NovlttatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET NEW YORK, N.Y. 10024 U.S.A. NUMBER 2699 JULY 11, 1980

ALEXANDER RITCHIE The Late Anaspid Genus Rhyncholepis from Oesel, , and Ringerike, Norway 0 AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2699, pp. 1-18, figs. 1-11 July 11, 1980

The Late Silurian Anaspid Genus Rhyncholepis from Oesel, Estonia, and Ringerike, Norway

ALEXANDER RITCHIE'

ABSTRACT Re-examination and further preparation of the Saarolepis (Robertson). There are sufficient dif- original material of the Silurian anaspid Saarole- ferences of specific importance between the forms pis oeselensis (Robertson) from the Ludlovian from Norway and from Oesel to justify the reten- Paadla Formation of Oesel, Estonia, has led to tion of Rhyncholepis oeselensis (Robertson) as a the discovery of an almost complete and un- separate species, the second of this genus to be crushed head of this species. In the structure of recognized. Details are also given of the discovery its cranial and ventral (gular) shields the Oesel of well-developed, paired, ventrolateral finfolds form closely resembles the well-known Rhyncho- in the Norwegian species, R. parvulus Kiaer, lepis parvulus Kiaer from the Upper Silurian of reinforcing earlier evidence that the presence of Ringerike, southern Norway, and it must now be such fins is the rule rather than the exception in assigned to that genus, with the suppression of anaspids.

INTRODUCTION The are a small, diverse order of 1899, 1905) Silurian, Scotland; Euphanerops Siluro- ostracoderm agnathans most (Woodward, 1900), Upper Devonian, Cana- closely related to the (cephalas- da; Ctenopleuron (Matthew, 1900), Silurian, pids) with which they are usually grouped to Canada; Pharyngolepis, Pterygolepis (Pter- form the , and to at least olepis), and Rhyncholepis (Kiaer, 1924), Up- one major division of the extant cyclo- per Silurian, Norway; Endeiolepis (Stensio, stomes-the Petromyzontida (lampreys). 1939), Upper Devonian, Canada; Saarolepis To date 10 genera of anaspids have been (Anaspis) (Robertson, 1941, 1945), Silurian, described: , (Traquair, Estonia; (White, 1946), Silurian,

'Curator of Palaeontology, The Australian Museum, 6-8 College Street, Sydney, New South Wales, Australia 2000.

Copyright ©) American Museum of Natural History 1980 ISSN 0003-0082 / Price $1.40 2 AMERICAN MUSEUM NOVITATES NO. 2699

Scotland. In addition to these there are oth- Stockholm and by T. S. Westoll, Newcastle, er, as yet undescribed, new anaspids re- England (Denison, 1952). covered from the Middle Silurian of Corn- In 1938 the collection was moved to Dart- wallis Island in the Canadian Arctic mouth College Museum, given the accession archipelago, from the Downtonian of Stone- number 38-71 and recatalogued. There it re- haven, eastern Scotland and from the mained until 1975 when the Dartmouth Col- Lower Old Red Sandstone (Lower Devoni- lege authorities decided to dispose of the sys- an) of Argyllshire, western Scotland. tematic collections and the Patten Collection Of the described genera the least known was presented in its entirety to the American is, unquestionably, Saarolepis oeselensis Museum of Natural History, New York, (Robertson) based on a handful of fragmen- where it has now been incorporated into the tary specimens collected by Patten in the palaeoichthyological collections of that insti- 1920s and 1930s. The present paper presents tution. new information gained during re-examina- The scarce anaspid specimens among the tion and subsequent preparation of the orig- Oesel remains were first noted by inal material of Saarolepis following which Robertson (1937) who later published a brief it became clear that the Estonian anaspid account and description of the very incom- material can now be assigned to Rhyncho- plete material (Robertson, 1941, p. 314, fig. lepis, the smallest of the three well-known 1). He assigned the form to a new genus and Norwegian Silurian anaspids from Ringerike species, Anaspis oeselensis, on the grounds near Oslo. The generic name Saarolepis that there was insufficient evidence to place must therefore be suppressed. It is proposed, it in an existing genus, and that this course however, that the Estonian form be retained would avoid confusion if further study of this as a separate species of Rhyncholepis-R. (or additional) material should result in an oeselensis (Robertson). identification with a known form. When it The small island of Oesel (or Saaremaa) was later realized that Anaspis was preoc- off the western coast of Estonia in the Baltic cupied Robertson (1945) proposed Saarole- has long been famous for the abundance, di- pis as an alternative, a name derived from versity, and preservation of its Silurian os- Saarema, the Estonian name for the island tracoderm faunas. One of the most active of Oesel. Saarolepis oeselensis (Robertson) early workers in this area was Professor Wil- is the only member of the group described liam Patten of Dartmouth College, New from the Silurian sediments of the Baltic to Hampshire, U.S.A. who made four separate date. visits to the island to organize and direct siz- I have been involved in a study of the able excavation projects in the vertebrate- Scottish and Norwegian anaspids for many bearing horizons; the most productive years years (Ritchie, 1960, 1964, 1968) and visited were 1930 and 1932, and a total of more than Estonia in 1976 to attend the Inter-Union Col- 3500 specimens was recovered and taken loquium on Middle Palaeozoic held in back to the United States for preparation and Tallinn. The collections of the Estonian study. The collection included superb mate- Academy of Sciences contain a few fragmen- rial of various cephalaspids (Tremataspis, tary specimens of Saarolepis but nothing Dartmuthia, Thyestes, Rotsikiillaspis, Wi- that added significantly to Robertson's ac- taaspis, Procephalaspis, etc.), thelodonts count. The return trip to Australia via the (Phlebolepis) and rare, fragmentary speci- United States provided a long-awaited op- mens of anaspids. Patten himself described portunity to examine some North American some of the fauna but after his death in 1932 ostracoderm collections and, during a brief the study and description of the remaining stay in New York, I was able to study the material was carried out by G. M. Robertson original Saarolepis specimens from the Pat- and later by R. H. Denison; some of the ma- ten Collection now housed in the American terial was investigated on loan by E. Jarvik, Museum of Natural History. It was observed 1980 RITCHIE: RHYNCHOLEPIS 3 that some of the specimens had not been Museum of Natural History numbers, as fol- completely prepared. With permission of Dr. lows: Bobb Schaeffer of the American Museum of Natural History, Patten's original Saarolepis An. 1 (a, b) = DCM 38-71-12648 = AMNH 6698 material was forwarded to Australia for fur- Holotype ther preparation and study. The results ex- An. 2 (a, b) = DCM 38-71-12649 = AMNH 6699 ceeded all expectations and clarified for the An. 3 (a, b) = DCM 38-71-12650 = AMNH 6777 first time, the taxonomic relationships of Complete "Saarolepis. " headshield An. 4 = DCM 38-71-12651 = AMNH 6771 ACKNOWLEDGMENTS An. 5 = DCM 38-71-12652 = AMNH 6772 An. 6 (a, b) = DCM 38-71-12653 = AMNH 6773 The present paper is dedicated to the late An. 7 = DCM 38-71-12654 = AMNH 6774 Professor Anatol Heintz, Oslo, who encour- An. 8 = DCM 38-71-12655 = AMNH 6775 aged and assisted me in my studies on the An. 9 = DCM 38-71-12656 = AMNH 6776 Scottish and Norwegian Silurian ostraco- derms and freely made available the Ringe- The most informative specimen of "Saa- rike material for study, loan, and prepara- rolepis" is not the holotype (An. 1) but tion. another specimen (An. 3) preserved in part Thanks are also extended to his daughter, and counterpart. The larger portion, a block Dr. Natascha Heintz, Palaeontological Mu- about 7 cm. square and 3-4 cm. thick, had seum, Oslo, for similar cooperation; to Dr. been struck to remove a smaller flake. This Bobb Schaeffer, American Museum of Nat- flake (fig. IC) displayed a narrow strip of an- ural History, New York, for permission to aspid dermal skeleton, seen in visceral view, borrow and prepare the Saarolepis material which extended from the dorsal to the ven- from the Patten Collection; to Dr. Elga tral margin. The exposed area lay between Mark-Kurik, Institute of Geology, Estonian the left orbit and the anterior end of the bran- Academy of Sciences, Tallinn, for the op- chial row. When the counterpart was exam- portunity to examine Saarolepis material in ined the writer observed that (a) the left der- the Estonian collections; to the Director and mal skeleton continued anteriorly into the Trustees of The Australian Museum, Syd- limestone matrix and (b) the opposite (right) ney, for financing the visit to Estonia and to side of the dermal skeleton was still present, the United States in 1976 which led to the buried deeper in the block and visible only discovery of the material described here. in cross-section. The 1973 Rhyncholepis discoveries were The two surfaces of the exoskeleton of the made possible by financial assistance from pre-branchial and head region were separat- the C.S.I.R.O. Science and Industry Devel- ed by 0.5-1.0 cm. of matrix and it seemed opment Fund, the Ian Potter Foundation, possible that as a result the entire head re- and the Bushell Trust. gion was preserved uncrushed in such a way that it could be prepared from both sides. DESCRIPTION Preparation of this specimen, carried out Over the years the various specimens of by the writer with the aid of a vibrotool and Oesel anaspids have been given a variety of pneumatic engraving pen, provided the first numbers which are listed here to avoid con- information on the dorsal, lateral, and ventral fusion. Robertson's preliminary numbers regions of the headshield of "Saarolepis." (An. 1-An. 9, based on the original generic The entire left surface of the head region name, Anaspis) were superseded in 1938 by was found to be preserved virtually intact Dartmouth College Museum numbers (DCM and in considerable relief (fig. 2A). The lime- 38-71-12648-12656). stone matrix in which the head was embed- They have now been assigned American ded was excavated to a depth of about 1 cm. 4 AMERICAN MUSEUM NOVITATES NO. 2699

10

FIG. 1. Rhyncholepis (Saarolepis) oeselensis (Robertson), Upper Silurian (Middle Ludlow) Oesel, Estonia, U.S.S.R. A. Holotype, AMNH 6698 (formerly DCM 38-71-12648). Left flank of trunk posterior to orbit, seen in visceral view. Pineal plate (p) visible (top right), postbranchial and pectoral plates (bottom center). B. Counterpart of holotype showing considerable portion of posterior cranial roof, right pectoral plate, visceral surface of right trunk scales, ventral scales. C. Left lateral surface of head and anterior portion of trunk, AMNH 6777b, with anterior portion rebuilt from flakes removed from counterpart, AMNH 6777a during preparation. Scale in mm. 1980 RITCHIE: RHYNCHOLEPIS 5 all round, leaving the anaspid head on a nar- in detail. The rostral area is covered with row pillar of matrix. The head was then several pairs of largish plates but it is difficult flaked off the main block in the hope that to determine with certainty whether the an- much of the still-buried right lateral surface teriormost plate is single or paired; the latter could be recovered both as an external and interpretation is depicted here (fig. 7B). an internal mould. The approach was suc- One of the most distinctive plates on the cessful, the anaspid head was detached safe- anaspid head, the large, Y-shaped pineal ly but it was discovered that the right surface plate, is present in the midline dorsal and had been crushed inward during burial (fig. slightly posterior to the left orbit which is 2D) and was less complete than the left side. completely preserved. The deep, rounded Despite the postmortem damage this beau- notch which formerly housed the pineal or- tifully preserved specimen provides suffi- gan is open anteriorly, suggesting that the cient information to enable a reconstruction anterior portion of the pineal plate which of the entire head region (fig. 7A-C). Al- normally separates the pineal aperture from though much of the original dermal skeleton the much larger, single, naso-hypophysial fo- was preserved as a translucent, deep brown ramen has become detached and lost (figs. 3, substance most of it had split through the 4). middle or flaked off the specimen in such a The pattern of largish plates surrounding way that it could not be recovered. It was the orbit and covering the postorbital region therefore decided to remove all remains of can be restored by combining the informa- the exoskeleton from the various parts of this tion from the external (fig. lA-C) and inter- specimen to obtain detailed impressions of nal (figs. 3, 4) impressions. The differences both the dermal and visceral surfaces. This in shape and proportions from the plates cov- seemed the course likely to yield the maxi- ering the same areas in Rhyncholepis par- mum amount of information. A latex cast of vulus are considered here to be mainly of the dorsal, left lateral, and ventral surface specific importance. was prepared, unrolled and attached to flat The ventral (gular) region of the exoskel- card for study and photography. eton is more completely preserved than the The most striking discovery was that the dorsal surface, with only a small area near entire anterior ventral surface-the gular re- the anterior end damaged as the result of a gion-of the head was covered with a dis- fracture (fig. 2C, figs. 3, 5). The plate pattern tinctive arrangement of very large plates and can be restored with confidence (fig. 7C). scales (fig. 2A, C, D, figs. 3, 5) quite unlike The central plate, a five-sided unit, is pointed the pattern seen in the other armored anas- posteriorly and tapers anteriorly to a narrow, pids such as Birkenia (Heintz, 1958, fig. 4b, blunt end; the anterolateral margins are con- c), Pharyngolepis (Kiaer, 1924, fig. 37; Rit- cave. In this latter feature and in the rela- chie, 1964, fig. la-c) andPterygolepis (Kiaer, tively smaller size of the median gular plate 1924, fig. 35). However, the gular pattern re- Rhyncholepis oeselensis differs from R. par- vealed in "Saarolepis" is so similar to that vulus (Kiaer 1924, figs. 24, 25, 39; this paper present in Rhyncholepis parvulus Kiaer fig. 6). The median gular plate is flanked an- (1924, figs. 24, 25, 36; this paper fig. 6), the teriorly, laterally, and posteriorly, respec- smallest and in many ways the most special- tively, by three pairs of plates of approxi- ized of the three anaspid genera in the Silu- mately the same, or slightly smaller, size. rian fauna from Ringerike, that it must now Dorsolaterally, the plates become longer and be considered to belong to that genus. narrower merging into the smaller, obliquely The left half of the dorsal surface of the directed scales covering the cheek area. headshield is virtually complete but the right The gular region includes several other half is mostly missing (fig. 2A, B, figs. 3, 4). median plates which should be noted. The The basic pattern is not unlike that of Rhyn- ventral shield is terminated posteriorly by a cholepis parvulus Kiaer but with differences large, transversely oriented, semicircular 6 AMERICAN MUSEUM NOVITATES NO. 2699

FIG. 2. Rhyncholepis (Saarolepis) oeselensis (Robertson), Upper Silurian (Middle Ludlow), Oesel, Estonia, U.S.S.R. AMNH 6777a. Four views of isolated head, preserved in three dimensions. A. Left lateral view. B. Dorsal view (anterior to top). C. Ventral view (anterior to top). D. Right lateral view. plate with a convex posterior margin. This entire ventral surface, however, is a relative- is followed posteriorly by several similar, but ly small, median plate which forms the an- much narrower, median plates or scales. terior ventral margin of the oral opening. One of the most interesting features of the This plate, exposed as a mould of the vis- 1980 RITCHIE: RHYNCHOLEPIS 7

rp

FIG. 3. Rhyncholepis (Saarolepis) oeselensis (Robertson). Dermal skeleton of head region of AMNH 6777a (fig. 2A-D), from latex cast, split along right lateral margin and unfolded in one plane. Cast shows visceral surface of cranial roof (left half only; cf. fig. 4), left cheek scale covering, and ventral (gular) shield (cf. fig. 5). mp = mandibular plate; p = pineal plate; rp = left and right rostral plates. Scale in mm. 8 AMERICAN MUSEUM NOVITATES NO. 2699

FIG. 4. Rhyncholepis (Saarolepis) oeselensis (Robertson). Cranial shield (left half), AMNF 6777a (fig. 2A, B), visceral surface as seen in latex cast (fig. 3). Portion of pineal plate anterior to pineal foramen has been lost and the right rostral plate has also been dislodged relative to the left rostral plate which remains in situ. Scale in mm. ceral surface, displays an intriguing arrange- the plate to an external position on the ven- ment of anteriorly converging grooves and tral oral margin (Ritchie, 1964, pp. 9-13, figs. ridges (fig. 3, fig. 5, mp, fig. 7C) and imme- la, b, 2a, b, pl. VI, figs. 1, 2). diately brings to mind the strange, well-de- The anterior median ventral plate of Rhyn- veloped T-shaped "mandibular plate" of cholepis oeselensis (Robertson) is almost Pharyngolepis oblongus Kiaer (1924, p. 48, certainly homologous to the large, mandib- figs. 22, 37, pl. X). Although this was later ular plate of Pharyngolepis oblongus Kiaer reinterpreted by Stensio (1958, p. 226, figs. and both probably served to reinforce the 126, 127; 1964, figs. 27, 28) as part of a rasp- lower buccal margin. The interpretation of ing tongue apparatus, I concluded, after ex- the buccal region and the dermal skeleton of amining the type Pharyngolepis material in Rhyncholepis parvulus Kiaer has posed a the Oslo collections, that Kiaer's original in- problem since it was first described but, after terpretation was the correct one and restored careful examination of all the original Rhyn- 1980 RITCHIE: RHYNCHOLEPIS 9

FIG. 5. Rhyncholepis (Saarolepis) oeselensis (Robertson). Ventral (gular) plates of head seen in visceral view (cf. fig. 2A, C, D ). Latex cast whitened with ammonium chloride, from AMNH 6777a (cf. fig. 3). Scale in mm. 10 AMERICAN MUSEUM NOVITATES NO. 2699 cholepis material in the Oslo and other mu- Norwegian R. parvulus (fig. 9). From the seum collections, I am convinced that, with position and proportions of the branchial row a few minor modifications, Kiaer's restora- and the associated plates it can be estimated tion is quite accurate (1924, fig. 36; this pa- that the original number of openings was per, fig. 10). Unlike most of the other ar- probably around eight, the number present mored anaspids, such as Birkenia (Heintz, in R. parvulus (and in Birkenia), and was 1958, fig. 4) and Pharyngolepis (Ritchie, certainly fewer than in Pterygolepis (10 ap- 1964, fig. la-c) which appear to have had a ertures) or in Pharyngolepis (12-15). vertical, oval buccal opening, Rhyncholepis The associated dermal skeleton is of the seems to present an example of an agnathan standard anaspid type, a long, narrow, L- which partly overcame the absence of true shaped postbranchial plate and a large, tri- jaws by a specialized development of the gu- angular pectoral plate (figs. IA, B, 7A); in lar and mandibular skeleton. The new ma- the holotype the latter has been further un- terial from Oesel confirms and enlarges our covered by me to expose the full extent of understanding of this region in Rhynchole- the long, needle-sharp pectoral spine. From pis. the evidence of Pharyngolepis oblongus The dorsal and ventral shields in Rhyncho- Kiaer and P. heintzi Ritchie (Ritchie, 1964, lepis oeselensis are separated laterally by an fig. la-c) it is now known that, in anaspids, area of obliquely directed, narrow, lancet- this pectoral spine originally formed the an- shaped scales covering the cheek region and terior margin of a paired ventrolateral finfold merging posteriorly into the scale covering which extended along part or all of the ven- of the branchial region and trunk (figs. IC, trolateral ridge between the branchial row 3). This pre-branchial scale-covered area is and the anus. considerably larger in Rhyncholepis oeselen- Evidence for such a paired fin in Rhyncho- sis than in R. parvulus but the basic pattern lepis was discovered by the writer during is similar. The lancet scales extend below, studies on the original Ringerike material in and considerably anterior to, the orbit, ter- the Palaeontologisk Museum, Oslo, during minating some 7 mm. short of the rostral and 1966 and 1973 and is described below. mandibular margins, leaving a deep unar- The remainder of the trunk and caudal mored notch in the lateral buccal margin. skeleton of Rhyncholepis oeselensis (Rob- Whilst this area may well have been bridged ertson) remains unknown. Only one of the in life by soft tissues not preserved in the dorsal ridge scales has been observed in the fossil it is also possible that the lateral buccal available material (Robertson, 1941, 316) but margin was indented deeply and that the ven- it was apparently similar to those present in tral shield and the mandibular plate were suf- the Norwegian species Pterygolepis. It is ficiently mobile to be brought against the unlikely that the caudal fin differed signifi- ventral rostral margin with a nipping or biting cantly from R. parvulus but it would be of action. interest to discover whether the anal fin in The lateral scale covering of Rhyncholepis R. oeselensis was quite as specialized as the oeselensis (Robertson) is developed basical- long, low, spinous structure observed in R. ly as in the other scaled anaspids and can be parvulus (figs. 9, 10). reconstructed from information provided by In the light of the evidence provided by the the holotype (AMNH 6698, fig. IA, B), the recently prepared head of the Oesel Silurian detached flake (AMNH 6777b, fig. IC) as- anaspid previously assigned to Saarolepis sociated with the head region and some of oeselensis (Robertson) there seems little the other fragments in the Patten collection. doubt that this form must now be consid- Two specimens have the branchial region ered to be a species ofthe better known genus preserved but with the surrounding scale Rhyncholepis from the Upper Silurian of covering disturbed and the branchial aper- southern Norway, which has precedence tures are not as clearly displayed as in the (Kiaer, 1924). In several respects (the longer 1980 RITCHIE: RHYNCHOLEPIS 11

FIG. 6. Rhyncholepis parvulus Kiaer, Late Silurian (Ludlow), Rudstangen, Ringerike, Norway. Palaeontological Museum, Oslo EO169. Silicone rubber cast ofgular plates and left cheek scale covering, whitened with ammonium chloride. Scale in mm.

pre-orbital region, the pattern of the cranial Silurian anaspids from the quarry at Rud- roof and the gular region, the greater de- stangen, Ringerike described by Kiaer [1924; velopment of lancet scales of the cheek area) Pharyngolepis, Pterygolepis (Pterolepis), the Baltic form is sufficiently distinct from Rhyncholepis] evidence for the presence of Rhyncholepis parvulus Kiaer to justify con- paired ventrolateral finfolds was discovered tinued separation in a second species of the in several specimens of Pharyngolepis, the same genus, R. oeselensis (Robertson). largest of the three Ringerike genera (Rit- chie, 1964, pp. 4-9, fig. la-c, pls. I-V). PAIRED FINS OF RHYNCHOLEPIS Ventrolateral finfolds had earlier been de- In 1962 when I first visited Oslo to carry scribed in the late Devonian Endeiolepis out comparative studies on the Norwegian from Scaumenac Bay, Quebec, Canada by 12 AMERICAN MUSEUM NOVITATES NO. 2699

A.

FIG. 7. Rhyncholepis (Saarolepis) oeselensis (Robertson), Late Silurian, Oesel, Estonia, U.S.S.R.; A. Reconstruction of head and anterior trunk, based largely on AMNH 6777a (fig. 2A), counterpart AMNH 6777b (fig. IC) and part and counterpart of the holotype, AMNH 6698a, b (fig. IA, B), branchial apertures restored after R. parvulus Kiaer (figs. 9, 10). B. Cranial shield, flattened; composite recon- struction from AMNH 6698a, b, AMNH 6777a, b. C. Ventral (gular) shield, flattened; reconstructed after AMNH 6777a, b (figs. 3, 5). Abbreviations: ba = branchial aperture; mp = mandibular plate; nhf = nasohypophysial foramen; o = orbit; p = pineal plate; pbp = postbranchial plate; pf = pineal foramen; pp = pectoral plate; ps = pectoral spine; rp = rostral plate. 1980 RITCHIE: RHYNCHOLEPIS 13

Stensio (1939, fig. 7, P1. 1, figs. 1, 2) but this Two other specimens of Rhyncholepis was the only anaspid known to possess such parvulus Kiaer in the Oslo collections structures. Kiaer had noted a small row of (EO128 and E0168) also display faint traces needle-sharp projecting spines along the ven- of the delicate finfold behind the pectoral trolateral ridge in Pterolepis (now Pterygo- spine, confirming the presence, probable lepis) nitidus (1924, p. 71, fig. 31b). Despite shape and proportions of the lateral fin of careful examination I was unable, in 1962, to Rhyncholepis as shown here (fig. 10) in a detect any evidence of fins similar to those modified version of Kiaer's (1924) restora- of Pharyngolepis in the associated material tion. of Pterygolepis or Rhyncholepis which con- It has now become obvious that the pres- sist of hundreds of specimens in various ence of well-developed, paired ventrolateral states of completeness and preservation. finfolds was the rule, rather than the excep- On two subsequent visits to the Palaeon- tion, among the anaspids, and the presence tological Museum, Oslo, in 1966 and 1973, of such features makes anaspid swimming re-examination of the Rhyncholepis material habits somewhat easier to understand than did reveal several specimens which, after had originally been the case. For a long time some preparation, were discovered to pos- after their first discovery, the swimming abil- sess faint, but unmistakable, remains of a ities of anaspids were the subject of contro- delicate fin membrane. In each case this lay versy (Heintz, 1935, Parrington, 1958, Wes- immediately behind the projecting pectoral toll, 1958, Nursall, 1962) because of the spine of the large, triangular pectoral plate, unusual combination of a hypocercal caudal the same relationship already known to be fin with the notochord extending into the present in Pharyngolepis oblongus Kiaer ventral (hypochordal) lobe and the apparent and Pharyngolepis heintzi Ritchie (Ritchie, absence of paired fins. Following the discov- 1964, fig. la-c). ery of paired finfolds in the geologically late The specimen of Rhyncholepis parvulus Endeiolepis aneri Stensio (1939) and, more Kiaer which best displays the ventrolateral recently, in Pharyngolepis oblongus Kiaer finfold is a small, almost complete individual and Pharyngolepis heintzi Ritchie (1964) it (Pal. Mus. Oslo, EO140, fig. 8A, B) seen became somewhat easier to understand the from the ventral aspect with the gular region swimming mechanics. However, it was mis- of the head twisted sideways. The ventral takenly assumed that the finfold in Pharyn- scales of the trunk are clearly visible, con- golepis increased positive pitch and assisted verging anteriorly and the long, sharp pos- the caudal fin which had a similar effect. terior spines of the pectoral plates project Hopson (1974, pp. 83-93), in a closely rea- clear of the trunk on both sides. The pectoral soned analysis, has reviewed the functional spine on the right margin (as the specimen significance of the hypocercal caudal fin in lies inverted in the rock) is the more fully the active, fusiform anaspids and has cleared extended. In the area immediately posterior up earlier misconceptions about the hypo- to this spine a faint but definite stain can be cercal tail-lateral finfold combination. He distinguished, most clearly when the fossil is showed that equilibrium in the pitching plane immersed in alcohol (as in fig. 8A). This stain could only be achieved by increasing nega- thus represents a delicate, membranous fin, tive pitch, and that this was what the finfold, some 2-3 mm. wide extending along the ven- because of its position, was adapted to do. tral margin for about 5 mm. and terminating For a fuller understanding one must take into short of the anal fin. The outer margin is account the relative positions of the center strongly convex. There are traces of a simi- of gravity, center of buoyancy, and center of lar, but less completely preserved, fin mem- lift (from the finfolds and from that portion brane behind the other pectoral spine on the of the trunk between them). same individual. In a heavier-than-water with a hetero- 14 AMERICAN MUSEUM NOVITATES NO. 2699

fr IL

ei B

FIG. 8. Rhyncholepis parvulus Kiaer, Late Silurian (Lower Ludlow), Rudstangen, Ringerike, Nor- way. Palaeontological Museum, Oslo, E0140, smallish individual seen mainly in ventral view, with remains of delicate fin membranes preserved immediately posterior to the posterolaterally projecting pectoral spines. A. Specimen photographed under alcohol; B. Interpretation of specimen shown in figure 8A. Abbreviations: af = anal fin scales; el = epichordal lobe (of caudal fin); fm = fin membrane; gp = gular plates; lps, rps = left and right pectoral spines; vs = ventral scales (of trunk). 1980 RITCHIE: RHYNCHOLEPIS 15

FIG. 9. Rhyncholepis parvulus Kiaer, Late Silurian (Lower Ludlow), Rudstangen, Ringerike, Nor- way. Palaeontological Museum, Oslo E044, figured by Kiaer (1924, pl. XIV, fig. 1). Specimen photo- graphed immersed in alcohol. Note the extension of the anterior cheek scale covering between the cranial shield and the lateral gular plates immediately ventral to the orbit, as in Rhyncholepis oeselensis (Robertson) (cf. figs. 2A, 7A), and horizontal orientation of preanal plate. Scale in mm.

cercal caudal fin (eg., a shark; Hopson, 1974, P. oblongus. The net result was to decrease fig. IA) these three centers are quite widely the dynamic stability of the fish but simul- separated and the paired fins lie well anterior taneously increase its maneuverability. The to the centers of gravity and buoyancy. In lateral fins, lying directly below the center of Pharyngolepis oblongus the situation is gravity (as in some higher , e.g., go- quite different (1974, fig. 1B) with all three bies) would have produced very small turn- centers placed close together. ing movements and slight adjustments of the The finfold condition now known to exist caudal fin would be relatively more effective. in Rhyncholepis parvulus (fig. 10) and, al- The evidence therefore suggests that most certainly also present in R. oeselensis, Rhyncholepis, like Pharyngolepis and most is closer to that in Pharyngolepis heintzi Rit- of the other anaspids, was an active, efficient chie (1964, fig. lc) than to that in P. oblon- swimmer able to operate and feed anywhere gus. The shorter finfold corresponds only to between the surface and the bottom, in con- the anterior portion of the fin in P. oblongus trast to the majority of the more heavily ar- but is relatively wider and subtriangular in mored heterostracans and cephalaspids shape. The center of lift force in Rhyncho- which appear to have been more suited to a lepis (and in P. heintzi) lay closer to the cen- bottom-dwelling mode of life. ter of gravity and thus the lift available for STRATIGRAPHIC SIGNIFICANCE OF countering any downward thrust by the tail RHYNCHOLEPIS was virtually nil (fig. 11). Hopson suggested that the tail of P. heintzi (which is not The recognition of the anaspid genus known) was probably nearly or actually iso- Rhyncholepis in the Oesel ostracoderm fau- batic (producing level movement) rather than nas provides a useful link between the late epibatic (producing upward movement) as in Silurian vertebrate faunas of Estonia and 16 AMERICAN MUSEUM NOVITATES NO. 2699

FIG. 10. Rhyncholepis parvulus Kiaer, Late Silurian (Lower Ludlow) Rudstangen, Ringerike, Nor- way. Reconstruction, modified after Kiaer (1924, fig. 36). Main differences are in the separation of the cranial and gular shields by the anterior insertion of lateral scales of the cheek, the addition of a paired "pectoral" finfold supported anteriorly by the long, slender flattened pectoral spine, and in the reori- entation of the preanal ridge plate. The anteriorly converging ventral scales of the trunk terminate anterior to the preanal ridge plate. southern Norway. Mark-Kurik (1969, pp. rema) which was originally thought to ex- 145-152) has reviewed the distribution of Si- pose the uppermost part of the (now lurian in Estonia whilst Heintz Rootsikiila) Stage (K1) and the basal part of (1969, pp. 23-25) has reviewed the faunal the overlying Paadla Stage (K2); this was and stratigraphic relationships of the Eston- thought to mark the boundary between the ian, Norwegian and Scottish Silurian ostra- Lower and Middle Ludlovian. Later inves- coderm-eurypterid assemblages. tigations by Aaloe (1963) indicated that the Patten's original material of Rhyncholepis ostracoderm fauna characterized by Trema- oeselensis (Robertson) came from one very taspis mammillata Patten (and also including productive quarry at Himmiste (cf. Mark- Rhyncholepis oeselensis) came from dolo- Kurik, 1969, fig. 1) in western Oesel (or Saa- mite layers higher up in the Paadla Forma-

'w FIG. 11. Rhyncholepis parvulus Kiaer. Suggested vertical forces acting on the anaspid R. parvulus swimming horizontally (compare with Hopson, 1974, fig. IA). A, represents upthrust of water displaced, acting through the center of buoyancy; B, represents lift produced by snout, pectoral fins and area between them; C, represents lift produced by tail; W, repre- sents fish's weight acting through center of gravity. 1980 RITCHIE: RHYNCHOLEPIS 17 tion of probable Middle Ludlovian age. As- Russian.) Eesti NSV tead. Akad. Geol. sociated with Rhyncholepis at Himmiste are Inst. Uutim., vol. 13, pp. 83-90. abundant thelodont scales (Thelodus laevis, Denison, R. H. T. schmidti, and Phlebolepis elegans), lo- 1952. Types and figured specimens of fossil fishes in the Patten Collection Dart- cally abundant articulated specimens of mouth College Museum, Hanover, New Phlebolepis, and various genera and species Hampshire. Amer. Midland Nat., vol. of cephalaspids (Mark-Kurik, 1969, fig. 3). 48, pp. 161-164. The Himmiste Beds (K2H) are thought to Heintz, A. represent sediments deposited in a semi- 1935. How the fishes learned to swim. Report closed basin, temporarily freshened due to Smithsonian Inst. for 1934, pp. 223-245. the discharge of a river (Obruchev, and Kar- 1958. The head of the anaspid Birkenia ele- atjute-Talimaa, 1967, pp. 6-7) whilst the pre- gans Traquair. In Westoll, T. S. (ed.), ceding and following units, the Beds Studies on fossil vertebrates. London, (K2S) and Beds (K2U), respective- The Athlone Press, pp. 71-85. 1969. New agnaths from Ringerike Sand- ly, are generally represented by normal shelf stone. Skrifter norsk Vidensk Akad. deposits (Mark-Kurik, 1969, fig. 2). Oslo, Mat.-Naturv. Kl., N.S. 26, pp. 1- Isolated anaspid scales attributable to 28. Rhyncholepis (=Saarolepis) are now known Hopson, J. A. to occur well below the Himmiste Beds and 1974. The functional significance of the hy- are locally abundant at Vesiku (K1Vs) in the pocercal tail and lateral fin fold of an- Rootsikula Stage near the base of the Lud- aspid ostracoderms. Fieldiana, Geol., lovian. vol. 33, pp. 83-93. The richest vertebrate assemblages in the Kiaer, J. late Silurian rocks of southern Norway come 1924. The Downtonian fauna of Norway, I. Anaspida. Skr. Vidensk Selsk. Christi- from the Ringerike Sandstone at Rudstangen ania Mat.-Naturv. Kl. no. 6, pp. 5-136. (Kiaer, 1924) and at Nes, 1.5 km. to the Mark-Kurik, E. northeast (cf. Heintz, 1969, fig. 1 for map). 1969. Distribution of vertebrates in the Silu- Both horizons apparently lie within Zone 10 rian of Estonia. Lethaia, vol. 2, pp. 145- of Kiaer, the lower part of which is now be- 152. lieved to be of Lower Ludlovian age. Matthew, G. R. Abundant well-preserved anaspid remains 1907. A new genus and a new species of Si- have only been found at the Rudstangen site, lurian fish. Trans. Royal Soc. Can., ser. not at Nes, and it was the former site which 3, vol. 1, pp. 7-11. produced Pharyngolepis, Pterygolepis and, Moy-Thomas, J. A., and R. S. Miles. 1971. Palaeozoic Fishes. London, Chapman of course, Rhyncholepis parvulus Kiaer. The and Hall. occurrence of species of Rhyncholepis at Nursall, J. R. Rudstangen, Norway and Himmiste, Oesel 1962. Swimming and the origin of paired ap- need not indicate that the two faunas are ex- pendages. Amer. Zool., vol. 2 (2), pp. actly contemporary, especially since the lat- 191-203. est findings from Oesel indicate a Lower to Obruchev, D. V., and V. Karatjute-Talimaa. Middle Ludlovian range for this little anaspid 1967. Vertebrate faunas and correlation of the genus, but the new findings do strengthen the Ludlovian-Lower Devonian in eastern evidence for an early Ludlovian age for the Europe. Jour. Linn. Soc. (Zool.), vol. Rudstangen ostracoderm fauna. 47, no. 311, pp. 5-14. Parrington, F. R. 1958. On the nature of the Anaspida. In Wes- LITERATURE CITED toll, T. S. (ed.), Studies on fossil ver- Aaloe, A. 0. tebrates. London, The Athlone Press, 1963. On the stratigraphic position and con- pp. 108-128. dition of sedimentation (of strata) con- Ritchie, A. taining Tremataspis mammillata. (In 1960. A new interpretation of Jamoytius ker- 18 AMERICAN MUSEUM NOVITATES NO. 2699

woodi White. Nature, vol. 188, pp. 647- 1964. Les cyclostomes fossiles ou ostraco- 649. dermes. In Piveteau, J. (ed.), Traite de 1964. New evidence on the morphology of the Paleontologie, tome 4, vol. 1, Paris, Norwegian Anaspida. Skrifter norsk Masson. Vidensk Akad. Oslo, Mat.-Naturv. KI. Traquair, R. H. N.S. 14, pp. 1-35. 1899. Report of fossil fishes collected by the 1968. New evidence on Jamoytius kerwoodi Geological Survey of Scotland in the White, an important ostracoderm from Silurian rocks of the south of Scotland. the Silurian of Lanarkshire, Scotland. Trans. Royal. Soc. Edinburgh, vol. 39, Palaeontology, vol. 11, pp. 21-39. pp. 595-602. Robertson, G. M. 1905. Supplementary report on fossil fishes 1937. Anaspida from the Ludlow of Oesel. collected by the Geological Survey of Jour. Palaeont., vol. 11, p. 256. Scotland. Trans. Royal Soc. Edinburgh, 1941. The ostracoderm order Anaspida, with vol. 40, pp. 879-888. description of some Upper Silurian ma- Westoll, T. S. terial. Trans. Kans. Acad. Sci., vol. 44, 1958. The lateral fin-fold theory and the pec- pp. 314-317. toral fins of ostracoderms and early fish- 1945. Saarolepis, new name for Anaspis Rob- es. In Westoll, T. S. (ed.), Studies on ertson 1941. Jour. Palaeont., vol. 19, p. fossil vertebrates. London. Athlone 77. Press, pp. 181-211. Stensio, E. A. White, E. I. 1939. A new anaspid from the Upper Devo- 1946. Jamoytius kerwoodi, a new nian of Scaumenac Bay in Canada, with from the Silurian of Lanarkshire. Geol. remarks on the other anaspids. K. sven- Mag., vol. 83, pp. 89-97. ska Vetensk Akad. Handl. ser. 3, vol. Woodward, A. S. 18, pp. 3-25. 1900. On a new ostracoderm (Euphanerops 1958. Les cyclostomes fossiles ou ostraco- longaevus) from the Upper Devonian of dermes. In Grasse, R.-P. (ed.), Traite Scaumenac Bay, Quebec, Canada. Ann. de Zoologie, vol. 13, pp. 173-425. Paris, Mag. Nat. Hist., ser. 7, vol. 5, pp. 416- Masson. 419.

a a