Americant MUSEUM PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

Home , Allothrissops, Dapedium, Furo (fish), Leptolepidae, Leptolepides, Moythomasia

NoqvitatesAMERICANt MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2796, pp. 1-86, figs. 1-39, tables 1-4 November 13, 1984

Jurassic Fishes from the Western United States, With Comments on Jurassic Fish Distribution

BOBB SCHAEFFER' AND COLIN PATTERSON2

CONTENTS

Abstract ...... 2 Introduction ...... 2 Methods and Materials ...... 3 Acknowledgments ...... 3 Abbreviations ...... 4 Geologic Occurrence ...... 5 Systematics ...... 13 Chondrichthyes ...... 13 Elasmobranchii ...... 13 Hybodontidae ...... 13 Hybodus sp...... 13 Holocephali ...... 14 Chimaeridae ...... 14 Ischyodus sp...... 14 Osteichthyes ...... 14 Actinopterygii ...... 14 Neopterygii ...... 14 Halecostomi Incertae Sedis ...... 14 Hulettia, New Genus ...... 14 Hulettia americana (Eastman) ...... 14 Halecostomi ...... 37 Semionotidae ...... 37 Lepidotes sp...... 37 Halecomorphi ...... 38 Caturidae ...... 38 Caturus dartoni (Eastman) ...... 38 'Curator Emeritus, Department of Vertebrate Paleontology, American Museum of Natural History. 2 Research Associate, Department ofIchthyology, American Museum ofNatural History; Senior Principal Scientific Officer, Department of Palaeontology, British Museum (Natural History).

Teleostei ...... 40 Ichthyodectiformes Incertae Sedis ...... 40 Occithrissops, New Genus ...... 40 Occithrissops willsoni, New Species ...... 40 Teleostei Incertae Sedis ...... 51 Todiltia, New Genus ...... 51 Todiltia schoewei (Dunkle) ...... 51 Discussion and Summary ...... 63 Appendix: The Distribution of Jurassic Fishes ...... 66 Literature Cited ...... 80

ABSTRACT Seven genera of fishes from the Jurassic (upper thyodectiform teleost, but not assignable to sub- Bathonian-Callovian) Sundance and Wanakah order. Todiltia schoewei (Dunkle), new genus from formations ofwestern United States are described the Wanakah, is a teleost compared with Ascala- and their relationships discussed together with a bos and Leptolepis, but its affinities remain un- review of their stratigraphic occurrence. The as- known. semblage includes isolated teeth ofthe chondrich- The problem of incertae sedis genera and species thyans Hybodus sp. and Ischyodus sp., in part from (e.g., the monotypic Hulettia) is discussed in re- unrecorded Sundance localities near Hulett, Wy- gard to identification and relationships ofJurassic oming. Most common is the generalized neopte- fishes from other parts of the world. Included ta- rygian Hulettia americana (Eastman), new genus bles and paleogeographic maps show temporal and of unknown relationship, which occurs in the spatial distribution of these fishes, but variable Sundance of Montana and Wyoming, and in the preservation, inadequate description, and super- Wanakah ofColorado (including the Pony Express ficial systematic analysis usually preclude detailed Limestone) and New Mexico (Todilto Limestone). comparisons of Jurassic taxa from the literature. Lepidotes sp. and Caturus dartoni (Eastman), which It is probable, however, that most Jurassic fish belong to a monophyletic species group within the assemblages, like those ofthe Sundance-Wanakah, genus Caturus, are both present in the Sundance are mixtures ofform genera and monotypic genera and the Wanakah. Occithrissops willsoni, new ge- whose relationships are imprecisely known. nus, new species from the Sundance, is an ich- INTRODUCTION Our knowledge ofdiversity among various has yielded a fish assemblage that approaches groups of fishes that existed during the Ju- the diversity of the European ones (Gregory, rassic is mostly derived from a few major 1923; D. H. Dunkle, personal commun.). collecting areas in England, France, and Ger- Unfortunately, this fauna is mostly unstud- many. Specimens obtained from these local- ied. More recently Arratia (1982), and others ities have been studied longer and more in- have reported on Oxfordian fishes from tensively than those from rocks ofsimilar age northern Chile. In regard to North America, in other parts of the world. But in order to fishes have long been known from the Sun- obtain a broader understanding of both the dance Formation in South Dakota (Eastman, elasmobranch and osteichthyan fishes during 1899b) and from the Todilto Limestone in this significant interval in their history, it is New Mexico (Koerner, 1930). The primary obviously desirable to search for localities purpose of the present paper is to describe outside of Europe. The European taxa will, and consider the relationships of the fishes of course, continue to play a major role in from the Sundance Formation in northeast- working out the subtleties ofrelationship and ern Wyoming and from the Wanakah For- diversification, but well-preserved specimens mation (including the Todilto), and equiva- from elsewhere should provide additional in- lent units, in northern New Mexico and sights into phylogeny and paleobiogeogra- Colorado. With assistance from members of phy. the United States Geological Survey, an effort For the Western Hemisphere, only the Late has also been made to clarify the age and Oxfordian Jagus Formation in western Cuba somewhat enigmatic stratigraphic relation- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 3 ships ofthe rock units containing these fishes. Mr. R. J. Greffenius of the U.S. Forest Ser- A second objective is to summarize the avail- vice provided information about the fish lo- able information on marine and non-marine cality in the Pony Express Limestone in Pie- Jurassic fish assemblages from various parts dra River Canyon, Colorado, and Dr. G. D. of the world in relation to their distribution, Johnson for the locality near Hot Springs, composition and to current problems ofiden- South Dakota. tification and relationship. The regional geology and stratigraphic relationships of the Sundance Formation and METHODS AND MATERIALS the Todilto Limestone and Pony Express Limestone members of the Wanakah For- We have not been successful in making to- from the mation have been clarified through the gen- tal acid preparations of specimens erous assistance of Drs. George Pipiringos American Jurassic localities mentioned in this (retired) and Robert B. O'Sullivan ofthe U.S. report. However, many morphological de- Geological Survey. Additional information tails have been enhanced by local application has been supplied by Drs. Jennie L. Ridgley offormic or acetic acid, by Smooth-On® peels and J. Platt Bradbury, also ofthe Survey, and obtained from natural or prepared negative by Dr. W. F. Tanner. Drs. Ralph Imlay, An- specimens and by the use of alizarin to dis- H. Callomon, Hans-Pe- the good to thony Hallam, John tinguish bone from matrix. Also, ter Schultze, and John Utgaard provided excellent preservation of various neopteryg- opinions on problems ofcorrelation and dat- ian specimens has permitted interpretation ing. Drs. David H. Dunkle and Joseph T. of the dermal skull, some deeper skull struc- Gregory supplied data regarding the Todilto tures, and certain aspects of the postcranial Limestone at the Bull Canyon Todilto fish skeleton. The catalogued specimens listed for locality in northeastern New Mexico. The each taxon have been selected for their in- writers are greatly indebted to all of these formation content, and are the ones on which colleagues for their assistance and coopera- the descriptions are based. tion. For valuable discussion about systematic ACKNOWLEDGMENTS problems we are obligated to Drs. Peter L. The research for this paper was facilitated Forey, Brian G. Gardiner, Lance Grande, by a grant from the National Geographic So- Philippe Janvier, John G. Maisey, Amy R. ciety to Dr. Schaeffer, and by financial sup- McCune, Paul Olsen, and Donn E. Rosen. port to Dr. Patterson from the Department Mr. Walter Sorensen accompanied Schaef- of Geology, Field Museum of Natural His- fer during the first few collecting trips in Wy- tory, during his tenure there as Visiting Sci- oming and New Mexico, and he has prepared entist. The American Museum of Natural most ofthe specimens. Mr. Gilbert F. Stucker History has provided resources for many of served in this capacity during the remainder the illustrations. of the field program. Additional field assis- The fieldwork involved in obtaining the tance was provided at various times by Ms. fish specimens from the Sundance Formation Marlyn Mangus, Dr. Gareth J. Nelson, Dr. around Hulett, Wyoming was made possible Richard Lund and Mr. Richard W. Schaeffer. through the kindness and cooperation of the The drawings were made by Mr. Juan C. late Mr. Earl Willson and his daughter, on Barberis of the Graphic Arts Department at whose property most of the collecting was the American Museum and by Dr. Colin Pat- carried out, and of Mr. Arthur Meike, who terson. The photographs were taken by permitted us to excavate at a nearby ranch Messrs. Emile Javorsky and Richard Sheri- road locality. The late Mr. John Callquist dan at the American Museum and by the late provided many courtesies and facilities dur- Dr. Eugene S. Richardson. ing numerous stays in Hulett. Various individuals and institutions have Mr. J. C. Merrill granted permission to col- kindly lent specimens from both the Sun- lect fishes in the Todilto Limestone on his dance and the Todilto. These include Dr. ranch in Quay County, New Mexico. Dr. John Nicholas Hotton and Mr. Robert Purdy, Na- P. Bradbury assisted in collecting fishes at tional Museum of Natural History, Smith- several Todilto localities north of Santa Fe. sonian Institution; Dr. Keith S. Thomson, 4 AMERICAN MUSEUM NOVITATES NO. 2796

Peabody Museum, Yale University; Dr. Pe- den, dentary ter Robinson, The Museum, University of dhc, dorsal hemichordacentrum Colorado; Dr. Philip R. Bjork, Museum of dpcl, dorsal postcleithrum Geology, South Dakota School ofMines; Dr. dpl, dermopalatine dpt, dermopterotic Barry S. Kues, Department ofGeology, Uni- dptl, descending lamina of dermopterotic versity of New Mexico, Albuquerque; Mr. dr, dorsal fin radial Peter L. Larson, Black Hills Institute ofGeo- dsp, dermosphenotic logical Research, Hill City, South Dakota; ecp, ectopterygoid and Dr. Hans-Peter Schultze, Museum of enp, endopterygoid Natural History, The University of Kansas. ep, epural The Black Hills Institute has generously pre- epi, epioccipital sented several specimens from the Bull Can- epn, epineural process yon locality of the Todilto Limestone to the esc, extrascapular American Museum of Natural History. Dr. ff, fringing fulcrum fica, foramen of internal carotid artery T. T. Zwick presented several specimens from fm, foramen magnum the Sundance in Clarks Fork Canyon, Wyo- fmcv, foramen of middle cerebral vein ming. An additional collection from Bull foa, foramen of occipital artery Canyon, which has not been examined for fotb, ventral otic fissure this study, is in the Museum ofPaleontology, fpsa, foramen of efferent pseudobranchial artery University of California, Berkeley (J. T. fr, frontal Gregory, personal commun.). frd, foramen for sensory canal branch oftrigeminal nerve ABBREVIATIONS frf, descending lamina of frontal frl, foramen of sensory canal nerves in frontal INSTITUTIONAL fst IX, foramen of supratemporal branch of glossopharyngeal nerve AMNH, American Museum of Natural History fst X, foramen of supratemporal branch of vagus BHI, Black Hills Institute of Geological Research gpn, groove for palatine nerve BM(NH), British Museum (Natural History) grao, groove for dorsal aorta KUMVP, University of Kansas, Museum ofNat- hh, hypohyal ural History hmf, hyomandibular facet MCZ, Museum ofComparative Zoology, Harvard hpu, preural haemal spine University hym, hyomandibula NMNH (USNM), National Museum of Natural hyp, hypural History, Smithsonian Institution ic, intercalar SDSM, South Dakota School of Mines and Tech- id, interdorsal (dorsal intercalary) nology, Museum of Geology io, infraorbital YPM, Peabody Museum, Yale University iop, interopercular ANATOMICAL iv, interventral (ventral intercalary) jg, jugular groove ang, angular lc, lateral commissure ao, antorbital le, lateral ethmoid aol, housing of aortic ligament 11, lateral line canal in scales apa, autopalatine md, mandible ar, anal fin radial mes, mesethmoid asc, ascending process of parasphenoid mpt, metapterygoid bexo, basi-exoccipital mx, maxilla bhc, buccohypophysial canal na, nasal bht, basihyal toothplate nc, notochordal calcification bpt, basipterygoid process of parasphenoid neu, neural arch br, brachiostegal ray not, notochordal pit bsp, basisphenoid npu, preural neural spine cl, cleithrum occ, occipital arch cor, coracoid op, opercular csc, caudal scute opo, opisthotic dch, distal ceratohyal ors, orbitosphenoid 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 5 pa, parietal iments lacking diagnostic pas, parasphenoid invertebrates, the pch, proximal ceratohyal salinity of the water occupied by the fishes ph, parhypural (haemal spine of pu 1) remains equivocal. But the possibilities are pmx, premaxilla set forth, and the correlations ofthe fish-bear- pop, preopercular ing units are discussed, mostly on the basis pro, prootic ofrecently published regional reports and field prp, parapophysis studies plus consideration of the scanty am- pto, pterotic monite and other invertebrate evidence. pts, pterosphenoid ptt, posttemporal (suprascapular) SUNDANCE FORMATION pu, preural centrum pv, pelvis The paleogeography and sedimentology of qj, quadratojugal the lower Sundance Formation in the Black qu, quadrate Hills region has been discussed recently by ro, rostral Peterson (1972), Wright (1973), Rautman rode, rostrodermethmoid (1975, 1978) and Imlay (1980). The basal san, supraangular sbo, suborbital Canyon Springs Sandstone Member and the scl, supracleithrum lower part of the partly contemporaneous sl, standard length Stockade Beaver Shale Member (both of smx, supramaxilla which contain fishes) were deposited during sn, supraneural a marine transgression over the Wyoming soc, supraoccipital shelf, an east-west trending tectonic feature sop, subopercular in east-central Wyoming that was uplifted spg, spiracular groove during the Bajocian. The cross-bedded sand- spic, spiracular canal stones ofthe Canyon Springs Sandstone sug- spo, autosphenotic gest deposition under tidal influence, as in sr, sclerotic bone sue, supraethmoid the present shallow North Sea. The Stockade suo, supraorbital Beaver Shale is a shallow, near-shore marine sym, symplectic deposit formed in quiet water, as indicated u, ural centrum by the fine grain-size and by regional facies ud, urodermal changes (see paleogeographic map in Raut- un, uroneural man, 1978, fig. 13). una, ural neural arch According to Imlay (1980, p. iii) the cal- vhc, ventral hemichordacentrum careous sands forming the Canyon Springs vo, vomer Sandstone were deposited in the southeastern vpcl, ventral postcleithrum part of the late Bathonian Sea in a shallow- IV, trochlear foramen VII, facial foramen water environment that was partly intertidal X, vagus foramen or supratidal, while the limy mud that be- came the Stockade Beaver Shale was laid GEOLOGIC OCCURRENCE down farther to the west and north, also in shallow water. The discovery of the ammo- This section includes a resume of the stra- nite Warrenoceras in the Canyon Springs tigraphy and sedimentology of the Sundance Sandstone in the Hartville uplift in south- and Wanakah formations with particular ref- eastern Wyoming indicates a "not younger erence to units (members) from which fossil than middle" Bathonian age for this part of fishes have been obtained (figs. 1, 2). Where the member (Imlay, 1980, p. 85; J. H. Cal- pertinent, comments on the history of dis- lomon, personal commun.). However, the covery are also included. Two objectives of Canyon Springs Sandstone becomes progres- this discussion are (1) to attempt some con- sively younger from Wyoming into Colorado clusion regarding the environments in which where it interfingers with the base ofthe Pine the fishes lived and (2) to provide an opinion Butte Member of the Sundance, which may on the age of the fish-bearing rock units. As be about middle Callovian (Imlay, 1980; Pi- is frequently the case with epicontinental sed- piringos and O'Sullivan, 1976, sect. 12). 6 AMERICAN MUSEUM NOVITATES NO. 2796

Mont. I '4 I Oreg. I Ig, I I I

I-.- I I ------L t1----~

0 I C0

FIG. 1. Map of western North America showing the approximate extent of the transgressive inland sea during the Middle Jurassic. Diagonal ruled area represents the Wanakah depositional area including depositional basin of the Todilto and Pony Express limestones (shaded region in Arizona, New Mexico, and Colorado). Modified from Anderson and Kirkland (1960) and Imlay (1980). The circles indicate the areas from which fishes have been collected in the Sundance and Wanakah formations. Details in text. 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 7

o (n o W EUROPEAN NE WYOMING SCLRD ENWMXC STAGES SWS DAKOTA Q- (I) Overlying units (Cretaceous)

Portlandian ? ? ? Morrison Fm Morrison Morrison WVndy Formation Formation Kimmeridgian Hill w Ss. Mb.' 1 0- ;-,---,--- I-,-I IvIW-- I I I I a.

c Oxfordian 0 0 Redwater E Shale Member 0 IL

C0 I IO Pine Butte Mb Callovian I Wanakah Fm I ? Bell Ranch , LokMb Ls Fm c I.Pony Exp. Mbl Todilto IS| (I) Hulett Ss. Mb. Entrada Entrada (Exeter) Sandstone Sandstone U) tockade Canon Beaver -rFTTT- Bathonian Springh!h Mb Mb rVTT7 0 Ss. 0

Bajocian Gypsum Spring FormationI___ i~~~~~~~~~~~~~~~~~~i Toarcian

Pliensbachian 3: 0 -J I Sinemurian Hettangian

FIG. 2. Stratigraphic correlation chart for fish localities around Hot Springs, South Dakota; Hulett, Wyoming; Piedra Canyon, Colorado; Guadalupe and Quay counties, New Mexico. (Based mostly on Pipiringos and O'Sullivan, 1976, 1978, and Imlay, 1980.)

Wright (1974, p. 38) notes that part of the Callovian (or late Bathonian according to Im- Wyoming shelf was raised during the early lay, 1980) to form a northeast-southwest 8 AMERICAN MUSEUM NOVITATES NO. 2796

trending submarine barrier (the Sheridan stratigraphic occurrence is supported by the Arch) roughly between Lander and Buffalo, facies maps of Rautman (1975, fig. 3; 1977, Wyoming. This barrier greatly influenced the fig. 4) and of Pipiringos (personal commun.). sedimentation around it, and Wright believes It should be noted, however, that the Canyon that brackish conditions developed east of Springs is replaced in part or entirely by the the barrier, particularly during Stockade Bea- shaly Stockade Beaver Member northwest of ver deposition. These sediments lack fora- Hot Springs. In regard to the depositional minifera but in places have a rich ostracod environment of the fish layer at Hot Springs, fauna along with a Meleagrinella bivalve as- Anderson and Kirkland (1960, p. 43) remark semblage. The Sundance fishes apparently that "Darton's fish were not directly associ- lived near shore but the bottom environment ated with a marine fauna, but were the only had become unfavorable for benthic inver- fossils found in a sandstone below the marine tebrates. sequence and about 3 feet above an eroded The fish localities in the Sundance For- surface of redbeds (Darton, 1899, p. 399)." mation are as follows: But it must be emphasized that, except for 1. Hot Springs area, South Dakota: The occasional specimens, marine fossil fishes are initial discovery of fishes in the Sundance rarely found in association with benthic in- Formation was made by N. H. Darton in vertebrates. 1898 at a locality described by him as "about From Darton's (1899, p. 389) description 1 mile ESE from the railroad station or one- of the fish occurrence, it was apparently re- half mile south of the Catholicon Springs stricted both horizontally and vertically. An Hotel" (foundation still visible in 1966) in effort to relocate the spot in 1966 was un- Hot Springs, Fall River County, South Da- successful. Outcrops of the Canyon Springs kota. It is further described as "in a small and the Stockade Beaver are rare except where draw which heads in the sandstone ridge lying gullies have eroded through the grass cover. east of the Red Bed valley" (Darton, 1899, On the basis ofDarton's small collection, now pp. 388-389). On the basis of these land- in the National Museum of Natural History, marks, it appears that the locality is near the Smithsonian Institution, and the Museum of base of the high ridge (Seven Sisters Range) Comparative Zoology, Harvard University, that extends in a nearly north-south direction specimen preservation is generally good immediately to the east of Hot Springs (G. enough to warrant a further search for the D. Johnson, personal commun.). Both the productive horizon. Canyon Springs Sandstone and the Stockade 2. Hulett area, Wyoming: In terms of spec- Beaver Shale members occur along the east- imen number and diversity the most impor- ern margin ofthe valley in which Hot Springs tant fish locality in the Sundance Formation is situated (see detailed geologic map of the is on the former Earl Willson Ranch in Burnt Hot Springs quadrangle in Wolcott, 1967). Hollow, which is about 6.5 mi (10.5 km) north The locality is therefore in the NW 1/4, Sec. of Hulett, Crook County, Wyoming in SE 1/4, 25, R5E, T7S, Fall River County, South Da- Sec. 19, T55N, R64W. The first specimens kota. were found in 1910-1912 by a Hulett resi- Comparison of the stratigraphic section dent. More were discovered in later years and measured by Darton (1899, p. 389) in the numerous well-preserved examples were un- vicinity ofthe Catholicon Springs Hotel with covered in the 1950s when a bulldozer was those by Wolcott (1967) and by Rautman employed by Earl Willson to build a stock (1975) near the town of Minnekahta (12 mi dam across the intermittent stream in Burnt or 19 km west of Hot Springs) suggests that Hollow. Some of these specimens were ex- the fish collected by Darton were derived from amined in 1957 by the late James D. Bump, the upper part of the Canyon Springs Sand- then director ofthe Natural History Museum stone. Darton noted that the specimens occur at the South Dakota School ofMines in Rap- in a soft, fine-grained, thin-bedded (laminat- id City, who informed Schaeffer about the ed) sandstone which is above the orange-to- locality. The first collection for the American red, cross-bedded sandstone that constitutes Museum was made in 1958, and additional most of this unit. This conclusion regarding specimens were obtained at intervals through 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 9

1975. The laminated, unbioturbated, calcar- fish specimen, but the skeleton is never py- eous siltstone containing the fishes may yield ritized. fairly complete specimens following weath- The laminated calcareous siltstone unit ering in place or in isolated blocks. A primary rests unconformably on a layer of ledge- reason for returning to this locality during forming limestone containing irregular masses successive field seasons was to increase the of chert, and below this there is a bed of diversity of the fish assemblage. Unfortu- reddish, fissile shale. In 1966, a reconnais- nately, this was not particularly successful. sance of the area between the Burnt Hollow Ofthe nearly 800 remains ofindividual fishes fish locality and the Belle Fourche River with collected at the Burnt Hollow locality, only G. W. Imlay and G. N. Pipiringos reinforced five taxa are represented. Of these about 90 the opinion that the laminated calcareous percent can be referred to the single new siltstone in Burnt Hollow has a local distri- monotypic genus Hulettia. bution, and that it overlies the Gypsum Spring The fish zone is in the lower part of a cal- Formation (see Pipiringos and O'Sullivan, careous siltstone unit about 20 feet (6.5 m) 1978, pp. A20-A21, in regard to the J-2 un- thick that is well exposed immediately to the conformity at the top of the Gypsum Spring east of the Willson stock dam in Burnt Hol- Formation). According to the earlier strati- low. The exposure is limited to the north side graphic studies of Imlay (1947), Mapel and ofthe intermittent stream that flows through Bergendahl (1956) and of Robinson, Mapel, the Hollow andjoins the Belle Fourche River and Bergendahl (1964), the fish zone would in Sec. 20, R65W, T55N. It extends eastward be placed in the lower part of the Canyon for about 300 feet (100 m) from the dam, Springs Sandstone Member of the Sundance where it is covered by grass-covered allu- Formation. As shown in the stratigraphic sec- vium. There is no evidence ofthis calcareous tions compiled by Robinson, Mapel, and siltstone unit elsewhere in Burnt Hollow, nor Bergendahl (1964, pl. 2), siltstone occurs is it present in an excellent exposure of the above the limestone in the Hulett area in a lower Sundance along the Belle Fourche Riv- unit they designate as the Canyon Springs. er opposite the mouth of Burnt Hollow, also However, field studies of the facies changes in Section 20. in the Sundance for northeastern Wyoming During the 1966 field season, the siltstone and the adjacent part of South Dakota (Im- unit was excavated from its contact with the lay, 1947, pp. 251-252; Pipiringos, 1968, p. overlying soft, greenish shale to the bottom D19 and fig. 19; Rautman, 1978) indicate of the fish zone. Lingula sp. and the bivalve that the Stockade Beaver Member uncon- Meleagrinella orbiculata Whitfield occur in formably overlies the Gypsum Spring For- the poorly bedded upper part ofthe unit along mation, and the Canyon Springs Sandstone with rare remains ofLepidotes and some iso- Member interfingers with the Stockade Bea- lated bone fragments and scales. About 4 feet ver well south of Hulett, and in South Da- (1.3 m) below the top ofthe massive siltstone kota, within a short distance north of Hot and through a thickness of approximately 3 Springs. On the basis ofthis evidence, there- feet (1 m), there are several concentrations fore, the fish zone in Burnt Hollow is placed ofdecapod crustaceans. According to Herrick by us in the lower part ofthe Stockade Beaver and Schram (1978), these include Antrimpos Shale Member. This interpretation is sup- sp., Bombus sp., Mecochirus sp., an aglypheid ported by the absence of cross-bedded sand- and some unidentified anomuran remains. stones in Burnt Hollow, which are charac- The fish zone, about 3 feet (1 m) thick, is teristic of the Canyon Springs Member. In readily distinguished from the massive silt- any case, the fish layers near Hot Springs, stone above it by being finely laminated. Lo- South Dakota and the fish zone in Burnt Hol- cally there is some fine cross-bedding. Inver- low are roughly contemporaneous (Imlay, tebrates are absent, but concentrations of 1947, fig. 3; Rautman, 1978, fig. 5; this paper, bedded, fine-grained black pyrite with a dis- fig. 1). tinctive dendritic or stellate pattern are com- In Bush Canyon, which is the next parallel mon and restricted to this zone. The pyrite drainage to the south of Burnt Hollow, the may occur on the same bedding plane as a limestone layer at the top of the Gypsum 10 AMERICAN MUSEUM NOVITATES NO. 2796

Spring Formation is overlain by a soft shale don Formation, which is the lateral equiva- and shaly limestone unit containing Melea- lent ofthe Canyon Springs and Stockade Bea- grinella and Lingula (see stratigraphic sec- ver members farther south in Wyoming. The tions in Imlay, 1947; Robinson, Mapel, and other area includes four localities in the vi- Bergendahl, 1964) as does the siltstone unit cinity of Gypsum Creek, Carbon County, in Burnt Hollow above the fish zone. These Montana; Sec. 28, 33, T9S, R27E and in Big two units are probably correlated, but the de- Horn County, Wyoming; Sec. 21, T58N, tailed sedimentary sequence is different, and R96W. According to J. Utgaard and M. L. no fish have yet been found in Bush Canyon. Dejarnett (personal commun.), who have in- Another limited exposure ofthe laminated vestigated the stratigraphy, sedimentology calcareous siltstone containing fishes is lo- and invertebrates ofthe Sundance Formation cated about 2.5 air mi (4 km) SSE ofthe Burnt in this region, the fishes occur in laminated Hollow locality. This layer forms the surface carbonates in the Hulett Sandstone Member. of a little-used road on the Meike (Everson) All the individuals found to date may be re- Ranch near a wide loop in the Belle Fourche ferred to Hulettia americana according to in- River in NE 1/4, Sec. 5, T54N, R64W. As formation and specimens kindly provided by preserved, it is about 1 foot (0.3 m) thick, Dr. H. P. Schultze. and lies directly on limestone containing In regard to the depositional environment chert. Extensive soil and grass cover make it of the fish zone around Hulett, it must have difficult to determine the total thickness of been mostly azoic. However, this refers to this siltstone unit away from the road. There the preservation environment for the fishes, is, however, a bed ofsoft, greenish fissile shale and not to the waters above the bottom where on the west side of the road that forms a the fishes must have lived. We have found gentle slope below the base of the Hulett no direct criteria to indicate whether this hab- Sandstone. In Burnt Hollow this same green- itat was marine, brackish or fresh water, even ish fissile shale, which is characteristic ofthe though the massive siltstone above the fish upper part of the Stockade Beaver Shale zone contains marine pelecypods and Lingu- Member (Robinson, Mapel, and Bergendahl, la. The presence ofLepidotes bones and scales 1964, p. 15), is about 137 feet (42 m) thick in association with benthic invertebrates in to the base ofthe Hulett Sandstone. It is easily the massive siltstone probably reflects the rel- recognized by its erosional pattern compared ative durability of these skeletal elements with the underlying, nearly vertical face of prior to burial. It is reasonable to assume, the siltstone unit near the Burnt Hollow stock however, that the fish occupied a marine en- dam and with the overlying cliff-forming Hu- vironment as the basal Stockade Beaver con- lett Sandstone. tains marine invertebrates to the north, south A third and also limited exposure of the and west. fish-bearing siltstone is present along the road leading to the original Willson Ranch build- ings. This is in Sec. 4, T54N, R64W and is WANAKAH FORMATION about a half-mile (0.8 km) north of State In this report the correlation of the fish- Highway 24. bearing Jurassic rock units in northeastern 3. Two additional Sundance fish collecting Wyoming and southwestern South Dakota areas have been brought to our attention. One with those in Colorado and New Mexico is is at the mouth of Clarks Fork Canyon, Park based on Imlay (1980) and other papers men- County, Wyoming in the NW 1/4, NW 1/4, Sec. tioned below. Ofparticular importance in this 5, T56N, R103W, Deep Lake 15' Quadrangle regard is the correlation of some 23 strati- (T. T. Zwick, personal commun.). The spec- graphic sections between Douglas, Wyoming imens have been found in a platy calcareous and the Ralston Reservoir, Colorado (Pipi- siltstone, and those recovered to date may be ringos and O'Sullivan, 1976, and their related referred to Lepidotes sp. According to the paper of 1978). Conclusions pertinent to this investigations of Pipiringos and O'Sullivan study are summarized below, hopefully in (1978, pl. 1, section E'-E, locality WI), the enough detail to provide a useful stratigraph- specimens occur in the lower part ofthe Rier- ic background. 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 11

The Morrison Formation is mostly or en- some confidence to the Wanakah, following tirely Kimmeridgian in age (Pipiringos and Imlay (1952, p. 961). The Wanakah For- O'Sullivan, 1978, p. A26). According to these mation (in part) may be equivalent to the Lak authors, the Ralston Creek Formation (Fred- Member of the Sundance Formation in Wy- erickson, 1956; Johnson, 1962) at the type oming (Pipiringos and O'Sullivan, 1978, pl. locality at Ralston Reservoir, Colorado, con- I and p. 24). sists of a lower unit, the Canyon Springs The Wanakah Formation in the San Juan Sandstone, and an upper unit which is ac- Basin area includes variable amounts of tually the lower part of the Morrison For- limestone, mudstone, siltstone, sandstone, mation. Fossils from the upper part of the and gypsum. The Todilto Limestone Mem- Ralston Creek Formation near the town of ber thickens eastward in McKinley and San- Morrison, Colorado are ofKimmeridgian age doval counties and is continuous at least as and are mostly identical with those from the far as Las Vegas in San Miguel County Morrison Formation elsewhere (Imlay, 1952, (Harshbarger, Repenning, and Irwin, 1957). p. 961; Scott, 1963, p. 92; see also Schultze It is nowhere more than 10 feet (3 m) thick. and Encisco, 1983). The gypsum beds, which attain a thickness There are no rocks ofOxfordian age in New ofover 100 feet (30 m) in the San Juan Basin, Mexico and most ofColorado. The Redwater overlie the limestone but may grade into it, Shale Member of the Sundance Formation, and into sandstones around the periphery of which is widespread in northern Wyoming, the evaporite basin. is middle Oxfordian in age (Imlay, 1980, p. Benthic invertebrates are generally absent 87), but unfortunately is not known to con- in the Todilto Limestone Member. However, tain fishes. near-shore sediments contain fresh or brack- The Wanakah Formation, of middle and ish water ostracods (Swain, 1946, p. 553), possibly lower Callovian age, is present in algal nodes (Perry, 1963), tracheids of vas- central Colorado and extends southward into cular plants (Anderson and Kirkland, 1960), New Mexico. The fish horizon in the Piedra and two families of aquatic hemipterid in- River Canyon, Colorado is in the basal part sects (Bradbury and Kirkland, 1966). Fishes of the Wanakah (in the Pony Express Lime- have been found at several places near the stone Member), as probably are the fish lo- edge as well as toward the middle of the To- calities in New Mexico east of Santa Rosa. dilto depositional area (Tanner, 1970 and The Todilto Limestone, which is regarded as personal commun.; Bradbury and Kirkland, the basal member of the Wanakah by Pipi- 1966 and personal commun.). These locali- ringos and O'Sullivan (1976, pl. 1, section ties include Lamy, Santa Fe County; La C-C'), immediately overlies the Entrada Liendre, a ghost town south of Las Vegas in Sandstone. In the Fort Wingate area, New San Miguel County; Hot Springs and Echo Mexico, the Cow Springs Sandstone overlies Amphitheater on U.S. Highway 84 between the Todilto Limestone (Kehler, 1975). There Abiquiu and Cebolla in Rio Arriba County. is now evidence (Peterson, 1974) that the Cow As noted above, the fish locality in the Piedra Springs is a bleached zone in the upper part River Canyon, about 5 mi (8 km) north of ofthe Entrada Sandstone. Iffurther field study the town of Piedra in Achuleta County, Col- confirms this possibility, then the Todilto be- orado, is in the Pony Express Limestone neath the Cow Springs could be lower Cal- Member ofthe Wanakah (Read et al., 1949), lovian. which is equivalent to the Todilto Limestone A fish-bearing unit north of Canon City, Member (R. B. O'Sullivan, personal com- Colorado (Schoewe, 1930; Dunkle, 1942) has mun.). been identified as part of the Ralston For- The Todilto Limestone has been mapped mation, although it is stratigraphically below eastward to Las Vegas, New Mexico by Baltz the Morrison (Frederickson, DeLay, and Say- (1972) and Johnson (1974). The fish-bearing lor, 1956, p. 2138). In view of the evidence limestone in Guadalupe and Quay counties cited above, plus the recent work on the ex- in northeastern New Mexico has the same tent and relationships of the Wanakah For- stratigraphic position within the Wanakah as mation, these beds may now be assigned with the Todilto farther west, and is lithologically 12 AMERICAN MUSEUM NOVITATES NO. 2796 very similar. In agreement with Imlay (1952) ronment. It is thus possible that the Todilto and Anderson and Kirkland (1960), this unit and Pony Express limestones were deposited is regarded here as Todilto. In northeastern under variable conditions. The lighter color New Mexico, Peterson (1972) and Dinwiddie ofthe unweathered Todilto Limestone in the and Clebsch (1973) refer to the Wanakah southwest part ofthe San Juan Basin together equivalent as the Bell Ranch Formation. with the concentrations of algal material in- The Todilto Limestone Member in Gua- dicate shallow, well-oxygenated water. In the dalupe and Quay counties, New Mexico con- central and eastern parts of northern New tains a fish assemblage identical with that in Mexico and in Colorado, the Todilto and the Todilto in the San Juan Basin and related Pony Express limestones are dark brown and areas. Many specimens have been collected rich in organic matter suggesting a lower oxy- in this limestone unit in Bull Canyon, Gua- gen bottom environment. The invertebrate dalupe County (Sec. 28, 32, and 33, T9N, remains (particularly the hemipterid aquatic R26E) following their discovery in 1929 insects), also from the southwestern margin (Koerner, 1930), and by Schaeffer et al., in of the Todilto depositional area, further in- 1964, on the nearby Merrill Ranch in Quay dicate the presence ofat least local freshwater County (sec. 5 and 6, T8N, R27E). and brackish environments. Specimens ofthe Interpretation of the Todilto depositional teleost Todiltia, new genus, have been found environment has been confused and contra- on the same bedding plane as the ostracods dictory. Tanner (1970, 1972, 1974 and per- (J. P. Bradbury, personal commun.). sonal commun.) has proposed that the To- The Sundance fish assemblage by compar- dilto Limestone in the San Juan Basin was ison with various European Jurassic assem- deposited in one or more large lakes that ex- blages may be regarded as marine (see table tended eastward to the Las Vegas area. The 3). Although only two Sundance fish taxa are evidence in favor of Todilto deposition in a also present in the Todilto (Hulettia, new ge- nearly closed gulf or embayment of the Late nus, and Caturus), there is reason for sug- Jurassic epeiric sea during Summerville-Cur- gesting that these fishes entered the Todilto tis time remains equivocal (Anderson and Basin from the western interior ("Sundance") Kirkland, 1960, p. 45). However, "prelimi- sea. Todiltia schoewei, which is common in nary carbon, oxygen and sulfur isotope stud- the Todilto and Pony Express limestones, but ies of bulk rock samples of the Todilto and is apparently absent in the Sundance, may Pony Express Limestone suggest a marine or- have been a freshwater form, but obviously igin for these formations. Carbon and sulfur this cannot be ascertained directly from the isotopes cluster within normal marine range. specimens. Oxygen isotopes are light; the variations are In summary, the fishes discussed in this a result ofdiagenesis" (J. L. Ridgley, U.S.G.S., report range in age from the last half of the personal commun.). Anderson and Kirkland Bathonian to about the middle of the Cal- have calculated that the area ofTodilto Lime- lovian. There is no present evidence that this stone deposition was about 34,600 sq mi time span will be narrowed, but at least the (89,600 kM2). Assuming a sea connection, regional correlation ofthe involved rock units they favor a mixed paralic basin environment appears to be much clearer than it was a de- with the evaporites derived mostly from the cade ago. Contrary to Schultze and Encisco surrounding drainage area. (1983), the lowest two members of the Sun- A paralic condition for at least part of the dance Formation (Canyon Springs Sandstone Wanakah depositional area is also implied in and Stockade Beaver Shale) have recently Imlay's (1980, figs. 32, 33) paleogeographic been pushed back into the Late Bathonian on maps for the western interior during the early the basis ofammonite evidence (Imlay, 1980; Callovian. The Wanakah "embayment" was J. H. Callomon, personal commun., 1982), presumably widely open to the interior sea and that the Hulett Sandstone Member, which during this interval (fig. 1), and marine or is in the zone of Kepplerites maclearni, may tidal flat conditions could have alternated with also be Bathonian (J. H. Callomon, personal an essentially freshwater, continental envi- commun., 1982). 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 13

FIG. 3. A. Ischyodus sp. AMNH 8727. Incomplete right mandibular toothplate. B. Hybodus sp. AMNH 10995. Isolated tooth crown. C. Hybodus sp. AMNH 8726. Isolated tooth crown. D. Hybodus sp. AMNH 8725. Isolated tooth crown. From Hulett area. x 2.75.

SYSTEMATICS collected by E. 0. Hovey in 1891 in "Jurassic beds" 2 mi (3.2 km) east ofthe Devils Tower CHONDRICHTHYES (AMNH 8726, fig. 3C; AMNH 8728) and 4 ELASMOBRANCHII mi (6.4 km) southwest of Hulett (AMNH HYBODONTIDAE OWEN, 1846 8725, fig. 3D). According to the map of this GENUS HYBODUS AGASSIZ, 1837 area in Robinson, Mapel, and Bergendahl DIAGNOSIS: See Woodward, 1916, p. 3; (1964), they were found in the Sundance, and Maisey, 1982, fig. 17. probably in the Stockade Beaver Member. TYPE SPECIES: Hybodus reticulatus Agassiz DISCUSSION: As in Hybodus grossiconus (designated by Woodward, 1916, p. 4). Agassiz, from the Bathonian and (according DISTRIBUTION: See Blot, 1969, p. 729, and to Priem, 1911) Oxfordian of Europe, the tables 3, 4. principal cusp is high, broad at the base, well separated from the lateral denticles, and bears Hybodus sp. close-packed striae extending about halfway Figure 3B-D up the labial face. With so little material and HORIZON AND LOCALITY: Stockade Beaver the problems ofspecies determination within Member, Sundance Formation, Burnt Hol- Hybodus, we prefer to leave our specimens low near Hulett, Crook County, Wyoming. as Hybodus sp. There are two pairs of lateral For details see section on geologic occur- denticles on AMNH 8728, and one pair on rence. AMNH 10995 and AMNH 8726. REFERRED SPECIMENS: AMNH 10995 (fig. The only previous record of Hybodus in 3B), isolated tooth from Burnt Hollow lo- the American Jurassic is a tooth from the cality at top of massive siltstone unit. Three "Atlantosaurus beds" near Piedmont, South additional teeth, also referred to Hybodus, Dakota, referred by Marsh (1899) to H. poly- 14 AMERICAN MUSEUM NOVITATES NO. 2796

prion Agassiz. Our specimens do not resem- of primitive and derived characters: brain- ble that species. case with persistent otico-occipital fissure; no supraoccipital; intercalar without membra- HOLOCEPHALI nous outgrowths; parasphenoid broad, toothed, with foramina for efferent pseudo- CHIMAERIDAE BONAPARTE, 1831 branchial and internal carotid arteries; vomer GENUS ISCHYODUS EGERTON, 1843 median; nasals separated by rostral. Nasal DIAGNOSIS: See Woodward, 1932, p. 96; processes of premaxillae simple, meeting in Heimberg, 1949, p. 76. midline and occasionally fusing; maxilla free TYPE SPECIES: Chimaera townsendii Buck- and mobile, extending beneath posterior part land, 1 835 (designated by Woodward, 189 1, of orbit, no supramaxilla. Quadratojugal in- p. 64). dependent and splintlike. Interopercular DISTRIBUTION: Middle Jurassic-Paleocene, present. Post-temporal (suprascapular) with Europe; Middle Jurassic, western interior sea internal process. Large, median basihyal ofNorth America; Cretaceous, New Zealand. toothplate; no gular. Vertebral column with separate neural arches and dorsal intercala- Ischyodus sp. ries throughout; separate ventral intercalaries Figure 3A present only in anterior part ofcaudal region; HORIZON AND LOCALITY: Unknown, but unpaired neural spines in posterior half of probably from the Sundance Formation column. Dorsal and ventral chordal hemi- "5-6 miles (8-10 km) sw ofSundance, Wyo." centra meeting as annular chordacentra in REFERRED SPECIMEN: AMNH 8727 (fig. mature individuals. No ossified ural/neural 3A). Collected by E. 0. Hovey in 1891. In- arches. At least seven epurals. Hypocaudal complete right mandibular toothplate. skeleton with five elongated preural haemal DISCUSSION: The toothplate is similar to spines and at least eight unmodified hypurals. those of the Bathonian Ischyodus emargi- No clavicle. Scales rhomboidal, composed of natus Egerton and the Callovian-Kimmer- enameloid and bony base; dentine absent. idgian I. egertoni (Buckland). According to Basal and fringing fulcra on all fins. Woodward (1891, p. 60) those two species Hulettia americana (Eastman) are "only provisionally retained distinct." Figures 4-18 Reliable species determination seems im- Pholidophorus americanus Eastman, 1899a, p. 642. possible in Ischyodus on mandibular tooth- Pholidophorus americanus Eastman; Eastman, plates alone (Woodward, 1891; Heimberg, 1899b, p. 398, pls. 45-57. 1949), and we leave the specimen under open Pholidophorus americanus Eastman; Knight, 1900, nomenclature. This Sundance specimen is p. 386 (name only). apparently the first record ofIschyodus in the Pholidophorus americanus Eastman; Gilmore, 1905, p. 89 (name only). western hemisphere. Pholidophorus americanus Eastman; Merrill, 1907, p. 15. OSTEICHTHYES Pholidophorus americanus Eastman; Koerner, ACTINOPTERYGII 1930, p. 463 (name only). NEOPTERYGII Pholidophorus americanus Eastman; Dunkle, 1942, HALECOSTOMI INCERTAE SEDIS p. 61 (name only). HULETTIA, NEW GENUS Pholidophorus americanus Eastman; Schultze and Enciso, 1983, p. 1056. TYPE SPECIES: Pholidophorus americanus Eastman. HOLOTYPE: NMNH 4788a (Eastman, DISTRIBUTION: Upper Bathonian, Wyo- 1899b, pl. 45, fig. 1; fig. 4A here) and coun- ming; Lower Callovian, New Mexico and terpart NMNH 4788b (Eastman, 1 899b, pl. Colorado. 46, fig. 1). From Canyon Springs Sandstone ETYMOLOGY: After Hulett, Crook County, Member, Sundance Formation, near Hot Wyoming. Springs, South Dakota. DIAGNOSIS: A generalized halecostome DIAGNOSIS: As genus, only species. neopterygian with the following combination REFERRED SPECIMENS: Partial and nearly 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 15 complete specimens of Hulettia americana SNOUT: The pattern of the dermal snout showing specific morphological features are has been mostly worked out on the basis of mentioned by catalogue number at pertinent specimens from the Hulett localities. The places in the description. The numbers listed ethmoid region is covered by a keystone- below by formation and locality represent shaped, median rostral bone that overlies the catalogued specimens examined for this study. nasal processes ofthe premaxillae and houses From the Sundance Formation, Canyon the prominent ethmoid commissure, which Springs Sandstone Member, near Hot Springs, extends transversely through the bone close South Dakota: NMNH 4784, 4787, 4789, to its anteroventral margin. The nasals, which 4790; MCZ 6623, 9696. From the Sundance are separated by the rostral, presumably fitted Formation, Stockade Beaver Shale Member, into the concave lateral margins of this ele- near Hulett, Wyoming: AMNH 10824, ment. The L-shaped antorbitals, readily iden- 10831, 10837, 10838, 10839, 10841, 10857, tified by the triradiate pattern oftheir sensory 10858, 10860, 10862, 10872, 10875, 10876, canals, are in contact with the rostral, the 10877, 10880, 10881, 10884, 10885, 10887, premaxillae and the infraorbital elements as 10888, 10897, 10903, 10917, 10918, 10919, shown in figure 11. There must have been 10921, 10922, 10923, 10926, 10928, 10931, space between the antorbital and the nasal 10932, 10940, 10941, 10943, 10944, 10948, for the anterior and posterior nasal openings. 10954, 10956, 10957, 10986, 11071, 11072, The snout pattern is similar in most re- 11073, 11075, 11077, 11437; SDSM 5780, spects to that of Acentrophorus, Dapedium, 5781, 5782, 54361, 59144, 59145. From and a parasemionotid as interpreted by Pat- Wanakah Formation, Pony Express Lime- terson (1975, figs. 134, 136, 137). The su- stone Member, Piedra River Canyon, Colo- praorbital sensory canal passes through the rado: AMNH 11454. From Wanakah For- nasal bone and terminates above the rostral mation, Todilto Limestone Member, Bull commissure. The sensory canal extension in Canyon area, New Mexico: AMNH 6339, the rostral process of the antorbital ends 10927, 10968; BHI P 5, P 1.25, 950F, P 1.10, blindly. The extent of the rostral-premaxilla YPM 8199, 8200. From Wanakah Forma- overlap in figure 11 has been determined by tion, Todilto Limestone Member, Merrill the ornamentation on the premaxillae, which Ranch, New Mexico: AMNH 11021, 11025, must have been exposed below the margin of 11028, 11443. the rostral. ROOF: The relatively short, broad frontal DESCRIPTION bones, which are not quite twice the length MEASUREMENTS AND PROPORTIONS: The of the parietals, give the skull roof a blunt, smallest measurable specimen (AMNH widened aspect much as in Acentrophorus 11075b) is about 50 mm SL, and the largest (Gill, 1923) and in many ofthe paleopterygi- (AMNH 10954) is 155 mm. The proportion- ans with a nearly vertical suspensorium al changes with increase in size are indicated (Schaeffer, 1973). The frontals are slightly in the graph (fig. 7). Although maximum body narrowed anteriorly and their anterior bor- depth is difficult to approximate because of ders are rounded and ruffled for the attach- compression, it is evident that the body form ment of the connective tissue that also con- remained relatively constant with increase in nected the snout elements. The well-spaced size-at least above about 50 mm. It is also pores ofthe supraorbital sensory canals have apparent that within this size range the skull elevated borders. On the ventral surface of is about one-quarter of the standard length. the frontal there is a prominent, rounded ridge According to this analysis, Hulettia is monotypic, which is verified by the scale and fin beneath the sensory canal that has ventral ray counts. The measured sample includes digitations in front of and behind the ossifi- specimens from the Sundance (Burnt Hollow cation center of the bone (fig. 12A), forming locality, Wyoming), the Pony Express Lime- a poorly defined descending lamina. stone (Piedra Canyon, Colorado), and the The rectangular parietals are indistinguish- Todilto Limestone (Bull Canyon, New Mex- ably fused in several specimens (fig. 12A) but ico). in most they are clearly separate. The anterior 16 AMERICAN MUSEUM NOVITATES NO. 2796

,: ., .;'-;. A

FIG. 4. Hulettia americana. A. NMNH 4788a. Holotype. x.90. B. SDSM 59144. x.68. C. AMNH 10921. x.70. D. AMNH 10903. x 1.1. From Hot Springs, S.D. and Hulett, Wyoming. pit lines are extensions of the supraorbital middle pit lines are evident in AMNH 10872. canals on the parietals. The rarely observed The dermopterotics require no particular 1984 SCHAEF:FER AND PATTERSON: JURASSIC FISHES 17

D FIG. 5. Hulettia americana growth series. A. AMNH 10860. x 1.32. B. AMNH 10837. x 1.50. C. AMNH 10838. x 1.30. D. AMNH 1088 1. x 1. 16. From Hulett, Wyoming.

comment except to note that the bone sur- OPERCULAR SERIES AND CHEEK: The oper- rounding the temporal sensory canal is thick- cular series, including the well-developed in- ened and forms a descending lamina pos- teropercular, has a characteristic, primitive teroventrally (figs. 8A, 12A). Also there is no neopterygian pattern. The narrow, nearly pronounced posterior extension of the der- vertical preopercular is covered dorsally by mopterotic around the canal, although this a large suborbital. There are about 13 bran- process is presumably present to some degree chiostegals. The gular plate is absent. in most primitive actinopterygians (Patter- The suborbital series includes one large ele- son, 1975, p. 552). ment (sbo1, fig. 1 I B), and a variable number 18 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 6. Hulettia americana. A. AMNH 11443. Throat region of specimen from Merrill Ranch, New Mexico. x 2.5. B. BHI P110. From Bull Canyon, New Mexico. x.80. C. AMNH 11454, from Piedra River Canyon, Colorado. x .88.

of smaller ones. There seems usually to be at supraorbitals in front ofthe dermosphenotic. least one small suborbital anterodorsal to the The last is sutured to the frontal and the der- large bone (sbo2, fig. 1 1B). In AMNH 1083 la mopterotic as in Perleidus, semionotids and there are three small bones in this position primitive teleosts, and does not form part of on the right side and two on the left. Several the skull roofas in Ophiopsis (Bartram, 1975), specimens show a suborbital (sbo3, fig. 1 1B) amiids and caturids (Patterson, 1973, pp. 244, ventral to the large one, but in other speci- 279). There are six infraorbitals, the third mens this is evidently absent. (counting back from the antorbital) enlarged The circumorbital series (fig. 1 1 B) is inter- as in parasemionotids, Ophiopsis and many rupted only in the region of the posterior na- ofthe primitive teleosts. The two narrow an- sal opening, between the dorsal process ofthe terior infraorbitals form little more than a antorbital and the foremost supraorbital. Hu- tube around the infraorbital sensory canal. lettia usually has three, sometimes only two In mature individuals all the cranial der- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 19

130- 120 A 110-

100- A A A mB 90- U 80- 70- U 60-

50- A A 40- ALj A a 30- 20 e ~~~0e 10-

I I F I - r I I I I 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 FIG. 7. Hulettia americana. Regressions in mm of SL (measured from the tip of snout, to posterior end oflateral line) on horizontal axis and: A, distance from snout to origin of anal fin; B, snout to origin of dorsal fin; C, estimated body depth; D, snout to posterior border of opercular bone. Sample includes specimens from South Dakota, Wyoming, Colorado, and New Mexico.

mal bones are covered to some extent with neopterygians generally and is known in some ganoin ridges and tubercles. On some bones paleopterygians. Sensory canal pores with such as the dentary, the density of these el- somewhat elevated rims are present on the evations gives a rugose appearance. On the frontals ofmature individuals. Pores are also opercular, interrupted ridges radiate from an evident along the mandibular sensory canal area near the anterior border. but are otherwise difficult to locate. SENSORY CANALS: As in most non-teleos- BRAINCASE: The available material of the tean actinopterygians, the supraorbital and braincase of Hulettia consists of five speci- infraorbital canals ofHulettia are not joined. mens: three isolated, dorsoventrally crushed The supraorbital canal extends into the pa- braincases, AMNH 10880 (fig. 8A), 11071, rietal where it is replaced by the anterior pit 11073, the last two lacking the occipital os- line (fig. 1 1). The middle pit line extends from sification; SDSM 54361, a specimen in which the parietal into the dermopterotic. In the the braincase and parasphenoid (fig. 8B) have snout region, the right and left infraorbital drifted free from the adjacent skeleton and canals are joined by the rostral commissure. roofing bones; and AMNH 10926 (fig. 9), an The dorsal spur of the infraorbital canal in isolated occipital ossification. Except for the the antorbital bone is characteristic of latter specimen, the braincases are severely 20 AMERICAN MUSEUM NOVITATES NO. 2796

2mm FIG. 8. Hulettia americana. Braincase. A. Restoration in ventral view, based mainly on AMNH 10880. Ethmoid ossifications omitted and vomer shown in outline only, since details are unknown. B. Interpretative sketch of braincase in left lateral view as preserved in SDSM 54361. crushed, so that only external structures are The occurrence of isolated occipital ossi- known in any detail. fications and of braincases lacking that por- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 21

f x ic bexo

grao fica FIG. 9. Hulettia americana. Occipital ossifications of braincase in posterior aspect. Based on AMNH 10926. x7.6.

tion shows that the braincase ofHulettia was FIG. 10. Hulettia americana. Parasphenoid in divided into occipital and pre-occipital sec- dorsal aspect. Mainly based on AMNH 10957. tions by a persistent cranial fissure. We are x6.7. unable to say with certainty what parts ofthe otico-occipital fissure were perichondrally lined, but AMNH 10926 and SDSM 54361 show that the dorsolateral part of the fissure orbit, an autosphenotic forming the postor- (epioccipital opposed by the pterotic) and the bital process, and a dorsolateral pterotic, re- region above the vagus foramen (intercalar) sembling that bone in, for example, Perleidus were so lined. A perichondrally lined otico- and parasemionotids (Patterson, 1975, figs. occipital fissure occurs in palaeoniscoids, 97, 98, 115, 116). AMNH 10926 shows that Australosomus, and certain pholidophorids. no supraoccipital was present and the only Parts ofthe fissure retain perichondral lining uncertainties concern the lower parts of the in Perleidus, some parasemionotids, pachy- otic and sphenoid regions, where in the gen- cormids, aspidorhynchids, and some pholi- eralized actinopt pattern (Patterson, 1975, p. dophorids and leptolepids (Patterson, 1975). 472) there is a basisphenoid, and paired We surmise that in Hulettia the extent of prootics and opisthotics. SDSM 54361 (fig. perichondral lining was about as it was in 8B) is not sufficiently well preserved to dem- Perleidus or Pholidophorus germanicus (Pat- onstrate those bones, but we infer their pres- terson, 1975, pp. 461, 417), from the vagus ence, an inference supported by an apparent canal up to the lateral part of the roof of the suture between the opisthotic and prootic in braincase. AMNH 10880 (fig. 8A). In the ethmoid re- The ossification pattern ofthe braincase in gion, there is a single ossification, recalling Hulettia is suggested by residual sutures sep- that in Acentrophorus (Patterson, 1975, fig. arating paired epioccipitals and intercalars 136), which probably ossified from paired from a basi-exoccipital in AMNH 10926 (fig. lateral ethmoid centers. Thus Hulettia pre- 9), and by SDSM 54361 (fig. 8B), which shows sents all the endocranial ossification centers an orbitosphenoid and pterosphenoid in the expected in generalized actinopterygians, and 22 AMERICAN MUSEUM NOVITATES NO. 2796

dpcl B brbr sym FIG. 1 1. Hulettia americana. Restoration of dermal skull in (A) dorsal and (B) lateral aspects. Snout bones in A spread in one plane. lacks a supraoccipital, a bone so far known so far as they can be estimated, Hulettia is only in pholidophorids and higher teleosts. generally similar to palaeoniscoids and pho- In the full-grown braincase of Hulettia the lidophorids (Patterson, 1975, p. 287) in hav- bones fused with one another within the otic ing a short occipital region, about 10 percent region and the occipital region, as in other of the total length of the braincase, and the primitive actinopts. Injuvenile specimens 35- postorbital and orbital length about equal, 45 mm SL, the braincase appears to be unos- roughly 40 percent ofthe total length. But the sified or very lightly so. By about 50 mm SL, braincase of Hulettia appears to be broader, at least the sphenotic and pterotic are ossified with the breadth across the postorbital pro- (AMNH 11076). cesses almost 80 percent of the total length, In form and proportions of the braincase, against about 50 percent in pholidophorids. 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 23

L-J:~~~~~~~~~~~~~

FIG. 12. Hulettia americana. A. AMNH 10824. Partial skull roof in ventral aspect with median parietal. x 1.8. B. AMNH 10862. Branchiostegals, mandibles and basihyal toothplate in ventral aspect. x 3.3. C. AMNH 10940. Dermal skull. From Hulett, Wyoming. x 1.8.

The ethmoid ossifications are nowhere well The basisphenoid region is nowhere well preserved, and no details can be described. preserved, but it appears to have been rather The orbitosphenoid (fig. 8) is confined to the extensive, and to contain paired vertical ca- dorsal part of the orbit, and there was no nals for the internal carotids (AMNH 1 1073). ossified interorbital septum. The pterosphe- The autosphenotic forms the postorbital pro- noid contains three foramina. There are two cess, as usual, and is limited posteroventrally near the center of the bone, the larger one by the hyomandibular facet, a space which dorsal to the smaller, presumably respective- was cartilage-filled in life. No foramina have ly for the middle cerebral vein and trochlear been found in the autosphenotic, and the spi- nerve (Patterson, 1975, p. 409). The third racular canal (fig. 8A) enters the prootic, a foramen is posteriorly directed and lies close condition otherwise known only in Lepidotes to the posterodorsal margin of the bone; it (Patterson, 1975, p. 399). probably carried a lateral line nerve to the In the prootic, the lateral commissure is skull roof. short-little more than a strut of cartilage 24 AMERICAN MUSEUM NOVITATES NO. 2796

hh-

FIG. 13. Hulettia americana. A. AMNH 11076. Basihyal toothplate, hypohyals and distal ceratohyals. B. AMNH 10885. Same. C. SDSM 5781. Dissociated snout bones, basihyal toothplate, and mandible. x4.7.

bone overlaid by the ascending process ofthe eminence on the ventral surface of the op- parasphenoid carrying the spiracular groove isthotic, probably marking the ampulla ofthe to the opening ofthe spiracular canal. A shal- posterior semicircular canal, or the posterior low jugular groove leads back from the jug- part of the external canal. The glossopha- ular canal beneath the lateral commissure, ryngeal foramen has not been identified. The and another groove leads dorsally, toward the pterotic is visible in only one specimen (fig. anterior part ofthe hyomandibular facet, im- 8B). It forms the posterodorsal shoulder of mediately behind the lateral commissure. The the otic capsule, and its upper surface has latter groove may mark the course ofthe hyo- thickened areas which contacted the parietal, mandibular nerve. There is no marked sub- separated by grooves for nerves and vessels. temporal fossa. The myodome remains un- On the lateral face of the pterotic, which known. The opisthotic is assumed to ossify presumably formed the medial wall ofa post- from a center over the external semicircular temporal fossa, there is a small, ventrally canal, at the posterior margin of the otic re- directed foramen transmitting the supratem- gion. Here the otic capsule meets the inter- poral branch of the glossopharyngeal to the calar in a rather well-ossified shelf, with skull roof. The posterior face of the pterotic grooves and a small foramen probably mark- appears to be perichondrally lined, and is op- ing the passage of nerves or vessels from the posite the epioccipital in the dorsolateral part vagus canal, which lies medial to this point. of the cranial fissure. In front of these grooves there is a ridge or In the occipital section of the braincase, 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 25

en-p mpt'

ich

hhX d FIG. 14. Hulettia americana. Bones of left palate and hyoid arch in lateral view. The hyomandibula and symplectic and bones of the hyoid bar are shown in their natural relationship, other bones shown as if disarticulated. Foramen for the hyomandibular nerve on the medial face of the bone is indicated by a broken line. Based on AMNH 10831, 10948, 11077, and SDSM 54361. the basi-exoccipital (figs. 8A, 9) shows no sign cartilage bone, with only rudimentary thick- ofseparate basi- and exoccipital components, enings ofmembrane bone rather than the ex- though we assume these existed in early on- tensive outgrowths that develop in most Ju- togeny. The notochordal pit is deep, and par- rassic actinopterygians. The intercalar has a tially lined by notochordal calcification. On posterior facet for tendinous attachment to the ventral surface, there is a deep aortic the descending limb ofthe post-temporal, and groove. The walls of this groove are peri- is perforated by a canal for the supratemporal chondrally lined except at the extreme pos- branch ofthe vagus. The epioccipital is a thin terior end, where a very short aortic canal vertical plate of cartilage bone, perichon- might have ossified later in life (Patterson, drally lined on both the anterior and poste- 1975, p. 319). In the roofofthe aortic groove rior face. On the dorsal surface of the epioc- there is a cuplike housing for the aortic lig- cipital there is a thickening which met the ament, as in other "holostean"-level fishes overlying post-temporal. (Patterson, 1975, p. 321). There is no more The sclerotic ring is ossified in two sections than one pair offoramina for occipital nerves (SDSM 54361), not four as in pholidophorids in the wall of the foramen magnum, again a and other primitive actinopterygians. primitive condition (Patterson, 1975, p. 318). In general, the braincase of Hulettia is that The intercalar consists almost entirely of ofa primitive "holostean"-grade fish (cf. Pat- 26 AMERICAN MUSEUM NOVITATES NO. 2796 terson, 1975, p. 563). Features ofmore prim- internal carotid entered posteriorly, in the itive grade, as in palaeoniscoids, include the transverse crevice behind the raised, toothed open, perichondrally-lined cranial fissure, the boss. pattern of the cartilage bones, especially the The basipterygoid process of the para- large opisthotic and endochondral intercalar sphenoid is moderately long, pointed and and their ontogenetic fusion, the canal for the formed of very thin bone. Presumably, the middle cerebral vein, the position of the metapterygoid articulated with a cartilagi- housing for the aortic ligament in front ofthe nous extension from the basisphenoid lying foramina for the occipital artery and the sin- in the groove on the dorsal surface of this gle occipital nerve. "Holostean"-level fea- dermal process. The ascending process ofthe tures include the slender lateral commissure parasphenoid is slender and passes up the and the presence ofnotochordal calcification lateral commissure to the opening of the spi- in the notochord pit in the basioccipital. De- racular canal in the prootic. Posteriorly, the rived features are the spiracular canal in the parasphenoid extends beyond the ventral otic prootic (otherwise known only in Lepidotes) fissure (the prootic/basioccipital suture), but and the bipartite sclerotic, as in teleosts and does not cover much of the ventral surface various fossil "holosteans" (Edinger, 1929). of the basioccipital. The posterior portion of PARASPHENOID AND VOMER: The para- the parasphenoid, behind the toothed area, sphenoid is exposed in dorsal view (fig. 10) forms roughly one-third ofthe total length of in several specimens, but only one isolated the bone, about as in Perleidus and primitive braincase (AMNH 10880) shows it in ventral pholidophorids. This part of the bone is rel- aspect (fig. 8A). Anteriorly, where it overlies atively shorter in palaeoniscoids, and longer the vomer, the parasphenoid tapers irregu- in other actinopts (Patterson, 1975, p. 527). larly. Beneath the orbit, the bone is excep- The vomer (figs. 8, 13C) is poorly known, tionally broad, and carries an elongate, raised but is exposed in dorsal view in SDSM 5781 patch ofrather large teeth which opposed the and 54361. The bone is median and un- basihyal toothplate. Injuveniles, the teeth are paired, and is an elongate triangle with the arranged in rows radiating from the opening tapering posterior part underlying the tip of of the buccohypophysial canal at the rear of the parasphenoid. Teeth are borne on the an- the toothed area. In the adult, the posterior terior part of the oral surface, but the extent part of the tooth patch, behind the opening of the toothed area is unknown. A median ofthe buccohypophysial canal, is raised on a longitudinal canal for the terminal branches roughly circular boss. Behind this boss there of the palatine nerves passes through the is a deep transverse crevice or pocket, as in anterodorsal part of the bone. Perleidus and parasemionotids (Patterson, The parasphenoid and vomer of Hulettia 1975, p. 536), interpreted as the site of in- show certain derived features. The vomer is sertion of the subcephalic muscles. The dor- median, as in teleosts (including pholidoph- sal surface ofthe parasphenoid (fig. 10) shows orids and Icthyokentema), Lepidotes, Dape- longitudinal grooves for the palatine nerves dium, pycnodonts and Bobasatrania (Patter- and arteries, and paired foramina for the ef- son, 1977, p. 513). In the last four of these, ferent pseudobranchial and internal carotid the vomer bears crushing teeth, whereas in arteries, with bilateral, curved grooves for Hulettia and generalized teleosts the vomer- commissures linking the two arteries, as in ine dentition is unmodified. The parasphe- Amia, Pholidophorus, and Leptolepis (Pat- noid of Hulettia contains foramina for the terson, 1975, figs. 142, 143). On the ventral internal carotid and efferent pseudobranchial surface of the parasphenoid there is no sign arteries. An internal carotid foramen is known ofthese vascular foramina. The efferent pseu- only in teleosts (including pholidophorids and dobranchial and internal carotid arteries ev- Ichthyokentema), Pachycormus, Dapedium, idently passed through the bone almost hor- Boreosomus, and possibly in Saurichthys. An izontally, so that their foramina are hidden efferent pseudobranchial foramen occurs only in ventral view. The efferent pseudobranchial in teleosts (including pholidophorids and artery entered laterally, through a crevice at some individuals of Ichthyokentema) and in the base ofthe basipterygoid process, and the Dapedium (Patterson, 1975). The basipter- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 27 ygoid process of the parasphenoid is rather skulls (e.g., AMNH 10831). The quadrato- large in Hulettia. The distance across the two jugal lies in a groove on the posterolateral processes is about 40 percent of the total face of the quadrate, and is broadest ven- length of the bone, compared with about 33 trally, where it ends in a swollen head but- percent in pholidophorids and 20 percent in tressing the articular condyle ofthe quadrate. leptolepids and Ichthyokentema. But this very The elongate endopterygoid bears small, high ratio in Hulettia is largely due to the pointed teeth over most of its oral surface. breadth of the parasphenoid. The projecting The ectopterygoid and autopalatine are not portion of the basipterygoid process is equal clearly distinguishable from one another in to about 10 percent of the total length of the the few specimens showing them, except by bone in Hulettia, compared with about 15 the contrast between cartilage bone (auto- percent in Pholidophorus germanicus, 8-9 palatine) anteriorly and dermal bone (ecto- percent in leptolepids and 5-6 percent in pterygoid) posteroventrally. It appears that Ichthyokentema. The large dermal basipter- the autopalatine was rather extensive, ex- ygoid process ofHulettia and these primitive tending posteriorly as an ossified area in the teleosts is a derived condition, but an equally palatoquadrate cartilage, which, in turn, lay large dermal basipterygoid process also oc- in the groove formed by the ridge along the curs in Lepidotes (Patterson, 1975, fig. 109). lower margin of the lateral face of the ectop- In other respects, the parasphenoid of Hu- terygoid, as in Amia. No ectopterygoid teeth lettia is relatively primitive. It is broad and are visible. The autopalatine ends in a broad heavily toothed; has a broad, transverse crev- anterodorsal process which articulated with ice for the subcephalic muscles, as in Perlei- the ethmoid. No dermopalatine has been ob- dus and parasemionotids; has a short poste- served in situ, but teeth are visible in the rior portion, like Perleidus and primitive dermopalatine position in some dorsoven- pholidophorids; has an open bucco- trally crushed specimens. Small toothplates hypophyseal canal, and a slender ascending adjacent to the autopalatine in disarticulated process, confined to the spiracular groove. specimens might be dermopalatines or cor- PALATE, HYOID, AND GILL ARCH SKEL- onoids. ETON: Most of the bones of the palate are No interhyal has been seen in the hyoid known by isolated examples in disarticulated arch. The proximal ceratohyal (fig. 14) is skulls (fig. 14). The hyomandibular is broad, semicircular; the distal ceratohyal is broad, curved, inclined somewhat backward, and of short and slightly waisted. There is a single generalized neopterygian type. It is perforat- nodular hypohyal. Many specimens show a ed by a canal for the hyomandibular nerve large, unpaired toothplate between the lower in the usual way. The hyomandibular and jaws (figs. 1 2B, 13). The toothplate is round- symplectic were evidently ossifications with- ed anteriorly, digitate posteriorly, and in ju- in the same cartilage, as in all neopterygians, veniles bears rows of teeth radiating from a since they remain in their natural relation- growth center near the anterior margin (fig. ship (fig. 14) in disarticulated skulls (e.g., 13A, B). The toothplate evidently opposed AMNH 10897a, SDSM 5780). The symplec- the parasphenoid dentition, and lies in front tic is long and tapers distally, where it lies ofthe hypohyals (fig. 13A, B) whose junction close behind the quadrate. The symplectic with the median copula marks the front end shows no sign of a condyle for articulation of the first basibranchial (Nelson, 1969, p. with the lower jaw, such as occurs in hale- 506). We therefore regard the toothplate as a comorphs (Patterson, 1973). basihyal rather than a basibranchial struc- The metapterygoid has a broad dorsal notch ture. There is no sign of an endoskeletal ba- between the medially inclined basal process sihyal (known only in teleosts) beneath the and the vertical posterior part of the bone. toothplate. Among non-teleostean actinopts, There is no sign ofa specialized articular sur- Lepisosteus has a series of paired basihyal face for the basipterygoid process, or of a toothplates, and the Triassic Bobasatrania dermometapterygoid. The triangular quad- (Nielsen, 1952) and Errolichthys (Nielsen, rate is not fused with the quadratojugal as the 1955) have a large median toothplate be- bones are separate in some disarticulated tween the lower jaws. In Bobasatrania and 28 AMERICAN MUSEUM NOVITATES NO. 2796

hyp8

epn 20 hc a$>hpU113(=pUgJ npu6 eph1

hpull 30(=pu9ivpu 6 ph FIG. 15. Hulettia americana. Restoration of axial skeleton (ribs omitted) and occipital ossification of braincase, based mainly on AMNH 10917, SL ca. 95 mm. Elements missing in that specimen (supraneurals, neural arches 12-28, epurals, hypurals 3-8) restored on the basis of specimens shown in figures 16 and 17. Cartilage bone stippled, notochordal calcifications hatched.

Errolichthys this toothplate, which may be not been found, but it was probably covered basibranchial or basihyal or both, opposes a for the most part by the prearticular. The large vomer and lies mainly behind the hy- number of marginal teeth on the dentary ex- pohyals over a single basibranchial ossifica- ceeds 20. tion. The dermal skull of Hulettia can only be Ossified, paired gill-arch elements are pre- characterized as generally neopterygian. The served in several specimens, but we have snout design, including the premaxillae, is learned nothing useful about their pattern. close to the pattern of a parasemionotid fig- No ossified basibranchials have been ob- ured by Patterson (1975, fig. 137), as well as served, or any gill-arch dentition or gill-rak- that of Acentrophorus, the semionotids, ers. Ophiopsis Bartram (1975) and Amia. Unfor- JAWS: The premaxillae meet in the midline tunately, the dermal snout pattern in the pa- throughout most of the length of their nasal laeonisciforms is not well understood, and a processes. In some large specimens they are reasonable approximation of the Hulettia fused together (fig. 13C). The apposition of condition is not known. the nasal processes suggests that the nasal Hulettia, Acentrophorus (Gill, 1923) and septum, which appears to separate these pro- Lepisosteus are the only primitive neopte- cesses in Acentrophorus, was less elevated in rygians known to lack a supramaxilla. The Hulettia (Patterson, 1975, figs. 136, 137). The absence of this superfluous-looking bone, thickened dentigerous portion supports about which is doubled in pholidophorids and in eight sharply pointed teeth. generalized teleosts, is still another primitive The maxilla (fig. 11 B) is entirely separated character and is hence oflittle systematic sig- from the cheek. The internal process, which nificance. In summary, there are no apparent articulates with the vomer and ethmoid, is derived characters in the Hulettia dermal skull attenuated and recurved in typical neopte- that indicate special affinity with any other rygian fashion. The maxilla expands poste- known actinopterygian. riorly to a slight degree, and extends in that AXIAL SKELETON: The axial skeleton is par- direction about as far as the apex ofthe man- tially visible in immature individuals, 35-50 dibular coronoid process. There is no indi- mm SL, in which the squamation is incom- cation ofa supramaxilla. About 13 maxillary pletely developed (fig. 16), and in two mature teeth have the same shape as the premaxillary specimens in which the squamation was ones but decrease in size posteriorly. stripped off postmortem, AMNH 10917, ca. The mandible has a well-developed coro- 95 mm SL (fig. 15 is mainly based on this noid process (fig. 13C). Laterally it exhibits specimen), and AMNH 10875, ca. 105 mm the usual three primitive elements-dentary, SL (fig. 17), which is poorly preserved. angular, and supra-angular (Patterson, 1982, The boundary between the ural and preural p. 248). A well-preserved medial surface has regions ofthe vertebral column is marked by 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 29

dr I

hy 1 hpu 14or8

pv ar

FIG. 16. Hulettia americana. Axial skeleton as preserved in three immature specimens: A, AMNH 10860, SL ca. 50 mm; B, AMNH 11075b, SL ca. 50 mm, scales drawn in transparency; C, AMNH 11072, SL ca. 41 mm. Bars equal 5 mm. the bifurcation of the dorsal aorta (Nybelin, sition in the first hypural (Patterson, 1968, 1963). Behind this point, the aortic canal in fig. 2B). In AMNH 10917, as in P. bechei, the the haemal arch is eliminated, and the (pre- first hypural is the fifth bone supporting lower ural) haemal spines are replaced by the (ural) caudal fin-rays. In other specimens of Hu- hypurals. In fossil non-teleost actinopts, the lettia we have recognized the preural/ural location of the bifurcation is not always ev- boundary by comparison with AMNH 10917. ident. In Amia (Nybelin, 1963, fig. 16) and On this basis, there are 38 preural vertebrae in some individuals of Lepisosteus (Nybelin, in AMNH 10917, and 38 or 39 in AMNH 1977, fig. 2) it is marked by an anterior em- 10875, the only two specimens in which the bayment in the proximal part of the first hy- entire column is visible. There was thus one pural with an anterior process below it. In vertebra to each scale row, a relationship also some fossil forms an anterior process also shown by the translucent squamation of im- occurs on several preceding preural haemal mature individuals (fig. 16B). In AMNH arches (e.g., Patterson, 1968, figs. 1-4; Ny- 10917, the 38 vertebrae comprise 25 abdom- belin, 1977, fig. 3). In Hulettia, mature in- inals and 13 caudals (with haemal arches and dividuals show similar anterior processes (figs. without ribs), but the number of caudals ap- 15, 17), and we designate the last bone with parently ranged from 12 (fig. 17) to 14 (fig. a process as the first hypural. In AMNH 10917 16C). The preural count of 38-39 is thus 25- (fig. 15) this bone contains a foramen at the 27 abdominals plus 12-14 caudals. level of the haemal canal. In Pholidophorus There is a marked difference between the bechei a foramen may also occur in this po- vertebrae of immature and mature individ- 30 AMERICAN MUSEUM NOVITATES NO. 2796

,..~ A-C *L--JD 2mm B D

epn epn FIG. 17. A-C. Hulettia americana, axial skeleton ofAMNH 10875, SL ca. 105 mm: A, interpretation of caudal vertebrae and hypurals as preserved; B, ventral hemichordacentrum and parapophyses of vertebra 22, from the posterior abdominal region, in anterior view; C, detached abdominal neural arches in lateral aspect, the first lying below vertebrae 8-9, the second above vertebrae 18-19, the third above vertebrae 15-16. D. Pholidophorus bechei, a neural arch from the posterior abdominal region ofan acid- prepared BMNH specimen, Lower Lias, Lyme Regis, Dorset. uals. Immature specimens (fig. 16) have no In the two adult specimens, the neural arches centra, no ossified interdorsals or interven- are missing in AMNH 10917 from vertebra trals (intercalaries), and the vertebrae appear 12 to 28 (from PU 11 forward), and are dis- monospondylous. In mature individuals (figs. articulated and scattered from PU 13 forward 15, 17) there are ossified interdorsals in front in AMNH 10875 (fig. 17). The first few neural of each neural arch except the first and the arches ofAMNH 10917 show traces of short last four or five. There are ossified interven- posterolateral processes. However, some of trals in the midcaudal region and crescentic the disarticulated neural arches in AMNH dorsal and ventral hemichordacentra, which 10875 (fig. 17C) have relatively longer pro- have a diplospondylous relationship. The cesses, suggesting that these outgrowths ofthe adult vertebral column will be described first, neural arches were present throughout much and ontogenetic differences in young individ- ofthe abdominal region. We assume that the uals will be noted afterwards. posterolateral processes are homologous with Above the notochord, the paired neural the short epineural processes on the anterior arches are separate from each other back to neural arches of various palaeoniscoids and the origin ofthe dorsal fin at vertebra 20-21, other fossil actinopts (Patterson, 1973, p. 237; and are fused distally into a median neural Rosen et al., 1981, p. 244), and with the much spine beyond that point (fig. 16B). There are longer epineural processes ofpholidophorids rodlike median supraneurals distal to the first (fig. 17D) and other primitive teleosts. The 20-21 neural spines, the last supraneural lying last few neural spines decrease progressively just in front ofthe first dorsal radial (fig. 16A). in length, and the diminutive last neural arch 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 31 and spine is that of PU2 (figs. 16B, 17A); in above), there are at least eight ossified hy- AMNH 10917 (fig. 15) the last five neural purals (figs. 16B, 17A). No distal caudal ra- arches are concentrated so that they oppose dials have been observed. only three haemal arches. There are no os- Paired interventrals (ventral intercalaries) sified ural neural arches. Epurals are pre- are present in front of the haemal arches of served only in one immature specimen (fig. PU6-1 1 (AMNH 10917, fig. 15) or PU5-10 16B), where there are seven. The first epural (AMNH 10875, fig. 17A). They are small appears to be aligned with the short neural blocks of cartilage bone resembling the in- spine of the last ossified neural arch. terdorsals. The paired interdorsals, which are small The centra of adult specimens consist of blocks of cartilage bone, are present in front separate dorsal and ventral hemichordacen- of each neural arch except the first and the tra. The ventral crescents are larger than the last four (AMNH 10875) or five (AMNH dorsal (figs. 15, 17A). A dorsal hemichor- 10917). Their relationship with the abdom- dacentrum joins each pair of interdorsals inal neural arches and parapophyses is not across the midline. The only neural arches clear and it is not possible to determine bearing hemichordacentra are those in the whether the abdominal interdorsals lay close midcaudal region between PU5 and PU10 in front of the neural arch bases as in Amia, (fig. 15). In these five segments the dorsal close behind them as in Australosomus (Niel- hemichordacentra are diplospondylous.3 The sen, 1949), or whether the arrangement was loss or disarticulation ofthe precaudal neural varied along the column. In figure 15 the arches in the adult specimens indicates that neural arch bases and interdorsals are shown they had no intimate connection with the no- as equally spaced. The abdominal interdor- tochordal sheath. sals either opposed or alternated with the par- Large ventral hemichordacentra are asso- apophyses beneath the notochord. As pre- ciated with each pair of parapophyses and served in the two adult specimens, they appear with the haemal arches from the occiput back to alternate, but this could be due to crushing to PU2 (AMNH 10917) or PU4 (AMNH following preservation. The arrangement 10875), except for the haemal arch of PU10 shown in figure 15 is therefore conjectural. (AMNH 10917) or PU9 (AMNH 10875) (see In the caudal region, the interdorsals lie mid- below). Each ofthe five pairs ofinterventrals way between the neural and haemal arches. in the midcaudal region, PUS-10 or PU6- There are paired parapophyses in the ab- 11, carries a hemichordacentrum. These five dominal region beneath the notochord. Ex- segments are thus diplospondylous, each with cept for the first pair, they all bear lateral two dorsal and two ventral hemichordacen- processes (fig. 1 7B) with an articular facet for tra. the long, rather stout pleural ribs. At the tran- In the middle of the diplospondylous re- sition from the abdominal to the caudal region, the dorsal and ventral hemichordacen- gion (ca. PU 14), vertebra 26 or 27 has a pair tra are in contact or nearly so (figs. 15, 17), of enlarged, elongated and waisted parapo- but are not actually fused to form a ring cen- physes without ribs. These appear to meet trum. The hemichordacentra are larger and distally and form a haemal canal in some specimens (fig. 16A, C), but to be separated in others (fig. 16B). We regard this vertebra 3By diplospondyly we mean two centra or two pairs as the first caudal one. The succeeding ver- ofhemicentra to each segment. In actinopts, diplospon- tebra has a shorter haemal arch bearing a dyly is confined to the midcaudal region, where one cen- separate median haemal spine, which lies be- trum or pair of hemicentra is associated with the inter- hind the proximal ends of the first two anal segmental neural and haemal arch and one with the intrasegmental intercalaries. This type ofdiplospondyly fin radials (fig. 1 6A, C). The remaining caudal does not occur in teleosts above the pholidophorid level. haemal arches bear long haemal spines, the It is known in the Triassic pholidopleurid Australosomus last four of which are expanded distally to (Nielsen, 1949), in Caturus (Rosen et al., 1981, fig. 59), support lower caudal fin-rays. If the ural/ in Amia and fossil amioids, in Notagogus (Saint-Seine, preural boundary is correctly identified (see 1949, fig. 86), and in pholidophorids (Patterson, 1968). 32 AMERICAN MUSEUM NOVITATES NO. 2796 thicker throughout the column in AMNH in few fossil lower actinopts, where it is cov- 10875 (ca. 105 mm SL, fig. 17A, B) than in ered by ganoid scales. the smaller AMNH 10917 (ca. 95 mm SL, Among the eight features listed above, the fig. 15). In the larger specimen the ventral interdorsals (1), interventrals (2), and mid- hemichordacentra in the first 10 to 12 ver- caudal diplospondyly are surely primitive tebrae extend well above the horizontal mid- features, widely distributed in lower actin- line of the notochord. It appears that calci- opts and other fishes (Schaeffer, 1967b; An- fication of the chordacentra began in the drews, 1977; Rosen et al., 1981). Hemichor- midcaudal region, where they are largest dacentra (3) and epineural processes (4) recall (AMNH 10917; fig. 15) and then progressed the pholidophorids, but hemichordacentra are backward from the occiput. In AMNH 10917 much more widely distributed (Australoso- the haemal arch of PU10 may lack a chor- mus-Nielsen, 1949; Turseodus-Schaeffer, dacentrum, and since the same may be true 1967a; macrosemiids-Bartram, 1977; Ca- for PU9 in AMNH 10875, this absence is turus -Rosen et al., 1981, fig. 59; and prob- unlikely to be an artifact ofpreservation. This ably Haplolepis- Baum and Lund, 1974). The condition indicates that PU9 or 10 was the epineural processes of Hulettia are much last centrum to calcify, marking the junction shorter than those ofpholidophorids (fig. 17C, between a possible rostrocaudal gradient from D), yet, at present, such outgrowths of the the occiput with a caudorostral gradient from neural arches are known to occur throughout the midcaudal region. the abdominal region only in pholidophorids The first vertebra is poorly preserved in and more derived teleosts (Patterson, 1977, both adult specimens, but it appears to have fig. 19). Epineurals are unknown in cladis- extensive bases to the neural arch and par- tically more primitive teleosts (sensu lato) apophysis that almost surround the noto- such as ichthyokentemids, pleuropholids, chord. aspidorhynchids and pachycormids. How- In young specimens (up to 50 mm SL, fig. ever, epineural processes occur in the lower 16), the vertebral column differs from the actinopterygians Mimia, Boreosomus, Aus- adult condition in lacking any trace of chor- tralosomus, and Caturus. In Mimia they are dacentra or ofinterdorsals and midcaudal in- present on the first 14 neural arches (Gardi- terventrals, so that the column appears mon- ner, 1984, p. 363). ospondylous. Interdorsals and interventrals The unpaired neural spines of Hulettia are were presumably present in cartilage. Epi- a halecostome feature (Patterson, 1973). The neural processes on the abdominal neural seven epurals, and the fact that the supra- arches are suggested in some immature spec- neurals do not extend beyond the dorsal fin imens (fig. 16A, middle ofthe abdominal re- origin, are derived characters relative to the gion), but are much less prominent than in numerous epurals and supraneurals of chon- the adult (fig. 17C). We assume that these drosteans and palaeoniscoids, but are gen- too were cartilaginous in the young. eralized within halecostomes (there are seven The most notable features ofthe axial skel- epurals in Pachycormus, Dapedium and Io- eton of Hulettia are: (1) a complete series of noscopus-Patterson, 1973). The absence of dorsal intercalaries; (2) ventral intercalaries ossified ural neural arches means only that in the midcaudal region only; (3) dorsal and their form cannot be used to relate Hulettia ventral hemichordacentra which are diplo- to actinopterygian groups such as teleosts or spondylous in the midcaudal region; (4) epi- caturids. neural processes on most of the abdominal In actinopterygians, vertebral centra may neural arches; (5) the series of supraneurals be composed entirely of chordal tissue (e.g., ends in front ofthe dorsal fin, and the neural Hulettia, Caturus, and Pholidophorus); of a spines are median from there onward; (6) at combination ofchordal and perichordal (e.g., least seven epurals; and (8) no ossified ural Aspidorhynchus, some leptolepids and some neural arches. It is notable that the first four larval teleosts); or only of perichordal (e.g., of these are undeveloped in immature indi- Polypterus, Amia, and some teleosts). Among viduals. Comparisons with other fishes are fossil paleopterygians, Pygopterus (Aldinger, difficult because the axial skeleton is known 1937); Haplolepis (Baum and Lund, 1974); 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 33

Turseodus (Schaeffer, 1967a); Australosomus PAIRED FINS: The endoskeletal shoulder (Nielsen, 1949); and Macroaethes (Wade, girdle of Hulettia remains unknown so im- 1935) have entirely chordal centra (as de- portant characters involving the mesocora- duced from the histology) along with Tetra- coid arch cannot be discussed (Jessen, 1972; gonolepis, Pachycormus (Patterson, 1973, p. Patterson, 1973). As shown in figure 11, the 274) and, according to Bartram (1977, p. 215), dermal elements-extrascapular, post-tem- Propterus, Catervariolus and Euthynotus. In poral (suprascapular), supracleithrum and our view, the distinction between chordal and cleithrum-require little comment. The ex- perichordal centra needs restudy in these and trascapulars meet in the midline. The post- other Mesozoic actinopterygians. Among the temporals diverge at about 45 degrees, have teleosts (as defined by Patterson, 1973), a well-developed internal process, and barely pholidophorids, pachycormids, and some touch just behind the extrascapulars. The aspidorhynchids and leptolepids have only cleithrum has a prominent internal lamina, chordal centra. Aspidorhynchus comptoni which is a primitive actinopterygian char- centra have an inner chordal ring and an out- acter. There is no evidence ofa clavicle or of er perichordal bony one (AMNH 6363). a "serrated appendage" (Bartram, 1977). The It is increasingly evident that the hemi- or dorsal and ventral postcleithra have about annular centra ofpaleopterygians, ifthey de- the same relative proportions as in, for in- velop at all during ontogeny, are invariably stance, Ichthyokentema (Griffith and Patter- chordal, apparently with no perichordal ad- son, 1963), except that the ventral one pro- dition. Among the neopterygians, as noted jects farther posteriorly and has a rounded above, this is also the case in Hulettia and in free border. various other Mesozoic forms. The pectoral fin is bordered anteriorly by In Hulettia, as shown in the restoration of about five basal fulcra and by fringing fulcra the axial skeleton (fig. 15), all the haemal that extend the entire length of the anterior- components are composed of replacement most ray. It contains about 13 principal rays, bone and are in contact with the narrow all jointed. hemichordacentra. The same is true for the The pelvic plates lie beneath vertebrae 18- dorsal and the relatively few ventral inter- 20, and are more or less hourglass-shaped calaries. It is of some interest that in diplo- (fig. 16). The pelvic fins have only a few basal spondylous actinopterygians where the entire fulcra, numerous fringing fulcra, and eight or vertebral column is known, chordal hemi- nine segmented rays. The origin ofthe pelvic centra are absent beneath the neural arch basis almost exactly halfway between the origin es except for those in the midcaudal dip- ofthe pectoral and the base ofthe hypocaudal lospondylous region. Chordal hemicentra are lobe. present, however, below all the dorsal inter- UNPAIRED FINS: The dorsal fin originates calaries. This indicates that, as in Caturus at about the level of vertebra 23, behind the (Rosen et al., 1981, fig. 59) and in pholido- pelvics. The fin contains eight or nine seg- phorids (sensu lato), the hemichordacentra mented rays, the first of which may or may in the abdominal region were derived entirely not be branched, preceded by four or five from the chordal tissue related to the inter- unsegmented basal fulcra. Fringing fulcra are calaries. In Recent teleosts, where intercala- interposed between the last two basal fulcra, ries are absent, the chordacentra develop only and are continued up the leading edge of the from ventral hemichordacentra (Fran9ois, first one or two segmented rays. The fin is 1966, 1967; Cavender, 1970). The absence preceded by an enlarged ridge scale, and a of chordal calcifications related to most of row of about six small, rounded scales ex- the neural arches in Hulettia probably rep- tends back along each side of the base of the resents the primitive actinopterygian condi- fin. The dorsal fin is supported by about 11 tion. The variations in intercalary and cen- radials (fig. 16A). The first radial has a large, trum development in the non-teleost platelike anterodorsal expansion, supports the actinopterygians present a fascinating prob- basal fulcra and the first ray, and ends proxi- lem in pattern formation, which we plan to mally behind the neural spine of vertebra 20 treat elsewhere. or 21. Ossified middle segments are pre- 34 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 18. Hulettia americana. Restoration ofentire fish. Squamation based mainly on AMNH 10903. served distal to the fifth and succeeding ra- scale row in the axial lobe, and there are no dials. epaxial fin-rays (cf. Bartram, 1977, p. 218). The anal fin originates at about the level The caudal is bordered dorsally by an en- of vertebra 29 (i.e., beneath PU10-1 1), mid- larged scute followed by seven or eight robust way between the pelvics and the base of the basal fulcra (unpaired, unsegmented) and by hypochordal lobe ofthe tail. The fin contains fringing fulcra. The ventral border of the fin seven or eight branched and segmented rays, also carries fringing fulcra and is preceded by preceded by three or four unsegmented basal a large scute and six or seven narrower basal fulcra and three enlarged ridge scales. Fring- fulcra, the last three segmented distally. ing fulcra are interposed between the last two The relatively low number of principal basal fulcra and continue up the leading edge caudal rays (19-20), with only eight in the ofthe first ray. The fin is supported by seven lower lobe, distinguishes Hulettia from pho- or eight radials (fig. 16B, C). The long first lidophorids, which have 20-27 principal cau- radial ends beneath the first haemal arch and dal rays, with 10 or more in the lower lobe supports the basal fulcra and the first ray. No (Patterson, 1968). In Ichthyokentema there ossified middle segments are preserved in the are 19 principal caudal rays (Griffith and Pat- anal fin. terson, 1963), as in Hulettia. There are fewer CAUDAL FIN: The hypochordal lobe of the rays in pleuropholids (ca. 15) and aspido- caudal fin is supported by four elongated pre- rhynchids (ca. 14), and many more in pa- ural haemal spines and eight hypurals (see p. chycormids (Wenz, 1968). In macrosemiids 29 for discussion of the problem of iden- there are always eight lower principal rays, tifying the ural/preural boundary and the de- as in Hulettia, but the upper lobe contains scription of the caudal skeleton). The occur- only three to eight (Bartram, 1977). Caturus rence of only four preural haemal spines has about 25 principal caudal rays, Dape- supporting caudal fin-rays in Hulettia ap- dium about 24 (Gill, 1923; Wenz, 1968), Lep- pears to be a rare condition among non-te- idotes and Semionotus ca. 17-19, and Acen- leost neopterygians, although the number may trophorus 17. range from three to nine (fig. 15). The caudal skeleton and fin ofHulettia may The caudal fin is nearly equilobate exter- be regarded as a primitive actinopterygian nally and includes 19 to 20 principal caudal one in terms of the little modified haemal rays, with 10 or 11 branched rays in the upper spines supporting the caudal rays, in the ab- lobe and seven in the lower. As in other prim- sence of epaxial fin-rays, and in the presence itive neopterygians, the uppermost caudal of well-developed basal and fringing fulcra. rays decrease in size upward, and the last one The absence of ural centra is probably also is difficult to distinguish from the axial lobe primitive. The absence ofossified ural neural squamation proximally. The uppermost ray arches, as noted above, is simply uninfor- is continuous proximally, with the longest mative. The occurrence of seven epurals is 1 984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 35

matched in a wide variety of neopterygians and Recent teleosts (van Oosten, 1957), the (e.g., Dapedium, Pachycormus, Ionoscopus). lateral line scales develop first: this is shown, This number is derived relative to the much for example, by AMNH 10919, ca. 45 mm larger number in some chondrosteans (e.g., SL, which has lateral line scales as far forward 15 in Pteronisculus, 30 in Acipenser), but as the origin ofthe dorsal fin, but shows scales primitive relative to the four or five in pho- above and below the lateral line only from lidophorids and the three in Recent teleosts. the middle of the anal fin backward. In SQUAMATION: The body scales (fig. 18) are Brookvalia Wade (1935) found that scales first rhomboidal, have well-developed peg-and- appeared dorsal and ventral to the lateral line socket articulations, a smooth posterior bor- scales at both ends of the trunk, but in Hu- der and lack a dentine layer. In the flank area lettia we have seen no signs ofanterior scales the scales are about twice as deep as wide. in incompletely scaled individuals, and the Those covering the abdominal area are much squamation appears to have developed from smaller and narrower. All the fins are bor- the tail forward, as is general in teleosts (van dered by both basal and fringing fulcral scales Oosten, 1957). In AMNH 10885, ca. 65 mm (see fin descriptions for fulcral counts). Dor- SL, the squamation is complete on the flank, sal ridge scales are absent. In well-preserved but from the anal fin forward the scales do specimens from both the Sundance and the not reach the dorsal or ventral borders of the Todilto the scale count along the lateral line trunk. By about 70-80 mm SL, the squa- is 37-38. The origin ofthe pelvic fin is behind mation is complete. the eleventh oblique scale row; that of the When scales first appear, they lack ganoin, dorsal and anal behind the twenty-first or as in Brookvalia (Wade, 1935, p. 27). On twenty-second oblique scale row. The appar- individual scales, ganoin first appears as a ent constancy in fin position in the Sundance spot at the posterior point or margin of the and Todilto specimens is also indicated by scale, and spreads forward from there the regressions ofbody length to fin position (AMNH 1 1075a). It develops first on the dor- (fig. 7), and by immature specimens in which sal and ventral caudal scutes, then on the vertebral counts can be made (fig. 16B). caudal lateral line scales, and spreads forward ONTOGENETIC CHANGES: Our material of dorsally and ventrally from there (AMNH Hulettia americana includes individuals 10885). ranging in standard length from about 35 mm The cranial dermal bones and the fin-rays, to over 150 mm. We have noted no differ- like the scales, lack ganoin in small individ- ences that can be interpreted as ontogenetic uals. On the frontal, for example, AMNH among specimens more than about 70 mm 11075b (ca. 50 mm SL) shows a small patch in standard length, but smaller specimens dif- ofsmooth ganoin close to the inferred growth fer from the adult condition in squamation, center of the bone, but medial to the sensory fulcra, ornamentation of the dermal bones, canal. On other bones, such as the cleithrum and in the axial skeleton (the latter described and supracleithrum, ganoin first appears as under that heading). dorsoventral strips or ridges, whereas on the In the smallest specimens (e.g., AMNH opercular it appears in ridges concentric with 11075a, ca. 35 mm SL), there are no scales the growth center at the articulation with the apart from the large scutes above and below hyomandibular. The fin-rays develop ganoin the caudal skeleton. Scales develop in larger at about the same time as the scales adjacent specimens from the axial lobe of the tail for- to them. Fringing fulcra seem to be absent ward. They extend to the posterior end ofthe on all the fins in the smallest specimens, and anal fin insertion in specimens that range from first appear at about 40-50 mm SL. 40 mm SL(AMNH 11072,11437)to 50 mm The fact that the scales, fin-rays and cranial SL (AMNH 11075b, fig. 16B). One individ- dermal bones of Hulettia are well formed in ual (AMNH 10860, ca. 50 mm SL, fig. SA) bone before any ganoin is added to them has scales as far forward as the origin of the matches Meinke's (1982) account ofthe der- anal fin. As in the Triassic redfieldiid Brook- mal skeleton in Polypterus, where ganoin de- valia (Wade, 1935), the early Cretaceous pho- velops after bone, and is initially separate lidophorid Wadeichthys (Waldman, 1971), from it. 36 AMERICAN MUSEUM NOVITATES NO. 2796

RELATIONSHIPS OF HULETTIA this interpretation of BMNH 37094) and teleosts; (4) a foramen for the internal carotid Although Hulettia is now one ofthe better in the parasphenoid, as in Pachycormus, Da- known Mesozoic actinopterygians, its rela- pedium, Boreosomus, and teleosts; (5) a long tionships remain enigmatic. To recapitulate dermal basipterygoid process, as in Lepidotes our comments in the descriptive sections, and teleosts; (6) no gular, which is also absent Hulettia retains many primitive features in- in Lepidotes, macrosemiids, pycnodonts, as- cluding: a large rostral separating the nasals; pidorhynchids, and many higher teleosts; (7) a fully open cranial fissure; an endochondral a large basihyal toothplate, as in Lepisosteus, intercalar without extensive membrane bone many teleosts, and probably in Bobasatrania outgrowths; no supraoccipital; a broad para- and Errolichthys. Among the fishes sharing sphenoid with an open bucco-hypophysial derived characters with Hulettia are Lepi- canal, a wide transverse crevice for subce- dotes (spiracular canal, vomer, basipterygoid phalic muscles and a short posterior part; no process, no gular), Dapedium (vomer, foram- supramaxilla; a complete series of dorsal in- ina in parasphenoid) and teleosts, including tercalaries; midcaudal diplospondyly; and a pholidophorids (vomer, foramina in para- long series of unmodified hypurals. These sphenoid, basipterygoid process). features have varied distributions among ac- In the currently accepted scheme of tinopterygians, but are found together only neopterygian relationships (Patterson, 1973, in "chondrosteans" such as Perleidus or Aus- 1977; Wiley, 1976, 1979; Lauder, 1982), with tralosomus. Amia as the sister group of teleosts among However, Hulettia has a number of de- Recent fishes, pholidophorid-level teleosts are rived features, which show that it belongs in widely separated from Lepidotes and Dape- the Neopterygii. These neopterygian features dium, so that the characters shared by Hu- include: premaxilla with nasal process; ant- lettia with pholidophorid-level teleosts and orbital with long rostral process; one large with Lepidotes or Dapedium would be treated suborbital plus one to three smaller ones; free as conflicting, and therefore uninformative. maxilla with an internal head; narrow lateral In an alternative scheme proposed by P. E. commissure; two-part sclerotic; vertical sus- Olsen (personal commun.), the splintlike pensorium with a narrow, crescentic pre- quadratojugal ofLepidotes, Semionotus, Da- opercular; an interopercular; symplectic de- pedium, Lepisosteus, macrosemiids, Hulettia veloped in the same cartilage as the and teleosts (where it is fused to the quadrate) hyomandibular, a splintlike quadratojugal; ar- is regarded as a synapomorphy relating these ticular with a coronoid process; unpaired fishes. neural spines from the origin ofthe dorsal fin We cannot evaluate Olsen's scheme in de- onward; notochordal hemicentra; no clav- tail here, but if it were accepted, the two fo- icle; fin-rays equal in number to their sup- ramina in the parasphenoid of Dapedium, ports in the dorsal and anal fin; axial lobe of Hulettia, and teleosts would also emerge as tail reduced; and no dentine layer in the scales. synapomorphous relative to the single fora- Most ofthe neopterygian characters are widely men in Pachycormus and the lack of foram- distributed among "holosteans," but three of ina in Lepisosteus and Lepidotes. We foresee them, the free maxilla, the interopercular, and problems with this pattern of relationships. the median neural spines are defining char- One relates to the jawjoint. The paleoniscoid acters of the Halecostomi (Patterson, 1973). Pteronisculus is described (Nielsen, 1942, p. Finally, Hulettia shows a few derived char- 175, figs. 35, 36) as having an articulation acters of more restricted distribution. These between the symplectic (interhyal of Patter- include: (1) spiracular canal entering prootic, son, 1982, fig. 1) and the posteromedial sur- as in Lepidotes; (2) vomer median, as in Lep- face of the articular. This might be taken as idotes, Dapedium, the pycnodonts, Bobasa- homologous with the symplectic-articular trania and teleosts; (3) a foramen for the ef- joint in Amia, caturids, and parasemionotids ferent pseudobranchial artery in the (Patterson, 1973). However, the contact sur- parasphenoid, as in Dapedium, perhaps Mac- faces of the two bones in Pteronisculus are rosemius (Bartram, 1977, p. 144; but we doubt perichondrally lined, not cartilage-faced as in 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 37

FIG. 19. Lepidotes sp. AMNH 11442. Partial, dissociated skull. x 0.8. From Hulett, Wyoming.

true joint surfaces. In the Devonian paleo- HALECOSTOMI niscoids Mimia and Moythomasia (Gardi- SEMIONOTIDAE WOODWARD, 1890 ner, 1984, p. 346) the interhyal (=symplectic GENUS LEPIDOTES AGASSIZ, 1832 in Nielsen's usage) does not articulate with DIAGNOSIS: See Woodward, 1895, 1916; the lower jaw. Since Gardiner finds that Lehman, 1966. Mimia and Moythomasia are successive sis- TYPE SPECIES: Cyprinus elvensis De Blain- ter taxa to the remaining actinopterygians, ville. their lack of a hyoid arch-lower jaw contact DISTRIBUTION: See Lehman, 1966, p. 160; appears to be the primitive actinopterygian 1982. condition. On this basis, we believe that the Wang, 1974; Arratia, 1981; Gayet, symplectic-articular joint in halecomorphs (amiids, caturids, parasemionotids) is syn- Lepidotes sp. apomorphic for this group. Figure 19 At present, we regard the characters that HORIZON AND LOCALITY: Stockade Beaver Hulettia shares with Dapedium, Lepidotes, Member, Sundance Formation, Burnt Hol- and the pholidophorid-level teleosts as mu- low, near Hulett, Crook County, Wyoming tually incongruent. This means that Hulettia, and Sundance Formation at mouth ofClark's although a halecostome, shows no unambig- Fork Canyon, Park County, Wyoming. For uous synapomorphy with any halecostome details see section on geologic occurrence. subgroup. Taxonomically, our options are to REFERRED SPECIMENS: From the Sundance leave it as a genus incertae sedis in the Hal- Formation, Stockade Beaver Shale Member, ecostomi, or to name a monotypic family or near Hulett, Wyoming: AMNH 11442, higher taxon for it, in order to emphasize its 11440. From Rierdon Formation, Clark's unique features. Given our present ignorance Fork Canyon, Wyoming: AMNH 11441. ofthe detailed structure of Mesozoic neopte- DISCUSSION: The tentative identification of rygians, we can see no advantage in naming these specimens as Lepidotes is based pri- a higher taxon. marily on the peglike, weakly tumid form and 38 AMERICAN MUSEUM NOVITATES NO. 2796 relative size of the teeth on the premaxilla, Caturus dartoni (Eastman) maxilla, dentary, and dermopalatine (AMNH Figures 20, 21 11442). In this genus both the inner and outer Amiopsis (?) dartoni Eastman, 1899b, p. 406. teeth are generally obtuse, and in some species Amiopsis dartoni Eastman; Boreske, 1974, p. 75. more tumid rather than acuminate, as in its probable sister genus Semionotus. Several HORIZON AND LOCALITY: Canyon Springs anterior elements in the infraorbital series are Sandstone Member, Sundance Formation preserved in AMNH 11442 and 11441 along near Hot Springs, South Dakota. Stockade with the c-shaped antorbital. The cheek area, Beaver Member, Sundance Formation, near including the suborbitals, is broken away. The Hulett, Wyoming. Todilto Limestone Mem- short, posteriorly broadened and rounded ber, Wanakah Formation, Bull Canyon, Gua- maxilla is characteristic ofboth Lepidotes and dalupe County, New Mexico. For additional Semionotus along with the robust premaxilla, details see section on geologic occurrence. which has an elongated, perforated nasal pro- HOLOTYPE: NMNH 4792. cess. The short mandible with its articulation REFERRED SPECIMENS: From the Sundance under the orbit is also characteristic of both Formation, Canyon Springs Sandstone taxa, as is the narrow, gently recurved pre- Member near Hot Springs, South Dakota: opercular. The scales in all three Sundance NMNH 4793, 4794; MCZ 9696. From the specimens are smooth and their posterior Sundance Formation, Stockade Beaver Shale borders are not serrated. Member, near Hulett, Wyoming: AMNH At present there is no possibility ofrelating 10975, 10994, 10996. From Wanakah For- the Sundance Lepidotes to any of the 96 mation, Todilto Limestone Member, Bull species of this genus listed by Woodward Canyon area, New Mexico: BHI 953F. (1895), plus numerous others described since DISCUSSION: The specimens from the that time. Woodward (1895) concluded that Sundance Formation and the Todilto Lime- the Rhaetic-Jurassic species tend to have sty- stone have been identified as Caturus based liform or weakly tumid, obtuse teeth with on thejaws, dentition, branchiostegal ray size slender pedicles, whereas the known Creta- and number (at least 24), the presence of ceous species have outer and inner teeth that hemichordacentra, the fusion of hypurals are more robust with decidedly tumid to mo- 1-3 (AMNH 10975), the relative size of the lariform crowns and relatively short pedicles. paired fins (pectorals much larger than pel- On the basis of Woodward's limited obser- vics, both with delicate fringing fulcra), and vations this appears to be plausible, but far thin cycloidal scales that break up into sin- more comparison is needed for substantia- uous rods (fig. 20B; Schultze, 1966, figs. 3, tion. 51; Bartram, 1977, p. 219). The pectoral fin has about 24 rays, the pelvic has eight to ten, the dorsal approximately HALECOMORPHI 20 and the anal about 13. These counts are CATURIDAE OWEN, 1860 all within the range of the European species, GENUS CATURUS, AGASSIZ, 1834 including the type species C. furcatus from the Kimmeridgian (Woodward, 1895; Saint- DIAGNOSIS: A caturid genus distinguish- Seine, 1949). The mostly complete Wanakah able from other genera in this family (Furo, specimen has a standard length of about 30 Heterolepidotus, and Osteorachis-see Pat- cm, with the distance between the origin of terson, 1973, 1975; Bartram, 1975) by the the pectoral and pelvic fins measuring about characters listed by Gardiner (1960, p. 293) 10 cm. In the Sundance individual this dis- plus the presence of multiple, irregular su- tance is about 13 cm. praorbitals arranged in two or three rows. According to Patterson (1973) and Bar- TYPE SPECIES: Caturusfurcatus Agassiz. tram (1975), Eoeugnathus, Sinoeugnathus, DISTRIBUTION: Lower Liassic to Wealden. Allolepidotus, Eurycormus, Callopterus, Lo- Western Europe, Cuba, Brazil, eastern phiostomus, and Macrepistius should be re- Greenland, western United States. moved from the Caturidae and assigned to 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 39

;::tzM0, i R t (

41. .. ." ",,loot ,,,I Ii..

.1 .- w A,

B 7-J FIG. 20. Caturus dartoni. A. BHI 953F. From Bull Canyon, New Mexico. x.38. B. AMNH 10994. Isolated scales from Hulett area, Wyoming. x 3.25. other families. The remaining five genera- observed fusion between the anterior hypu- Caturus, Furo, Heterolepidotus, Osteorhach- rals in C. smithwoodwardi White (1925) from is, and Neorhombolepis-may represent a the German Lias (Holzmaden). However, the grade group related to the amiids or the first three hypurals are clearly separate in three ophiopsids, or they may turn out to be a Kimmeridgian species; C. furcatus (type- monophyletic group, if they all share with species), C. pachyurus Agassiz from the Ba- Caturus paired, blocklike ural-neural arches. varian lithographic limestone, and C. driani Caturus is apparently distinguished from the from Cerin, France, and in one Lower Cre- other genera in this assemblage by having taceous species; C. tarraconensis Sauvage more than one row of supraorbitals, and from Lerida, Spain. Among over 20 other sometimes a mosaic of extremely small ones nominal species of Caturus, the caudal skel- (e.g., C. porteri Rayner, 1948, fig. 1; C. driani eton is known in only five (C. insignis Kner, Thiolliere, Lehman, 1966, fig. 133). Unfor- Upper Triassic, Austria; C. giganteus Wag- tunately, the supraorbitals cannot be seen in ner, C. velifer Thiolliere and C. bellicianus the American specimens. [Thiolliere] from the Kimmeridgian litho- Although C. dartoni remains poorly known, graphic limestones of France and Germany; it shares one derived character, fusion of hy- and C. deani Gregory from the Oxfordian of purals 1-3, with the English Liassic C. het- Cuba). We have not seen material of these erurus (Patterson, 1973, fig. 21). We have also five species. Pending a long overdue revision 40 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 21. Caturus dartoni. AMNH 10975. Caudal skeleton. x 1.5. From Hulett, Wyoming. of Caturus, we suggest that there may be two Occithrissops willsoni, new species species-groups in Caturus-one character- Figures 22-29 ized by fused anterior hypurals (including C. HOLOTYPE: AMNH 10964 (figs. 23A, 24A, heterurus, C. smithwoodwardi, and C. dar- 30), a large fish, 200 mm SL, lacking the tip toni), and the other by a mosaic ofnumerous of the snout and parts of the anal and caudal very small supraorbitals (including C. fur- fins. catus, C. porteri and C. driani). DISTRIBUTION: Same as for genus. All the specimens have been found in the lower part of the Sundance Formation. TELEOSTEI ETYMOLOGY: In honor ofthe late Earl Will- ICHTHYODECTIFORMES INCERTAE SEDIS son of Hulett, Wyoming, on whose property OCCITHRISSOPS, NEW GENUS most of the material was collected. TYPE SPECIES: Occithrissops willsoni, new DIAGNOSIS: As genus, only species. Reach- species. ing about 200 mm SL; vertebrae 58 (33- DISTRIBUTION: Upper Bathonian, Wyo- 34 + 22-23 + 2 ural); D iii, 10-11; A iv-v, ming. 18-20; P ?, 14; V 9; C viii, I, 9, 8, I, vii. ETYMOLOGY: Refers to the occidental oc- REFERRED SPECIMENS: From the Sundance currence of the genus and its relationship to Formation, Stockade Beaver Shale Member, Thrissops. near Hulett, Wyoming: AMNH 10867, DIAGNOSIS: Ichthyodectiform fishes in 10873, 10949, 10960, 10966, 10973, 10974, which the depth ofhead, length ofsnout, and 10976, 10982, 10984, 10990, 10992, 10993, size of mouth are intermediate between Al- 11434, 11435, SDMS 54362. lothrissops and Thrissops; chordacentra only, without perichordal ossification in individ- DESCRIPTION uals up to about 80 mm SL; uroneurals do not cover the lateral surfaces of preural cen- This species is described mainly by com- tra, anterior tips of uroneurals 1-4 regularly parison with species of Allothrissops (Nybe- spaced; supraneurals confined to abdominal lin, 1964; Taverne, 1975a; Patterson and Ro- vertebrae; anal fin strongly falcate, with 18- sen, 1977) and Jurassic species of Thrissops 20 radials. (Nybelin, 1964; Taverne, 1977), since the 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 41

4? -A" .1..1--1v .,,-I

jlo.l , !- ,.1 A .> 4o,

K;4

FIG. 22. Occithrissops willsoni. A. AMNH 10964. Holotype. x.44. B. AMNH 10990. x 0.78. From Hulett, Wyoming. material of Occithrissops does not permit a Allothrissops mesogaster and Thrissops for- comprehensive account. mosus (fig. 24). In the specimens of Occi- SKULL: The skull of Occithrissops willsoni thrissops willsoni these bones are not clearly (fig. 23) is, so far as it is known, generally visible, but SDSM 54362 suggests that they similar to that of Allothrissops mesogaster were like those in Allothrissops. The propor- (Agassiz) (fig. 24A; Patterson and Rosen, tions and structure of the ethmoid region are 1977, figs. 5, 7-9), but resembles Thrissops also markedly different in Allothrissops me- formosus Agassiz (fig. 24B; Taverne, 1977, sogaster and T. formosus (cf. Patterson and figs. 6, 7) in certain ways. The head of 0. Rosen, 1977, fig. 7A, and Taverne, 1977, fig. willsoni is proportionally deeper than that of 9-the "rhinal" of Taverne is the "ethmo- A. mesogaster, the mouth is larger and more palatine" of Patterson and Rosen). The pro- upturned, and the lower limb of the pre- portions, if not the details, of the ethmoid opercular is shorter. In all these traits, 0. region of 0. willsoni are traceable in AMNH willsoni tends toward the condition in Thris- 10873 (fig. 26A), and here again 0. willsoni sops formosus. The ratio of head depth to appears more like Thrissops than like Allo- head length is about 1.2-1.3 in 0. willsoni, thrissops. The length of the snout is approx- whereas in Allothrissops mesogaster this ratio imately 75-80 percent of the diameter of the is about 1.45-1.5, and in Thrissops it is 1.3 orbit in A. mesogaster, about 60-65 percent in T. formosus, 1.05 in T. subovatus v. Miun- in T. formosus, and about 65-70 percent in ster, and 1.0 in T. cirinensis Nybelin (mea- 0. willsoni. Nybelin (1964, p. 5) gives it as surements of Thrissops from figures in Tav- 60-80 percent in T. formosus. erne, 1977). The ratio of the lower limb of In Thrissops (Taverne, 1977, and BMNH the preopercular to the upper limb (measured specimens) there is a large supraoccipital crest. along the sensory canal) is about 1.75-1.9 in Occithrissops willsoni (fig. 25) resembles Al- 0. willsoni, 1.4-1.5 in A. mesogaster, 2.1 in lothrissops, where the crest is small (Patterson T. formosus, 3.7 in T. subovatus, and 2.3 in and Rosen, 1977, fig. 5). According to Pat- T. cirinensis (Thrissops measured, as before, terson and Rosen (1977, p. 100), another dif- on Taverne's 1977 figures). ference between Allothrissops and Thrissops There is a marked difference in form ofthe is that there is a suborbital in the latter but antorbital and first two infraorbitals between none in the former. In 0. willsoni no speci- 42 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 23. Occithrissops willsoni. Skull in lateral aspect. A. AMNH 10964. x 1.65. B. SDSM 54362. x2.5.

men is sufficiently well preserved to judge condyle lies beneath the posterior third ofthe whether there was a suborbital. The quadrate orbit in 0. willsoni, as in A. mesogaster; in 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 43

FIG. 24. Skull restorations. A. Allothrissops mesogaster (from Patterson and Rosen, 1977). B. Thris- sopsformosus (from Taverne, 1977).

Thrissops the condyle is situated farther for- p. 17). Nybelin (1964) and Taveme (1977) ward, beneath or in front of the center of the noted that in Thrissops the teeth are smaller orbit (Taveme, 1977). in T. formosus than in the other Jurassic Judging by illustrations in Nybelin (1964), species, and, in detail, it is not easy to dis- Taverne (1975a, 1977), and Patterson and tinguish the dentition of T. formosus from Rosen (1977), there is a marked difference in that of Allothrissops or Occithrissops. In 0. the jaws and dentition between Allothrissops willsoni (fig. 27A, B) there are about a dozen and Thrissops. The former has small, Lep- moderately large teeth on the premaxilla, and tolepis-like teeth and a lower jaw with a long, the bone has two dorsal processes, one at its high coronoid process; Thrissops has larger, medial border, and a larger one, more lat- fanglike teeth set in longjaws (Taveme, 1 977, erally placed. In T. formosus (fig. 27D) the 44 AMERICAN MUSEUM NOVITATES NO. 2796

in size and number, but those of A. mesogaster may match those of 0. willsoni. The dentary teeth of 0. willsoni (fig. 27C) are roughly equal in size along the margin of the bone, as they are in Allothrissops (fig. 27K, L), whereas in T. formosus (fig. 27I) they are somewhat enlarged in the middle ofthat margin. Although Taverne (1977) stated that Thrissops lacks a coronoid process on the dentary, none ofthe specimens recorded from the French or German lithographic stone shows the dentary margin in its entirety. Fig- ure 28J shows an isolated Thrissops dentary which certainly possesses a coronoid process, and we assume that a similar process is present in the intact lithographic stone specimens. Thus the coronoid process is similar FIG. 25. Occithrissops willsoni. Sketch resto- in Thrissops, Allothrissops and Occithrissops, ration ofskull roof, based on AMNH 10973. x 3.5. but is placed farther back in Thrissops than in Allothrissops. In the position of the coro- premaxillary teeth are fewer (five to eight) noid process and the length of the toothed and larger, and the bone is ovoid in outline, border of the dentary, Occithrissops is inter- with a poorly marked dorsal process. In Al- mediate between Allothrissops and Thrissops. lothrissops (fig. 27E-H) the premaxilla may In A. mesogaster(Patterson and Rosen, 1977, have a medial process (A. regleyi, fig. 27E) or fig. 9A) the dentary tooth row is equal to a broad dorsal process (A. mesogaster, fig. about 40 percent of the length of the ventral 27G), and the premaxillary teeth are variable border of the bone, whereas in 0. willsoni

fr A fr rode rode sue le sue vo vo pas pas

FIG. 26. Sketches to show proportions ofthe snout (ethmoid region ofbraincase) in A, Occithrissops willsoni (based on AMNH 10873); B, Allothrissops mesogaster (after Patterson and Rosen, 1977, fig. 7A); C, Thrissopsformosus (based on Taverne, 1977, fig. 6), with additions after Cladocylus (Patterson and Rosen, 1977, fig. 6A). 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 45

C ---Jr, - A

G vV I' - ' Lt~AAlU\J\JV\Jj I

H K -44A4ANRI 'LS) FIG. 27. Comparison ofdentitions in ichthyodectiforms. A-C. Occithrissops willsoni: A, premaxillae of AMNH 10976; B, left premaxillary teeth of AMNH 10984; C, dentary margin of AMNH 10973. D. Thrissopsformosus, right premaxilla (preserved in impression) ofBMNH P. 3684. E. Allothrissops regleyi, right premaxilla ofBMNH P. 918. F-H. Allothrissops mesogaster: F, right premaxillary teeth from below of BMNH P. 915; G, posterior part of right premaxilla of BMNH P. 3680b; H, right premaxillary teeth and anterior part of dentary margin of BMNH P. 3680a. I. Thrissops formosus, right dentary margin (medial view, reversed) of BMNH P. 3684. J. Thrissops sp., margin of isolated left dentary (reversed) of BMNH P. 54599. K, L, Allothrissops mesogaster: K, margin of right dentary (medial view, reversed) ofBMNH P.916; L, margin ofleft dentary (reversed) ofBMNH P.915. A-C, from Sundance Formation, Hulett, Wyoming; D, F-I, K, L; from Lithographic Stone. Bavaria; E, from Lithographic Stone, Cerin, Ain, France; J, from Kimmeridge Clay, Encombe, Dorset, England. Not to scale. (fig. 28A) this ratio is about 45 percent and In both Thrissops and Allothrissops the dis- in T. formosus about 55 percent. tal ceratohyal is deep, with a large fenestra 46 AMERICAN MUSEUM NOVITATES NO. 2796

L I

FIG. 28. Occithrissops willsoni. A. Left dentary in lateral view, based on AMNH 10973. B. Outline of distal ceratohyal, from AMNH 10962. Bar equals 1 mm.

(Patterson and Rosen, 1977, fig. 9B; Tav- whereas A. mesogaster has 11-14 (Patterson erne's (1977, fig. 10) restoration of the cer- and Rosen, 1977, fig. 5; Nybelin, 1964, pl. 9, atohyal of T. formosus lacking a fenestra is fig. 3). As in both Thrissops and Allothrissops, incorrect). Occithrissops willsoni also has a there is a marked posterior expansion of the deep, fenestrated ceratohyal (fig. 28B). In 0. preopercular at the angle in 0. willsoni. willsoni, AMNH 11435 shows 17 branchio- AXIAL SKELETON: In the smallest available stegal rays, 13 on the distal ceratohyal, two individual of 0. willsoni (AMNH 10982, 49 or three on the gap between distal and prox- mm SL) there are about a dozen ventral imal ceratohyals, and only one or two on the hemichordacentra extending from PU6 for- proximal ceratohyal (there may have been ward to the origin of the anal fin. The two more branchiostegals posteriorly). The first ural chordacentra are also calcified, but there 11 or 12 branchiostegals are slender and are no other centra. In somewhat larger spec- threadlike, and only the last three are spa- imens (AMNH 10873, 64.5 mm SL, and thiform. In A. mesogaster there are 18 bran- AMNH 11435, estimated SL 80 mm) there chiostegals, 11 on the distal ceratohyal, and are cylindrical chordacentra throughout the the last three are spathiform (Patterson and column. In still larger specimens (SDSM Rosen, 1977, p. 102). Thrissopsformosus has 54362, 93 mm SL, and AMNH 10949, es- 20 branchiostegals (Taveme, 1977, fig. 10), timated SL 96 mm) there is evidence ofperi- 12 on the distal ceratohyal, and the last four chondral bone on the surface of the chorda- are spathiform. centra, and in the largest specimens (Holotype In the ventral limb of the preopercular, 0. AMNH 10964, 200 mm SL, figs. 22A, 29) willsoni has eight to 10 sensory canal branch- the centra are well ossified. In Allothrissops es. Thrissops formosus also has about 10, there are perichordal additions to the chor- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 47 dacentra in specimens over about 30 mm SL; and segmented only toward the tip, where the in the Jurassic Thrissops small individuals second and third rays split up into fine are not recorded (Nybelin, 1964). branches. The pectoral fin occupies about half In 0. willsoni there are 58 vertebrae, 33- the distance between its insertion and that of 34 abdominal, 22-23 caudal, and two ural. the pelvic. In Allothrissops mesogaster there are 57-61 In 0. willsoni the pelvic fin originates be- (31-33 + 24-27 + 2) vertebrae, and in neath vertebra 26-28, and contains nine rays Thrissops formosus there are 57-59 (33- which extend over the length of five or six 35 + 24-26 + 2). In 0. willsoni all the-neur- -veitebrae. The pelvic is similar in A. meso- al arches, haemal arches, and parapophyses gaster and T. formosus. are apparently autogenous. In A. mesogaster UNPAIRED FINS: The dorsal fin of 0. will- most of the caudal haemal arches are fused soni orginates behind the anal, opposite ver- with the centra in large individuals. Patterson tebra 39-40, and with its first radial ending and Rosen (1977, p. 115) used this character proximally behind neural spine 36. The fin as a synapomorphy to distinguish the sub- contains three unbranched, and 10 or 11 order Allothrissopoidei (Allothrissops only) branched rays that insert on 10 or 11 radials. from the Ichthyodectoidei, including Thris- The first is bifid proximally and has an an- sops. However, we suspect that some of the terior process extending from its head. The caudal haemal arches may fuse with the cen- dorsal fin is not significantly different in A. tra in large individuals of T. formosus and T. mesogaster and T. formosus (Nybelin, 1964; subovatus. Taverne, 1977). In 0. willsoni there are long epineurals on The anal fin of 0. willsoni originates be- the abdominal and first two or three caudal neath vertebra 37-38, with its first radial end- neural arches, a series of supraneurals from ing proximally in front of the haemal spine the occiput back to about the twenty-seventh of vertebra 34-35. The fin contains four or vertebra, and the neural spines are paired from five unbranched, and 18-20 branched rays the occiput back to about the twenty-eight that insert on 18-20 radials. The fin is falcate, vertebra. In A. mesogaster and T. formosus with the longest ray equal in length to seven the paired neural spines and supraneurals ex- or eight caudal vertebrae. In A. mesogaster tend farther posteriorly. There are supra- the anal contains four or five unbranched rays neurals back to the first or second caudal ver- and about 25 branched rays, inserting on 24- tebra in A. mesogaster, and in BMNH 37043 27 radials. The first radial ends in front of the first median neural spine is on vertebra the haemal spine of vertebra 32-33, and the 35. Thrissopsformosus has supraneurals back fin originates under vertebra 36-37. In T. to vertebra 37 (Taverne, 1977, fig. 11). Ac- formosus, the anal contains four or five un- cording to Nybelin (1964, p. 35), there is a branched rays and about 30 branched rays, difference between Allothrissops and Thris- which insert on 29-32 radials. The first radial sops in the form ofthe ribs, which he believed ends in front of the haemal spine of vertebra are stout and thickened dorsally in Thrissops 33-35, and the fin originates under vertebra and slender throughout in Allothrissops. 38-42 (Taveme, 1977). The anal is falcate in However, we are not convinced that there is both Allothrissops and T. formosus, but more a real difference here; the ribs appear to be strongly so in the latter (Nybelin, 1964): the similar in Allothrissops, Thrissops, and Oc- longest anal ray is equal in length to about cithrissops. eight caudal vertebrae in T. formosus and to PAIRED FINS: There are no notable differ- about four in A. mesogaster, A. salmoneus, ences between the pectoral girdle and fin of and A. regleyi. The anal fin of 0. willsoni (1 8- A. mesogaster (Patterson and Rosen, 1977, 20 radials) is therefore shorter than that of fig. 10) and T. formosus (Taveme, 1977, fig. Allothrissops (24-30 radials) and T. formosus 5). In 0. willsoni the girdle is similar, so far (29-32 radials), but closer in form to that of as it is known, and shows the same massive T. formosus, since its longest rays are equal coracoid as in Allothrissops and Thrissops. in length to seven or eight vertebrae. The pectoral fin of 0. willsoni contains at CAUDAL SKELETON AND FIN: The caudal least 14 rays, the first three ofwhich are stout skeleton of 0. willsoni is best preserved in 48 AMERICAN MUSEUM NOVITATES NO. 2796 the holotype (AMNH 10964, fig. 29), but parts In 0. willsoni the two lower hypurals show ofit are also visible in AMNH 10873, 10949, the configuration typical in ichthyodecti- and 10990. The caudal skeleton is generally forms (cf. fig. 29 and Patterson and Rosen, similar to that of A. mesogaster (Taveme, 1977, p. 106). AMNH 10964 has eight hy- 1975a, fig. 14; Patterson and Rosen, 1977, purals (fig. 29), again as in Allothrissops and figs. 17, 18) and T. formosus (Patterson and Thrissops, and there are three epurals as in Rosen, 1977, fig. 14; Taverne, 1977, figs. 14, those genera. 15) except for a striking difference in the form The caudal fin of 0. willsoni contains 19 ofthe uroneurals. One synapomorphy of the principal rays, of which 17 are branched. Ichthyodectiformes recognized by Patterson There are eight upper and seven lower pro- and Rosen (1977, p. 115) is described as "six current rays (fig. 29). The fin formula is sim- or seven uroneurals, the first three or four ilar to that in Allothrissops and Thrissops. extending anteroventrally to cover the entire There is a single urodermal in 0. willsoni lateral surface of the first, second, or third (AMNH 10966), as in T. formosus and some preural centra." As figure 29 shows, this con- individuals of A. mesogaster (Patterson and dition is not found in 0. willsoni, for the Rosen, 1977, p. 113). According to Nybelin lateral surface ofthe preural centra is exposed (1964), the caudal fin is larger and more deep- beneath the uroneurals, which occupy the ly forked in Thrissops than in Allothrissops, dorsolateral position normal in leptolepids and the lower lobe is longer than the upper and other generalized teleosts (Patterson and in Thrissops, whereas the two lobes are equal Rosen, 1977). Among the four specimens of in Allothrissops. In 0. willsoni the caudal fin 0. willsoni showing the uroneurals, there is is weakly forked as in Allothrissops, but in all one (AMNH 10949) in which the anterior five specimens where the tail is intact, the ends ofthe first four uroneurals do cover the lower lobe is longer than the upper. The ex- lateral surfaces of U1-PU2, but the bones cess length of the lower lobe is only 2-4 per- appear to be disarticulated post mortem. The cent that of the upper in AMNH 10992 and other two specimens show the same config- 10966, but is 13-15 percent in AMNH 10960, uration as in figure 29. In all four specimens 10974, and 10993. However, in Allothrissops the tips of the first four uroneurals are reg- we find a similar excess of 10-15 percent in ularly spaced, each ending close to the an- the length of the lower caudal lobe. terior margin of a centrum in the sequence SQUAMATION: The scales of 0. willsoni are Ul and PU1-3 (fig. 29). This condition also not particularly well preserved, but show contrasts with that in Allothrissops and Thris- nothing to differentiate them from those of sops, where three or four bunched uroneurals A. mesogaster and T. formosus. As in those extend forward to PU2. Regular spacing of species, 0. willsoni had one scale row to each the tips of the uroneurals is general in lep- vertebra. tolepids and in Pachythrissops (Patterson and DIET: Gut contents are preserved in several Rosen, 1977, figs. 33, 35, 45, 46, 51). specimens of 0. willsoni. As in Allothrissops The neural arches of the ural and last few (Patterson and Rosen, 1977, p. 1 5), they are preural centra are poorly preserved in 0. will- in the form ofamorphous phosphatic matter, soni, and only AMNH 10964 (fig. 29) shows suggesting a microphagous diet. Swallowed them at all completely. In that specimen these prey are known in Thrissops (Nybelin, 1958), neural arches are difficult to assign to centra indicating a predatory habit. since the last three (labeled NPU1-NPU3) appear to end above the junctions between centra. In Allothrissops and Thrissops the RELATIONSHIPS OF OCCITHRISSOPS neural arches and spines are variable in this Patterson and Rosen (1977, p. 115) have region (Patterson and Rosen, 1977, p. 110; defined the order Ichthyodectiformes by five Taverne, 1977, p. 26), but since there is never characters: (1) floor of nasal capsule with an a full-length neural spine on PU 1, we assign ethmo-palatine ossification that articulates the last such spine in AMNH 10964 to PU2, with the palatine; (2) six or seven uroneurals and assume that this centrum (or PU3) bore with the first three or four extending antero- two neural arches. ventrally to cover the sides ofthe first to third 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 49

FIG. 29. Occithrissops willsoni. Caudal skeleton as preserved in the holotype, AMNH 10964. x 7.5. From Hulett, Wyoming. preural centra; (3) teeth in a single series in Occithrissops, Allothrissops, and/or Thris- the jaws; (4) coracoid enlarged and meeting sops in the dentition (fig. 27) and fin structure, its fellow in a midventral symphysis; and (5) which provide evidence that Occithrissops is, anal fin long, falcate, and opposed by a short, in fact, an ichthyodectiform. remote dorsal fin. Occithrissops willsoni shares Within the ichthyodectiforms Patterson characters 3, 4, and 5 with the ichthyodec- and Rosen (1977) have established two sub- tiforms, but lacks character 2. The condition orders, the Allothrissopoidei (the Upper of character 1 remains unknown. Unfortu- Jurassic Allothrissops only) and the Ichthyo- nately, the characters shared by Occithrissops dectoidei (Upper Jurassic-Cretaceous Thris- and the ichthyodectiforms are not unique to sops and 10 other Cretaceous genera). If these fishes and thus do not demonstrate that these two suborders are sister groups, there Occithrissops is an ichthyodectiform. There are only three possible positions for Occi- are, however, detailed resemblances between thrissops within Ichthyodectiformes: it is an 50 AMERICAN MUSEUM NOVITATES NO. 2796

Q) £9 0 0e

A B CkX FIG. 30. Alternative cladograms of Occithrissops and other ichthyodectiforms. Cladogram A is supported by the proportions of the head and the falcate median fins; cladogram B by the number of supraneurals and anal fin-rays. We have insufficient information on the structure of Occithrissops to resolve this conflict, and therefore accept the trichotomous cladogram C. allothrissopoid, an ichthyodectoid, or the sis- tary tooth row longer than in Allothrissops ter group of both. The Allothrissopoidei are (and the leptolepids). characterized by three features: no suborbital 2. Chordacentra only are developed up to bone, infraorbital canal ending blindly in the about 80 mm SL, that is, later than in Al- lacrimal, and caudal haemal arches fused with lothrissops (or leptolepids). The only ichthy- the centra. Occithrissops is unknown regard- odectoid of comparable size known to us is ing the first of these, imperfectly known re- BMNH P. 12463, ?Thrissops sp., from the garding the second, and we now doubt the Lower Purbeck beds, ca. 60 mm SL. This reality of the third, since large individuals of specimen has well-ossified centra, with only Thrissops formosus and T. subovatus may a narrow notochordal perforation. show fusion ofcaudal haemal arches and cen- 3. There are paired neural spines and su- tra. The Ichthyodectoidei are defined by six praneurals from the occiput back to vertebra features. The condition in Occithrissops is 27-28. In Allothrissops supraneurals extend unknown for four ofthese (which include two back to vertebra 30-31 (A. regleyi), or 34-35 braincase characters, plus the palatine mal- (A. salmoneus, A. mesogaster), and in Thris- leolus and the composition of the joint sur- sops to vertebra 32 (T. cirinensis), or 35-38 face in the lowerjaw). Occithrissops also lacks (T. formosus, T. subovatus) (Taverne, 1977). one oftwo remaining ichthyodectoid features 4. The uroneurals do not cover the lateral (enlarged supraoccipital associated with for- surfaces of Ul-PU3, and their ends are reg- ward displacement of the parietals), and ularly spaced (as in leptolepids). In Allo- probably lacks the other (ball-and-socketjoint thrissops, Thrissops formosus and T. subov- between first hypural and first ural centrum; atus, the uroneurals show the usual cf. fig. 29; Patterson and Rosen, 1977, figs. ichthyodectiform configuration (Patterson 13-20). and Rosen, 1977). However, the holotype Evidence for the position of Occithrissops (and only specimen) of Thrissops curtus within Ichthyodectiformes is therefore (Woodward, 1919; BMNH P. 10612) from the Lower Purbeck beds, shows the same con- equivocal (fig. 30), given the subordinal char- dition as Occithrissops willsoni. acters proposed by Patterson and Rosen. 5. The anal fin is shorter than in Allothris- However, the following characters of Occi- sops or Thrissops. thrissops seem relevant in regard to its rela- 6. The dorsal and anal fins are strongly tionships among the ichthyodectiforms: falcate, as in Thrissops. 1. The head is deeper, the snout and lower Reviewing these characters, there is no in- limb ofthe preopercular shorter, and the den- dication of special relationship between Oc- 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 51 cithrissops and Allothrissops, but characters fin in the middle of the back with about 16 1 and 6 imply that Occithrissops may be clos- rays; anal originates beneath posterior edge er to Thrissops and the ichthyodectoids than ofdorsal, with about 14 rays; pelvics beneath is Allothrissops, whereas characters 3, 4, 5, dorsal origin, with about 12 rays. No bone- and perhaps 2, suggest that Allothrissops may enclosed rostral commissure or rostral pit- be closer to Thrissops and other ichthy- line; supraorbital sensory canal not meeting odectoids than is Occithrissops, or, in other infraorbital, and with three medial and two words, that Occithrissops is the sister group lateral branches above the rear of the orbit. of Allothrissopoidei + Ichthyodectoidei. The Anterior pit line on parietal, but probably no presence of character 4 in Thrissops curtus middle pit line. Infraorbital sensory canal with indicates that this feature is not universal in four branches in lacrimal, none in second in- ichthyodectoids (T. curtus appears to be an fraorbital, two or three in third, two in fourth, ichthyodectoid because it has a triangular su- and one in fifth. No suborbital. Eight to 12 praoccipital crest and a serrate-margined bas- sensory canal branches in preopercular. Max- al sclerotic bone; cf. Patterson and Rosen, illa toothed throughout its length; teeth larger 1977, p. 115). Character 2 (persistence of on premaxilla than on maxilla. In lower jaw, chordacentra) can hardly be considered a dentary with a few teeth and a high coronoid primitive feature in Occithrissops, as this process; angular and retroarticular fused; sen- evaluation would imply that perichordal cen- sory canal enters medial face ofpostarticular tra arose independently in ichthyodectiforms process. About 15 branchiostegal rays, no gu- and other teleosts. If characters 2 and 4 are lar. No pelvic splint. Paired neural spines and excluded, we have two characters relating Oc- unpaired supraneurals present from occiput cithrissops to ichthyodectoids (proportions of back to beneath dorsal fin, so that supra- head and falcate median fins), and two re- neurals lie between the first few dorsal radi- lating Allothrissops to ichthyodectoids (long als; epineurals on abdominal and first few series of supraneurals and long anal fin). The caudal neural arches; five to seven pairs of first pair of characters seems more likely to epipleurals present in region of abdominal/ be significant, but we are unwilling to resolve caudal transition. Caudal skeleton with two the incongruence without more knowledge of free ural centra, six uroneurals, eight or nine Occithrissops and of other Late Jurassic and hypurals, three epurals, one dorsal fringing Early Cretaceous ichthyodectiforms. For the fulcrum and two urodermals. Two scale rows present, we place 0. willsoni as Ichthyodec- to each vertebra. tiformes incertae sedis. We justify erection of the new genus Occithrissops for the reason Todiltia schoewei (Dunkle) that it shows no indication of close relation- Figures 31-36 ship to any other ichthyodectiform genus, and Leptolepis schoewei Dunkle, 1942, p. 62, pl. 6. differs from all by the characters given in the "Leptolepis" schoewei Dunkle; Schultze and En- diagnosis above. ciso, 1983, p. 1053, fig. 3. HOLOTYPE: KUMVP No. 784, a juvenile TELEOSTEI INCERTAE SEDIS fish 36 mm in SL from Felch Creek, 15 m TODILTIA, NEW GENUS (24 km) north of Canon City, Colorado. TYPE SPECIES: Leptolepis schoewei. DIAGNOSIS: As for genus; only species. DISTRIBUTION: Lower Callovian, Colo- REFERRED SPECIMENS: From Wanakah rado, and New Mexico. Formation, Pony Express Limestone Mem- ETYMOLOGY: For the Todilto Limestone. ber, Piedra River Canyon, Colorado: AMNH DIAGNOSIS: Middle Jurassic teleostean 11457. From Wanakah Formation, Todilto fishes of leptolepid grade, but differing from Limestone Member, near Hot Springs, New similar fishes in having only chordacentra Mexico: AMNH 9832, 11489, 11490. From until late in growth. The chordacentra receive Wanakah Formation, Todilto Limestone perichordal additions only in the midcaudal Member, Bull Canyon area, New Mexico: region, and only in the largest specimens. NMNH 8551,17899,17903,17907; BHI P3, About 50 vertebrae, 30 abdominal. Dorsal P4, P5, 951. From Wanakah Formation, To- 52 AMERICAN MUSEUM NOVITATES NO. 2796

Al"

D FIG. 31. Todiltia schoewei. Growth series (SL). A, AMNH 17899, 40 mm; B, NMNH 9832,37 mm; C, AMNH 11487,48 mm; D, AMNH 11488,66 mm. A, from Bull Canyon, N.M.; B, from Hot Springs, N.M.; C and D, from Merrill Ranch, N.M. dilto Limestone Member, Merrill Ranch, New and with "Leptolepis" species. Our usage Mexico: AMNH 11487, 11488, 14495, needs explanation. The Leptolepididae are 11496, 11508, 11516. known to be a polyphyletic grade taxon (e.g., COMMENTARY: Throughout the following Patterson and Rosen, 1977). Traditionally, description and discussion of Todiltia we Leptolepis and the leptolepids are generalized compare Todiltia schoewei with leptolepids Jurassic and Early Cretaceous teleosts, with 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 53 small, feebly-toothed mouths, thin, cycloid sen, 1977, p. 54), and, in our view, serves no scales, and unmodified fins. At least 50 nom- useful purpose. By "leptolepid," we do not inal species of Leptolepis have been de- mean Leptolepididae sensu Nybelin, or sensu scribed. During the last two decades, the trend Taverne (e.g., 1975a, 1975b). has been toward assigning these species to Summarizing our usage: new or revived genera, mostly monotypic, 1. Leptolepis = L. coryphaenoides and L. including: Aethalionopsis Gaudant, Ascala- normandica. bos Miinster, Clupavus Arambourg, Lepto- --2-. "Leptolepis" = other species of Lepto- lepides Nybelin, Nybelinoides Taverne, lepis sensu Nybelin (1974) and Taverne Paraclupavus Saint-Seine and Casier, Patter- 1975a, 1975b). sonella Taveme, Pholidolepis Nybelin, Pro- 3. leptolepid = Leptolepis sensu lato, not leptolepis Nybelin, Pseudoleptolepis Taverne, Leptolepididae sensu Nybelin or Tav- Seefeldia Nybelin, Tharsis Blainville, and erne. Wenzichthys Taverne. Some of these genera have been shown to be related to higher teleosts (Clupavus, Leptolepides, Nybelinoides, DESCRIPTION Pattersonella, Tharsis, Wenzichthys). Others MEASUREMENTS AND PROPORTIONS: See ta- are less closely related to Recent teleosts than ble 1. is the type species ofLeptolepis, the Toarcian SKULL: A sketch restoration of the head is L. coryphaenoides Bronn (Pholidolepis, Pro- shown in figure 32. Nothing is known of the leptolepis, Seefeldia). There remains a residue braincase, except that there were separate of nominal species of Leptolepis, comprising pterosphenoids and an orbitosphenoid, a well- two suites. The first includes comparatively ossified lateral ethmoid, and a mesethmoid well-known species (reviewed by Nybelin, of typical leptolepid form (Patterson, 1975, 1974; Taverne, 1975a, 1975b; Patterson and figs. 127, 128, 130, 131). The dermal com- Rosen, 1977) whose relationships are "some- ponent (rostrodermethmoid) has a pair of where in the region" of L. coryphaenoides. strong posterolateral processes forming the Yet even in this restricted sense, Leptolepis anterior wall and the floor of the nasal pit. remains polyphyletic (Patterson and Rosen, There is no sign of a rostral commissure or 1977). Finally, there is a second suite ofpoor- pitline. It is not known whether the para- ly known Leptolepis species whose relation- sphenoid bore teeth and a basipterygoid pro- ships are unknown. Our study of L. schoewei cess, as it does in other leptolepids. removes it from this second suite, but shows The skull roof is imperfectly known, but that it falls into the first, Leptolepis sensu appears similar to that of other leptolepids Nybelin (1974). Since we believe that genus (Cavender, 1970, fig. 1; Nybelin, 1974, figs. to be polyphyletic, we have two alternatives. 1, 4, 23, 29; Patterson, 1975, fig. 147; Pat- The first is to refer to L. schoewei and other terson and Rosen, 1977, figs. 34, 49). The members of Leptolepis sensu Nybelin (apart supraorbital sensory canal in the frontal runs from the type species) as "Leptolepis"; the to the posterior margin ofthe bone, as is usual second is to continue the trend of dismem- in leptolepids. Above the posterior part of bering "Leptolepis" into monophyletic or the orbit the canal gave off three medial and monotypic genera. We have decided to make two lateral branches. These are fewer branch- the new genus Todiltia for L. schoewei, and es than in Tharsis dubius (Patterson and Ro- to refer only L. coryphaenoides (and the sen, 1977, fig. 147) or in the Callovian "Lep- closely related or synonymous L. norman- tolepis" (Patterson, 1975, fig. 147), but more dica) to Leptolepis sensu stricto. We will refer than in Leptolepides sprattiformis (Patterson to Leptolepis sensu Nybelin as "Leptolepis," and Rosen, 1977, fig. 49). There was evi- and to Leptolepis sensu lato as "leptolepids." dently no postorbital connection between the Nybelin (1974) restricted the Leptolepididae infra- and supraorbital canals. On the pari- to the three genera Leptolepis sensu Nybelin, etal, the line of the supraorbital canal is con- Proleptolepis, and Tharsis. That assemblage tinued by a shallow anterior pit line groove. is, however, polyphyletic (Patterson and Ro- There is no sign of a middle pit line but the 54 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 32. Todiltia schoewei. Restoration of skull in lateral aspect. preservation in this area is poor. The nasal canal has one bone-enclosed branch in io 5, is a slender, tubular bone. two in io 4, two (USNM 17907) or three The circumorbital bones are fairly well pre- (AMNH 11508) in io 3, none in io 2, and served in several specimens, and follow the four in the lacrimal. There is no indication standard pattern ofa triangular dermosphen- of a suborbital behind io 5, and we assume otic (dsp), three broad infraorbitals (io 3-5) that this bone was absent. behind the orbit, a slender one (io 2) below The upper jaw also follows the usual lep- it, a triangular lacrimal (io 1), and a single tolepid pattern, with a small premaxilla and large supraorbital (suo). The small antorbital a long, curved maxilla, each with teeth along present between the supraorbital and lacri- the oral margin. The teeth are larger on the mal in other leptolepids has not yet been rec- premaxilla than on the maxilla. There are two ognized in Todiltia. The infraorbital sensory supramaxillae, the anterior long and slender,

TABLE 1 Todiltia schoewei, Counts and Measurements (SL in millimeters, other measurements as percentage of SL.)

7a~~~~~a 0

NMNH 17899 37 124 18 35 24 ?30 18+2 5 iv, 13 13 iii, 11 12 AMNH 11487 48 123 24 32 22 - 18+2 7 iv, 12 13 iii, 11 12 AMNH 11508 60 123 25 34 25 28+ 18+2 8 iv, 13 14 iii, 11 12 AMNH 11488 66 120 29 32 23 30 18+2 8 iv, 13 14 iii, 11 12 NMNH 17907 78.5 118 22 31 24 - 18+2 7 iv, 13 - - 12 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 55 and the posterior with a pointed process above AXIAL SKELETON: Whereas the skull and fin the anterior. The visible parts of the palate skeleton of Todiltia show nothing remarkable show nothing remarkable. No teeth are evi- in comparison with leptolepids, the axial dent on the palate, but preservation may be skeleton differs strikingly from that of most inadequate. In the lowerjaw, the dentary bears species ascribed to "Leptolepis." In two spec- a few teeth, about equal in size and number imens where the vertebrae are accurately to those on the premaxilla. Behind these teeth, countable (AMNH 11488 and NMNH the dentary rises in the long, high coronoid 17907), there are 50 vertebrae, 30 abdomi- process normal in leptolepids. The mandib- nal, 18 caudal, and two ural centra. All spec- ular sensory canal opens by three pores on imens show 18 caudal vertebrae. There is the lateral surface of the anterior part of the marked variation, assumed to be ontogenet- dentary, two large pores on the ventral sur- ic, in the vertebral centra. In the smallest face of the tube in the dentary, a pore at the specimens, such as AMNH 11495 (SL 28 mm) suture between dentary and angular, and one there are no centra at all. In slightly larger small pore on the angular. The angular has a specimens, such as AMNH 9832 and NMNH well-developed postarticular process, and the 17899, both SL 37 mm long, only the second point of entry of the sensory canal is on the ural centrum in AMNH 9832 is calcified. medial face of this process (cf. Patterson and There are two cylindrical ural centra in Rosen, 1977, fig. 32). The retroarticular and NMNH 17899 plus separate dorsal and ven- angular appear to be fused in the only spec- tral hemicentra in PU 1, a slender centrum or imen in which the inner face ofthe lowerjaw hemicentrum tapering dorsally above HPU3 is partially visible (AMNH 11488, a large (fig. 34), and eight delicate centra extending individual, SL 66 mm). We assume that this from PU5 to PU12. Those of PU7-8 are the fusion was ontogenetic (cf. Patterson and Ro- widest ofthis series.4 AMNH 9832 (fig. 3 IA) sen, 1977, fig. 32). (SL 40 mm) has slender ventral hemicentra Nothing is known ofthe hyoid or gill arch- extending forward from the ural centra to PU es. There are at least 13 branchiostegal rays 11, but has no obvious dorsal hemicentra, (AMNH 11488) and possibly as many as 16 except perhaps in PU6. AMNH 11487 (fig. (AMNH 11508). There is no indication of a 31C) (SL 48 mm) has cylindrical centra for- gular, despite Dunkle's (1942, p. 62) mention ward from U2 to PU 16. Those ofPU6-8 are of "a relatively large median gular plate." the widest centra in this series, and as The shape of the preopercular and opercular preserved, PU3-9 or 10 show a vertical split bones is shown in figure 32. The number of in the middle of the centrum. Also in this branches ofthe preopercular sensory canal is specimen, PU 17 is represented by two hem- variable, probably in part as the result ofon- icentra, preserved as left and right, and in togenetic multiplication of branches, as in- PU18 there are two minute, spindle-shaped ferred by Nybelin (1974) in other leptolepids, hemicentra, again preserved as left and right. and as described by Allis (1889) in Amia. In In all these small individuals, the uncalcified NMNH 17899 (37 mm SL) there are eight part ofthe vertebral column was occupied by branches, six of them on the lower limb of a wide, unconstricted notochord. From the the bone, and in AMNH 11487 (48 mm SL) mode ofcalcification ofthe centra (which oc- there are also six branches on the lower limb. curs long after ossification of the neural and In AMNH 11508 (60 mm SL) there are 11 haemal arches in the form ofcrescentic hem- branches in all, eight on the lower limb, icentra), and their dense, acellular structure, whereas in AMNH 11488 (66 mm SL) there are also 11 branches in all, but nine on the lower limb. NMNH 8551 (64 mm SL) also 4 It is possible that the pattern of centra in NMNH 17899, with a gap between PUl and PU5 occupied only has nine branches on the lower limb, and by one incomplete centrum, is abnormal, since the neural NMNH 17907 (78.5 mm SL), the largest and haemal arches are "out-of-step" (fig. 34). The cen- specimen, has 10 branches on the lower limb. trum above HPU3 passes toward a neural arch, which There is no indication ofa suprapreopercular shows features of both NPU2 and NPU3 in other spec- bone. imens and has a gap behind it, so that it counts as NPU2. 56 AMERICAN MUSEUM NOVITATES NO. 2796

we assume that they are chordacentra (cf. Fran9ois, 1966, on Salmo). In larger specimens (59-78 mm SL) (e.g., fig. 3 1 D) the whole vertebral column is occupied by cylindrical chordacentra, and at least those in the caudal region (those first calcified) are slightly hourglass-shaped, so that the notochord was slightly constricted in the middle of each centrum. As noted, the centra in the midcaudal region, roughly PU3-10, usually show a vertical split in the middle, evidently accentuated by crushing in preservation. In AMNH 11508 (SL 60 mm) there is a double first ural centrum, the first and second hypurals articulating with separate cylindrical chordacentra. In the largest specimens, such ° as NMNH 17907 (SL 78.5 mm) the midcaudal centra, about PU3-15, show perichordal ossification in the form of cellular bone with surface relief, superficial to the smooth, acellular chordacentra. 0 Vertical splits in the midcaudal centra, 0 similar to those in Todiltia, have also been 0 reported in "L." talbragarensis Woodward o by Cavender (1970, p. 23), who suggested <, that a groove was present even in uncrushed centra. We have observed splits in crushed midcaudal centra of several other "leptolepids" (e.g., young Tharsis dubius (Blainville) 1

i FIG. 34. Todiltia schoewei. NMNH 17899. Caudal skeleton. Bar equals 2 mm. pelvic radial. The pelvic fins are best pre- by 12 radials. The first two radials lie in front served in AMNH 11488, where the left fin of the first caudal haemal spine, and support contains 12 rays and the right 1 1; the first ray the three unbranched rays. Ossified middle is unbranched, and all are segmented. There segments are present distal to the fourth and is no pelvic splint. The pelvic fin extends succeeding radials. about three-fourths the distance between the CAUDAL SKELETON AND FIN (figs. 34-36): pelvic and anal origins. The neural spines of PU2-4 are broader dis- UNPAIRED FINS: The dorsal fin originates tally than their predecessors, and decrease behind the pelvic fin origin and above the progressively in length. The neural spine of twenty-seventh or twenty-eighth vertebra. PU2 is about two-thirds as long as that of The fin base occupies the length of nine ver- PU3. PUl bears a normal but reduced neural tebrae. The fin contains four unbranched rays, arch with a short neural spine except in the first unsegmented, and 12 or 13 branched NMNH 17899, fig. 34, where the neural arch- rays, supported by 13 or 14 radials (table 1). es are "out-of-step," and U1 bears an even The first radial is bifid, with a platelike an- smaller arch and spine. Also in NMNH 17899 terior limb divided into finger-like processes. the arch and spine of U1 resemble those of This radial supports the first three un- PU1 in other specimens. The haemal spines branched rays. Ossified middle segments are ofPU 1-3 are long, broad and in contact with present distal to the third and succeeding ra- each other: they support the ventral procur- dials in larger individuals (fig. 33). rent rays, and the parhypural (haemal spine The anal fin arises opposite the hind end of PU1) supports the lowermost four prin- of the dorsal fin base, and its base occupies cipal rays. the length of five or six vertebrae. The fin There are two ural centra, the first sup- contains three unbranched rays, the first un- porting hypurals 1 and 2, as usual, and the segmented, and 11 branched rays, supported second hypurals 3-5, except in AMNH 1508, 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 59

FIG. 35. Todiltia schoewei. AMNH 11487. Caudal skeleton. Bar equals 4 mm. where the first ural centrum is double. There and 35 could not, we will assume that the are six uroneurals: the first extends foward to undifferentiated state is representative of the PU2 and ends posteriorly over U2; the sec- species. There are eight or nine hypurals (figs. ond extends from PU1 to above the head of 35, 36B) and three epurals (fig. 36B). the seventh hypural; the third extends for- The caudal fin contains 19 principal rays ward to U1; the fourth reaches U2; and the with nine branched in the upper lobe and fifth and sixth are small, slender and largely eight in the lower. This is the generalized hidden by the fin-rays. In AMNH 11488 (fig. teleost condition. The bases of the first six 36B) and third uroneural appears to be di- upper branched rays form a bundle that ex- vided into two roughly equal parts by an tends across several hypurals. The bases of oblique suture, which seems to be natural and the seventh and eighth rays have dorsal pro- not an artifact. Patterson and Rosen (1977, cesses. The broad, elongate base ofthe ninth, p. 129) noted a differentiation of the uro- which reaches hypural 3, is typical of lepto- neural series in certain leptolepids and in lepids and certain pholidophorids (Patterson higher teleosts where the first four uroneurals and Rosen, 1977, figs. 31, 35). The upper are elongate (e.g., Tharsis dubius, ibid., fig. principal rays are preceded by 10 procurrent 35), and the last three uroneurals are shorter rays, the last three segmented distally. There and more horizontal, forming a series that is one dorsal fringing fulcrum (fig. 36A). The lies lateral to the hind ends of the anterior lower principal rays are preceded by six pro- uroneurals. In Todiltia there is no such dif- current rays, all but the first segmented. There ferentiation in NMNH 17899 (fig. 34) or in are no ventral fringing fulcra. Two elongate AMNH 11487 (fig. 35), but in AMNH 11488 urodermals (figs. 34, 35) lie on the two up- (fig. 36) the last two uroneurals appear to be permost branched principal rays. There is a more horizontal than, and differentiated from, large caudal scute in front of the upper and the anterior four uroneurals. This might rep- lower procurrent rays. resent intraspecific variation, but since the The caudal fin is deeply forked, and its condition in figure 36 could be due to post- length is about 24 percent of the standard mortem disturbance, and that in figures 34 length. 60 AMERICAN MUSEUM NOVITATES NO. 2796

FIG. 36. Todiltia schoewei. AMNH 11488. A. Caudal skeleton. B. Enlarged portion ofcaudal skeleton with basal fulcra omitted. Bar for A equals 5 mm; for B 2 mm.

SQUAMATION: The scales are preserved as sionally visible, but it is impossible to count a uniform film, in which circuli are occa- circuli or scales with any accuracy. The lateral 1984 SCHAEFFIER AND PATTERSON: JURASSIC FISHES 61 line scales, visible in the region of the un- the characters ofL. coryphaenoides and lower calcified notochord ofNMNH 17899, appear forms (nos. 1-24), three of the characters of to outnumber the vertebrae two to one, sug- Tharsis dubius and higher teleosts (nos. 31, gesting that there were about 90 scale rows. 32, and 34): no middle pit line groove on der- SoFT ANATOMY: In NMNH 17903, much mopterotic, no suborbital and epipleurals of the epaxial and caudal hypaxial muscu- present in the middle part ofthe trunk. Among lature is preserved as a yellowish mass within the other characters of Tharsis dubius and which fibers and their alignment are clearly higher groups, Todiltia is known to lack visible. In the small NMNH 17899 (fig. 31A) sculptured centra constricting the notochord, and in AMNH 9832 (fig. 31B), pigment is differentiated anterior and posterior uro- preserved (in addition to the retinal pigment neurals (cf. figs. 34-36 with Patterson and preserved in most specimens) in the form of Rosen, 1977, fig. 35), and all subsequent dark spots, assumed to be chromatophores. characters in their list. The main purpose of There is a dark band of pigment along the the Patterson-Rosen scheme was to deter- dorsal midline, and an even sprinkling of mine the relationships of the Mesozoic chromatophores over the head and anterior ichthyodectiforms. Those fishes were found part of the abdomen, with a denser patch to lack three of the characters of Tharsis du- around the pectoral fin. bius and higher groups: absence of the sub- The gut content is preserved in nearly all orbital bone, presence ofepipleurals, and the specimens, as a white or yellowish, feature- uroneurals in two series. less, phosphatic mass. This suggests a mi- Thus Todiftia, on the basis of present crophagous diet. The combined information knowledge, has two derived characters not on gut form from all specimens, shows that found in the ichthyodectiforms-absence of there was a large stomach anteriorly, sepa- the suborbital and presence of epipleurals, rated by a sphincter from an intestine which and lacks one derived character present in looped ventrally, rose dorsally again above these fishes-well-developed perichordal the pelvic fins, then passed obliquely down centra. However, the ichthyodectiform Oc- to the anus, in front of the anal fin. cithrissops, as noted above, also lacks well- developed perichordal centra until it exceeds RELATIONSHIPS OF TODILTIA the apparent maximum size of Todiltia. Among the leptolepids, Patterson and Ro- Dunkle (1942) distinguished Leptolepis sen (1977) found two taxa, the poorly known schoewei from other leptolepids by "an ex- "Leptolepis" macrophthalmus Egerton and ceptionally weak vertebral development," and "Leptolepis" talbragarensis Woodward, by five other features that are either general which have approximately the same char- in leptolepids, such as autogenous neural and acter combination as Todiltia. The mono- haemal arches and caudal scutes, or are shown typic Ascalabos voithii Munster was also by better material to have been misinter- found to fit at about this level. "Leptolepis" preted. The relatively late calcification ofthe caheni (Saint-Seine and Casier, 1962), which chordacentra in Todiltia (at about 30 mm Taverne (1975c) has compared with "L." tal- SL), their absence in the abdominal region bragarensis, also requires further comment. below about 50 mm SL, and the feeble de- "Leptolepis" macrophthalmus, "L." tal- velopment ofperichordal centra in the largest bragarensis, Ascalabos voithii, and Todiltia individuals (about 80 mm SL), remain the schoewei have a broadly similar caudal skel- most striking characters ofthe species. These eton (cf. figs. 24-36 with Patterson and Ro- characters are discussed again in comparing sen, 1977, figs. 45, 46, 53). Todiltia and "L." Todiltia with "L." talbragarensis. macrophthalmus have an "empty triangle" In an outline classification of certain pho- beneath the dorsal caudal scute and above lidophorids, leptolepids, and other Mesozoic the epurals that is partly or completely filled lower teleosts, Patterson and Rosen (1977) in both "L." talbragarensis and Ascalabos by listed 48 characters which specified previ- elongate preural neural spines. This same ously unrecognized monophyletic groups. empty triangle occurs in L. coryphaenoides Comparing Todiltia with that list, it has all (Taverne, 1982, fig. 5), "L." caheni (Taverne, 62 AMERICAN MUSEUM NOVITATES NO. 2796

1975c, figs. 10, 11), and in Tharsis (Patterson supraorbital-infraorbital sensory canal junc- and Rosen, 1977, fig. 35; Taverne, 1975a, fig. tion behind the orbit (absent in Todiltia), one 9). "L." macrophthalmus and "L." talbra- branch of the infraorbital sensory canal in garensis apparently lack epipleurals, al- infraorbital 3 (two or three in Todiltia), no though these elements are present in Todiltia canal branch in infraorbitals 4 or 5 (two in and Ascalabos.s Little has been learned about infraorbital 4, and one in 5 in Todiltia), a the detailed structure of "L." macrophthal- pelvic splint (absent in Todiltia), about three mus, but with its long, slender body, numer- fewer vertebrae than in Todiltia, and three ous preopercular sensory canal branches or four upper caudal fringing fulcra (one in (Patterson and Rosen, 1977, fig. 35b), and Todiltia). The significance of most of these well-ossified centra, it shows no particular features is unknown, but together with the similarity to Todiltia. In vertebral structure, absence ofepipleurals in "L. " talbragarensis, "Leptolepis" talbragarensis is like Todiltia. they distinguish the two species. We have The smallest specimens of L. talbragarensis found no special similarities to justify a sister available to us are about 30-35 mm SL. The group relationship between them. earliest stages ofvertebral development show Ascalabos voithii (Nybelin, 1974; Patterson ventral hemichordacentra in the first few ver- and Rosen, 1977) differs from Todiltia in tebrae (7-8 in BMNH P. 12426, 33 mm SL) having annular chordacentra throughout the and in the midcaudal region (PU5-8 in P. column (at 35 mm SL), larger teeth, longer 12426). In larger specimens, ca. 40 mm SL, jaws, about 10 fewer vertebrae, and a pred- most centra are rings rather than hemicentra, atory habit (swallowed prey observed by Ny- but there is still agap in vertebral calcification belin [1974, p. 180], and by us in BMNH at the rear of the abdominal region and in specimens). According to Nybelin, one ofthe the last few caudal centra (e.g., BMNH P. most striking features of A. voithii is that the 37972a, ca. 39 mm SL, lacks PU21-29 and distance between the origins of pelvic and PU4-U1; P. 12447b, 41mm SL, lacks the anal fins is only 62-70 percent of that be- last five abdominal centra and PU1-3). By tween the pectoral and pelvic origins. In To- about 50 mm SL, all centra are represented diltia this ratio is even lower- 52-62 percent. by rings, but PU1 is the last to calcify. The Although Todiltia shares with A. voithii such pattern and sequence of vertebral calcifica- features as epipleurals, a similar number of tion as estimated from SL in"L." talbragar- preopercular sensory canal branches, and a ensis thus matches that in Todiltia. single dorsal caudal fringing fulcrum (contra However, there are other differences be- Nybelin, 1974, p. 179), there seems to be no tween "L." talbragarensis and Todiltia. Ac- good reason for regarding the two species as cording to Cavender's (1970) and Nybelin's related, or for including schoewei in the genus (1974) accounts, "L." talbragarensis has: a Ascalabos. "Leptolepis" caheni, from the Kimmeridg- s Taverne (1975b) published an account of Ascalabos ian ofZaire, differs from Todiltia and resem- voithzi based principally on a specimen in the Rijksu- bles "L." talbragarensis in having a supraor- niversiteit, Gent, Belgium. This individual differs from bital/infraorbital canal junction and no those described by Nybelin (1974) and Patterson and epipleurals. "Leptolepis" cahenihas some dif- Rosen (1977) in lacking epipleural bones, in having 43 ferentiation of the uroneurals into anterior vertebrae (including the ural centra) rather than 38-39, and posterior series, approaching the condi- in having four rather than three uroneurals extending tion in Tharsis and higher teleosts. The centra forward to PU1, and in a few other details. Taverne's of "L." caheni were described by Taverne specimen is small, 43 mm SL. BMNH P. 3671, labeled (1975c, p. 837) as lacking relief, but their "Leptolepis polyspondylus Agassiz" by von Munster is unknown (chordal or peri- (Nybelin, 1974, p. 174), is about 32 mm SL, and shows composition the same features as Taverne's specimen. Nybelin refers chordal). to the BMNH specimen as "too badly preserved to allow These comparisons are summarized in ta- a safe identification," but we suggest the possibility that ble 2. As the table shows, at our current level neither Taveme's specimen nor BMNH P. 3671 may be of knowledge of these five species, compar- A. voithii. isons are restricted to superficial features, are 1984 SCHAEFFER AND PATTERSON: JURASSIC FISHES 63

conflicting, and are rendered questionable by individual variation within species and by problems with inadequate preservation and/ we N-,Cl or observation. So our conclusion, that skeJ-ug o - the nearest relative of Todiltia, OTAI3d _-__- cannot specify -- is in part only a reflection of our ignorance of this and other leptolepids. N- -4 00 r I -- en1f _-- DISCUSSION AND SUMMARY sAtui-uij m In comparing the Sundance and Wanakah l'BUV -4 fishes with those from other areas, we will r- Cl treat them as a single assemblage.6 This can be justified in our present state of knowledge 'IC and -2 '0 on the grounds ofpaleogeography (fig. 1), ri 1- by the general distribution of Hulettia from t,.ilnj 'BuTgu!jj l 0 -I' C - southern Montana to New Mexico. The ap- Ispna,3 o ToqtunN 1- o parent restriction of Caturus and Occithris- s1muospoin c) " " Cl4 - Cl 0n sops to the Sundance and of Todiltia to the jpo laqtunN Wanakah may have some unresolved paleo- sIL,.mdAll 00 0l *^ l O 00 as noted in the section on TaqvunN 00 00 ecological basis, 0 jpo geological occurrence. A c) s siunououspundXipu c0 The Sundance-Wanakah assemblage in- *e jo J3quunN 'IO CO *4 cn cludes: -4 Oalael liuea C'. cl. c'. 0r Hybodus sp. 0 t - C l 00 Ischyodus sp. CC 00 O *0-. Hulettia americana (Eastman) +1+1 ,-I dR No -4 N peuta dod 6 cl. 0 Lepidotes sp. Q Caturus dartoni (Eastman) Po I+ CO uoTl3unf Occithrissops willsoni new genus, new cl* + JItqOi/OS cK. E) species Todiltia schoewei (Dunkle) 0 slt,.n;aldTdg + + _- This list is obviously limited in compari- -l son to assemblages from rocks ofcomparable There are CO age and area in Europe (table 3). C c) '0 mO several possible explanations for this. One is C'3 0 0 0 Jo Cis CO that exploration for Jurassic fishes in North uqo_n 10 0 _C America is still inadequate by Europen stan- Q~~~ CO sOs;r"r E) dards. Representatives of most taxa listed 0) ,cO above were found before 1900 (excepting Oc- uo TPqunr 00 cithrissops and Todiltia) and no concerted ef- 0, t *, fort was made to improve this situation before 1959, although several collections were q6) t X ^ ^ t~~~~~~0 e) .o

6 The term fauna is hardly applicable here. We are concerned at best with biased samples of the fish fauna j !2, EEg that occupied the western interior sea and possibly some 11 ofthe surrounding drainages during the Middle Jurassic. There is no way ofknowing how representative any fossil fish assemblage may be in terms ofthe original diversity or the original habitats. 64 AMERICAN MUSEUM NOVITATES NO. 2796 obtained later from the Todilto. Intensive C. dartoni. All ofthese taxa are characterized prospecting at the Hulett Sundance fish lo- by having fused lower hypurals, as discussed calities during eight field seasons failed, for above. The relationships of this subgroup to reasons unknown, to increase the diversity. other nominal Caturus species remain un- Another is that suitable preservation envi- known. Thus in identifying new fossils as ronments for Jurassic fishes were seemingly members ofgenera such as these, and in com- more restricted on this continent than in Eu- paring them with their presumed occurrence rope. But regardless of this apparent low di- elsewhere, one must be careful not to draw versity, these specimens must be regarded as unwarranted inferences that go beyond the a valid and meaningful sample of the fishes anatomical and systematic data. Names alone, that inhabited the western interior sea area without supporting data, are a poor guide to during mid-Jurassic times. relationship. Perhaps the most striking aspect of this The three new monotypic genera, Hulettia, assemblage is the inclusion of four familiar Occithrissops, and Todiltia in themselves genera (Hybodus, Ischyodus, Lepidotes, and represent three different levels of resolu- Caturus), plus three new monotypic genera tion-that is the descending categorical level (Hulettia, Occithrissops, and Todiltia) whose at which each must be regarded as incertae relationships are obscure. The first four taxa sedis: Hulettia in the Halecostomi, Todiltia are represented only by one or a few frag- in the Teleostei and Occithrissops in the mentary specimens, whereas the last three are Ichthyodectiformes. Analogous examples are known from relatively abundant and com- also cited in the comparisons of Jurassic fish plete material. This spectrum of occurrence assemblages in the Appendix. and degree of completeness is related to dif- Stratigraphically and geographically, the ferent levels ofresolution in our understand- closest assemblage to the Sundance-Wana- ing of these fishes. These levels are manifest kah one is that from the Oxfordian of Cuba in terms ofpreservation, relative abundance, (Gregory, 1923; White, 1942; Dunkle, per- diversity and, of course, the evident prob- sonal commun.; see Appendix). Here we are lems in systematic interpretation. immediately faced with the problem oftrying The chondrichthyan genera Hybodus and to evaluate assemblage resemblances-relat- Ischyodus are undoubtedly form genera with ed mostly to inadequate description and long temporal and wide geographic ranges. comparison. Although this Cuban assem- Unfortunately, they are usually represented blage has been known for more than 60 years, by isolated teeth. Lepidotes and Caturus also and several collections of fish-containing have long temporal and wide geographic nodules exist in museums, no detailed study ranges, but their representation in the fossil of the fishes has yet been made. Apart from record is considerably better. Certain species the pycnodont Gyrodus, Gregory's (1923) are well-known anatomically, and there is identifications (Caturus, ?Sauropsis, the new abundant, well-preserved material in the "pachycormid" genus Eugnathides and Lep- British Museum (Natural History) and a few tolepis) seem as likely to mislead as to inform, other collections. Lepidotes is not a form ge- if taken at face value. White's (1942) mono- nus in the sense that Hybodus and Ischyodus typic "leptolepid" genus Luisichthys may be are, and in our opinion the only question a synonym ofPachythrissops, but critical de- about its monophyly concerns possible syn- tails are lacking in his description. Until a onymy with Semionotus (Olsen and Mc- modern, systematic study is published, it ap- Cune, Ms.). At present, the many nominal pears that the Cuban assemblage differs species of Lepidotes and/or Semionotus are mainly from that of the Sundance/Wanakah monophyletic at some undetermined level, in including pycnodonts and in lacking chon- and are interrelated in unknown ways. drichthyans. We also suspect that Caturus is a form ge- South America (see Appendix) is less well nus, either para- or polyphyletic. But the cau- explored than North America in regard to dal skeleton of C. dartoni implies that it be- Jurassic fishes. To date, the only coherent longs to a monophyletic subgroup comprising knowledge from this continent concerns the at least C. heterurus, C. smithwoodwardi and Oxfordian assemblage from Domeyko, in 1984 SCHAEF'FER AND PATTERSON: JURASSIC FISHES 65 eastern Chile (Arratia, 1982). This contains analysis has shown, or suggested, that tradi- undescribed Lepidotes, pycnodonts and tional Mesozoic actinopterygian groups such amioids, and the teleosts Pholidophorus do- as the parasemionotids, semionotids, catur- meykanus Arratia et al. (1 975a), "Leptolepis" ids, pholidophorids, and leptolepids are non- opercularis Arratia et al. (1975c), and three monophyletic, as are many of their included monotypic endemic genera, Protoclupea Ar- genera. Other traditional groups, such as the ratia et al. (1975b), Varasichthys Arratia macrosemiids (Bartram, 1977), pachycor- (1981) and ChongichthysArratia (1982). Here mids and aspidorhynchids (Patterson, 1977; we have a problem similar to that represented Mainwaring, 1979), are found to be mono- by the Sundance/Wanakah assemblage. Lep- phyletic, but of unknown or uncertain rela- idotes, Pholidophorus and "Leptolepis" are tionship even at a very high taxonomic level. widespread and long-ranging "phenetic ab- Given the cladistic method, the tendency, stractions." As the last two are certainly non- with increasing knowledge ofmorphology, is monophyletic, they are essentially valueless to divide extinct para- or polyphyletic groups for assemblage comparison. Among the three (form taxa) into smaller monophyletic or monotypic genera, Varasichthys and Chong- monotypic units. These may be left in limbo, ichthys have been well described, so far as either temporarily or permanently. Examples the material allows, and their structure has in this paper are the species group of Caturus been thoroughly compared with that ofother proposed above, as well as the genera Hu- Jurassic fishes by cladistic analysis. The con- lettia, Todiltia, Varasichthys, and Chong- clusion reached by Arratia (1981, 1982) is ichthys. that both taxa are incertae sedis within the One could take this situation to mean that Teleostei; that is, their relationships are un- paleontological critics of cladistics, such as known. They therefore pose the same system- Campbell (1975) or Van Valen (1978), are atic problem as Hulettia and Todiltia -as correct in stating that the cladistic method is comparatively well-known fossil neopteryg- unsuited to the study of fossils, in which sig- ians that are valueless for purposes ofassem- nificant characters may be difficult or im- blage comparison because each is, at present, possible to find. But alternatives to cladistic unique and without known relatives. hypotheses of relationship, however fragile, Discrimination of these two sets of taxa are "horizontal groups," which are acknowl- (wide-ranging, non-monophyletic, and unique edged to be para- or polyphyletic (Campbell, monotypic) is a result of the introduction of 1975, p. 95), or "segments ofclades bounded cladistics into paleontology. Before cladistics, by adaptive changes" (Van Valen, 1978, p. fossil fishes such as Varasichthys, Chong- 293). As is now evident, vague concepts like ichthys, and Todiltia would have been treated these guided most of us in pre-cladistic days as members of a grade taxon, probably des- in assigning fossil fishes to form genera such ignated as Leptolepis or Leptolepidae. They as Hybodus and Leptolepis, or injudging that could have been used in faunal comparisons, fossils previously assigned to these genera perhaps as stratigraphic indicators, and as merited new generic status. Campbell (1975, contributions to the phylogeny and historical p. 95) claims that taxa conceived in this way biogeography of teleosts. But those compar- "cannot be used for detailed biogegraphical isons, and the conclusions drawn from them, work," but "can be used as indicators ofsim- would have been as devoid of phylogenetic ilar climatic belts, broad geographic prov- meaning as the statement that Varasichthys, inces, and comparable habitats." As the ta- Chongichthys, and Todiltia are teleosts ofun- bles in the Appendix show, however, known relationship. Para- or polyphyletic Hybodus, Lepidotes or Leptolepis sensu lato, grade taxa, like Leptolepis sensu lato, and the imply virtually cosmopolitan distribution, Leptolepididae, are taxonomic artifacts, and and so are not helpful. To date, we lack cri- opinions on stratigraphy, phylogeny, or bio- teria by which Hybodus, Lepidotes, or Lep- geography based on them are equally artifi- tolepis might be broken up into more useful cial. (in Campbell's sense) subgroups. The problem outlined above is a growing In systematics, the two horns of the pale- one, and one that will not disappear. Cladistic ontologist's dilemma are cladistic analysis 66 AMERICAN MUSEUM NOVITATES NO. 2796

versus the traditional or evolutionary meth- example, the Sundance endemic Occithris- od (e.g., Mayr, 1982). The first leads to un- sops willsoni is the earliest ichthyodectiform certainty, expressed in liberal use of incertae known, and could be treated as ancestral to sedis. The second leads to spurious certainty, other members ofthe group, or at least to the when phenetic or "adaptive" groupings are ichthyodectoids. Ancestry is thus one way of treated as real. interpreting the meaning of a monotypic tax- As for "detailed biogeographical work" on in a trichotomous cladogram like figure (Campbell's 1975 phrase), it begins for the 30C. One could then develop a narrative of cladist with the question "what is the sister- ichthyodectiform dispersal, with the Sun- group, and where does it live?" The question dance Sea as the center of origin. But just as is asked for many taxa in a diversified fauna. evolutionary systematics may lead to spu- In our Sundance/Wanakah assemblage, the rious certainty, so evolutionary biogeography question cannot be answered for any taxon, may lead to spurious stories. Our inability to and analytical biogeography cannot be start- resolve the relationships of Occithrissops is ed. For the traditionalist, biogeography can surely no basis for an inferred center oforigin. be sketched out with a broader brush. For

APPENDIX: THE DISTRIBUTION OF JURASSIC FISHES A resume of Jurassic fish distribution has been Forcing all Jurassic fish localities into either a ma- included in this Appendix in the form of two ta- rine or a nonmarine setting involved some judg- bles, one for marine (table 3) and the other for ments that may be incorrect, but in our view that nonmarine (table 4), and by a series of paleogeo- possibility is preferable to lumping all occurrences, graphic maps (figs. 37-39) showing most of the or to introducing a third set of tables for indeter- areas in which Jurassic fishes have been found. In minable or possible brackish-water occurrences. our present state of knowledge, this compilation The basis for the classification is necessarily a must be regarded as tentative. The reasons for this combination of the "classical" and phylogenetic have been set forth in the preceding discussion, approaches. For the Chondrichthyes, the recent which emphasizes some ofthe problems that arise studies of Thies (1983), Maisey (personal com- in analyzing and comparing any fossil fish assem- mun.), and Patterson (1965) have been utilized. blages of about the same age from different areas The actinopterygian classification is mostly based or parts of the world. Our justification for includ- on Gardiner (1967) for the chondrosteans, on Pat- ing these tables is simply that no effort of this sort terson (1973) for the non-teleost neopterygians, has been attempted before, and that the tables will and on Patterson and Rosen (1977) for the teleosts. provide a summary of current information about Taxa of unknown or uncertain status are framed both the systematics and the distribution of Ju- by quotation marks or are placed in an incertae rassic fishes. sedis category. The preparation of worldwide taxonomic, The abbreviations in the marine and non-ma- stratigraphic, and geographic records ofextinct an- rine tables represent the stages of the Jurassic, as imals, such as are represented in these tables, nec- follows: essarily involves data compression and various Portlandian Po arbitrary decisions. If such compilations are un- Upper Kimmeridgian K critical, they cease to be informative. Doubtful or Oxfordian Ox questionable identifications, which are often per- petuated in tables such as these, can mislead the Callovian Ca specialist as well as the non-specialist. On the other Middle Bathonian Bt hand, to omit a record because of doubt about Bajocian (incl. Aelenian) Bj provenance or identification can also lead one astray. We have tried to steer a middle course, Toarcian T using specialist knowledge judiciously, checking Pliensbachian P1 specimens where possible, and omitting records Lower Sinemurian S only when we believe that they lack any real factual Hettangian H basis. The distinction between the marine and non- In some areas, and particularly for nonmarine marine tables has been based on the literature, on deposits, the Jurassic has been divided only into paleogeography, and on our own opinion and/or Lower (L), Middle (M) and Upper (U), and in a field experience of the associated fauna and flora. few places only a Jurassic age (J) has been recog-