AMERICAN MUSEUM

Norntates

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2728, pp. 1-34, figs. 1-38 June 4, 1982

A Revision of the Fossil Genus tDiplomystus, With Comments on the Interrelationships of Clupeomorph Fishes

LANCE GRANDE'

ABSTRACT Several primitive fossil clupeomorphs are de- new genus, is proposed for tClupea brevissimus scribed, and a hypothesis of clupeomorph inter- Blainville, 1818. The relationship of tArmigatus relationships is proposed. The genus tDiplomys- brevissimus, new genus, to other clupeomorphs tus is revised and the type species for the genus, is not clear, and it forms an unresolved trichoto- tD. dentatus Cope, 1877, is redescribed. tDiplo- my with tellimmichthyids and clupeiforms. mystus dentatus is closely related to tD. birdi Many other species erroneously assigned to the Woodward, 1895, and tD. dubertreti Signeux, genus tDiplomystus are removed to make the ge- 1951, and all three species form the sister group nus monophyletic and thus useful in systematic to tEllimmichthys longicostatus (Cope, 1886). and comparative anatomical studies. Because the dorsal scutes of tE. longicostatus The comparative morphology of clupeomorph lack the pectinate posterior border diagnostic of dorsal scutes is discussed. It is found that "double tDiplomystus, Jordan's (1910 and 1919 [in Jordan armor" in clupeiforms is not restricted to a small and Gilbert, 1919]) removal of this species from specialized group, but rather is a widespread tDiplomystus is considered valid. Both tDiplo- character, occurring in clupeids, engraulids, tel- mystus and tEllimmichthys are placed in tEllim- limmichthyids, and tArmigatus. Detailed mor- michthyidae, new family (tDiplomystidae is phological study of the dorsal scute in clupeo- preoccupied by a family of South American cat- morphs shows a complex of several characters, fishes). The tEllimmichthyidae, new family, is the some of which can be used to define monophyletic sister group to the Clupeiformes. tArmigatus, groups within the Clupeomorpha.

INTRODUCTION The fossil herring-like fish tDiplomystus is anatomical comparisons with other teleosts one of the most widely cited fossil teleosts by at least 40 different authors and in more in the literature. It has been cited or used for than 75 different papers. Yet, as currently

1 Department of Ichthyology, American Museum of Natural History.

Copyright © American Museum of Natural History 1982 ISSN 0003-0082 / Price $3.15 2 AMERICAN MUSEUM NOVITATES NO. 2728 defined this is a non-monophyletic group and The morphology of most of the species therefore of questionable value in systematic (including the type, tD. dentatus) is poorly studies. known. "tDiplomystus" brevissimus (Blain- The genus has largely been a repository ville, 1818) is the only species described in for most any articulated clupeomorph fossil much detail, mainly by Patterson (1967); but with dorsal scutes, since its description by no evidence has ever been presented that Cope in 1877. Even Cope included a modern would indicate that "i'D." brevissimus and type of clupeoid (today called tKnightia) in tD. dentatus are closely related. Patterson the same genus with tDiplomystus. And (1967) showed that although "i'D." brevis- since Cope about 15 additional taxa have simus is a clupeomorph, it is not a clupeoid been erroneously placed into the genus. This as had been previously thought. has caused serious problems in papers using Several authors (Woodward, 1892, p. 413; tDiplomystus for anatomical and/or out- Schaeffer, 1947, p. 24; Greenwood, 1968, p. group comparisons. 265 and others) have proposed that "double- The genus tDiplomystus was described by armored" clupeiforms (those clupeiforms Cope (1877) as a clupeid genus including sev- with both dorsal and ventral scutes) consti- eral fossil species from the Green River For- tute a monophyletic group. Nelson (1970b) mation. He divided the genus into two "sec- challenged this concept by pointing out that tions." In section 1 he included tD. dentatus Clupanodon thrissa (a gizzard shad) and Cope, 1877, tD. analis Cope, 1877, and tD. Ethmidium (Chilean shad), which are both pectorosus Cope, 1877; and he included tD. double-armored, are not closely related to "humilis" (Leidy, 1856)2 and fD. altus (Lei- other double-armored clupeids. Nelson (per- dy, 1873) in section 2. He placed a sixth sonal commun.) also drew my attention to species from the Green River Formation, the fact that many engraulids are also dou- tD. theta (Cope, 1874), as "intermediate be- ble-armored (although they are peculiar in tween ... the two sections" (Cope, 1877, p. having only one or two dorsal scutes). A 811). Section 2 was removed from the genus pristigasterin (Pristigaster cayana) was also tDiplomystus by Jordan (1907) and placed in found to have dorsal scutes (see below). a new genus tKnightia Jordan, 1907. The The main objectives of the present work genus tKnightia is revised elsewhere are: to provide a detailed description of (Grande, in press). All of Cope's tDiplomys- tDiplomystus dentatus (the type species of tus species from the Green River Formation the genus); to find derived characters uniting except for tD. theta were found to be junior the type with other species, thus defining the synonyms of tD. dentatus Cope, 1877, by genus more rigorously; and to examine the Grande (1980). It is not possible to comment relationship of tDiplomystus to i'Knightia on the position of tD. theta because there is and other groups within the Clupeomorpha. no known holotype for the species, and its A comparative morphological study of the description is insufficient. dorsal scutes in clupeomorphs is also includ- More than 15 species have been assigned ed as an appendix. to the genus tDiplomystus in addition to Cope's species, but reasons for most of these ACKNOWLEDGMENTS assignments are erroneous or tenuous. All species known to the author were reviewed I thank Mr. Guido Dingerkus, Drs. Donn during this project. E. Rosen, Gareth Nelson, Colin Patterson, and C. Lavett Smith for reading and com- menting on various drafts of this manuscript. 2 Because tClupea humilis Leidy, 1856 was a hom- Dr. Gareth Nelson also assisted in translat- onym of tC. humilis von Meyer, 1848, Leidy's name is unavailable. Cope's (1870) tClupea pusilla is unavail- ing some type descriptions from French to able for a replacement because that name is preoccupied English, and by sharing his discovery that by C. pusilla Mitchill, 1814. Therefore, the valid name many anchovies have dorsal scutes. Mr. Gui- for this taxon is tKnightia eocaena Jordan, 1907 (and do Dingerkus aided with some of the pho- not tK. humilis Leidy as given in Grande, 1980). tographic work. 1982 GRANDE: tDIPLOMYSTUS 3

I also thank Dr. Daniel Goujet (Museum All other counts and measurements follow national d'Histoire naturelle) and Dr. Colin Grande (1980) and Hubbs and Lagler (1949). Patterson (British Museum, Natural History) Fossils were prepared using needles under for their help during my visit abroad. a dissecting microscope. Dorsal scutes and Part of the fieldwork and travel expenses skeletons of recent teleosts were cleared and were paid for by grants to the author, from stained following Taylor (1967). All original the Theodore Roosevelt Memorial Fund, drawings were made using a Leitz camera American Museum of Natural History. A lucida on a Bausch and Lomb stereomicro- grant from City College of New York helped scope. All illustrated specimens of tDiplo- pay for drawing equipment and photographic mystus dentatus are from locality F- 1 of supplies. Grande (1980) except for the F-2 specimen in figure lB. The names of all fossil taxa mentioned in the text are preceded by a MATERIALS AND METHODS dagger (t). Drawings of dorsal scutes are of the dorsal Fossils used here are deposited in the fol- surface unless otherwise stated. Explanation lowing institutions: Department of Verte- of morphological terminology used here for brate Paleontology, American Museum of dorsal scutes is given in figure 38. In all Natural History, New York (AMNH); De- drawings, the anterior margin faces left partment of Paleontology, Science Museum (drawings of some fossils have been re- of Minnesota, St. Paul (SMMP); British Mu- versed). seum (Natural History), London (BMNH); The taxonomic subdivisions (subfamilies) Museum national d'Histoire naturelle, Paris of clupeids and engraulids used in the Ap- (MNHN); University of Michigan, Museum pendix follow Whitehead (1968). of Paleontology (UMMP); and Department of Paleontology, California Academy of Sci, ences (CAS). Recent material examined is in ABBREVIATIONS USED IN FIGURES the Department of Ichthyology, of the Amer- Aa, angulo-articular ican Museum of Natural History, and the Bt p, basipterygoid process of parasphenoid Australian Museum, Sydney, New South C, circulus (or ridge) Wales (AM). Some uncatalogued specimens Ch, anterior ceratohyal (=ceratohyal of Mc- from the Smithsonian Oceanographic Sorting Allister, 1968) Center, Washington, D.C. (SOSC) were also De, dentary examined and accession numbers for these Dsp, dermosphenotic are given in the text. Eh, posterior ceratohyal (=epihyal of McAllister, like most 1968) tDiplomystus, clupeomorphs, Enpt t, endopterygoid teeth has two vertebrae in advance of those bear- Ep, epural ing full length ribs. These are often not vis- Fr, frontal ible in fossils because they are normally cov- HS, haemal spine ered by the superficial bones of the skull. Hy, hypural Therefore the number of anterior vertebrae lo, infraorbital in all fossils was determined by counting the lop, interopercle pairs of ribs and then adding two. Only Mx, maxilla preural centra are included in the vertebral N, nucleus (or focus) counts. Standard length was measured from NS, neural spine the anterior tip of the snout to the Op, opercle posterior Pa, parietal end of the third hypural plate. Principal ray PH, parhypural counts of the dorsal and anal fins (="major Pop, preopercle fin rays" of Grande, 1980) include those rays Pmx, premaxilla which extend all the way to the outer fin Ps, parasphenoid margin. The term predorsal bones is used in Ps t, parasphenoid teeth the same sense as in Smith and Bailey (1961). Pt, post-temporal 4 AMERICAN MUSEUM NOVITATES NO. 2728

Pto, pterotic RELATIONSHIP OF tDIPLOMYSTUS Pu, preural centrum TO R, retroarticular tKNIGHTIA AND OTHER Smx, supramaxillary bone CLUPEIFORMS Sop, subopercle Much of the confusion that surrounds U, ural centrum tDiplomystus stems from two misconcep- Un, uroneural tions: (1) that tDiplomystus and tKnightia are closely related; and (2) that tDiplomystus TAXONOMIC PLACEMENT OF is a clupeoid. One or both of these ideas have "tDIPLOMYSTUS" BREVISSIMUS been implied or stated by Cope (1877, et tClupea brevissimus Blainville, 1818, was seq.), Jordan (1907), Thorpe (1938), Schaef- placed in tDiplomystus by Woodward in fer (1947), Cavender (1966), Greenwood 1888 (see below). Since then this species has (1968), Nelson (1970a), Taverne (1976), been studied in more detail (primarily by Pat- Uyeno (1979), Grande (1980), and others. It terson, 1967) than any other species in this is found here that tDiplomystus is not a clu- genus. It is removed from tDiplomystus here peoid (or even a clupeiform), although for reasons discussed below, and placed into tKnightia is (see fig. 19 for explanation of tArmigatus, new genus. taxonomic groups as used here). tDiplomystus dentatus is examined here and has [unlike Clupeiformes; and like all tellimmichthyids (see below), tArmigatus tARMIGATUS, NEW GENUS brevissimus, new combination (see above), TYPE SPECIES: tClupea brevissimus Blain- and tOrnategulum sardinioides (Pictet) (see ville, 1818. Forey, 1973)] (1) no recessus lateralis (fig. 7); GENERIC DIAGNOSIS: Clupeomorph fishes (2) the parietals in contact between the su- that differ from all others in the following praoccipital and the frontals; (3) an Osteo- combination of characters. Unlike Clupei- glossum-like tooth patch on the parasphe- formes, tArmigatus has (1) an osteoglossid- n,oid and a basipterygoid process (fig. 6) like tooth patch on the parasphenoid; (2) (although no parasphenoid tooth patch is parietals which are in contact between the known in tOrnategulum), and (4) a large fo- supraoccipital and frontals, and (3) a large ramen in the anterior ceratohyal (fig. 8A) foramen in the anterior ceratohyal. Unlike (=the "Beryciform foramen" of McAllister, tellimmichthyids and tOrnategulum, tAr- 1968, p. 6). tDiplomystus, unlike clupeoids, migatus has a series of subtriangular dorsal has no uroneural fused to the first preural scutes, which extends only two-thirds the centrum, and has the parhypural fused to the distance from the anterior end of the dorsal first preural centrum (see figs. 10, 12, and fin to the posterior end of the dorsocranium. 17). These are all primitive clupeomorph ETYMOLOGY: armigatus, bearer of armor characters. The recessus lateralis, separation (from Latin); gender masculine. of the parietals by the supraoccipital, reduc- DISCUSSION: Patterson (1967), after study tion of size and number of endopterygoid of "tDiplomystus" brevissimus Blainville, teeth, loss of Osteoglossum-like parasphe- 1818, determined that "tDiplomystus" was noid tooth patch, and loss of the "beryciform not a clupeoid. This observation was not foramen" are considered to be derived char- based on the type species, but rather on a acters of the Clupeiformes (Patterson, 1967, primitive clupeomorph which is probably and others). Fusion of the first uroneural to only distantly related to tDiplomystus. It is the first preural centrum and separation of removed from the genus here (see below) the parhypural from the first preural centrum and placed in tArmigatus, new genus, be- (fig. 19) are considered here to be derived cause it does not form a monophyletic group characters of clupeoids. It is evident that with tDiplomystus (tD. dentatus, tD. du- tDiplomystus is not a clupeiform (as this bertreti, and tD. birdi; see fig. 20). group is used here-see fig. 20). 1982 GRANDE: tDIPLOMYSTUS 5

tKnightia, on the other hand, is a clu- scutes, elongated laterally (figs. 9, 13, 18, peoid. It was found (Taverne, 1975; Grande, and Appendix). in press) to have a recessus lateralis; pari- DISCUSSION: These are primitive clupeo- etals separated by the supraoccipital; no ba- morph fishes that lack a recessus lateralis sipterygoid process or parasphenoid teeth; and several other clupeiform characters (see and no "beryciform foramen" (all clupei- fig. 20). The tEllimmichthyidae contains form characters). Also unlike tDiplomystus, tDiplomystus dentatus Cope, 1877, tD. birdi tKnightia has the following clupeoid char- Woodward, 1895, tD. dubertreti Signeux, acters: reduction in relative size of ural cen- 1951, and tEllimmichthys longicostatus trum 1, uroneural 1 fused with the preural (Cope, 1886). The family name "tDiplomys- centrum 1, and the parhypural is not fused tidae" (as used by Patterson, 1970, and Pat- to the preural centrum 1 (compare fig. 19 to terson and Rosen, 19774) is preoccupied by figs. 10, 12, and 17), and no lateral line a South American catfish family (type genus scales. tKnightia also has only one supra- Diplomystes-first pointed out by Nelson, maxillary bone (tDiplomystus has two), and 1976, p. 75). Because tEllimmichthys, Jor- it has a lower number of branchiostegals and dan, and tDiplomystus Cope (see below) are dorsal scutes (Grande, in press), and a dif- sister taxa (see fig. 20), we can use the genus ferent type of dorsal scute, than tDiplomys- tEllimmichthys to form a family name tus. tKnightia is examined in detail else- (tEllimmichthyidae) which includes both of where (Grande, in press). the genera. By using tEllimmichthys rather Another point of confusion seems to be than tDiplomystus an unoccupied family which species are in the tDiplomystus group, name can be constructed. and which are in the tKnightia group. Forey Because the tEllimmichthyidae is the sis- (1973, pp. 1309, 1314), for example, com- ter group to the Clupeiformes, and because pares tOrnategulum to "tDiplomystus hu- additional species have been found (work in milis," which in fact is not tDiplomystus but progress) that may be closely related to tel- rather the type species of tKnightia.3 Also, limmichthyids, it seems advisable to place Nelson's (1973, pp. 12, 25) "tDiplomystus" this family in its own order, the tEllimmich- sp. (AMNH 4300) is actually a specimen of thyiformes, new order. The diagnosis for this tKnightia eocaena. Uyeno (1979) seemed to order is the same as for the tEllimmichthyi- use the generic names tKnightia and tDiplo- dae (because it presently contains only that mystus interchangeably, when he (p. 22) sug- family) until further revision of the group is gested Hyperlophus "[is] the closest relative made. of tDiplomystus" [probably meaning Some morphological features of tellim- tKnightia] and described two new clupeo- michthyids include: usually seven predorsal morph species as tDiplomystus spp. He gave bones with thin laminar expansions (see figs. information that would not warrant their 1 and 5); pelvic fins in advance of dorsal fin placement in tDiplomystus but that might (except in tD. dubertreti where position of place them in tKnightia. the pelvic fin is unknown); and ventral scutes numerous (about 24 to 34), running along SYSTEMATICS AND PHYLOGENY ventral midline, from isthmus back to anus; OF THE tELLIMMICHTHYIDAE, NEW two supramaxillary bones present; parhy- FAMILY; REDESCRIPTION OF pural fused to vertebral column; and lateral tDIPLOMYSTUS DENTATUS COPE line scales present. These characters (other than those in the diagnosis) are not unique tELLIMMICHTHYIDAE, NEW FAMILY that differ from DIAGNOSIS: Clupeomorphs 4Although Whitehead (1967, p. 87), Greenwood all others in having subrectangular dorsal (1968, p. 265) and others have suggested that tDiplo- mystus be given familial rank, Patterson (1970, p. 179) JThe valid name for this species is tKnightia eocaena was the first to use formally the name "tDiplomysti- Jordan, 1907 (see fn. 2). dae" to include "tDiplomystus." 6 AMERICAN MUSEUM NOVITATES NO. 2728 to the tEllimmichthyidae and are discussed TYPE: AMNH 2477, illustrated in Cope, elsewhere in this paper. The extreme devel- 1884, plate 10, figure 1. opment of the laminar expansions of the pre- REFERENCE SPECIMENS USED HERE: dorsal bones in adult tellimmichthyids and AMNH 2477 (the type), 763, 2480, 2883, tArmigatus may be a shared derived 2979, 8109, 8168, 10465, 10466, 10469-10471, character uniting the two taxa; but in my 10473-10476. opinion, not enough is known about laminar HORIZON AND LOCALITY FOR HOLOTYPE: expansions in clupeiforms (pellonulins and The late Early Eocene Fossil Butte Member other groups) to warrant such grouping at of the Green River Formation, near the town this time. The problem is currently under of Fossil, Wyoming (=locality F-i of study. Grande, 1980). A freshwater limestone de- posit. tDIPLOMYSTUS COPE, 1877 KNOWN GEOGRAPHIC AND STRATIGRAPH- IC RANGE: Known tCopeichthys Dollo, 1904, an objective ju- from Early to Middle nior synonym (Jordan, 1907, p. 137). Eocene Green River Formation sediments in TYPE SPECIES: tDiplomystus dentatus Wyoming and possibly Utah. Most common Cope, 1877 (by original designation). at localities F-i and F-2 of Grande, 1980. REVISED GENERIC DIAGNOSIS: tEllim- REVISED DIAGNOSIS: A large species of michthyids that differ from all others in hav- tDiplomystus (reaching a total length of ing dorsal scutes with a pectinate posterior about 65 cm.) with more anal fin rays (usu- border (figs. 9 and 13) along the dorsal mid- ally 38 to 40), anal pterygiophores (usually line from the posterior edge of the skull to 38-41), ribs (17-18) and vertebrae (41-43) insertion of the dorsal fin; and having a than any other species of tDiplomystus (or higher number of dorsal scutes (usually 22- any other tellimmichthyid); more dorsal 36) than any other tellimmichthyid. The su- scutes (about 33 to 36) than any other known praoccipital crest is very well developed in species of tDiplomystus (or any other clu- tDiplomystus (see fig. 1, for example). peomorph); greater number of spines along The diagnostic information above was also posterior border of the dorsal scutes (in used by Cope (1877) in his original diagnosis adults) than any other tDiplomystus species for "section 1" of tDiplomystus [his "sec- (see fig. 9); elongate fusiform in shape, less tion 2" is what Jordan (1907) renamed as deep bodied than any other tDiplomystus species (compare 1 with tKnightia]. Cope (1877, et seq.) placed fig. figs. 11 and 14), tDiplomystus in the Clupeidae. It is correlated with increased number of caudal concluded here that although tKnightia is a vertebrae and associated elements. true clupeoid (Grande, in press), tDiplomys- ETYMOLOGY: dentatus, toothed (from tus is not even a clupeiform (see above and Latin); gender masculine. below). ETYMOLOGY: (not given by Cope) diplo, DESCRIPTION AND DISCUSSION double (from Greek); mystus, hidden? or GENERAL FEATURES: A laterally com- possibly from the word myst meaning a mys- pressed clupeomorph with a long anal fin and tic (from Greek); gender masculine. which has a known maximum total length of about 65 cm. The description below com- pares tD. dentatus to tOrnategulum, tAr- tDiplomystus dentatus Cope, 1877 migatus, new genus, other tellimmichthyids, Figures 1-10 and clupeiforms. This species is the only tDiplomystus analis, Cope, 1877, and tD. known freshwater tDiplomystus. pectorosus Cope, 1877, and probably tD. ROOFING BONES AND FONTANELLES: Be- theta (Cope, 1874); the first two, and prob- cause of lateral compression and lack of dor- ably the third are subjective junior synonyms soventrally oriented specimens, little can be (see Grande, 1980, pp. 90-91). said about the skull roof. AMNH 10467 1982 GRANDE: tDIPLOMYSTUS 7

I I/

A

Bt

FIG. 1. tDiplomystus dentatus Cope, 1877. A, large specimen with fin margins preserved, total length 48 cm. (from Grande, 1980); B, specimen missing the scales and showing the post cranial skeleton (SMMP 78.9.14), total length 8.5 cm.; negative reversed. From the Eocene Green River Formation. shows that the parietals are in advance of the occipital). The sensory canal branches on the supraoccipital, and appear to meet at the parietal (visible on AMNH 10465). At least dorsal midline of the skull as in tOrnategu- some of the canals of the skull roof appear lum, tArmigatus, and all tellimmichthyids to be contained in "crests" as described for (and unlike clupeiforms which have the pa- tArmigatus brevissimus by Patterson (1967, rietals completely separated by the supra- p. 104). The supraoccipital crest (well pre- 8 AMERICAN MUSEUM NOVITATES NO. 2728

served on AMNH 10467) is relatively large and well developed. ORBITAL REGION: AMNH 763 (fig. 7) shows the dermosphenotic and the last three infraorbitals with the tube for the sensory canal preserved intact. The infraorbital canal was clearly branched in the dermosphenotic, and did not merge with the preopercular ca- nal as it does in clupeiforms thus indicating the absence of a recessus lateralis. The sen- R sory canal also branches in the lowermost FIG. 3. tDiplomystus dentatus Cope, 1877. preserved infraorbital bone (io 3?). A com- The lower jaw, medial view, drawn from AMNH plete count of the infraorbital bones was not 10468 (s.l. 147 mm.), teeth missing. Reversed. possible, but there were at least six infraor- bitals, counting the dermosphenotic (Forey, 1973, counted six in tOrnategulum; Patter- son, 1967, counted seven in tArmigatus). and all tellimmichthyids have strong endo- PARASPHENOID AND ENDOPTERYGOID: pterygoid teeth. Like tArmigatus, Osteo- tDiplomystus dentatus, like other tellim- glossum, and various other teleosts and un- michthyids and tArmigatus, and unlike clu- like clupeiforms, tD. dentatus has a peiforms and tOrnategulum, has a patch of basipterygoid process (fig. 7 in Patterson, conical teeth on the posterior end of the 1967, for tArmigatus, fig. 6, here, for tD. parasphenoid (fig. 6) which looks very much dentatus). The full size of this process could like the parasphenoid tooth patch in Osteo- not be established, because it was broken in glossum. tDiplomystus dentatus has strong all specimens of tDiplomystus where this endopterygoid teeth, much like those of Os- feature was visible. teoglossum. tOrnategulum, tArmigatus, JAWS: The dentary, maxilla, and premax- illa of tDiplomystus bear a single row of small, numerous conical teeth (usually not visible on very small specimens). There are two supramaxillary bones. Unlike tOrnateg- ulum, but like most higher clupeomorphs, Smx 1 there is no sculpturing on the supramaxillary Smx 2 bones. The dentary, maxilla, premaxilla, and supramaxillary bones are illustrated in fig- ures 2 and 3. OPERCULAR SERIES AND HYPOBRANCHIAL APPARATUS: The opercular bones are smooth without noticeable sculpturing. The preoper- cular canal has four branches in the lower arm and about three in the upper arm (tAr- PmxMx migatus brevissimus has about 21 branches Pmx total, tOrnategulum has at least seven, and the condition is not well known in other clu- peomorphs). The vertical arm of the pre- opercle is considerably longer than the lower FIG. 2. tDiplomystus dentatus Cope, 1877. arm. The bones of the opercular series are External dermal bones of the upper jaw. Mx, 4. Smx, and Smx2 all drawn from AMNH 8168 (s.l. illustrated in figure 57 mm.). Pmx drawn from AMNH 10465 (s.l. 118 Branchiostegal rays number about 12; mm.) and drawn smaller to approximately match curved and rodlike anteriorly, becoming the Mx in proportionate size. Reversed. more spatulate posteriorly. The anterior cer- 1982 GRANDE: tDIPLOMYSTUS 9

pansions [of the] anterior neural spines" that Cope (1884, p. 77) described are actually laminar expansions of the predorsal bones (figs. 5 and 1). This is apparent in young in- dividuals that are not fully developed (e.g., AMNH 10470). Similar laminar expansions can also be found in tArmigatus, all tellim- Q michthyids and some clupeoids (e.g., Pota- Pop malosa). The dorsal fin has 11 or 12 principal rays, the first of which is unbranched, and is the longest ray. The principal dorsal rays are preceded by one or two accessory rays.5 The

Iop Sop 5For definition of "accessory rays" see Grande, 1980. FIG. 4. tDiplomystus dentatus Cope, 1877. Bones of the opercular series, drawn from AMNH 8168 (s.l. 57 mm.). Drawing of OP reversed so Bt p that anterior for all bones to the left. Dashed lines indicate margin preserved by impression only. atohyal has a large foramen just above the midsection of the center (=the "Beryciform foramen" of McAllister, 1968, p. 6, fig. 8A). Enpt t VERTEBRAL COLUMN, DORSAL, AND ANAL Ps t FINS: The specimens examined here had 41- FIG. 6. tDiplomystus dentatus Cope, 1877. 43 vertebrae (n = 6), usually 41 or 42, ex- Endopterygoid teeth and part of parasphenoid cluding the two ural centra. The third ver- showing the Osteoglossum-like tooth patch and tebra through the twentieth or twenty-first part of the basipterygoid process. Drawn from bear ribs. The "antero-posterior laminar ex- AMNH 10468 (s.l. 143 mm.). Reversed.

A B C (27) (35) (46)

FIG. 5. tDiplomystus dentatus Cope, 1877. Growth series showing the development of predorsal bones 4 and 5. Number in parentheses refers to s.l. of each fish in millimeters. A through C were drawn from the following specimens: A, AMNH 10470; B, AMNH 10471; and C, AMNH 10469. C is reversed. 10 AMERICAN MUSEUM NOVITATES NO. 2728

0 / II I/ I

Pop

"I A B

FIG. 7. tDiplomystus dentatus Cope, 1877. Preopercle and infraorbitals, with canal system pre- served: A, AMNH 763 (s.l. 82 mm.); B, drawing of same specimen. The preopercular canal and the infraorbital canal clearly do not merge, indicating the absence of a recessus lateralis. Black dashed line indicates areas preserved by impression in matrix only. Reversed. dorsal fin is supported by usually 11 or 12 The predorsal scutes of tDiplomystus den- pterygiophores. The dorsal fin margin is not tatus continued to develop throughout life curved. (fig. 9). The subrectangular scute with the The anal fin usually has 39 or 40 principal pectinate posterior border is thought to be a rays, but the possible range is somewhat derived character of tD. dentatus, tD. birdi, larger (Grande, 1980, found 36 to 42 rays in and tD. dubertreti, because no other known specimens examined there). The first prin- animal has this type of dorsal scute6 (see cipal anal ray is unbranched, is the longest Appendix). The dorsal scutes in tDiplomys- ray, and is preceded by one or two accessory rays, the first of which is usually unseg- mented. The anal fin is supported usually by 6 The predorsal scales of Brevoortia tyrannus super- 39 or 40 pterygiophores, but the range can ficially resemble the dorsal scutes of tDiplomystus, but be from 35 to 42 The anal fin close examination reveals that the condition in Bre- (Grande, 1980). voortia is merely a series of pairs of medially overlap- margin is very slightly falcate anteriorly. ping scales (Monod, 1961; for example, fig. 6). Dorsal PREDORSAL BONES AND SCUTES: There scutes, on the other hand, are unpaired, median ele- are seven or eight (usually seven) predorsal ments which are more heavily ossified than the scales bones with thin anteroposterior laminar ex- and which lie under the skin (usually themselves cov- pansions (explained above). These laminar ered with scales). Brevoortia has no dorsal scutes. expansions are best developed in adult spec- There is no evidence to indicate that dorsal scutes are imens (figs. 5 and 1). derived from scales. 1982 GRANDE: tDIPLOMYSTUS I1I tus usually number between 33 and 36 and run from the posterior edge of the supraoc- cipital to the dorsal fin, along the dorsal mid- line. The abdominal scutes are similar to those found in other tellimmichthyids, tArmiga- tus, and many clupeiforms. They run from the isthmus back to the anus, along the ven- tral midline (fig. 1). The most anterior ab- dominal scutes are small; they increase in size posteriorly to near the insertion of the pectoral fin after which they are all approx- imately the same size. They usually number C",V 32 to 34. PAIRED FINS: The pectoral fin usually has 14 or 15 rays, although some of the smaller rays are frequently missing (possibly due to lack of preservation) resulting in counts as low as 12 (Grande, 1980). The pelvic fin inserts slightly anterior to the dorsal fin (fig. 1) and has usually seven FIG. 8. tDiplomystus dentatus Cope, 1877. A) (rarely six) rays, the first of which is un- The branchiostegal support apparatus showing branched. The pelvic fin supports (visible on the "Beryciform foramen" of McAllister (see text AMNH 10469) are long and thin and contact for explanation). Drawn from AMNH 10468 (s.l. each other posteriorly, but are free anterior- 147 mm.). Reversed. B) Isolated scale, drawn ly. from holotype AMNH 2477 showing the curved, CAUDAL SKELETON AND FIN: Cavender vertically arranged circuli. (Growth lines not (1966) described the caudal skeleton of tDip- shown here, but illustrated in Grande, 1980, fig. lomystus dentatus (=his "tDiplomystus II.49b.) Many circuli between the five drawn are sp.") in detail, and the specimens examined omitted. Reversed. here agree with his description (fig. 10). There are six hypurals; the first articulates principal rays, one unbranched and nine with, but is not fused to, the first ural cen- branched in the upper lobe, and eight trum (as in tOrnategulum, tArmigatus, and branched and one unbranched in the lower all tellimmichthyids). In most clupeiforms, lobe. the first hypural does not articulate with the SCALES: Scales small; circuli (surface) vertebral column (fig. 19). As in all clupeo- semicircular to almost vertical across the morphs, the second hypural is fused to the scale (see fig. 8B) as in most clupeiforms (not first ural centrum. The parhypural is fused to concentrically arranged around the nucleus the first preural centrum (as in all other tel- as stated in Forey, 1973, p. 1314). The circuli limmichthyids, tOrnategulum, tArmigatus, are well preserved on the holotype (AMNH and denticipitoids; and unlike clupeoids). 2477) and the pattern appears to be the same The ural centra are relatively large and well in USNM 4005 (holotype of tD. analis-syn- developed as in tOrnategulum, tArmigatus, onym of tD. dentatus). There are growth and all tellimmichthyids. There are three rings which are arranged concentrically epurals. Of the three uroneurals, none is around the nucleus (fig. II 49b in Grande, fused to the vertebral column. In clupeoids, 1980). In all known clupeomorphs, the circuli the first uroneural fuses anteriorly with the (in the sense of Lagler, 1947, p. 150) are not first preural centrum. The caudal fin is sharp- coincident with the growth rings of the scale. ly forked, and the lower lobe is slightly lon- Although growth rings (when visible) are ar- ger than the upper (fig. IA). It contains 19 ranged concentrically around the nucleus, 12 AMERICAN MUSEUM NOVITATES NO. 2728

A B D E (27) (35) (62) (84)

cz_3zz

F G H I J (124) (174) (223) (319) (393)

FIG. 9. tDiplomystus dentatus Cope, 1877. Ontogenetic series showing the development of the dorsal scute. Number in parentheses refers to s.l. of each fish in millimeters. Drawings A through J were made from the following specimens: A, AMNH 10470; B, AMNH 10471; C, AMNH 2883; D, AMNH 10473; E, AMNH 10474; F, AMNH 10475; G, AMNH 2979; H, AMNH 2480; I, AMNH 8109; and J, AMNH 10476. Scutes oriented so that anterior to the left (as are all illustrations in this paper).

the fine surface ridges (circuli) are either 10465. The lateral line runs all the way back semicircular [as in tDiplomystus (fig. 8B), to the caudal region. Radii or "grooves" ab- Potamalosa and many other clupeiforms] or sent. nearly vertical (as in Chirocentrus and some other clupeoid species). Bardack's (1965, p. 51, fig. F or G?) drawing of a scale from Chi- tDiplomystus birdi Woodward, 1895 rocentrus dorab does not show the pattern Figures 11-13 of circuli (the concentric lines drawn are pos- sibly the growth lines). TYPE: BMNH P 96, illustrated in Wood- There are approximately 82-90 scales ward, 1895, plate I, figure 3. along the lateral line, 18 scale rows above REFERENCE SPECIMENS USED HERE: the vertebral column and 27 scale rows be- AMNH 5745, 5798, 6113, 10188, 10189, and low. Lateral line scales are present (unlike 11106. clupeoids) and are well preserved on AMNH HORIZON AND LOCALITY FOR HOLOTYPE: 1982 GRANDE: tDIPLOMYSTUS 13

The Upper Cretaceous marine limestone de- posits at Hakel, Mount Lebanon, Lebanon. Pu1 U2 NS Ep LOCALITIES FOR REFERENCE SPECIMENS: Reference specimens from Hajula, another -6 Upper Cretaceous marine limestone deposit in Lebanon and Hakel. REVISED DIAGNOSIS: A small, fairly deep- bodied fish that differs from all other tellim- ~Ja michthyids in having about 17 principal dor- H2~V Hy sal fin rays (16-18); usually 16 dorsal pteryg- iophores (15-17); and about 30 preural vertebrae. 12 ETYMOLOGY: birdi-named for Rev. Wil- liam Bird, a noted student of Syrian geology. ~PH DESCRIPTION AND DISCUSSION: tDiplo- s mystus birdi is the smallest known species of FIG. 10. tDiplomystus dentatus Cope, 1877. tDiplomystus (the largest specimen known Caudal skeleton, after a drawing by Cavender to the author has a total length slightly more (1966) of UMMP 52891 (s.l. 84 mm.). Hypurals, than 11 cm.). Like fD. dentatus and tD. du- epurals and first ural centrum colored black. bertreti, the dorsal scutes are pectinate along the posterior border and the number of scute because the limestone containing these fos- spines increases with the age of the fish (see sils is often quite strained. Normally the fig. 13). body shape resembles that shown in figure The body depth and shape are variable, 11 and in Woodward (1895, figs. 3 and 4).

FIG. 11. tDiplomystus birdi Woodward, 1895. Nearly complete specimen, AMNH 10188 (s.l. 62 mm.). From the Upper Cretaceous of Hakel, Lebanon. 14 AMERICAN MUSEUM NOVITATES NO. 2728 NS U2 Ep Un

Hy

t2 (19) (65) FIG. 13. tDiplomystus birdi Woodward, 1895, dorsal scutes; number in parentheses refers to s.l. HS fPH of the fish in millimeters; juvenile scute from AMNH 10189 and adult from AMNH 11106 (the FIG. 12. tDiplomystus birdi Woodward, 1895, adult specimen is broken so s.l. is an estimate). caudal skeleton, drawn from AMNH 10188 (s.l. Dashed lines represent restoration. 65 mm.), hypurals, epurals, and first ural centrum colored black.

rays. tDiplomystus dubertreti is also a larger species than its closest relative, tD. birdi; it Dorsal scutes number about 23 and there reaches a total length of at least 16.5 cm. are seven predorsal bones. The dorsal fin has (Diagnosis after Signeux, 1951.) about 17 principal rays plus two accessory DESCRIPTION AND DISCUSSION: This rays; and there are about 16 or 17 dorsal pte- species is quite rare and is known by only a rygiophores. The anal fin has about 23 prin- few specimens. The body appears to have cipal rays preceded by one or two accessory been extremely deep (fig. 14 here and fig. 1 rays. There are about 24 anal pterygiophores in Signeux), but this is based on only two (23-25, usually 24). The small pelvic fin is specimens, and the rock from Sahal Alma slightly anterior to the dorsal fin. There are (the only known locality for this species) is about 30 preural vertebrae, and or 12 pairs often quite strained. of ribs. Abdominal scutes number about 24; This species has dorsal scutes very similar branchiostegals about 12. For further de- to those of tD. birdi (with pectinate borders scription see Woodward, 1895, and figures bearing fewer spines than fD. dentatus). 11-13 here. Dorsal scutes number about 20, and there are seven predorsal bones. The dorsal fin has tDiplomystus dubertreti Signeux, 1951 about 21 to 23 principal rays preceded by two Figure 14 small accessory rays. The anal fin has 27 TYPE: MNHN 1946-18-17, figure 14. principal rays (supported by 27 pterygio- REFERENCE SPECIMENS: MNHN 1946-18- phores). There are about 33 or 34 preural vertebrae and 12 or 13 pairs of ribs. For fur- 248 (fig. 1, Signeux, 1951). ther see HoRIZON AND LOCALITY FOR HOLOTYPE: description Signeux (1951). The Upper Cretaceous marine chalk deposits at Sahel Alma, Lebanon. This species has tELLIMMICHTHYS JORDAN, 1919 not been reported elsewhere. TYPE SPECIES: tDiplomystus longicosta- REVISED DIAGNOSIS: A deep-bodied fish tus Cope, 1886. Designated by Jordan 1919 that differs from all other tellimmichthyids [in Jordan and Gilbert, 1919, p. 27]. in having about 21 to 23 principal dorsal fin GENERIC DIAGNOSIS: Clupeomorph fishes 1982 GRANDE: tDIPLOMYSTUS 15

. -

I w At t s ~~~~~~~~~~i

FIG. 14. tDiplomystus dubertreti Signeux, 1951; holotype (MNHN 1946-18-17) (s.1. 13 cm.). From the Upper Cretaceous of Sahel Alma, Lebanon. closely related to tDiplomystus, sharing a lomystus is quite justifiable. Its placement subrectangular dorsal scute shape, and dif- into tEllimmichthys also permits construc- fering from that genus in lacking the pecti- tion of an unoccupied family name for the nate scute border. The middle posterior bor- tDiplomystus-tEllimmichthys group (see der of the scute is recessed (fig. 18). above). ETYMOLOGY: ellim-from tEllimma Jor- dan (1913); Ellimma-etymology not speci- fied by Jordan; ichthys-a Greek word for tEllimmichthys longicostatus fish; gender masculine. (Cope, 1886) DISCUSSION: Because the dorsal scutes of Figures 15-18 "tD." Iongicostatus lack the pectinate pos- tDiplomystus longicostatus Cope, 1886; terior border diagnostic of tDiplomystus (see and tEllipes longicostatus Jordan, 1910. Jordan, 1907, p. 136; and above and below), First name changed here for reasons ex- Jordan's removal of this species from tDip- plained above in discussion of tEllimmich- FIG. 15. tEllimmichthys longicostatus (Cope, 1886); slab with two nearly complete specimens (AMNH 734). Upper specimen (dotted in outline) is designated as neotype. Scale = 5 cm.; specimen coated with ammonium chloride. From the Lower Cretaceous of Bahia, Brazil. 1982 GRANDE: tDIPLOMYSTUS 17

FIG. 16. tEllimmichthys longicostatus (Cope, 1886); closeup of the left side of the skull roof from the neotype (from fig. 15) which shows sculpturing diagnostic of this species. thyidae, new family; tEllipes is preoccupied (although Cope, 1886, p. 3, states that they by Ellipes Scudder, 1902, a genus of crick- are longer than wide). There is ornamenta- ets. tion on the dorsal surface of the scute. Al- TYPE: Lost (first reported lost by Schaef- though the dorsal surface of the scutes are fer, 1947); should be in Department of Ver- embedded in the rock, the dorsal ornamen- tebrate Paleontology, AMND, but still can- tation is visible due to the translucency of not be located. the scutes. REFERENCE SPECIMEN AND NEOTYPE: There are seven predorsal bones, and the AMNH 734, a slab with two nearly complete dorsal scutes number about 12. fish, illustrated in figure 15. The upper, more The body depth is extremely deep (as in complete specimen is here designated as the tD. dubertreti) being about 63 percent of the neotype. standard length. HORIZON AND LOCALITY FOR REFERENCE SPECIMEN, NEOTYPE, AND LOST TYPE: From Lower Cretaceous deposits along the coast near Itacaranha, Province of Bahia, U2 NS Brazil. Preserved in a black sandstone (con- U1 sidered to be marine by Cope, 1886, p. 4, and others). REVISED DIAGNOSIS: An extremely deep- PUi bodied fish (greatest body depth about 63 percent of standard length) that differs from all other tellimmichthyids in having strongly sculptured skull roofing bones; only 10 prin- cipal anal fin rays and about nine anal pte- rygiophores; and 22 or 23 pairs of ribs. ETYMOLOGY: longicostatus, having long ribs (from Latin); gender feminine. Pu2 1 DESCRIPTION AND DISCUSSION: The dorsal scutes are well preserved on AMNH 734 and one scute is illustrated in figure 17. Although HS PH the posterior border is interrupted near the FIG. 17. tEllimmichthys longicostatus (Cope, center ("emarginate" of Cope, 1886, p. 3) 1886); caudal skeleton, after Patterson and Rosen the overall shape is still subrectangular. The (1977) (from BMNH P.7109); hypurals, epurals, dorsal scutes are definitely wider than long and first ural centrum colored black. 18 AMERICAN MUSEUM NOVITATES NO. 2728

Pu2 Un

3 Hy t2 1 FIG. 18. tEllimmichthys longicostatus (Cope, 1886) new combination; dorsal scute drawn from AMNH 734 (neotype; s.l. 87 mm.). Anterior to left. HS FIG. 19. tKnightia eocaena Jordan, 1907, The anal fin has 10 principal rays preceded showing the typical clupeoid caudal skeleton by an accessory ray; and there are nine anal (note fusion of Un, to PU.2); drawn from AMNH 10461 (s.l. 104 mm.); hypurals, epurals, pterygiophores. The dorsal fin has 10 or I 1 and first principal rays and two accessory rays; there ural centrum colored black. are about 12 dorsal pterygiophores. The pel- vic fin, although broken, appears to be very small (lower specimen in fig. 15) in AMNH tus on the basis of its dorsal scutes, and "all 734 and is in advance of the dorsal fin. the typical characters of tDiplomystus" There are 36 or 37 preural vertebrae (with (characters which he does not specify and only about 10 caudal vertebrae); and 22 or which are evidently plesiomorphous). This 23 pairs of ribs. The abdominal scutes extend taxon cannot be placed in tDiplomystus be- backward from the isthmus to the anus, and cause it would make that genus paraphyletic. number about 28. It also cannot be retained in the genus Clu- TAXA REMOVED FROM THE pea for the same reasons (it is not even a clupeiform as shown by Patterson, 1967). GENUS tDIPLOMYSTUS The species is here placed in tArmigatus, Nearly all types of dorsal scutes found in new genus (see above). teleosts are examined here (see Appendix "tDiplomystus" primotinus Uyeno, 1979, and figs. 9, 13, and 18). On the basis of this and "tD." kokuraensis Uyeno, 1979, both information, tDiplomystus dentatus Cope, from Early Cretaceous rocks of Japan, are 1877, tD. birdi Woodward, 1895, and tD. removed from tDiplomystus here because no dubertreti Signeux, 1951, are more closely character information was given to warrant related to each other than to any other placement in that genus. On the basis of the known species; and "tD." iongicostatlus scute drawings given by Uyeno (1979, fig. Cope, 1886, is proposed as the sister group IA-D) and Uyeno and Yabumoto (1980, figs. to these three species (explained below) and 2 and 3), and a reduced centrum UI (illus- is placed in tEllimmichthys Jordan for rea- trated in Uyeno and Yabumoto, 1980, fig. 1), sons explained above (fig. 20). these appear to be clupeoids. Better pre- No shared derived characters were found served skull, dorsal scute, and caudal skel- to indicate that "tDiplomystus" brevissimus eton material is needed to further classify (Blainville, 1818) (originally described as these fossils. tClupea brevissimus Blainville, 1818) is an Schaeffer (1947, p. 22) put tEllimma el- tellimmichthyid. Woodward (1888, p. 134) modenae Jordan and Gilbert, 1919 (from the placed this species in the genus tDiplomys- Miocene Monterey Formation of southern 1982 GRANDE: tDIPLOMYSTUS 19

California) into tDiplomystus, without ex- Gaudant, 1971 (from the Upper Cretaceous plaining why (although he refers it to of Tunisia) was placed in tDiplomystus on tKnightia elsewhere in that paper). The type the basis of primitive characters common to (CAS 55404) was examined and found to be most major clupeomorph groups and of its a clupeoid, rather than a species of tDiplo- subtriangular dorsal scutes (p. 158). The sub- mystus. triangular dorsal scute is thought here to be "tDiplomystus" tenuissimus de Stefano, the primitive type of dorsal scute because it 1918, was found by Arambourg (1927, p. 42) occurs in primitive clupeomorphs (tArmi- to be a myctophid. gatus), clupeids (Ethmidium), and several tClupea vectensis Newton, 1889 (from the engraulids (see Appendix). "tD." solignaci Oligocene of the Isle of Wight) was placed is removed from tDiplomystus and classified into tDiplomystus by Woodward (1889), but as Clupeomorpha, incertae sedis. this was because of similarities to tKnightia tHistiurus elatus Costa, 1850 (see Costa, ("tD. humilis" at that time). Examination of 1864), tH. serioloides Costa, 1864, and tH. the dorsal scutes and other skeletal elements ventricosus Costa, 1865 (all from Upper Cre- suggest placement in tKnightia or in a group taceous rocks of Italy) were placed in tDip- closely related to tKnightia. lomystus by Woodward (1901, p. 146). He "tDiplomystus" marmorensis Woodward gave no reason for doing so, and the original (in Newton, 1904) (from the Miocene of Tur- descriptions and illustrations do not indicate key) was originally placed in the tKnightia an assignment in tDiplomystus. They are group by Woodward, and was found here to therefore regarded as Clupeomorpha, incer- be a clupeoid. No dorsal scutes were ob- tae sedis. served on the holotype (BMNH P 10015). tDiplomystus coverhamensis from Upper "tDiplomystus" dartevellei Casier, 1965 Cretaceous deposits of New Zealand was (from Lower Cretaceous deposits of the Af- described (Chapman, 1918, p. 26) as resem- rican Congo) was placed in tDiplomystus on bling [tArmigatus] brevissimus Blainville. the basis of its deep body, prepelvic scutes There are no characters to warrant its place- and other primitive characters, not diagnos- ment into tDiplomystus as used here. It is tic of tDiplomystus, but of a larger group removed from tDiplomystus and placed in including tellimmichthyids, clupeiforms and Clupeomorpha, incertae sedis. tArmigatus. It is therefore removed from One extant taxon from Australia, Hyper- tDiplomystus and placed as Clupeomorpha, lophus sprattelides Ogilby, 1892 [a junior incertae sedis until it can be classified. synonym of H. vittatus (Castelnau, 1875)] "tDiplomystus" goodi Eastman, 1912 was placed in tDiplomystus by Woodward (from Cretaceous deposits of West Africa) (1892, p. 413). Hyperlophus is a clupeoid. was examined (AMNH 6146, 6151, 6162, There are no known extant tellimmichthy- 6166, and 6168), and none of the specimens ids. observed were well enough preserved to classify accurately. Eastman's description INTERRELATIONSHIPS OF has no character information to justify place- CLUPEOMORPHS ment in tDiplomystus. Taverne (1975) de- tOrnategulum sardinioides (Pictet) is de- scribed "tD." goodi as having a well-devel- scribed in detail by Forey (1973) and tAr- oped first ural centrum, parasphenoid teeth, migatus brevissimus (Blainville) by Patter- basipterygoid process, parietals in contact son (1967). A hypothesis of interrelationships on skull, and a free first uroneural; but these for clupeomorphs is given by the cladogram are all primitive clupeomorph characters and in figure 20. The numbers in the cladogram do not indicate that the taxon belongs in refer to shared derived characters listed in tDiplomystus. Therefore, "tD." goodi is re- the classification below. The classification is moved from tDiplomystus and placed in Clu- arranged so that each taxon is followed by peomorpha, incertae sedis. the members it includes (given in brackets), "tDiplomystus" solignaci Gaudant and and the synapomorphies for the group (de- 20 AMERICAN MUSEUM NOVITATES NO. 2728

(1)~ I I 11~~

,/4;'s/- t o ~y

SI ~ ~~~~~~~0 t.

12,13,14,15

ivision 1 Division 2 4,5,6 Clupeomorpha 1,2,3 Clupeocephala FIG. 20. A cladogram of proposed interrelationships of tellimmichthyids with other clupeomorph fishes. The numbers on the cladogram refer to synapomorphic characters listed in the text. The tEllimmichthyidae is contained in the order, tEllimmichthyiformes (not shown). SD = subdivision. For definition of Clupeocephala and Euteleostei see Patterson and Rosen (1977).

rived characters shared by all the members containing tDiplomystus as revised here, of the group). For example, the Clupei- and the genus tEllimmichthys) is not in Clu- formes contains the Denticipitoidei and the peiformes. Therefore tKnightia and several Clupeoidei which share derived characters other clupeoid species placed in tDiplomys- 12, 13, 14, and 15. Because monotypic taxa tus by various authors (discussed above) have no synapomorphies, no characters are must be removed from tDiplomystus. Con- listed for them; instead, there is a reference sequently, anatomical comparisons by Nel- to a specific diagnosis which contains the son, Forey, and others using the so-called autapomorphies which define these species. "tD. humilis" as representative of tDiplo- Some taxonomic groups remain unnamed mystus must be re-evaluated because "tD. here (Division one and two of the Clupeo- humilis" is, in fact, the type species of the morpha, for example) until a study of the in- clupeoid tKnightia (see fn. 2). terrelationships of fossil and recent clu- Also, no character information could be peoids (in progress) can be completed. A discovered to resolve a trichotomy involving biogeographic cladogram of the groups tArmigatus, tellimmichthyids and Clupei- shown in figure 20 is given in figure 21. formes; but all three groups make up the sis- The proposed theory of interrelationships ter group to tOrnategulum. Based on mate- in figures 20 and 21 indicates several things. rial observed here, the North American tel- First of all, the tEllimmichthyidae (a group limmichthyiform, tD. dentatus, is more 1982 GRANDE: tDIPLOMYSTUS 21

0 ~0" 0oC AZ:. £z. 0~ 'Ir CV V - 7 ..N . 1 . 0IY. w.Z) (IT(ry / , C. O; 'Q Z ,b e WS' A~c0 "- 01~- 0,co ",,9 s,, ,9

FIG. 21. The geographic localities and known ages of the groups shown in figure 20. L.C. = Lower Cretaceous, U.C. = Upper Cretaceous, E. = Eocene, 0. = Oligocene and R. = Recent. closely related to an unnamed group from neus-a gonorynchid; and tPhareodus-an Syria and Lebanon (containing tD. birdi and osteoglossoid) in that it has no close relatives tD. dubertreti) than to any other known in the Recent North American fish fauna. taxon. If tellimmichthyiforms and clupeiforms Clupeomorpha [contains divisions 1 and 2 as used exclusively shared a common ancestor, then here]. (1) Hypural 2 fused with the first ural cen- trum at all stages of development (and hypural I divergence from that common ancestor took free from first ural centrum). (2) Supratemporal place at least as early as Lower Cretaceous. commissural sensory canal primitively passing This would be indicated by the fact that both through parietals and supraoccipital (see Patter- tellimmichthyiforms and clupeiforms are son and Rosen, 1977). (3) Otophysic connection known as Lower Cretaceous fossils (tellim- involving a diverticulum of the swim-bladder that michthyiforms by the Lower Cretaceous penetrates the exoccipital and extends into the tEllimmichthys discussed here, and clupei- prootic within the lateral wall of the braincase forms by "Clupavus sp.," illustrated in figs. (Patterson and Rosen, 1977). 10 and 11 of Taverne, 1977). tDiplomystus Division 1 [monotypic, contains tOrnategulum dentatus the most recent sardinioides (Pictet)-see Forey, 1973]. is species of tellim- Division 2 [contains tArmigatus brevissimus michthyid known, and it is known only from (Blainville), new genus, tEllimmichthyi- the Early and Middle Eocene of North formes, new order, and Clupeiformes]. (4) America. tDiplomystus dentatus is similar to The development of dorsal scutes with a me- several other fish species of the Eocene dian keel primitively subtriangular in shape Green River Formation (such as tNotogo- (see Appendix). (5) The development of pre- 22 AMERICAN MUSEUM NOVITATES NO. 2728

pelvic and postpelvic abdominal scutes, with separates parietals (vs. parietals meeting in median spines. (6) The presence of a well- middle of skull in lower clupeomorphs). defined pre-epiotic fossa (assumed to be sec- (14) Reduction in size and/or number of ondarily lost in Denticeps, possibly obliter- teeth on the endopterygoid (vs. strong nu- ated by expansion of pterotic bulla-see merous teeth). (15) Loss of so-called "Be- Greenwood, 1968, p. 232) (Forey, 1973). ryciform foramen" (see McAllister, 1968) tArmigatus brevissimus [see Patterson, 1967, of the anterior ceratohyal. for description]. Denticipitoidei [contains the monotypic tEllimmichthyiformes, new order [contains genera Denticeps and tPaleodenticeps]. one family (tEllimmichthyidae) with three (16) The presence of odontodes (denti- species of tDiplomystus and one species of cles) covering the dermal bones of the tEllimmichthys. (7) Lateral expansion of skull (Greenwood, 1968). (17) Reduction dorsal scute "wings" which give scute a in number of uroneurals to I (vs. 2 or 3 subrectangular shape (see text). in other clupeomorphs). (18) The pres- tEllimmichthys [monotypic, contains tE. ence of a pelvic plate (Greenwood, 1968, longicostatus (Cope) new combination- p. 269); several other characters given see text]. in Greenwood (1968). tDiplomystus [contains tD. dentatus, tD. Clupeoidei [contains Dussumieriidae, Clu- birdi and tD. dubertreti-see above]. (8) peidae, Engraulidae and Chirocentridae; Spines on edge of posterior dorsal scutes and includes about 310 Recent and about (see text). (9) Increase in number of dor- 90 fossil species]. (19) Reduction in rel- sal scutes (to 22 to 36) and reduction or ative size of the first ural centrum. (20) loss of median recess in the posterior Fusion of the first uroneural with the edge of the scute. first preural centrum. (21) Loss of lateral Subdivision I [contains tD. dentatus line scales. (22) Separation of the par- only-see text]. hypural from the first ural centrum Subdivision 2 [contains tD. birdi and (within clupeoids, the parhypural is tD. dubertreti]. (10) Increase in num- known to be fused to the centrum only ber of dorsal fin rays (to about 17 or in Dussumieria acuta-Gosline, 1960, more) and dorsal pterygiophores (to fig. 7). about 16 or more). (11) Decrease in Following Whitehead (1968) and others, number of preural vertebrae (to 34 or less). the Clupeidae includes five subfamilies: Clu- Clupeiformes [contains Denticipitoidei and peinae, Pellonulinae, Alosinae, Dorosoma- Clupeoidei]. (12) The presence of a reces- tinae, and Pristigasterinae; and the Engrau- sus lateralis (infraorbital canal merges with lidae includes two subfamilies: Engraulinae preopercular canal rather than passing up and Coilinae. into a long dermosphenotic stretching well Clupeoidei also contains the fossil family forward above orbit as in lower clupeo- tClupavidae if Taverne's (1977) restorations morphs). (13) Supraoccipital completely are accurate. APPENDIX DORSAL SCUTES IN CLUPEOMORPH tDiplomystus FISHES: THE "DOUBLE-ARMORED tD. dentatus Cope, 1877 [fig. 9] HERRINGS" tD. birdi Woodward, 1895 [fig. 13] Clupeiformes SYSTEMATIC LIST OF CLUPEOMORPH DORSAL Clupeidae SCUTES ILLUSTRATED HERE Clupeinae tEllimma Clupeomorpha-Division 2 tE. branneri (Jordan, 1910) [fig. 23] tArmigatus brevissimus (Blainville, 1818) [fig. Pellonulinae 22] tKnightia tEllimmichthyiformes (includes tEllimmi- tK. eocaena Jordan, 1907 [fig. 26A] chthyidae only) tK. alta (Leidy, 1873) [fig. 26B] tEllimmichthys tK. new species A (Grande, in press) tE. longicostatus (Cope, 1886) [fig. 18] [fig. 26C] 1982 GRANDE: tDIPLOMYSTUS 23

Potamalosa P. richmondia (Macleay, 1879) [fig. 24] Hyperlophus H. vittatus (Castelnau, 1875) [fig. 25A] H. translucidus McCulloch, 1917 [fig. 25B] incertae sedis "tClupea" vectensis Newton, 1889 (40) (70) [fig. 28] tNew genus and species B (Grande, FIG. 22. tArmigatus brevissimus (Blainville, in press) [fig. 27] 1818). The dorsal scutes in two specimens. The Alosinae specimen on the left was drawn from AMNH 5373 Ethmidium and the specimen on the right was drawn from E. maculatum (Valenciennes, 1847) AMNH 5354. Numbers in parentheses refer to s.l. [fig. 29] of fish in millimeters. Dorosomatinae Clupanodon C. thrissa (Linnaeus, 1758) [fig. 30] C. dussumieri Valenciennes, 1848 [fig. Pristigasterinae 36A] Pristigaster C. mystus (Linnaeus, 1758) [fig. 36B] P. cayana Cuvier, 1829 [fig. 31] C. nasus Gunther, 1868 [fig. 36C] Engraulidae (nomenclature follows Wongra- C. neglecta Whitehead, 1968 [fig. 36D] tana, 1980) Engraulinae Most dorsal scutes are relatively delicate Stolephorus structures and are often not preserved or are S. macrops Hardenburg, 1933 [fig. poorly preserved in fossil species. There- 32A] fore, the dorsal scute counts given below for S. tri (Bleeker, 1852) [fig. 32B] fossil species are estimates based on obser- Thrissina vations of many T. baelama (Forskal, 1775) [fig. 33] specimens. Thryssa tArmigatus: The only known species, tA. T. brevicauda Roberts, 1978 [fig. 34A] brevissimus, has a subtriangular scute (fig. T. dussumieri (Valenciennes, 1848) 22). There are 11 or 12 scutes which extend [fig. 34B] from the anterior end of the dorsal fin for- T. hamiltoni (Gray, 1835) [fig. 34C] ward, about two-thirds of the way to the T. kammalensis (Bleeker, 1849) [fig. skull (dorsal scutes appear to extend all the 34D] way to the skull in all other dorsal-scuted T. mystax (Schneider, 1801) [fig. 34E] clupeomorphs discussed here, except for en- T. purava (Hamilton-Buchanan, 1822) graulids and Pristigaster).7 The length of [fig. 34F] T. setirostris (Broussonet, 1782) [fig. each scute is slightly shorter than that of a 34G] preural centrum. T. vitrirostris (Gilchrist and Thomp- tEllimmichthyidae: tEllimmichthyids all son, 1908) [fig. 34H] have the lateral wings of the dorsal scute Setipinna elongated and blunted at the lateral edges, S. breviceps (Cantor, 1850) [fig. 35A] giving the scute a subrectangular outline. S. melanochir (Bleeker, 1849) [fig. tEllimmichthys longicostatus (fig. 18) has a 35B] complex pattern of sculpture on the dorsal S. papuensis Munro, 1964 [fig. 35C] surface. The length of each scute is about S. gilberti Jordan and Starks, 1905 [fig. equal to the length of one preural centrum 35D] the width is much S. phasa (Hamilton-Buchanan, 1822) (although greater than the [fig. 35E] length as in all tellimmichthyids); and they S. godavari Babu Rao, 1961 [fig. 35F] number about 12. S. taty (Valenciennes, 1848) [fig. 35G] tDiplomystus is highly derived in having Coilinae Coilia 7See addendum. 24 AMERICAN MUSEUM NOVITATES NO. 2728

A FIG. 23. tEllimma branneri (Jordan, 1910). Drawing of two anterior dorsal scutes from AMNH 10040 (s.l. 93 mm.). the posterior border of the scute pectinate; pectination increases with increase in body B size (figs. 9 and 13). tDiplomystus birdi (fig. FIG. 25. Hyperlophus dorsal scutes. A, H. 13) has about 23 dorsal scutes, and each vittatus (Castelnau, 1875) drawn from AMNH scute is slightly shorter in length than the 3050 (s.l. 60 mm.); B, H. translucidus McCulloch length of a centrum. tDiplomystus dentatus 1917, drawn from AM 1. 18464-001 (paratype) (s.l. is further specialized in having a higher num- 48 mm.). ber of dorsal scutes than any other known clupeomorph (33 to 36), and a higher number of scute "spines" (fig. 9) in the adult stage, than any other species of tDiplomystus. The (Grande, in press).] tEllimma branneri has length of the scute in tD. dentatus is less a complex pattern of sculpture on the dorsal than that of a preural centrum. Clupeinae: tEllimma branneri (Jordan, 1910) was the only7 clupeine species found to have dorsal scutes. [This species is not in (63 the genus tKnightia, as placed by Schaeffer (1947) and others, because, among other things, it has two supramaxillary bones (29) A (63)

B C FIG. 26. tKnightia dorsal scutes. A, tK. FIG. 24. Potamalosa richmondia (Macleay, eocaena Jordan, 1907, young individual and an 1879). Dorsal scute drawn from AMNH 28513 (s.l. adult. Numbers in parentheses refer to s.l. in mil- 156 mm.). Above, lateral view showing median limeters. Young individual is AMNH 10420, adult crest running from anterior edge to beyond pos- is AMNH 2499; B, tK. alta (Leidy, 1873), drawn terior edge. The median crest runs the entire from AMNH 10433 (s.l. 86 mm.); C, tK., new length of the scute, along the dorsal surface in species A (description, Grande, in press) from the most clupeomorphs (except for Clupanodon late Middle Paleocene Tongue River Formation, thrissa, Pristigaster cayana, and engraulids). Be- Powder River Co., Montana. Drawn from AMNH low, dorsal view. 10404 (s.l. 70 mm.). 1982 GRANDE: tDIPLOMYSTUS 25

FIG. 27. tNew genus and species of clupeid FIG. 28. "tClupea" vectensis Newton, 1889, from the early Middle Eocene Laney Member of dorsal scute drawn from BMNH P.6854 (s.l. 27 the Green River Formation, Wyoming. Drawn mm.). from AMNH 10458 (s.l. 28 mm.). and each is slightly shorter than the length surface (fig. 23) of the scutes. The scutes of a preural centrum. Hyperlophus translu- number 11 or 12, and the length of each scute cidus (fig. 25B) has about 17 or 18 dorsal is about equal to the length of a preural cen- scutes and each is about the same length as trum. a preural centrum. Pellonulinae: tKnightia, Potamalosa, Hy- Potamalosa has a somewhat elongate dor- perlophus, and possibly "tClupea" vecten- sal scute (about 1.5 centra in length) which sis and the undescribed species in figure 26 narrows anteriorly (fig. 24). Dorsal scutes in are the only known pellonulines with dorsal P. richmondia (the only valid species- scutes. Their scutes are distinctive in having Grande, in prep.) usually number 14. a symmetrical oval to circular shape,8 with Alosinae: Ethmidium maculatum has a median crest that extends over both the about 24 dorsal scutes, and all are subtrian- anterior and the posterior edge of the scute. gular in shape except the most anterior scute Although this scute type occurs in all of the (as in fig. 29) and slightly less in length than "Hyperlophini" (a group name used by a preural centrum. The first (most anterior) Whitehead, 1973, p. 13, to include pellonu- dorsal scute is more rounded and about 50 lines with dorsal scutes), it also occurs in a percent larger than the rest. probable clupein fossil (fig. 27), in some clu- Dorosomatinae: The dorsal scutes of Clu- peoid fossils whose relationships are un- panodon thrissa have an asymmetrical known, and in some Recent clupeins which shape, and the median crest on the dorsal bear only a single scute-see addendum. surface does not reach its anterior or poste- All species of tKnightia (fig. 26) have rior edge (fig. 30). They number about 20 and about 12 to 14 dorsal scutes, which are each are slightly shorter in length than a preural about the length of one preural centrum. The centrum. scutes often are difficult to see on specimens Pristigasterinae: Pristigaster cayana because of incomplete preservation, or a (AMNH 10186 SW, two specimens 95 and 84 covering of the thick scales characteristic of mm. s.l.) has two asymmetrical scutes (fig. this group. "tClupea" vectensis (fig. 28) has 36) between the supraoccipital crest and the a dorsal scute morphology similar to first predorsal bone. The scutes have no me- tKnightia and has about 13 scutes, each dian crest and have a slightly concave dorsal about as long as a centrum. The undescribed surface. The second scute shows a clupeid (description, Grande, in press) very whose scute is illustrated in figure 27, has about 12 or 13 scutes, each about as long as a centrum. Hyperlophus has a scute morphology very similar to that of tKnightia. Hyperlophus vit- tatus (fig. 25A) usually has 29 dorsal scutes

8 The shape is sometimes variable (from oval to cir- FIG. 29. Ethmidium maculatum (Valen- cular) even within a single individual. The normal scute ciennes, 1847), dorsal scute drawn from AMNH types were used here for illustrations. 7738 (s.l. 165 mm.). 26 AMERICAN MUSEUM NOVITATES NO. 2728

A B FIG. 32. Stolephorus dorsal scutes. A, S. ma- crops Hardenburg, 1933, drawn from CAS 46939 (s.l. 50 mm.); B, S. tri (Bleeker, 1852), drawn FIG. 30. Clupanodon thrissa (Linnaeus, 1758). from USNM 204235 (s.1. 62 mm.). Dorsal scute drawn from AMNH 17738 (s.l. 165 mm.). Above, lateral view showing median crest present only on part of the dorsal surface of the scale, and reaching neither the anterior nor the posterior edge; Below, dorsal view. found that Pristigaster does indeed have dor- sal scutes. They are not merely the tops of predorsal bones, and are in advance of the 5 predorsal bones found in this species. The small posterior spine. The anterior (and larg- scutes were observed in both specimens (list- est) scute is slightly less than one PUC in ed above) and were removed from the larger length (PUC = preural centrum). specimen to make the drawing in figure 36. There has been some confusion in the lit- The position of the dorsal scutes in Pris- erature concerning the occurrence of dorsal tigaster is unusual. All other dorsal-scuted scutes in Pristigaster. Valenciennes (1847a, clupeomorphs that lack a complete series of p. 335) gives the first indication of them when scutes from the head back to the dorsal fin he states: (such as tArmigatus and engraulids) lack an- "Les interepineux dorsaux font une petite terior scutes (those just behind the head). saillie au-dessus des muscles, transversent Pristigaster is unique7 in lacking posterior la peau et sortent par deux petites pointes scutes (those just anterior to the dorsal fin). qui rendent cette partie du corps dentel- Signeux (1964) describes a fossil pristigas- lee." [The predorsal bones make small terine, tGasteroclupea branisai from the bumps above the muscles, penetrating the Upper Cretaceous El Molino Fm. of central skin and projecting by two small points Bolivia (also illustrated in Schaeffer, 1963, which make this part of the body dentic- fig. 6). It appears to have about 25 dorsal scutes, but they were badly crushed on the ulated.] specimen observed here. Unlike Pristigas- Bertin and Arambourg (1958, p. 2227) also ter, and like most clupeomorphs with dorsal cite Pristigaster as having predorsal scutes, scutes, the scutes are arranged in a series but Whitehead (1967, p. 101) suggests that along the dorsal midline, running from just the scutes were identified "possibly through behind the head back to the dorsal fin origin. a misinterpretation of the pre-dorsal bones." Engraulidae: Dorsal-scuted engraulids are In the two specimens examined here it was peculiar in having reduced the number of

FIG. 33. Thrissina baelama (Forskal, 1775), FIG. 31. Dorsal scutes in Pristigaster cayana dorsal scute drawn from CAS 29385 (s.l. 110 Cuvier, 1829. Drawn from AMNH 10186 sw (s.l. mm.). The scute spine on this species is very 95 mm.). poorly developed. 1982 GRANDE: tDIPLOMYSTUS 27

A B C D

E F G H FiG. 34. Thryssa dorsal scutes. A, T. brevicauda Roberts, 1978, drawn from AMNH 38190 (s.l. 39 mm.); B, T. dussumieri (Valenciennes, 1848), drawn from AMNH 38191 (s.l. 103 mm.); C, T. hamiltoni (Gray, 1835), drawn from AMNH 38188 (s.l. 90 mm.); D, T. kammalensis (Bleeker, 1849), drawn from AMNH 38189 (s.l. 66 mm.); E, T. mystax (Schneider, 1801), drawn from AMNH 18295 (s.l. 72 mm.); F, T. purava (Hamilton-Buchanan, 1822), drawn from CAS 47099 (s.l. 113 mm.); G, T. setirostris (Broussonet, 1782), drawn from AMNH 38192 (s.l. 88 mm.); H, T. vitrirostris (Gilchrist and Thompson, 1908), drawn from CAS 33932 (s.l. 102 mm.). dorsal scutes to 1 or 2.9 They also have a does have a dorsal scute. It is interesting to dorsoposteriorly pointing spine near the an- note that only the Indo-Pacific engraulids terior end of the scute (fig. 37). There is no have a dorsal scute; and that all Indo-Pacific median crest on engraulid scutes (also a de- engraulids, except for Lycothrissa and some rived feature). Among engraulids, only the Stolephorus species, have a dorsal scute. genera Thrissina, Thryssa, Setipinna, Coilia The dorsal scutes of Setipinna (fig. 35) are and some of the Stolephorus species appear more elongate than those of any other known to have a dorsal scute.10 Papuengraulis (with clupeomorph (about equal in length to 13/4 to one species, P. micropinna Munro, 1964) 51/2 preural centra). Scute length (measured was not available for study, but the illustra- from the anterior edge of the scute to the tion given in Munro (1967) indicates that it posterior tip of the spine) for each species, given in preural centra lengths (PUCs) are 9 Although the author has not observed any engrau- approximately as follows: S. breviceps (51/2 lids with two dorsal scutes, G. Nelson (person com- PUCs); S. melanochir (3 PUCs); S. papuen- mun.) has observed two on rare occasions. sis (31/2 PUCs); S. gilberti (2 PUCs); S. 10 The dorsal scute should not be confused with the phasa (2½2 PUCs); S. godavari (2 PUCs); anterior lateral expansion of the first pterygiophore and S. taty (13/4 PUCs). Except for the which is sometimes spatulate enough to resemble a species with the most elongated scutes, the scute, often bears a small paired ray, and is present in scute in the genus Setipinna is subtriangular both scuted and unscuted engraulids. In dorsal-scuted in outline. engraulids, this pterygiophore expansion underlies the posterior edge of the scute making it difficult to see the The dorsal scutes of Thryssa (fig. 34) are posterior border of the scute (unless the specimen is all basically subtriangular in outline, and cleared and stained). Engraulid dorsal scutes, unlike the range in length from about 1 to 11/2 PUCs, pterygiophore expansions, are unpaired structures except for T. purava which is about 2 PUCs which are not part of any other osteological element, in length. In addition to the species listed in and bear no paired rays. figure 32, T. rastrosa Roberts, 1978 (CAS 28 AMERICAN MUSEUM NOVITATES NO. 2728

A B-

C~~~

D E F G FIG. 35. Setipinna dorsal scutes. A, S. breviceps (Cantor, 1850), drawn from AMNH 17703 (s.l. 134 mm.); B, S. melanochir (Bleeker, 1849), drawn from AMNH 9525 (s.l. 220 mm.); C, S. papuensis Munro, 1964, drawn from AMNH 17551 (s.l. 87 mm.); D, S. gilberti Jordan and Starks, 1905, drawn from CAS 47091 (s.l. 142 mm.); E, S. phasa (Hamilton-Buchanan, 1822) CAS (SU) 25624 (s.l. 155 mm.); F, S. godavari Babu Rao, 1961, drawn from SOSC 4 (s.l. 78 mm.); G, S. taty (Valenciennes, 1848), drawn from SOSC 4 (s.l. 98 mm.).

[SU] 41548); T. scratchleyi (Ramsay and (CAS [SU] 25161); S. buccaneeri Strasburg, Ogilby, 1887) (USNM 217035); and T. mal- 1960 (CAS 30100); S. indicus (van Hasselt, abarica (Bloch, 1795) (USNM 217040) were 1823) (AMNH 18291); S. bataviensis Har- also examined and were found to have dorsal denberg, 1933 (AMNH 27550); S. andhraen- scutes 1 to 11/2 PUCs in length and of the sis Babu Rao, 1966 (USNM 204230); S. hol- same general morphology as illustrated here odon (Boulenger, 1902) (CAS [SU] 31337); for the other species of Thryssa. and S. commersonii Lacepede, 1803 (CAS The dorsal scutes of Coilia are all nearly [SU] 38399) were also examined, and none identical in the four species studied (fig. 36). had dorsal scutes. They are all subtriangular and fairly small The dorsal scute of Thrissina is quite dis- (about 1 PUC). tinctive (fig. 33). Thrissina baelama (For- Within the genus Stolephorus, only some skal, 1775) and T. encrasicholoides (Bleeker, of the species have dorsal scutes. Those ob- 1852) both have the same general scute served here were in S. macrops and S. tri shape. The spine is quite reduced in these (fig. 32). Stolephorus tri shows a typical sub- species and the posterior edge of the scute, triangular shape, and S. macrops has a very unlike that of any other known engraulid, is rounded subtriangular shape, with a very convexly rounded in outline. The length of large spine. The scutes of both species are the scute is about 1 PUC. about 1 PUC. OTHER ACTINOPTERYGIANS WITH DORSAL Stolephorus heterolobus (Riippell, 1837) SCUTES: Some euteleosts (e.g., the alepisau- 1982 GRANDE: tDIPLOMYSTUS 29

the clupeomorph dorsal scute (primitively subtriangular with a median crest extending over the entire length of the dorsal surface) was independently derived for Division 2 of B the Clupeomorpha (fig. 20). A DISCUSSION: All known clupeomorphs with dorsal scutes also have ventral scutes and therefore, all clupeomorphs with dorsal scutes are "double-armored." The presence , I D FIG. 36. Coilia dorsal scutes. A, C. dussu- mieri Valenciennes, 1848, drawn from CAS (SU) 68196 (s.l. 108 mm.); B, C. mystus (Linnaeus, 1758), drawn from AMNH 10322 (s.l. 210 mm.); A~ C, C. nasus Gunther, 1868, drawn from AMNH 10321 (s.l. 125 mm.); D, C. neglecta Whitehead, 1968, drawn from CAS 33904 (s.1. 134 mm.). roids, tEurypholis illustrated in Piveteau, 1966, p. 204, tEnchodus illustrated in Goody, 1969, p. 92, and tSaurorhamphus il- lustrated in Goody, 1969, p. 124), acipenser- ids, and tpycnodontiforms (e.g., tMacro- mesodon, tMicrodon, and tCoelodus, illustrated in Piveteau, 1966, pp. 176, 178, and 179) have dorsal scutes, but these are FIG. 37. Typical engraulid dorsal scute morphologically distinct from clupeomorph (Thryssa kammalensis, drawn from AMNH dorsal scutes. Euteleosts primitively do not 38189) showing the peculiar "spine." A, lateral have dorsal scutes, and it is proposed that view; B, dorsal view.

mc

Iw

FIG. 38. Explanation of morphological terminology used for dorsal scutes. Anterior faces left. A, Potamalosa scute (upper = lateral view; lower = dorsal view); B, tDiplomystus scute (dorsal view). lw = lateral wing; mc = median crest; pb = pectinate border. 30 AMERICAN MUSEUM NOVITATES NO. 2728 of "double armor" is found, at least primi- in number (to one or rarely two), have lost tively, in most major clupeomorph groups the median crest on the dorsal surface, and (Clupeidae, Engraulidae, tEllimmichthyi- have acquired a prominent spine. Some dae, and tArmigatus). Therefore, the pres- species of Setipinna are further specialized ence of double armor (or dorsal scutes) is not in having a greatly elongated dorsal scute synapomorphic for any group more specific (see above). Clupeids have several scute than Division 2 of the Clupeomorpha as types, including a primitive subtriangular shown in figure 20 (which includes all known type in Ethmidium, and several types pro- clupeomorphs except tOrnategulum sardi- posed here as more specialized (less general nioides). Therefore, I agree with Nelson in occurrence) such as the symmetrical cir- (1970b) that the "double-armored herring" cular-to-oval shape in hyperlophins and group proposed by Schaeffer (1947, p. 24) some clupeins, the asymmetrical scute with and implied by Woodward (1892), Eastman an incomplete median crest in Clupanodon (1912), and others is an unnatural (non- thrissa and the slightly concave, asymmet- monophyletic) group. It is proposed here to rical scute in Pristigaster which lacks any be non-monophyletic unless it includes all median crest. Complex sculpturing of the members of Clupeomorpha Division 2. dorsal scute surface appears to have devel- We find when looking at the detailed mor- oped independently in tellimmichthyids phology of dorsal scutes in clupeomorphs, (tEllimmichthys longicostatus) and clupeids that they are complex, and that certain mor- (tEllimma branneri). The pattern of sculp- phological types are indicative of smaller turing is quite different between the two (less general) clupeomorph groups (de- groups. scribed above). A subtriangular scute with a median crest on the dorsal surface is pro- posed as the primitive scute type for Clupeo- ADDENDUM morpha, Division 2 because it is found in tArmigatus, clupeids, and engraulids. I After this paper went to press, Gareth Nel- disagree with Uyeno's (1979, p. 22) state- son and I found that several clupeins and at ment that "the predorsal scutes [which are] least one alosin (at least some species of round at both ends . .. appear to be the most Herklotsichthys, Opisthonema, Sardinella, primitive form, since the predorsal scutes and Alosa) have a single dorsal scute just must have originated from cycloid scales behind the skull and before the first predorsal covering the mid-dorsal line of the predorsal bone (similar in position to the two dorsal region." The dorsal scutes in most clupeo- scutes of Pristigaster). The clupein scute is morphs are embedded in the skin, which is similar in morphology to the "hyperlophin" itself covered with scales. There is no evi- type of scute (see figs. 24-28); oval in shape dence, developmental or otherwise, to indi- (longest diameter in the anterior-posterior cate that the scutes originated from cycloid direction) with a median crest and no sculp- scales along the midline; and the most par- ture on the lateral wings. The occurrence of simonious interpretation of dorsal scute mor- this scute within clupeids is currently being phological characters in combination with studied. other types of clupeomorph characters, sug- gests that the subtriangular scute with a me- LITERATURE CITED dian dorsal crest is the primitive type of dor- sal scute for clupeomorphs. Arambourg, C. 1927. Les poissons fossiles d'Oran. Materiax A subrectangular shape due to lateral ex- pour la carte geologique d'Algerie. Al- pansion is seen as a derived character of giers (1, Pal.), no. 6, pp. 1-298. tellimmichthyids; the appearance of a pec- Bardack, D. tinate posterior border is a derived character 1965. Anatomy and evolution of chirocentrid of tDiplomystus. Engraulid dorsal scutes are fishes. Paleont. Contrib. Univ. Kansas, specialized in that they are greatly reduced Vertebrata, art. 10, pp. 1-88. 1982 GRANDE: tDIPLOMYSTUS 31

Bertin, L., and Arambourg, C. Cuvier, G. 1958. Super-ordre des t6leosteens (Teleostei). 1829. Regne Animal. 2nd edition, no. 2, Paris, In Traite de Zoologie, vol. 13, no. 3. 406 pp. Blainville, H. de Stefano, G. 1818. Sur les ichthyolites ou les poissons fos- 1918. I pesci fossi di Licata. Sicilia Mem. siles. Paris, Nouv. Dict. Hist. Nat., Serv. Desc. Carta Geol. Italia, VII (1), XXXVII, pp. 310-395. pp. 1-102. Casier, E. Dollo, L. 1965. Poissons fossiles de la serie du Kwango 1904. Resultats du voyage du S. Y. Belgica. (Congo). Ann. Mus. Roy. African Poissons. 40, Antwerp, pp. 1-239. Centr., ser. in-8°, Sci. geol., no. 50, pp. Eastman, C. R. 13-16. 1912. Tertiary fish-remains from Spanish Castelnau, C. Guinea in West Africa. Ann. Carnegie 1875. Researches on the fishes of Australia. Mus., vol. VIII, pp. 370-378. Intercol. Exhib. Essays, Victorian Forey, P. L. Dept., 1875, no. 2. 1973. A primitive clupeomorph fish from the Cavender, T. Middle Cenomanian of Hakel, Leba- 1966. The caudal skeleton of the Cretaceous non. Canadian Jour. Earth Sci., vol. 10, teleosts Xiphactinus, Ichthyodectes, no. 8, pp. 1302-1318. and Gillicus, and its bearing on their re- Gaudant, M., and J. Gaudant lationship with Chirocentrus. Occas. 1971. Une nouvelle espece de Diplomystus Pap. Mus. Zool., Univ. Michigan, no. (Poisson teleosteen) dans le Cretac6 su- 650, pp. 1-15. perieur du Sud tunisien. Bull. soc. geol. Chapman, F. de France (7), XIII, 1971 n° 1-2, pp. 1918. Cretaceous and Tertiary fish-remains of 156-159. New Zealand. New Zealand Geol. Goody, P. Surv., Paleont. Bull., no. 7, pp. 1-45. 1969. The relationships of certain Upper Cre- Cope, E. D. taceous teleosts with special reference 1870. Observations on the fishes of the Ter- to the myctophoids. Bull. Brit. Mus. tiary shales of Green River, Wyoming (Nat. Hist.), (Geol.), Suppl. 7, pp. 1- Territory. Proc. Amer. Philos. Soc. XI, 255. pp. 380-384. Gosline, W. A. 1874. Fishes from the freshwater Tertiaries of 1960. Contributions toward a classification of the Rocky Mountains. Bull. U.S. Geol. modern isospondylous fishes. Bull. Brit. and Geog. Surv. of the Terr., vol. 1, no. Mus. (Nat. Hist.) Zool., vol. 6, no. 6, 2, pp. 49-51. pp. 325-365. 1877. A contribution to the knowledge of the Grande, L. ichthyological fauna of the Green River 1980. The paleontology of the Green River shales. Bull. U.S. Geol. and Geog. Formation, with a review of the fish fau- Surv., vol. 3, art. xxxiv, pp. 807-819. na. Bull. Wyoming Geol. Surv., vol. 63, 1884. The Vertebrata of the Tertiary forma- pp. 1-334. tions of the West. U.S. Geol. Surv. 1982. A revision ofthe fossil genus tKnightia, Terr., vol. 3, pp. 1-1009. with a description of a new genus from 1886. A contribution to the vertebrate paleon- the Green River Formation (Teleostei, tology of Brazil. Proc. Amer. Phil. Soc., Clupeidae). Amer. Mus. Novitates. vol. 23, no. 121, pp. 3-4. Greenwood, P. H. Costa, 0. G. 1968. The osteology and relationships of the 1864. Paleontologia del regno di Napoli con- Denticipitidae, a family of clupeomorph tenente la descrizione e figura di tutti gli fishes. Bull. Brit. Mus. (Nat. Hist.), avanzi organici fossili racchiusi suolo di (Zool.), no. 16, pp. 213-273. questo regno. Atti Acad. pont. Naples, Hubbs, C. L., and K. F. Lagler V, 128 pp. 1949. Fishes of the Great Lakes region. 1865. Nuove osservazione e scoperte intorno Bloomfield Hills, Michigan, Cranbrook ai fossili della calcarea ad ittioliti di Pie- Press, 186 pp. traroja. Atti Acad. Sci. Naples, II, pp. Jordan, D. S. 1-12. 1907. The fossil fishes of California; with sup- 32 AMERICAN MUSEUM NOVITATES NO. 2728

plementary notes on other species of Nelson, G. J. extinct fishes. Bull. Dept. Geol., Univ. 1970a. The hypobranchial apparatus of teleos- California, vol. 5, pp. 95-145. tean fishes of the families Engraulidae 1910. Description of a collection of fossil fish- and Chirocentridae. Amer. Mus. Novi- es from the bituminous shales at Riaco tates, no. 2410, pp. 1-30. Doce, state of Alaqoas, Brazil. Ann. 1970b. Dorsal scutes in the Chinese gizzard Carnegie Mus., vol. 7, no. 1, pp. 23-24. shad Clupanodon thrissa (Linnaeus). 1913. Ellimma, a genus of fossil herrings. Japanese Jour. Ichthyol., vol. 17, pp. Proc. Biol. Soc. Washington, vol. 26, p. 13 1-134. 79. 1973. Notes on the structure and relationships 1919. New genera of fossil fishes from Brazil. of certain Cretaceous and Eocene te- Proc. Acad. Nat. Sci. Philadelphia, vol. leostean fishes. Amer. Mus. Novitates, 71, pp. 208-210. no. 2524, pp. 1-3 1. Jordan, D. S., and J. Z. Gilbert Nelson, J. S. 1919. II. Fossil fishes of the Miocene (Mon- 1976. Fishes of the world. New York, John terey) Formations. In Fossil fishes of Wiley and Sons Inc., 416 pp. southern California. Leland Stanford Newton, E. T. Junior University Publications. Univ. 1889. Description of a new species of Clupea series, no. 38, pp. 16-64. (C. vectensis) from Oligocene strata in Lagler, K. F. the Isle of Wight. Quart. Jour. Geol. 1947. Lepidological studies 1. Scale charac- Soc., vol. XIV, pp. 112-117. ters of the families of Great Lakes fish- Newton, R. B. es. Transactions of the American Mi- 1904. Notes on the post-Tertiary and Tertiary croscopical Society, vol. LXVI, no. 2, fossils obtained by Col. English from pp. 149-171. the district surrounding Dardanelles. Linnaeus, C. Quart. Jour. Geol. Soc. London, vol. 1758. Systema naturae. 10th edition, 824 pp. LX, pp. 277-292. Macleay, W. Ogilby, J. D. 1879. On the Clupeidae of Australia. Proc. 1892. On some undescribed reptiles and fishes Linn. Soc. New South Wales, vol. 4, pt. from Australia. Rec. Australian Mus., 3, pp. 363-385. vol. 2, pp. 23-26. McAllister, D. E. Patterson, C. 1968. Evolution of branchiostegals and clas- 1967. Are the Teleosts a polyphyletic group? sification of teleostome fishes. Nat. In Problemes actuels de paleontologie Mus. Canada Bull., no. 221. (evolution des vertebres). Coll. Intern. McCulloch, A. R. Centre Nat. Rech. Scient., no. 163, pp. 1917. Studies in Australian fishes. Rec. Aus- 93-109. tralian Mus., vol. 1, no. 7, pp. 163-188. 1970. A clupeomorph fish from the Gault Mitchill, S. L. (Lower Cretaceous). Jour. Linnean of L. Soc. (Zool.), vol. 49, no. 3, pp. 161-182. 1814. Report, in part Samuel Mitchill, Patterson, C., and D. E. Rosen M.D., Professor of Natural History, on 1977. A review of ichthyodectiform and other the fishes of New York. Reprinted by Mesozoic teleost fishes, and the theory Theodore Gill (ed.), Washington, 1898, and practice of classifying fossils. Bull. pp. 1-30. Amer. Mus. Nat. Hist., vol. 158, art. 2, Monod, T. pp. 81-172. 1961. Brevoortia Gill 1861 et Ethmalosa Re- Pictet, F. J. gan 1917. Bulletin de l'I.F.A.N. T. 1850. Description de quelques poissons fos- XXXIII, ser. A, n° 2. siles du Mont Liban. Geneve, J. G. Munro, I. S. R. Fick, pp. 1-59. 1967. The fishes of New Guinea. Dept. of Piveteau, J. Agr., Stock and Fisheries, Port Mores- 1966. Traite de Paleontologie, Tome IV, vol. by, New Guinea. Victor C. N. Blight, 3, Actinopterygiens, Crossopterygiens, Govt. Printer, Sydney, New South Dipneustes. Paris, Masson et Cie, pp. Wales, pp. 1-650. 1-442. 1982 GRANDE: tDIPLOMYSTUS 33

Rosen, D. E. Thorpe, M. R. 1973. Interrelationships of higher euteleos- 1938. Wyoming Eocene fishes in the Marsh tean fishes. In Greenwood, P. H., R. S. Collection. Amer. Jour. of Sci., vol. 36, Miles, and C. Patterson (eds.), Interre- pp. 279-295. lationships of fishes. London, Academic Uyeno, T. Press, pp. 397-513. 1979. Early Cretaceous freshwater fishes from Schaeffer, B. northern Kyushu, Japan. I. Description 1947. Cretaceous and Tertiary actinopterygi- of two new species of the clupeid genus an fishes from Brazil. Bull. Amer. Mus. Diplomystus. Bull. Kitakyush Mus. Nat. Hist., vol. 89, no. 1, pp. 1-40. Nat. Hist., vol. 1, pp. 11-24. 1963. Cretaceous fishes from Bolivia, with Uyeno, T., and Y. Yabumoto comments on Pristid evolution. Amer. 1980. Early Cretaceous freshwater fishes from Mus. Novitates, no. 2159, pp. 1-20. northern Kyushu, Japan. Bull. Kitak- Signeux, J. yushu Mus. Nat. Hist., vol. 2, pp. 25- 1951. Notes paleoichthyologiques. V. Diplo- 31. mystus dubertreti, une nouvelle espece Valenciennes, A. du Senonien du Liban. Bull. Mus. Natl. 1847a. Histoire naturelle du hareng, compre- Hist. Nat., t. 23, no. 2, pp. 692-693. nant la description zoologique et ana- 1964. Gasteroclupea branisai, clupeide nou- tomique de cet important poisson, et veau du Cretace Superieur de Bolivie. une histoire detaillee de sa peche an- Bull. Mus. Natl. Hist. Nat., vol. 36, no. cienne et moderne. Extrait du tome 20 2, pp. 291-297. de l'Histoire naturelle des poissons. Smith, C. L., and R. M. Bailey Paris, pp. 1-260. 1961. Evolution of the dorsal fin supports of 1847b. Histoire naturelle des poissons. Vol. 20, percoid fishes. Papers Michigan Acad. Paris, 472 pp. of Sci., Arts, and Letters, vol. XLVI, von Meyer, H. pp. 345-363. 1848. Fossile fische aus dem Tertiiirthon von Stoll, N. R. Unter-Kirchberg an der Iller. Palaeon- 1961. International code of zoological nomen- tographica, vol. 2, pp. 85-113. clature by the XV International Con- Whitehead, P. J. P. gress of Zoology. Internatl. Trust for 1967. The clupeoid fishes described by La- Zool. Nomenclature, pp. 1-176. cepede, Cuvier and Valenciennes. Bull. Taverne, L. Brit. Mus. (Nat. Hist.), Zool., suppl. 2. 1975. A propos du genre fossile Knightia Jor- 1968. A new genus for the South American dan, D. S., 1907, Clupeid& a double ca- clupeid fish, Lile platana Regan. Jour. rene de l'Eocene des Etats-Unis et du Nat. Hist., vol. 2, pp. 477-486. Bresil. Biol. Jahrb. Dodonaea, vol. 43, 1973. Clupeoid fishes of the Guianas. Bull. pp. 246-251. Brit. Mus. (Nat. Hist.), Zool., suppl. 5, 1976. Les Teleosteens fossiles du Cretace pp. 1-227. Moyen de Kipala (Kwango, Zaire). Wongratana, T. Ann. Mus. Roy. African Centr., ser. in- 1980. Systematics of Clupeoid fishes of the 80, Sc. Geol., no. 79, pp. 1-50. Indo-Pacific region. Ph.D. thesis, Univ. 1977. Osteologie de Clupavus maroccanus of London. (Cretace Superieur du Maroc) et con- Woodward, A. S. siderations sur la position systematique 1888. Notes on some vertebrate fossils from et les relations des Clupavidae au sein the province of Bahia, Brazil, collected de l'ordre des Clupeiformes sensu stric- by Joseph Mawson, Esq., F. G. S. Ann. to (Pisces, Teleostei). Geobios, vol. 10, Mag. Nat. Hist., ser. 6, vol. 2, no. 8, fasc. 5, pp. 697-722. pp. 132-136. Taylor, W. R. 1889. Prof. von Zittel on palaeoichthyology. 1967. An enzyme method of clearing and Geol. Mag., vol. 6, no. 3, pp. 227-232. staining small vertebrates. Proc. U.S. 1892. Doubly-armoured herrings. Ann. Mag. Natl. Mus., vol. 122, no. 3596, pp. 1- Nat. Hist., ser. 6, vol. 10, pp. 412-413. 17. 1895. On two deep-bodied species of the clu- 34 AMERICAN MUSEUM NOVITATES NO. 2728

peoid genus Diplomystus. Ann. Mag. suborders Isospondyli (in part), Nat. Hist., ser. 6, vol. 15, pp. 1-3. Ostariophysi, Apodes, Percesoces, 1901. Catalogue of the fossil fishes in the Brit- Hemibranchii, Acanthopterygii, and ish Museum. Vol. 4: Containing the Anacanthini. London, Taylor and Fran- actinopterygian Teleostomi of the cis, pp. 1-636.