt Hunneberg, from the localities Tunhem and Storeklev. At the latter locality it was found at the same level as the richest finds of the H. copiosus fauna. Large-sized stipe fragments with thick cortex cover are present also in the T. phyllograptoides Zone at Mt Hunneberg (one specimen from Mossebo, SGU collections) as well as at Galgeberg, Oslo (Bergen Museum, Monsen collection 231). The only other Clonograptus s.l. species of a similar outline is C. trochograptoides Harris and Thomas, 1939. The pattern of branching and cortex cover is identical, but stipes of a given order are somewhat shorter, giving a more compact rhabdosome. C. trochograptoides was said to have thecae of Clonograptus s.s. type, 8-9 in 10 mm. Subgenus clonograptus (clonograptus) Nicholson, 1873 Diagnosis. As for genus, but with thecae straight or slightly curved simple tubes, overlapping one- third to two-thirds of their length; proximal development isograptid, dextral. Remarks. A number of genera were synonymized with Clonograptus ( Clonograptus ) by Lindholm and Maletz (1989), most importantly Temnograptus Nicholson, 1876. Clonograptus ( Clonograptus ) magnus sp. nov. Text-fig. 10 a, c, d Name. Latin magnus , big. Material. One specimen with proximal part preserved (holotype PMO 108.564-108.565; Text-fig. 10a) and one distal stipe fragment (paratype PMO 108.561-108.562) from 0- 5-1-2 m above the Ceratopyge Limestone at Slemmestad crossroads. Additionally there are four stipe fragments from 3-42-3-60 m ( + a gap of unknown extent, probably 0-5-2 m) above the Ceratopyge Limestone at Grundvik, Slemmestad. Associated species. H. copiosus , reclined Tetragraptus indet. (juvenile). Stratigraphic range. H. copiosus Zone, possibly also T. phyllograptoides Zone. Diagnosis. A very robust Clonograptus s.s. with a considerable cortical thickening in mature specimens. Tetragraptid proximal part, second-order stipe length approximately 10-40 mm, third and higher order generally over 40 mm. There are at least five stipe orders. 9—10 thecae in 10 mm, thecal overlap two-thirds or more, profile stipe width 1 -5—2 0 mm. Description. Details of the proximal development are unknown. In the only specimen with proximal part (Text-fig. 10 a), first-order stipes consist of one theca each and second-order stipes are from 10 to 38 mm long. The specimen does not show complete third-order stipes, but the five longest fragments are 38-54 mm long. 308 PALAEONTOLOGY, VOLUME 34 The angle between the second-order stipes is rather high, c. 105°, whereas the angle between the third-order stipes is low, c, 45-70°. There are 9-10 thecae in 10 mm. A third-order stipe is preserved in partial profile view, giving an estimate of profile stipe width of I -5 to 2-0 mm. As far as can be seen the thecae are rather long and narrow. All of this specimen is covered by cortex, giving a lateral stipe width of approximately 4 mm for the first-order stipes, 2-5— 2-8 mm for second-order stipes, and 1 -5-2-3 mm for third-order stipes. The largest stipe fragment (Text-fig. 10c, d) has less cortical cover than the previous specimen. The fragment has four orders of stipes, the most proximal one probably of third or higher order. The length of the two middle orders of stipes is 47 and 77 mm. There are 9-5 thecae in 10 mm. The thecae are slightly curved, with about two-thirds to three-quarters of overlap. The thecal apertures are concave. The profile stipe width is 18 mm and the lateral stipe width is 1-3-1 -5 mm. There are no bithecae. The remaining stipe fragments contain only one dichotomy each. They all have 9-10 thecae in 10 mm, and a profile stipe width of 1 -5—2-0 mm. Remarks. The only Clonograptus species of a comparable size are C. multiplex (Nicholson, 1868) and C. magnificus (Pritchard, 1892). Neither of these have the massive cortical thickening characteristic of this species. Also, its branching angles differ from those of these two species. A stipe fragment from Taimyr, identified by Obut and Sobolevskaya (1962) as Temnograptus aflf. noveboracensis Ruedemann, may be conspecific with C. magnus. Form genus tetragraptus Salter, 1863 (= tetragraptus s.I.) Tetragraptus longus sp. nov. Text-figs 1 1 a-d and 12 Name. Latin longus , long, referring to the length of the stipes. Material. 121 specimens in all, most of them more or less broken. Nearly all of them are from Galgeberg, east- central Oslo (found on PMO 58.969, 58.970); 14 specimens come from Slemmestad (PMO 97.702, 97.708 and one specimen, LR 1, in the Lund collections), and one from the Krapperup core ( 1 29-46—1 29-54 m). The holotype is found on PMO 58.970 (text-fig. 1 1A), the paratypes on PMO 58.969 and 97.708. Associated species. H. copiosus , H. tjernviki, T. cf. krapperupensis. Stratigraphic range. H. copiosus Zone. Diagnosis. A thin-stiped (0-7 1 I mm) horizontal tetragraptid, with small central disc in mature specimens. The divergence angle between second-order stipes is 90° or less. There are 9-5-1 1 thecae in 10 mm. The stipes may become extremely long. Description. The species has a normal tetragraptid proximal part (Text-fig. 1 1 c), i.e. the first-order stipes are composed of one theca each. All specimens are preserved horizontally, and thus do not reveal any details of proximal development or the possible presence of a sicular bitheca. Many stipes are preserved in relief and show total absence of bithecae. The stipes are 0-7-0-9 mm wide in profile view, up to 1 1 mm in very large specimens. The lateral width is about 0-4— 0-5 mm, but the stipes very often show the profile view. The longest stipe fragments encountered were 710 and 680 mm respectively (Text-fig. 12; all specimens on the slab are fragmented - cf. Text-fig. 10b - but there appears to be no tectonic distortion). They probably both belong to one specimen. The thecae are straight tubes, about three times as long as wide and with straight apertures. There are 9-5-1 1 thecae in 10 mm and they overlap for one half of their length or slightly less. The thecal inclination is about 20°. Mature specimens develop a small central disc (Text-fig. 1 1 a, b). The largest one seen is approximately 2 by 4 mm. No more than 2-3 thecae per second-order stipe are encroached upon by the disc. Cortical thickening has not been noticed along the second-order stipes, but is likely to be present to some degree, considering the relative straightness of most stipes seen in Text-figure 12. The second-order stipes normally make an angle of 80-90°, slightly more in a few specimens. The long stipe fragments in Text-figure 12 seem to have been curved by rotational movement during post-mortem descent to the sediment surface. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 309 text-fig. 1 1. a-d, Tetragraptus longus sp. nov., Galgeberg, east-central Oslo, a, holotype, PMO 58.970, with a well-developed central disc; judging from the size of the disc, the stipes must have been very long, b-d, PMO 58.969; b, a specimen with the beginnings of a disc; c, specimen partly preserved in relief in the proximal part, showing a normal tetragraptid branching pattern; d, typical specimen, without central disc, e-h, Gen. et sp. mdet. 1, all from Slemmestad. E, PMO 108.599, a complete immature specimen. F, PMO 108.566, a specimen showing the proximal branching pattern inside the central disc. G, PMO 108.567, a specimen with thecae in profile view and an immature central disc, h, PMO 108 . 599, proximal fragment of a presumably large specimen with a well-developed central disc. Remarks. Thin horizontal tetragraptids are most commonly lumped together under the name T. quadribrachiatus . As was shown by Williams and Stevens (1988), even the type collection of J. Hall contains specimens of two unrelated taxa of different age. Their recommendation was that the name should not be used until the taxon was redefined. T. longus differs from the original description of T. quadribrachiatus in having a central disc in mature specimens. J. Hall (1865) noted that he had never seen one in T. quadribrachiatus. Also, the stipe divergence angles of T. longus are somewhat unusual, being more often below than above 90°. The great length of the stipes is also unique. The number of specimens found might suggest that the species is a common one. This is not the case -it appears to be an invasion species, found covering surfaces in almost monotypic assemblages, and being very rare in intervening beds. 310 PALAEONTOLOGY, VOLUME 34 text-fig. 12. Tetragraptus longus sp. nov., PMO 97.708, Slemmestad. Illustration of part of a very big slab, showing ten (of a total of twelve) specimens with proximal part, and two pairs of very long stipes of the species, possibly both belonging to one specimen. The arrows point in the distal direction, and are placed along the ventral side of the stipe. Associated fauna is not shown. Tetragraptus krapperupensis sp. nov. Text-fig. 13a, c, e; cf. Text-fig. 1 3 f Name. From the Krapperup core. Material. 6 specimens, all from the Krapperup core, three of them at 140 87 m, the other three at 140-30 m. Holotype LO 5988T (140-87 m; Text-fig. 13 c), paratypes LO 5987t and LO 5989t (both from 140-30 m). One specimen of T. cf. krapperupensis (LO 5990t) is present at 1 29-46—1 29-54 m. LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 311 Associated species. A. murrayi , Didymograptus sp. 1. Stratigraphic range. A. murrayi Zone, possibly also H. copiosus Zone. Diagnosis. A three-stiped species of slightly declined habit. Sicular length 2-2— 2-5 mm, stipe width 1-3 mm proxinially and up to 21 mm distally, 9-5-11 thecae in 10 mm, thecal overlap about one half, distal thecal inclination 35^45°. Description. The sicula is 2-2-2- 5 mm long and about O4-0-7 mm wide at the aperture. No specimen shows any details of proximal development. Only stipe1 has a second dichotomy, after the first theca, resulting in 3 final stipes. The thecae are straight or slightly curved, about 2-3 times as long as wide. They are inclined at about 35^45° to the dorsal margin of the stipes. Well-preserved thecae are somewhat denticulate and have a concave aperture. The aperture is generally inclined at 60-70° to the dorsal margin of the stipes. The thecal overlap is mostly difficult to see, but appears to be approximately one half. There are 9-5-1 1 thecae in 10 mm, the lower value being found in the largest specimen. The proximal stipe width is about 1 -3—1 -4 mm, the maximum stipe width varies with the length of the stipes, from T5 mm in small specimens to 2-1 mm in the largest one. Across a specified theca the stipes are somewhat wider in larger specimens, possibly suggesting a certain amount of continued thecal growth. Two specimens (see Text-fig. 13 a) indicate a plaited thecal structure - or triad budding : due to the relatively low relief, the presence or absence of bithecae along the stipes cannot be verified. The species could belong to the transitional forms with only traces remaining of an anisograptid structure (Lindholm and Maletz 1989). The stipes are declined in their proximal part. They are straight throughout or have a slight dorsally concave curvature. In life, the stipes were probably slightly declined. One specimen, found a few metres higher in the core, has somewhat narrower stipes and more closely set thecae (Text-fig. 1 3 f). Until more material of the species is known, I refer to it as T. cf. krapperupensis. Remarks. No four-stiped rhabdosomes have been observed in the beds containing T. krapperupensis and the assignment of the species to the dichograptid form genus Tetragraptus is based on the fact that no bithecae have been identified, and that there are only two orders of dichotomy; there appears to be no other existing dichograptid genus for three-stiped forms. However, since preservation precludes observation of bithecae, these could in fact be present, in which case the species would have to be referred to an anisograptid genus. Triograptus is the only three-stiped anisograptid genus. It has three ‘primary’ stipes, i.e. the second-order dichotomy follows the first without intervening unicalycal theca (Cooper and Fortey 1983), although one specimen (Text-fig. 1 3 h) in the collection forming the basis of Monsen’s (1925) original description of the type species, Triograptus osloensis , appears to have two primary stipes, one of them branching after theca 1, just like the species here described. However, the thecal morphology of Triograptus osloensis (Text-fig. 1 3 g) makes it very unlikely that the two species are closely related. Near the base of the Krapperup core there is one specimen probably belonging to another Triograptus species (Text-fig. 1 3 1). Also this species has a thecal shape quite unlike that of T. krapperupensis. Three-stiped rhabdosomes of roughly the same shape are found also in younger beds in southern Scandinavia. Five specimens were found with the H. copiosus fauna in Slemmestad (PMO 108.566, 108 . 569-108 . 570, 108 . 598; Text-fig. 13 d). The thecal morphology agrees reasonably well with that of T. krapperupensis , but the sicula is much stouter and longer. This form is found together with four-stiped specimens. Three-stiped forms are especially common in the overlying T. phyllo- graptoides Zone (more than 150 specimens from Mt Hunneberg in RM, SGU, and Lund collections have been investigated), where three typical shapes can be seen among the declined to slightly reclined forms. Some additional specimens are preserved horizontally, so that thecal characteristics are obscured. This fauna has not yet been studied in enough detail to see if there is a continuous range of variation among its members or not, but it seems possible that there are distinct forms, some or all of which may be related to four-stiped forms, i.e. merit the name Tetragraptus. One form (Text-fig. 13b, j) is very similar in outline to T. krapperupensis. I hesitate to synonymize them since there are indications of a plaited thecal structure in T. krapperupensis , whereas specimens of the younger fauna have normal dichograptid stipes. Perhaps they formed part of a three-stiped lineage with separate bithecal reduction. Where three- and four-stiped specimens occur in the same beds it 312 PALAEONTOLOGY, VOLUME 34 text-fig. 13. Tetragraptus krapperupensis sp. nov. and comparative material, a, c, e, T. krapperupensis sp. nov., all from the Krapperup core; a, LO 5987t, 140-30 m, the left-hand stipe shows plaited thecal structure; c, holotype, LO 5988T, 140-87 m, the largest specimen; the drawing is a combination of counterparts; e, LO 5989t. 140-30 m, a smaller specimen with narrower maximal width, b, j, Tetragraptus sp. 1, Mossebo, Mt LINDHOLM: SCANDINAVIAN ORDOVICIAN OR APTOLITES 313 is often difficult to see if they are conspecific since the proximal parts of the four-stiped specimens tend to be preserved horizontally (dorsoventral view), so that the sicula and proximal width of stipes etc. cannot be seen. Three-stiped rhabdosomes of various species appear to be present through most of the Arenig of southern Scandinavia. In the T. phyllograptoides Zone, in addition to the forms discussed above, there is a pendent three-stiped form which has no associated four-stiped pendent specimens. According to S. H. Williams (pers. comm.) it is identical to P. cf. pendens from Newfoundland (Williams and Stevens 1988). The Newfoundland fauna, however, contains both three- and four- stiped specimens. In beds above the T. phyllograptoides Zone in Scandinavia, practically all three- stiped forms have reclined rhabdosomes. Some three-stiped forms from the T. phyllograptoides Zone at Mt Hunneberg were described by Maletz (1987) under the name of T. triograptoides (nomen nudum; junior homonym of T. triograptoides Harris and Thomas, 1938), but had been observed already by Tornquist (1904, pi. 1, fig. 20), who grouped them with four-stiped forms as T. serra ( = T. amii according to current usage). Non-triograptid, more or less horizontal, three-stiped forms of Tremadoc-Arenig age are known also from other areas. Tetragraptus otagoensis and T. decipiens (three-stiped form) from New Zealand were shown by Bulnran and Cooper (1969) to have the same branching pattern as the Scandinavian forms. T. otagoensis is of La 2 zone age, and is therefore roughly coeval with T. krapperupensis , but has considerably narrower stipes than the latter. The three-stiped form of T. decipiens is somewhat younger. La 3 zone, approximately coeval with the T. phyllograptoides Zone fauna of Mt Hunneberg. It appears to have stipes narrower than the mature Scandinavian specimens, but it is worth noting that immature Scandinavian specimens, with stipes of comparable length to that of the New Zealand specimens, also have a comparable stipe width. As in the Scandinavian specimens, the second-order dichotomy in the New Zealand specimens is based on stipe1 (the stipe developed on the th l1 side), quoted erroneously (R. A. Cooper pers. comm.) by Bulman and Cooper (1969) and Cooper (1979) as the stipe2 side. The three-stiped form of T. decipiens has not yet been reported from Australia (R. A. Cooper pers. comm.). Harris and Thomas (1938) described Tetragraptus triograptoides from the lowermost part of the Bendigonian of Victoria. This is a very slender form, belonging to the sigmagraptines, judging by its thecal characters. Chen el al. (1983) reported a specimen of a three-stiped extensiform species, Adelograptus rohustus , from Jiangxi, South China, associated with T. approximate . Its dimensions, apart from the comparatively broad proximal part of the stipes, are not far from those of certain specimens found at Mt Hunneberg, but it has bithecae along the stipes. A probably middle Arenig form was described from Czechoslovakia (T. postlethwaitii\ Kraft 1987). It resembles the form illustrated in Text-figure 13b except in having slightly narrower stipes. The species contains both three- and four-stiped forms. Form genus didymograptus M‘Coy, 1851 (= didymograptus sd.) Didymograptus sp. 1 Text-fig. 14a, b v cf. 1986 Corymbograptus sp. 1 Gutierrez Marco, pp. 445-447, text-fig. 39d-m; pi. 14, figs 2, 4, 5. v cf. 1988 Didymograptus cf. sinensis Lee and Chen; Molyneux and Rushton, p. 66, fig. 9a, b. Hunneberg; b, SGU Type 8020; J, RM Cn 1838, the biggest specimen found, d, Tetragraptus sp. 2, PMO 108.569-108.570, Slemmestad; the drawing is a combination of counterparts, f, T. cf. krapperupensis , LO 5990t, Krapperup core 1 29-46—129-54 m. g-h, Triograptus osloensis Monsen, both on PMO 59.215, Stensberggaten, central Oslo, Ceratopyge Shale, 155-180 cm below the Ceratopyge Limestone; G, part of a stipe fragment showing shape of thecae; h, an aberrant specimen with two primary stipes and a second-order dichotomy, i, Triograptusl sp. 1, LO 60 1 5t, Krapperup core, 1 51-45-151 -46 m. 314 PALAEONTOLOGY, VOLUME 34 text-fig. 14. a, b, Didymogrciptus sp. 1, Krapperup core 137-72-1 37-76 m, two associated specimens; a, LO 599 It, the largest specimen found, slightly deflexed; B, LO 5992t, a seemingly declined specimen, c, f, Didymogrciptus sp. 2, Krapperup core, 1 18-50-1 18-54 m, two associated specimens; c, LO 5993t; f, LO 5994t. d, E, G, H, Tetragraptus phyllograptoides from near the base of the range of the species; all have considerably narrower stipes than the typical form; from two localities in the Slemmestad area; d, LO 5995t, immature Phyllograptus-hke specimen, a-b pair; dotted circle indicates the broken connection towards the sicula; e, LO 5996t, a probably three-stiped specimen ; the drawing is a combination of counterparts ; G, LO 5997t, the largest specimen found, a-b pair; as in d, the connection towards the sicula is broken; h, LO 5998t, a probably three- stiped specimen showing few thecae in the conjoined part of the stipes, i, Tetragraptus sp. 3, PMO 1 12.967, 112.969, Slemmestad, associated with a Hunnegraptus fauna; a specimen in medium relief with the proximal part preserved as a mould; sicular outline drawn from counterpart. Material. 10 specimens from the interval 147-33-137-70 m of the Krapperup core. Most specimens are small, showing no more than 4 thecae per stipe. Associated species. A. murrayi , T. krapperupensis. Stratigraphic range. A. murrayi Zone, and possibly higher beds. Diagnosis. A thin (c. 07-0-8 mm) deflexed to declined didymograptid with sicular length about 1-4 mm and around 12 thecae in 10 mm. Thecae straight, inclined at c. 30°. Description. All specimens are too flattened to show any details of proximal development or the possible presence of bithecae. The sicula is straight, 1-2-1 -6 mm long and about 0-3 mm wide at the aperture. It protrudes about 07-0-9 mm above the dorsal margin of the rhabdosome. The thecae are almost straight tubes, inclined at about 30°. Their apertures are straight or slightly concave, inclined at 70-80 ° to the dorsal margin of the rhabdosome. There are normally 12 thecae in 10 mm, but the total variation seen is 11-14. The stipes are 0-6-0-7 mm wide proximally, widening to about 0-8 mm or, rarely, 1-Omrn distally. The shape of the rhabdosome is slightly deflexed or declined, with a stipe divergence angle of 120-145 °. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 315 Remarks. The assignment of this species to Didymograptus s.l. is based on the fact that no bithecae have been seen along the stipes. Considering the age of the fauna, however, it is possible that bithecae are present, but not detectable due to the flattened state of the rhabdosomes. If so, the species will have to be referred to Kiaerograptus. The general shape of the rhabdosome of this species is common to several species throughout higher parts of the Arenig and even the lower part of the Llanvirn, but the only similar species of roughly the same age are Corymbograptus sp. 1 Gutierrez Marco, 1986, and Didymograptus cf. sinensis described by Molyneux and Rushton (1988). Corymbograptus sp. 1 differs only in having a somewhat shorter sicula and narrower proximal width, as well as a more accentuated deflexed shape in some of the specimens. D. cf. sinensis is, in my opinion, a synonym of that species, differing only in having slightly more thecae in 10 mm, well within the normal limits of variation of a species of that age. There are only a couple of reasonably large Swedish specimens, not necessarily representative of the mean of the population, hence further finds may prove the Swedish species to be conspecific with the Spanish/English one. Gen. et sp. indet. 1 Text-fig. 1 1 e-h Material. 4 specimens from Slemmestad crossroads, c. 0-5-1 -0 m above the Ceratopyge Limestone, found on PMO 108.566, 108.567 and 108.599. Associated species. H. copiosus , Clonograptus s.s. sp. indet., 3-stiped extensiform tetragraptids, P. tenuis , P. cf. rarus. Stratigraphic range. H. copiosus Zone. Diagnosis. A five-stiped rhabdosome, with a tetragraptid proximal part. Stipe width c. 1-6 mm, 9-5-1 1 thecae in 10 mm. A central disc is found in mature specimens. Description. All four specimens are preserved horizontally, obscuring details of proximal morphology. The proximal part is tetragraptid, however, with one theca per first-order stipe. A well-developed central disc is found in the most mature specimens. The two largest discs measure 2 by 4 and 2 by 5 mm respectively (Text- fig. 1 1 f, h). The only specimen showing thecae in profile view (Text-fig. 1 1 G), has faint beginnings of a disc. The specimen is preserved in low relief and shows possible plaited thecal structure along one stipe (alternatively, it represents a compression structure). A characteristic of the species is that one of the second- order stipes divides consecutively. There are no indications in the available material of any further dichotomies, and it may be presumed that the final number of stipes is five, especially since there seems to be a certain amount of readjusting of the stipe angles, to even out the distances between the stipes. However, this is unfortunately difficult to prove due to the fragmentary state of the rhabdosomes. The thecae are slightly curved and somewhat expanded tubes, and slightly denticulate. The thecal overlap is about 60% and there are 9-5-1 1 thecae in 10 mm. The profile stipe width is 1-6 mm, the lateral width 0- 5-0-7 mm. Remarks. The three orders of stipes present in these specimens would suggest the use of the genus name Dichograptus. A reduction of the final stipe number is known to occur within Dichograptus. However, for the following reasons I prefer not to assign these specimens to that genus. Firstly, the Slemmestad form is older than any reported species of the genus and, secondly, I have not come across any definite Dichograptus specimens from the lower Arenig of Scandinavia. Additionally Dichograptus (as well as Tetragraptus) is likely to be a form genus and its type species, D. sedgwicki , has never been properly described (see Salter 1863; Elies and Wood 1902) and apparently has no associated fauna confirming its age. It was referred to as a subspecies of D. octobrachiatus by Elies and Wood (1902). Compared with D. octobrachiatus , the present form is much less robust and, due to the lack of details known from the proximal part of either taxon, their phylogenetic relationship is unclear. The beds containing my five-stiped form are of an age when great changes took place in 316 PALAEONTOLOGY, VOLUME 34 the graptolite fauna, including not only the loss of bithecae along several lineages but, as far as I have seen, also an instability in the number of stipes present in a specimen. It thus seems more probable to me that the present form has a derivation separate from that of the later Dichograptus species. Family sinograptidae Mu, 1957 Subfamily sigmagraptinae Cooper and Fortey, 1982 Diagnosis (from Fortey and Cooper 1986): Dichograptinids with sigmagraptine proximal region. Remarks. The Sigmagraptinae was originally described as a subfamily of the Dichograptidae with included species united by the characteristic proximal part and generally slender thecae. The taxon was raised to family rank by Fortey and Cooper (1986), consisting of the nominate subfamily only. Williams and Stevens (1988) lowered the rank back to subfamily level, and included it in the family Sinograptidae. I follow the classification of Williams and Stevens (1988), and also their concept of the content of the family (sinograptines, sigmagraptines, and the previously ‘obscure' Kinnegraptus). The use of the name Sinograptidae follows priority rules, even though a sigmagraptid is the ancestor of the sinograptines. The kinnegraptids were raised to family rank by Mu (1974) and were used at this level for Paradelograptus by Erdtmann et al. (1987). Also Williams and Stevens (1988), though temporarily including them in the Sigmagraptinae, considered the possibility that further study might show that the kinnegraptids merit family rank. However, my own investigations have shown them to be very close to the main stock of sigmagraptines. Acrograptus gracilis has an equally prolonged prosicula, and the exaggerated apertural lip (rutellum ; Williams and Stevens 1 988) of thecae and sicula is found among some of the Paradelograptus species, as well as in some specimens of A. tenellus (Hutt 1974, fig. 8a), the species which must be considered the best candidate for an ancestor of the Sigmagraptidae. Genus paradelograptus Erdtmann, Maletz and Gutierrez Marco, 1987 Diagnosis. See Erdtmann et al. (1987). The most important features mentioned are biradiality, irregular dichotomies, isograptid development, asymmetrical proximal part, and a characteristic thecal shape with long thin prothecae and expanding metathecae, sometimes provided with ‘lappets' [here meaning ventral prolongation]. Bithecae were not observed. Remarks. When defined by Erdtmann et at. (1987), the genus Paradelograptus was referred to the family Kinnegraptidae Mu, 1974. However, the authors base this family only on the shape of the thecae, disregarding features of the proximal end (1987, p. 113): ‘This character [shape of the proximal part], however, is not a discriminating factor for Paradelograptus alone nor for the Kinnegraptidae and Sigmagraptinae [of the Dichograptidae], as was suggested by Cooper and Fortey (1982, p. 259), but it is observed quite frequently in many other dichograptids, dating back to the ancestral Adelograptus tenellus (Hutt, 1974, fig. 5 b, Maletz and Erdtmann 1987) and to other adelograptinid forms (i.e. to Choristograptus Legrand, 1964). Therefore , no taxonomic significance may be attached to this feature alone [my italics] '. With this statement I disagree. In my opinion, they have defined a group of genetically related taxa. Further, figure 2 of Erdtmann et al. ( 1 987), showing the ‘phyletic relations' of taxa, disagrees with the text. The text states that the concepts of Cooper and Fortey ( 1982) have been used, that is, that Sigmagraptinae is a subunit of Dichograptidae. The figure, on the other hand, shows a possibly diphy letic Sigmagraptinae branching off the Kinnegraptidae, and possibly also the Clonograptinae. Paradelograptus differs from Adelograptus solely in the absence of bithecae along the stipes. It includes both two-stiped and multi-stiped taxa. Among the Paradelograptus species described by Erdtmann et al. (1987) the proximal development is known only for the type species, P. onubensis. LINDHOLM: SCANDINAVIAN ORDOVICIAN G R APTOLITES 317 It is quite possible that the genus Paradelograptus , with the constituent species as given by Erdtmann et al. (1987, p. 115; 15 species in all, those mentioned below and P. sedecimus , P. ranis , P. smithi , P. ramulosus , P. chapmani , /\ ? tenuiramis, P. ? clarkefieldi, P. ? bulmani, C. tenellus var. problematica Harris and Thomas, and C. tenellus sd. Cooper and Steward), is a polyphyletic assemblage of similar-looking forms, which have responded in a similar way to peculiarities of the environment. However, the external shape of the proximal parts, with an adelograptid type of sicula, in P. onubensis , P. antiquus , P. pritchardi and P. mosseboensis , and P. elongatus and P. tenuis described here, is so similar as to make it likely that at least this group is monophyletic. P. kinnegraptoides appears from illustrations not to have an adelograptid sicula. The proximal development of P. smithi was not seen in the specimens from Mt Hunneberg, and the inclusion of that species by Erdtmann et al. (1987) seems to be based on thecal morphology alone. Two new species are described here. A number of other species present in the Scandinavian Lower Hunneberg beds are mentioned under ‘Other species’ (p. 320). Paradelograptus elongatus sp. nov. Text-fig. 15 c, g-i Name. Latin elongatus , elongated, referring to the long first-order stipes. Material. 16 specimens, 15 of which come from Slemmestad (14 of them on PMO 108.568-108.570, Slemmestad crossroads; 1 specimen, LR 2, from the base of the T. phyllograptoides Zone at Hagastrand, Lund collections). One specimen was found in the Krapperup core (124-87-1 24-89 m). A questionable specimen was found at Storeklev (2-32 nr, Lund collections). Both the holotype (Text-fig. 1 5 1 ; the only mature specimen) and the paratypes are found on PMO 108.570. Associated species. A. murrayi, H. copiosus , P. antiquus. Stratigraphic range. H. copiosus Zone, and at least the basal beds of the T. phyllograptoides Zone. Diagnosis. Biradiate, declined to pendent in profile view, branching dichotomously at irregular intervals; first-order stipes consist of more than one theca. Proximal development probably isograptid; both sinistral and dextral forms occur. A sicular bitheca has been observed. Metathecae somewhat Hared, but less so than in P. mosseboensis. Dimensions close to the latter. Description. The sicula is straight and tube-like, 1 -9-2-0 mm long and 03-0-4 mm wide at the aperture. Theca 1 1 originates close to the apex of the sicula. The development is probably isograptid, the prothecal part of th 21 is seen in the specimen in Text-figure 15g, and can be traced back almost to the point of origin of theca l2. A sicular bitheca is present on the obverse side of the sicula (Text-fig. 15 c). There are both dextral and sinistral forms. The thecae are 2 mm long or longer, have relatively thin prothecal parts and somewhat flaring metathecae, which are sometimes seen to have a short denticle. Thecal width at the apertures reaches 0-4— 0-5 mm. There are 8-9 thecae in 10 mm and the thecal overlap is about 40-50% (the point of origin of thecae is commonly obscure). The profile stipe width is 0-6-0-7 mm in proximal parts, up to 0-9 mm in distal parts. The first-order stipes vary in attitude from almost horizontal to pendent, if seen in profile view. The first dichotomy occurs at th 3 or later, sometimes considerably later: in one unbranched specimen, first-order stipes have 8 and 14 thecae. The holotype is a mature specimen showing five orders of stipes, the greatest number known. There is a considerable amount of cortical strengthening of proximal stipes in the holotype, giving it a much more rigid appearance than the associated smaller specimens. No bithecae have been observed along the stipes, but the material is mostly of low to no relief. Remarks. P. elongatus most closely resembles P. mosseboensis , which occurs at a considerably higher level, around the lower boundary of the D. balticus Zone at Diabasbrottet, Hunneberg (Erdtmann et al. 1987, fig. 1). The dimensions of the two species are virtually the same, but the ventral thecal processes are less pronounced in P. elongatus. The latter is widely variable in the position of the second-order branching, whereas the corresponding range for P. mosseboensis was 318 PALAEONTOLOGY, VOLUME 34 text-fig. 15. For legend see opposite. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 319 indicated as theca 2-3 (Erdtmann et al. 1987, p. 120). This range might be considerably wider, since the species was only represented by three incomplete specimens. The holotype of P. elongatus very much resembles that of P. kinnegraptoides in general shape, but not in size and thecal details. Paradelograptus tenuis sp. nov. Text-fig. 1 5 a— b, D— F ? 1979 Clonograptus tenellus Linnarsson s./.\ Cooper and Stewart, pp. 785-786, text-fig. 8 m. Name. Latin tenuis , thin, small-sized. Material. 15 specimens, 13 of which come from various localities in the Slemmestad area (12 of them, some with counterpart, are found on PMO 73.188-73.189, 73.191-73.192, 108.557-108.560, 108.566, 112. 966-1 1 2 . 969, 113.031; one specimen, LR 3, from the base of the T. phyllograptoides Zone at Hagastrand, Lund collections). The other two specimens are from Storeklev (TUB HUN-S/2. 18-2.3/006 + 030 and 036). Two additional questionable specimens from Storeklev were found on PU Vg 127 and one slab in the Lund collections (from 2-32 m). The holotype is a mature specimen on PMO 108.566 (Text-fig 15a), the four paratypes, all illustrated in Text-figure 15, are found on PMO 108.557 108.559. Associated species. A. murrayi , H. copiosus , C. alT. multiplex , Isograptus sp., horizontal tetragraptids (‘quadribrachiatus’-type). Stratigraphic range. H. copiosus Zone and at least the basal beds of the T. phyllograptoides Zone. Diagnosis. A small, thin paradelograptid with tetragraptid proximal part and frequent branchings. Sicula 1 -6-1 -9 mm long, 7-5-8 thecae in 10 mm, profile stipe width 0-5-0-65 mm, lateral width 0-2 mm or more. Description. The sicula is of general paradelograptid shape, 1-6-1 -9 mm long and 0-3-0-4 mm wide at the aperture. The two stipes diverge from the sicula at different levels, stipe1 at sicular mid-length or slightly closer to the aperture, stipe2 leaving 0-2 mm or less protruding on the ventral side of the stipe. The first-order stipes consist of one theca each (resulting in a tetragraptid proximal plan), the second-order ones of 1-3 thecae. The following orders each get a little longer, but in general aspect, the rhabdosome is very thin-stiped and rather densely branching. Six orders of stipes were found in the largest specimen. The thecae have very low inclination, their ventral margins are concave, and they are denticulate. The apertural margins are straight to markedly concave, making an angle of 90° or more with the dorsal margin of the stipes. The profile stipe width is 0-5-0-65 mm and there are 7-5—8 thecae in 10 mm. The amount of thecal overlap could not be determined. The lateral stipe width is variable, 0 2-0-6 mm, depending above all on the amount of cortex overgrowth. A noticeable amount of cortex cover is only found in the most mature specimen where, due to slight pyritization, the outline of the stipes can be traced inside the cortex. No specimen was well enough preserved to verify presence or absence of bithecae. Badly preserved stipe fragments appear as thin branching ‘threads’ with no thecae visible. Remarks. The size and shape of sicula and thecae are very close to those of P. elongatus (see Text- fig. 15). However, the two species differ in branching density and the position of the second dichotomy (tetragraptid proximal part only in P. tenuis). text-fig. 15. a-b, d-f, Paradelograptus tenuis sp. nov., Slemmestad; a, holotype, PMO 108.566, the largest specimen, with considerable cortical thickening; b, PMO 108.557-108.558, horizontally preserved specimen; the drawing is a combination of counterparts; d, PMO 108.559, immature specimen with (secondarily?) pendent proximal part; e, PMO 108.557; f, PMO 108.559, combination of counterparts, c, G-i, Paradelograptus elongatus sp. nov., Slemmestad, all on PMO 108.570; c, specimen showing presumed sicular bitheca; G, H, specimens showing variation in proximal stipe attitude; i, holotype, the only mature specimen found; the sicula points downwards into the sediment, the two shortest second-order stipes point slightly upwards. 320 PALAEONTOLOGY, VOLUME 34 The general aspect of the species is very close to that of C. tenellus s.l. sensu Cooper and Stewart, 1979, from the La 2 zone of Victoria, Australia. The distal stipe width of P. tenuis is somewhat broader and, also, no cortical thickening was mentioned for C. tenellus s.l. There is a certain resemblance in shape also to Adelograptus altus Williams and Stevens (1991) but, due to the indifferent preservation of the mature specimens of that species and the generalized shape of the proximal part (similar to P. elongatus , P. tenuis and probably other species), no closer comparison can be made. Other species In the Early Hunneberg fauna there are several species in addition to the ones described above. Here, they are only briefly discussed as some of them are quite well known from other areas, others are hard to identify due to a fragmentary preservation, and still others are very rare and are not diagnostic of the fauna, e.g. horizontal tetragraptids. Clonograptids s.l. Several specimens of a thin-stiped Clonograptus species with slightly prolonged first-order stipes (Text-fig. 10b) are present on PMO 97.708, from the Slemmestad area, together with Tetragraptus longus sp. nov. The species is very similar in outline to Clonograptus rigidus , but it appears to be thinner and the thecae are not well enough preserved for a definite identification. Clonograptus ( Clonograptus ) aff. multiplex (PMO 108.557-108.559) occurs at Slemmestad and in the T. phyllograptoides Zone of Mt Hunneberg. It was described by Lindholm and Maletz (1989). Robust stipe fragments probably belonging to Clonograptus s.s. (Text-fig. 1 0 e) occur at a couple of levels low in the Krapperup core. A cortex-covered specimen was found in the Slemmestad area (PMO 113.032). It may belong to Clonograptus norvegicus. Horizontal tetragraptids. These are extremely rare below the T. phyllograptoides Zone in the Krapperup succession: two very badly preserved specimens of tetragraptid outline ('quadribrachiatus’-type) were found at 152-89-93 m (LR 4-5), a level which probably equals a very early Hunneberg age. The longest stipe of the larger specimen is 15 mm. No thecal details are visible in this specimen, but a stipe of the other specimen, preserved in relief, seems to show a plaited thecal structure. Apart from this, only a possible immature specimen was found at 131 -70—1 31-73 m. In the Storeklev section at Mt Hunneberg I have not found any tetragraptids. However, two specimens of ‘ Eotetragraptus sp. 1 ’ were reported by Maletz (1987; the stratigraphic level most likely corresponds to low T. phyllograptoides Zone). Thecal characters were not observable. Tetragraptids are somewhat more frequent along with the Hunnegraptus fauna of Norway. Six specimens of varying stipe width have been found in the Slemmestad area (PMO 108.560, 108.569+ 108.570, 112.968, 1 12.969, 120.751, and one specimen in the Lund collections). Text-figure 1 4 1 illustrates the broadest specimen found. It is in moderate relief, but the irregularities seen in the lower right of the figure are hard to interpret: do they represent a plaited thecal structure or merely the effects of compression? A 3 mm wide stipe fragment of tetragraptid appearance was found on PMO 108.599. Pendent didymograptids. A small pendent (or immature deflexed?) didymograptid species (Text-fig. 14c, f) has been found in the highest beds of the early Hunneberg fauna, just below the base of the T. phyllograptoides Zone. It is the most diagnostic species of this interval. It occurs at 1 18.54-1 12.75 m in the Krapperup core (47 specimens; the majority of them associated on a couple of surfaces and too badly preserved to form the basis of a description), and a few specimens were also found both at Mt Hunneberg and in the Slemmestad area. Isograptids. Primitive isograptids have been found in the H. copiosus Zone in the Slemmestad area. These are apparently the oldest isograptids found anywhere. They will be described in a separate paper (Lindholm in prep.). The oldest specimens have isograptid symmetry, but much less reclined stipes than the majority of isograptids. They also possess a sicular bitheca. An additional specimen was found at 125.67-69 m of the Krapperup core. Paradelograptids (Erdtmann et al. 1987). Par adelograptus is represented by several species, especially in the Krapperup core, see Text-figure 16. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 321 text-fig. 16. Paradelograptus species present in the topmost Tremadoc and lower Hunneberg of Scandinavia, but not described in this paper, a, d, P. pritchardi (T. S. Hall), two associated specimens, Krapperup core 1 34-20— 1 34-26 m; a, LO 5999t ; d, LO 6000t, the stipes are slightly twisted, b, P. antiquus (T. S. Hall), LO 6301 1, Storeklev. c, P. sp. I, PMO 1 12.967, Slemmestad. e, P. onubensis Erdtmann, Maletz and Gutierrez Marco, LO 600 1 1, Krapperup core 151-46—151-50 m. F, P. sp. 2, LO 6002t, Krapperup core, 153 20-153-29 m. G, h, P. sp. 3, LO 6003t and 6004t, Krapperup core 151-96-451-99 m; the specimens are associated, i, P. cf. rarus (Harris and Thomas), LO 6302t, Storeklev. j, k, P. sp. 4; j, LO 6005t, Krapperup core 141-75 m; k, LO 6006t, Krapperup core 150-71-150-75 m. l, P. sp. 5, LO 6007t, Krapperup core 1 53-20 1 53-29 m. m, n, P. sp. 6; m, LO 6008t, Krapperup core 148-79 m; n, LO 6009t, Krapperup core 1 50-7 1—1 50-75 m. One specimen of P. onubensis was found at 151 -46—1 5 1-50 m. It is rather immature, with only two stipes, but the shape of the proximal part is unmistakable (Text-fig. 1 6 e). A second specimen was found at 1 1 1 40-1 1 1 45 m. P. pritchardi ( Text-fig. 16 a, d) occurs in the 135 09 1 18-50 m interval. P. antiquus (Text-fig. 16 b) was found between 1 17-88 and 114-17 m of the Krapperup core. The species is also represented at Storeklev (LO 630 It, TUB HUN-S/2. 18-2.3/023, PU Vg 124, and one specimen in the Lund collections), at Toyen (GP1 T4, T6), and at Slemmestad (PMO 108.568. 108.572, one specimen in the Lund collections). Stipe fragments indistinguishable from P. rarus (Text-fig. 1 6 1 ) were found at Storeklev (LO 6302t and counterpart ; one specimen in the Lund collections) and Slemmestad (PMO 108 . 567, 109 . 148). Stipes probably belonging to the same species are not uncommon, but the thecal outline needed for identification is seldom seen. In addition to these species, unidentifiable proximal parts and stipe fragments occur in the Krapperup core 322 PALAEONTOLOGY, VOLUME 34 text-fig. 17. Graptoloidea indet. spp. Relief specimens, all (and more) associated on one surface, near the base of the A. murrayi Zone, Krapperup core 1 47-66—147-72 m, LO 60 1 Ot— 60 1 4t. a-d show the sicular bitheca; d shows a bitheca in stipe 1. e could be interpreted as triradiate. c was drawn from a latex cast; e was combined from both counterparts. All illustrations were made under vertical light. in the interval 1 53-29— 1 34-20 m. Some of them are illustrated in Text-figure 16f-h, j-n. One of the species is minute - its sicula is only 0-3 mm long (Text-fig. 16 m, n). Relief specimens of unknown affinity. Text-figures 17 and 18 show some of the immature relief specimens of various kinds that have been found at two levels in the Krapperup core, 1 47-66—147-72 m and 1 32-63—1 32-66 m, i.e. close to the bases of the A. mwravi and H. copiosus Zones. At both levels all specimens in obverse view show a sicular bitheca. The stipes are mostly too incomplete to show presence or absence of bithecae. Specimens with bithecate stipes are present at the lower level (Text-fig. 17d), as well as possibly triradiate forms (Text-fig. 1 7 h). The specimens are of extensiform, declined, and pendent types. Because stipes are incomplete, it is also difficult to say how many branchings the mature specimens would have had, but some of the pendent forms may belong to A. murrayi (Text-fig. 18c). Kiaerograptids or didymograptids? The lower part of the Krapperup core, mainly below the level of the Hunnegraptus fauna, contains several badly preserved specimens that are declined to deflexed. They are seldom very big, mostly containing 5 thecae per stipe or less, but it seems unlikely that they would have branched further, had they lived longer. Because of their flatness (bithecae undetectable) and the short stipes (immaturity), their identity is uncertain. The earliest T. phyllograptoides. This species does not belong in the fauna under discussion, but is present in the succeeding T. phyllograptoides Zone. It appears right at the base of its zone in Slemmestad, but some of the earliest specimens found there, in the lowermost metre, deviate from the typical form described by Cooper and Lindholm (1985). As seen in Text-figure 14e, h, some of the specimens could be three-stiped. Preparation gave no evidence of a fourth stipe. The atypical specimens also differ in having considerably narrower stipes ( T3-T6 mm) and fewer thecae (2-4) in the conjoined part of the stipes. Only one specimen with normal width of stipes was found in the lowermost horizon at Grundvik, Slemmestad. Some specimens also have slightly less strongly reclined stipes. So far, only 13 specimens (all belonging to the Lund collections) have been found this low, at three different localities, in the Slemmestad area: 3 specimens from 6-25 m above the Ceratopyge Limestone at Hagastrand; 8 specimens from 2-50 m above the missing part at Grundvik ( c . 10 cm higher than Hagastrand) and one specimen 12 cm higher; finally one specimen from about 80 cm higher than the lowest Grundvik level at the Rortunet section. In the Krapperup core, some weakly reclined tetragraptids are found a couple of metres below the first find of T. phyllograptoides. They are mostly very short-stiped, and no species identification has been attempted. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 323 text-fig. 18. Relief specimens, all (and more) associated on one surface, near the base of the H. copiosus Zone, Krapperup core, 1 32-63-1 32-66 m, LO 6085t-6095t. a, b, d, e, g, i, k, Graptoloidea indet. spp; b shows a typical symmetrical pair formed by the sicula and th l1, with the bitheca in the centre; e has no bithecae along the stipes; i shows th l1 growing around the sicula; in k, the proximal part is a mould; all specimens in obverse view show a sicular bitheca, c, 'lAraneograptus murrayi (J. Hall), the stipes are of dichograptid type, f, j, H. copiosus sp. nov. h, cf. H. copiosus., the apex of the sicula points somewhat downwards. J and k were made from latex casts, c and e from combinations of counterparts. All illustrations were made under vertical light. 324 PALAEONTOLOGY, VOLUME 34 Acknowledgements . I thank Roger Cooper and Anita Lofgren for valuable discussions and linguistic help, and Gerhard Regnell for advice on Latin names. David Bruton, Adrian Rushton and Jorg Maletz made useful comments on the manuscript. Nils Spjeldnaes and Bernd-Dietrich Erdtmann kindly made their collections available for study, and Henry Williams gave me access to a manuscript prior to publication. The following have assisted in the loan of specimens: David Bruton and Gunnar Henningsmoen (PMO), Bjorn Neuman (Bergen), Valdar Jaanusson (RM), Sven Laufeld and Sven-Ola Nilsson (SGU), and Solveig Stuenes (PU). Jorg Maletz provided the photograph. I also want to thank all my field assistants through the years. Financial help has been given by the Swedish Natural Science Research Council, NFR (project ‘Early Ordovician Biostratigraphy’), Lunds Geologiska Faltklubb, and Gyllenstiernska Krapperupsstiftelsen. REFERENCES barnhs, c. r., johnston, d. i., pohler, s. l. m. and williams, s. h. 1988. Lower Ordovician chronostratigraphy ; global stratotype sections in the Cow Head Group, Western Newfoundland. 6. In williams, s. h. and barnes, c. R. (eds). Fifth International Symposium on the Ordovician System , Program and Abstracts. St John’s, Canada, 117 pp. berry, w. b. n. 1960. Graptolite faunas of the Marathon region. West Texas. University of Texas Publication , 6005. 179 pp. bockelie, j. f. 1982. The Ordovician of Oslo-Asker. 106-121. In bruton, d. l. and williams, s. h. (eds). Field excursion guide , IV International Symposium on the Ordovician System. Paleontological Contributions from the University of Oslo, 279, 217 pp. brogger. w. c. 1882. Die Silurischen Etagen 2 und 3 im Kristianiagebiet und aufEker, ihre Gliederung, Fossilien , Schichtenstorungen und Contactmetamorphosen. Universitatsprogramm 2. Semester 1882. A. W. Brogger, Kristiania [Oslo], 376 pp, 12 pis, 1 map. bulman, o. m. b. 1950. Graptolites from the Dictyonema shales of Quebec. Quarterly Journal of the Geological Society of London, 106. 63-99, pis 4-8. — 1954. The graptolite fauna of the Dictyonema shales of the Oslo Region. Norsk geologisk Tidsskrift, 33, 1-40, pis 1-8. — 1970. Graptolithina. In teichert, c. (ed. ). Treatise on invertebrate paleontology , Part V. 2nd Edition. Geological Society of America and University of Kansas Press. Boulder, Colorado, and Lawrence, Kansas, i-xxxii + 163 pp. — and cooper, r. a. 1969. On the supposed occurrence of Triograptus in New Zealand. Transactions of the Royal Society of New Zealand , 6, 213-218. chen xu, yang da-quan, han nai-ren and li luo-zhao. 1983. Graptolites from the Tetragraptus ( Etagraptus) approximate Zone of the lowermost Ningkuo Formation in Yushan, NE Jiangxi. Acta Palaeontologica Sinica, 22, 324-330, pi. I . cooper, r. a. 1979. Ordovician geology and graptolite faunas of the Aorangi Mine area, north-west Nelson, New Zealand. New Zealand Geological Survey Paleontological Bulletin, 47, 1-127, pis 1-19. — and fortey, R. a. 1982. The Ordovician graptolites of Spitsbergen. Bulletin of the British Museum (Natural History ), (Geology), 36, 157-302, pis 1-6. — 1983. Development of the graptoloid rhabdosome. Alcheringa, 7, 201-221. — and lindholm, K. 1985. The phylogenetic relationships of the graptolites Tetragraptus phyllograptoides and Pseudophyllograptus cor. Geologiska Fbreningens i Stockholm Forhandlingar , 106. 279-291. - 1990. A precise worldwide correlation of early Ordovician graptolite sequences. Geological Magazine, 127, 497-525. — and stewart, i. r. 1979. The Tremadoc graptolite sequence of Lancefield, Victoria. Palaeontology, 22, 767-797. destombes, j., sougy, j. and willefert, s. 1969. Revision et decouvertes paleontologiques (brachiopodes, trilobites, et graptolites) dans le Cambro-Ordovicien du Zemmour (Mauritanie septentrionale). Bulletin de la Societe geologicpie de France, (7), 11, 185-206. elles, g. l and wood, e. m. r. 1902. A monograph of British graptolites. Part 2. Palaeontograpliical Society Monograph, pp. i-xxviii, 55-102, pis 5-13. erdtmann, b.-d. 1965a. Eine spat-tremadocische Graptolithenfauna von Toyen in Oslo. Norsk geologisk Tidsskrift, 45, 97-1 12, pis 1-5. 1965b. Outline stratigraphy of graptolite-bearing 3b (Lower Ordovician) strata in the Oslo region, Norway. Norsk geologisk Tidsskrift, 45, 481-547. LINDHOLM: SCANDINAVIAN ORDOVICIAN GRAPTOLITES 325 — 1988. The Baltic Hunneberg Series: missing link between Tremadoc and Arenig? 29. In williams, s. h. and barnes, c. r. (eds). Fifth International Symposium on the Ordovician System , Program and Abstracts. St John’s, Canada, 117 pp. maletz, j. and Gutierrez marco, J. c. 1987. The new Early Ordovician (Hunneberg stage) graptolite genus Paradelograptus (Kinnegraptidae), its phytogeny and biostratigraphy. Paldontologische Zeitschrift , 61, 109-131. — and vandenberg, a. h. m. 1985. Araneogrciptus gen. nov. and its two species from the late Tremadocian (Lancefieldian, La2) of Victoria. Alcheringa , 9, 49-63. fortey, r. a. and cooper, r. a. 1986. A phylogenetic classification of the graptoloids. Palaeontology, 29, 631-654. — and owens, r. m. 1987. The Arenig Series in South Wales: stratigraphy and palaeontology. 1. Bulletin of the British Museum (Natural History), (Geology), 41, 69-307. Gutierrez marco, j. c. 1982. Descubrimiento de nuevos niveles con Graptolitos ordovicicos en la unidad ‘Pizarras con Didymograptus’ - Schneider 1939 -(Prov. Huelva, SW de Espana). Communicagoes dos Servigos Geologicos de Portugal , 68, 241-246. — 1986. Graptolitos del Ordovtcico Espahol. Unpublished thesis. Department of Palaeontology, Complutense University, Madrid. — and acenolaza, f. g. 1987. Araneogrciptus murrayi (Hall, 1865) (Graptoloidea, Anisograptidae) : su identidad con “ Dictyonema yaconense" Turner, 1960 y distribution en Espana y Sudamerica. Decimo Congreso Geologico Argentino, San Miguel de Tucumdn , Adas, 1, 321-334. hall, j. 1858. Descriptions of Canadian graptolites. Geological Survey of Canada, Report for 1857, 11 I 145. — 1865. Graptolites of the Quebec group. Geological Survey of Canada, Canadian organic remains, dec. 2, 1 151, pis a and b, 1-21. hall, T. s. 1891. On a new species of Dictyonema. Proceedings of the Royal Society of Victoria (New Series), 4, 7-8, pis I and 2. — 1897. Victorian graptolites. Part 1. Proceedings of the Royal Society of Victoria (New Series), 10, 13-16. Harris, w. j. and thomas, d. e. 1938. Victorian graptolites (new series). Part V. Mining and Geological Journal, 1(2), 70-81. — 1939. Victorian graptolites. Part VE Some multi-ramous forms. Mining and Geological Journal, 2, 55-60. hede, J. e. 1951. Boring through Middle Ordovician-Upper Cambrian strata in the Fagelsang district, Scania (Sweden). 1. Succession encountered in the boring. Lunds Universitets Arsskrift N. F. Avd. 2, 46(7), 1-85. henningsmoen, g. 1973. The Cambro-Ordovician boundary. Lethaia, 6, 423-439. herrmann, o. 1883. Vorlaufige Mittheilung iiber eine neue Graptolithenart und mehrere bisher noch nicht aus Norwegen gekannte Graptolithen. Nyt Magazin for Naturvidenskaberne, 27, 341-362. — 1885. Die Graptolithenfamilie Dicliograptidae, Lapw. mit besonderer Beriicksichtigung von Arten aus dem norwegischen Silur. Mallingske Bogtrykkeri, Kristiania [Oslo], 94 pp. holm, g. 1881. Tvenne nya slagten af familjen Dichograptidae Lapw. Ofversigt af Kongliga Vetenskaps- Akademiens Forhandlingar, 1881(9), 45-51, pis 12 and 13. Stockholm. hutt, j. 1974. The development of Clonograptus tenellus and Adelograptus hunnebergensis. Lethaia, 7, 79-92. jaanusson, v. 1976. Faunal dynamics in the Middle Ordovician (Viruan) of Balto-Scandia. 301-326. In bassett, m. g. (ed. ). The Ordovician System. University of Wales Press, Cardiff, 696 pp. — 1982. Introduction to the Ordovician of Sweden. 1-9. In bruton, d. l. and williams, s. h. (eds.). Field excursion guide , IV International Symposium on the Ordovician System. Paleontological Contributions from the University of Oslo, 279, 217 pp. kraft, j. and mergl, m. 1979. New graptolite fauna from the Klabava Formation (Arenig) of the Ordovician of Bohemia. Vestnik Ustredniho ustavu geologickeho , 54, 291-295, pis 1 and 2. kraft, p. 1987. Graptolite fauna of the Klabava Formation (Ordovician, Arenig) from Teskov near Rokycany. Casopis pro mineralogii a geologii, 32, 59-71, pis 1-6. lapworth, c. 1873. On an improved classification of the Rhabdophora. Geological Magazine, 10, 500-504, 555-560, table 1. legrand, p. 1964. Un graptolite remarquable de l’Ordovicien inferieur du Sahara algerien Choristograptus louhai nov. gen., nov. sp. Bulletin de la Societe geologique de France, (7), 5, 52-58, pis 3 and 4. lindholm, k. 1984. The graptolite faunas of the Lower Ordovician Hunneberg Substage. 27th International Geological Congress, Abstracts, 1, 102-103. Moscow. — 1991. Hunnebergian graptolites and biostratigraphy in southern Scandinavia. Lund Publications in Geology, 95. 326 PALAEONTOLOGY, VOLUME 34 — and maletz, j. 1989. Intraspecific variation and relationships of some Lower Ordovician species of the dichograptid, Clonograptus. Palaeontology, 32, 711-743. m'coy, f. 1851. Description of the British Palaeozoic fossils. In sedgwick, a and m'coy, f. A synopsis of the classification of the British Palaeozoic rocks by the Rev. Adam Sedgwick. Part 1 . Cambridge University Press, London and Cambridge, 184 pp., pi. 1a-l. maletz, j. 1987. Biostratigraphie and Graptolithenfauna ini Unteren Ordovizium des Hunneberges in Vastergotland (Westliches Zentralschweden). Unpublished thesis, Institute of Geology and Palaeontology, University of Gottingen. — and erdtmann, b.-d. 1987. Adelograptus tenellus (Linnarsson 1871): its astogenetic development and its stratigraphical and palaeogeographical distribution. Bulletin of the Geological Society of Denmark , 35, 179-190. Matthews, s. c. 1973. Notes on open nomenclature and on synonymy lists. Palaeontology, 16, 713-719. molyneux, s. G. and dorning, K. J. 1989. Acritarch dating of latest Tremadoc-earliest Arenig (early Ordovician) sediments in the Carmarthen district, south Wales. Geological Magazine, 126, 707-714. — and rushton, a. w. a. 1988. The age of the Watch Hill Grits (Ordovician), English Lake District: structural and palaeogeographical implications. Transactions of the Royal Society of Edinburgh: Earth Sciences, 79, 43-69. monsen, a. 1925. Uber eine neue Ordovicische Graptolithenfauna. Norsk geologisk Tidsskrift, 8, 147-187, pis 1-4. — 1937. Die Graptolithenfauna im Unteren Didymograptusschiefer (Phyllograptusschiefer) Norwegens. Norsk geologisk Tidsskrift. 16, 57-263, pis 1-20. mu, a. t. 1957. Some new or little-known graptolites from the Ningkuo Shale. Acta Palaeontologica Sinica, 5, 369 — 437, pis 1-8. — 1974. Evolution, classification and distribution of Graptoloidea and Graptodendroids. Scientia Sinica, 17, 227-238. song li-sheng, li jin-seng, xu bao-zheng and zhang you-kui 1982. Hemichordata. 294-347, pis 73-83. In Xian Institute of Geology and Mineral Resources (ed. ). Paleontological atlas of Northwest China. Shaanxi-Gansu-Ningxia Volume, Part I. Precambrian and Early Paleozoic. [In Chinese]. nicholson, H. a. 1868. The graptolites of the Skiddaw Series. Quarterly Journal of the Geological Society of London, 24, 125-145, pis 5 and 6. 1873. On some fossils from the Quebec group of Point Levis, Quebec. Annals and Magazine of Natural History, (4), 11, 133-143. — 1876. Notes on the correlation of the graptolitic deposits of Sweden and Britain. Geological Magazine, 13, 245-249, pi. 9. obut, a. m. 1961. Graptolity tremadokskie i smezhnykh s nimi otlozheniy Aktyubinskoy i Orenburgskoy oblastey. In keller, b. m. (ed.). Ordovik Kazakhstana 4, Trudy geologicheskogo Instituta, 18, 146-149, pi. 1. [In Russian], — and sobolevskaya, r. f. 1962. Graptolity rannego Ordovika na Taymyre (Early Ordovician graptolites of Taymyr). In Problems of oil and gas occurrences in the Soviet Arctic: Palaeontology and biostratigraphy. Trudy Nauchnoissledovatelskogo Instituta Geologii Arktiki, 127, 65-85, pis 1-5. [In Russian]. Pritchard, G. b. 1892. On a new species of Graptolitidae ( Temnograptus magnificus). Proceedings of the Royal Society of Victoria (New Series), 4, 56-58, pi. 6. regnell, g. 1960. The Lower Palaeozoic of Scania. 1—43. In regnell, g. and hede, j. e. The Lower Palaeozoic of Scania - The Silurian of Gotland. 21st International Geological Congress, Excursion Guide A 22 and C 17. Geological Survey of Sweden, Stockholm, 89 pp. ruedemann, r. 1947. Graptolites of North America. Memoir of the Geological Society of America, 19, 1-652. rushton, a. w. a. 1981. A polymorphic graptolite from concealed Tremadoc rocks of England. Geological Magazine, 118, 615-622. — 1985. A Lancefieldian graptolite from the Lake District. Geological Magazine, 122, 329-333. Salter, j. w. 1863. Note on the Skiddaw Slate fossils. Quarterly Journal of the Geological Society of London, 19, 135-140. skevington, d. 1966. The lower boundary of the Ordovician system. Norsk geologisk Tidsskrift, 46, 1 1 1-1 19. spjeldnaes, n. 1963. Some Upper Tremadocian graptolites from Norway. Palaeontology, 6, 121-131, pis 17 and 18. stouge, s. and bagnoli, g. 1988. Early Ordovician conodonts from Cow Head Peninsula, western Newfoundland. Palaeontographia Italica, 75, 89-179. strandmark, j. e. 1902. Undre graptolitskiffer vid Fagelsang. Geologiska Foreningens i Stockholm Forhandlingar, 23 [for 1901], 548-556, pi. 17. LINDHOLM: SCANDINAVIAN ORDOVICIAN GR APTOLITES 327 thoral, M. 1935. Contribution a I’etude paleontologique de I’Ordovicien inferieur de la Montague Noire et revision sommaire de la faune Cambrienne de la Montague Noire. Imprimerie de la Charite, Montpellier, 362 pp., 35 pis. tjernvik, t. E. 1956. On the Early Ordovician of Sweden. Stratigraphy and fauna. Bulletin of the Geological Institutions of the University of Uppsala , 36, 107-284, pis 1-11. — and johansson, j. v. 1980. Description of the upper portion of the drill-core from Finngrundet in the south Bothnian Bay. Bulletin of the Geological Institutions of the University of Uppsala (new series), 8, 173-204. tornquist, s. L. 1901. Researches into the graptolites of the lower zones of the Scanian and Vestrogothian Phyllo-Tetragraptus beds, part 1. Lunds Universitets Arsskrift, 37(2: 5), 1-26, pis 1-3. — 1904. Researches into the graptolites of the lower zones of the Scanian and Vestrogothian Phyllo- Tetragraptus beds, part 2. Lunds Universitets Arsskrift , 40(1: 2), 1-29, pis 1-4. tullberg, s. a. 1880. Nagra Didymograptusarter i undre graptolitskiffer vid Kiviks-Esperod. Geologiska Fbreningens i Stockholm Forhandlingar , 5, 39-43, pi. 2. wang gang. 1981. On the discovery of new graptolites from the Tungtze Formation (Lower Ordovician) in Gulin of Sichuan. Acta Palaeontologica Sinica, 20, 349-352, pi. 1. [In Chinese with English abstract], WHITTINGTON, H. B., DEAN, W. T., FORTEY, R. A., RICKARDS, R. B., RUSHTON, A. W. A. and WRIGHT, A. D. 1984. Definition of the Tremadoc Series and the Series of the Ordovician System in Britain. Geological Magazine , 121, 17-33. williams, s. H. and stevens, R. k. 1988. Early Ordovician (Arenig) graptolites of the Cow Head Group, western Newfoundland, Canada. Palaeontographica Canadiana , 5, 1-167. 1991. Late Tremadoc graptolites from western Newfoundland. Palaeontology , 34, 1^17. zalasiewicz, j. a. 1986. Graptolites from the type Arenig Series. Geological Magazine , 123, 537-544. KRISTINA LINDHOLM Department of Historical Geology and Palaeontology Typescript received 5 December 1989 Solvegatan 13, Revised typescript received 5 March 1990 S-223 62 Lund, Sweden TRILOBITES FROM THE ORDOVICIAN OF PORTUGAL by M. ROMANO Abstract. The following trilobite species from the Llanvirn to Llandeilo of north and central Portugal are recorded or described and their stratigraphical ranges are discussed: Colpocoryphe aff. rouaulti (Henry), C. cf. thoralis conjugens Hammann, C. grandis (Snajdr), Salterocoryphe salleri salteri (Rouault), Prionocheilus mendax (Vanek), P. cf. pulcher (Barrande) and Valongia wattisoni (Curtis). Actinopeltis tejoensis sp. nov. and Prionocheilus costai (Thadeu) from the upper Ordovician of central Portugal are described. Salterocoryphe lusitanica (Thadeu) is put into synonymy with Salterocoryphe salteri salteri ; Prionocheilus cf. pulcher is recorded for the first time from Portugal; authorship of Prionocheilus costai is here attributed to Thadeu and a lectotype is chosen. The faunas show similarities with those in central Iberia and northwest France. Calymenid, cheirurid and bathycheilid trilobites form an important element of the Ordovician faunas of Portugal. As early as 1849, Sharpe noted the presence of ‘ Calymene Tristani ' and ‘ Cheirurus ' from Valongo (Text-fig. 1), and some years later Ribeiro (1853) recorded ‘'Calymene Tristani' and ‘ Calymene Arago ' from Buqaco. Delgado (1897, p. 28; 1908, pp. 57, 80, 106) listed six species of ‘ Calymene' from the Ordovician of Buqaco, Amendoa/Maqao and Valongo, but did not describe or figure any of the material. Subsequently, Costa (1942) published a short account on the Calymenidae in which he figured ‘ Calymene Tristani' and ‘ Calymene Salteri' from Valongo. More recently Thadeu (1947) revised some of the upper Ordovician trilobites from Buqaco, among which he described and figured the following species of ‘ Cheirurus' : ‘ {l)Bocagei, claviger , gelasinosus, ( ?) Venceslasi, aff. completus and aff. verrucosus', as well as ‘ Pharostoma costai'. Two years later Thadeu (1949) revised the Portuguese calymenids and recognized five species of ‘ Synhomalonotus' C Aragoi , Salteri , Tristani , transiens, lusitanica') and two of ' Pharostoma ' (‘ Costai , pulchra'). South of the River Douro, along an extension of the Valongo outcrops, Thadeu (1956) again recorded the species ‘ aragoi ' and 'tristani', as well as ' Calymeme cf. duplicata' . Curtis (1961) described Actinopeltis wattisoni from the Valongo area. The present paper revises the taxonomy and distribution of the following genera from the Ordovician of Portugal: Colpocoryphe , Salterocoryphe , Prionocheilus , Actinopeltis and Valongia. Most of the material is restricted stratigraphically to beds of Llanvirn to early Caradoc age; only Prionocheilus costai (Thadeu, 1947) is of late Caradoc-?Ashgill age. Material used is housed in the collections of the Geological Survey offices, Lisbon (prefixed SG or MR) and Earth Sciences Unit, University of Sheffield (prefixed P or RC). Further material was kindly made available by Dr A. H. Cooper (Cooper 1980) and Dr T. P. Young (Young 1985; prefixed ABO, CST, LOR, MDC, PEN, PG and QXP, at present in the Geology Department, University College of Cardiff). STRATIGRAPHY The material studied is mainly from the major outcrops of fossiliferous rocks in Portugal; namely Valongo to Arouca, Buqaco to Rio Ceira, and Domes to Amendoa/Maqao (Text-fig. 1). The simplified stratigraphic columns in Text-figure 2 illustrate the major lithotypes, formations and members of these regions. More detailed descriptions of the rock units may be found in Romano IPalaeontology, Vol. 34, Part 2, 1991, pp. 329-355, 4 pls.| © The Palaeontological Association 330 PALAEONTOLOGY, VOLUME 34 ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 331 text-fig. 2. Generalized lithostratigraphic sections for north Portugal (Valongo-Arouca) and central Portugal (Bugaco, Bupaco-Rio Ceira, Dornes-Amendoa/Mapao). 332 PALAEONTOLOGY, VOLUME 34 and Diggens (1976), Henry, Nion et al. (1976), Mitchell (1974), Cooper (1980), Brenchley et al. (1986) and Young (1985, 1988). A brief resume is given here of the successions shown in Text-figure 2. The Llanvirn Llandeilo sequence at Valongo to Arouca is a monotonous sequence of mudrocks overlying the Armorican Quartzite. The mudrocks (Valongo Formation) are abruptly succeeded by quartzites, above which are pebbly siltstones and sandstones of the Sobredo Formation. This latter unit is of glaciogenic origin and is probably of late Ordovician (Hirnantian) age. The region between Bugaco and Rio Ceira shows a general homogeneity and differs significantly in detail from the Valongo sequence. Young (1985, 1988) has recently revised the lithostratigraphy of this region above the Monte da Sombadeira Formation (Brenchley et al. 1986) and his terminology is incorporated in Text-figure 2. Graptolitic mudstones of Llanvirn age are known from the Brejo Fundeiro Formation (Cooper 1980) and diverse Llandeilo faunas occur in the Fonte da Horta Formation. The base of the Llandeilo is within the upper part of the Brejo Fundeiro Formation, and the Carregueira Formation is probably of early Caradoc age (Young 1988). The Louredo Formation is entirely Caradoc in age; the fossiliferous basal Favagal Bed is considered to be of early Caradoc age (Henry and Thadeu 1971 ; Henry, Nion et al. 1976; Paris 1979, 1981), while the faunas from the uppermost mudstone unit (Galhano Member) indicate an upper Caradoc age (Paris 1979, 1981 ; Young 1988). The overlying Porto de Santa Anna Formation contains a rich fauna in the basal Leira Ma Member. Mitchell ( 1974) attributed an early Caradoc age to this assemblage but later authors have suggested a late Caradoc/early Ashgill age. Young (1988) suggested a possible Rawtheyan age for the upper part of the Porto de Santa Anna Formation. In the southern part of the Bugaco to Rio Ceira region the Porto de Santa Anna Formation is replaced by a sequence of massive dolomites; in the extreme south around Rio Ceira, clastic sequences overlie an attenuated Porto de Santa Anna Formation and are succeeded by pebbly siltstones of the Casal Carvalhal Formation. The final column in Text-figure 2 represents the sequences around Domes and Amendoa/Magao. The units here, up to the Favagal Bed, are essentially similar to those around Bugaco. In the lower part of the Cabego do Peao Formation, however, is a richly fossiliferous unit, termed the Queixoperra Member (Young 1988), of early Caradoc age (includes the Bryozoa Beds of Cooper 1980). Poorly fossiliferous sequences overlie the Cabego do Peao Formation in this southern region, but the pebbly siltstones of the Casal Carvalhal Formation may be correlated with those of the Rio Ceira section and probably the Sobredo Formation at Valongo. The upper part of the Vale da Ursa Formation (Cooper 1980; Young 1988) contains graptolites indicating an early Llandovery age. REMARKS ON THE VERTICAL RANGES AND GEOGRAPHICAL DISTRIBUTIONS OF THE TRILOBITE FAUNAS Vertical ranges Colpocoryphe aff. rouaulti Henry, 1970. This species is first known from the upper Llanvirn where it is present approximately 30 m above the top of the 'Armorican Quartzite’ in the Domes area (Cooper 1980) and persists at least into the Lower Llandeilo (Text-fig. 3). A broadly similar range is known for C. rouaulti in Spain (Hammann 1983; Gutierrez-Marco et al. 1984), and in Brittany it is known to occur from the upper Llanvirn to the Marrolithus bureaui biozone (Henry 1 980c/ and pers. comm.) where it is common south of Rennes (Traveusot Formation) but rare in the upper part of the Postolonnec Formation in the Crozon Peninsula (Henry 1980a). Colpocoryphe cf. thorali conjugens Hammann, 1983. This species first appears less than one metre above the Armorican Quartzite Formation in the Bugaco syncline where it is of early Llanvirn age (Romano et al. 1986); its upper range limit has not yet been established in Portugal. In northeast Portugal it has been recorded from near Moncorvo (Text-fig. 1 ) in beds low down in the Xistenta Formation (Rebelo and Romano 1988) where it is probably of Llanvirn age. Thadeu (1956, p. 19, pi. 6, fig. 1) recorded C. aragoi from the Canelas quarries at Arouca, south-east of Valongo. The specimen figured by Thadeu is poorly preserved but Henry (1970, p. 13) tentatively assigned it to ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 333 s/suaofai smadou/fotf jaqo/nd p sn/iaqoououd sipuoub aqdAjoood/oj ds dqdAjooodjOj H/nDnoj wo aqdAjoood/oj dds snidDJBovuApiQ O '< CP < < o 0 z IU < 1 CO LU ds aqd/joood/oj c~ — r sipuDJd apd/joaod/OQ suabnfuoo i/djol h p aqd/joood/OQ CL O O uayyaq (SnidDJbooouuiip) iuossjduuii srqdDJbouuj/fg dds sn/dojboujApiQ =. iDtsoo snuaqaououj sipuoub aqdrfjoood/oQ ppui ds ^aqd/fjoood/oj i//nonoj noapdAjoaod/OQ suabnfuoo nojoqi p aqdAjooodtoo C' -T - — o- < CL LU O 0 CL 1 CO O o < CP 3 DO ijayyaq (SnidDuboooujiiQ sn/nosnqaja/ snidDuboidiC/Q snujao p sn/dDJbotd/Cig snnbuidojd snjdDJboqjJOt iuossjduuii snfdDJbouuj/CQ dds snidouboLU/fpiij Z O o < CP 3 m lUOS/lfDM DlbUO/D/\ xDpuatM sniiaqoououd uanDS ijai/DS aqdAjooojanos ds aqd/CjooouauDS ys srqDJDO/DO snfdojboqiJO dds sn/doJbow/CpKj S3(i|0(doj6 ouoysua(X9 o O O z o 3 < > text-fig. 3. Sections showing ranges of graptolite and trilobite species. Abbreviations refer to lithostratigraphic units shown in Text-figure 2. 334 PALAEONTOLOGY, VOLUME 34 C. rouaulti. I have not seen Thadeu's specimen but have collected further material from the quarries which include Salterocoryphe sp. (possibly 5. salteri salteri). Thadeu mentioned the presence of furrows on the pleural lobes of the pygidium of the Arouca specimens which suggests that it may be better assigned to Salterocoryphe ; the details of the preglabellar area are not easy to distinguish from Thadeu's figure but the apparently bell-shaped glabella is reminiscent of S. salteri salteri. The age of the Canelas quarries assemblage was regarded as Llandeilo by Thadeu but the presence of Hungioides bohemicus (see Rabano 1983), Bathycheilusl castilianus and Nobiliasaphus caudiculatus, as well as poorly preserved pendent didymograptids, suggests a Llanvirn and possibly early Llanvirn age (Gutierrez-Marco et al. 1984). Courtessole et al. (1981) assigned C. thorali thorali (Dean, 1966), from the Lower Arenig of the Montagne Noire, to Salterocoryphe but this is not accepted here. Colpocoryphe grandis (Snajdr, 1956). This species in Portugal appears to be restricted to the lower Caradoc and first makes its appearance in the Carregueira Formation in central Portugal. It is last recorded from the Queixoperra Member of the Cabeqo do Peao Formation, some 15-20 m above the Favagal Bed. The species has a greater stratigraphic range in Brittany (Henry 1980a) where it is known from the top of the Postolonnec Formation, Schistes de Raguenez and Riadan Formation. In Spain, Rabano (1984) records it only from the Caradoc but I have collected a deformed cephalon of Colpocoryphe from just above the Los Rasos sandstones (equivalent to the Monte de Sombadeira Formation) in the Guadarranque area, Toledo Mountains, central Spain, which is very close to C. grandis. Hammann (1983, unit 6 in Guadarranque section) considered this horizon to be of Llandeilo age and, if the identification proves to be correct, thus possibly extends the range of this species in Spain to approximately equal to that in Bohemia, where it is of Llandeilo-Caradoc age (Dobrotiva, Liben and Letna Formations, see Havlicek and Vanek 1966). Salterocoryphe salteri salteri (Rouault, 1851). This species definitely occurs in beds of Llandeilo age from the upper part of the Valongo formation in north Portugal, but as yet I have not recorded undoubted specimens from the Llanvirn. However, at Arouca, a single pygidium from the lower part of the Valongo Formation (probably early Llanvirn) is tentatively identified as S. salteri salteri. Delgado (1908, pp. 134, 137 and 138) questionably identified the species from the ‘Schistes a Didymograptus ’ (lower part of the Valongo Formation and considered by Gutierrez-Marco, fide Hammann et al ., 1986 to be of early Llanvirn age), but this material has not been seen by the author. In Spain the species occurs in the Llanvirn at Guadarranque and much of the Llandeilo of Corral de Calatrava and El Centenillo (Hammann 1983; Gutierrez-Marco et al. 1984), while in Brittany it appears to be restricted to the Llandeilo (Henry 1980a) where it occurs south of Rennes, on the northern flank of the Laval syncline and only very rarely in the western part of the median syncline. Prionocheilus mendax (Vanek, 1965). The range of this species in Portugal parallels that of S. salteri salteri ; it is of Llandeilo age at Valongo but possibly extends down into the Llanvirn (Delgado 1908, pp. 57, 106). In Bohemia it ranges from the Llandeilo to lower Caradoc (Vanek 1965) while in central Spain and Brittany it is exclusively of Llandeilo age (Rabano 1984; Henry 1980a). Prionocheilus cf. pulcher (Barrande, 1846). This species has so far only been recorded from the Caradoc of the Dornes-Amendoa/Ma?ao region and as far as I am aware does not occur in Spain. In Brittany it has only tentatively been recorded from the lower Caradoc although Dr J.-L. Henry informs me that there are differences in that the French specimen shows shorter and straighter spines on the cephalic border than either P. pulcher or P. verneuili. In Bohemia it has a range throughout much of the Caradoc. Prionocheilus costai (Thadeu, 1947). This species in central Portugal is so far only known from beds of late Caradoc or early Ashgill age and is probably of a similar age in Spain (Hammann 1983) where it is known from the ‘Bancos mixtos’. It is also present in dropstones from the basal part of ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 335 the Casal Carvalhal Formation at Domes (Dr T. P. Young pers. comm.)- In Brittany it occurs in the lower part of the Rosan Formation. Actinopeltis tejoensis sp. nov. This species is only known from the Queixoperra Member, Amendoa/Maqao region, of Caradoc age. The genus is recorded from Spain, from the Caradoc (Flammann 1972; Rabano 1984), and also from the Rosan Formation of Brittany (J.-L. Henry pers. comm.). Valongia wattisoni (Curtis, 1961). This monospecific genus is at present only known from the Llandeilo of Valongo. Geographical distributions The distribution of the species described in this paper substantiates the previously documented contrast found in the composition of the trilobite faunas in north (Valongo-Arouca, Marao, ?Moncorvo) and central (Buc;aco-Amendoa/Maqao) Portugal throughout much of the Ordovician (Hammann and Henry 1978; Henry and Romano 1978; Romano 1982). For example, during the lower Llandeilo Salter ocoryphe salteri saltern Prionocheilus mendax and Valongia wattisoni are only known from the 'northern’ region, while Colpocoryphe aff. rouaulti is apparently restricted to the 'southern' region (although an imperfectly preserved specimen of IColpocoryphe is known from the upper part (Llandeilo) of the Valongo Formation in the north). During Caradoc-Ashgill times, trilobite faunas are at present unknown in north Portugal but C. aff. rouaulti , C. grandis , P. cf. pulcher , P. costai and A. tejoensis occur further south. However, in the lower Llanvirn, C. cf. thorali conjugens appears to have had a wider distribution and is recorded from Moncorvo and Bu<;aco Rio Ceira. Within Brittany and Spain, trilobite associations also show restricted distribution (Henry 1980u; Rabano 1984) and it has been frequently noted that, for example, sequences and faunas in the Crozon Peninsula have more in common with the Bugaco area in Portugal (Henry and Thadeu 1971 ; Paris 1981 ; Young 1989, 1990) than with the Ordovician succession south of Rennes (Henry and Morzadec 1968; Henry, Melou et al. 1976; Henry, Nion et al. 1976). Two maps are presented (Text-fig. 4) of France and Iberia during early Llanvirn and early Llandeilo times which show the distribution of the trilobites described in this paper. These distributions are now briefly discussed. Early Llanvirn. C. thorali conjugens , S. salteri salteri and S. sampelayoi are known to occur in the Montes de Toledo and Sierra Morena of central Spain. Elsewhere in Iberia and Brittany their presence appears to be patchy. 5. salteri salteri possibly occurs in the Valongo-Arouca region while C. cf. thorali conjugens is so far only known from Bm;aco and Moncorvo in Portugal and probably a similar form is present in the Traveusot Formation, south of Rennes in Brittany (J.-L. Henry pers. comm.). S. sampelayoi (Hammann, 1977) has only definitely been recorded in Spain to date. Early Llandeilo. As in the lower Llanvirn, central Spain appears to have been environmentally homogeneous in that C. rouaulti , S. salteri salteri and P. mendax are known across most of the region, and are also present in eastern Portugal. At Valongo C. rouaulti is probably absent although, as indicated above, a poorly preserved IColpocoryphe may belong to this species. In the Bugaco area the author has not seen specimens of either S. salteri salteri or P. mendax although Delgado ( 1908) recorded ‘ Calymene pulchra' from the Brejo Fundeiro Formation, Louredo Formation and probably Porto de Santa Anna Formation. In Brittany all three species are known from the median syncline, Domfront and south of Rennes, but P. mendax does not occur in the Crozon Peninsula. The distribution of the above species is informative in terms of environmental differences within the Central Iberian Zone ( sensu Hammann et al. 1982) and Brittany. During early Llanvirn times all the areas appear to show a general similarity in that mud/silt was deposited over a broad shelf 336 PALAEONTOLOGY, VOLUME 34 text-fig. 4. Maps showing distribution of trilobite species during early Llanvirn and early Llandeilo times in Iberia and north-west France. Stippled areas represent outcrop of Ordovician/Silurian rocks. following the post 'Armorican Quartzite ’ transgression. Little direct evidence regarding water depth or proximity to shore can be ascertained either from the lithofacies or faunas, and one of the few indications that there was a change in conditions across the Iberian region is seen in the lower diversity of the trilobite faunas from south to north. A similar situation probably existed during early Llandeilo times, although in terms of the species considered here few convincing differences can be seen. However, when consideration is given to a larger sample of the trilobite faunas (Romano 1982), as well as to the lithofacies (Brenchley et al. 1986) the differences are considerably more marked. The major contributing factor to the differences in the trilobite assemblages is probably water depth, with its accompanying control on energy level/light/temperature and/or food supply. It was suggested by Brenchley et al. (1986) that the Ordovician shelf in central and western Iberia deepened towards the north in Llandeilo times. This picture fits in well with the observed taphonomy of the trilobite assemblages from Valongo, with their relatively high proportion of ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 337 complete specimens (Romano 1976) and frequent dispersal within the rock (low energy conditions) compared with the often disarticulated exoskeletons and common bedding plane accumulations at Bugaco (higher energy conditions), particularly above the Monte da Sombadeira Formation. SYSTEMATIC PALAEONTOLOGY The classification of Colpocoryphe and Salter ocoryphe has recently been comprehensively discussed by Henry (19806) and Hammann (1983). Henry pointed out reasons for excluding the former from the Homalonotidae (Sdzuy 1957; Bergstrom 1973; Thomas 1977) and included it, with Salter ocoryphe, in the Calymenidae. Henry further suggested that Salterocoryphe could be placed in the Flexicalymeninae (Siveter 1977), supported by the fact that Flexicalymene ( Onnicalymene ) jemtlandica and Salterocoryphe salteri have almost identical hypostoma, and that Colpocoryphe should be restricted to Colpocoryphinae. Hammann (1983), however, favoured the inclusion of both Salterocoryphe and Colpocoryphe within the Colpocoryphinae of the Calymenidae and considered that Prionocheilus of the subfamily Pharostomatinae should be included in the Bathycheilidae. The suprageneric level of classification is not further discussed here and in the following section the genera are not grouped into higher ranks. Genus colpocoryphe Novak in Perner, 1918 Type species. Calymene arago Rouault, 1849. Colpocoryphe aff. rouaulti Henry, 1970 Plate 1, figs I -13, 15, 17 ?1908 Calymene Aragoi Rouault; Delgado, p. 57. 1949 Synhomalonotus Aragoi (Rouault); Thadeu, pi. 1, fig. 1. Material. Two cephala; two cranidia; eight pygidia, with or without attached thoracic segments; thirteen complete, or nearly complete specimens; all preserved as internal and/or external moulds. Horizon and locality. SG 142 (Thadeu 1949, p. 1, fig. 1 ), 7143, 500 m N 40° E of Cacemes, Bugaco. SG 144, 450 m S 70° E of Louredo, Bugaco, probably Llanvirn. SG 1326, 100 m N 40° E of Beloi chapel, probably Llanvirn. SG 2190 and MR 59-64, section from Zuvinhal to Santa Ant. do Cantaro, unit 25, Bugaco (Delgado 1908, p. 35), Llanvirn. MR 41, 42, section through Val. San Jorge, unit 20, Bugaco. MR 43-48. same section as specimens 41-42, unit 21, Bugaco (Delgado 1908, p. 42). MR 49-51, Palheiros, Bugaco. MR 52-58, 900 m S 65° E of Venda Nova, Poiares. The material occurs in beds from the Brejo Fundeiro Formation (Llanvirn) to the Fonte da Horta Formation (Llandeilo). MR 41-48 are of Llandeilo age, MR 49-58 are Llanvirn. Discussion. The material is certainly very close to C. rouaulti but differs from it in several respects. In the Portuguese specimens the glabella converges forwards more markedly and the straight anterior margin of the glabella is shorter. The swollen posterior lobe to the central body of the hypostoma is more like that figured by Hammann ( 1983) while the internal posterior notch is closer to that of Henry’s (1970, 1980u, b) material. The pygidium shows slight differences in the shallower axial and vincular furrows and smaller side lobes. At this stage the author prefers to identify the Portuguese material as C. aff. rouaulti. Further, the Portuguese material suggests that there may be slight differences between the Llanvirn and Llandeilo forms assigned here to C. aff. rouaulti. Although the cephala are virtually indistinguishable the pygidial axis of the stratigraphically lower specimens tends to carry less well-defined ring furrows and the vincular furrows are less strongly indented than in the Llandeilo forms. It is possible that the Llanvirn material may prove to be subspecifically distinct from the Llandeilo form, but this must await more and better preserved material. 338 PALAEONTOLOGY, VOLUME 34 Colpocoryphe cf. thorali conjugens Hammann, 1983 1986 Colpocoryphe cf. thorali conjugens Hammann; Romano et al., p. 429, pi. 1, figs 2-5. 1988 Colpocoryphe cf. thorali conjugens Hammann; Rebelo and Romano, p. 54, pi. I, figs 8-11; pi. 2, fig. 5. Material. Four cranidia; two cranidia, with part thorax; three pygidia; all preserved as internal and/or external moulds. Horizon and locality. PI 57/7, L2 m above lingulid bed at top of Armorican Quartzite Formation, road section south of River Mondego, Penacova. RC1/2, 10 cm above lingulid bed at top of Armorican Quartzite Formation, track section, north of River Ceira, Vila Nova do Ceira. SG 1 154/1-4, 6, 10, Xistenta Formation, 3-5 km ESE of Mos. 13 km east of Moncorvo. P157/7 and RC1/2 are Lower Llanvirn, SG 1154/1^4, 6, 10, probably Llanvirn. Discussion. For description and discussion of the above material see Romano et al. (1986) and Rebelo and Romano (1988). Nothing new can be added. The subspecies is known from lower Llanvirn of the Sierra Morena (Hammann 1983; Rabano 1984; Gutierrez-Marco et al. 1984). Colpocoryphe grandis (Snajdr, 1956) Plate 2, figs 1-3, 7, 8, 1 1 ? 1 908 Calymene Aragoi Rouault; Delgado, pp. 41, 57. *1956 Calymene ( Colpocoryphe ) grandis Snajdr; p. 529, pi. 3, figs. 1-9. 19806 Colpocoryphe grandis (Snajdr, 1956); Henry, text-fig. 3, pi. 2, figs 3 and 4. (for full synonymy see Flenry 1980a, p. 64; and Hammann 1983, p. 85). Material. Three cephala; three cranidia; two cephala, with part of thorax; five pygidia, with part of thorax; two pygidia; two complete or nearly complete specimens; all preserved as internal and/or external moulds. Horizon and locality. LOR 1.007-9, Louredo Formation, Favagal Bed, Louredo village. LOR 2.001, less than 2 m below Favagal Bed, Louredo village. PEN 1.001-2, less than 10 m below Favagal Bed, quarry 320 m ENE EXPLANATION OF PLATE 1 Figs 1-13. 15, 17. Colpocoryphe aft', rouaulti Henry, 1970. 1-3, SG 142; internal mould of cephalon, dorsal, anterior and lateral views, x 2, Brejo Fundeiro Formation, Bugaco, Llanvirn. 4, MR 41 ; internal mould of cranidium, dorsal view, x F4, Fonde da Horta Formation, Bugaco, Llandeilo. 5, MR 42; internal mould of cranidium, dorsal view, x 2, Fonte da Horta Formation, Bugaco, Llandeilo. 6-8, MR 52; internal mould of cranidium, dorsal, anterior and lateral views, x F8, Brejo Fundeiro Formation, Poiares, Llanvirn. 9, MR 44; internal mould of cranidium, dorsal view, x 1, ?Fonte da Horta Formation, Bugaco, Llandeilo. 10, SG 144; internal mould of pygidium, dorsal view, x 2, Brejo Fundeiro Formation, Bugaco, ?Llanvirn. 11, SG ? 1 43 ; internal mould of pygidium, dorsal view, x 2, Brejo Fundeiro Formation, Bugaco, ?Llanvirn. 12, MR 49; internal mould of pygidium, dorsal view, x 2, Brejo Fundeiro Formation, Bugaco, Llanvirn. 13, MR 53; internal mould of pygidium, dorsal view, x2, Brejo Fundeiro Formation, Poaires, Llanvirn. 15, MR 46; internal mould of nearly complete specimen, dorsal view, x 1, ?Fonte da Horta Formation, Bugaco, Llandeilo. 17, MR 45; internal mould of nearly complete specimen, x 1, ?Fonte da Horta Formation, Bugaco, Llandeilo. Fig. 14. Colpocoryphe sp. MD 2.001/2; internal mould of incomplete cranidium, dorsal view, x 6, Carregueira Formation, Domes, Caradoc. Fig. 16. Colpocoryphe ? sp. indet. SG 146; partly enrolled specimen with 6 thoracic segments and pygidium, dorsal view of pygidium, x 1.2, Fonte da Horta Formation, Bugaco, Llandeilo. Fig. 18. Salterocoryphe salteri salteri (Rouault, 1851). SG 1 681 ; internal mould of pygidium, dorsal view, x 09, Valongo Formation, Valongo, Llandeilo. PLATE 1 ROMANO, Colpocoryphe , Colpocoryphel, Salter ocoryphe 340 PALAEONTOLOGY, VOLUME 34 of east end of bridge over River Mondego, east of Penacova. QXP 2.001-5, 40, and Magao specimen of Cooper (1980), ‘1700 mN 57° E de pyr. de Queixoperra, Magao', probably from 'Bryozoa Beds’ (Cooper 1980; Romano 1982) within unit 7 of the ‘Schistes a Orthis Berthoisi' (Delgado 1908, p. 92), Queixoperra Member of Cabego do Peao Formation. CST 2.001M, 1400 mN 62° E of Pereiro, Magao, oolitic beds probably equivalent to basal oolite (Favagal Bed) of Louredo Formation. ABO 9.001, Aboboreira, Carregueira Formation, from less than 3m below oolitic beds, and ABO 10.001, basal ‘Bryozoa Beds’, Queixoperra Member, both approximately 1 km WNW of Carregueira, Magao. T. Young collection, unnumbered specimens from 2 km SSE of Aboboreira [20710, 28915], and west of Pereiro [20975, 29090]; all from Favagal Bed. Domes material (loc. 70), grid reference 18992 31368, from ‘Bryozoa Beds’ of Cabego do Peao Formation. Fragmental material is also known from near the top of the Carregueira Formation at Rio Ceira (Young 1985). All specimens are probably of early Caradoc age. Discussion. The present material agrees in all important respects with that described and figured by Snajdr (1956), Destombes (1966), Henry (1980a) and Hammann (1983). Colpocoryphe ? sp. indet. Plate 1, fig. 16 71908 Calymene transiens Verneuil and Barrande; Delgado, p. 57. 1949 Synhomalonotus transiens (Verneuil and Barrande); Thadeu, pi. 1, fig. 6. Material. SG146. Internal mould of enrolled specimen with six thoracic segments and pygidium (Thadeu 1949, pi. 1, fig. 6). Horizon and locality. ‘ 100 m S 80° E of Foredo’, Bugaco. Delgado (1908, p. 57) records the species from the ‘Schistes a Homalonotus oehlerti' (Fonte da Horta Formation) of Flandeilo age. Description. Thoracic segments of Colpocoryphe type (see Henry 1980a, pi. 7, fig. la) carrying sculpture of small tubercles. Pygidial axis wide anteriorly, narrowing evenly backwards and with shallow axial furrows, posterior part not preserved. Eight visible axial rings seen separated by shallow, complete ring furrows. Small triangular pleural lobes extend back to eighth axial ring and carry up to four poorly defined ribs. Fateral borders have wide, open furrows with no trace of segmentation on lower surfaces. Sculpture slightly coarser than on thorax. Discussion. The absence of segmentation on the lateral borders is typical of the genus Colpocoryphe as distinct from Salterocorvphe. The ribs on the pleural lobes are more obvious than in C. rouaulti and C. grandis and there is no median shallowing of the axial ring furrows in the pygidium as in the latter species, although this feature is more apparent on external moulds (Henry 1980a, pi. 7, figs 6a, b and 7; pi. 8, fig. 2 a-d). EXPLANATION OF PLATE 2 Figs 1-3, 7, 8, 11. Colpocoryphe grandis (Snajdr, 1956). 1. QXP 2.0001; internal mould of incomplete cranidium, dorsal view, x 1, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 2, 7, QXP 2.040; internal mould of pygidium, dorsal and posterior views, x 1 and x 1-4 respectively, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 3, PEN 1.002; internal mould of pygidium, dorsal view, x 1, Carregueira Formation, Amendoa/Magao, Caradoc. 8, CST 2.004; internal mould of cephalon, anterior view, xO-6, Favagal Bed, Magao, Caradoc. 11. CST 2.003; internal mould of fragmentary cephalon and hypostoma, anterior view, x 0-7, Favagal Bed, Magao, Caradoc. Figs 4-6, 9, 10, 12. Salterocoryphe salteri salteri (Rouault, 1851). 4, SG 1687.1 ; internal mould of incomplete specimen, dorsal view, x 1, Valongo Formation, Valongo, Flandeilo. 5, SG 149; internal mould of complete specimen, dorsal view, x 1-3, Valongo Formation, Valongo, Flandeilo. 6, SG 1686.1; internal mould of partly enrolled specimen, dorsal view, x 1-2, Valongo Formation, Valongo, Flandeilo. 9, SG 1325; internal mould of specimen with hypostoma, dorsal view, x 1-4, Valongo Formation, Valongo, Flandeilo. 10, 12, SG 1687; internal mould of nearly complete specimen, dorsal view, x 1, anterior view, x 14, respectively, Valongo Formation, Valongo, Flandeilo. PLATE ROMANO, Colpocoryphe, Salter ocoryphe 342 PALAEONTOLOGY, VOLUME 34 The incomplete specimen precludes specific identification. Delgado and Thadeu referred the species to ‘ Calymene transiens ’ (Verneuil and Barrande 1855, p. 974, pi. 25, fig. 5) from Almaden, Spain, but Verneuil and Barrande’s figure and description do not permit a close comparison. Tromelin and Lebesconte (1876, p. 629) reinvestigated the type of transiens and regarded it as belonging to ‘ Calymene salteri \ Henry (1970, p. 22) pointed out that Salteroeoryphe salteri is present at Almaden which tends to support Tromelin and Lebesconte’s suggestion. Hammann (1983, p 90) reported that the type of Calymene transiens could not be found and that the species cannot reliably be attributed to either Colpocoryplie or Salteroeoryphe. However, Hammann considered it to be closer to the latter. Colpocoryplie sp. Plate 1, fig. 14 Material. MDC 2.001/2, part and counterpart of incomplete cranidium. Horizon and locality. Carregueira Formation, Domes, type section (Young 1985), about 5 m below oolitic horizon (at base of Cabego do Peao Formation). Lower Caradoc. Description. Glabella (excluding occipital ring) slightly longer than wide, sides gently converging anteriorly. Occipital furrow forwardly flexed, occipital ring carries small median tubercle which is considerably fainter on external mould. Anterior margin of glabella gently rounded with short, straight median part. S2 furrows are short, straight and inclined only slightly backwards; S3 very short and indistinct. Anterior notch of cranidium is broad with widely diverging sides. Cranidium is finely tuberculate. Discussion. The specimen is small (about 4-5 mm long) and may represent a meraspid stage. The open anterior notch suggests affinities with Colpocoryplie grandis but this species does not possess such a narrow glabella, at least in adult specimens, or a median occipital tubercle. The short S2 and S3 furrows and fine sculpture are features of Salteroeoryphe salteri salteri (Hammann 1977; Henry 1980u) and the juvenile of this species bears a median occipital tubercle (Hamman 1983, pi. 1 1, fig. 110). However the structure of the anterior cephalic border is typical of Colpocoryplie and until juvenile specimens of C. grandis are described it is preferable to leave the Portuguese specimen in open nomenclature. Genus salterocoryphe Hamman, 1977 Type species. Calymene salteri Rouault, 1851. Salterocoryphe salteri salteri (Rouault, 1851) Plate 1, fig. 18; Plate 2, figs 4-6, 9, 10, 12; Plate 3, fig. 9. *1851 Calymene salteri Rouault, p. 358. 1949 Synhomalonotus salteri (Rouault) Thadeu, pi. 1, figs 2-3. 1949 Synhomalonotus lusitanica Delgado; Thadeu, p. 131, pi. 1, figs 7-9; pi. 2, figs 1 and 2. 1960 Neseuretus lusitanica (Thadeu); Whittard, p. 145. 1966 Colpocoryplie lusitanica (Thadeu); Dean, p. 309. 1982 Neseuretus lusitanicus (Thadeu); Fortey and Morris, p. 70. 1982 Salterocoryphe lusitanica Romano, p. 96. 1982 Salterocoryphe salteri Romano in Hammann, Robardet and Romano, p. 40. (for full synonymy see Henry 1970, p. 18; and Hammann 1983, p. 90). Material. Five complete or nearly complete specimens; fourteen other specimens; all preserved as internal or external moulds. ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 343 Horizon and locality. SG 147 (figured by Thadeu 1949, pi. 2, fig. 2), ‘ vizinhangas de Valongo'. SG 148 (Thadeu 1949, pi. 1, fig. 9), ‘800 m NE de Boloi’. SG 149 and 150 (Thadeu 1949, pi. 1, figs 7 and 8), ‘ Ribeira da Murta, Valongo’. SG 1688, (Thadeu 1949, pi. 2, fig. 1) ‘ 1650 m (non 1680 m, Thadeu 1949, p. 131 ) S 20° W da piramide de Santa Justa’. SG 1323 and 1324, ‘ 1000 m S 30° E da igreja de Covelo, Valongo’. Information taken from labels on specimens, slightly modified using Thadeu (1949). Other specimens from Beloi, Covelo, Santa Justa and Penha Garcia (see Text-fig. 1). PG 6.001 from approximately 11m above Monte da Sonrbadeira Formation, Penha Garcia. All north Portuguese material is probably from upper part of the Valongo Formation (Schistes a Uralichas Ribeiroi of Delgado 1908) of Llandeilo age. Discussion. The species has been recently described and figured by Henry (1970) and Hammann (1983). Henry (1980u) distinguished two subspecies of S. salteri of which the Portuguese material may be assigned to 5. salteri salteri. Since Thadeu (1949) first described Salter ocoryphe lusitanica , various authors (Whittard 1960; Dean 1966) have briefly referred to it in discussions relating to generic assignment. Henry (1970, p. 21) discussed the status of the species when he suggested that it bore a striking resemblance to S. salteri in that the form of the glabella, eye position and number of axial rings were identical. The only difference that Henry noted was that the pleural lobes on the pygidium were more clearly segmented in the Portuguese material, but this he thought could well be the result of deformation. A study of additional material indeed confirms Henry’s suggestion that the preservation of the furrows varies with the deformation. The anterior part of the pygidial axis tends to be relatively narrower in the Portuguese specimens but this does not seem to be an important criterion for separating the two forms. Hence I prefer to put lusitanica into synonymy with salteri. Hammann (1983, p. 93) regarded lusitanica as a distinct species but his criteria are not accepted here; for example the range in glabella length : width ratios of salteri and lusitanica are virtually identical and the ornament of specimens assigned to lusitanica is similar to that of salteri. Genus prionocheilus Rouault, 1847 (Syn. Pharostoma Hawle and Corda, 1847) Type species. Prionocheilus verneuili Rouault, 1847. Remarks. The question of the priority of Prionocheilus Rouault, 1847 or Pharostoma Hawle and Corda, 1847 has been discussed by Dean (1964, 1966, p. 300; 1971, p. 42), Whittington (1965, p. 56), Ingham (1977, p. 103), Siveter (1977, pp. 339, 393), Owen and Bruton (1980, p. 2), Henry ( 1980n, p. 79) and Hammann ( 1983, p. 51 ). For the present paper I prefer to accept Dean’s argument and follow his suggestion for using Prionocheilus. Prionocheilus mendax (Vanek, 1965) Plate 3, figs 1-5, 8 1908 Calymene pulchra Barrande; Delgado, pp. 106, 134, 7138. 1942 Calymene pulchra Barrande; Costa, p. 93. 1949 Calymene pulchra Barrande; Thadeu. p. 129, pi. 2, figs 3-5. *1965 Pharostoma pulchrum mendax ; Vanek, pp. 30—32, fig. 6; pi. 2, fig. 10; pi. 3, figs 6 and 7; pi. 4, figs 2-5. (for a full synonymy see Vanek 1965, p. 30; Henry 1980n, pp. 80-81; and Hammann 1983, p. 53). Material. Three cephala with part thorax; one hypostoma; two free cheeks; twenty-three complete or nearly complete specimens; all preserved as internal and/or external moulds. Horizon and locality. SG 151, ‘Valongo’. SG 1692, 1692.1-3, 1693.1 and MR 38-40, 1400 m S 32° E of Covelo church, Valongo. SG 1327, 1327.1, 1000 nr S 30° E of Covelo church, Valongo. SG 1691, 1691.1-3, 800 m S 26° W of ‘ermida de Santa Justa, Valongo’. Delgado (1908) records the species from the ‘Schistes a Didymograptus' (Llanvirn) to the ‘Schistes a Uralichas Ribeiroi ’ (Llandeilo) of the Valongo Formation 344 PALAEONTOLOGY, VOLUME 34 (Romano and Diggens 1976; it is not possible to relate his specimens to exact horizons. The author has collected this species from beds of Llandeilo age at Valongo but has not recorded it from the Llanvirn. Description. A full description is not given since the species is well documented. Cephalon semicircular in outline with evenly rounded anterior and lateral margins. Glabella subtriangular in outline, with evenly curved anterior margin and slightly curved sides. Maximum width of cephalon about two and a half times that of posterior glabellar width. Glabella from 0.7-0. 8 times as long as cephalon. Three pairs of glabellar lobes and furrows. Faint oval areas situated on inner side of posterior branch of IS. Glabella gently convex dorsally. Axial furrows generally deep, expanding into small, crescent-shaped paraglabellar areas abaxial to 1L. Palpebral lobes prominent, situated opposite 2L and nearer to axial furrow than lateral margin. Faint eye ridge runs to 2S. Free cheeks with long, posteriorly directed genal spines. Lateral and anterior margin of cephalon carry at least fifty downwardly directed, slightly curved (posteriorly) spines. Sculpture of small tubercles of uniform size, absent in paraglabellar areas. Hypostoma consists of gently convex, subovate middle body, longer than wide, with shallow inwardly directed furrows defining a posterior crescent-shaped lobe; lobe consists of two oblique lateral lobes. Anterolateral margins with small outwardly directed pointed wings. Posterior margin of hypostoma has a shallow open notch; posterior projections are rounded. Thorax consists of thirteen segments. Axial furrows gently outwardly curved ; axis widest at about third axial ring where it is over one and a half times as wide as at posterior end. Axial rings lobate laterally. Broad (trans.), horizontal inner parts of pleural region, outer parts bent sharply down. Pleural furrows deep, wide (exsag.) and straight, starting at anterolateral corner of axial ring and running approximately parallel to pleural margins. Posterior border slightly wider (exsag.) than anterior. At geniculation, furrows swing forwards and die out before reaching rounded tip of segment. Thorax finely tuberculate like cephalon. Pygidium semicircular in outline. Anterior end of axis about one-third maximum width of pygidium. Axis narrows evenly backwards, not reaching posterior margin. Axis carries five rings (sometimes with faint suggestion of a sixth) and a semicircular terminal piece which stands higher than rest of axis. Ring furrows shallow and narrow (sag.) posteriorly. Up to five nearly straight ribs, separated by deep furrows which curve strongly backwards distally. First, three/four ribs carry short furrows extending from axial furrow. Sculpture similar to that of cephalon and thorax. EXPLANATION OF PLATE 3 Figs 1-5, 8. Prionocheilus mendax (Vanek, 1965). 1, MR 38.3 and 38.5; latex cast of external mould, dorsal view, x 1, Valongo Formation, Valongo. Llandeilo. 2, MR 38.6-8; latex cast of external mould, dorsal view, x08, Valongo Formation, Valongo, Llandeilo. 3, MR 39; internal mould of incomplete specimen, dorsal view, x 1, Valongo Formation, Valongo, Llandeilo. 4, MR 38.1 ; latex cast of external mould, dorsal view, x 1, Valongo Formation, Valongo, Llandeilo. 5, SG 1692.1; latex cast of internal mould showing hypostoma, dorsal view, x 1, Valongo Formation, Valongo, Llandeilo. 8, MR 38.13; latex cast of external mould of free cheek, dorsal view, x 1, Valongo Formation, Valongo, Llandeilo. Figs 6, 10. 12, 13. Prionocheilus cf. pulcher (Barrande, 1846). 6, QXP 2.017; latex cast of external mould of cranidium, dorsal view, x 3, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 10, SG 152; internal mould of cranidium, dorsal view, x 1-5, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 12, QXP 2.014; internal mould of cranidium, dorsal view, x 2, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 13, QXP 2.009; latex cast of external mould of pygidium, dorsal view, x 2, Cabego do Peao Formation, Amendoa/Magao. Caradoc. Fig. 9. Salterocoryphe salteri salteri (Rouault, 1851). PC 6.001; latex cast of external mould of cranidium, dorsal view, x F75, Fonte da Florta Formation, Penha Garcia, Llandeilo. Figs 7, 1 1, 14-16. Actinopeltis tejoensis sp. nov. 7, QXP 2.041 ; internal mould of pygidium, dorsal view, x 2, Cabego do Peao Formation, Amendoa/Magao, Caradoc. II, QXP 2.043; internal mould of incomplete pygidium, dorsal view, x 3, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 14, QXP 2.006; internal mould of cephalon, dorsal view, x 2, Cabego do Peao Formation, Amendoa/Magao, Caradoc. 15 and 16, SG 225 (holotype); internal mould and latex cast of corresponding external mould, dorsal views, x 1-4 and F8 respectively, Cabego do Peao Formation, Amendoa/Magao, Caradoc. PLATE 3 ROMANO, Prionocheilus, Salter ocoryphe, Actinopeltis 346 PALAEONTOLOGY, VOLUME 34 Discussion. The Portuguese material is very similar to that figured by Vanek (1965) from the Dobrotiva Formation (Llandeilo) to Letna Formation (lower Caradoc) of Bohemia. A minor difference is the shallower posterior axial ring furrows and pleural furrows. This was also noted by Flenry (1980«, p. 81) who recorded the species from the upper part of the Traveusot and Andouille formations (Llandeilo) of Brittany. Flenry (1980a) figured an in situ hypostoma and commented on the difference between it and that figured by Vanek (1965, pi. 2, fig. 10). Vanek’s specimen of a hypostoma is incomplete and contrasts with the notched posterior margin of specimens from Brittany (Henry 1980a, fig. 31, pi. 14, fig. 3 a) and Spain (Hammann 1983, text-fig. 14). Although the two Portuguese in situ hypostomata are not well-preserved, they both show the notched posterior margin; the posterior projections, however, have more rounded outlines than in the Brittany specimen. The minor differences mentioned above are considered insufficient to separate the Portuguese species from Prionocheilus mendax. Prionocheilus cf. pulcher (Barrande, 1846) Plate 3, figs 6, 10, 12, 13 1908 Calymene pulchra Barrande; Delgado, pp. 57, 80. Material. Thirteen cranidia; one free cheek; two cephala with part of thorax; four pygidia; all preserved as internal and/or external moulds. Horizon and locality. SG 152, ‘Aboboreira, 300 m N 60° W (Maqao)’, from the 'Schistes a Orthis ( = Svobodaina) Berthoisi ’ (Delgado 1908, p. 80, but not listed in following descriptions of stratigraphic sections). QXP. 2.008-2.028, ‘ 1700 m N 57° E de pyr. de Queixoperra’, Maqao; ‘Bryozoa Beds’, Queixoperra Member, ‘Schistes a Orthis Berthoisi'. ABO 10.002, approximately 1 km WNW of Carregueira, Maqao; basal ‘Bryozoa Beds’, Queixoperra Member, Cabeqo do Peao Formation. All material of Caradoc age (probably early). Description. Glabella subtriangular in outline with nearly straight anterior margin, length just over three- quarters the basal width. Glabella nearly two-thirds as long as cranidium. Occipital ring about same length as anterior border medially. Behind LI, occipital ring is constricted and swings forwards where at posterolateral corner of LI it is half of its median length. Occipital furrow shallow and straight behind central glabellar lobe; at inner posterior corner of LI furrow deepens and remains so to axial furrow. Three pairs of unequal glabellar lobes. LI largest, length just under half that of glabella, with nearly straight lateral and posterior margin, and angular anterolaterally. SI shallow near axial furrow, deepest at inner anterior corner of LI where furrow bifurcates. Posterior branch runs backwards and curves inwards; anterior branch shallow and short, directed inwards and forwards. L2 just over half the length of LI, with more or less straight anterior and posterior margins. S2 straight, shallow near axial furrow, directed inwards and backwards at a smaller angle to the midline than SI. LI and L2 separated from central glabellar lobe by very faint furrow. L3 very small and delimited anteriorly by very faint S3. Oval areas situated adaxial to posterior branch of SI. Glabella gently convex (trans. and sag.). Axial furrow gently curved, convex outwards, shallowest opposite L2 and at posterior end of LI . Outside LI , axial furrow expanded into crescent-shaped paraglabellar areas. Anterior pit associated with slightly inwardly placed large tubercle situated on outer side of axial furrow, just anterior to S3. Preglabellar field separated from glabella by narrow, shallow furrow; preglabellar field of same length as anterior border and slopes gently backwards. Prominent, convex (sag.) anterior border separated from preglabellar field by well marked furrow which shallows abaxially. Posterior border narrow (exsag.) at axial furrow, widening abaxially. Back of palpebral lobe level with where SI meets axial furrow, anterior margin of palpebral lobe approximately level with anterior corner of L2. Palpebral lobe slopes inwards and merges with fixed cheek. Faint eye ridge running inwards and forwards from palpebral lobe to meet axial furrow just behind anterior pit. Anterior branch of facial suture runs in slightly sigmoidal curve to cut anterior margin in-line approximately with outer part of paraglabellar areas (preservation poor). Posterior branch of facial suture runs outwards and backwards (posterolateral parts of fixed cheeks not preserved). Free cheek narrow, extending into long, posteriorly directed genal spine reaching back to at least 6th thoracic segment. At least nineteen ventrally directed and slightly curved spines (just under 0 5 mm long) situated along border. Sculpture on cephalon of fine tubercles, about twenty per square mm; absent in furrows and very sparse on preglabellar field. Hypostoma not known. Thorax of Prionocheilus type, tuberculate except in furrows. Pygidium strongly curved anteriorly, gently curved posterior margin. Pygidium two and a half times as wide as anterior part of ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 347 axis; latter narrows evenly posteriorly and about four-fifths length of pygidium. Seven axial rings narrowing (sag.) posteriorly, 6th and 7th axial ring furrows weak to absent medianly. Terminal piece about one-quarter axial length, broadly rounded and sloping down steeply posteriorly. Six pairs of ribs separated by deep furrow which terminate just before lateral margins. Surface of pygidium tuberculate. Discussion. In most features the Maqao specimens are very close to Prionocheilus pulcher (Barrande 1852, pi. 19, figs 1-3, 6; see also Vanek 1965, p. 31, pi. 3, figs 3-5, pi. 4, fig. 1, text-fig. 5). Minor differences include the straighter anterior border in the Portuguese specimens and the less dense ornament, particularly on the fixed cheeks (cf. Vanek 1965, pi. 3, fig. 4). P. pulcher is known from the Vinice, Zahorany and Bohdalec Formations of Bohemia (Havlicek and Vanek 1966) but is only tentatively recorded from NW France (Henry 1980a, p. 187) where, however, the very similar P. verneuili Rouault is known from beds of Caradoc age to the south of Rennes (Riadan Formation) and possibly in the Crozon Peninsula (top of Postolonnec Formation) (Henry 1980a, p. 80). Henry stated, as Dean (1966, p. 303) had noticed, that the deformation of the Riadan specimens made some of the distinguishing features less certain; thus the status of verneuili is still in doubt. Prionocheilus costai (Thadeu, 1947) Plate 4, figs 1-9 1908 Calymene Costai Delgado, p. 57. *1947 Pharostoma Costai (Delgado); Thadeu, p. 218, pi. 2, figs 5-10. 1949 Pharostoma Costai (Delgado); Thadeu, p. 130. 1960 Pharostoma Costai (Delgado); Whittard, p. 138. 1976 Prionocheilus costai (Delgado); Hammann, p. 39, pi. 4, figs 46-51 ; pi. 5, figs 52-58; text-fig. 7. 1980a Prionocheilus costai (Delgado); Henry, p. 80. 1982 Prionocheilus costai ; Hammann et a/., p. 23. 1983 Prionocheilus costai (Delgado 1908); Hammann, p. 55, pi. 3, figs 34—36. 1984 Prionocheilus costai (Delgado); Rabano, p. 272. Material. Designated lectotype: SG 160 (Thadeu 1947, pi. 2, fig. 8). Paralectotypes ; SG 161-163 (Thadeu 1947, pi. 2, figs 7, 5 and 6 respectively). Three cephala with part of thorax; two nearly complete specimens; all preserved as internal and/or external moulds. Horizon and locality. All specimens are listed by Delgado (1908) as occurring in the 'Schistes culminants et schistes diabasiques', 250-300 m N 40° W of Louredo. Porto de Santa Anna Formation, late Caradoc- Ashgill age. The specimens are probably from the lower part of the unit, the Leira Ma Member. Description. Since the species has already been described by Thadeu (1947) and a further full account was given by Hammann (1976), only additional notes will be given here. Thadeu noted that only two pairs of glabellar furrows are present. However, on specimens SG 160 and 161, short, shallow inwardly directed S3 start level from where the eye ridge meets the axial furrow. The L3 thus defined are very short (exsag.) and less than half the length of L2. Over sixty downwardly directed short spines are present on the convex cephalic doublure and are continuous around the anterior margin of the cephalon. The axial rings and posterior and anterior bands on the pleurae carry numerous large tubercles. On specimen SG 162 these tubercles are seen to form the bases of short posterodorsally directed spines up to 0-5 mm in length. The pygidium is nearly three times as wide as long. The axis consists of five rings and a terminal piece, and four (five) backwardly directed pleural furrows with less distinct interpleural furrows. Discussion. Delgado first used the specific name costai in a faunal list (1908, p. 57). Although he did not describe or figure the species, later authors (see synonymy above) have credited it to Delgado, following the practice of Thadeu (1947) who was the first to formally describe the species. In this work authorship is attributed to Thadeu. All the differences between P. costai and P. pulcher (Barrande) (authorship attributed to Beyrich by Whittard 1960, p. 134) listed by Whittard (1960, p. 138) are now known not to be valid. Thus P. costai does possess a spinose cephalic border and preglabellar field (although in Hammann 1976, 348 PALAEONTOLOGY, VOLUME 34 text-fig. 7 the spines do not appear to be continuous around the anterior margin). Also the glabella has straighter sides in the Iberian species, there are fewer pygidial ribs and the granular sculpture is coarser. Genus actinopeltis Hawle and Corda, 1847 Type species. Actinopeltis globosa (Barrande, 1852). Actinopeltis tejoensis sp. nov. Plate 3, figs 7, 1 1, 14—16; Text-fig. 5 1908 Cheirurus sp. n.; Delgado, p. 80. Diagnosis. Species of Actinopeltis with the following characteristics: large, inflated, spherical anterior part of glabella; small isolated basal glabellar lobes separated from inflated glabellar lobe by long (x<3g.) furrow. Eyes situated far back, with eye ridge running to just anterior of S2. Pygidium with four pairs of spinose pleurae; posterior pair short to nearly as long as third pair. Type material. Holotype: SG 225, part and counterpart of nearly complete individual. Paratypes: QXP 2.006, 2.007, 2.026, internal moulds of incomplete cephala. QXP 2.041/2, 2.043/4, parts and counterparts of pygidia. Horizon and locality. SG 225 from ‘500 mN 52° E do logar do Pereiro (Magao)’. Other specimens from ' 1700 m N 57° E de pyr. de Queixoperra’, Macao. All specimens from 'Schistes a Orthis BerthoisV (Delgado 1908, p. 80), Queixoperra Member of the Cabego do Peao Formation, of Caradoc age. Derivation of name. From the Portuguese name Rio Tejo (River Tagus), into which drain the rivers of the Magao area. Description. The total length of specimen SG 225 is 28 mm of which the cephalon constitutes nearly 8 mm and the thorax about 12 mm. The specimen is obliquely deformed and crushed; the right side has been damaged. Cephalon dominated by large, approximately spherical anterior part of glabella, which is slightly wider than long and covered with small, closely spaced tubercles (barely visible on internal mould). Narrow (trans.) posterior part of glabella (though varies with preservation) comprises pair of small, nodular basal lobes, anterior to which a broad furrow separates them from inflated anterior part of glabella. Occipital furrow indistinct, merging with transverse furrow anterior to LI which are thus isolated at abaxial portions of broad (sag.) furrow. Occipital ring convex (sag. and trans.) and carrying similar ornament to glabella. Faint, shallow S2 start just posterior to where eye ridge meets axial furrow; S2 possibly directed slightly forwards but fracturing of glabella makes this uncertain. Short, shallow S3 situated approximately level to where lateral border furrow meets axial furrow. Axial furrows deep, without sculpture, and widely divergent. Cheeks small, triangular in outline, highest part lying adjacent to basal glabellar lobes. Lateral border strongly convex, of more or less constant width, extending with posterior border into long genal spine back to at least seventh thoracic segment. Genal spine oval(?) in cross-section and covered with small, densely EXPLANATION OF PLATE 4 Figs 1-9. Prionocheilus costai (Thadeu, 1947). 1-3, SG 160 (designated lectotype); internal mould and latex cast of external mould of nearly complete specimen, dorsal and anterodorsal views, x 2-7, x 3 and x 3 respectively, Porto de Santa Anna Formation, Bugaco, Caradoc/ Ashgill. 4, SG 2849; latex cast of external mould of incomplete cephalon and thorax, dorsal view, x 3, Porto de Santa Anna Formation, Bugaco, Caradoc/Ashgill. 5, SG 162; internal mould of cephalon and part thorax, dorsal view, x 3, Porto de Santa Anna Formation, Bugaco, Caradoc/Ashgill. 6, 9, SG 161 ; latex cast of external mould and internal mould of incomplete specimen, dorsal views, x 2-4, Porto de Santa Anna Formation, Bugaco, Caradoc/Ashgill. 7 and 8, SG 163; internal mould and latex cast of external mould of nearly complete specimen, dorsal views, x 3 and x 34 respectively, Porto de Santa Anna Formation, Bugaco, Caradoc/Ashgill. PLATE 4 ROMANO, Prionocheilus 350 PALAEONTOLOGY. VOLUME 34 text-fig. 5. Actinopeltis tejoensis sp. nov. Reconstruction of cephalon and pygidium. spaced tubercles. Prominent eye on short stalk, situated on highest part of cheek and fairly close to posterior border furrow. Eye lenses visible on QXP. 2.002. Well marked, low, eye ridge runs from eye to axial furrow just anterior to where S2 starts. Anterior branch of facial suture runs anterolaterally from the eye, approximately parallel to axial furrow, to margin. Posterior branch curves outwards and then backwards in even curve to cut lateral margin just anterior to base of genal spine. Lateral and posterior border furrows deep, except around base of genal spine. Posterior border convex (exsag.); narrowest near midline, widening evenly and gradually to genal spine. Free and fixed cheek covered with coarse pits of irregular size and distribution. Thorax consists of eleven segments. Axis narrow, strongly convex, and delimited by rather weak axial furrows. Axial rings gently curved forwards, broadest (sag.) along mid-line. Rings consist of convex (sag. and trans.) posterior band which broadens laterally. Anterior part of ring consists of broad, nearly flat band which narrows toward axial furrow where there is a shallow apodeme. Convex articulating half-ring separated from axial ring by marked change of slope. Pleurae consist of inner part (approximately one-third their transverse width) which is flat lying, and an outer spinose part which is outwardly inclined. Inner part of pleural segment consists of wide (exsag.), convex band bounded by narrow anterior and posterior bands. Wide band carries rows of pits, generally about 6, along midlength (appears as almost continuous groove on internal mould). Posterior pleural band is constricted at the fulcrum, where there is a prominent fulcral process and socket. Axial rings carry fairly dense ornament of faint tubercles while spinose parts of pleurae are very sparsely tuberculate. Pygidial axis about one and a half times as long as wide, delimited by shallow furrows which become less well defined posteriorly. Axis consists of 4 rings which decrease in length posteriorly. Ring furrows shallow medially. Posterior to fourth axial ring are pair of short, longitudinally aligned furrows which lie in series with the deeper inner parts of pleural furrows. Pleurae consist of inner flatfish part in which, from front to back, pleurae are progressively directed more posteriorly. Pleural furrows end in deep apodemal pits at axis. Outer parts of pleurae are slender spines; the first two being long, slightly curved and approximately of equal length; the third is a little shorter and curved proximally, while the fourth pair are much shorter (about one-quarter as long as second pair and less than half as long as third pair), more slender and directed backwards. In one specimen (QXP. 2.029a, b) the fourth pair of spines are considerably longer than in the other two examples and ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 351 extend posteriorly to terminate level with the tips of the third pair. Axis and flat pleural region carry a few scattered tubercles while the spines have an ornament similar to that of the glabella. Discussion. The genus had previously been recorded in Portugal from Valongo and Bugaco. At Valongo, Curtis (1961) recorded A. wattisoni sp. nov. (referred to Valongia wattisoni by Pfibyl and Vanek 1984) which differs from the Magao species in possessing 12 thoracic segments, a less swollen median glabellar lobe, more forwardly placed eyes and shorter genal spines. The ‘faint horizontal rib furrow’ which Curtis (1961, p. 9) described appears to consist of a row of faint pits as in the present material (see Curtis 1961, plate 4). Also, the pygidial rib furrows on wattisoni may be deformational features. At Bugaco, Delgado (1908, p. 57) listed a number of ‘ C he i rums ’ species from the upper Ordovician, some of which have been more recently described by Thadeu (1947). Of these, A. aff. completa (Barrande) (Thadeu 1947, pi. 3, figs 6 and 7) from the Porto de Santa Anna Formation has been recently compared to A. vercingetorix Pfibyl and Vanek (1969) by Hammann (1974, p. 105). Both of these species show a more forwardly placed eye and coarser sculpture on the glabella than Actinopeltis tejoensis. A. spjeldnaesi (Hammann 1972, p. 372, pi. I, fig. 3; 1974, p. 102, pi. 12, figs 200--207, text-fig. 38; 1976, p. 65, pi. 5, figs 64-68) from the upper Caradoc-lower Ashgill (Hammann et at. , 1982, p. 23) of Almaden (Sierra Morena, Spain) is fairly close to A. tejoensis. Rabano (1984) considered A. spjeldnaesi to be of Caradoc age. The Spanish species possesses more forwardly placed eyes, basal glabellar lobes which are not clearly delimited adaxially, and lacks the broad (sag.) furrow posterior to the swollen part of the glabella. The anterior pygidial spines are also more outwardly flexed in A. spjeldnaesi. It is of interest to note, however, that in both A. spjeldnaesi and A. tejoensis the fourth pair of pygidial spines are of variable length (see Hammann 1974, pi. 12, fig. 203; 1976, fig. 65; and PI. 3, figs 7, 1 1 herein). This characteristic appears to be independent of preservation. Delgado (1908) first recorded the Magao species as ‘ Cheirurus sp. n. aff. gryphus Barrande’ (Barrande 1872, pi. 3, figs 10-17) but the Bohemian specimens clearly differ from the Portuguese material in having less prominent eyes situated closer to the glabella and considerably shorter genal spines. Among other Bohemian species of Actinopeltis. the type species, A. globosa (Barrande 1852, pi. 35, figs 1-7, pi. 40, figs 26 and 27, pi. 43, fig. 27; Whittington 1968, text-fig. 7, p. 104), and the closely related A. rivanol (Snajdr, 1982) have less well delimited LI adaxially, and shorter genal and pygidial spines; the latter being well rounded distally. A. insocialis (Barrande, 1852, pi. 40, figs 28-31) does not possess genal spines, and the pygidial spines are shorter and rounded at the ends. Kielan (1959) assigned the specimens of A. insocialis from the Kraluv Dvur beds to a new species, A. barrandei, which also differs from Actinopeltis tejoensis in the absence of genal spines, the very small eye and the only slightly pointed pygidial spines. Actinopeltis sp. ‘a’ from the S. clavifrons Zone of Poland (Kielan 1959, pi. 24, fig. 4, text-fig. 36) has a similar structure to the pygidial axis as Actinopeltis tejoensis but the tuberculation is coarser and more densely spaced. Genus valongia Pfibyl and Vanek, 1984 Type species. Actinopeltis wattisoni Curtis, 1961. Valongia wattisoni (Curtis, 1961) *1961 Actinopeltis wattisoni sp. nov.; Curtis, p. 8, pi. 3, fig. 2, pi. 4, fig. 1. 1974 Actinopeltis wattisoni Curtis; Hammann, p. 105. 1982 Actinopeltis wattisoni ; Romano, p. 96. 1984 Valongia wattisoni (Curtis); Pfibyl and Vanek, p. 126, fig. 4, 3. Material. In 49184, holotype. part and counterpart of nearly complete specimen. Horizon and locality. Upper part of Valongo Formation, near Covelo; Llandeilo. 352 PALAEONTOLOGY, VOLUME 34 Discussion. The species was described and figured by Curtis (1961) and no further material has been found. Curtis assigned the species to Actinopeltis. Recently Pribyl and Vanek (1984) erected a new genus, Valongia , for this species since they considered the specimen showed important morphological features which distinguished it from those assigned to Actinopeltis. These included size of free cheek, course of facial suture, position of palpebral lobes, number of axial segments and structure of pygidial axis. The present author is in agreement with Pribyl and Vanek that Curtis’ species shows significant differences from those of Actinopeltis , but is more reluctant to follow their procedure of erecting a new genus, based on a single deformed specimen. However, for the present, their proposal is followed here. Acknowledgements. Drs J.-L. Henry and I. Rabano provided information on Armorican and Spanish material respectively; Dr M. Ramalho, Portuguese Geological Survey, allowed me to use material housed in Lisbon; Drs A. H. Cooper and T. Young gave me access to their personal collections. Miss P. Mellor typed the manuscript, Mr M. Cooper redrew Text-figures 1—4, and Miss G. Thompson photographed the specimens. The work was carried out with the aid of NERC Grant GR3/3786. REFERENCES barrande, J. 1846. Notice preliminaire sue le systeme Silurien et les trilobites de Boheme. Leipzig, vi + 97pp. - 1852. Systeme Silurien du centre de la Boheme. lere partie. Recherches paleontologiques. Vol. 7. Crustaces , Trilobites. Prague and Paris, xxx + 935 pp. - 1872. Systeme Silurien du centre de la Boheme. lere partie. Recherches paleontologiques , supplement au Vol. 1. Prague and Paris, xxx + 647 pp. Bergstrom, s. 1973. Organization, life and systematics of trilobites. Fossils and Strata. 2, 1-69. brenchley, p. j., romano, m. and Gutierrez- marco, J. c. 1986. Proximal and distal hummocky cross-stratified facies on a wide Ordovician shelf in Iberia. 241-255. In knight, r. j. and mclean, j. r. (eds). Shelf sands and sandstones. Canadian Society of Petroleum Geologists, Memoir 2. 1-347. cooper, a. h. 1980. The stratigraphy and palaeontology of the Ordovician to Devonian rocks of the area north of Domes (near Figueiro dos Vinhos), central Portugal. Unpublished Ph.D. Thesis, University of Sheffield. costa, j. c. da 1942. Notas sobre a familia Calymenidae. Boletim da Sociedade Geoldgica de Portugal , 1 (2), 91-100. courtessole, R., pillet, j. and vizca'ino, d. 1981. Nouvelles donnees sur la biostratigraphie de POrdovicien inferieur de la Montagne Noire. Revision des Taihungshaniidae de Megistaspis ( Ekeraspis ) et d ' Asaphopso- ides (Trilobites). Memoires de la Societe des Etudes Scientifiques de I’Aude, 32 pp. curtis, m. l. k. 1961. Ordovician trilobites from the Valongo area, Portugal. Cheiruridae, Pliomeridae. Boletim da Sociedade Geoldgica de Portugal , 14, 1-16. dean, w. t. 1964. The status of the Ordovician trilobite genera Prionocheilus and Polyeres. Geological Magazine , 101, 95-96. - 1966. The Lower Ordovician stratigraphy and trilobites of the Landeyran Valley and the neighbouring district of the Montagne Noire, south-western France. Bulletin of the British Museum ( Natural History), (Geology), 12, 247-353. - 1971. The trilobites of the Chair of Kildare Limestone (Upper Ordovician) of eastern Ireland. I. Palaeontographical Society Monograph , 60 pp. delgado, J. F. n. 1897. Fauna Silurica de Portugal. Novas observayoes acerca de Lichas ( Uralichas ) Ribeiroi. Diregcao dos trabalhos geologicos de Portugal, 34 pp. - 1908. Systeme Silurique du Portugal. Etude de stratigraphie paleontologique. Commission du Service geologique du Portugal , 245 pp. destombes, j. 1966. Quelques Calymenina (Trilobitae) de l'Ordovicien moyen et superieur de l'Anti-Atlas (Maroc). Notes et memoires du Service geologique du Maroc , 188, 33-52. fortey, R. a. and morris, s. F. 1982. The Ordovician trilobite Neseuretus from Saudi Arabia, and the palaeogeography of the Neseuretus fauna related to Gondwanaland in the earlier Ordovician. Bulletin of the British Museum ( Natural History), (Geology), 36, 63-75. gutierrez-marco, j. C., rabano, i., prieto, M. and martin, j. 1984. Estudio bioestratigrafico del Llanvirn y Llandeilo (Dobrotiviense) en la parte meridional de la Zona Centroiberica (Espana). Cuadernos geologic t Iberica , 9, 289-321. ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 353 hammann, w. 1972. Neue prepare Trilobiten aus dem Ordovizium Spaniens. Senckenbergiana lethaea , 53, 371-381. 1974. Phacopina und Cheirurina (Trilobita) aus dem Ordovizium von Spanien. Senckenbergiana lethaea , 55. I 151. 1976. Trilobiten aus dem oberon Caradoc der ostlichen Sierra Morena (Spanien). Senckenbergiana lethaea , 57, 35-85. 1977. Neue Calymenacea (Trilobita) aus dem Ordovizium von Spanien. Senckengergiana lethaea , 58, 91-97. 1983. Calymenacea (Trilobita) aus dem Ordovizium von Spanien; ihre Biostratigraphie, Okologie und Systematik. Abhancllungen senckenbergischen naturforschenden Gesellschaft , 542. 1 -177. — and henry, j.-l. 1978. Quelques especes de Calymenella, Eohomalonotus et Kerfornella (Trilobita, Ptychopariida) de l'Ordovicien du Massif Armoricain et de la Peninsule Ibcrique. Senckenbergiana lethaea , 59, 401-429. rabano, i. and Gutierrez- marco, J. c. 1986. Morfologia funcional del exosqueleto del genero Selenopeltis Hawle & Corda, 1847 (Trilobita, Odontopleurida ; Ordovicico). Paleontologia I Evolucio , 20. 203-211. — robardet, M. and romano, m. 1982. The Ordovician System in southwestern Europe (France, Spain and Portugal). International Union of Geological Sciences, 11. 1-47. havlicek, v. and vanek, j. 1966. The biostratigraphy of the Ordovician of Bohemia. Sbornik geologickych Ved, paleontologie, 8, 7-69. hawle, i. and corda, a. j. c. 1847. Prodrom einer Monographic der bohmischen Trilobiten. Prague, 176 pp. henry, j.-l. 1970. Quelques Calymenacea (Trilobites) de l’Ordovicien de Bretagne. Annales de Paleontologie, 56, 1-27. — 1 980«. Trilobites ordoviciens du Massif Armoricain. Memoires du Bureau de Recherches geologiques de Bretagne, 22, 1-250. — 19806. Evolution and classification of some Ordovician Calymenina (Trilobita). Geological Magazine, 117, 351-362. — melou, m., nion, j., Paris, f., robardet, m., skevington, d. and thadeu, d. 1976. L'apport de graptolites de la zone a G. teretiusculus dans la datation de faunas benthiques lusitano-armoricaines. Annales de !a Societe geologique du Nord , 96, 275-318. — and morzadec, p. 1968. Sur la presence du sous-genre Phacopidella ( Prephacopidella ) Destombes, 1963 (Trilobites) dans les schistes ordoviciens du Portugal. Compte rendu sommaire des seances de la Societe geologique de France, 5, 158-159. — nion, j., Paris, f. and thadeu, d. 1976. Chitinozoaires, Ostracodes et Trilobites de l'Ordovicien du Portugal (serra de Bugaco); essai de comparaison et signification paleogeographique. Commission du Service geologique du Portugal , 57, 303-345. — and romano, m. 1978. Le genre Dionide Barrande, 1847 (Trilobite) dans l'Ordovicien du Massif Armoricain et du Portugal. Geobios, 11, 327-343. — and thadeu, d. 1971. Interet stratigraphique et paleogeographique d'un microplancton a Acritarchs decouvert dans l’Ordovicien de la Serra de Bupaco (Portugal). Comptes rendus hebdomadaires des seances de I'academie des sciences, 272. 1343-1346. ingham, j. k. 1977. The upper Ordovicien trilobites from the Cautley and Dent districts of Westmorland and Yorkshire. Part 3. Palaeontographical Society Monograph, 89-121. K.IELAN, z. 1959. Upper Ordovician Trilobites from Poland and some related forms from Bohemia and Scandinavia. Palaeontologia Polonica, 11, 1-198. Mitchell, w. I. 1974. An outline of the stratigraphy and palaeontology of the Ordovician rocks of central Portugal. Geological Magazine, 111. 385-396. owen, a. w. and bruton, d. l. 1980. Late Caradoc-early Ashgill trilobites of the central Oslo Region. Palaeontological Contributions from the University of Oslo , 245, 1-63. Paris, F. 1979. Les chitinozoaires de la formation de Louredo, Ordovicien superieur du synclinal de Bupaco (Portugal). Palaeontographica, Abteilung A. 164, 24-51. — 1981 . Les chitinozoaires dans le Paleozo'ique du sud-ouest de l’Europe. Memoires de la Societe geologiques et miner alogique de Bretagne, 26, 1^112. pribyl, f. and vanek, j. 1969. Uber einige Trilobiten des mittelbohmischen Ordoviziums. Vestnik Ustfedniho ustavu geologickeho, 44, 365-374. — and 1984. Observations on some Bohemian and foreign cheirurid trilobites. Paldontologische Zeitschrift, 58, 1 19-130. 354 PALAEONTOLOGY, VOLUME 34 rabano, i. 1983. The Ordovician trilobite Hungioides Kobayashi (Asaphina, Dikelokephalinidae) from Spain. Geobios , 16. 431-441. - 1984. Trilobites ordovicicos del Macizo Hesperico espanol: una vision bioestratigrafica. Cuademos de geologia Iberica, 9, 267-287. rebelo, J. and romano, m. 1988. A contribution to the lithostratigraphy and palaeontology of the Lower Palaeozoic rocks of the Moncorvo region, northeast Portugal. Comunicagoes dos Servigos Geologicos de Portugal , 72 (for 1986), 45-57. ribeiro, c. 1853. On the Carboniferous and Silurian formations in the neighbourhood of Bussaco in Portugal. With notes and a description of the animal remains by D. Sharpe, J. W. Salter and T. R. Jones; and an account of the vegetable remains by C. J. F. Bunbury. Quarterly Journal of the Geological Society of London, 9, 135-161. romano, m. 1976. The trilobite genus Placoparia from the Ordovician of northern Portugal. Geological Magazine, 113, 11-28. 1982. The Ordovician biostratigraphy of Portugal - A review with new data and re-appraisal. Geological Journal, 17,89-110. brenchley, p. j. and mcdougall, n. d. 1986. New information concerning the age of the beds immediately overlying the Armorican Quartzite in central Portugal. Geobios, 19, 421-433. - and diggens, j. n. 1976. The stratigraphy and structure of Ordovician and associated rocks around Valongo, north Portugal. Comunicagdesi dos Servicos Geologicos de Portugal, 57 (for 1973-74), 22-50. rouault, m. 1847. Extrait du Memoire sur les Trilobites du departement d’Hle-et-Vilaine. Bulletin de la Societe geologique de France, 4, 309-328. 1849. Memoire: 1° sur la composition du test des Trilobites; 2° sur les changements de forme dus a des causes accidentelles, ce qui a pu permettre de confondre des especes differentes. Bulletin de !a Societe geologique de France, 6, 67-89. - 1851. Memoire sur le terrain paleozoi'que des environs de Rennes. Bulletin de la Societe geologique de France, 8, 358-399. sdzuy, k. 1957. Bermerkungen zur Familie Homalonotidae (nrit der Beschreibung einer neuen Art von Calymenella). Senckenbergiana lethaea, 38, 275-290. sharpe, d. 1849. On the geology of the neighbourhood of Oporto, including the Silurian Coal and Slates of Vallongo. Quarterly Journal of the Geological Society of London, 5, 142-153. siveter, d. j. 1977. The Middle Ordovician of the Oslo region, Norway. 27. Trilobites of the family Calymenidae. Norsk Geologisk Tidsskrift, 56, 335-396. snajdr, m. 1956. The trilobites from the Drabov and Letna Beds of the Ordovician of Bohemia. Sbornik Ustfedniho ustavu geologickeho , 22, 477-533. - 1982. New trilobites from the Bohdalec Formation (Berounian) in the Barrandium. Vestnik Ustfedniho ustavu geologickeho , 57, 227-230. thadeu. D. 1947. Trilobites do Silurico de Louredo (Bugaco). Boletim da Sociedade Geologica de Portugal, 6, 217-236. - 1949. Calimemdeos portugueses. Boletim da Sociedade Geologica de Portugal, 8. 129-134. - 1956. Note sur le silurien beiro-durien. Boletim da Sociedade Geologica de Portugal, 12, 1-38. thomas, a. 1977. Classification and phylogeny of homalonotid trilobites. Palaeontology. 20. 159-178. tromelin, G. and lebesconte, p. 1876. Essai d’un catalogue raisonne des fossiles silurien des departements de Maine-et-Loire, de la Loire-Inferieure ct du Morbihan, avec des observations sur les terraines paleozoiques de l’ouest de la France. Compte rendu 4eme session de /’association frangaise pour I’avancement des sciences, 601-661. vanek, j. 1965. New species of the suborder Calymenia Swinnerton, 1915, (Trilobita) from the Barrandian area. Sbornik geologickych Ved, paleonto/ogie, 6, 21—37. verneuil, p. e. p. and barrande. j. 1855. Descriptions des fossiles trouves dans les terrains silurien et devonien d'Almaden, d’une partie de la Sierra Morena et des montagnes de Tolede. Bulletin de la Societe geologique ; de France, 12. 904-1025. whittard, w. f. I960. The Ordovician trilobites of the Shelve Inlier, West Shropshire. Part 4. Palaeonto- graphical Society Monograph, 117-162. Whittington, h. b. 1965. The Ordovician trilobites of the Bala area, Merioneth. Part 2. Palaeontographical Society Monograph, 33-62. - 1968. The Ordovician trilobites of the Bala area, Merioneth. Part 4. Palaeontographical Society Monograph, 93-138. ROMANO: PORTUGUESE ORDOVICIAN TRILOBITES 355 young, t. p. 1985. The stratigraphy of the upper Ordovician of central Portugal. Unpublished Ph.D. Thesis, University of Sheffield. - 1988. The lithostratigraphy of the upper Ordovician of central Portugal. Journal of the Geological Society of London , 145, 377-392. 1989. Eustatically controlled ooidal ironstone deposition: facies relationships of the Ordovician open- shelf ironstones of Western Europe. 51-63. In young, t. p. and taylor, w. e. g. (eds). Phanerozoic Ironstones. Geological Society Special Publication, 46, 251 pp. 1990. Ordovician sedimentary facies and faunas of southwest Europe: palaeogeographic and tectonic implications. 421-430. In mckerrow, w. s. and scotese, c. r. (eds). Palaeozoic Palaeogeography and Biogeography . Geological Society Memoir, 12, 435 pp. MICHAEL ROMANO Earth Sciences Unit University of Sheffield Typescript received 7 August 1989 Beaumont Building, Brookhill Revised typescript received 11 April 1990 Sheffield S3 7HF, UK NOTE ADDED IN PROOF After this manuscript was completed the author was kindly sent an important paper by Rabano (1990) on middle Ordovician trilobites of the Central Iberian Zone in Spain. Rabano recognized S. lusitanica (Thadeu, 1949) as a distinct species and pointed out that it was distinguishable from S. salteri by a number of cephalic features (Rabano 1990, pp. 120-122). However, the variability of the sculpture and the frequently deformed nature of the Portuguese specimens do not allow confident separation of these two forms. Rabano also stated that S. lusitanica is a characteristic species of the Lower Llanvirn of the Central Iberian Zone, and designated the specimen figured by Thadeu (1949), pi. 2, fig. 1) as the lectotype. This specimen (SG 71688) was collected from ‘ 1650 m [not 1680 m] S 20° O da piramide de Santa Justa (Valongo)'. From the same locality, Eccoptochile almadenensis, E. cf. clavigera and Eodolmanitina Ideslombesi destombesi occur. The known ranges of these three species indicate a Llandeilo age for this assemblage. REFERENCE rabano, i 1990. Trilobites del Ordovicico Medio del sector meridional de la zona Centroiberica espanola. Publicaciones especiales de! Boletin Geologico y Minero , 100 (for 1989), 1-233. CAMBROCLAVES AND PARACARINACHITIDS, EARLY SKELETAL PROBLEMATICA FROM THE LOWER CAMBRIAN OF SOUTH CHINA by s. conway morris and CHEN menge Abstract. Cambroclaves are a major group of sclerite-bearing metazoans, known from the Lower Cambrian of China (south China, Xinjiang), USSR (Kazakhstan) and Australia. Zhijinites longistriatus Qian is redescribed on the basis of abundant material from the Hongchunping Formation at Maidiping, Sichuan. Sclerites show extensive morphological variability and have a taphonomic history of endolithic infestation and diagenetic phosphatization, the latter leading to replication of wall ultrastructure. Deiradoclavus trigonus gen. et sp. nov. and Deltaclavus graneus gen. et sp. nov. are younger cambroclaves recovered from the Guojiaba Formation near Kuanchuanpu, Shaanxi, and the Shuijingtuo Formation at Taishanmiao, Hubei. Both taxa bore a cataphract scleritome, composed of interlocking sclerites. In Deltaclavus articulated series of sclerites include ‘arm-like’ structures. Paracarinachitids may be related to cambroclaves, and are described on the basis of isolated sclerites of Paracarinachites spinus Yu from the Yuhucun Formation of Meishucun, Yunnan. Protopterygotheca leshanensis Chen from the Hongchunping Formation of Maidiping is included in the paracarinachitids, and is described on the basis of isolated sclerites bearing prominent flanges on either side of the spatulate axis. The primary function of the scleritomes of cambroclaves and paracarinachitids may have been protective, but in the absence of intact scleritomes both the palaeoecology and affinities of these groups are uncertain. The new class Cambroclavida is proposed. The irruption of skeletal faunas close to the Precambrian-Cambrian boundary (Conway Morris 1987, 1989a) has attracted wide attention on two principal counts. First, there is debate as to whether the acquisition of skeletons (a) was mediated by extrinsic factors, such as changes in ocean chemistry, and/or (b) represents a biological response such as providing a defensive cover against durophagous predators and offering greater support to soft tissues. The second point of discussion is the part these early skeletal faunas played in the major adaptive radiations that are often referred to as the ‘Cambrian explosion'. Evidence for metazoan diversification is readily apparent from both the record of trace fossils (Crimes 1989) and Burgess Shale-type soft-bodied assemblages (Conway Morris 1989^), but by taphonomic necessity the bulk of the evidence must come from skeletal remains. It has become apparent that many of the earliest of these forms are of problematic affinity, bearing no clear relationship to known phyla. Although some taxa continue to languish in a taxonomic limbo, recent work has demonstrated the presence of several major groups. All are extinct, but their status probably deserves the cognomen, in terms of orthodox taxonomy, of phylum on account of their distinctive body-plans (but see Conway Morris 1989c). Such groups now include the tommotiids, coeloscleritophorans, anabaritids, cambroclaves and the possibly related paracarinachitids, the last two of which are the subject of this paper. With the exception of the tubicolous anabaritids, all these groups share a skeletal arrangement of sclerites that presumably coated the exterior body to form the scleritome. Reconstruction of the original scleritome ideally relies on articulated material such as might occur in a Konservat-Lagerstatte. With the halkieriids (Coeloscleritophora) comparisons with the Burgess Shale Wiwaxia allowed the latter to act as a model for scleritome reconstruction (Bengtson and Conway Morris 1984), and this may now be tested further on account of the discovery of articulated halkieriids in the Lower Cambrian of north Greenland (Conway Morris and Peel 1990). In the remaining cases, however, sclerite arrangement must be inferred from either rare specimens showing fusion or functional analysis of areas of articulation between adjacent sclerites. I Palaeontology, Vol. 34, Part 2, 1991, pp. 357-397, 9 pls.| © The Palaeontological Association 358 PALAEONTOLOGY, VOLUME 34 CAMBROCLAVES Cambroclaves are represented by a distinctive group of sclerites that consist of a circular to oval base that bears a spine, usually elongate. They are recorded from the Lower Cambrian of China (Text-fig. 1), Kazakhstan, and Australia. However, they appear to be unknown from other parts of the world, including the equivalent-aged sections in Mongolia, the Siberian platform, and Canada. In the past, it has been found useful to make a distinction between sclerites with a sub-circular base bearing a prominent spine (zhijinitid morph) and those with a more elongate base, often having a dumb-bell shape, with the spine arising from the anterior half (cambroclavid morph). The orientation of cambroclaves follows that outlined by Mambetov and Repina (1979, fig. 8), with the prominent spine assumed to arise from the anterior end of the dorsal surface. In the absence of any articulated scleritomes, overall sclerite attitude with respect to the entire animal, assumed to be a bilaterally symmetrical metazoan, is not known. Scleritomes that appear to have consisted of either entirely zhijinitids or predominantly cambroclavids with a small proportion of zhijinitids are both known. The former type is documented in some detail here on the basis of Chinese material of Zhijinites longistriatus Qian, while cambroclavids have received extensive study on the basis of well-preserved material of a new species of Cambroclavus from South Australia (Bengtson et al. 1990). Reconstructions of cambroclave scleritomes may also be inferred with varying degrees of confidence from descriptions in the literature. However, the wide morphological variability of the sclerites has resulted in excessive use of form-taxa by some authors, and one aim of the extensive synonymy list proposed here for Z. longistriatus is to encourage a classification designed to lead to more reliable scleritome reconstructions. In addition to the two basic sclerite types mentioned above, two other variants are reported on the basis of scleritomes inferred to have consisted of (a) oval sclerites bearing a ridge- like spine ( Deiradoclavus gen. nov.), and (b) sclerites with a predominantly triangular outline (Deltaclavus gen. nov.). The first report of cambroclaves was by Zhong [Chen] (1977), who mentioned and illustrated material from Guizhou and Sichuan provinces as Zhijinites sp. However, as none of the formalities of his taxonomic description accords to those laid down by the International Commission for Zoological Nomenclature, this reference to Zhijinites must be taken as a nomen nudum. Formal descriptions of Zhijinites (Z. longistriatus , Z. minutus ), on the basis of material from near the town of Zhijin, west of Guiyang in Guizhou Province (Text-fig. 1), were made available shortly afterwards by Qian ( 1978a, see also 1978/)). This and adjacent localities in Guizhou have continued to provide abundant material of Zhijinites (Chen 1979; Qian and Yin 1984 a, b; Wang et al. 1984a, b), as have other localities (Text-fig. 1) in Sichuan (Chen 1979; Yin et al. 1980a, b ; Yang et al. 1983; He et al. 1984), Yunnan (Jiang 1980, 1982; Luo et al. 1984a) and Hubei provinces (Chen 1979; Qian et al. 1979). Numerous species of Zhijinites have been erected, most of which are probably synonymous (see below). In addition the genera Heterosculpotheca Jiang, 1982 and Parazhijinites Qian and Yin 1984/), are both regarded as junior synonyms of Zhijinites. Furthermore, what are evidently specimens of Zhijinites have been referred to the conodont-like Fomitchella (Yin et al. 1980a), the halkieriid Sachites (Yin et al. 1980a) and the hyolith Allatheca (Yang et al. 1983). Notwithstanding the taxonomic confusion that appears to accompany our existing understanding of Zhijinites from south China, in the majority of samples it seems that the sclerites are derived from the dispersal of a scleritome composed exclusively of zhijinitid morphs. There is, however, evidence text-fig. 1. Distribution of cambroclaves and paracarinachitids in China. The numbers refer to localities from which material described herein has been recovered: 1, Meishucun, Yunnan (see Text-fig. 6); 2, Maidiping, Sichuan (see Text-fig. 3); 3, Taishanmiao, Hubei (see Text-fig. 5); 4, Xuanjiangping, Shaanxi; 5, Liangshan, Shaanxi (for both see Text-fig. 4). In addition to these occurrences cambroclaves are reported also from Xinjiang province, China (Qian and Xiao 1984), Kazakhstan, USSR (Mambetov and Repina 1979) and Australia (Bengtson et al. 1990). Paracarinachitids are reported also by Kerber (1988) from southern France. CONWAY MORRIS AND CHEN : CAMBRIAN PROBLEMATIC A 359 360 PALAEONTOLOGY, VOLUME 34 that in some stratigraphically higher cambroclaves such morphs persisted as a minor component in a scleritome otherwise composed of cambroclavid sclerites. For example, in reconstructing an Australian cambroclave the occasional zhijinitid morphs were regarded as an integral part of the scleritome (Bengtson et al. 1990). This decision was based on the presence of intermediate forms which showed a progressive reduction of the posterior end, general similarities of ornamentation, and consistent co-occurrence with the more abundant cambroclavid morphs. It is for this reason that a number of Chinese species placed in Zhijinites are regarded tentatively as more probably being derived from a cambroclavid scleritome, rather than once constituting a scleritome composed of only zhijinitids. One such instance might be Z. intermedins , and perhaps Z. claviformis , from the Lower Cambrian Yurtus Formation of Xinjiang Province (Qian and Xiao 1984) that could be attributable to the same scleritome that yielded elements referred to Cambroclavus ( = Sugaites ) bicornis. Similarly, zhijinitid morphs (Z. claviformis , Z. cordiformis) co-occurring with cambroclavid sclerites in the upper Lower Cambrian (Damao Group) of Yaxian County, Hainan Island (Text- fig. 1 ) (Jiang and Huang 1986) may all be derived from the same scleritome. Duan (1984) described zhijinitid-morphs, which he attributed to Tcinbaoites ( T . porosus , T. spiculosus) from the Xihaoping Formation of Hubei Province (Text-fig. 1 ). However, they co-occur with a plethora of cambroclavid taxa (nine species placed in Cambroclavus , Sinoclavus and Phyllochiton) that may all be derived from a single scleritome. What may be similar cambroclavids have also been recorded from strata of Atdabanian age from a section at Xiaoyangba in Zhenba County, Shaanxi (Text-fig. 1) (Xie 1988, fig. 1). It is clear, therefore, that until a more rigorous approach to scleritome reconstruction in Chinese cambroclaves is undertaken, little headway can be expected in draining the swamp of form- taxonomy that mires present efforts to introduce a degree of order. This is also exemplified in the interpretation of cambroclaves from Maly Karatau and Talassky Ala-Too, Kazakhstan (Mambetov in Mambetov and Repina 1979; see also Missarzhevsky and Mambetov 1981). Mambetov provided the first description of the cambroclave morph in the form of Cambroclavus antis. In addition a zhijinitid-morph was distinguished as C. undulatus , which both Jiang (1982) and Duan (1984) transferred to Zhijinites. As the stratigraphic range of C. antis and Z. undulatus only partly overlap, it is possible that each was derived from a separate scleritome. It is also conceivable, however, that the concept of Z. undulatus is incompletely understood. Zhijinitid-morphs that co-occur with C. antis (Mambetov in Mambetov and Repina 1979, pi. 13, figs 2, 10, 13) appear to differ from those collected from a separate horizon (Mambetov in Mambetov and Repina 1979, pi. 13, figs 1, 4, 11, 12), and it may be that only the latter belong to Zhijinites s.s., having been derived from a scleritome composed solely of zhijinitids. Yet another type of cambroclave, Pseudoc/avus singularis (Mambetov in Mambetov and Repina 1979, pi. 14, figs 5, 7, 10, 11) represents a distinctive variety of cambroclave, apparently unique to Kazakhstan. The status of the remaining taxon, C. clavus , described by Mambetov in Mambetov and Repina (1979, pi. 13, figs 3, 5, 7-9), is somewhat uncertain, but it seems to be a zhijinitid-like morph with a conspicuous spine and diminished base. Although sclerites are almost invariably found isolated owing to both post-mortem decay of any intervening soft tissue and the exigencies of the extraction technique of acid digestion and sieving, their original arrangement may be preserved in rare instances as fused associations. These were noted first by Mambetov (in Mambetov and Repina 1979, pi. 14, figs 6, 8, 9) who depicted sclerites of C. antis deployed in orderly rows. Amongst the well-preserved cambroclavids from the Atdabanian Ajax Limestone of South Australia rare examples of two sclerites fused in a longitudinal direction were noted (Bengtson et al. 1990). Such specimens confirm the function of the anterior ventral and posterior dorsal facets. In addition, outline shape of these Australian cambroclavids shows how they would have interlocked to form a cataphract (chainmail-like) arrangement (Text-fig. 1 1 b), and this lends credence to the idea that the sclerites might have formed a protective coating on a metazoan. Analogies to the wiwaxiids and halkieriids, where entire specimens are known, might suggest that cambroclaves were also worm or slug-like. However, unusual examples of articulated series of sclerites in Deltaclavus graneus gen. et sp. nov. (see below) and lack of knowledge of associated soft parts makes these analogies distinctly tentative. CONWAY MORRIS AND CH EN : C A M B R I AN PROBLEMATICA 361 The analysis of the new Australian species of Cambroclavus also revealed the sclerites to have a very wide degree of morphological variability, encompassing not only radical differences in typical cambroclavid types, but also reduction towards a zhijinitid condition of sub-circular base surmounted by prominent spine (Bengtson et a/. 1990). While the possibility cannot be dismissed that any one individual bore a restricted range of sclerite morphs, it is considered more likely that mutual accommodation in shape between adjacent sclerites would have led to extensive variation across the body. For the most part, published illustrations of cambroclaves are insufficient to gauge reliably the extent of morphological variability. However, the likelihood of extensive synonymies within suites of sclerites, ostensibly belonging to a plethora of nominal taxa, from localities such as Hubei (Duan 1984; Qian and Yin 1984/)) and Xinjiang provinces (Qian and Xiao 1984) argue for morphological variability being widespread in cambroclaves. PARACARINACH1TIDS These distinctive sclerites have been reviewed critically by Qian and Bengtson (1989), who recognized four species (the type species Paracarinachites sinensis , and also P. columellatus, P. parabolicus and P. spinus). Paracarinachitids have a narrow spatula-like form with a median row of spines, and evidently grew incrementally. Although best known from South China (Qian and Bengtson 1989), sclerites from the Montagne Noire of southern France are important because they show also a flange (Kerber 1988). Here, we describe material probably attributable to P. spinus , but differing in occurring as single sclerites rather than articulated associations where the sclerites are arranged in a row. In Protopterygotheca leshanensis Chen in Qian et al ., 1979 sclerite form is especially clear on account of well-developed flanges, but the diagnostic paracarinachitid spines are only seldom evident. Ultimately it may transpire that Protopterygotheca Chen, 1979, should be taken as a senior synonym of Paracarinachites Qian and Jiang, 1982 (see Qian and Bengtson 1989), but this is premature on existing evidence. Qian and Bengtson (1989) proposed that para- carinachitids and zhijinitids are closely related. This may well be correct for P. spinus , but the remaining three species of Paracarinachites (and the related sclerites of Protopterygotheca and Scoponodus) may be better treated as a group distinct from the cambroclaves. In addition, the possibility that Ernogia (see Qian and Bengtson 1989, pp. 100-102, figs 64 and 65) be included in the roster of paracarinachitids may also bear further consideration. Qian and Bengtson (1989) considered this option briefly, on account of both overa'l shape and growth incrementation. The nodular ornamentation on the exterior of Ernogia stands in contrast to the median spines of Paracarinachites , but this difference may be of relatively minor importance given the more or less smooth appearance of most sclerites of P. leshanensis described herein. STRATIGRAPHY AND LOCALITIES The material described herein comes from the following horizons and localities: 1. Zhijinites longistriatus. All the material illustrated here was obtained from Beds 36 and 37 of the Maidiping Member, Hongchunping Formation (Text-fig. 2), exposed at the Maidiping section near to Emei, Sichuan Province (Text-fig. 3). This section is one of many Precambrian-Cambrian Boundary sections located around Emei Mountain (see He et al. 1984, fig. 4-1), but its stratigraphy and fossil content have received particularly detailed attention by several workers (e.g. Yin et al. I980«, />; He and Yang 1982; He et al. 1984). Zhijinitids were reported by Zhong [Chen] (1977, p. 123), Chen (1979, p. 281), and Yin et al. (1980«, pp. 178-179; see also synonymy list for apparently erroneous assignments to Fomitchella and Sachites hastatus). Comparable material was obtained from the Dananguo section, Liangshan (Text-fig. 4), located 3 km from the Oriental Instrument Plant factory and 10 km north-west of Hanzhong (Fu 1983, fig. 1 ; Ding et al. 1983, fig. 2). Here the Yangjiakon member of the Dengying Formation (Text-fig. 2) (Fu’s placement of this part of the section in the Guojiaba Formation is less likely, because correlations suggest it to be equivalent to the deeper water Kuanchuanpu Member exposed west of Hanzhong in the Ningqian area (see Xing and Yue 1984; Conway Morris and Chen 1989)) is composed near its top of sandy glauconitic limestones. It yields abundant zhijinitids (Fu 1983; Ding et al. 1983) which, however, are generally somewhat smaller than those recovered 362 PALAEONTOLOGY, VOLUME 34 ^^^Region Strata East Yunnan (Meishucun) West Hubei (Taishanmiao) Southwest Shaanxi (Xuanjiangping & Liangshan) Southwest Sichuan (Maidiping) Qiongzhusi Yu anshan Upper Upper Stage Qiongzhusi Member Shuijingtuo Guojiaba Member Jiulaodong Member Fm. Badaowan Fm. Fm. Lower Fm. Lower Z < Member Member Member DC DO S < O 0> cn CD 55 c 3 O Dahai Member Tianzhu - Maidiping Member 3 SZ 0) Zhongyicun shan Member SE Member Dengying (Huangshan- Dengying Kuanch- Hongchun- Yuhucun dong uanpu ping Fm. Xiaowai - toushan Member Fm. Member) Fm. Member Fm. z < z 0) cn CD 55 CD X Baiyanshao Member Biamatuo Member Maoergang Member w c ’>* cn c 0) -Jiucheng Member Shibantan Member o text-fig. 2. Summary stratigraphic chart for the Sinian-Cambrian boundary sequence in South China (see Text-fig. I for position of localities). Asterisked arrows refer to stratigraphic horizons of occurrence of cambroclaves and paracarinachitids, see text for further details. Based on table 8-1 of Ding et al. (1984). from Maidiping. In addition to zhijinitids the productive sample yielded also abundant Hyolithellus, rare Protohertzina (cf. Fu 1983) and spicules. From the overlying shales of the Guojiaba Formation (equivalent to the Qiongzhusi Formation) at Liangshan Chen (1985) has described trilobites. 2. Deiradoclavus trigonus gen. et sp. nov. Abundant specimens were recovered from a calcareous horizon (thin limestones overlying beds with calcareous concretions) in the otherwise largely clastic Guojiaba Formation (Text-fig. 2) (see Chen et al. 1975, fig. 2). The outcrop forms part of the Xuanjiangping section, and is located in a stream about 1 500 m southeast of the hamlet of Xuanjiangping (Text-fig. 4). The underlying strata, especially of the Kuanchuanpu Formation, have received extensive attention on account of their abundant small skeletal fossils and proximity to the Precambrian-Cambrian Boundary (e.g. Shizhonggou section, Xing and Yue 1984; Xing et al. 1984; Piaojiaya section (Text-fig. 4), Conway Morris and Chen 1989). However, apart from reports of Tommotia (Qin and Yuan 1984), the small skeletal fossils of the Guojiaba Formation appear to have received little attention. In addition to Deiradoclavus the sample yielded Tannuolina zhangwentangi (Conway Morris and Chen 19906; Qin and Yuan’s (1984) report on Tommotia may refer to this taxon), Pelagiella sp., Actinotheca , bradoriids, unidentified tubes, and other problematica. 3. Deltaclavus graneus gen. et sp. nov. Abundant material was recovered from the lower Shuijingtuo Formation (Text-fig. 2; see also Zhou and Xu 1987), exposed in the Taishanmiao section, near the village of Taishanmiao, Hubei Province (Text-fig. 5). The sample was a fallen block in a roadside quarry about 250 m south of the Precambrian-Cambrian Boundary as locally defined within the Tianzhushan member of the CONWAY MORRIS AND CHEN: CAMBRIAN PROBLEMATICA 363 text-fig. 3. Locality map of region around Emei (Sichuan), showing location of Maidiping section (asterisked). Based on fig. 1 of Yin et al. (19806) and fig. 4.1 of He et al. (1984). Dengying Formation (see Conway Morris and Chen (1990a) for further details of locality and associated fauna, including the problematic Blastu/ospongia). Despite a variety of stratigraphic terms, for the most part correlation of the Lower Cambrian around the Yangtze platform of South China (Text-fig. 2) is reasonably straightforward (e.g. Xing el al. 1984). Accepting the existing scheme of correlations it would appear that the zhijinitids from Maidiping (Sichuan) and Meishucun (and Liangshan, Shaanxi) are of approximately the same age, falling into the lower part of the Paragloborilus-Siphogonuchites assemblage. Deiradoclavus gen. nov. from the lower Guojiaba Formation (Shaanxi) and Deltaclavus gen. nov. from the lower Shuijingtuo Formation are somewhat younger, and about the same age as one another. These two genera appear to be closely related, and it is to be expected that similar canrbroclaves will be found in due course in equivalent strata such as the Badaowan member of the Qiongzhusi Formation (Yunnan) and the Jiulaodong Formation (Sichuan) (Text-fig. 2). These are predominantly clastic units, but so far investigation of carbonate horizons in both formations (Conway Morris and Chen, 19906; see also Qian and Bengtson 1989) has not revealed any material. 4. Paracarinachites spinas. Numerous specimens were recovered from dolomites of Bed 7 of the Zhongyicun member, Yuhucun Formation (Text-fig. 2) exposed at the Xiawaitoushan section of the Kunyang Phosphorite Mine, at Meishucun, Yunnan Province (Text-fig. 6) (Luo et al. 1980, 1982, 1984a, b). The Meishucun locality has attracted widespread interest on account of it being the Chinese stratotype candidate for the Precambrian-Cambrian Boundary. To date, this taxon has been chiefly documented from articulated series of sclerites from this horizon and the overlying Bed 8 of the Dahai Member (Text-fig. 2) (see Qian and Bengtson 364 PALAEONTOLOGY, VOLUME 34 text-fig. 4. Locality map of region near Hanzhong (Shaanxi), showing locations of section near Liangshan (asterisk on main map) and near Xuanjiangping (asterisk on inset map), and also Piaojiaya section (see Conway Morris and Chen 1989). Based on fig. 6-1 of Xing and Yue (1984). 1989; He and Xie 1989), and the abundant isolated sclerites are less well known (He and Xie 1989, pi. I, figs 20 and 22). 5. Protopterygotheca leshanensis. This material, which only seldom has the median spines diagnostic of other paracarinachitids present, was obtained from the same horizons at Maidiping, Emei (Text-fig. 3) as the specimens of Z. longistriatus (see above). The type material of this taxon comes from the Tianzhushan section in Hubei (Text-fig. 5). SYSTEMATIC DESCRIPTIONS Class CAMBROCLAVIDA nov. Diagnosis. Calcareous) ?) sclerites with variously shaped base bearing an elongate spine, forming a scleritome that ranged from articulated cataphract array to individual sclerites studding surface, apparently separated by unmineralized tissue. CONWAY MORRIS AND CHEN: CAMBRIAN PROBLEMATICA 365 text-fig. 5. Locality map of region around Yichang (Hubei), showing location of Taishanmiao and Tianzhushan sections (asterisked). Based on fig. 2-1 of Chen et al. (1984). Family zhijinitidae Qian, 1 978c/ Emended diagnosis. Sclerites composed of base and elongate spine, now hollow, formerly filled with soft tissue? Base may be subcircular to elongate. In the latter case it may have prominent constriction imparting dumb-bell shape, and usually articulatory facets. Dorsal surface or- namentation frequently of radiating low ridges. Spine elongate, usually recurved posteriorly, transverse section varies from circular to elongate. Original composition probably calcareous. Component genera. (1) Zhijinites Qian, 1978a (junior synonyms include Sachites sensu Yin et al ., 1980a, non Meshkova, 1969; Fomitchella sensu Yin et al ., 1980a, non Missarzhevsky in Rozanov et al ., 1969; Heterosculpotheca Jiang, 1982; Parazhijinites Qian and Yin, 19846). (2) Cambroclavus Mambetov in Mambetov and Repina, 1979 (junior synonyms include Phyllochiton Duan, 1984; Sinoclavus Duan, 1984; Sugaites Qian and Xiao, 1984; and probably Isoclavus Qian and Zhang, 1983; Tanbaoites Duan, 1984; Wushichites Qian and Xiao, 1984; and Zhijinites (in part) sensu Qian and Xiao, 1984; and Jiang and Huang, 1986; see above). (3) Deiradoclavus gen. nov. (4) Deltaclavus gen. nov. zhijinites Qian, 1 9 7 8 cv Type species. Zhijinites longistriatus Qian, 1978a. 366 PALAEONTOLOGY, VOLUME 34 text- fig. 6. Locality map of region around Meishucun (Yunnan), showing location of Bed 7 of the Xiaowaitoushan section (asterisked). Based on figs 1 and 2 of Luo et al. (1984/)). Diagnosis. Sclerite base subcircular to somewhat elongate, ventral base gently concave, dorsal surface convex. Margin usually entire, occasionally with prominent cleft that may be enclosed to leave perforation. Spine elongate, inclined, more or less straight. Transverse section of spine variable from subcircular to concavo-convex, ornamentation variable, including prominent transverse ridges, longitudinal ribbing to more or less smooth. Zhijinites longistriatus Qian, 1 978c/ Plates 1-3; Text-figs 7 and 1 la 1977 Zhijinites sp. Zhong [nomen nudum], p. 123, pi. 3, fig. 7; pi. 4, figs 22-27. 1978a Zhijinites longistriatus Qian, p. 34, pi. 2, fig. 5a, b. 1978 b Zhijinites longistriatus Qian, p. 350, pi. 142, fig. 5a, b. 1978a Zhijinites minutus Qian, p. 34, pi. 2, fig. 6 a-c. 1978/) Zhijinites minutus Qian, p. 350, pi. 142, fig. 4a-c. 1979 Zhijinites annae Chen [nomen nudum], p. 281, fig. id. 1979 Zhijinites costatus Chen [nomem nudum], p. 281, fig. 3 b. 1979 Zhijinites dictyoformise Chen [nomen nudum], p. 281, fig. ie. 71979 Zhijinites lubricus Qian et al., p. 225, pi. 4, figs 14 and 15. 1979 Zhijinites minutus Lu. pi. 1. figs 6 and 7. 1979 Zhijinites longistriatus Lu, pi. 1, figs 19 and 20; pi. 2, figs 2 and 1 1. 71980 Zhijinites lubricus Zhao et al., p. 49, pi. 3, fig. 21. 71980a Zhijinites lubricus Yin et al., p. 178, pi. 19, figs 1 1 and 12. 1980a Zhijinites longistriatus Yin et al., p. 179, pi. 19, figs 8 and 10. 1980a Zhijinites minutus Yin et al., p. 179, pi. 19, fig. 9. 71980a Fomitchella sp. Yin et al., pi. 19, fig. 7. 1980a Sachites hastatus Yin et al., p. 195, pi. 18, fig. 9 (?non pi. 18, fig. 30). 71980/) Zhijinites lubricus Yin et al., p. 65. 1981 Zhijinites minutus Xiang et al.. pi. 1, fig. 15. 1982 Zhijinites longistriatus Yin et al., pp. 287, 291. CONWAY MORRIS AND CHEN: CAMBRIAN PROBLEMATICA 367 1982 Zhijinites minutus Yin et al ., pp. 287, 291. 1982 Zhijinites longistriatus Jiang, p. 181, pi. 17, figs 8, 9, 19. 71982 Zhijinites lubricus Jiang, p. 182, pi. 17, fig. 13. 1982 Zhijinites minutus Jiang, p. 182, pi. 17, figs 6, 7, 20. 71982 Zhijinites undulatus Jiang, p. 182, pi. 17, fig. 1 1 (non fig. 12, \2a 7 = Paracarinachites spinus). 71982 Zhijinites umbelletes Jiang, p. 182, pi. 17, fig. 10, 10a. 1982 Heteroscu/potheca pheneres Jiang, p. 166, pi. 13, figs 23, 23a, 24, 25. 71983 Zhijinites lubricus Fu, p. 416, pi. 1, figs 12 and 13. 71984 Zhijinites lubricus Qian, pi. 3, fig. 8. 1984 Zhijinites minutus Qian, pi. 3, fig. 7. 71984 Zhijinites sp. Qian, pi. 3, fig. 16. 71983 Allatheca nanjiangensis [nomen nudum] Yang et a/., p. 95, pi. 1, fig. 14. 1984a Zhijinites longistriatus Qian and Yin, pi. 4, fig. 14. 1984a Zhijinites minutus Qian and Yin, pi. 5, fig. 14. 1984 b Zhijinites longistriatus Qian and Yin, pp. 215, 218, pi. 1, figs 12-15. 1984 b Zhijinites cordiformis Qian and Yin, pp. 215, 219, pi. 1, figs 20-23. 19846 Zhijinites minutus Qian and Yin, pp. 215, 218, text-figs 1.3 and 3c. 1, 2; pi. 1, figs 1-11 19846 Zhijinites panduriformis Qian and Yin, pp. 215, 218, 219, pi. 1, figs 16-19. 19846 Zhijinites triangularis Qian and Yin, pp. 215, 219, pi. 2, figs 14-21. 719846 Parazhijinites quizhouensis [sic] Qian and Yin, p. 220, text-fig. 3.3; pi. 2, figs 1-8. 1984a Zhijinites longistriatus Wang et al ., pi. 22, figs 13 and 14. 1984a Zhijinites minutus Wang et al., pi. 22, figs 14. 1984a Zhijinites panduriformis Wang et al.. p. 177, pi. 22, figs 9-12. 71984a Parazhijinites guizhouensis Wang et al., p. 177, pi. 22, figs 5-8. 19846 Zhijinites longistriatus Wang et al., pi. 4, fig. 9. 19846 Zhijinites minutus Wang et al., pi. 5, fig. 14. 1984a Zhijinites minutus Luo et al., pi. 10, fig. 24. 1984 Heterosulpotheca [sic] phaneres [sic] Jiang, fig. 4.3. 1984 Heterosulpotheca [sic] phoneres [sic] Jiang, pi. 2, fig. 8. 1984 Zhijinites minutus Jiang, pi. 3, fig. 11. 71987 Parazhijinites quizhouensis [sic] Liu, fig. 3a. 71987 Zhijinites triangularis Liu,, fig. 3b. 1989 Zhijinites sp. Chen, pi. 1, fig. 1. Diagnosis. As for the genus. Holotype. Institute of Geology and Palaeontology, Academia Sinica, Nanjing, ASN 33676 (Qian 1978a, pi. 2, fig. 5). Material illustrated here. Institute of Geology (Beijing), Academia Sinica IGAS-BC-88-30079-301 12. Stratigraphic horizon. Beds 36 and 37, Maidiping Member, Hongch unping Formation, Meishucun Stage, Lower Cambrian (see Yin et al. 1980a, 6 and He et al. 1984 for further details). Locality. Maidiping section, Emei, Sichuan. Preservation. The majority of specimens consist of sclerites with a phosphatized wall enclosing a central cavity (Text-fig. 7). Petrographic sections demonstrate that the extent of phosphatization varies quite widely, and may include obvious spherulitic ingrowths on the interior of the spine cavity (Text-fig. 76). The dorsal side of the base is, apart from radial furrows, often relatively smooth (e.g. PI. 1, figs 1, 3, 16, 17; PI. 2, figs 4, 8, 13). It is frequent on the spine, however, for surfaces to be more irregular, consisting of a fibrous ultrastructure running parallel to the long axis of the spine (PI. 1, figs 5, 11-14, 16; PI. 2, figs 1 and 10). This fibrosity is interpreted as diagenetic phosphatization of an originally calcareous wall. In addition, endolithic borings are also abundant in some specimens. These may be visible as openings on the surface of the base (PI. 1, figs 8 and 9), or as steinkerns of tubes that run along the spine (PI. 3, figs 1-3, 9, 12). The tubes consist of two distinct size classes (c. 3 /mi and 7 /mi diameter respectively). The larger category possesses a series of swellings, sometimes locally pronounced, that impart a beaded appearance (PI. 3, figs 10 and 13) and may end blindly 368 PALAEONTOLOGY, VOLUME 34 text-fig. 7. Petrographic thin sections of Zhijinites longistriatus Qian, 1978 from Beds 36 (a-c, e , g and h) and 37 (.‘ TRIMPLEY gxdNL'ER HEREFORD & WORCESTER ' ■•'.. •Ross on Wye o • ■ • Monmouth J YJ GWENT Newport 0 10 20 30 40 50 scale in km text-fig. 1. Known phialaspidid localities in the Anglo-Welsh Lower Old Red Sandstone, indicated by symbols as shown in the key. Mountain District; (19) Eastham Brook, The Trimpley Inlier; (20) Heath Farm, Wolferlow, The Trimpley Inlier; (21) Holheach House Stream, The Trimpley Inlier; (22) House of the Wood Quarry, Garnon’s Hill, Heightington ; (23) Hurtlehill Farm Quarry, Heightington ; (24) Llan Farm, Dorstone, Black Mountain District; (25) Man Brook, The Trimpley Inlier; (26) Mary Moors, The Trimpley Inlier; (27) Merbach Brook, Ledbury; (28) Park Atwood Stream, The Trimpley Inlier; (29) Ross Motorway, M50 section; (30) Sapey Brook, Thrift Farm; (31) Shatterford, Boundry Brook, The Trimpley Inlier; (32) Westhope Hill, near Hereford; (33) Witchery Hole, Clifton on Teme. Phialaspis symondsi found at all localities except 28 and 31, at which Toombsaspis pococki was found. Both species present at localities 18 and 33, in the same horizon only at the latter. Powys'. (34) Onen, Court Wood Quarries; (35) Pen-y-lan, Crwcws Wood Quarries. Phialaspis symondsi found at both localities. Gwent '. (36) Alteryn Quarry, Toombsaspis pococki ; (37) Coed-y-coedcae, Phialaspis symondsi ; (38) Penrhos Farm Quarry, Phialaspis symondsi. Gloucestershire'. (39) Lydney, Phialaspis symondsi ; (40) Sharpness Docks, Toombsaspis sabrinae. Dyfed: (41) Caldy Island, Phialaspis symondsi , Toombsaspis pococki (several horizons); (42) Freshwater West, Phialaspis symondsi; (43) Manorbier Bay, Phialaspis symondsi. TARRANT: LOWER DEVONIAN OSTRACODERM 401 MATERIALS AND METHODS To obtain information about their outer surfaces, the Toombs and Rixon (1950) transfer method was used on several specimens. This entailed mounting the specimens on clear resin, and removing the matrix with acetic acid. Although several specimens prepared reasonably well, the results were mixed. The larger plates had a tendency to be destroyed by the acid owing to the calcite infill of their cancellous spaces. A limited amount of success in tracing the sensory canal system came from impregnating certain specimens with oil of aniseed and viewing them in transmitted light. Often it was necessary to remove the aspidin with dilute hydrochloric acid (White 1935, 1946). Many specimens were not prepared because of the risk of damage to their inner surfaces. The bulk of material is new and is housed in Ludlow Museum, Shropshire, SHRCM.G -235 specimens, and in the National Museum of Wales, Cardiff, NMW - 6 specimens. Other specimens studied are from established collections housed in the following museums: British Geological Survey Museum, Kegworth, Notts., BGS (GSM); British Museum (Natural History), London, BMNH; University of Birmingham Geology Museum, BU; National Museum of Canada, Ottawa, NMC; Princeton University Geological Museum, New Jersey, USA, PU; Royal Museum of Scotland, Edinburgh, RSM ; W. F. Whittard collection, S. STRATIGRAPHY AND PALAEOECOLOGY Most of the heterostracans described in this work came from the Upper Downton Group, Lower Old Red Sandstone, Anglo-Welsh Region. The stratigraphy and sedimentology of this area have been documented by Ball and Dineley (1961); Allen and Tarlo (1963); Allen (1964, 1974r/, 19746, 1985); and Allen and Williams (1978, 1981). It is dominated by red mudstones, which are interspersed with discrete beds of upwardly fining, current-influenced units of sandstones and intraformational conglomerates. The conglomerates usually hold the largest concentrations of vertebrate fossils. Most recent workers in the field have considered that they represent infilled freshwater channel complexes, within an extensive deltaic floodplain (Ball and Dineley 1961; Allen and Tarlo 1963; Allen 1964, 1974 a, 19746). The area is dissected by the Psammosteus Limestone, a pedogenic feature (see Allen 1974u, 1985), which divides the Downton from the overlying Ditton. Although rare specimens of Phialaspis symondsi have been found above the Psammosteus Limestone, this horizon marks a distinctive faunal change, where the phialaspids are replaced by pteraspidiforms (White 1950r/; Ball and Dineley 1961). White (1950(3) used Phialaspis symondsi as a zone fossil marking the uppermost Downton, and Toombsaspis pococki to mark the underlying zone. The base of the range of Phialaspis symondsi is about 30 m below the Psammosteus Limestone. Recent field studies (Rowlands and Tarrant, unpublished data), would suggest that the top of its range is considerably less than Ball and Dineley’s (1961) claim of c. 53 m above the Psammosteus Limestone. The bulk of Toombsaspis pococki material studied by White (1946), came from 5 m below the Psammosteus Limestone at Gardener’s Bank, Shropshire, which is the top of its range. Squirrel and Downing ( 1969) collected fragments which they considered to belong to this species from 158 m below the Psammosteus Limestone at Ateryn Quarry, Gwent, which may be the bottom of its range. However, it would appear that the two species substantially overlapped in time. They have only been recorded together at the Witchery Hole, Clifton on Teme, Hereford and Worcester (Ball and Dineley 1961), where the material was in loose blocks and may have originated from different horizons (M. A. Rowlands, pers. comm.). As Ball and Dineley suggested, this could indicate that they occupied different environments. Following this, the two species are mainly found with different vertebrate faunas. Phialaspis symondsi is characteristically found with Tesseraspis tesselata , Anglaspis macculloughi , Corvaspis kingi, Turinia pagei , cephalaspids, Ischnacanthus wickhami and other acanthodians (Ball and Dineley 1961; Turner 1973). It is also occasionally 402 PALAEONTOLOGY, VOLUME 34 found with Protopteraspis gosseleti , Pteraspis rostrata and Nodonchus sp. Toombsaspis pococki is associated with Tesseraspsis tesselata, Didymaspis grindrodi and other cephalaspids, the Goniporus- Katoporus thelodont fauna, Ischnacanthus kingi and other acanthodians (White 1946; Turner 1973). It would seem that the overlapping vertebrate assemblages of the horizons subjacent to the Psammosteus Limestone are related as much to varying ecological conditions as they are to time. Indeed, Karatajute-Timalaa (1978) and Blieck (1984) proposed that the zones of Traquairaspis symondsi and Traquairaspis pococki should be amalgamated into a single Traquairaspis zone. However, because of the reclassification of these species in this work, it is proposed that it should be renamed the Phialaspis symondsi-Toombsaspis pococki zone. As Ball and Dineley (1961) observed, the vertebrate remains are often fragmented, and concentrated in pockets with individual specimens of a similar size, buoyancy, or weight, suggesting that they were probably originally transported, water selected, and in some cases may have been reworked. Their preservation is usually good, often showing fine details, and the vascular cancellous layers are normally not crushed, because of calcite infilling. SYSTEMATIC PALAEONTOLOGY Subclass heterostraci Lankester, 1868 Order traquairaspidiforme Tarlo, 1962 Diagnosis. (After Dineley and Loeffler 1976). Dorsal shield comprises either single plate or single dorsal disc, rostral and pineal plates, and paired orbital, branchio-cornual or branchial and cornual plates. Orbital, pineal and branchial openings enclosed. Ornamentation of dorsal shield often of elevated, laterally serrated tubercles, commonly with narrow interstitial tubercles or ridges, mainly arranged in cyclomoriform units, sometimes with outer adult plate growth. Ventral disc ovate to elongate, with lateral ornamentation similar to dorsal shield, becomes broader and flatter towards longitudinal midline, or replaced by smooth, flat, ovate central area. Lateral line system variable, ranging from pattern of longitudinal canals and transverse commissures to anastomosing network. Discussion. Although Weigeltaspis may prove to be a traquairaspidiform (Obruchev 1964; Blieck 1983), this has yet to be established. It is possible that the Canadian ? Traquairaspis and Nat/aspis Dineley and Loeffler, 1976, with ornamented ventral central regions, may prove to represent different evolutionary lineages from those species with smooth ventral central regions. White (1950a) realised that the specimens he had described as Phialaspis pococki subsp. cowiensis White, 1946 were ventral discs of Traquairaspis campbelli , and he reclassified Phialaspis pococki and Phialaspis symondsi as members of the genus Traquairaspis. However, following Halstead’s (1982) retention of the name Phialaspis , the British species can be divided into two distinct morphological groups. They are considered in this work to represent two distinct families, the Traquairaspididae and the Phialaspididae. Family phialaspididae White, 1946 Type genus. Phialaspis Wills, 1935 Other genera assigned. Toombsaspis gen. nov., Munchoaspis gen. nov. Diagnosis. Dorsal shield usually comprises seven separate plates. Dorsal disc quadrate, vaulted posteriorly, with median row of large cyclomoriform units on posterior half forming a keel, and usually a dorsal vane. Ventral disc flattening and widening anteriorly with raised, smooth, coffin- shaped central area, situated more posteriorly than anteriorly and enclosed by ornamented margin of disc. Two rows of longitudinally running, large cyclomoriform units on each lateral side of dorsal disc, one row on each lateral side of ventral disc, another row on dorsal side of each branchio- TARRANT: LOWER DEVONIAN OSTRACODERM 403 cornual plate. Regions of cyclomoriform adult growth on anterior and lateral edges of dorsal and ventral discs. Paired lateral plates, with, quite frequently, separate paired post oral plates. Genus phialaspis Wills, 1935 Type species. Cyathaspis ( ? ) symondsi Lankester, 1 868 Diagnosis. Large advanced Phialaspididae. Dorsal discs more vaulted than ventral disc. Dorsal vane large with two cyclomoriform units, median keel with one. Rostrum enlarged. Branchio- cornual plates with lateral keels terminating posteriorly in lateral vanes. Ventral disc with non- unital cyclomoriform growth between longitudinal units and smooth central region, with three units positioned behind and sometimes fused to its posterior edge. Ornament of stellated tubercles, which are often ringed on the reticular layer by a groove or shelf. Phialaspis symondsi (Lankester, 1868) Plates 1-4; Plate 5, figs 2-5; Plate 6; Text-figs 1 10, 13, 14a,b, 15a-g. 16; Table I 1868 Cyathaspis (?) symondsi Lankester, p. 27, pi. 6, fig. 5. 1898 Psammosteus anglicus Traquair, p. 67, pi. 1, figs 1 and 2. 1935 Phialaspis symondsi (Lankester); Wills, pp. 439^444, pis 5-7; text-fig. 4. 1948 Traquairaspis symondsi (Lankester); White and Toombs, p. 7. Holotype. BGS(GSM)31 380, ventral disc. Horizon and localities. Upper Silurian/Lower Devonian. Uppermost Downton and Lowest Ditton Groups, Anglo-Welsh region (see Text-fig. 1). Referred material. SRCH.G: 213 from Devil’s Hole, 13 from Little Oxenbold, 6 from Earnstry Brook, 2 from Barnsland Farm Quarry, 1 from Oak Dingle; NMW : 2 from Cusop Dingle, 1 from Lydney and 3 from Manorbier Bay; material in the BMNH, BGS(GSM), and BU. This material consists of 27 dorsal discs, 14 table 1 . Maximum dimensions of adult Phialaspis symondsi plates in millimetres. Abbreviation : pop, post- oral process. Range Average Length Width Ratio of width to length Length Width Ratio of width to length Dorsal discs 50-70 40-60 0-67-0-96 60 49 0-80 Orbital plates 26-34 12-22 0-44-0-56 30 16 0-52 Pineal plates 9-20 10-20 0-85-1-33 14 15 1-07 Rostrums 13-17 21-24 1-23-1-84 15 22 1 46 Branchio-cornuals 60-80 35-37 0-43-0-55 68 35 0-51 Ventral discs 68-100 42-64 0-54-0-75 81 50 0-61 Lateral plates + pop 26-43 14-24 0-48-0-61 36 20 0-56 Lateral plates — pop 20-31 16-23 0.67-0.90 25 19 0-76 Curvital dimensions Dorsal discs 51-77 50-74 0-76-1 08 64 60 0-93 Branchio-cornuals 94-116 100 Ventral discs 69-101 44-74 0 61 83 59 0-71 0.78 404 PALAEONTOLOGY. VOLUME 34 dorsal vanes, 20 orbital plates, 10 pineal plates, 7 rostral plates, 26 branchio-cornual plates, 74 ventral discs, 20 lateral plates, 5 oral plates, plus scales and fragments. Diagnosis. As for genus. Description. The dorsal disc (PI. 1, figs 1 and 2; PI. 2, fig. 2; Text-fig. 2) is highest and widest about halfway along its length. Its lateral margins are gently scalloped, and the anterior margin is sometimes angled and slightly indented to match the contact with the posterior margin of the pineal plate and the dorso-posterior margins of the orbital plates. The two lateral rows of units are most pronounced at the posterior of the first, most medially-placed row, and along the second row, internal impressions marking their edges can often be seen (PI. 1, fig. 2; Text-fig. 2b). The cancellae are enlarged under the apex of each unit causing the exoskeleton text-fig. 2. Phialaspis symondsi (Lankester), dorsal discs, a, part superimposed on counterpart to show ornamentation and incomplete sensory canals, SHRCM.G08137/1-2. B, incomplete and mainly internal mould in plan and lateral views, a-a, b-b, lines of cross section, SHRCM.G08138. c, part superimposed on counterpart to show sensory canal system, SHRCM.G08139/1-2. d, incomplete juvenile showing developing ornament, c-c, line of cross section, SHRCM.G08140. e, incomplete specimen with part superimposed on to counterpart to show sensory canal system, SHRCM.G08141/1-2. Abbreviations: ag, adult growth region; dv, dorsal vane; imu, internal impressions of units; ltc, lateral transverse commissure; mdc, medial dorsal longitudinal canal; mtc, medial transverse commissure; sp, sensory pore. to swell from 1 to 2 mm in thickness. With the exception of the large tubercle or frequently large tubercles, capping the apex, the tubercles are small and irregular on the units. This contrasts with the larger and more equilateral tubercles found on the peripheral adult zone (Text-fig. 2 a), where a longitudinal fold can sometimes be observed on each lateral side of the larger dorsal discs. This is so vestigial that it could hardly be described as a row of units. TARRANT: LOWER DEVONIAN OSTRACODERM 405 text-fig. 3. Phialaspis symondsi (Lankester), dorsal vanes, a, lateral view SHRCM.G08142. b, cross section at a-a. c, short high specimen, SHRCM.G08143. d, adolescent specimen, SHRCM.G08144. e, long low specimen, showing ornamental details, SHRCM.G08145. f, ditto, posterior view. Abbreviations; ae, anterior element; dr, developing region; pe, posterior element. The dorsal vane (PI. 1, figs 3-5; Text-fig. 3) is triangulate and varies in proportion, ranging from 34 mm long x 21 mm high to 19 mm long x 26 mm high and is 7-10 mm thick at its base. Its two specialized units are often in tandem, with a doubled and thickened cancellous layer which narrows towards the tip and divides at the base to merge with the disc. The rear unit is normally largest. However, the dorsal vane of SHRCM.G08137 (PI. 1, fig. 6) has an atrophied rear unit, and is mainly formed from the front unit. Although broken at its anterior end, the dorsal vane SHRCM.G08140 (PI. 1, fig. 3; Text-fig. 3d) is small, only 17 mm, in height. A depressed region running longitudinally just above its base may indicate an area of growth. An incomplete and immature dorsal disc SHRCM.G08140 measures 17 mm long x 24 mm double half width (PI. 2, fig. 2; Text-fig. 2d). Its dorsal vane is shown in section and is 5 mm high x 3 mm thick at the base. The medial longitudinal sensory canals have been exposed and are much closer together than on the larger dorsal discs. All levels of its exoskeleton were present. The surface is pitted with openings on the more complete left lateral and anterior edges. This grades inwards via developing tubercles (Text-fig. 2d) to well-formed large tubercles not arranged in cyclomoriform patterns. The orbital plates (PI. 2, fig. 1 ; Text-fig. 4a-c) are elongate and irregularly diamond-shaped, with usually concave dorsal edges to accommodate the pineal plate. They are curved to present dorsal and lateral sides towards the front, and become flattened towards the back to slope at a dorso-lateral angle. The orbital opening ranges from 2 to 3 mm in diameter; it is on the angle of the dorso-lateral fold, and is usually slightly nearer the anterior edge of the plate than the posterior. The pineal plate (PI. 2, fig. 3; Text-fig. 4d-g) is distinguished by its flat and more or less rhomboid shape. There is a centrally placed pineal foramen, which is approximately 1 mm in diameter. The tubercles on the pineal and orbital plates are arranged in concentric rings around the pineal and orbital openings. The dorsal side of the rostrum (PI. 2, fig. 6; Text-fig. 5 c) is rounded and highest at the posterior edge which is three-pointed, with two concave edges which would have accommodated the front of the orbital plate. The plate tapers towards a tip formed by large horizontally running tubercles. The ventral pre-oral surface (PI. 2, fig. 4; Text-fig. 5a) has a raised, flat central region, which probably represents an area of a similar kind to the pre-oral field found on certain pteraspidiforms. With the exception of several long tubercles traversing the anterior half, it is ornamented with small and atrophied tubercles. Posterior to the pre-oral surface and rimmed with a maxillary flange on the angle of ascent, the pre-oral wall ascends vertically to join the posterior undersurface of the plate. The basal laminated layer on the posterior undersurface of SHRCM.G08161 (Text-fig. 5a) is folded and contorted on each side of a shallow median groove. The larger specimens have proportionally longer pre-oral regions. On the smallest rostrum, NMW88.32G. (PI. 2, fig. 5; Text-fig. 5d), the pre-oral region is 0-28 times the length and 0-60 times the width of the pre-oral region of SHRCM.G08161 . It is broken along its posterior edge, and has all exoskeletal layers present. Its small size and proportions indicate it was from an immature animal. A strong depression on each side of the pre- oral surface shows regions of possible active growth. The basal laminated layer of the posterior undersurface, along its junction with the pre-oral wall, is perforated with vascular foramina often set within large depressions (Text-fig. 5d) which appear to match the contorted conditions found in this region on SHRCM.G08161. 406 PALAEONTOLOGY, VOLUME 34 text-fig. 4. Phialaspis symondsi (Lankester). a-c, orbital plates; a, right plate, part superimposed on counterpart to show sensory canal system, a-a, b-b, c-c, d-d, lines of cross section, SHRCM.G08147/1-2; b, right plate showing sensory canal system, NMW88.32G.2; c, right plate, part superimposed on counterpart to show ornamentation and sensory canal system, SHRCM.G08146/1-2. d-g, pineal plates; d, small plate, part superimposed on counterpart to show part of sensory canal system, a-a, b-b, lines of cross section, SHRCM.G08148/1-2 ; e, small plate, part superimposed on counterpart to show sensory canal system, SHRCM.G08149/1-2 ; f, fragmentary large plate, part superimposed on counterpart to show part of sensory canal system, SHRCM.G0815/1-2; G, fragmentary large plate showing pineal opening and ornamental details, SHRCM.G08 1 51 . Abbreviations: cor, circum-orbilal canal; ior, inter-orbital canal; ldc, dorsal longitudinal canal; ltc. lateral transverse commissure; mdc, medial dorsal longitudinal canal; or, orbit; pi, pineal organ; sor, supra-orbital canal; sp, sensory pores. The branchial opening, located about three-fifths of the way along the length of the branchio-cornual plate (PI. 3, figs 4 and 5; Text-fig. 6) is dorsally facing and ovate, and ranges in size from 8x4 mm to 11x7 mm. The lateral keel embraces the lateral side of the branchial duct and encloses the front of the branchial opening. Its vascular cancellous layer is greatly thickened and individually variable, ranging from 5-12 mm wide x 4-10 mm thick regardless of the size of the rest of the plate. Elongated and longitudinally running rows of tubercles are found on both sides of this region. On its lateral edge, closely spaced, 1 mm thick tubercles overlie smaller primary tubercles (Text-fig. 6 b). Occasionally, there are regions of abrasion on the ventral side (Text-fig. 6d). The lateral vane occupies from the back, one half to one third the length of the branchio-cornual plate and joins the lateral keel. It is solid, triangulate, and dorso-ventrally flattened, with two greatly thickened vascular cancellous layers. Measuring 7 mm thick at its base on SHRCM.G08194, it forms the postero-lateral edge of the branchial opening. The whole vane is tilted postero-laterally, with an elongated region of small and EXPLANATION OF PLATE 1 Figs 1-6. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland. 1, 2, 6, dorsal discs, x2; 1, SHRCM.G08166/1, dorsal view of external mould; 2, SHRCM.G08243/1, dorsal view of small internal mould showing impressions; 6, SHRCM.G08166, lateral view of silicon rubber impression showing malformed dorsal vane. 3-5, dorsal vanes, x2; 3, SHRCM.G08140, external cast of immature vane; 4, SHRCM.G08143, external cast; 5, SE1RCM.G08145, external cast. PLATE 1 -ASfFZiSt; ■ • PHV. - . « ' .^Ss| re^^'«K? ,#;.*•**« ■ :A^r^. -v3- ’ P}:: *' •' Sim TARRANT, Phialaspis symondsi 408 PALAEONTOLOGY, VOLUME 34 text-fig. 5. Phialaspis symondsi (Lankester), rostral plates, a, ventral view showing details of tip and posterior undersurface, a-a, line of cross section, SHRCM.G08160. b, fragmentary specimen, showing pre-oral wall and part of pre-oral surface, SHRCM.G08161. c, dorsal view showing ornamentation, b-b, line of cross section, SHRCM.G08162. d, immature specimen, part superimposed on to counterpart to show dorsal side with details of ornamentation, and ventral side with details of posterior under surface, NM W88.32G. 1 a/b. Abbreviations : dr, developing region; por, pre-oral rim/maxillary brim; pos, pre-oral surface; pow, pre-oral wall. irregularly shaped tubercles running from the tip to the branchial opening and dividing the dorsal side. On the antero-dorsal side and edge, the tubercles have a tendency to form weak ornamental units and run in rows around the postero-dorsal and ventral sides. Towards the tip, they become elongated and reach up to 1 mm in thickness. The lateral vane is usually terminated at the back by a cyclomoriform unit, which forms a horizontal flange (see Text-fig. 6a) measuring 10 mm long x 5 mm wide on SHRCM.G08153 and SHRCM.G08154. Where the branchial-cornual plate slopes upwards to meet the lateral edge of the dorsal disc, it is composed of a longitudinal row of units (Text-fig. 7b,e). These cover the dorsal side of the branchial duct, encircle the medially facing side of the branchial opening, and are terminated posteriorly by a large unit (Text- fig. 6a, d). They often leave internal impressions marking their edges (PI. 3, fig. 4; Text-fig. 6b). On the basal laminated layer of SHRCM.G08155 growth ridges run longitudinally between the units and the lateral keel (Text-fig. 6 e). The ornament on the ventral side of the branchio-cornual plate (Text-fig. 6c) curves transversely from the front of the lateral vane, to run parallel with the ventral edge, which is concave to accommodate the lateral edge of the ventral disc. A zone of growth runs parallel with the ventral and dorsal anterior edges, which are angled to match the ventro-posterior edge of the orbital plate and the posterior edge of the lateral plate. Three elongate and approximately diamond-shaped plates (PI. 3, figs 1 and 2; Text-fig. 6f,h,i) appear to represent juvenile branchio-cornual plates. Their sizes are: SHRCM.G08157 33 mm longxl3mm wide; SHRCM.G08156 26 mm long x 10 mm wide; SHRCM.G081 58 21 mm long x 8 mm wide. Each is bowed along its length, angled at its front, and tapered towards the back, where a region about 5 mm long projects about 2-5 mm from the lateral side of the plate. This apparently represents the developing lateral vane. SE1RCM.G08156 has been prepared to show typical P. symondsi tubercles in various stages of eruption and development (Text-fig. 6 f). The inner surface of SHRCM.G08158, shows recently enclosed spaces which form blister-like regions with centrally-placed pores. A 1-2 mm wide margin around the edges is a maze of openings surrounded by enclosing basal laminated growth (PI. 3, fig. 3; Text-fig. 6 1). The longitudinal row of units is explanation of plate 2 Figs 1-6. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland. 1, SHRCM.G08147/2, right orbital plate, mostly internal view. 2, SHRCM.G08140, immature dorsal disc, in part external mould. 3, SHRCM.G08250, cast of pineal plate. 4, SHRCM.G08160, cast of ventral surface of rostrum. 5, NMW88.32G.lu, cast of ventral surface of immature rostrum. 6, SHRCM.G08162/1, cast of dorsal surface of rostrum. All x 4. PLATE 2 TARRANT, Phialaspis symondsi 410 PALAEONTOLOGY, VOLUME 34 superimposed on counterpart, showing sensory canal system, SHRCM.G08152/1-2. b, right plate, dorsal side showing ornamentation, with dorsal half of counterpart showing inner surface, SHRCM.G08153/1-2. c, right plate, ventral side with part superimposed on counterpart, SHRCM.G08154/1-2. d, left plate, lateral view showing details of worn ornament, BMNH31146. e, fragmentary right plate, lateral view of inner surface, SHRCM.G08155. f,g, juvenile right plate; f, ventral view with detail of ornamentation; G, lateral view; a-a, line of cross section. SHRCM.G08156. H, juvenile right plate, lateral view, SHRCM.G08157. i, juvenile left plate, showing inner surface, SHRCM.G08158/1-2. j, ‘adolescent' left plate, lateral view, mainly internal, SHRCM.G08159. Abbreviations: b', bite; bpu, large posterior unit; brd, branchial duct; bro, branchial opening; grd, growth ridges; hf. horizontal flange; im, inset margin; imu, internal impressions of units; ldc, lateral dorsal longitudinal canals; lk, lateral keel; lv, lateral vane; vp, vascular pores. missing from these plates, and they are much flatter than the adult branchio-cornual plates. Nevertheless, their branchial ducts run their entire length, which shows that the branchial openings were posteriorly placed and not enclosed. Although the superficial layer was destroyed during preparation, the lateral side of SHRCM.G08157 is 4 mm in thickness, corresponding to the enlargement of the lateral keel. SHRCM.G08159 (Text-fig. 6 j) is an internal mould of an early stage of development of a branchio-cornual EXPLANATION OF PLATE 3 Figs 1-5. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland, branchio-cornual plates. 1 , SHRCM.G08 158/1 , cast of inner surface of immature left plate, x 4. 2, SHRCM.G081 56, cast of ventral surface of immature right plate, x4. 3, detail of 1, x 10. 4, SHRCM.G081 53/1, dorsal view of right plate, in part internal mould, x 2. 5, SHRCM.G08152/2, dorsal view of external mould of left plate, x2. PLATE 3 mi in TARRANT, Phialaspis symondsi 412 PALAEONTOLOGY, VOLUME 34 i i 20mm epm text-fig. 7. Phialaspis symondsi (Lankester), dorsal shield, SHRCM.G08164/1-2. a, internal plan view, with detail of anterior, b, part superimposed on to counterpart to show ornamentation, a-a, line of cross section. Abbreviations : epm, estimated position of posterior margin ; igbcp, regions of intergrowth between dorsal disc and branchio-cornual plate; igorp, regions of intergrowth between dorsal disc and orbital plate; igpip, regions of intergrowth between dorsal disc and pineal plate; ltc, lateral transverse commissure; mdc, medial dorsal longitudinal canal; su, sutures. plate. Although the whole plate only measures 30 mm long, its branchial duct is strongly dorso-ventrally folded as in adult specimens. A cross-section at the front shows all levels of the exoskeleton to be present. The large dorsal posterior unit is present, and is estimated to be two-thirds the size of the equivalent area on the adult plates. A depressed margin, 1-2 mm wide, around its dorsal and posterior edges, and running along the dorsal side of the branchial duct, appears to show regions of active outward growth. An open-ended notch 2 mm wide dissects the margin at the anterior of the posterior unit and appears to represent the start of the enclosure of the branchial opening. An incomplete dorsal headshield, SHRCM.G08164 (PI. 4, figs 1 and 2; Text-fig. 7), which is somewhat distorted by compression, consists of the dorsal disc, the inner halves of the branchio-cornual plates, the back of the pineal plate and the dorso-posterior part of the orbital plates, with the omission of the pineal, orbital, and branchial openings. In contrast to the tubercles found in the regions of adult growth of isolated dorsal discs, the tubercles in the regions of adult growth of SHRCM.G08164 vary considerably in size and shape (Text-fig. 7 b). Prior to the formation of peripheral adult growth, the dorsal disc acquired the longitudinal units of the branchio-cornual plates, then fused with the back of the pineal plate and the dorso-posterior part of the orbital plates, where sutures can be observed on the inner surface (Text-fig. 7 a). As there is no evidence of adult growth on the dorsal edges of the branchio-cornual plates of P. symondsi , it appears that during adulthood, the dorsal disc SHRCM.G08164 encroached and intergrew with the longitudinal units of the branchio-cornual plates. Due to a large range in size, it would appear that the pineal and orbital plates of P. symondsi were capable of adult growth, and probably on SHRCM.G08164, they contributed to the intergrowth and kept the encroachment of the dorsal disc in check. In the ventral discs (Text-fig. 8), the flat central area stands proud by 1-2 mm and its surface consists of a TARRANT: LOWER DEVONIAN OSTRACODER M 413 text-fig. 8. Phialaspis symondsi (Lankester), ventral disc, a, internal view, showing attached posterior units, SHRCM.G08165. b, external view showing sensory canal system, a-a, b-b, lines of cross section, SHRCM.G08166/1-2. c. immature plate, a-a, line of cross section, SHRCM.G08167/1-2. d, detail of ornament on left anterior corner, SHRCM.G08168. e, detail of worn ornament on the anterior, BMNH46712. f, abraded posterior, SHRCM.G08169; G, transverse view across midline, showing part of abnormal longitudinal rows of units, SHRCM.G08170. h, part of left side showing longitudinal rows of units, SHRCM.G08171/ i, specimen developed to show sensory canal system, SHRCM.G08171. Abbreviations: b , bite; hf, healed fracture; lpu, lateral posterior unit; lru, longitudinal row of units; mpu, medial posterior unit; poc, post-oral sensory canal; sea, smooth central area; sp, sensory pores; vie, ventral longitudinal sensory canal. smooth sheet of dentine. Regardless of the size of the rest of the disc, it varies considerably in size and proportions, ranging in length from 31 to 60 mm, and in width from 1 1 mm to an estimated and exceptional 40 mm in BU759. The posterior edge in SHRCM.G08169 is worn, and the adjacent tubercles are abraded and merge with the smooth dentine (Text-fig. 8 f). The tubercles of the ventral disc, in contrast to those of the dorsal disc, are usually of a similar size and normally moderately elongated. Tarlo (1962) recognised that bands of ornamented growth joined each lateral side of the smooth central region to a row of longitudinally running units. These units are cyclomoriform and raised centrally. Although internal impressions marking their edges are sometimes observed, they often protrude internally. There are usually five or more a side, and in SHRCM.G08170 there are two rows crowded together on each side (Text-fig. 8g). A single unit can range in size from 4-15 mm long x 4-10 mm wide. In certain specimens of ventral discs, the ornament at the front runs horizontally, matching underlying growth ridges. In other cases, it runs at right angles to the growth ridges, before curving round the anterior edges of the longitudinal rows of units, where on SHRCM.G08168 the tubercles join together to become 414 PALAEONTOLOGY, VOLUME 34 elongated (Text-fig. 8d). This region on BMNH46712 has broadened and abraded tubercles (Text-fig. 8e). The flow of ornament running from the posterior of the smooth central region is also variable, but does not overlie any growth ridges. Three units forming the posterior end of the ventral disc and uniting the two longitudinal rows of units are only observed clearly as internal impressions on SHRCM.G08165 (PI. 4, fig. 4; Text-fig. 8a), and probably contacted the antero-ventral ridge scale and the two antero-ventral scales. The medial posterior unit measures 7 mm long x 9 mm wide and the two lateral posterior units both measure 1 1 mm long x 17 mm wide. As they flatten out at an angle from the vaulted posterior end of the disc, it is likely that they would have been lost after death, and were seemingly often independent of the disc in the younger animals. Four specimens of immature ventral discs are represented by smooth areas with narrow ornamented margins and no attached units. Each central area is of adult proportions, but as observed on SHRCM.G08167 (PI. 4, fig. 3; Text-fig. 8 c), it rests only slightly proud of the rest of the disc. The anterior and posterior ends of the discs are observed to be flattened internally by a thickening of the cancellous layer. A specimen (BU77) described by Wills (1935) as cf. Ctenaspis , is actually a fragmentary ventral disc from P. symondsi. With the exception of a small region of smooth dentine measuring 1 x 2 mm, the superficial layer is missing, leaving the cancellae exposed. As White (1946) found on T. pococki , P. symondsi had two lateral plates. The lateral plate is approximately triangular in shape (PI. 5, figs 2 and 5; Text-fig. 9a-d). It is widest at its anterior end, and tapers towards its posterior edge where it met the branchio-cornual plate. The plate is folded, forming anterior, lateral, and ventral sides, and it is deepest at their junction. The bulk of the plate is ventral in position, where it is flattest. One edge is concave to embrace half of the anterior edge of the ventral disc. On the opposite edge, the plate is folded longitudinally at an angle of 60-90°, to form the 3-4 mm wide laterally facing side, which is angled to meet the ventral edge of the orbital plate. The anteriorly facing side folds inwards at between 20° and 50° from the main body of the plate. It is cradled by a concave region, which at one end forms a projection that would have met the postero-ventral part of the rostrum, and at the other end forms either a truncated mesial edge, or a post-oral process (Text-fig. 9a,c), The post-oral process is partly square in outline, flattened at its free end, and measures about 5 mm long x 10 mm wide. It is too short to occupy the space between the lateral plates, the front of the ventral disc, and the oral region, and must therefore have been paired. About one third of the lateral plates collected from Devil’s Hole have post-oral processes, but the rest show no sign of this structure and must have possessed separate post-oral plates. Older animals may have fused plates. Although the lateral plate, SHRCM.G08174 (PI. 5, fig. 5; Text-fig. 9d), has an exoskeleton of adult thickness, it measures 19 mm long x 10 mm wide, and is so small that it must represent an immature plate, yet it possesses a well-formed post-oral process. The ornament on the lateral plates is cyclomoriform, and the post-oral process was formed from a separate cyclomoriform unit. The tubercles in some specimens (Text-fig. 9 a) are enlarged and joined together, and tubercles occasionally run at right angles to the main ornamental direction. Because of their large size, it seems that two anterior lateral plates and one median oral plate were the full complement of oral plates present in P. symondsi. The posterior end of the median oral plate seems to have been as wide, if not slightly wider than the posterior margin of the oral cavity, and somewhat longer than the length of the oral cavity. Therefore it would appear to have rested inside the mouth, where, laterally and posteriorly, it was overlapped by the anterior lateral plates. It ranges in width from 16 to 11 mm, in length from 14 to 12 mm, and is about 6 mm high. It is scoop-shaped, and is ornamented on its outer side and smooth on its inner side and edges (PI. 5, fig. 4; Text- fig. 9e,f). The inner side (Text-fig. 9 f) has an elongated and gently convex central area, which strengthens and widens towards the back. The inner side behind the edge turns outwards to form a narrow lip, which is matched by a thickening of the exoskeleton. On the outside the tubercles are small and narrow and they generally run longitudinally, although they sometimes curve and run at right angles to the main direction. The ornament is abraded in places, in particular on the right lateral side of SHRCM.G08177, and there is a large callus 5 mm EXPLANATION OF PLATE 4 Figs 1-4. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland. 1 and 2, SHRCM.G08164/1/2, dorsal headshield, dorsal views of internal mould and silicon rubber impression of external surface, respectively. 3 and 4, ventral discs; 3, SHRCM.G08167/1, external mould of immature disc; 4, SHRCM.G08165, internal view of cast. All x 1. PLATE 4 TARRANT, Phialaspis symondsi 416 PALAEONTOLOGY, VOLUME 34 10mm text-fig. 9. Phialaspis symondsi (Lankester), a-d, lateral plates; a, external view of right plate, showing ornamentation, a-a, b-b, lines of cross section, SHRCM.G08173/1-2; b, internal view of left plate, without post-oral process, SHRCM.G08176/1-2; c, right plate developed to show sensory canal system, SHRCM.G08175; d, internal view of immature left plate, SHRCM.G08174. e-h, oral plates; e, internal view of median oral plate; F, ditto, external view, a-a, b-b, lines of cross section, SHRCM.G08177/1-2; g, right posterolateral corner of median oral plate, showing region of abrasion, SHRCM.G08178. H, external view of left anterior lateral plate, c-c, line of cross section, SHRCM.G08179. i-u, scales; i-k, dorsal ridge scale, SHRCM.G08180/1-2; I, external view; j, internal view; K, lateral view; l-m, ventral ridge scale, SHRCM.G08181 ; l, external view; m, lateral view; n,o, internal views of ventral ridge scales, a-a, line of cross section, SHRCM.G08182, 08183; p,q, large flank scales, b-b line of cross section, SHRCM.G08190, 08191; R, external view of flank scale, SHRCM.G08185 ; s, internal view of flank scale, SHRCM.G08184; t, ?caudal scale, SHRCM.G08187; u, incomplete ?ventral lateral scale, c-c, line of cross section, SHRCM.G08189. Abbreviations: as, anterior side; bsp, broken spine; ca, calus; cn, concaved notch; cr, convexed central area; esc, exit pores for sensory canals; fo, foramina; poc, post-oral sensory canal; pop, post-oral process; vie, ventral longitudinal sensory canal. EXPLANATION OF PLATE 5 Fig. 1. Toombsaspis pococki (White), lower Devonian, Welsh Borderland, BU.2098/1, cast of dorsal headshield, x 4. Figs 2-5. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland. 2, SHRCM.G08179, cast of right anterior lateral plate. 3, SF1RCM.G08173/1, cast of right lateral plate. 4, SF1RCM.G08177/1, cast of median oral plate. 5, SHRCM.G08174, internal view of cast of immature left lateral plate. All x4. PLATE 5 TARRANT, Toombsaspis pococki , Phialaspis symondsi 418 PALAEONTOLOGY, VOLUME 34 long and 2 mm thick, at the posterior end of the right lateral side in SHRCM.G08178 (Text-fig. 9g). The worst of these abrasions appear to have been caused by friction against the anterior lateral plates. The anterior lateral plate is represented by one specimen SHRCM.G08179 (PI. 5, fig. 2; Text-fig. 9h). Although much larger, this resembles BMNH24788, a specimen that White (1946) described as possibly an anterior lateral plate from T. pococki. It measures 14 x 10 mm and is semicircular and bowed in shape. It rises to 4 mm, around a concave notch on one edge, and flattens towards the opposite edge. The inner surface is smooth, and the tubercles on the outer surface are progressively more abraded towards the raised region, where they are missing. The exoskeleton is pierced by foramina, which radiate in three rows around the concave notch. The largest of these foramina are ovate, 1 mm in diameter, and are angled to point towards the flattened part of the plate. These oral plates presumably lay freely edge to edge, with their raised concave notches lying antero-laterally and facing the anterior edges of the lateral plates. This supposition is based on the shape of the oral cavity, as manifested by its surrounding plates, and the shape and dimensions of the anterior lateral plates. As White (1946) observed with Phialaspis , the sensory canal system was variable, irregular, often asymmetrical, and in P. symondsi sometimes segmented. It also ranged considerably in depth within the exoskeleton. Grooves underlying sensory canals can be seen with varying clarity on the internal moulds. These are not to be confused with impressions marking the edges of units (Text-figs 2b and 6b). In places, rows of pores can be traced across the external surface of several specimens (Text-figs 2 a, 4c, 8i). White (1946) considered that T. pococki had paired, medial and lateral dorsal longitudinal canals joined by medial and lateral transverse commissures, although he suggested that the lateral dorsal longitudinal canals may have been incomplete or sometimes absent. It is possible that the lateral transverse commissures were occasionally partly joined at their lateral extremities by a longitudinal canal, but I have found no evidence in T. pococki or P. symondsi to support the presence of lateral dorsal longitudinal canals in the positions suggested by White. Instead, it would appear that they were isolated, except at their anterior end, from the lateral transverse commissures, and are represented by the branchial canals described in T. pococki by White (1946). In P. symondsi , these run under the longitudinal rows of units on the branchio-cornual plates (Text-fig. 6 a). The inter-orbital canal forms a crescent on the pineal plate encircling the posterior and lateral sides of the pineal organ (Text-fig. 4e,f). It joins the medial dorsal longitudinal canals. On each side, it meets a supra- orbital canal and a transverse canal which run on to the orbital plate (Text-figs 4 and 13). and join before meeting the circum-orbital canal (Text-figs 4a-c, 13, 14). The circum-orbital canal completely encircles the orbit. Radiating from it, are the anterior lateral transverse commissure, the lateral dorsal longitudinal canal, and a canal which runs ventrally on to the lateral plate (Text-figs 4c and 9c) and joins the post-oral and ventral longitudinal canals. White (1946) observed no post-oral canals in the P. symondsi ventral discs he studied. This is often the case, as the post-oral canals were frequently short and confined to the lateral plates, although sometimes they were present on the ventral discs and V-shaped (Text-fig. 8i). As White noted, the ventral longitudinal canals underlaid the longitudinal rows of units and were varied, often segmented posteriorly (Text-fig. 8b,i). Because specimens of the rostrum and anterior lateral plates are rare, no attempt has been made to expose possible sensory canals in these regions. However, no evidence has been found for sensory canals either leading to or on them, and it would appear that these were either absent or not linked to the main sensory canal system. A variety of scales has been found, scattered thinly among the larger P. symondsi plates. Most are superficially pteraspid-like, but the anterior flank scales are proportionally much larger. Dorsal ridge scales measure 8 mm long x 5 mm wide to 17 mm long x 16 mm wide. They are rounded and flattened anteriorly and rise towards the back to create an overlapping region where, although broken on SHRCM.G08177 (Text-fig. 9i-k), they appear to have been crowned with a low spine. The undersurface is gently dished, with two exit pores near the back for the medial dorsal longitudinal sensory canals. The ventral ridge scales range from 8 mm long x 4 mm wide to over 17 mm long x 9 mm wide. They are elongated and spinate (Text-fig. 9l-o), with a small anterior region of attachment angled at about 60° from a hollowed, posterior undersurface. This shows that they were considerably raised and possibly, overlapped strongly. Several flank scales, including two very large specimens (Text-fig. 9p,q), are asymmetrical and somewhat flattened, and range in size from 11x11 mm to 16x18 mm. A depressed region running across one anterior corner indicates an overlapped region, the majority of flank scales collected are smaller, ranging from 7 mm long x 8 mm wide to 10 mm longxl3mm wide. They are diamond shaped (Text-fig. 9r,s) and folded longitudinally to leave a slightly raised posterior and a somewhat flattened anterior corner, suggesting regions of overlap. Several possibly incomplete, anterior ventral lateral scales are asymmetrical, elongated, and folded longitudinally, to present two unequal sides (Text-fig. 9u). They range from 15 mm long x 10 mm wide to over 15 mm long x 15 mm wide. Two small crescent scales are possibly caudal in origin (Text-fig. 9t). TARRANT: LOWER DEVONIAN OSTRACODERM 419 text-fig. 10. Phialaspis symondsi (Lankester), regions of injury, a-d, on anterior parts of ventral discs; a,b,c, SHRCM.G3802/G08168/G08195.1-2 respectively; d, BGSGSM31380. E, deformed lateral plate, SHRCM.G08196. F, scar on longitudinal unit of ventral disc, SHRCM.G38197. Abbreviations: as, anterior side; dar, damaged region; der, deformed region; hf, healed fracture; sea, scar; vi, vascular impressions. Injuries and predation scars. White (1946, figs 53 and 54) observed the impression of a healed fracture in the left anterior corner of the P. symondsi ventral disc BMNH194. The injury had healed perfectly showing, as White suggested, that the injury must have happened some time before the animal’s death. Of ventral discs collected from Devil’s Hole which were complete enough to observe the anterior portion, 44% showed injuries like the fracture described by White (PI. 6, fig. 5), which suggests a common specific kind of injury. Although many of these injuries consist of healed fractures, on several specimens one or both antero-lateral corners are missing (see PI. 6, figs 3-5; Text-fig. 10 a-d). This shows a failure of the broken components to knit. In certain individuals (Text-fig. 10 b), narrow regions of outward growth, running across these more severe injuries, show that the animals died soon after their occurrence. In contrast, other specimens (Text-fig. 10c,d) show a reasonable amount of post-injury plate growth. These injuries are extreme in the type specimen BGS(GSM)31380 (Text-fig. 10d), where both antero-lateral corners and a medially placed segment at the front are missing. A deformed, right lateral plate (Text-fig. 1 0 e) has the usual concave contact edge with the ventral disc straight. This may correspond to the injuries on the ventral discs. A ventral disc SHRCM.G08170 and a branchio-cornual plate SHRCM.G08154 (Text-figs 6g and 8c) are both pierced by a 1-5 mm wide circular hole. Tarlo (1966) observed a hole of a similar kind in the branchial plate of Psammosteus praecursor, which he considered was caused by a crossopterygian bite. It is possible that the holes in P. symondsi may have been caused by the bite of a large ischnaeanthid. Much the same may apply to a well-healed, semi-circular scar on a longitudinal unit of the ventral disc SHRCM.G38197 (Text-fig. 1 0 f). Remarks. Dineley ( 1964) described a dorsal disc, NMC10373, from the Knoydart Formation, Nova Scotia, Canada, which he considered was sufficiently close to the specimens from the Anglo-Welsh region to call it Traquairaspis symondsi. However, until more Nova Scotian material is described, it is unjustifiable to consider that this specimen belongs to P. symondsi , and it is probably best referred to as Phialaspis sp. Genus toombsaspis gen. nov. Etymology. In remembrance of the late Mr H. A. Toombs, and asp is, Greek, shield. 420 PALAEONTOLOGY, VOLUME 34 Type species. Phialaspis pococki White, 1946 Other species assigned. Toombsaspis sabrinae (White, 1946) Diagnosis. Small phialaspidids with lateral keels. Low dorsal vane. Ventral, longitudinal cyclomoriform units against each lateral side of ventral, smooth central area. Ornament of stellated, equilateral and elongated tubercles divided by fine ridges. Ventral disc tubercles elongated on sides, in stacked V-shapes at back. Toombsaspis pococki (White, 1946) Plate 5, fig. I ; Text-figs 1,11, 14c-g, 15h; Table 2 1946 Phialaspis pococki White, pp. 217-229, pi. 12, fig. 1 ; figs 1, 3-8, 12-19, 22-27, 31-35, 39, 40M4, 55. 1948 Traquairaspis pococki (White); White and Toombs, p. 55, pi. 7, fig. 1. Holotype. BMNH24511 dorsal disc. Horizon and localities. Upper Silurian/Lower Devonian, Upper Downton Group. The Lower Old Red Sandstone of the Anglo-Welsh region (Text-fig. 1). table 2. Maximum dimensions of Toombsaspis pococki plates in millimetres. Abbreviation : pop, post-oral process. Range Average Length Width Ratio of width to length Length Width Ratio of width to length Dorsal discs 22-31 22-29 0-73-1 00 28 25 0-89 Orbital plates 11-13 7 0-53-0-63 12 7 0-58 Pineal plates 5-7 6-7 1-00-1-20 6 6 1-00 Rostrums 2 9 4-5 Branchio-cornuals 25-29 6-7 0-24 27 6-5 0-24 Ventral discs 32-39 18-25 0-53-0-69 35 22 0-63 Lateral plates 4- pop 8 7 0-87 Lateral plates — pop 14 7 0-50 Curvital dimensions Dorsal discs 22-31 24-31 0.89-1.09 28 27 0-96 Ventral discs 33^40 24-30 0-68-0-80 36 28 0-77 Referred material. Specimens housed in the BMNH (especially BMNH24751 and BMNH24568-9) and BU (especially BU2096-2102). Diagnosis. Dorsal and ventral discs approximately evenly vaulted. Dorsal vane small with one cyclomoriform unit, dorsal median keel with two. Rostrum short. Dorsal disc tubercles long on units, equilateral on periphery. Description. Internal impressions marking the edges of cyclomoriform units are not usually found, although White (1946) observed internal impressions in the dorsal disc BMNH24512, which he considered were left by sensory canals. These appear to resemble the internal impressions marking the edges of units in the dorsal discs of P. symondsi. TARRANT: LOWER DEVONIAN OSTRACODERM 421 text-fig. 1 1 Toombsaspis pococki (White), a, pineal plate fused to orbital plate, BMNH24568-9. b, dorsal view of rostral plate, BMNH24751. c, ditto, anterior view, with detail of tip ornamentation, d, right lateral plate, a-a, b-b, lines of cross section, BU2097. e, internal view of left anterior side of dorsal disc showing sensory canals, BU2100o. f, fragmentary headshield, showing left orbital plate, and part of pineal plate and dorsal disc, BU2098. G, right lateral plate showing sensory canal system, BU481A h, internal view of fragmentary dorsal disc showing sensory canals, BU2101. Abbreviations: cor, circum-orbital sensory canal; dd, dorsal disc; ?ior, possible inter-orbital sensory canal; ltc, lateral transverse sensory commissures; mdc, medial dorsal longitudinal sensory canal; mtc, medial transverse sensory commissures; or, orbital opening; orp, orbital plate; pi, pineal opening; pip, pineal plate; poc, post-oral sensory canal; vie, ventral longitudinal sensory canal. The dorsal disc is similar in shape to that of P. symondsi and two longitudinal rows of cyclomoriform units are found on each lateral side. The dorsal vane ranges from 2 to 3 mm in height and 6 to 9 mm in length. The orbital plate (Text-fig. 11a,f), which is more gently curved than that of P. symondsi , is approximately ovate to diamond shaped. The orbital opening ranges from 1 to T5 mm, in diameter. The pineal plate (Text-fig. 1 1 a) is triangular in shape, with the greatest width at the posterior end. The pineal organ is centrally placed and penetrates the surface of the plate. The ornament on both the pineal and orbital plates is cyclomoriform, encircling the openings. BU2098 (PI. 5, fig. 1 ; Text-fig. 1 1 f) is a fragment of a dorsal headshield showing plate fusion between the dorsal disc, left orbital plate and pineal plate. The pineal organ can only be seen as an internal impression on the counterpart. The orbital plate forms a ridge at its anterior edge, where it would have met the rostrum. BMNH24568-9 (Text-fig. 1 1 a), identified by White (1946) as an orbital plate, is a pineal plate fused to a right orbital plate. The rostrum BMNH24751 (Text-fig. 1 1 b,c) was originally considered to have been a pineal plate (White 1946). It is proportionally much shorter than the immature P. symondsi rostrum, and more closely resembles that of ITraquairaspis adunata Dineley and Loeffler, 1976, with elongated tubercles running in rows across its anterior end and no prominent anterior apex. The branchio-cornual plates (White 1946) are proportionally flatter and less massive than those of P. symondsi and, with no lateral vanes, their shape is generally closer to those of Traquairaspis campbelli. The branchial opening is about 2-5 mm long x T5 mm wide. It faces dorso-laterally and is located at about three- fifths along the length of the plate from the front. The tubercles on the dorsal side tend to be elongate, and a row of cyclomoriform units overlies the lateral dorsal longitudinal sensory canal. Excluding the smooth central area, the ventral disc (White 1946) is more vaulted than that of P. symondsi and, on average it has proportionally larger central area. This ranges from 24-27 mm in length to 10 to 12 mm in width. There is a well-defined row of cyclomoriform units resting against each side of the central area, with no intervening rows of ornamented growth. Although much smaller, the lateral plates (Text-fig. 1 1 d,g) closely resemble those of P. symondsi. They 422 PALAEONTOLOGY, VOLUME 34 usually have an attached post-oral process. Nevertheless, White (1946) illustrated a specimen, BU4816, with the post-oral process apparently missing (Text-fig. I1g). White also described a possible anterior lateral plate BMNH24788 and a ridge scale BMNH24759. These appear to be similar to their corresponding parts in P. symondsi. The sensory canal system (Text-figs 1 1 e,f,h and 14) is incompletely known, but appears to be arranged similarly to that of P. symondsi. However, the dorsal lateral transverse commissures are longer, and the inter- orbital canal may have extended into the anterior edge of the dorsal disc (Text-fig. 1 1 e). Remarks. T. pococki , which retains more in common with the earlier Traquairaspididae than P. symondsi , must be considered as a more primitive phialaspidid. The specimens of a traquair- aspidiform from the Red Bay Series, Fraenkelryggen Formation, Spitsbergen, considered by Blieck (1983) as Traquairaspis cf. pococki , are considerably larger than comparable Anglo-Welsh T. pococki specimens, and are provisionally assigned to Traquairaspidiform farm, gen. et sp. indet. Toombsaspis sabrinae (White, 1946) 1946 Phialaspis pococki var. sabrinae White, pp. 217-229, pi. 12, figs 2^4; figs 2, 9-1 1, 20, 21, 28-30, 56. Holotype. S4, dorsal disc (White 1946). Type horizon and locality. Upper Silurian/Lower Devonian, Upper Downton Group, Lower Old Red Sandstone, Sharpness, Gloucestershire, England (Text-fig. 1). Diagnosis. Dorsal disc approximately 30 mm long with equilateral tubercles. Dorsal vane large with long median tubercle, continuous with dorsal keel. Genus munchoaspis nov. Etymology. After Lake Muncho, British Columbia, and as pis, Greek, shield. Type species. Traquairaspis denisoni Dineley, 1964. Diagnosis. Dorsal disc approximately ovate, attaining length of 100 mm, with median keel, no dorsal vane, longitudinal carina on each lateral side marking the change in vaulting and double cyclomoriform whorl on the anterior. Ornament in long fine ridges which run parallel to the anterior and lateral edges. Munchoaspis denisoni (Dineley, 1964) comb. nov. 1964 Traquairaspis denisoni Dineley, pp. 211-215, pi. 38; text-figs 1-4. Holotype. NMC 10371, dorsal disc. Type horizon and locality. Silurian, Ludlow/Pridoli, North West of Lake Muncho, British Columbia, Canada. Diagnosis. As for genus, the only known species. Remarks. Dineley (1964) described several incomplete ventral discs from Canada, which he considered were indistinguishable in outline from the British ones. The smooth ventral central region, surrounded by a gently sloping ornamented brim with peripheral adult growth impressions, is a further typical phialaspidid characteristic. The early occurrence of this species would appear to strengthen Dineley and Loeffler’s (1976) claim for a traquairaspidiform evolutionary centre in Western and Arctic Canada. TARRANT: LOWER DEVONIAN OSTRACODERM 423 Family traquairaspididae Kiaer, 1932 Type genus. Traquairaspis Kiaer, 1932 Other genus assigned. Rimasventeraspis nom. nov. Diagnosis. Ventral disc with narrow ornamented margins, steep lateral sides each with a longitudinal row of elongated tubercles surrounded by cyclomorial fine ridges. The posterior edge sometimes medially notched. Large ventral central area extending to posterior edge, either totally smooth, partly subdivided, or with irregular dentine ridges, and on the anterior half, ventral medial commissures and post-oral sensory canals. Genus traquairaspis Kiaer, 1932 Type species. Cyathaspis campbelli Traquair, 1913 Diagnosis. Dorsal disc not fused to adjacent plates, ornamented with twelve or more, alternating, longitudinally running rows of small cyclomoriform units. Low dorsal, postero-medial keel. The branchio-cornual plates narrowly keeled behind the enclosed branchial openings. Two distinct types of ventral discs; type 1- smooth central area extending the length of plate (White 1946), type 2- posterior margin deeply notched, median region with a maze of dentine ridged units (Tarlo 1960). Traquairaspis Campbell i (Traquair, 1913) Text-fig. 12; Table 3 1911 Cyathaspis n.sp. Traquair in Campbell, p. 66. 1913 Cyathaspis campbelli Traquair in Campbell, p. 932. 1932 Traquairaspis campbelli (Traquair); Kiaer, pp. 25-26, pi. 11. 1946 Phialaspis pococki subsp. cowiensis White, p. 239, figs 36-38. table 3. Maximum dimensions of Traquairaspis campbelli plates in millimetres. Average Length Width Ratio of width to length Dorsal disc 39 27 0-69 Branchio-cornual 39 1 1 0-28 Ventral disc type 1 44 27 0 61 Ventral disc type 2 49 22 0-44 Holotype. RSM1960. 14. 1. Horizon and locality. Upper Silurian, Pridoli, Stonehaven, Kincardineshire, Scotland. Referred material. Specimens in the BMNH. Diagnosis. As for genus. Description. The dorsal disc is four-sided, vaulted posteriorly, flattened anteriorly, its lateral and posterior edges are gently convex, its anterior edge is indented, and it has slightly raised tubercles at the posterior edge. A broken plate (Text-fig. 1 2 d) located on the slab BMNH27388, measures 8 x 7 mm, and is perforated 424 PALAEONTOLOGY, VOLUME 34 medially by a 2 mm wide foramen. Although Dineley and Loeffler (1976) have described distinctive pineal foramina in several Canadian traquairaspidiforms, the large size of the opening shows that this specimen probably represents an orbital plate. The branchial opening is about 4 mm long x 3 mm wide, postero-laterally facing, and located at about two- fifths along the length, from the front of the branchio-cornual plate (Text-fig. 1 2 e). Fine elongated tubercles run longitudinally behind, and curve across the plate in front of, the branchial opening. Although the two types of ventral disc could prove to indicate two distinct species, they have identical lateral ornamentation and proportional overlap (Table 3; Text-fig. 12a,b). Also, problems occur with categorizing the other plates into two types. This may indicate that the two types of ventral discs are dimorphic, possibly sexual, examples of the same species. White (1946) showed the segmented longitudinal sensory canals running in association with the longitudinal row of tubercles on each lateral side of ventral disc type 1. Pores show the te (T-fig. 12. Traquairaspis campbelli (Traquair). a, ventral disc type 1, a-a, b-b, lines of cross section, BI/1NH37379, b, ventral disc type 2, a-a, b-b, lines of cross section, BMNH27037. c, ditto, detail of ornamentation on central area. D, fragmentary orbital plate, p.27388. E, branchio-cornual plate, plus cross section, BMNH43544. f, detail of dorsal disc ornamentation, BMNH43523. G, flank scale, on BMNH43525. h, ridge scale, on BMNH43525. Abbreviation: sea, smooth central area. presence of post-oral canals running on to the anterior half of the smooth central area. These also show the positions of probable ventral medial sensory commissures. On the slab BMNH43525, there are scales of two types. The ridge scales (Text-fig. 1 2 h) lack the pronounced spine or spinal process of phialaspidids, and instead they have a medially placed elongated tubercle. The flank scales (Text-fig. 1 2 g) are very wide compared to those of P. symondsi. The most complete ridge scale measures 6 mm long x 4 mm wide and the most complete flank scale measures 5 mm long x 10 mm wide. Remarks. The morphological similarity and contemporaneity with the Canadian traquair- aspidiforms supports Dineley and Loeffler’s (1976) assignment of its part of Scotland to the North American Silurian continent. This arrangement is shown by Scotese el al. (1985) on their Silurian and Devonian base maps. Genus rimasventeraspis nom. nov. Etymology. Rimas venter , Latin, fissured belly, and as pis. Greek, shield. Type species. ? Traquairaspis angusta Denison, 1963. Remarks. The previous generic name is pre-occupied ( Yukonaspis Kobayshi, 1936). TARRANT LOWER DEVONIAN OSTRACODERM 425 Diagnosis. Ventral disc; 80-85 mm long x 35-5 mm wide, with medially notched posterior edge. Ventral smooth central area covers nearly all the disc, is partly subdivided into units which grade into tubercles on antero-lateral edges. Ornament of stellated tubercles became elongated and divided by fine ridges on lateral sides. Rimasventeraspis august a (Denison, 1963) 1963 ITraquairaspis angusta Denison, pp. 132-135, figs 78 and 79. 1964 Yukonaspis angusta (Denison); Obruchev, p. 63; Stensio, p. 364, fig. 123a. Holotype. PU 17388, ventral disc. Type horizon and locality. Silurian, Ludlow/Pridoli, Beaver River, South-eastern Yukon, Canada. Diagnosis. As for genus, the only known species. Remarks. A ventral disc fragment described by Dineley and Loetfler (1976) as ‘Traquairaspididae indet. Type 1 ’, from the Pridoli of the Delorme Formation, Mackenzie, Canada, may be conspecific with, or closely related to, R. angusta. Discussion. The Traquairaspidinae are readily distinguished from the Phialaspidinae, by a ventral disc with steep sides and a large ventral central region extending to the posterior edge. The ornamentation of small cyclomoriform units suggests a more scale-like dermal arrangement than is found on the phialaspidinids. This, together with the wide flank scales and the absence of a specific pattern of adult growth, would suggest a more primitive condition, in comparison with undifferentiated very scale-like ornamentation and extremely wide, spindly scales of the Ordovician heterostracon Arandaspis (Ritchie and Gilbert-Tomlinson 1977). RESTORATION OF PHIALASPIDID CARAPACES The reconstructions of phialaspidid carapaces (Text-figs 13 and 14) are based upon average measurements (Tables 1 and 2) because of the large proportional range of individual plates, in particular in P. symondsi. They have been based upon specimens showing plate fusion, the matching of similarly shaped and sized edges, the similarity of alignment and type of ornament, the matching of the sensory canal system and plate orientation in other heterostracans. Impressions were taken of individual specimens of each component plate, and models were made for both P. symondsi and T. pococki. This has shown that the plate arrangements in both genera were identical, with the exception of the junction of orbital plates of P. symondsi between the rostrum and the pineal plate. In certain regions one edge is often more strongly angled than its corresponding plate margin, suggesting (White 1946) the former presence of small areas of connective tissue. INTERNAL ANATOMY Impressions of internal organs on the inside of the plates tend to be obscured by impressions of plate growth. Partly because of this, with the exception of the pineal organ, there are no distinguishable impressions of the brain, the semicircular canals, or nasal sacs. Impressions of the vascular system. The impressions of vessels and possibly nerve fibres, in the basal laminated layer of the exoskeleton, can be seen on many of the specimens of P. symondsi from Devil's Hole. These are clearest where radiating from the centres on the interiors of the ventral discs (Text-fig. 10c). The impressions are too incomplete to observe their general ramification, but, at frequent intervals along their lengths, branches leave at right-angles to run through the basal laminated layer into the exoskeleton. As Janvier and Blieck ( 1979) have observed, these are usually seen as small foramina in the basal laminated layer in heterostracans. 426 PALAEONTOLOGY, VOLUME 34 bro mdc 20m m IP /*/ I.W'I *J I ft/:‘ ■'.->) V • poc .vie MPi / ; / |»p^ / / vv ^ ✓* m“T text-fig. 13. Phialaspis symondsi (Lankester), reconstruction ot headshield. a,b, dorsal view showing ornamentation and sensory canal system. c,d, ventral view showing ornamentation and sensory canal system. Abbreviations: alp, anterior lateral plate; bep, branchio-cornual plate; bro, branchial opening, cor, circum- orbital sensory canal; dd, dorsal disc; dv, dorsal vane; ior, inter-orbital sensory canal; ldc, lateral dorsal longitudinal sensory canal; lp, lateral plate; ltc, lateral transverse sensory commissure; mdc, medial dorsal longitudinal sensory canal; mop, median oral plate; mtc, medial transverse sensory commissure, or, oibit, orp, orbital plate; pi. pineal opening; pip, pineal plate; poc, post-oral sensory canal; ro, rostrum, sea, smooth central area; sor, supra-orbital sensory canal; vd, ventral disc; vie, ventral longitudinal sensory canal. TARRANT: LOWER DEVONIAN OSTRACODERM 427 B text-fig. 14. Reconstruction of headshields. a,b, Phialaspis symondsi (Lankester); a, lateral view; b, lateral view of anterior portion with sensory canal system. c-G, Toombsaspis pococki (White); c,d, dorsal view showing ornamentation and sensory canal system; e,f, ventral view showing ornamentation and sensory canal system; G, lateral view. (See Text-fig. 13 for symbols) A cone-shaped structure (Text-fig. 1 5 F) found only on the branchio-cornual plate, SHRCM.G081 52/1, runs into the exoskeleton of the lateral keel, from near the branchial opening on the branchial duct. It is 6 mm long, 4 mm wide at its base, tapers to I mm wide at its tip, and lies at an antero-lateral angle of 50° from the branchial duct. Impressions of vessels adjoin it in places, in particular at the tip. Branchial structures. The internal, paired and ovate impressions, running in longitudinal rows along heterostracan dorsal and ventral shields, are generally considered to have been made by gill pouches, as originally suggested by Woodward (1891). Stensio (1958) interpreted longitudinal grooves on the ovate impressions as gill lamellae. Tarlo and Whiting (1965) considered that the paired impressions were made by head somites, which were used to pump the gills. In contrast, Janvier and Blieck (1979) considered that the cephalic somatic musculature was much reduced or absent in the Heterostraci, and its place filled by the branchial apparatus, and that the impressions they observed represented branchial and extrabranchial divisions of the gill pouches, visceral arches with attachment points to the exoskeleton, and an arrangement of nerves closely resembling those found on the branchial regions of the Osteostraci, and the ammocete larva. White (1946) recognised paired branchial impressions on the anterior parts of a ventral disc of T. camphelli. Although he was uncertain about the ‘lobes’ originally found by Wills (1935), on the smooth, ventral central area of P. symondsi , he observed a pattern of rounded ridges on the external surface of that region, in the type specimen of ‘ Psammosteus anglicus ’. The ventral discs from Devil’s Hole show that these impressions run around the shapes of three usually strong and commonly found internal impressions (Text-fig. 15a-c). The most anterior of these is medially 428 PALAEONTOLOGY, VOLUME 34 text-fig. 15. Internal impressions, a-g, Phialaspis symondsi (Lankester); a, anterior part of ventral disc showing branchial impressions, SHRCM.G3339; b, immature ventral disc showing branchial impressions, SHRCM.G3527A; c, internal impression of anterior part of ventral smooth central region, SHRCM.G08144; d, imperfect dorsal disc showing branchial impressions, SHRCM.G08144 ; E, internal view of lateral plate, SHRCM.G08173/2; f, internal view of branchio-cornual plate showing impressions on branchial duct, with detail of vascular structure, SHRCM.G08152/I g, internal view of orbital plate, SHRCM.G3387. h, Toombsaspis pococki (White), anterior of ventral disc showing internal impressions, BU2099. Abbreviations: as, anterior side; asca, anterior edge of smooth central region; bmb, branchial muscle block; brd, branchial duct; bro, branchial opening; bv, blood vessel; ci, central impression; gr, growth ridge; hbm, hypobranchial muscles; lsg, groove for longitudinal sensory canal; msc, muscle scars; or, orbital opening; pea, pre-branchial central impression; plb. posterior limit of branchial region; pop, post-oral process; pva, points of vascular attachment; sea, smooth central area; tvm, transverse muscles; va, visceral arch. EXPLANATION OF PLATE 6 Figs 1-5. Phialaspis symondsi (Lankester), lower Devonian, Welsh Borderland. 1, SHRCM.G3339/1, anterior of internal mould of ventral disc, x2. 2, SHRCM.G08152/1, internal impression, detail of branchial duct, x4. 3-5, regions of injury on ventral discs; 3, SHRCM.G08168, external right anterior side of cast, x 1 -5 ; 4, SHRCM.G08 195/1, anterior of internal mould, x L5; 5, SHRCM.G3302/1 . internal anterior of cast, x 1. PLATE 6 TARRANT, Phicilcispis symondsi 430 PALAEONTOLOGY, VOLUME 34 placed and rounded, with an average diameter of 10 mm. At its posterior end, the other impressions form a pair, join medially and fan out antero-laterally on each side, to define the antero-lateral edges of the smooth central region. The average measurements of each of these impressions are about 14 mm long x 7 mm wide. Branchial impressions can be best seen on the internal mould of the ventral disc, SHRCM.G3339 (PI. 6, fig. 1 ; Text-fig. 15a) and these run from the three centrally placed impressions to the antero- lateral edges of the disc. Their posterior edges are clearly defined, and Wills (1935) described these in his specimens, as grooves of indeterminate origin. Lines of beaded, 2 mm wide and raised impressions, divided by lines of pits, are contained on each side within a fan-shaped area. Seven or possibly eight rows are on the left side. On an immature ventral disc SHRCM.G08192 (Text-fig. 15b) the beaded impressions are found closer beneath the internally flattened, antero-lateral corners of the smooth central area, and the two sides are closer together. This resembles the arrangement on the ventral disc of T. pococki (Text-fig. 1 5 h). In the dorsal discs, only the distorted specimen SHRCM.G08194 (Text-fig. 1 5 D). shows any branchial impressions, and these are incomplete, but are of the same beaded type as those on the ventral discs. A row of Y- or U-shaped impressions, running along the dorsal side of the heterostracan branchial duct, and corresponding to the more medially placed branchial impressions, have been interpreted as part of the gill pouches (Kiaer 1930; Kiaer and Heintz 1935; Wills 1935), as impressions marking the positions of branchial pouch openings (Watson 1954; Stensio 1958, 1964; Tarlo and Whiting 1965; Jarvik 1980), or of visceral arches (Halstead 1982). The branchial duct in P. symondsi can be detected running longitudinally, from below the orbit and the deepest part of the lateral plate, to the branchial opening (Text-fig. 15e,g,f). It is seen most clearly on the branchio-cornual plate SHRCM.G08152/I (PI. 6, fig. 2; Text-fig. 1 5 f), where well- defined impressions run transverse across it, along the length of its dorsal side, and most strongly near the branchial opening. With the possible exception of a large blood vessel on the branchial duct of the orbital plate SHRCM.G3387 (Text-fig. 1 5 G), no obvious impressions of branchial blood vessels or nerves have been detected. The rows of beaded and depressed impressions undoubtedly represent the positions of visceral arches. The incompleteness and inconsistency of the impressions appears to indicate that the main respiratory movements were endoskeletal, and were mostly made by the branchial region when it was fully expanded. The flexibility and elasticity of the cartilaginous visceral arches would have been an important factor in the extension and contraction of the branchial regions. This explains the rows of beaded impressions, which would represent the positions of branchial muscle plates overlying the visceral arches, and transverse muscles running in between. The large paired impressions, usually found under the smooth ventral central area, have all the appearance of two large hypobranchial muscles, which would have served to raise and lower the branchial regions. The impressions on the branchial duct of P. symondsi could hardly be described as Y- or U- shaped, but rather as bands joining the more medial branchial regions, and swathing the branchial duct. It is unlikely that impressions left by the extrabranchial atria would be found on the lateral branchial region, since they would have been positioned away from the exoskeleton. The impressions in P. symondsi appear more like muscle bands, which would have strengthened the internally hollowed and bulky lateral exoskeleton, and could have forced water out through each branchial opening by longitudinal waves of compression, to aid in steering and in controlling pitch and roll. Janvier and Lund (1983) argued that hypobranchial somatic musculature, found on the myxinoids, anaspids, and to a lesser extent on the lampreys, mobilized the anterior parts of the body, compensating for the lack of paired fins. The same was possible for a juvenile P. symondsi at a stage prior to plate growth, as was suggested for the Heterostraci by Tarlo and Whiting (1965). These same muscular contractions could have been used by the adults, to control jet-aided steering and balance. It seems odd that the Heterostraci did not need paired fins; it seems likely that they had evolved TARRANT: LOWER DEVONIAN OSTRACODERM 431 their own substitute. P. symondsi , with its streamlined shape, large dorsal and lateral vanes, which indicate an active existence, and its obvious ability to frequent narrow meandering channels, must have manoeuvred more efficiently than is supposed for the Heterostraci, despite its rigid carapace. Water under pressure, forced out of the branchial opening on one side, would push the same side downwards, causing the animal to roll. If this coincided with a yaw in the same direction, the animal would bank, using its wide undersurface to effect a turn. If water was expelled with force from both branchial openings at the same time this would raise the anterior end, which could direct the animal upwards, and slow it down, or stop its forward motion, using the underside as a brake. It seems likely that this proposed method of jet-aided steering could have originally developed as a method of expelling debris from the large and enclosed branchial regions. Jarvik ( 1980) suggested that water expelled through the branchial openings of the pteraspidiforms would have aided the forward movement of the animals to some extent, as is known for modern actinopterygians. As many heterostracans are streamlined, especially so with certain large and advanced pteraspidiforms, it must be assumed that efficient manoeuvrable free-swimming must have been achieved, despite the inflexibility of the carapace. As the branchial openings on most species are directed posteriorly, it is probable that forward movements were jet-aided. The independent expulsion of water to aid in steering in these animals would be less efficient, and would have worked in the opposite way, to the method suggested for P. symondsi. Oral and olfactory apparatus and feeding methods. It has been generally accepted that two circular impressions found internally on the anterior edge of the heterostracan dorsal headshield indicate the position of nasal sacs, as first described by Jaekel ( 1903). Rostral spaces, medially divided to various degrees, have been found in certain pteraspidiforms (Stensio 1927, 1932a; Heintz 1962; Denison 1964, 1970) and in the cyathaspidid Torpedaspis (Broad and Dineley 1973). With the exception of Stensio (1958, 1964, 1968), who considered that the spaces were filled with cartilage, it has been generally agreed that they would have housed the anterior part of the nasal sacs. Although it has been considered that in some heterostracans the olfactory organ or organs opened into the buccal cavity (White 1935), notches on the anterior edge of the dorsal armour have been described as external nares (Kiaer and Heintz 1932; Watson 1954; Novitskaya 1975). Paired grooves on the rostral under surface of certain pteraspidiforms have been described as olfactory grooves (Zych 1931 ; Tarlo 1961), or as impressions indicating the position of tentacles (Stensio 1958; Janvier 1974; Jarvik 1980). Stensio (1927, 1958, 1964) was the first to suggest a close affinity between the Myxinoidea and the Heterostraci. In order to do this, he considered that the Heterostraci had a palatosubnasal lamina with 'upper labial plates’ against which the oral plates worked, separating the oral cavity from a single medially-placed olfactory organ duct and opening. As no fossil evidence of 'upper labial plates’ has been found, Denison (1960), White ( 1961 ), Tarlo ( 1961 ), Heintz (1962), Halstead (1973), and Novitskaya (1975), disagreed with Stensio’s suggested parts. Stensio was supported by Jarvik (1980) and by Janvier (1974) who later rejected a close relationship between the two classes, mainly because the Myxinoidea have a single semicircular canal and that the Heterostraci had two (Janvier and Blieck 1979), although they still maintained that the Heterostraci had a 'palatosubnasal lamina’, and favoured for most Heterostraci, a medial position for a single olfactory opening, duct and organ. In contrast, Halstead (1973) and Novitskaya (1975) considered that there were two olfactory organs, as in gnathostomes. The small size and the positions of phialaspidid orbital openings suggest a limited range of vision. Therefore, there must have been a heavy reliance upon well-developed olfactory organs, and possible tactile taste organs, to detect food. In P. symondsi , the folded and contorted under-surface of the back of the rostrum indicates a likely continuation of the external skin that covered the ventral pre-oral surface, and an attachment area for the soft dorsal parts of the mouth. The absence of rostral spaces, the large median oral plate which would have filled the oral cavity medially, plus the likely soft supportive and muscular structures of the oral region, suggest the anterior absence of a palatosubnasal lamina, and a more 432 PALAEONTOLOGY, VOLUME 34 text-fig. 16. Phialaspis symondsi (Lankester). a, ventral view of oral region, b, anterior part of head with mouth closed, c, ditto, with mouth open. Abbreviations: alp, anterior lateral plate; ba, barbels; lp, lateral plate; mop, median oral plate; na, narial opening; ns, nasal sac; pos, pre-oral surface. posteriorly placed olfactory complex, than is accepted on the pteraspidiforms. This indicates lateral positions for possible inhalant openings. It seems that the raised notch on the anterior lateral plate represents an inhalant opening, indicating that P. symondsi had paired inhalant olfactory ducts, the foramina surrounding the raised notch might suggest the positions of tactile and possible taste organs. These could possibly be extensions of the olfactory apparatus, as in the myxinoids (Janvier 1974). The raised notch and foramina may have served to house a large tentacle on each side of the oral cavity. The abrasions on the sides of the median oral plate appear to have been caused by friction against the overlapping anterior lateral plates, indicating that the latter were hinged at their posterior edges, and would have swung open as the median oral plate was extruded. This action, taking into account the shape of the front of the lateral plates, could have been restrained by such tactile organs. It seems likely that the nasal sacs would have been separated, and have rested under the anterior of the orbital plates. A more medial position for a single olfactory organ would have meant that it had to rest under the telencephalon, which would have involved excessive cranial flexure and where there would have been insufficient room. Georgieva et al. (1979) considered the ‘sensory buds’ on the barbels of Myxine glutinosa resembled the taste buds of the gnathostomes, and Baatrup (1983) described sensory buds in larval lampreys akin to the taste buds of other vertebrates. Therefore, it is feasible that P. symondsi may have possessed similar structures, in particular on its tactile organs. Various suggestions have been made about the oral workings of Heterostraci, particularly the pteraspidiforms and certain cyathaspidiforms. Kiaer (1928) considered that the oral plates bit against the maxillary brim, on the ventral margin of the rostral region. Stensio (1932) and Janvier (1974) thought that they worked in a myxinoid-like manner. White (1935) considered that the oral plates were connected together by the epidermis, and would have moved down and forwards, to form a scoop or shovel, and Denison (1961) further suggested that the protrusible mouth could have selected and picked up food, including small invertebrates. This could have been aided by inhalant respiratory currents. Dineley and Loeffler (1976) described a large plate in the oral region of Poraspis cf. polaris , which they interpreted as a large single oral plate used as a scoop. P. symondsi had far fewer oral components than the pteraspidiforms, and it is inconceivable to imagine its large median oral plate retracting, Myxine -fashion, into its gullet. The shape of the median oral plate indicates that it would have worked in the way that White (1935) and Denison (1961) described for the pteraspidiforms. The elongate and convex area on its inner side indicates an attachment area for protractor and retractor muscles, and this suggests that the median oral plate could have, if needed, worked rapidly, snapping shut with force. The smooth edges show that TARRANT: LOWER DEVONIAN OSTRACODERM 433 it had no grasping or cutting facilities, although it may have worked against the maxillary flange and pre-oral surface. The size of the oral cavity, surrounded by rigid lateral plates, limited the size of food engulfed. Nevertheless, the oral region of P. symondsi has all the appearance of working like an efficient trap, with its scooping median oral plate embraced by anterior lateral plates. The shape of P. symondsi , albeit constricted by an inflexible carapace, has the lines of an active feeder, rather than a sluggish animal swallowing mud and filtering organic substances, as has often been supposed for the Heterostraci (Halstead 1985). The apparent lack of wear on the tip of the median oral plate appears to substantiate this. The small size and structure of the oral region would have prevented total filter feeding in open water. As White (1946) suggested for Phialaspis , the smooth ventral central area could have been used as a sliding plane and fulcrum, while the animal wriggled across the surface of the substrate. Taking the dorso-ventrally flattened, and anteriorly heavy, carapace into consideration, plus occasional abrasion observed on the anterior part of the ventral discs and undersurface of the branchio-cornual plates, the crenulated tip of the rostrum, and the ventral position of the oral region, it seems likely that P. symondsi was mainly a benthic feeder, rooting in the substrate. This, plus its common and wide occurrence, indicates that it was not a highly specialized feeder, but more of an opportunist, feeding on a wide range of animal and vegetable matter, both dead and alive. Its small mouth rules out any extensive predatory role, but it appears well-equipped to snap up small animals, which it would have disturbed out of the substrate. T. pococki had a more evenly vaulted cephalothorax and a short rostrum. Its oral region was more terminal in position (Text-fig. 14e-g), which indicates that it may have fed not so much within the substrate, but more on or possibly somewhat above its surface. GROWTH AND ONTOGENY Despite divergent views on heterostracan exoskeletal growth, evidence is patchy. From studies on elasmobranch scales, 0rvig (1951) developed the Lepidomorial Theory, which Stensio (1958) used to interpret heterostracan exoskeletal growth. This, he considered, was achieved in two ways: (1) cyclomorial growth, in which peripheral concentric growth took place around an initial primordium, and (2) synchronomorial growth, in which calcification was achieved simultaneously, to produce a completed part of the carapace. This was mainly based on the assumption that, once a part of the carapace mineralized, it remained unchanged, and that the mode of growth could be deduced from the form of dentine patterning. However, Dineley and Loeftler (1976) discovered concentric growth impressions in association with synchronomorial dentine patterns in certain cyathaspidiform shields. From this, they argued that the Lepidomorial Theory was not applicable to heterostracan exoskeletal growth, and was only useful to describe cyclomoriform and synchronomoriform ornamental pattern. An example of phylogenetic heterostracan exoskeletal growth can now be demonstrated, since Elliott (1984) has shown that the pteraspidi forms were derived from the cyathaspidiforms. The superficial layer formed prior to the underlying layers in the cyathaspid (Denision 1964), and during early ontogeny in the pteraspidiforms (Denison 1973; White 1973). The cyathaspidiform shield did not form until the animal had achieved its definitive size (Denison 1964; Dineley and Loeffler 1976), whereas the pteraspidiform shield grew as separate peripheral plates, which fused together at maturity (Heintz 1938; White 1958). This latter process was progressively delayed in later forms (White 1958). As Dineley and Loeffler (1976) argued, it is likely that the earliest traquairaspidiforms had an undivided dorsal shield, although how this was formed is open to speculation. Nevertheless, to aid in synchronous growth between the animal and its exoskeleton, later forms attained a mode of plate division parallel to the pteraspidiforms. The orbital, pineal and lateral plates mainly grew cyclomorially by peripheral additions. Much the same could be said about the regions of mature growth in the other major plates. Nevertheless, it would appear likely that some plate remodelling may have been required for fusions and to sustain the proportional vaulting and matching of peripheral plate contacts. Similar speculations 434 PALAEONTOLOGY, VOLUME 34 have been made about resorption and regrowth in the pteraspidiforms (Halstead 1969; Denison 1973 ; White 1973), although it has never been demonstrated. However, Tarlo (1965) has shown that certain heterostracans were capable of resorption and regrowth within their middle exoskeletal layers, to aid normal plate enlargement. It would appear that P. symondsi was at least capable of using resorption and regrowth to repair broken exoskeletal components (PI. 6, fig. 5; Text-fig. 10). Secondary formation of tubercles overlying and replacing primary tubercles has been described in various heterostracans. These formed to repair worn or damaged regions (Tarlo and Tarlo 1965; Tarlo 1965, 1966; Denison 1973), were preceded by resorption rather than wear (Gross 1961), or represented a normal process of growth (0rvig 1976). In P. symondsi the large tubercles overlying smaller tubercles on the lateral edges of the branchio-cornual plates and front part of the rostrum are located on likely regions of abrasion. Secondary tubercles forming to fill spaces between primary tubercles as a normal part of plate growth have not been described in the heterostracans. Despite this, the evidence of erupting and developing tubercles over the surface of an immature dorsal disc and branchio-cornual plate (Text- figs 2d and 6f) appears to show a ready ability to develop tubercles within the main body of the plates, as part of the general growth. It would seem to follow that this mode of growth of the superficial layer could have been accompanied by resorption and regrowth of the underlying exoskeletal layers. Tarlo (1962) considered that the traquairaspidiform units grew as isolated tesserae which ultimately fused with the main plates. Although no recognisable isolated tesserae have been found in the beds containing P. symondsi , it is conceivable that the units initially developed in Tarlo’s suggested fashion, as is apparent from the posterior units on the ventral disc. In certain instances, they may not have fused with the main plates until their growth had ceased. However, evidence of non-cyclomorially arranged developing tubercles in an immature dorsal disc (see Text-fig. 2d) and the small posterior unit in an immature branchio-cornual plate (Text-fig. 6 j ), suggests that each unit was also capable of growth whilst attached to its neighbouring units and the main plate, thus providing a more or less unified mode of outward plate growth. This contained method of growth could have caused the basal laminated layer to fold inwards at the regions of contact between each unit and their contact with the main plate, thus leaving the internal ‘constriction’ impressions often observed at the edges of the units (PI. 1, fig. 2; PI. 3, fig. 4; Text-figs 2b and 6b). This mode of growth enabled the ontogenetic and phylogenetic development of the folded units forming the dorsal vane. This would suggest that the depressed region in the small dorsal vane (PI. 1, fig. 3; Text-fig. 3d) may have contained recessed epithelial tissue, in which new tubercles would have formed. As is evident from the most immature branchio-cornual plates (PI. 3, figs 1 and 2; Text-fig. 6f-h), the longitudinal units were either isolated or not formed during early ontogeny. All the known earlier traquairaspidiform rostral plates (Dineley and Loeffler 1976) could be described as an antero-dorsal unit, formed mainly by cyclomorial peripheral growth, with enlarged tubercles at the anterior apex, where it folds to cover the dorsal margin of the oral cavity. However, in P. symondsi , there would appear to have been a new centre of cancellous and superficial layer growth within the main body of the plate, forming the ‘pre-oral field’. As Tarlo (1962) suggested, it would appear that the ventral, smooth central region achieved full size and developed an enclosing band of ornamented growth prior to its fusion with, or formation of, the ventral longitudinal units (PI. 4, fig. 3; Text-fig. 8c). The development of the ventral smooth central region is not seen in any specimens, and its formation is open to speculation. The ventral central ornamentation on certain Canadian Pridolian traquairaspidiforms (Dineley and Loeffler 1976) may illustrate the mode of origin. This grades from the unspecialized ventral tubercles in certain forms, to flattened and broad ventral ornamentation, which approaches the subdivided condition in the ventral smooth central region of Rimasventeraspis. The abraded regions in P. symondsi show (Text-figs 6d and 8 e) broad flattened tubercles like the ventral pattern in the Canadian traquairaspidiforms. Persistent abrasion on the ventral surface of active benthic animals might have triggered selection for a permanently smooth ventral central region. This would have greatly aided movement over the substrate, and might have evolved independently in different TARRANT: LOWER DEVONIAN OSTRACODERM 435 lineages. The anomalous and non-abraded ornamentation of the T. campbelli Type 2 ventral disc may suggest a different lifestyle, and that the tubercle formation was still inherent, despite the possible ancestral formation of a smooth ventral central region. The large size range of the phialaspidid adult plates is mainly due to the amount of outer peripheral growth. In the dorsal and ventral discs, growth ridges become more numerous as the region extends. This shows that the animals were capable of growth throughout life. The growth ridges influenced all exoskeletal levels and are seen most clearly as folds in the nasal laminated layer. By folding, the exoskeleton would have been able to have kept itself moulded to the animal, and it appears that the exoskeleton continued to grow for a time. The resulting excess of exoskeletal growth forced the growing edges of a plate downwards, then upwards at the resumption of underlying growth, to form a growth ridge. These corrugations not only mark the rhythmic growth cycles, but would also have strengthened the plates. The range in proportions of the dorsal vanes, lateral keels, branchial openings, smooth ventral regions, and the number and size of units, appears to have had nothing to do with adult plate growth. No consistent variation can be observed in these parts, and it seems unlikely that they represent species or sexual differences. From the most immature specimens, it is possible to estimate that P. symondsi developed its dermal plates when it was about one-third the length of the mature animal. This shows that an amocoete-like lifestyle was impractical, since a borrowing worm-shaped body, unimpeded by immobilizing plates of a carapace, would have been needed. Regardless of the great size range of orbital and pineal plates, the orbital and pineal openings show a small range in size. This suggests that the orbits and the pineal organ had probably reached full size at the onset of dermal plate development. Also, the posteriorly directed branchial ducts on the juvenile branchio-cornual plates suggest that a relatively large area of the thorax was free of the headshield. At this stage of development, the small animal would have needed sufficient mobility and field of vision to detect and evade predators. It is possible that it first fed upon planktonic organisms in the relative safety of shallow water, and moved into deeper water to consume larger food as it developed its armour and increased in size. CONCLUSIONS The morphology of Phialaspis symondsi and Toombsaspis pococki is sufficiently different from that of Traquairaspis campbelli , to necessitate the selection of two families, the Phialaspididae and the Traquairaspididae, within the order Traquairaspidiformes. Internal impressions on the phialaspidid plates are interpreted as branchial musculature swathing the branchial duct in association with the visceral arches, which could have been used to facilitate jet-aided manoeuvrability to compensate for the lack of paired fins. A conspicuous notch surrounded by foramina in an anterior lateral plate of P. symondsi suggests the occurrence of paired olfactory ducts in association with clusters of tactile and taste sensory organs. Dorsal and lateral swimming stabilizers and a smooth central ventral sliding plane in the Anglo- Welsh phialaspidids, suggests an active and mainly benthic lifestyle. These, their common occurrence, and the workings of the oral region in P. symondsi , would imply that these species were probably opportunist feeders, well able to catch and consume small benthic animals. Acknowledgements. My thanks go to Dr C. J. Cleal, Ms M. A. Rowlands and Mr A. M. Tarrant for valuable help in the field and to the GCR unit of the Nature Conservancy for its excavation of the Devil’s Hole stream section. For hospitality in their various museum departments, I thank Dr P. Forey and Ms S. Young, British Museum (Natural History); Dr D. White, The British Geological Survey; Mr P. Osborne, University of Birmingham Geology Museum; and, in particular, Mr J. Norton, Ludlow Museum. I wish to acknowledge useful correspondence with Dr M. M. Smith, Unit of Anatomy in relation to Dentistry, Guy’s Hospital, London and Dr J. D. D. Smith, International Commission on Zoological Nomenclature. Helpful advice and criticism were provided by Dr A. Blieck, Professor D. L. Dineley, Dr L. B. Halstead, Dr P. Janvier, and Dr E. J. Loeffier. The photographic illustrations are the work of Messrs T. Foxall and I. Miller. 436 PALAEONTOLOGY, VOLUME 34 REFERENCES allen, j. r. l. 1964. Studies in fluvial sedimentation: six cyclothems from the Lower Old Red Sandstone, Anglo-Welsh Basin. Sedimentology, 3, 163-198. — 1974n. Studies in fluvial sedimentation: implications of pedogenic carbonate units, Lower Old Red Sandstone, Anglo-Welsh outcrop. Geological Journal, 9, 181-204. — 19746. Sedimemtology of the Old Red Sandstone (Siluro-Devonian) in the Clee Hills area, Shropshire, England. Sedimentary Geology, 12, 73-167. — 1985. Marine to fresh water: the sedimentology of the interrupted environmental transition (Ludlow-Siegenian) in the Anglo-Welsh region. Philosophical Transactions of the Roval Society of London , Series B, 309, 85-104. — and tarlo, l. b. 1963. The Downtonian and Dittonian facies of the Welsh Borderland. Geological Magazine , 100. 1 29- 155. — and williams, b. p. j. 1978. The sequence of the earlier Lower Old Red Sandstone (Siluro-Devonian) north of Milford Haven, southwest Dyfed (Wales). Geological Journal , 13, 13-36. - 1981. Sedimentology and stratigraphy of the Townsend Tuff Bed (Lower Old Red Sandstone) in South Wales and the Welsh Borders. Journal of the Geological Society , 138, 15-29. baatrup, e. 1983. Ciliated receptors in the pharyngeal terminal buds of larval Lampetra planeri (Bloch) (Cyclostomata). Acta Zoologica , 64, 67-75. ball, h. w. and dineley, d. l. 1961. The Old Red Sandstone of Brown Clee and the adjacent area. 1. Stratigraphy. Bulletin of the British Museum (Natural History ), Geology, 5, 176-242. blieck, a. 1983. Biostratigraphie du Devonian inferieur du Spitsberg: donnees complementaires sur les heterostraces (Vertebres Agnathes) du Groupe de Red Bay. Bulletin du Musee national d'Histoire naturelle, Paris, 5, 75-111. — 1984. L,es heterostraces pteraspidiformes : systematique, phylogenie, biostratigraphie, biogeographie. Cahiers Paleontologiques (Vertebres), C.N.R.S. , 1-198. broad, d. s. and dineley, d. l. 1973. Torpedaspis : a new Upper Silurian and lower Devonian genus of Cyathaspididae (Ostracodermi) from Arctic Canada. Bulletin of the Geological Survey of Canada, 222, 53-90. denison, r. h. 1956. A review of the habitat of the earliest vertebrates. Fieldiana Geology, 11, 359-457. — 1961. Feeding mechanisms of Agnatha and early gnathostomes. American Zoologist, 1, 177-181. 1963. New Silurian Heterostraci from Southeastern Yukon. Fieldiana Geology, 14, 105-135. — 1964. The Cyathaspidae, a family of Silurian and Devonian jawless vertebrates. Fieldiana Geologv, 13, 309-473. — 1970. Revised classification of Pteraspididae with description of new forms from Wyoming. Fieldiana Geology, 20, 1-41. — 1971. On the tail of the Heterostraci. Forma et Functio , 4, 87-99. — 1973. Growth and wear of the shield in the Pteraspididae (Agnatha). Palaeontographica, Abteilung A, 143, 1-10. dineley, d. l. 1964. New specimens of Traquairaspis from Canada. Palaeontology, 1, 210-219. — and gossage, d. w. 1959. The Old Red Sandstone of the Cleobury Mortimer area, Shropshire. Proceedings of the Geologists' Association, 70, 221-238. — and loeffler, E. J. 1976. Ostracoderm faunas of the Delorme and associated Siluro-Devonian Formations, North West Territories, Canada. Special Papers in Palaeontology , 18, 1-214. elliott, d. k. 1984. A new subfamily of the Pteraspididae (Agnatha, Heterostraci) from the upper Silurian and lower Devonian of Arctic Canada. Palaeontology, 27, 169-197. georgieva, v., patzner, r. a. and adam, h. 1979. Transmissions- und Rasterelektronenmikroskopische Untersuchung an den Sinnesknospen der Tentakeln von Myxine glutinosa L. (Cyclostomata). Zoologica Scrip ta, 8. 61-67. gross, w. 1961. Aufbau des Panzers obersilurischer Heterostraci und Osteostraci Norddeutschlands (Geschiebe) und Oesels. Acta Zoologica, 42, 73-150. halstead, L. b. 1969. Calcified tissues in the earliest vertebrates. Calcified Tissue Research, 3, 107-134. - 1971. The presence of a spiracle in the Heterostraci (Agnatha). Journal of the Linnean Society (Zoology), 50, 195-197. - 1973. Affinities of the Heterostraci. Biological Journal of the Linnean Society, 5, 339-349. - 1982. Evolutionary trends and the phylogeny of the Agnatha. 156-196. In joysey, k. a. and Friday, a. e. (eds). Problems of phylogenetic reconstruction. Academic Press, London and New York, 442 pp. TARRANT: LOWER DEVONIAN OSTR ACODER M 437 — 1985. The vertebrate invasion of fresh water. Philosophical Transactions of the Royal Society of London, Series B, 309, 243-258. Harris, j. e. 1936. The role of the fins in the equilibrium of the swimming fish. 1 . Wind-tunnel tests on a model of Mustelus canis Mitchell. Journal of Experimental Biology, 13, 476-493. heintz, a. 1938. Uber die altesten bekannten Wirbeltiere. Naturwissenschaft , 26, 49-58. — 1962. Les organs olfactifs des Heterostraci. Collogues international du Centre national de Recherche Scientifiques, 104, 13-29. jaekel, o. 1903. Die Organisation und systematische Stellung der Asterolepiden. Zeitschrift der deutschen geologischen Gesellschaft, 55, 41-60. janvier, p. 1974. The structure of the naso-hypophysial complex and the mouth in fossil and extant cyclostomes, with remarks on the amphiaspiforms. Zoologicci Scripta, 3, 193-200. — and blieck, a. 1979. New data on the internal anatomy of the Heterostraci, with general remarks on the phylogeny of the craniates. Zoologica Scripta, 8, 287-296. — and lund, r. 1983. Hardistiella montanensis n. gen. et sp. (Petromyzontida) from the Lower Carboniferous of Montana, with remarks on the affinities of the lampreys. Journal of Vertebrate Paleontology, 2, 407 — 4 1 3. jarvik, E. 1980. Basic structure and evolution of vertebrates. Academic Press, London and New York, 575 pp. karatajute-talima, v. N. 1978. Telodont y silura i devona SSSR i Spicbergena. Mosklas publications, Vilnius, 334 pp. [In Russian]. kermack, k. a. 1943. The functional significance of the hypocercal tail in Pteraspis rostrata. Journal of Experimental Biology, 20, 23-27. kiaer, j. 1928. The structure of the mouth of the oldest known vertebrates, pteraspids and cephalaspids. Palaeobiologica, 1, 117-124. — 1930. Ctenaspis, a new genus of cyathaspidian fishes. Skrifter Svalbard Ishavet, 33, 1-7. — 1932. The Dowtonian and Devonian vertebrates of Spitzbergen. IV. Suborder Cyathaspida. Skrifter Svalbard Ishavet, 52, 1-26. — and heintz, a. 1935. The Downtonian and Devonian vertebrates of Spitsbergen. V. Suborder Cyathaspida, Part 1, Tribe Poraspidei, Family Poraspidae Kiaer. Skrifter Svalbard Ishavet, 40, 1-138. kobayshi, t. 1936. Cambrian and Lower Ordovician trilobites from North West Canada. Journal of Paleontology , 10, 157-167. lankester, e. r. 1868. The fishes of the Old Red Sandstone of Britain. Monograph of the Palaeontographical Society, 1. 1-62. Novitskaya, l. i. 1971. Diagnostic evaluation of the ornamentation of Agnatha and Pisces. Palaeontological Journal, 4, 494-506. — 1975. Sur la structure interne et les liens phylogenetiques des Heterostraci. Collaques international du Centre national des Recherches Scientifiques, 218, 31 — 40. Obruchev, D. v. 1964. Subclass Heterostraci (Pteraspides). Osnovy Paleontologii, 11, 45-82. — and karatjute-talimaa, v. m. 1967. Vertebrate faunas and correlation of the Ludlovian-Lower Devonian in Europe. Journal of the Linnean Society (Zoology), 47, 5-14. orvig, t. 1951. Historic studies of placoderms and fossil elasmobranchs. The endoskeleton, with remarks on the hard tissues of lower vertebrates in general. Arkiv for Zoologi, 2, 231-454. — 1976. Paleohistological notes. 3. The interpretation of pleromin (pleromic hard tissue) in the dermal skeleton of psammosteid heterostracans. Zoologica Scripta, 5, 35-47. ritchie, a. and gilbert-tomlinson, j. 1977. First Ordovician vertebrates from the southern hemisphere. Alcheringa, 1, 351-368. scotese, c. R., van der voo, r. and barrett, s. f. 1985. Silurian and Devonian base maps. Philosophical Transactions of the Royal Society of London, Series B, 309, 57-77 . squirrel, h. c. and downing, r. a. 1969. The geology of the South Wales Coalfield. Part 1. Memoirs of the Geological Survey of Great Britain, 249, 1-133. stensio, e. a. 1927. The Downtonian and Devonian vertebrates of Spitsbergen. Part 1. Family Cephalaspidae. Skrifter Svalbard, Nordishavet, 12, 1-391. — 1932. The cephalaspids of Great Britain. British Museum (Natural History), London, 220 pp. — 1958. Les cyclostomes fossiles. 173-420. In grasse, p. p. (ed.). Traite de zoologie, 13. Masson, Paris, 2758 pp. — 1964. Les cyclostomes fossiles ou ostracoderms. 96-385. In piveteteau, j. (ed.). Traite de Paleontologie , 4, Part 7. Masson, Paris, 387 pp. — 1968. The cyclostomes with special reference to the diphyletic origin of the Petromyzontida and Myxinoidea. Nobel Symposium, 4, 13-71. 438 PALAEONTOLOGY, VOLUME 34 tarlo, B. J. and tarlo, L. B. H. 1965. The origin of teeth. Discovery , 26, 20-26. tarlo, l. b. h. 1960. The Downtonian ostracoderm Corvaspis kingi Woodward, with notes on the development of dermal plates in the Heterostraci. Palaeontology, 3, 217-226. — 1961. Rhinopteraspis cornubica (McCoy) with notes on the classification and evolution of the pteraspids. Acta palaeontologica polonica , 6, 367-400. — 1962. The classification and evolution of the Heterostraci. Acta palaeontologica polonica, 7, 90-249. — 1965. Psammosteiforms (Agnatha). A review with descriptions of new material from the Lower Devonian of Poland. I. General Part. Palaeontologica polonica, 13, 1-135. — 1966. Psammosteiforms (Agnatha). A review with descriptions of new material from the Lower Devonian of Poland. II. Systematic Part. Palaeontologica polonica, 15, 1-168. — 1967. The tessellated pattern of dermal armour in the Heterostraci. Journal of the Linnean Society (Zoology), 47, 45-74. — and tarlo, b. j. 1961. Histological sections of the dermal armour of the psammosteid ostracoderms. Proceedings of the Geological Society of London, 1593, 3-4. — and whiting, h. p. 1965. A new interpretation of the internal anatomy of the Heterostraci (Agnatha). Nature, 206, 148-150. toombs, h. a. and rixon, a. e. 1950. The use of plastics in the ‘transfer method’ of preparing fossils. Museums Journal , 50, 105-107. traquair, r. h. 1898. Notes on Palaeozoic fishes. Annals and Magazine of Natural History, 11, 67-70. — 1911. In Campbell, r. Preliminary note on the geology of south-eastern Kincardineshire. Geological Magazine, 8, 63-69. — 1912. Note on the fish-remains collected by Messrs R. Campbell, W. T. Gordon and B. N. Peach in Palaeozoic strata at Cowie, Stonehaven. Reports of the British Association for the Advancement of Science, 1911, 463. — 1913. 923-960. In Campbell, r. The geology of south eastern Kincardineshire. Transactions of the Royal Society of Edinburgh , 48, 37 pp. turner, s. 1973. Siluro-Devonian thelodonts from the Welsh Borderland. Journal of the Geological Society of London. 129, 557-584. watson, d. m. s. 1954. A consideration of ostracoderms. Philosophical Transactions of the Royal Society of London , Series B , 238, 1-25. westoll, t. s. 1945. A new cephalaspid fish from the Downtonian of Scotland, with notes on the structure and classification of ostracoderms. Transactions of the Royal Society of Edinburgh , 61, 341-357. — 1967. Radotina and other tesserate fishes. Journal of the Linnean Society (Zoology), 47, 83-98. white, e. i. 1935. The ostracoderm Pteraspis Kner and the relationships of the agnathous vertebrates. Philosophical Transactions of the Royal Society of London, 225, 381-457. — 1946. The genus Phialaspis and the ‘ Psammosteus ’ Limestones. Quarterly Journal of the Geological Society of London, 101, 207-242. — 1950a. The vertebrate faunas of the Lower Old Red Sandstone of the Welsh Borders. Bulletin of the British Museum (Natural History), Geology, 1, 51-67. — 1950A Pteraspis leathensis White, a Dittonian zone-fossil. Bulletin of the British Museum (Natural History), Geology, 7, 69-89. — 1958. Original environment of the craniates. 212-234. In westoll, t. s. (ed.). Studies on fossil vertebrates. Athlone Press, London, 52 pp. — 1961. The Old Red Sandstone of Brown Clee Hill and the adjacent area. 2. Palaeontology. Bulletin of the British Museum (Natural History), Geology, 5, 243-310. — 1973. Form and growth in Belgicaspis (Heterostraci). Palaeontographica, Abteilung A, 143, 11-24. — and toombs, h. a. 1948. Guide to excursion C16. Vertebrate Palaeontology. International Geological Congress 18th Session, Great Britain, 4-14. wills, l. j. 1935. Rare and new ostracoderm fishes from the Downtonian of Shropshire. Transactions of the Royal Society of Edinburgh, 58. 427-447. woodward, A. s. 1891. Catalogue of the fossil fishes in the British Museum (Natural History). British Museum (Natural History), London, 567 pp. zych, w. 1931. Fauna ryb Devonii i Downtoni Podola. Pteraspidomorphi : Heterostraci. Palaeontologicheskii Sbornik, Lwow, 91 pp. [In Polish]. PETER REX TARRANT 8 St Gregory’s Close Typescript received 5 January 1990 Morville, Nr Bridgnorth Revised typescript received 20 March 1990 Shropshire WV16 4RL, UK THE RHYNCHONELLIDE BRACHIOPOD EOCOELIA FROM THE UPPER LLANDOVERY OF IRELAND AND SCOTLAND by E. N. DOYLE, A. N. HOEY and D. A. T. HARPER Abstract. Biometrical description oflarge samples (N > 300) of the rhynchonellide brachiopod Eocoelia from the Kilbride Formation (upper Telychian) in the west of Ireland and the Lower Camregan Grits (lower Telychian) of the Girvan district, south-west Scotland, suggests the refinement of the stratigraphically important Eocoelia lineage in the upper Llandovery. The Irish and Scottish species have previously both been assigned to Eocoelia curtisi Ziegler. However, the Girvan population is significantly different from type and topotype specimens from Tortworth and from the Irish material. The Scottish form is accorded separate subspecific status, Eocoelia curtisi immatura subsp. nov., whereas the Irish form is included in the nominate subspecies. The Irish and Scottish subspecies are within the upper and lower parts of the range of E. curtisi s.l. respectively. Interpolation within the lineage confirms some of the established morphological transpecific trends and may permit more precise correlation within the upper Llandovery. The distinctive rhynchonellide brachiopod Eocoelia Nikiforova, 1961 (in Nikiforova and Andreeva 1961) has been of considerable importance in studies of Silurian benthos. First, the genus is the eponymous component of the widespread Eocoelia Community which occupied nearshore environments during the late Llandovery (Ziegler 1965), and secondly the well-documented Eocoelia lineage has been effectively used in biostratigraphical correlation within lower Silurian shelly facies (Ziegler 1966). Detailed biometrical analysis of Eocoelia from the west of Ireland and south-west Scotland has permitted a significant refinement of the existing Llandovery part of the Eocoelia lineage which has some bearing on correlation at and near the base of the Telychian. The analysis, however, confirms some problems in the application of conventional Linnean nomenclature in such gradualist lineages (e.g. Sheldon 1987). DISTRIBUTION OF EOCOELIA IN IRELAND AND SCOTLAND Eocoelia is relatively widespread throughout the lower Silurian of the Anglo-Welsh area but its distribution is comparatively more localized across Ireland and Scotland. The occurrences in the west of Ireland and Girvan are the only records of the genus from the Midland Valley of Scotland and its Irish equivalent (Text-fig. 1). Two species of Eocoelia have been recorded from the West of Ireland. E. curtisi Ziegler dominates shell beds within the lower part of the Kilbride Formation along the Silurian outcrop of North Connemara and on the Kilbride Peninsula (Piper 1972); E. angelini occurs in the lower Wenlock Lough Muck Formation (E. sulcata in Laird and McKerrow 1970). Large new collections of E. curtisi have been made from three localities within the lower part of the Kilbride Formation along the Llandovery outcrop in north Connemara as follows: 11, Lough Fee (IGR L 609790); 12, Lettershanbally (IGR L 584836); 13, Lee (IGR L 570 885) (see also Doyle 1989). In the Girvan district, SW Scotland, Eocoelia has long been known from the Lower Camregan Grits of the Main Silurian Outcrop, south of the Girvan Valley, in Penwhapple Burn and adjacent areas (Davidson 1867). A.N.H. has made substantial new collections from three localities within the Lower Camregan Grits in the Penwhapple Burn area as follows: SI, (NGR NX 2271 9807); S2, (NGR NX 2230 9799); and S3, (NGR NX 2254 9805); and a new occurrence of the genus is recorded from the Craighead inlier where it occurs with Pentameroides. | Palaeontology, Vol. 34, Part 2, 1991, pp. 439^454.| © The Palaeontological Association 440 PALAEONTOLOGY. VOLUME 34 text-fig. 1. Location of upper Llandovery Eocoelia in the Midland Valley of Scotland and its Irish equivalent. Abbreviations: ci, Clare Island; cp, Croagh Patrick; ng. North Galway; ch, Charlestown; 1, Lisbellaw; p, Pomeroy; c, Craighead inlier; g, Main Outcrop, Girvan; le, Lesmahagow; h, Hagshaw Hills; pt, Pentland Hills; g.g.f.. Great Glen fault; h.b.f.. Highland Boundary fault; s.u.f., Southern Upland fault; i.s., putative track of Iapetus suture; M.V., Midland Valley. Occurrence of Eocoelia indicated by asterisk. MORPHOLOGICAL ANALYSIS Large samples of Eocoelia from Girvan (N = 342) and Connemara (N = 419) together with a more limited sample of topotype material of E. curtisi from the Tortworth inlier were analysed with reference to a set of continuous variates defined below and illustrated on Text-figure 3a-c. The ribbing patterns of all three samples were investigated with the aid of frequency histograms and non-parametric inferential statistics. All graphical and statistical analyses were processed by the PALSTAT package (Harper and Ryan 1987) implemented on a BBC B microcomputer. Measurements taken (in mm) were si, sagittal length; mw, maximum width; pm, position of maximum width measured from posterior margin; pt, position of maximum depth measured from posterior margin; pd, position of deflection measured along sagittal length; nr, total number of ribs; lc, total length of crural fossettes; me, maximum separation of crural fossettes; sc, sagittal length of crural fossettes; mt, maximum separation of teeth; mb. maximum separation of distal ends of the brachiophores; bl, maximum length of brachiophores ; In, length of notothyrial platform. Matrices of sample sizes are shown in Table 1. Pooled samples both of the brachial and pedicle valve exteriors and interiors from Connemara, Girvan and Tortworth were investigated for size-independent variation within and between samples using the multivariate technique of Principal Component Analysis (PCA); the relationships between the continuous variates, defined above, are described by a correlation matrix from which the appropriate eigenvalues and eigenvectors have been extracted. The rib counts, defined as the total number of costae, for all three samples are displayed as frequency polygons (Text-figs 4 and 5) compared using the non-parametric Kolmogorov-Smirnov test, whilst the rib strength (height/ width ratio calculated as a percentage) was similarly investigated by histograms together with parametric and non-parametric inferential statistics (Text-figs 6 and 7). Three features of shell morphology yielded taxonomically significant results: (i) size-independent shape variation between samples of valve exteriors, (ii) the total number of ribs, and (iii) the strength of ribs. DOYLE ET A L. : UPPER LLANDOVERY BRACHIOPOD 441 text-fig. 2. Locality details and stratigraphies for the Eocoelia- bearing horizons sampled in the West of Ireland (a) and Girvan, SW Scotland (b). 442 PALAEONTOLOGY. VOLUME 34 text-fig. 3. Location of measurements made on the exteriors (a) and ventral (b) and dorsal (c) interiors of Eocoelia. Abbreviations and definition of measurements given in text. table 1. Matrices of sample sizes. Abbreviations: PVE, pedicle valve exterior; PVI, pedicle valve interior; BYE, brachial valve exterior; BVI, brachial valve interior. Girvan Connemara Locality Valve SI S2 S3 Total 11 12 13 Total PVE 47 37 33 117 33 35 33 101 PVI 15 1 1 18 44 32 32 32 96 BVE 33 35 35 102 38 40 33 111 BVI 25 19 35 79 33 45 33 111 Pooled samples of the Connemara, Girvan and Tortworth specimens were investigated by PCA: both pedicle and brachial valve exteriors and interiors were analysed with reference to the variates defined above. The investigation of comparative internal morphology, with reference to the following variates - si, mw, mb, bl and In for brachial valves and si, mw, lc, me, sc and mt for pedicle valves - yielded no apparent differences between the three samples when each specimen was plotted relative to the second and subsequent (size-independent) eigenvectors. The Irish specimens have markedly larger scores on the first eigenvector, confirming their relatively larger size. However, multivariate examination of the valve exteriors based on the variates si, mw and pm suggests the samples may also be differentiated with reference to their scores on the second eigenvector (direction cosines: —0 278, —0423 and 0 862) of this analysis; the Irish specimens had significantly lower scores on this eigenvector indicating an outline with a maximum width, on average, nearer the posterior margin (Text-fig. 8). Significant differences in the rib counts were detected between the Girvan material and the specimens from both Connemara and Tortworth (Text-fig. 5). The Tortworth sample appears to DOYLE ET AL.\ UPPER LLANDOVERY BRACHIOPOD 443 text-fig. 4. Frequency polygons of total rib numbers on valves of E. curtisi from Connemara, Tortworth and Girvan. TORTWORTH CONNEMARA text-fig. 5. Comparison of the cumulative frequency polygons of the total rib numbers on valves of E. curtisi from Connemara, Tortworth and Girvan. plot between the Irish and Scottish samples on the frequency polygon (Text-fig. 4) and significant differences were detected using the Kolmogorov-Smirnov test (at 1 % level) between it and the material from Connemara (D = 0 718) and Girvan ( D = 0-620). Although Ziegler (1966, p. 530) considered the modal rib density did not appear to behave consistently with time, despite the small samples in many collections, there is in fact a decrease in the number of ribs with time along this part of the lineage: a trend true, in general terms, for the lineage as a whole. Moreover, Ziegler’s claim that the stratigraphically older E. hemisphaerica (reported modes of 14 and 16) has fewer ribs than E. intermedia (reported modes of 16 and 18) is not supported by the data in his table 3. Larger samples and counts of discrete rather than grouped rib numbers may help tighten this putative 444 PALAEONTOLOGY, VOLUME 34 text-fig. 6. Frequency polygons of rib-strength indices for brachial valves of E. curtisi from Girvan, Tortworth and Connemara. text-fig. 7. Cumulative frequency polygons of rib-strength indices for brachial valves of E. curtisi from Girvan, Tortworth and Connemara. DOYLE ET A L.: UPPER LLANDOVERY BRACHIOPOD 445 SCORES ON SECOND EIGENVECTOR text-fig. 8. Comparison of the scores on the second eigenvector (direction cosines: —0-278, —0-423, 0-862) for a PCA of variates si, mw and pw for brachial valve exteriors of the Irish and Scottish Eocoelia. The Connemara specimens, E. c. curtisi have significantly smaller scores on this eigenvector (D > 0-23 at 1 % level - Kolmogorov-Smirnov test). trend. Sheldon (1987) has shown that evolutionary reversals are possible in otherwise unidirectional evolutionary trends, so it is conceivable that the overall trend of loss of ribs in the Eocoelia lineage may be influenced by periods of no loss or possible rib gain. The rib strength of the taxa from Connemara, Girvan and Tortworth also displayed significant contrasts (Text-figs 6 and 7). Clearly, in view of the probability of some abrasion of the ribs during postmortem transport and modification with compaction, diagenesis and subsequent dissolution, this feature must be treated with some caution. Nevertheless, the clear decrease in rib strength with time is confirmed within the area of the lineage investigated here. Data from the Tortworth (Ziegler 1966, table 6), Connemara and Girvan specimens were compared statistically by F and t tests and the rib strengths of all three samples were compared with those of the stratigraphically older E. intermedia. But although the direction of the trend is confirmed, the timing of events within the trend would appear to be slightly retarded. The Irish Eocoelia ribs are significantly stronger than those of the nominate subspecies from Tortworth, having rib strengths similar to those of E. intermedia from Norbury, whereas the Girvan specimens have rib strengths similar to those of the stratigraphically younger E. intermedia from May Hill. Many of the significant differences may be artefacts of sample comparisons of discrete and spatially isolated segments of a gradualist lineage. Further interpolation within the lineage will clearly strain the existing Linnean framework established for Eocoelia and will require a complete revision of its taxonomy or the recognition of categories of lesser rank than the subspecies (see also Sheldon 1987). 446 PALAEONTOLOGY, VOLUME 34 REFINEMENT OF THE EOCOELIA LINEAGE The Eocoelia lineage, first established in detail by Ziegler (1966) and later modified by Cocks (1971), has more recently been summarized by Bassett (1984) in its entirety. Cocks et al. (1984) have, in revising the type Llandovery Series, documented the lower part of the lineage, summarized on Text- figure 9. Ziegler (1966) identified a number of clear morphological trends during the phytogeny of Eocoelia : (i) strengthening of the articulating mechanisms with a trend towards deeper fossettes, stronger hinge plates and more robust teeth; (ii) reduction in the development of lips and deflections; and (iii) decline in rib strength. All three samples investigated have well-developed crural fossettes, strong hinge plates and robust teeth, but umbonal chambers are lacking. Although these features are difficult to compare quantitatively, the samples from Girvan, Tortworth and the West of Ireland are consistent with those of E. curtisi Ziegler, 1966, confirming their inclusion in that species. Fourteen percent (N = 180) of pedicle and brachial valves from the West of Ireland possessed a deflection whereas only 0-01 % (N = 220) of pedicle and brachial valves from Girvan possessed the same feature. Clearly this interpolated trend is contrary to that seen in the lineage as a whole where there is an increase in the development of the deflection and lip. Although Ziegler (1966), in establishing a trend in this aspect of the Eocoelia shell, implied that the development of such features is not related to the size of the individual, deflections are in fact most commonly recorded from larger specimens. In the samples investigated by Ziegler ( 1966), the stratigraphically older species are LU 1 Graptolite Zones LITHOSTRATIGRAPHY Llandovery Connemara Girvan Tortworth Evolutionary Brachiopod Lineages Eocoelia Stricklandud Pentamerid O O > cc LU B Q Z < Centrifugus Crenulata Cwernfelen Fm. Lettergesh Fm. Gowlaun Mbr. Tonalee Fm. Griestoniensis Crispus Turriculatus Kilbride Fm. Sedgwickii Convolutus Cerig Fm. Wormwood Fm. Rhydings Fm. Knockgardner Blair Shale Tortworth Beds Drumyork Flags Up. Trap Damery Beds Lauchlan Fm. Proto. Grits Penkill Fm. U.C.G. Max. M.st. f'/ood Burn Fm L.C.G. Pencleuch Shale IT ® i E i £ ' 4- UJ + „t It text-fig. 9. Lithostratigraphy of the Connemara, Girvan and Tortworth successions together with brachiopod lineages of correlative value displayed relative to the current graptolite biostratigraphy and chronostratigraphy for relevant parts of the lower Silurian. Abbreviations: L.C.G., Lower Camregan Grits; Max. Mst., Maxwellston Mudstones; U.C.G., Upper Camregan Grits; Proto., Protovirgularia, Up., Upper. DOYLE ET AL.: UPPER LLANDOVERY BRACHIOPOD 447 generally larger than those occurring high in the lineage. Ziegler (1966, table 7) reported a total absence of deflections on the shells of E. curtisi. The Irish samples are however larger (mean widths of 110 and 114mm for dorsal and ventral exteriors, respectively) than the largest mean width (6 68 mm) reported in Ziegler (1966, table 5). Within the Eocoelia lineage (Ziegler 1966) the transition between E. intermedia and E. curtisi is taken to occur in the lowest part of the turriculatus Biozone at a level correlated with the Aeronian/Telychian junction (Cocks et al. 1984). Therefore implicit in the definition of this boundary is the coincidence of the base of the turriculatus Biozone with the base of the Telychian. In the type area of the Llandovery Series diagnostic graptolite and shelly fossils are rare or absent at and adjacent to the Aeronian/Telychian stratotype boundary. Graptolites are, in fact, absent within the basal part of the Telychian in the type area, although faunas considered diagnostic of the lower half of the turriculatus Biozone (Cocks et at. 1984, p. 168) are reported 10 km from the stratotype section (Temple 1988, p. 879). Ascending the type section, E. curtisi is first encountered in the lower part of the Cerig Formation, 35 m above the basal Telychian stratotype (Cocks et al. 1984, fig. 67), although it is assumed to occur throughout the lower part of the formation (Cocks et al. 1984, fig. 69), and thus within the lower part of the turriculatus Biozone. Cocks et al. (1984, p. 168) considered that the distribution of the relevant Eocoelia species and graptolites in the Penwhapple Burn section, near Girvan (Cocks and Toghill 1973) confirmed the coincidence of the base of the Telychian with the base of the turriculatus Biozone. However, at Girvan specimens hitherto assigned to E. curtisi (see Cocks and Toghill 1973) and assigned here to E. curtisi immatura subsp. nov. occur with Pentamerus oblongus in the Lower Camregan Grits. Although Cocks and Toghill (1973) do not record an in situ graptolite fauna from the overlying Wood Burn Formation, some slabs in the Gray Collection (British Museum of Natural History) from the Penkill locality suggest a correlation with the upper sedgwickii Biozone (Cocks and Toghill 1973, p. 226). The succeeding Maxwellston Mudstones contain a lower turriculatus Biozone fauna (Cocks and Toghill 1973, p. 227). Thus, rather than suggesting E. curtisi occurs with turriculatus Biozone graptolites, the Girvan section indicates a co- occurrence with graptolites of the sedgwickii Biozone. Clearly, the lack of graptolite control across the Aeronian/Telychian boundary stratotype invites correlation of the upper part of the Wormwood Formation with either the lower turriculatus or upper sedgwickii biozones. However, in the absence of more equivocal faunal control in the type area, the faunal data from Girvan suggest the base of the Telychian may be better correlated with a horizon within the upper part of the notional sedgwickii Biozone. Moreover there is a gap of some 65 nr in the Eocoelia lineage between the last occurrence of E. intermedia in the upper part of the Wormwood Formation and the first occurrence of E. curtisi in the Cerig Formation. If the lineage is as cosmopolitan as previous and current documentation suggests, then forms similar to E. c. immatura , together with graptolites of the sedgwickii Biozone, might be expected within this faunal hiatus. North of the Girvan valley, in the Craighead inlier Cocks and Toghill (1973) have documented the co-occurrence of E. curtisi (E. c. immatura herein) and Pentamerus oblongus within the Lower Camregan Grits. Extension of the pit in Craigfin Wood (Cocks and Toghill 1 973, p. 217) has yielded a new fauna in higher strata: a species of Pentameroides occurs with a number of poorly preserved specimens assigned, on the basis of their outlines and rib numbers, to E. cf. curtisi curtisi. Poor exposure and complex faulting in this part of the inlier presents considerable stratigraphical diflficulties; nevertheless, the co-occurrence of Pentameroides and a form approximating to the nominate subspecies of E. curtisi supports the correlations presented in Text-figure 9. In the west of Ireland, E. curtisi curtisi occurs with Costistricklandia lirata and Pentameroides within the Kilbride Formation, presumably near the top of its range. Although no diagnostic fossils are present in the overlying Tonalee Formation (Doyle et al. 1990) the succeeding Benbeg Mudstones contain crenulata Biozone graptolites (Rickards 1973). 448 PALAEONTOLOGY. VOLUME 34 SYSTEMATIC PALAEONTOLOGY Family trigonirhynchiidae McLaren, 1965 Genus eocoelia Nikiforova, 1961 (in Nikiforova and Andreeva 1961) Remarks. Cocks (1978, p. 149) transferred Eocoelia from its traditional site within the Atrypida, on the basis of an undescribed species of Eocoelia , from the Idwian (Aeronian) of Shropshire, with similarities to Rostricellula. Eocoelia curtisi curtisi Ziegler, 1966 Text-fig. 10 1867 Atrypa? hemisphaerica J. de C. Sowerby; Davidson, p. 136 (pars), pi. 13, figs 24-30a, non fig. 23.’ 1966 Eocoelia curtisi Ziegler, p. 537 (pars), pi. 83, figs 7 and 8; pi. 84, figs 12-17. Holotype. OUM C3241 ; an internal mould of a pedicle valve from the Damery Beds (Telychian) of the Tortworth Inlier, Gloucestershire; pi. 84, figs 15, 16, 17 of Ziegler 1966. Material. About 200 pedicle valves and 200 brachial valves, all virtually complete, none conjoined. Diagnosis. Nominate subspecies of E. curtisi with 7-16 (mode 1 1 - Connemara or 12 - Tortworth) strong ribs developed on brachial valve exteriors; maximum width posterior to midvalve length and rib strengths about one-quarter. Description Exterior. Medium-sized, planoconvex valves of transversely subquadrate to subelliptical outline with maximum width posterior to mid-valve length. Hinge line about four-fifths maximum width, cardinal extremities obtuse and rounded. Pedicle valve about three-quarters as long as wide and about one-quarter as deep as long; anterior and lateral profiles convex medianly but in later growth stages growth vectors change to produce anterolateral flattening. Maximum depth occurs between one-fifth and two-fifths valve length. Brachial valve about three-quarters as long as wide, essentially flat with faint median sulcus and flatly convex flanks. Deflection of valve profile present on 14% of pedicle and brachial valves ( N = 180). Ornament of strong costae of evenly rounded, semicircular profile, numbering 7-15 on 1,3,2,20,34,29,12,7,2 valves and with mean (variance) rib strength (height/width * 100) of 26-4 (43-5) for 40 brachial valves; costae subdued or absent posterolaterally. Concentric growth lines absent posteriorly but accentuated anteriorly. Ventral interior. Delthyrial chamber moderately deep with faint pedicle callist rarely developed in posterior half. Large cyrtomatodont teeth, oval to triangular in dorsal view with rounded anterior surfaces. Dental plates absent; teeth attached directly to shell wall. Large fossettes cut deeply into medial face of teeth extending into shell wall. Muscle scars not impressed. Dorsal interior. Socket plates large, well developed, almost rectangular in ventral view, diverging at 55-75 degrees and supported posteriorly on broad, raised notothyrial platform ; cardinal process very rarely ( < 1 %) present. Sockets deep, conical and widely divergent. Median ridge arising anterior to notothyrial platform and extending to about one-half valve length. Muscle scars feebly impressed. Measurements and statistics Brachial valve exteriors Variates si mw pm Sample size 110 111 no Means 7-71 11-0 2-46 Variance-covariance matrix 2-29 2-87 0-58 4-92 0-74 0-39 DOYLE ET A L.: UPPER LLANDOVERY BRACHIOPOD 449 J K L text-fig. 10. Eocoelia curtisi curtisi Ziegler, from the lower part of the Kilbride Formation, north Connemara. a, external mould of brachial valve, JMM BrlOOO, x 3. b, external mould of brachial valve, JMM BrlOOl, x 3. c, latex cast of pedicle valve exterior, JMM Brl002, x 2. d, latex cast of pedicle valve exterior, JMM Brl003, x 3. E, latex cast of brachial valve interior, JMM Brl004, x 3. f, internal mould of pedicle valve, JMM Brl005, x 2. G, internal mould of pedicle valve, JMM Brl006, x 2. h, internal mould of pedicle valve, JMM Brl007, x 3. i, internal mould of brachial valve, JMM Brl008, x 3. J, latex cast of pedicle valve interior, JMM Brl009, x 3. k, latex cast of pedicle valve interior, JMM Br 1010, x 3. L, internal mould of brachial valve, JMM BrlOl 1, x 3. 450 PALAEONTOLOGY. VOLUME 34 Brachial valve interiors Variates si mw mb bl In Sample size 104 105 1 1 1 111 111 Means 8-00 10-8 1 90 1-24 0-82 Variance-covariance matrix 1-76 2-47 0-43 0-23 013 4-98 0-81 0-40 019 0-20 009 0-04 0-08 004 0-03 Pedicle valve exteriors Variates si mw pm Pt Sample size 98 100 100 100 Means 8-06 1 14 2-82 2-80 Variance-covariance matrix 2-06 2-17 0-61 0-57 3-95 0-66 0-53 0-32 0-23 0-36 Pedicle valve interiors Variates si mw lc me sc mt Sample size 96 96 96 96 96 96 Means 8-21 11-2 1-23 2-01 1 09 2-42 Variance-covariance matrix 2-07 2-37 0-18 0-35 0-21 0-44 4-02 0-25 0-54 0-26 0-65 006 0-06 004 0-07 015 0-06 0 16 0-05 007 0-21 Remarks. The description of E. curtisi curtisi, presented here, is based exclusively on material from the lower part of the Kilbride Formation, which crops out along the northern margin of Connemara. The Irish specimens are considered morphologically identical to the type and topotype material of the nominate subspecies from Tortworth except for the development of the ribs. The ribbing strengths of the various E. curtisi morphs have been discussed above. However, analysis of the rib counts of the Tortworth and Connemara specimens presents taxonomic difficulties. A Kolmogorov-Smirnov test, as noted above, indicates a significant difference between the two frequency distributions. The Irish material has a modal value of 11 ribs, that from Tortworth has a mode of 12. However, the sample from Tortworth is disproportionately smaller than that from Ireland. Moreover it is probable that the Connemara specimens are from slightly younger horizons than those from Tortworth, thus confirming the trend of decreasing rib number with decreasing time. Since the Irish E. curtisi agrees in all other aspects with the nominate subspecies it is not separated on the basis only of the modal rib counts. However, it may be suggested the two represent chronological morphs of the same subspecies which a more rigorous investigation of more material may confirm or reject. Eocoelia curtisi Ziegler, 1966 immatura subsp. nov. Text-fig. 1 1 1867 Atrypa? hemisphaerica J. de C. Sowerby; Davidson, p. 136 (pars), pi. 13, figs 25, 27-30. 1973 Eocoelia curtisi Ziegler; Cocks and Toghill, p. 225, pi. 3, figs 1-3. Name. Latin immature a youthful morphological characteristics. DOYLE ET AL. : UPPER LLANDOVERY BRACHIOPOD 451 M N O P text-fig. 1 1. Eocoelia curtisi Ziegler immatura subsp. nov. from the Lower Camregan Grits, Penwhapple Burn, Girvan. a, external mould of brachial valve, JMM Br 1012, x 3. B, external mould of brachial valve, JMM Br 1013, x 2. c, external mould of brachial valve, JMM Brl014, x 2. d, external mould of brachial valve, JMM Br 1015, x3. e, latex cast of pedicle valve exterior, JMM Br 1016, x 2. F, latex cast of pedicle valve exterior, JMM Br 1017, x 3. G, latex cast of brachial valve interior, JMM Br 1018, x 3. h, latex cast of brachial valve interior, JMM Brl019, x 2. i, internal mould of pedicle valve, JMM Brl020, x 2. j, internal mould of pedicle valve, JMM Brl021, x 2. k, internal mould of brachial valve, JMM Brl022, x 2. l, internal mould of brachial valve, JMM Brl023, x 2. m, latex cast of pedicle valve interior, JMM Brl024, x 2. n, latex cast of pedicle valve interior, JMM Brl025, x 2. o. External mould of pedicle valve, JMM Brl026, x 2. p, external mould of pedicle valve, JMM Brl027, x3. All type and figured specimens are deposited in the James Mitchell Museum, University College Galway, Ireland. Holotype. JMM Brl020; an internal mould of a pedicle valve from the Lower Camregan Grits, Penwhapple Burn, Girvan, SW Scotland. Material. About 150 pedicle valves and about 180 brachial valves, all virtually complete, none conjoined. Diagnosis. Small subspecies of E. curtisi with 13- 19 (mode 15) strong ribs developed on brachial valve exteriors; maximum width at or near midvalve length and rib strength of about one-third. 452 PALAEONTOLOGY, VOLUME 34 Description Exterior. Small, planoconvex valves of subquadrate outline with maximum width at or near mid-valve length. Hinge line about three-quarters maximum width with obtusely rounded cardinal extremities. Pedicle valve about four-fifths as long as wide and about one-quarter as deep as long with maximum depth at about one- third valve length. Delthyrium relatively wide and open. Brachial valve about four-fifths as long as wide and essentially flat. Ventral and dorsal interareas obsolete. Ornament of relatively strong ribs of evenly rounded profile developed over entire valve surface and numbering 1 3-1 9 on 1 1 , 34, 40, 1 3, 4, 0, 1 valves with rib strength of mean (variance) 35-8 (47-4) for 38 brachial valves. Ventral interior. Relatively deep delthyrial chamber flanked by large cyrtomatodont teeth, oval in dorsal view with rounded anterior surfaces; dental plates absent. Dental fossettes cut deeply into medial face of teeth. Muscle scars not impressed. Dorsal interior. Large, robust socket plates, elongately rectangular in ventral view and distal parts anteriorly divergent on low notothyrial platform. Deep, divergent conical sockets. Broad median ridge extending anteriorly from margin of notothyrial platform. Muscle scars not impressed. Measurements and statistics Brachial valve exteriors Variates si mw pm Sample size 103 103 103 Means 5-61 6-90 2-30 Variance-covariance matrix 0-73 0-80 0-24 114 0-32 014 Brachial valve interiors Variates si mw mb nl In Sample size 79 79 79 79 79 Means 5-84 7-08 1-44 114 0-54 Variance-covariance matrix 0-82 100 0 12 010 0-05 1-49 01 9 015 0-08 0-08 004 001 004 001 0-01 Pedicle valve exteriors Variates si mw pm pd Pt Sample size 117 117 1 17 117 117 Means 5-35 6-47 2-34 1 62 13-2 Variance-covariance matrix 219 1 94 0-72 0-52 0-39 2-86 0-91 0-69 0-66 0-43 0-27 018 0-28 010 1 53 Pedicle valve interiors Variates si mw lc me sc Sample size 44 44 44 44 44 Means 5-99 6-80 0-66 1-33 0-55 mt 44 1-55 DOYLE ET A L.: UPPER LLANDOVERY BRACHIOPOD 453 Variance-covariance matrix 0-89 0-7 1 061 013 010 003 012 01 1 002 003 009 0-07 002 0-01 002 014 013 002 0-03 0-01 0-04 Remarks. The Scottish material, hitherto referred to E. curtisi by Cocks and Toghill (1973), differs in two main respects from the nominate subspecies. First the maximum width is at or near the mid- valve length and secondly it has more and stronger ribs. Taken together, the morphological contrasts may be interpreted as specific differences; however, the Girvan, Connemara and Tortworth samples are characterized by well-developed crural fossettes, strong hinge plates and robust teeth; umbonal chambers are absent. This association of characteristics conventionally describes E. curtisi ; the differences, therefore, are accorded only subspecific status. Acknowledgements . We thank A. Davis for assistance, P. Powell for access to material in the Oxford University Museum and W. S. McKerrow for many wide-ranging discussions. Doyle and Hoey were financed by Postgraduate Fellowships at University College Galway and Harper is grateful to the Royal Irish Academy for help with field expenses. bassett, m. G. 1984. Lower Palaeozoic Wales -a review of studies in the past 25 years. Proceedings of the Geologists' Association , 95, 291-311. cocks, l. R. m. 1971. Facies relationships in the European Lower Silurian. Memoires du Bureau de Recherches Geologicjues et Minieres, 73, 223-227. — 1978. A review of British Lower Palaeozoic brachiopods, including a synoptic revision of Davidson’s monograph. Monograph of the Palaeontographical Society , 131 (549), 1-256. — and toghill, p. 1973. The biostratigraphy of the Silurian rocks of the Girvan district, Scotland. Journal of the Geological Society of London , 129, 209-243, pis 1-3. woodcock, n. h., rickards, R. b., temple, j. t. and lane, p. d. 1984. The Llandovery Series of the type area. Bulletin of the British Museum ( Natural History), (Geology), 38, 131-182. davidson, t. 1867. A monograph of the British fossil Brachiopoda. Part VII. No. II. The Silurian Brachiopoda. Monograph of the Palaeontographical Society, 3, 89-168, pis 13-22. doyle, e. n. 1989. The biostratigraphy and sedimentology of the Lower Silurian (Llandovery) rocks of north Galway. Unpublished Ph.D. thesis. National University of Ireland. — harper, d. a. t. and parkes, m. a. 1990. The Tonalee fauna: a deep-water shelly assemblage from the Llandovery rocks of the West of Ireland. Irish Journal of Earth Sciences, 11, 127-143. harper, d. a. t. and ryan, p. d. 1987. PALSTAT - a statistical package for palaeontologists. Palaeontological Association and Lochee Publications, Dundee, Scotland. laird, M. G. and mckerrow, w. s. 1970. The Wenlock sediments of northwest Galway, Ireland. Geological Magazine , 107, 297-317. mclaren, d. j. 1965. Family Trigonirhynchiidae McLaren, n. fam. H559-H562. In moore, r. c. (ed. ). Treatise on invertebrate paleontology. Part H. Brachiopoda. Geological Society of America and Kansas University Press, Boulder, Colorado and Lawrence, Kansas, 927 pp. Nikiforova, o. i. and andreeva, o. n. 1961. Stratigraphy of the Ordovician and Silurian of the Siberian Platform and its palaeontological basis (Brachiopods). Biostratigraphiya Paleozoya Sibirskov Platformy, Leningrad, 1, 1-412, pis 1-56. piper, d. J. w. 1972. Sedimentary environments and palaeogeography of the late Llandovery and earliest Wenlock of north Connemara. Quarterly Journal of the Geological Society of London, 128, 33-51. rickards, R. b. 1973. On some highest Llandovery red beds and graptolite assemblages in Britain and Eire. Geological Magazine, 110, 70-72. sheldon, p. r. 1987. Parallel gradualistic evolution of Ordovician trilobites. Nature, 330, 561-563. temple, J. T. 1988. Biostratigraphical correlation and the stages of the Llandovery. Journal of the Geological Society of London, 145, 875-879. REFERENCES 454 PALAEONTOLOGY. VOLUME 34 ziegler, A. m. 1965. Silurian marine communities and their environmental significance. Nature , 207, 270-272. - 1966. The Silurian brachiopod Eocoelia hemisphaerica (J. de C. Sowerby) and related species. Palaeontology , 9. 523-543, pis 83 and 84. E. N. DOYLE1, A. N. HOEY2 and D. A. T. HARPER Department of Geology University College Galway, Ireland Present addresses: 1 Department of Geology University of the West Indies Mona, Kingston 7, Jamaica Typescript received 2 January 1990 Revised typescript received 7 March 1990 2 Department of Geology University College Belfield, Dublin, Ireland THE ROLE OF PREDATION IN THE EVOLUTION OF CEMENTATION IN BIVALVES by ELIZABETH M. HARPER Abstract. The independent appearance of many taxa of cementing bivalves during the early Mesozoic coincided with the marked increase in predation pressure described by Vermeij (1977, 1987). A causal link is implied by experimental work in which predators were offered the choice of byssate or cemented bivalve prey: cementation confers a significant selective advantage by inhibiting manipulability. The example illustrates the potential value to palaeontology of studies in behavioural ecology. Epifaunal bivalves attach to the substratum by two means: cementation by one valve or, more commonly, anchorage by byssal threads produced by the foot. Yonge (1962) believed that most, if not all, living bivalves possess a byssus in the larval stage, and that this structure was retained in some adults, for example the Mytilacea and the Arcacea, by neoteny. It would seem that the cemented habit in bivalves was evolved in stocks already possessing a functional adult byssus; indeed most living cementing bivalves, e.g. the Spondylidae and Hinnites , pass through a byssate stage in early ontogeny. EXPERIMENTAL WORK A series of experiments was designed to establish the relative vulnerability to predation of byssate and cemented bivalve prey. Asteroid and crustacean predators were offered the choice of bivalves attached both bysally and by cementation. Mytilus edulis was used for both prey types, so that any preference expressed would be due to mode of attachment only, rather than on the basis of different nutritional quality. Mussels with established byssal threads were collected intertidally in Dunstaffnage Bay, Oban, and cementation was simulated using an epoxy resin (Araldite Rapid - Ciba Geigy) to fix the shell by one valve to large blocks of substratum. These ‘cemented’ Mvti/us fed normally and even produced superfluous byssus threads and hence behaved identically to the byssate individuals. Many byssate individuals were daubed with epoxy in order to monitor any inhibitory effect on predator behaviour (e.g. masking metabolite cues from the prey): no such effect was apparent. Treated and untreated specimens were eaten in equal proportions. The experiments were run in outdoor running seawater tanks (1-5 x 0-8 m), each set up with a random distribution of the byssate and ‘cemented’ prey. A number of individuals of Asterias ruhens , Cancer pagur us or Carcinus maenas were introduced into each tank, having previously been starved for at least four days. Regular observations were made on the feeding behaviour of the predators and any prey item taken was replaced with an identically attached individual. Hence the relative numbers of prey types were held constant. RESULTS If cemented and byssate prey were indistinguishable to predators, one might expect that they would be eaten in the proportions in which they occur in the tank (the null hypothesis). The results were in fact very different: a much higher proportion of prey taken was byssate (see Text-fig. 1). Chi- squared one-sample analysis of these results reveals that the preference for byssate prey over cemented was highly significant, rejecting the null hypothesis for Asterias and Cancer (P • * 0 » • • r * . • • -10 1 -10 1 PC 1 PC 1 text-fig. 3. Scatterplots of the 80% of the specimens for each group lying closest to the group centroid. Only groups with five or more specimens are shown. Axes as in Text-figure 1. Key: Lower Cambrian: A, Corynexochida ; □ , Eodiscina; Olenellina; O. Ptychopariacea ; Middle Cambrian: A, Corynexochida ; □, Marjumiacea; O, Ptychopariacea; 0. Solenopleuracea ; Upper Cambrian: B. Anomocaracea ; i . Illaenuracea ; A- Komaspidacea; □, Marjumiacea; Raymondinacea ; A, Proetida; O. Ptychopariacea; 0. Solenopleuracea; Ordovician 1 : i , Asaphacea; M Cheirurina; □, Conocoryphacea ; >K Cyclopygacea ; A- Komaspidacea ; 0. Olenacea; A- Proetida; O. Scutelluina ; Ordovician 2: Cheirurina ; □, Odontopleurida ; A. Proetida; Remopleuridacea ; O. Scutelluina; A. Trinucleacea; Ordovician 3: i , Asaphacea; other symbols as for Ordovician 2. 470 PALAEONTOLOGY, VOLUME 34 in dispersion evident within higher taxa as well? Do higher taxa represent morphotypes, as was implicitly assumed above? The dispersion within and among higher taxa is depicted graphically in Text-figure 3. Here, only the 80% of the specimens lying closest to the morphologic centroid (in principal-component space) for each group are presented. As above, the purpose of this culling procedure is to remove the visual effect of extreme specimens. To keep the graphs simple, only groups with sample sizes of at least five are plotted. Two patterns are evident here: I There is no obvious tendency for within-group dispersion to increase through time. (Note that the scatterplots for different intervals are drawn at different scales.) At all times there are groups encompassing a large range of morphology, as well as morphologically more restricted taxa. This is true even though some of the higher taxa are at the level of the order. 2. The separation among groups clearly increases through time. This pattern is most striking when the Cambrian as a whole is compared to the Ordovician as a whole, but the trend is also evident within the Ordovician. Cambrian trilobites are difficult to partition into suprageneric groups that correspond to well-defined morphotypes, while at least some Ordovician taxa correspond to morphologically well defined units. This is in accord with previous observations (e.g. Rasetti 1954, 1961; Palmer 1958; Whittington 1966). It is likely that if more dimensions (i.e. morphologic variables) were added to this analysis, the Cambrian groups would become easier to discriminate. However, the fact that discrimination has historically been relatively difficult suggests that the difference between the Cambrian and the Ordovician is real. Dispersion within groups shows no obvious trend, while dispersion among groups increases. This suggests that the overall morphologic diversification among the trilobites is tied to patterns at higher taxonomic levels. This is not meant to imply that there are superfamily-level evolutionary processes that differ fundamentally from evolutionary mechanisms within populations. Quantitative analysis of higher taxa The patterns depicted in two dimensions appear striking, but should be quantified in the 12- dimensional space. I emphasize that all subsequent analyses in this paper are based on the complete , 12-dimensional Fourier space , not the principal-component space. This quantification requires the use of multivariate measures of dispersion. There has been much discussion about how to measure morphologic dissimilarity (e.g. Van Valen 1974; Ashton and Rowell 1975; Atchley et al. 1982; Cherry et al. 1982). In principle, variances (e.g. Pearson 1926) and covariances (e.g. Atchley et al. 1982) should be taken into account when describing morphologic distances among groups. In practice, however, it has been found that simple distance measures that do not consider variances and covariances are more reliably estimated (Atchley et al. 1982; Cherry et al. 1982). Atchley et al. (1982) point out that simple distance measures may be more precise (i.e. more reliably estimated) but may be further from the morphologic ‘truth’. For purposes of this study, it is more important that distance measures be reliable so that they can be compared among taxa and among times. Therefore, simple Euclidean distance is used here as a measure of morphologic dissimilarity. If there are p variables, then the Euclidean distance between two specimens is given by d12 = (Xn-Xj2)* Lj-l (1) where Xn and Xj2 are the values of variable j on specimens 1 and 2. Three dispersion indices were defined for the 12-dimensional Fourier space. W is the weighted mean of all within-group distances, and gives a measure of the morphologic variability within higher taxa. (Methods of weighting are discussed below.) A is the weighted mean of the distances among group centroids, and provides a measure of the morphologic variability among higher taxa. (The group centroid is an imaginary point representing the average morphology of the group, i.e., the arithmetic average for each of the variables measured on all specimens within a group.) Intuitively, it seems that the less dispersion there is within taxa and the greater the distance among taxa, the FOOTE: TRILOBITE DIVERSIFICATION 471 better defined or more distinct those taxa are. Therefore, discreteness, D, is defined as A/W. D is qualitatively similar to Mahalanobis’ generalized distance, D 1 (Davis 1986, p. 486). D differs from Mahalanobis’ D 2 in that it does not take into account variable correlations (which may not be reliably estimated for small sample sizes (Atchley et al. 1982; Cherry et al. 1982)), and does not assume a homogeneous variance-covariance structure. In computing IV, the number of pairwise comparisons increases with the square of the group sample size rather than with the sample size itself. This implies that large groups contribute disproportionately to the average distance. A method of weighting was used to correct for this. Within-group distances were weighted so that each group contributes to W according to its sample size rather than the number of comparisons made within that group. This method of weighting is explained below. If: G is the number of groups; ni is the number of specimens in group z (z = 1, ; G' is the number of groups with ni > 1 (i.e. the number of groups in which comparisons can be made); c, is the number of pairwise comparisons in group z (equal to zz((zz;. — l)/2; N is the total number specimens; N' is the total number of specimens in groups with nt > 1 (i.e. the total number of specimens in groups in which comparisons can be made); dtjk is the Euclidean distance between specimens j and k in group z; and d \ is the mean of all pairwise distances within group z, (equal to dijk/cd ; then W is defined as follows: W=~i:dini (2) 1=1 where the sum is only over those groups where n- > 1 . If A were computed without weighting, then a group with a large sample size, i.e. a group whose centroid is very reliably determined, and a group with a small sample size, i.e. a group whose centroid is less reliably determined, would make the same contribution to the average distance among groups (and therefore to the determination of discreteness, D). A method of weighting was used so that each group contributes to A in proportion to its sample size. Thus, groups whose position in morphospace is better determined have greater weight. This is explained below. If: i V, G, and ni are defined as above; z7 is the average group sample size (equal to N/G); M is the number of comparisons among groups (equal to G(G— 1 )/2); and d(} is the distance between the centroids of groups i and j ; then A = 1 2/zM G G V V i=l j=i+ 1 dn{ni+nj). (3) W, A, and D were computed for each of the six stratigraphic intervals. Two questions were addressed regarding temporal changes in dispersion indices. First, does the Cambrian as a whole differ from the Ordovician as a whole? This approach stresses the transition from the Cambrian to the Ordovician. Second, is there a monotonic trend in the dispersion indices? This approach stresses the continuity of the patterns. Some means of comparing these dispersion indices among the intervals is needed. This involves the estimation of how well constrained the indices are, i.e. the estimation of the standard error. Jackknifing (Sokal and Rohlf 1981, p. 795) was used to obtain unbiased estimates of W, A, and D and to determine the variability associated with these estimates. By this method one group is omitted and W, A and D are recomputed. (Because W is not defined for a group with a sample size of one, it is recomputed only if the group omitted has a sample size greater than one.) If G, is the number of groups in the zth interval, then a pseudovalue, Ys, is calculated as Yj = Gi(Y) — (Gi — 1 ) (Xj), where A is the original value (i.e. W, A, or D), and X- is the value calculated when the /th group is omitted. (When calculating pseudovalues corresponding to W, G- is substituted for Gt.) Each group is left out in turn, and the mean of all the Yj provides an unbiased estimate of A. The standard error of the Yj provides an unbiased estimate of the standard error of A. 472 PALAEONTOLOGY, VOLUME 34 table 2. Dispersion indices and their standard errors. In this and all subsequent tables, G is the number of higher taxa relevant to the calculation of A and D, G is the number of higher taxa relevant to the calculation of W, and SE stands for ‘standard error’. Abbreviations: LCAM, Lower Cambrian; MCAM, Middle Cambrian; UCAM, Upper Cambrian; ORD1, Ordovician 1; ORD2, Ordovician 2; ORD3, Ordovician 3. Interval G G' W SE A SE D SE LCAM 6 4 2-57 0-62 2-33 0-72 0-90 0-20 MCAM 9 6 2-40 0-50 2-47 0-39 0-94 0-39 UCAM 14 10 2-62 018 2-39 0-28 0-91 0 12 ORD1 10 10 2-82 0-37 3-80 0-41 1 34 0-06 ORD2 10 9 3-55 0-76 5-39 116 1 46 0-40 ORD3 8 8 2-14 0-26 3-98 0-49 1 84 0-22 text-fig. 4. Unbiased estimates of within- and among-group dispersion plotted against stratigraphic position. Error bars give one standard error on either side of dispersion index. Abbreviations as in Text-figure 2. The unbiased estimates of W, A , and D are given with their standard errors in Table 2 and are shown in Text-figure 4. A method of comparing values through time is needed. One could use parametric statistical approaches, for example, making multiple comparisons among the values, or using the standard errors for analysis of variance. Using the standard errors estimated with jackknifing is analogous to treating each pseudovalue as if it were a single observation. Non- parametric statistical approaches are developed below, but this same approach is used: each pseudovalue is treated as a single datum. FOOTE: TRILOBITE DIVERSIFICATION 473 To test for differences between the Cambrian and the Ordovician, the Kruskal-Wallis statistic, H (a non-parametric analogue to analysis of variance), was computed (Sokal and Rohlf 1981, p. 430). This method treats each observation (pseudovalue) as a ranked variate. For example, there are 57 observations (pseudovalues) computed for the analysis of d. In a ranking from lowest to highest, the six observations for the Lower Cambrian have ranks of 30, 2, 1, 14, 47, and 41, corresponding to the pseudovalues calculated when the groups Eodiscina, Corynexochida, Ptychopariacea, Solenopleuracea, Olenellina, and Redlichiina, respectively, are omitted. In the statistical testing of H , the distribution of ranks among categories (i.e. stratigraphic intervals) is compared to the distribution expected for a random partitioning of ranks. H is distributed approximately as /2 for a random partitioning (Sokal and Rohlf 1981, p. 432). To test for monotonic changes in the dispersion indices, Kendall’s rank correlation coefficient, t, was computed (Sokal and Rohlf 1981, p. 602). The observations are ranked as above, and each stratigraphic interval is ranked from lowest to highest. Statistical tables were constructed by randomization. For example, in the testing of A there are six intervals with 6, 9, 14, 10, 10, and 8 groups, respectively. Thus the total number of observations is 57. The ranks 1 to 57 are randomly assigned to the six intervals with the constraint that the number of ranks assigned to each interval be equal to the actual number of observations in that interval, r is then computed for the randomized ranks. This procedure is repeated 1000 times to construct a distribution of values of r that would be expected by chance (Table 3). If an observed value of r exceeds, say, 95% of the values obtained by randomization, this observed value is considered significant at p = 0-05 and a monotonic trend is inferred. For the data studied here and for the culled data sets discussed below, distributions of r were constructed and compared to the normal approximation (Burr 1960; Sokal and Rohlf 1981, p. 606; Rohlf and Sokal 1981, p. 77) (Table 3). Inspection of the results reveals that the distributions constructed by randomization are generally conservative for statistical testing, i.e., the null hypothesis of lack of monotonicity is less likely to be rejected. table 3. Critical values of r, the rank correlation coefficient, generated by randomization. ‘Tables’ refers to other tables in the text to which these values are relevant. ‘Indices’ refers to dispersion indices in the relevant tables for which these values are used. Subscripts for r refer to the significance levels generated by randomization, /’-values give the corresponding significance level obtained using the normal approximation. Tables Indices Th>5 P him P ^0 001 P 4 W 0-215 0-033 0-272 0-0072 0-347 00006 4, 11 A , D 0190 0-037 0-247 0-0068 0-285 000014 9 W, A, D 0-397 0-021 0-506 00034 0-599 0-0006 1 1 W 0-235 0038 0-286 0-0062 0-323 0002 As would be expected from the two-dimensional representations of higher taxa (Text-fig. 3), there is no significant change in within-group dispersion through time (Table 4). This result holds whether the Cambrian as a whole is compared to the Ordovician as a whole, or whether all six intervals are compared sequentially for monotonic changes. Thus, the obvious increase in total morphological dispersion among all trilobites does not result from the increase in the diversity of forms within an existing suprageneric taxon. Also in agreement with the view presented in Text-figure 3, there is a significant increase in among-group dispersion (Table 4). The Cambrian as a whole differs from the Ordovician as a whole, and the changes among the six intervals indicate a monotonic trend. The total increase in dispersion among all trilobites is therefore linked to evolutionary patterns at taxonomic levels above that of the genus. This increase in among-group dispersion may result from either (1) the first appearance of new higher taxa that are morphologically well removed from their ancestors, or (2) the morphological divergence of established higher taxa, or some combination of these two. These 474 PALAEONTOLOGY, VOLUME 34 table 4. Kruskal -Wallis statistics and Kendall rank correlation coefficients. In this and all subsequent tables, * indicates statistically significant at P < 0-05, ** means significant at P < 0 01, and *** means significant at P < 0 001. All statistical tests in this study are two-sided. Index H x W 0-600 0-029 A 14-676*** 0-325*** D 14191*** q-349*** alternatives are discussed below. Finally, given the significant increase in among-group dispersion and the lack of pattern in within-group dispersion, the morphologic discreteness of higher taxa increases through time (Table 4). This is in accord with previous observations that post-Cambrian trilobites are easier to classify into suprageneric taxa than are Cambrian forms (e.g. Whittington 1966). Reality of morphotypes In addition to investigating temporal changes in dispersion among taxa, it is important to determine whether, for a single stratigraphic interval, the taxa have some reality as morphotypes. One way to test this is to determine whether the discreteness value observed for a single stratigraphic interval differs significantly from discreteness values that would be expected for a random arrangement of specimens into groups. For each interval there are G groups with sample sizes nt, i = 1,...,G. Groups were artificially constructed so that the specimens were randomly divided among the G groups with the corresponding sample sizes. The discreteness, D , was then calculated for this random arrangement. One hundred unique randomizations were constructed for each stratigraphic interval, yielding a distribution of values of D that would be expected by chance. Comparison between observed values of D and the distributions of randomized values for each interval indicates that, with the possible exception of the Lower Cambrian, the arrangement of specimens into higher taxa is morphologically non-random (Table 5). Higher taxa of trilobites are thus shown to represent morphotypes, at least with respect to the shape of the cranidium. table 5. Number of randomized discreteness values greater than observed. Based on 100 randomizations. Interval N Lower Cambrian 6 Middle Cambrian 0 Upper Cambrian 0 Ordovician 1 0 Ordovician 2 0 Ordovician 3 0 Analysis of persistent taxa To determine whether new higher taxa are morphologically displaced from their ancestors, or established higher taxa move away from each other in morphospace, all higher taxa that appear in but a single interval were first removed from the data set, leaving all taxa that persist for two or more intervals. These remaining taxa were then arranged into sets of coexisting, persistent taxa to form smaller sets of data. Five such data sets were constructed and analysed as above (Tables 6-10). The rank correlation coefficient was computed only if the number of stratigraphic intervals was greater than two. FOOTE: TRILOBITE DIVERSIFICATION 475 table 6. Dispersion indices for Eodiscina, Corynexochida, Ptychopariacea and Solenopleuracea in Lower Cambrian and Middle Cambrian. Lower Cambrian Middle Cambrian H G 4 4 G' 3 4 W (SE) 2-06 (0-28) 2-48 (0-55) 0-50 A (SE) 1-42 (0-62) 212 (0-66) 0-33 D (SE) 0-66 (0-34) 0-88 (0-32) 0-08 table 7. Dispersion indices for Asaphiscacea, Crepicephalacea, Marjumiacea, Norwoodiacea, Ptychopariacea and Solenopleuracea in Middle Cambrian and CIpper Cambrian. Middle Cambrian Upper Cambrian H G 6 6 G' 4 5 W (SE) 1-89 (0 06) 2-62 (0-25) 4-86* A (SE) 2-21 (0-59) 1 98 (0-41 ) 0-41 D (SE) 0-79 (0-36) 0-76(0-17) 2-56 table 8. Dispersion indices for Proetida, Komaspidacea, and Olenacea in Upper Cambrian and Ordovician 1 . Upper Cambrian Ordovician 1 H G 3 3 G' 2 3 W (SE) 2-88 (0-39) 2-81 (0-52) 0-33 A (SE) 1-72 (0-08) 4-48 (1-56) 3-86* D (SE) 0-56 (0 14) 1-75 (0-48) 3-86* table 9. Dispersion indices for Scutelluina, Cheirurina, Proetida, Asaphacea, Remopleuridacea and Trinucleacea in Ordovician 1, Ordovician 2, and Ordovician 3. G' is equal to G for all intervals. H measures the overall heterogeneity among the three intervals. ORD1 ORD2 ORD3 H r G 6 6 6 W (SE) 2-90(0-63) 3-57 (M3) 2-30 (0-32) 1-91 -0-27 A (SE) 3-90 (0 64) 4-81 (MO) 4-46 (0-51) 0-22 0-21 D (SE) 1-33(011) 1 26 (0-28) 1 92 (0-24) 3-94 0-428* table 10. Dispersion indices for Scutelluina, Odontopleurida, Cheirurina, Proetida, Asaphacea, Remo- pleuridacea and Trinucleacea in Ordovician 2 and Ordovician 3. G' is equal to G for both intervals. Ordovician 2 Ordovician 3 H G 7 7 W (SE) 3-48 (0-81) 2-30 (0-26) 1-80 A (SE) 4-38 (1-07) 4-13 (0-54) 0-20 D (SE) 1-25 (0-15) 1-78 (0-24) 2-55 476 PALAEONTOLOGY, VOLUME 34 If established higher taxa diverged morphologically, one would expect an increase in among- group dispersion within the subsets of persistent taxa. This is generally not the case. The only exception is the transition from the Upper Cambrian to Ordovician 1. Here a significant increase in among-group dispersion is marked by changes in taxonomic composition within the higher taxa. Komaspidacea in the Upper Cambrian is dominated by the Elviniidae, and in the Lower Ordovician by the Komaspididae. Perhaps more importantly, the Upper Cambrian Proetida are dominated by plethopeltids, and the Lower Ordovician Proetida by hystricurids. That higher taxa tend to occupy a relatively fixed place in morphospace is also evident from inspection of Text-figure 3. Discussion Since persistent higher taxa do not diverge appreciably, the significant increase in among-group dispersion is tied to the origin of new higher taxa. This might be seen as an inevitable consequence of the practice of classification. When forms show significant morphological divergence, they are perforce assigned to new higher taxa, leaving a paraphyletic residue. The phylogenetic relationships among higher taxa of trilobites are not sufficiently well known to state with certainty which groups are paraphyletic. However, the following discussion of higher taxa used in this study suggests that, at the least, we can be confident that paraphyly is more prevalent among Cambrian taxa than among post-Cambrian taxa. Either Redlichiina or Olenellina would appear to be paraphyletic. If opisthoparian sutures are primitive, then Redlichiina may be seen as the paraphyletic ancestor of Olenellina (Eldredge 1977). If, on the other hand, lack of dorsal sutures is the primitive condition, then Olenellina may be the paraphyletic ancestor of Redlichiina (Eortey and Whittington 1989). Eodiscoids are probably derived relative to polymeroid trilobites, and primitive relative to agnostoids (Eldredge 1977; Fortey and Whittington 1989). This suggests that Eodiscina is the paraphyletic ancestor to holophyletic Agnostina. Lane and Thomas (1983), in expressing their belief in the relationship between Corynexochida and Scutelluina, left open the question of whether the corynexochids are a paraphyletic ancestor of Scutelluina, or a holophyletic sister group. Paraphyly appears to be quite common among the ptychoparioid superfamilies. Robison (1987, p. 231) believes that ‘many or most families [of trilobites] arose independently from an unspecialized stock (ptychoparian) ...’ As Eldredge (1977) points out, most similarities among trilobite groups represent symplesiomorphies, and many of the diagnoses of ptychoparioid superfamilies in the Treatise (Harrington et al. 1959) read like descriptions of a generalized trilobite. Of the superfamilies considered here, Asaphiscacea, Crepicephalacea, Komaspidacea, Leiostegiacea, Marjumiacea, Ptychopariacea, and Solenopleuracea seem to fit the description of a generalized ptychoparioid trilobite. On the other hand, a few ptychoparioid superfamilies are characterized by features that may be seen as valid synapomorphies. Conocoryphaceans lack eyes, norwoodiaceans are characterized by proparian or gonatoparian sutures, olenaceans have free cheeks that are fused or separated by a median suture, and raymondinaceans are characterized by cedariiform sutures (Harrington et al. 1959). Phylogenetic analysis of the Asaphina (Fortey and Chatterton 1988) suggests that paraphyly is much less common in this predominantly post-Cambrian suborder. While Fortey and Chatterton believe the Asaphacea and Anomocaracea to be paraphyletic, Cyclopygacea, Dikelocephalacea, Remopleuridacea, and Trinucleaca appear to be holophyletic (Fortey and Chatterton 1988). Although not supported completely by formal phylogenetic analysis, it would seem that other post- Cambrian taxa are quite homogeneous and well derived, so that they are likely to be holophyletic. These include Harpina, Lichida, Odontopleurida, Phacopina, Proetida, and Scutelluina. While the greater prevalence of paraphyletic taxa in the Cambrian no doubt contributes to patterns of within- and among-group dispersion, one observation suggests that this bias is not alone responsible. If taxonomic practice forced among-group dispersion to increase in the way outlined above, it could be argued that the increase should be rather regular. Instead, there is a large jump from the Upper Cambrian to Ordovician 1 , and even within the Ordovician the increase can be seen. But within the Cambrian there is virtually no change in among-group dispersion. There is FOOTE: TRILOBITE DIVERSIFICATION 477 something about the distribution of forms in the Ordovician that allows systematists to define groups in such a way that newer groups are morphologically far removed and distinct relative to older taxa. If the separation of younger taxa were merely the result of this taxonomic artifact, then one would expect to see the pattern within the Cambrian, if the distribution of Cambrian forms allowed this taxonomic practice to be exercised. Taxonomic artifact of another sort must also be considered. As discussed above and elsewhere (e.g. Whittington 1954; Foote 1988), it is possible that Cambrian and post-Cambrian genus concepts are not wholly compatible. The sampling methods employed here were designed to circumvent this bias. However, if taxonomic concepts were disparate at higher levels as well, this difference could, in part, cause the patterns seen here. The results shown above could conceivably tell more about changes in taxonomic practice than in the occupation of morphospace. However, changes in taxonomic practice are not independent of changes in the distribution of forms. It seems reasonable to suppose that if genera in the Cambrian showed a distribution of forms that would allow them to be arranged into discrete suprageneric taxa, then they would have been. Simply put, the results of this quantitative analysis are in agreement with what students of trilobites have long known regarding the distinctness of higher taxa (e.g. Rasetti 1954, 1961 ; Whittington 1954, 1966; Palmer 1958). The pattern of increasing taxonomic separation is clearly linked to the overall morphological diversification of the trilobites. It is conceivable that Cambrian forms are difficult to arrange into discrete suprageneric groups because the total amount of morphospace occupied is so small. It is also possible that taxonomic separation is high in the Ordovician because of the influence of a few extreme groups. Ordovician taxa in the inner regions of morphospace might be similar in distinctness to Cambrian taxa. If so, the increase in average separation could be caused by the large among-group distances associated with the morphologically peripheral taxa. However, the observed pattern is not the result of these two factors, as shown by the following analysis. The morphologic centroid (in the complete, 12-dimensional space) was calculated for each stratigraphic interval. A morphologic distance was chosen that defines a hypersphere centred on the Middle Cambrian centroid, and within which 90% of the Middle Cambrian data happen to fall. (This choice is somewhat arbitrary, but is justifiable. A much smaller volume would exclude too much of the Ordovician data. For example, the volume containing 80% of the Middle Cambrian data includes only 14% of the data of Ordovician 2, and therefore makes statistical analysis dubious. On the other hand, a much larger volume would include too much data, and therefore make the analysis nearly identical with that presented above.) The same volume is placed in turn in each of the six stratigraphic intervals, centred on the morphologic centroid for that interval. This constant volume contains 79% of the Lower Cambrian data, 90% for the Middle Cambrian, 77% for the Upper Cambrian, 59% for Ordovician 1, 44% for Ordovician 2, and 58% for Ordovician 3. Analyses of the data within the constant volume indicates the same pattern as the unculled data. There is no significant change in within-group dispersion, but among-group dispersion and discreteness increase significantly. This implies that the pattern is not caused by extreme taxa, and can be detected at a smaller scale. With respect to taxonomic practice, we can conclude that Cambrian forms are difficult to classify into discrete higher taxa not because the total amount of morphospace occupied is smaller, but because the Ordovician morphospace is occupied in a more discontinuous manner. BIASES IN DATA COLLECTION AND STRATIGRAPHIC CLASSIFICATION Several analyses are presented below to correct for various potential biases in data collection and stratigraphic classification. These analyses involve subsets of data that are culled from the original data set. Space limitations preclude detailed presentation of results, but all further analyses yield patterns in general agreement with those presented above. More detailed treatment can be found in Foote ( 1 989 Paleorhinus parvus Mehl, p. 142, figs 1 and 2; pi. 37. I 10; pi. 38, 1-7; pi. 39. 1 2, 4. 1949 Paleorhinus scurriensis Langston, p. 325, figs 1-3. Holotype. FMNH UC 632, skull (Williston 1904, fig. 6; Lees 1907, fig. 1-7). Locality and Horizon. Popo Agie Formation (Upper Triassic) at Squaw Creek, southeast corner of Township 3 South, Range 1 East, Fremont County, Wyoming. Referred specimen. PPM P217, a partial skull (Text-fig. 4). 492 PALAEONTOLOGY, VOLUME 34 Description of referred specimen. PPM P217 is a phytosaur skull that lacks the rostrum anterior to the external nares and portions of the maxillae lateral to them, and all palatal elements anterior to the posterior portion of the basisphenoid. The maximum length of the skull is 414 mm, with a maximum width of 357 mm, which has been increased by flattening of the skull. The skull is relatively undeformed, although the basioccipital has been pushed forward about 20 mm, and the ventral portions of the quadrates have been pushed posteriorly. The main deformation is the dorsoventral flattening evident in the orientation of the quadrates. The external nares are well forward of the antorbital fenestrae, as is the case with ‘ Par asuchus' hislopi (Chatterjee 1978), Paleorhinus magnoculus (Dutuit 1977) and Paleorhinus bransoni (Lees 1907). The antorbital fenestrae are relatively small. Although the anterior margins of both antorbital fenestrae are broken, the curvature of the upper and lower margins of the right fenestra indicates its original size. In dorsal view, the posterior margin of the skull appears very wide, but this is due to dorsoventral distortion. The external nares are inclined anteriorly. The lateral temporal fenestrae are roughly square in shape and have dorsal margins that are longer than the antorbital fenestrae. The quadratic foramina are large (19 mm maximum diameter) and visible in dorsal view because of the flattening of the skull. The sutural pattern is consistent with other specimens of Paleorhinus (e.g. Langston 1949; Chatterjee 1978). PPM P217 is assigned to Paleorhinus on the basis of having external nares anterior to the antorbital fenestrae, external nares whose dorsal margins incline anteriorly, and the possession of large quadratic foramina. This specimen is assigned to Paleorhinus bransoni because of the small size of the orbits (cf. P. magnoculus) and inclusion of the jugal in the antorbital fenestrae (cf. ‘ Par asuchus' hislopi’. Chatterjee 1978, text-fig. 3 a). The only morphological difference between P. bransoni and P. neukami is in the length of the rostrum, a feature not preserved in the new specimen. On the relatively weak grounds of geographic proximity, the new skull is thus identified as Paleorhinus bransoni. PALEORHINUS TAXONOMY AND DISTRIBUTION USA Wyoming. Williston (1904) named Paleorhinus (type species P. bransoni) for a skull from the Popo Agie Formation at Squaw Creek in the Wind River Mountains of western Wyoming (Mehl 19286). Williston (1904) briefly described the genoholotype of P. bransoni , and subsequently Lees (1907) described it in detail. Mehl (19156, 19286), Jaekel (1910) and Langston (1949) criticized several of Lees’ (1907) interpretations of the structure of the Paleorhinus skull, but did not doubt its generic distinctiveness. Mehl (1915a, 19156) demonstrated that an ilium assigned to Paleorhinus by Lees (1907) actually pertains to the rauisuchian Poposaurus. Mehl (19286) described a second partial skull and skeleton of Paleorhinus (MU 530), which he named P. parvus , from the Popo Agie Formation at Sage Creek in the same area of Wyoming as the type locality of P. bransoni. The skull and lower jaw of Paleorhinus parvus , which is now in three pieces, show no major differences from P. bransoni. Mehl (19286, pp. 155-156) cited principally differences in the length of the rostrum and the degree of downward deflection of the rostral tip to distinguish P. parvus. However, he ignored the large size difference between the skulls of the two putative species. Colbert (1947) documented that relative rostral length is proportional to skull size in phytosaurs. In addition, the deflection of the rostral tip of the holotype skull of P. parvus is probably the result of post-burial deformation. Thus, we consider P. parvus a subjective junior synonym of P. bransoni. Texas. Case (1922) named Promystriosuchus ehlersi for a badly fractured skull from the Tecovas Formation of Crosby County, Texas. Subsequently, Gregory (1962) included this taxon in Paleorhinus. ‘ Promystriosuchus ’ differs from Paleorhinus bransoni in lacking a posterior squamosal hook in lateral view, but this could be the result of damage to the Texas skull. The holotype skull of Promystriosuchus ehlersi (UMMP V7487) is badly distorted and broken anteriorly along the midline so that, in ventral view, the right tooth row is directed ventrally, but the left tooth row is oriented laterally. Also, the lateral aspect of the left external naris is visible along the split midline of the skull. Gregory (1962, pp. 671-673) criticized Case’s (1922) diagnosis of Promystriosuchus ehlersi in detail. We agree with Gregory (1962, pp. 672-673) that all the differences between HUNT AND LUCAS: NON-MARINE UPPER TRIASSIC 493 Paleorhinus bransoni and Promystriosuchus ehlersi cited by Case (1922) are either errors of interpretation or are characters now recognized as variable within phytosaur taxa. P. ehlersi apparently differs from P. bransoni in having a median narial septum which is not visible in lateral view. However, the holotype is so badly distorted and fractured that we consider P. ehlersi a nomen dubium at the species level, although its holotype clearly is a specimen of Paleorhinus. Langston (1949) described Paleorhinus scurriensis from the Camp Springs Member of the Tecovas Formation at the base of the Dockum Group in Scurry County, Texas. The holotype (TTVP 539) is a partial skull that is similar to P. bransoni in having a more anterior placement of the nares than in Promystriosuchus ehlersi , but this is a variable feature within the genus Paleorhinus (Gregory 1962). Langston (1949, p. 325) used qualitative criteria to distinguish this species, including exceptionally large palatine foramina, moderately elongate posttemporal fenestrae, and dorso-ventral flattening of the skull. We do not consider these characters diagnostic, because the palate of most species of Paleorhinus is poorly known, fenestral shape is subject to postmortem deformation, and most Paleorhinus skulls are dorso-ventrally flattened. We are thus unable to diagnose P. scurriensis as a species separate from P. bransoni. Six other undescribed skulls of Paleorhinus are known from Texas, one from the ?lower Tecovas of Borden County (LIT 31213) and five from the lower Dockum Group of Howard County (UT 31100-453, 31100-101, 31100-239, 31100-418, 31025-172; Gregory 1962; Shelton 1984). Shelton (1984) referred all these specimens to P. scurriensis. However, as argued above, P. scurriensis and Promystriosuchus ehlersi are both conspecific with P. bransoni. Indeed, we have examined these specimens, and conclude that all the Texas Paleorhinus material represents one taxon, P. bransoni. Arizona. A small fragment of a Paleorhinus skull (MNA V2698) has been collected from the Downs quarry in the lowermost levels of the Petrified Forest Member of the Chinle Formation in Apache County (Murry and Long 1989). This specimen has external nares anterior to the antorbital fenestrae, but cannot be identified beyond Paleorhinus sp. New Mexico. Toepelman (1916, fig. 1) described a partial phytosaur rostrum from the Bluewater Creek Member of the Chinle Formation at Fort Wingate, McKinley County (Lucas and Hayden 1989) as ? Paleorhinus. However, this fragment is not diagnostic below the subordinal level (Hunt and Lucas 1989). Thus, there are no known occurrences of Paleorhinus in New Mexico. Eastern USA. Paleorhinus has not been reported from the Newark Supergroup of eastern North America (USA and Canada). Historically, and recently, most phytosaur specimens from the Newark have been assigned to Rutiodon , regardless of how fragmentary the material is (e.g. Olsen 1989u. fig. 9.7). However, much of the Newark is Carnian in age, so it is possible that some of the fragmentary phytosaur material represents Paleorhinus. More complete specimens will be needed to evaluate this possibility. Morocco Dutuit (1977) described a nearly complete phytosaurid skull from the Argana Formation as Paleorhinus magnoculus . This species differs from other species of Paleorhinus in the enormous size of the orbits and, possibly, in the exclusion of the jugal from the antorbital fenestra. India Huxley (1870) used the name Parasuchus in a table, and this taxon was validated, and the species P. hislopi named, by Lydekker (1885) for fragmentary reptilian fossils from the Maleri Formation of the Pranhita-Godavari Valley. Subsequently, Huene (1940) identified one of these fragments as a basicranium of the rhynchosaur Paradapedon. The phytosaur specimens have since been referred to aff. Brachysuchus maleriensis by Huene (1940) and to Phytosaurus maleriensis by Colbert (1958). Gregory (1962) concluded that the type specimens of Parasuchus hislopi were generically indeterminate. Chatterjee (1974) designated a phytosaur rostral fragment from among the syntypes 494 PALAEONTOLOGY, VOLUME 34 as the lectotype of Parasuchus hislopi , but this specimen is generically indeterminate, and thus the taxon is a nomen dubium. Virtually complete skeletons of a phytosaur have been collected from the Maleri Formation, and other specimens have been obtained from the Tiki Formation of the Son-Mahanadi Valley (Chatterjee 1967, 1978). Chatterjee (1978) referred these specimens to Parasuchus , but we follow Ballew (1989) in assigning them to Paleorhinus. Although the lectotype of Parasuchus hislopi may be a nomen dubium , we provisionally use the binominal Paleorhinus hislopi for all relevant specimens of ‘ Parasuchus' pending a restudy of all the Indian specimens. The Indian species (P. hislopi) apparently differs from other species of Paleorhinus in lacking interpterygoid vacuities in the palate. West Germany Kuhn (1932, 1936) erected two genera and four species of phytosaurs from the Carnian Blasensandstein at Ebrach in Franconia. These taxa, Francosuchus broilii , F. latus , Ehrachosuchus angustifrons and E. neukami , are morphologically very similar to Paleorhinus. Indeed, Gregory (1962) and Westphal (1976) placed these taxa in a subgenus Francosuchus of the genus Paleorhinus. Chatterjee (1978) considered the specimens Kuhn described to represent the genus Francosuchus , which he placed in a different subfamily from Paleorhinus. However, the only major difference between the Ebrach specimens and other specimens of Paleorhinus is rostral length (Gregory 1962). Francosuchus broilii was originally reconstructed with a short snout (Kuhn 1932), but Kuhn (1936, p. 65) later realised that a portion of the snout was missing. The holotypes of F. latus (Kuhn 1932, fig. 5) and E. angustifrons both lack complete rostra, but that of E. neukami has a very elongate rostrum (Kuhn 1936, pi. 8, la-e). It is principally on the basis of the elongate rostrum of E. neukami that Gregory (1962) and Westphal (1976) placed all the Ebrach phytosaurs in a distinct subgenus from other Paleorhinus specimens. However, rostral length is a variable feature among phytosaurs (Gregory 1962), and we consider it a feature of taxonomic value only at the species level. Therefore, we do not uphold separate generic or subgeneric status for Ehrachosuchus or Francosuchus. Chatterjee (1978) considered Francosuchus to be a separate genus on the basis of the position of the external nares relative to the antorbital fenestrae, which is a variable character (Gregory 1962), and the absence of posterior squamosal processes. However, the Ebrach skulls that have undamaged posterior margins exhibit posterior squamosal processes (Kuhn 1936, pi. 8, la; pi. 10, 5). Thus, we believe the taxonomic disposition of the Ebrach skulls should be to consider Paleorhinus neukami a distinct species based on its elongate rostrum, and to refer the other nominal taxa to Paleorhinus sp. because they lack diagnostic features. Kuhn (1936) established another new phytosaur taxon from Ebrach, Ebrachosaurus singularis , that is obviously a Stagonolepis-hke aetosaur, as noted by Benton and Walker (1985) (compare Kuhn 1936, pi. 13, 4 with Walker 1961, fig. 16, and Kuhn 1936, pi. 11, 1-3 with Walker 1961, fig. 20a-o). Kuhn also identified a lower jaw from Ebrach as Mystriosuchus , a Norian genus, but this specimen is indeterminate (Gregory 1962). In 1910, Jaekel described a phytosaur skull, supposedly from the Buntsandstein (Lower Triassic) of Bernberg, as a new genus, Mesorhinus. There are several problems with this taxon. The holotype was found to have a label that read Trematosaurus (a labyrinthodont taxon: Jaekel 1910). The specimen is undoubtedly a phytosaur, but the label with the specimen indicated that it was from the Early Triassic, whereas all other phytosaur taxa are restricted to the Late Triassic (Jaekel 1910; Gregory 1962). Also, the holotype was destroyed in the Second World War (Gregory 1962). Jaekel (1910) attempted to verify the locality data on the label by examining the matrix around the specimen and concluded that it was, indeed, from the Buntsandstein, but this cannot now be checked. Kuhn (1961) substituted the name Mesorhinosucluts for Mesorhinus because this name was preoccupied by that of a South American fossil mammal (Ameghino 1885). Recent authors have either considered Mesorhinosuchus Kuhn, 1961 ( = Mesorhinus Jaekel, 1910) a tentative synonym of Paleorhinus (Gregory 1962; Westphal 1976) or as indeterminate (Chatterjee 1978). Mesorhinosuchus is undoubtedly a phytosaur (Walker 1968, p. 11; contra Gregory 1962, p. 675) of Pa/eorhinus-\ike HUNT AND LUCAS: NON-MARINE UPPER TRIASSIC 495 morphology. It differs from all other adult phytosaurs in retaining a small pineal foramen (Jaekel 1910; Camp 1930), although the holotype of P. scurriensis has a shallow pit in this region (Langston 1949). Mesorhinosuchus is best considered as Paleorhinus sp. on the basis of the anterior placement of the nares. The age of the specimen must be considered indeterminate. A ustria Huene (1939) coined the name cf. Francosuchus trauthi for a skull fragment of a phytosaur collected in 1905 near Lunz (about 1 10 km southwest of Vienna), Austria. This specimen was derived from dark gray to black shale of the lower part of the Opponitzer Schichten (Opponitzer Kalk of some authors) (Trauth 1948, p. 90). The Opponitzer Schichten of the Northern Alps are a predominantly marine-limestone unit of late Carnian (Tuvalian) age (Janoscheck and Matura 1980; H. Zapfe, written comm., 1989). The skull fragment Huene named cf. Francosuchus trauthi clearly pertains to Paleorhinus and thus establishes a link between the nonmarine occurrence of Paleorhinus and Triassic marine biochronology. This skull fragment is number 1905/13 in the collection of the Naturhistorisches Museum of Vienna (a sharp resin cast is NMMNM P-12960) and was illustrated by Huene (1939, fig. la-c), Trauth (1948, fig. 14; pi. 12, figs 6 and 7) and Westphal (1976, fig. 7c). It is assignable to Paleorhinus because the external nares are obviously forward of the antorbital fenestae and lie on the posterior portion of the rostrum. This specimen, nevertheless, exhibits no other diagnostic characters and is here referred to Paleorhinus sp. THE PALEORHINUS BIOCHRON Cope (1875) first used fossil vertebrates to determine the age of red beds in the American West that we now recognize to be of Late Triassic age. Subsequently, Huene (1922a, 19226, 1926) established a crude biochronology within these red beds, based principally on phytosaurs. Huene (1926, pp. 3, 4) noted that ‘parasuchians such as Palaeorhinus [sic]..., having a supratemporal fenestra with a high posterior border, are relatively primitive and could not possibly be of Upper Triassic age’ in contrast to phytosaurs from higher stratigraphic levels such as ‘ IPhytosaurus dough ti ” [sic] which he thought were of Late Triassic age. Camp (1930), in the course of his revision of the phytosaurs, realized that Huene’s two faunas were both of Late Triassic age and refined the biochronology to recognize four successive faunas. Camp (1930), like Huene (1926), realized that Paleorhinus is more primitive than other phytosaurs, although he failed to recognize that Promystriosuchus is congeneric with Paleorhinus. Gregory (1962) published the next revision of the phytosaurs, and, in a series of articles (Gregory 1956, 1969; Colbert and Gregory 1957), he outlined a worldwide biochronology for the Late Triassic based on vertebrate faunas. Gregory (1956, 1969) recognized four faunas in North America, the oldest of which was defined by the co-occurrence of the phytosaurs Paleorhinus and Angistorhinus. Gregory (1956, 1969) correlated this fauna with the Blasensandstein of Germany. Chatterjee (1978) and Ballew (1989) subsequently used the occurrence of Paleorhinus to correlate the Argana Formation of Morocco and the Maleri and Tiki Formations of India with the Blasensandstein. Paleorhinus occurs with faunas that are distinct from those of overlying or underlying strata and which vary geographically. The phytosaur Angistorhinus occurs in several faunas with Paleorhinus (Popo Agie, lower Dockum, Argana). The co-occurrence of the rhynchosaur Hvperodapedon with Paleorhinus in the Maleri Formation (Benton 1983) is strong evidence to suggest that the Lossiemouth Sandstone Formation of Scotland and the Wolfville Formation of Nova Scotia, which also contain Hvperodapedon (Benton 1983; Olsen 19896), are also of the same age ( contra Cooper 1982). A complicating factor is that the Lossiemouth fauna also includes the aetosaur Stagonolepis (Walker 1961; Benton and Walker 1985) which occurs in North America (Calyptosuchus of Long and Ballew 1985) with post- Paleorhinus phytosaurs (Murry and Long 1989). 496 PALAEONTOLOGY, VOLUME 34 The Ischigualasto Formation of Argentina and the upper Santa Maria Formation of Brazil contain the rhynchosaur Scaphonyx which is very similar to Hyperodapedon (Benton 1983) and are also probably of the same age. The Ischigualasto, Santa Maria, and Lossiemouth Sandstone Formations contain terrestrial faunas that lack semiaquatic taxa such as phytosaurs. Other tetrapod taxa that are found in Paleorhinus- bearing or equivalent faunas are aetosaurs (lower Dockum - Longosuchus; Blasensandstein - Ebrachosuchus ; Maleri - undescribed; Ischigualasto/Santa Maria - Aetosauroides ; Lossiemouth/lower Chinle - Stagonolepis ), metoposaurs (Popo Agie, lower Dockum, Blasensandstein, Maleri, Argana), dicynodonts (Popo Agie, Argana, Ischigualasto), and rauisuchians (Popo Agie, lower Dockum, Maleri, Ischigualasto). Few of these taxa aid in correlation with the North American Late Triassic, but the rauisuchian Poposaurus occurs in the Popo Agie and the lower Dockum, and indistinguishable metoposaurs (Hunt 1989c/) occur in the lower Dockum (. Buettneri howardensis ), Maleri (Metoposaurus maleriensis ), Argana ( Metoposaurus azerouali) and the Wolfville and Camp Springs ( Buettneria bakeri : Case 1932; Baird 1986). In addition, the dicynodont Moghreberia from the Argana (Dutuit 1988) is very similar to Placerias from the lowermost Chinle (Camp and Welles 1956) and they may be congeneric (Lucas 1990). Thus, the aetosaurs, metoposaurs, dicynodonts and rauisuchians that occur in Paleorhinus-beanng strata or their equivalents are distinct from taxa in underlying and overlying strata. In the western United States and Germany, Paleorhinus-bearmg faunas are succeeded by faunas dominated by other phytosaur taxa. There are only two occurrences of overlap between Paleorhinus and more derived phytosaurs. At the Downs’ quarry in the lower Petrified Forest Member of the Chinle Formation in northeastern Arizona, a single skull fragment of Paleorhinus (MNA V 2698) co-occurs with the phytosaur Rutiodon. The remainder of the Chinle phytosaur fauna is dominated by fossils of Rutiodon and Pseudopalatus. At Home Creek in Crosby County, Texas, Case (1922) reported the presence of Paleorhinus (= Promystriosuchus ), but, subsequently, only specimens of Rutiodon have been found in this area (Gregory 1972). However, Case (1922) did not give exact geographic or stratigraphic information about his locality, and the Paleorhinus-bearmg Camp Springs Member does crop out in this area (Finch and Wright 1983; Finch et al. 1976). Thus, the two taxa of phytosaurs may not co-occur in the same fauna in West Texas. Paleorhinus occurs with faunas distinct from those of overlying and underlying strata that can be correlated throughout much of the world, and this taxon exhibits negligible stratigraphic overlap with other phytosaurs (Text-fig. 5). Therefore, we recognize a Paleorhinus biochron (Lucas and Hunt 1989) that has biochronological utility across Pangaea. The faunas that contain Paleorhinus have been considered Carnian in age by all recent authors (Murry 1982, 1986, 1989; Lucas et al. 1985 ; Chatterjee 1986; Olsen and Sues 1986; Lucas and Hunt 1989; Ballew 1989). Data from palynology (lower Dockum, Blasensandstein), radiometric dating W-CENTRAL WYOMING (USA) ST. JOHNS ARIZONA (USA) RANDALL & CROSBY CO. TEXAS (USA) HOWARD CO. TEXAS (USA) SOUTHERN WEST GERMANY AUSTRIA MOROCCO CENTRAL INDIA POPO AGIE FORMATION 4m TRUJILLO FM. UNTERER BURGSANDSTEIN HAUPTDOLOMIT main body OPPONITZER SCHICHTEN Camp Springs Member DOCKUM GROUP (undivided) BLASENSANDSTEIN LUNZ SCHICHTEN ARGANA FORMATION MALERI FORMATION • PALEORHINUS OCCURRENCES text-fig. 5. Correlation of Upper Triassic Paleorhinus-bearing strata of Pangaea. See text for discussion. HUNT AND LUCAS: NON-MARINE UPPER TRIASSIC 497 (Ischigualasto) and marine invertebrates (Opponitzer Beds) have the potential of giving a more refined age for these faunas. Paleorhinus sp. from the Opponitzer Beds from near Lunz in Austria was found associated with an upper Carnian brackish marine fauna (Huene 1939; Westphal 1976). This specimen thus can be correlated into the standard marine sequence of the Alpine province, via ammonites and pollen, that indicate it is of Tuvalian age (late Carnian: Janoscheck and Matura 1980). This ties the Paleorhinus biochron to Triassic marine biochronology. The Ischigualasto Formation of the Ischigualasto-Ischichuca basin of northwestern Argentina is associated with basalt and diabases which yield radiometric ages with a mean of 224 + 5 Ma (Gonzales and Toselli in Valencio et al. 1975). This date may be judged as approximately mid Carnian in age (Forster and Warrington 1985), but the spread of radiometric dates from the Ischigualasto is from early Carnian to early Norian (Forster and Warrington 1985). The age relationships of strata of the Middle Keuper in Germany are somewhat controversial despite palynological studies (Benton 1986). The Blasensandstein, which contains Paleorhinus , is equivalent to part of the Rote Wand of southwestern Germany which has been considered earliest Norian or early late Carnian in age (Fisher 1972; Fisher and Bujak 1975; Kozur 1975; Gall et al., 1977; Schroeder 1982). We prefer the latter correlation, as we believe that the unconformity at the base of the Stubensandstein may correlate with unconformities in other parts of the world that reflect a major eustatic fall of sea-level at the Carnian-Norian boundary (Embry 1988). Dunay (1972) attempted to compare the palynology of the Paleorhinus and ‘ Phytosaurus' ( Rutiodon ) zones of Gregory in the Tecovas Formation. However, Dunay’s (1972) samples from the Paleorhinus zone were from Crosby County where, as he noted, Rutiodon is also found, and there are no good locality data for the older collections. Therefore, he may have sampled a Paleorhinus fauna, a Rutiodon fauna, or a transitional fauna that contains both (cf. Downs’ quarry). However, Dunay (1972; Dunay and Fisher 1979) was certain that the palynofloras of the Tecovas Formation and the overlying Trujillo Formation were late Carnian in age. Litwin (1986) concluded that the lower Chinle Formation in Arizona that contains Rutiodon , a taxon characteristic of post- Paleorhinus strata in Texas, was also late Carnian in age. Palynological evidence thus suggests that the Paleorhinus biochron is of late-middle (middle Tuvalian), but not latest Carnian age. Ash (1980) reviewed the biochronology of megafossil plants in North America and proposed a number of ‘floral zones’. The only Paleorhinus- bearing stratigraphic unit that also contains megafossil plants is the Popo Agie Formation which Ash (1980) placed in his Eoginkgoites ‘floral zone’ of middle Carnian age. This age determination was based on palynological studies of the Newark Supergroup in Eastern North America (Cornet 1977) and vertebrate correlations. However, Ash (1980) only tentatively placed the Popo Agie flora in this zone, and the name-bearing taxon is only represented by ? Eoginkgoites. Thus, we have little confidence in assigning a middle Carnian age to the Popo Agie from the megafossil plants. Instead, we conclude that the Popo Agie is of late Carnian age. Dutuit (1983) explained the cosmopolitan nature of Late Triassic faunas dominated by phytosaurs and metoposaurs as being due to marine dispersal by these animals. However, there is no evidence that these animals lived in marine conditions, and there are terrestrial rather than marine connections between most occurrences of these faunas (Buffetaut and Martin 1984). Paleorhinus was a cosmopolitan taxon in the late Carnian, but phytosaur taxa in the Norian are more restricted in their distribution (Ballew 1989). A similar situation is seen in metoposaurs (Hunt 19896) and these changes reflect increased provincialization of faunas towards the end of the Late Triassic. Acknowledgements. We thank B. Harrison (PPM) for allowing us to borrow PPM P217, R. A. Long (University of California Museum of Paleontology) for bringing some literature to our attention, H. Zapfe (Osterreich Akadamie der Wissenschaft) for information about the provenance of cf. ‘ Francosuchus trauthi', H. A. Kollman (Naturhistorisches Museum, Wien) for a cast of cf. ‘ Francosuchus trauthi ’, two anonymous reviewers for their comments, M. J. Benton for editorial assistance, and the New Mexico Museum of Natural History for support. 498 PALAEONTOLOGY, VOLUME 34 REFERENCES ameghino, f. 1885. Nuevos restos de mamiferos fosiles oligocenos recogidos por el profesor Pedro Scalabrini y pertenecientes al Museo provincial de la cuidad del Parana. Boletin Academia Ciencia (Cordoba), 8, 5-205. ash, s. R. 1980. Upper Triassic floral zones of North America. 153-170. In dilcher, d. l. and taylor, t. n. (eds). Biostratigraphy of fossil plants. Dowden, Hutchinson and Ross, Stroudsburg, Virginia, 259 pp. baird, d. 1986. Some Upper Triassic reptiles, footprints and an amphibian from New Jersey. Mosasaur , 3, 125-153. 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, New Mexico, 414 pp. benton, M. J. 1983. The Triassic reptile Hyperodapedon from Elgin: functional morphology and relationships. Philosophical Transactions of the Royal Society of London. Series B, 302, 605-717. 1986. The Late Triassic tetrapod extinction events. 303-320. In padian, k. (ed. ). The beginning of the age of dinosaurs : faunal change across the Triassic-Jurassic boundary. Cambridge University Press, Cambridge, 378 pp. — and walker, a. d. 1985. The palaeoecology, taphonomy and dating of the Permo-Triassic reptiles from Elgin, northeast Scotland. Palaeontology , 28, 207-234. buffetaut, e. and martin, m. 1984. Continental vertebrate distribution, faunal zonation and climate in the Late Triassic. 3rd Symposium on Mesozoic Terrestrial Ecosystems, Short Papers, 25-29. 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, 1-174. — and welles, s. p. 1956. Triassic dicynodont reptiles. Part 1. The North American genus Placerias. Memoirs of the University of California, 13, 255-304. case, e. c. 1922. New reptiles and stegocephalians from the Upper Triassic of western Texas. Carnegie Institution of Washington Publication, 321. 1-84. — 1932. A collection of stegocephalians from Scurry County, Texas. Contributions from the Museum of Paleontology of the University of Michigan, 4, 1-56. — and white, t. e. 1934. Two new specimens of phytosaurs from the Upper Triassic of western Texas. Contributions from the Museum of Paleontology of the University of Michigan, 4, 133-142. chatterjee, s. 1967. New and associated phytosaur material from the Upper Triassic Maleri Formation of India. Bulletin of the Geological Society of India, 4, 108-1 10. 1974. A rhynchosaur from the Upper Triassic Maleri Formation of India. Philosophical Transactions of the Royal Society of London, Series B, 267, 209-261. 1978. A primitive parasuchid (phytosaur) from the Upper Triassic Maleri Formation of India. Palaeontology, 21, 83-127. 1986. The Late Triassic Dockum vertebrates: their stratigraphic and paleobiogeographic significance. 140-150. In padian, K. (ed.). The beginning of the age of dinosaurs : faunal change across the Triassic-Jurassic boundary. Cambridge University Press, Cambridge, 278 pp. Colbert, e. H. 1947. Studies of the phytosaurs Machaeroprosopus and Rutiodon. Bulletin of the American Museum of Natural History, 88, 53-96. — 1958. Relationships of the Triassic Maleri fauna. Journal of the Palaeontological Society of India, 3, 68-8 1 . — and Gregory, j. t. 1957. Correlation of continental Triassic sediments by vertebrate fossils. Bulletin of the Geological Society of America, 68, 1456-1467. cooper, m. r. 1982. A mid-Permian to earliest Jurassic tetrapod biostratigraphy and its significance. Arnoldia (Zimbabwe), 7, 77-103. cope, e. d. 1875. Report on the geology of that part of New Mexico examined during the field-season of 1874. Annual Report of the Wheeler Survey, 61-116. CORNET, B. 1977. The palynostratigraphy of the Newark Supergroup. Unpublished Ph.D. thesis, Pennsylvania State University. dunay, r. E. 1972. The palynology of the Triassic Dockum Group of Texas, and its application to stratigraphic problems of the Dockum Group. Unpublished Ph D. thesis, Pennsylvania State University. — and fisher, m. j. 1979. Palynology of the Dockum Group (Upper Triassic), Texas. Reviews of Palaeobotany and Palynology , 28, 61-92. dutuit, J. M. 1977. Paleorhinus magnoculus, phytosaure du Trias superieur de l’Atlas Marocain. Geologie Mediterrane'enne, 4, 225-238. 1983. Homogeneite et repartition des metoposaurides et phytosaurides; sont-elles en relation avec les HUNT AND LUCAS: NON-MARINE UPPER TRIASSIC 499 espaces marins du Trias superieur? Comptes rendus hebdomadciires des Seances de l 'Academie des Sciences , Serie II, 296, 1465-1468. 1988. Osteologie cranienne et ses enseignements, apports geologique et paleoecologique, de Moghreberia nmachouensis, dicynodonte (Reptilia, Therapsida) du Trias superieur marocain. Bulletin du Museum national d'Histoire naturelle (Paris), Serie 4, IOC, 227-285. embry, a. f. 1988. Triassic sea-level changes: evidence from the Canadian Arctic archipelago. Society of Economic Paleontologists and Mineralogists Special Publication , 42, 249-259. finch, w. i. and wright, j. c. 1983. Measured stratigraphic sections of uranium-bearing Upper Triassic rocks of the Dockum basin, eastern New Mexico, West Texas, and the Oklahoma Panhandle with brief discussion of stratigraphic problems. United States Geological Survey Open-File Report, 83-701, 1-118. and davis, b. e. 1976. Unevaluated preliminary geologic cross section of uranium-bearing Upper Triassic rocks extending from Palo Duro Canyon across the Matador Arch, through the type locality of the Dockum Group, to the White River Reservoir, Crosby County, Texas. United States Geological Survey Open-File Report, 76-376, 1. fisher, m. j. 1972. The Triassic palynofloral succession in England. Geoscience and Man, 4, 101-109. — and bujak, j. 1975. Upper Triassic palynofloras from Arctic Canada. Geoscience and Man, 11, 78-94. forster, s. c. and Warrington, G. 1985. Geochronology of the Carboniferous, Permian and Triassic. 99-1 13. In snelling, N. j. ( ed . ). The chronology of the geological record. Geological Society of London, London, 340 pp. gall, j.-c., durand, m. and muller, e. 1977. Le Trias de part d'autre du Rhin. Correlations entre les marges et le centre du bassin germanique. Bulletin du Bureau de Recherches geologiques et minieres, 3, 193-204. Gregory, j. t. 1956. Significance of fossil vertebrates for correlation of Late Triassic continental deposits of North America. International Geological Congress, 20 (2), 7-25. 1962. The genera of phytosaurs. American Journal of Science 260, 652-690. - 1969. Evolution und interkontinentale Beziehungen der Phytosauria (Reptilia). Palaontologisches Zeitschrift , 43, 37-51. - 1972. Vertebrate faunas of the Dockum Group, Triassic, eastern New Mexico and West Texas. New Mexico Geological Society Guidebook, 23, 120-123. huene, f. von 1922u. Neue Beitrage zur Kenntnis der Parasuchier. Jahrbuch Preussische Geo/ogische Landesanstalt, 42, 59-160. 1922h. Kurzer Ueberblick fiber die triassische Reptilordnung Thecodontia. Centralblatt fur Mineralogie, Geologie und Paldontologie, 1922, 408-415. 1926. Notes on the age of the continental Triassic beds in North America with remarks on some fossil vertebrates. Proceedings of the United States National Museum, 69 (18), 1-10. 1939. Ein primitiv Phytosaurier in der jungeren nordostalpin Trias. Zeitschrift fiir Mineralogie, Geologie und Paldontologie, 1939, 139-144. 1940. The tetrapod fauna of the Upper Triassic Maleri beds. Palaeontologia Indica , 32, 1-42. hunt, a. p. 1989n. Comments on the taxonomy of Late Triassic metoposaurs and a preliminary phylogenetic analysis of the family Metoposauridae. 293-300. 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. I989A The biochronological significance of Late Triassic metoposaurid labyrinthodonts. New Mexico Journal of Science, 29, 1 17-1 18. — and lucas, s. G. 1989. Late Triassic vertebrate localities in New Mexico. 72-101. 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. huxley, t. h. 1870. On the classification of Dinosauria, with observations on the Dinosauria ot the Trias. Quarterly Journal of the Geological Society of London, 26, 32-51. jaekel, o. 1910. Ueber einen neuen Belodonten aus dem Buntsandstein von Bernburg. Sitzungsberichte Gesellschaft naturforschender Freunde zu Berlin, 5, 197-229. janoscheck, w. R. and matura, A. 1980. Austria. 1-88. In anonymous (ed.). Geology of the European countries. Dunod, Paris, 433 pp. kozur, h. 1975. Probleme der Triasgliederung und Parallelisiering der germanischen und tethyalen Trias. Teil II: Anschluss der germanischen Trias an die internationale Triasgliedering. Freiberger Forschungshefte , Reihe C, 3, 58-65. kuhn, o. 1932. Labyrmthodonten und Parasuchier aus dem mittleren Keuper von Ebrach in Oberfranken. Neues Jahrbuch fiir Mineralogie, Geologie und Paldontologie, 69B, 94—144. 500 PALAEONTOLOGY, VOLUME 34 — 1936. Weitere Parasuchier und Labyrinthodonten aus dem Blasensandstein des mittleren Keuper von Ebrach. Palaeontographica , Abteilung A , 83, 61-98. — 1961. Die Familien der rezenten und fossilen Amphibien und Reptilien. Verlagshus Meisenbacli KG, Bamberg, 79 pp. langston, w. jr. 1949. A new species of Paleorhinus from the Triassic of Texas. American Journal of Science , 247, 324-371. lees, J. H. 1907. The skull of Paleorhinus. Journal of Geology , 15, 121-151. litwin, R. J. 1986. The palynostratigraphy and age of the Chinle and Moenave formations, southwestern USA. Unpublished Ph D. thesis, Pennsylvania State University. long, R. a. and ballew, k. l. 1985. Aetosaur dermal armor from the Late Triassic of the southwestern North America, with special reference to material from the Chinle Formation of Petrified Forest National Park. Museum of Northern Arizona Bulletin , 54, 35-68. lucas, s. G. 1990. Toward a vertebrate biochronology of the Triassic. Albertiana , in press. — and hayden, s. N. 1989. Triassic stratigraphy of west-central New Mexico. New Mexico Geological Society Guidebook , 40. 191-211. — and hunt, a. p. 1989. Vertebrate biochronology of the Late Triassic. 28th International Geological Congress Abstracts , 2, 335-336. — and morales, m. 1985. Stratigraphic nomenclature and correlation of Triassic rocks of east-central New Mexico: a preliminary report. New Mexico Geological Society Guidebook 36, 171-184. lydekker, R. 1885. Maleri and Denwa Reptilia and Amphibia. Palaeontologia Indica , Series 4 , 1, 1-38. Matthews, w. a. in. 1969. The geologic story of Palo Duro Canyon. Bureau of Economic Geology , University of Texas, Guidebook , 8, 1-51. mehl, m. G. 1915a. Poposaurus gracilis, a new reptile from the Triassic of Wyoming. Journal of Geology, 23, 516-522. 19156. The Phytosauria of the Trias. Journal of Geology, 23, 129-165. 1922. A new phytosaur from the Trias of Arizona. Journal of Geology, 30. 144-157. - 1928a. Pseudopalatus pristinus , a new genus and species of phytosaur from Arizona. University of Missouri Studies, 3, 1-22. — 19286. The Phytosauria of the Wyoming Triassic. Denison University Journal of Scientific Laboratories, 23. 141-172. murry, p. a. 1982. Biostratigraphy and paleoecology of the Dockum Group (Triassic) of Texas. Unpublished Ph D. thesis. Southern Methodist University. 1986. Vertebrate paleontology of the Dockum group, western Texas and eastern New Mexico. 109-137. In padian, K. (ed.). The beginning of the age of dinosaurs : faunal change across the Triassic -Jurassic boundary. Cambridge University Press, Cambridge, 378 pp. 1989. Geology and paleontology of the Dockum Formation (Upper Triassic), West Texas and eastern New Mexico. 102-144. In lucas, s. g. and hunt, a. p. (eds). Dawn of the age of the dinosaurs in the American Southwest. New Mexico Museum of Natural History, Albuquerque, New Mexico, 414 pp. — and long. r. a. 1989. Geology and paleontology of the Chinle Formation, Petrified Forest National Park and vicinity, Arizona and a discussion of vertebrate fossils of the southwestern Upper Triassic. 29-64. 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. olsen, p. e. 1989a. Stop 9.1 : St Mary’s Bay, Rossway, NS. 135-137. In olsen, p. e., schlische, r. w. and gore, p. J. w. (eds). Tectonic , depositional and paleoecological history of early Mesozoic rift basins, eastern North America. American Geophysical Union, Washington, 174 pp. 19896. Stop 11.1: Carrs Brook near Lower Economy. 149-150. In olsen, p. e., schlische, r. w. and gore, p. J. w. (eds). Tectonic, depositional and paleoecological history of early Mesozoic rift basins , eastern North America. American Geophysical Union, Washington, 174 pp. — and sues, h.-d. 1986. Correlation of continental Late Triassic and Early Jurassic sediments and patterns of the Triassic-Jurassic tetrapod transition. 321-351. In padian, k. (ed.). Beginning of the age of dinosaurs: faunal change across the Triassic-Jurassic boundary. Cambridge University Press, Cambridge, 378 pp. Schaeffer, b. 1967. Late Triassic fishes from the western United States. Bulletin of the American Museum of Natural History. 135, 287-342. — and Gregory, j. t. 1961. Coelacanth fishes from the continental Triassic of the western United States. American Museum Novitates, 2036, 1-18. schroeder. b. 1982. Entwicklung des Sedimentbeckens und Stratigraphie der klassischen Germanischen Trias. Geologische Rundschau, 71, 783-794. HUNT AND LUCAS: NON-MARINE UPPER TRIASSIC 501 shelton, s. v. 1984. Parasuchid reptiles from the Triassic Dockum Group of West Texas. Unpublished M.Sc. thesis, Texas Tech University. toepelman, w. C. 1916. Phytosaur remains from New Mexico. Bulletin of the University of Oklahoma 103, 26-28. trauth, f. 1948. Geologie des Kalkalpenbereiches der zweiten Wiener Hochquellenleitung. Abhandlungen der Geologie B un de sans t alt, 26, 1-99. valencio, d. a., mendia, j. e. and vilas, J. f. 1975. Paleomagnetism and K-Ar ages of Triassic igneous rocks from the Ishigualasto-Iscichuca basin and Puesto Viejo Formation, Argentina. Earth and Planetary Science Letters 26, 319-330. walker, a. d. 1961. Stagonolepis, Dasygnathus and their allies. Philosophical Transactions of the Royal Society of London, Series B, 248, 103-204. — 1968. Protosuchus , Proterochampsa and the origin of phytosaurs and crocodiles. Geological Magazine, 105, 1-14. westphal, f. 1976. Phytosauria. 99-120. In charig, a. j., krebs, b., sues, h.-d. and westphal, f. Thecodontia : Handbuch der Palaeoherpeto/ogie Teil 13. Gustav Fischer Verlag, Stuttgart, 136 pp. williston, s. w. 1904. Notice of some new reptiles from the Upper Trias of Wyoming. Journal of Geology 12, 688-697. Typescript received 31 January 1990 Typescript accepted 5 April 1990 ADRIAN P. HUNT AND SPENCER G. LUCAS New Mexico Museum of Natural History, Post Office 7010, Albuquerque, New Mexico 87194-7010, USA I NOTES FOR AUTHORS The journal Palaeontology is devoted to the publication of papers on all aspects of palaeontology. Review articles are particularly welcome, and short papers can often be published rapidly. A high standard of illustration is a feature of the journal. Four parts are published each year and are sent free to all members of the Association. Typescripts should conform in style to those already published in this journal, and should be sent to Dr Dianne Edwards, Department of Geology, University of Wales College of Cardiff CF1 3YE, who will supply detailed instructions for authors on request (these are published in Palaeontology 1990, 33, pp. 993-1000). Special Papers in Palaeontology is a series of substantial separate works conforming to the style of Palaeontology . SPECIAL PAPERS IN PALAEONTOLOGY In addition to publishing Palaeontology the Association also publishes Special Papers in Palaeontology. Members may subscribe to this by writing to the Membership Treasurer: the subscription rate for 1991 is £45-00 (U.S. $80) for Institutional Members, and £20-00 (U.S. $36) for Ordinary and Student Members. A single copy of each Special Paper is available on a non-subscription basis to Ordinary and Student Members only , for their personal use, at a discount of 25 % below the listed prices: contact the Marketing Manager. Non-members may obtain Nos 35—43 (at cover price) from Basil Blackwell Ltd., Journal Subscription Department, Marston Book Services, P.O. Box 87, Oxford OX2 0DT, England, and older issues from the Marketing Manager. For all orders of Special Papers through the Marketing Manager, please add £1-50 (U.S. $3) per item for postage and packing. PALAEONTOLOGICAL ASSOCIATION PUBLICATIONS Special Papers in Palaeontology For full catalogue and price list, send a self-addressed, stamped A4 envelope to the Marketing Manager. Numbers 2-30 are still in print and are available together with those listed below: 31. (for 1984): Systematic palaeontology and stratigraphic distribution of ammonite faunas of the French Coniacian, by w. j. Kennedy. 160 pp., 42 text-figs., 33 plates. Price £25 (U.S. $50). 32. (for 1984): Autecology of Silurian organisms. Edited by m. g. bassett and s. d. lawson. 295 pp ., 75 text-figs., 13 plates. Price £40 (U.S. $80). 33. (for 1985): Evolutionary Case Histories from the Fossil Record. Edited by J. c. w. cope and p. w. skelton. 202 pp., 80 text-figs., 4 plates. Price £30 (U.S. $60). 34. (for 1985): Review of the upper Silurian and lower Devonian articulate brachiopods of Podolia, by o. i. Nikiforova, t. l. modzalevskaya and m. g. bassett. 66 pp., 6 text-figs., 16 plates. Price £10 (U.S. $20). 35. (for 1986): Studies in palaeobotany and palynology in honour of N. F. Hughes. Edited by d. j. batten and d. e. g. briggs. 178 pp., 29 plates. Price £30 (U.S. $60). 36. (for 1986): Campanian and Maastrichtian ammonites from northern Aquitaine, France, by w. j. Kennedy. 145 pp., 43 text-figs., 23 plates. Price £20 (U.S. $40). 37. (for 1987): Biology and revised systematics of some late Mesozoic stromatoporoids, by rachel wood. 89 pp., 31 text- figs., 7 plates. Price £20 (U.S. $40). 38. (for 1987): Taxonomy, evolution, and biostratigraphy of late Triassic-early Jurassic calcareous nannofossils, by p. r. bown. 118 pp., 19 text-figs., 15 plates. Price £30 (U.S. $60). 39. (for 1988): Late Cenomanian and Turonian ammonite faunas from north-east and central Texas, by w. j. Kennedy, 131 pp., 39 text-figs., 24 plates. Price £30 (U.S. $60). 40. (for 1988): The use and conservation of palaeontological sites. Edited by p. r. crowther and w. a. Wimbledon. 200 pp., 31 text-figs. Price £30 (U.S. $60). 41. (for 1989): Late Jurassic-early Cretaceous cephalopods of eastern Alexander Island, Antarctica, by p. j. howlett. 72 pp., 9 text-figs., 10 plates. Price £20 (U.S. $40). 42. (for 1989): The Palaeocene flora of the Isle of Mull, by m. c. boulter and z. kvacek. 149 pp., 23 text-figs., 23 plates. Price £40 (U.S. $80). 43. (for 1990): Benthic palaeoecology of the Late Jurassic Kimmeridge Clay of England, by p. r. wignall, 74 pp., 50 text- figs. Price £30 (U.S. $60). Field Guides to Fossils and Other Publications These are available only from the Marketing Manager. Please add £1-00 (U.S. $2) per book for postage and packing plus £1-50 (U.S. $3) for airmail. Payments should be in Sterling or in U.S. dollars, with all exchange charges prepaid. Cheques should be made payable to the Palaeontological Association. 1. (1983): Fossil Plants of the London Clay, by M. E. collinson. 121 pp., 242 text-figs. Price £7-95 (U.S. $16) (Members £6 or U.S. $12). 2. (1987): Fossils of the Chalk, compiled by e. owen; edited by a. b. smith. 306 pp., 59 plates. Price £11-50 (U.S. $23) (Members £9-90 or U.S. $20). 3. (1988): Zechstein Reef fossils and their palaeoecology, by n. hollingworth and T. Pettigrew, iv + 75 pp. Price £4-95 (U.S. $10) (Members £3.75 or U.S. $7-50). 1982. Atlas of the Burgess Shale. Edited by s. conway morris. 31 pp., 24 plates. Price £20 (U.S. $40). 1985. Atlas of Invertebrate Macrofossils. Edited by j. w. Murray. Published by Longman in collaboration with the Palaeontological Association, xiii + 241 pp. Price £13-95. Available in the USA from Halsted Press at U.S. $24-95. © The Palaeontological Association, 1991 m Palaeontology VOLUME 34 ■ PART 2 ;(x 77 CONTENTS A spider and other arachnids from the Devonian of New York, and reinterpretations of Devonian Araneae P. A. SELDEN, W. A. SHEAR and P. M. BONAMO 241 Ordovician graptolites from the early Hunneberg of southern Scandinavia K. LINDHOLM ’283 Trilobites from the Ordovician of Portugal M. ROMANO 329 Cambroclaves and paracarinachitids, early skeletal problematica from the Lower Cambrian of South China S. CON WAY MORRIS and CHEN MENGE 357 The ostracoderm Phialaspis from the Lower Devonian of the Welsh Borderland and South Wales P. R. TARRANT 399 The rhynchonellide brachiopod Eocoelia from the Upper Llandovery of Ireland and Scotland E. N. DOYLE, A. N. HOEY and D. A. T. HARPER 439 The role of predation in the evolution of cementation in bivalves E. M. HARPER 455 Morphologic patterns of diversification: examples from trilobites M. FOOTE 461 The Paleorhinus biochron and the correlation of the non-marine Upper Triassic of Pangaea a. p. hunt and s. G. LUCAS 487 ; . Printed in Great Britain at the University Press , Cambridge ISSN 0031-0239 X X o z o z > W" £ > "'W*' 2 > z 07 z 07 * z to __ avaan ubraries Smithsonian institution NouniiiSNi nvinoshxiws ssiavaan librj 07 —r 07 — 07 1 iTmsgX o ^ vot>^ _ O p Ngiusgx o " q ITUTION 2 N01XnXllSNI~WlN0SHXWS S3 I BVB an LI B RAR I ES^SMITHSONIAN^INSTITUTlON^NOIXn. z 3 > z r~ z 3 z 7 3: 5 XX>St'}X >' 07 A‘ Z ^ •• Z 07 Z 07 ITUTION NOimillSNI NVINOSHXIWS S3 I avaan LIBRARIES SMITHSONIAN INSTITUTION N0I1DJ 07 X > 07 ~ 07 X 07 OJ CO „„ , ’*/ XI m xqmdcjz <3^ m X^ouT^X m X ^ m w — (n ± (_n \ X 07 ITUTION NOlinilJLSNI NVINQSHXIWS S3 laVBQ IT LIBRARIES SMITHSONIAN INSTITUTION NOIIO ... 07 ^ 07 Z 07 2: 07 SS 2 .*? Z^!S2^X 2 , 2; 2 /<5^lioXv Z O >5^.. m xfTO^oX ° 21 w/ — c r? zz u* avaan libraries Smithsonian institution NouniiiSNi nvinoshiiws S3iavaan libra Z 07 z -* 07 Z 07 Z , 2 X&R5X < VWlX 2 ^ ^ 07 A- Z W * z oS Z 07 ITUTION NOIXnillSNI NVINOSHXIINS S3iaVH9n LIBRARIES SMITHSONIAN INSTITUTION NOIXflJ 07 d > 07 X 07 X 07 o > TW" 2 ^ 2 > 3: (0*3 to ' 3 to 3 SMITHSONIAN INSTITUTION NQIimilSNI NVIN0SH1HNS S3IHVy8H LIBRARIES SMITHSONIAN_ IN: i to :r . to to 2: uj •■^p ^ •p’ *i j ^ i .✓*' NVINOSHIHNS^SB I U VH H IT LI B RAR I ES_SMITHSONIANt_INSTITUTION ^NOIlfUIJLSNI^NVINOSHlIWS^Sa I I I s => /*# 'M ? /£# o I##. *? 3 /*£ 5 73 m x^yos^ ~ m x^voswix ^ m _ CO “ CO — to SMITHSONIAN INSTITUTION NOIlfUIJLSNI NVIN0SH1IINS SHIdVilSIl LI B RAR I ES SMITHSONIAN IN „ Z > to 3 to 3 > ^ ~ ^ ? ,**gK < E jffi/A 5 ^ 5 NVIN0SH1ISNS S3 I M VH 8 n LIBRARIES SMITHSONIAN INSTITUTION NOIinillSNI_NVINOSHllWS SI to = CO ~ CO —r’t to to CO o 3 "wye. ■St .; ' c / v ' -j O “ ' O SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHillNS S3IH¥a8!l LIBRARIES SMITHSONIAN IN ■ fvmmr* § /^Sg&v 5 ® $ / * •< ro -n t— _ iJXSTxXv t— /c>WtwA -n /sc®?rA<\ .x *•../ 03 73 > 73 m v/ m vi xtQiiisgx rn ^Tpc^ ™ to to ± CO \ 5 to — to nvinoshiii/ms saiHvyen libraries Smithsonian institution NOliniliSNI nvinoshxiiais s: CO 3 .... to z ... CO z . 3 .* 5E o 3 ✓ 2 ^ > XOjjiiS^X *s > j 3 to * 3 to s- 3 to SMITHSONIAN INSTITUTION NQIinilXSNf NVIN0SH1IWS S31HVH8n LIBRARIES SMITHSONIAN _ I IN 3 ^ to — v CO ^ 5 ^ _ to W >jW I CO ^ E to £ CO _ SMITHSONIAN INSTITUTION NOliniliSNI NVIN0SH1IWS S3IHVy8n LI B RAR I ES SMITHSONIAN ir 3 7 . CO 3 CO 3 ■„ t/> 3 4^ i | | | /#% i a • %* ' ' 5 ^ I S Jo 3 CO 3 to ■** 3 CO NVIN0SH1IWS S3iyVH8ll LIBRARIES SMITHSONIAN INSTITUTION NOIiniliSNI_NVINOSHlllNS S to — to ” CO 21 to