Freshwater Stingrays of the Green

Home , Aetobatus, Butterfly ray, Dasyatis, Gymnura japonica, Mobula, Myliobatis

FRESHWATER STINGRAYS OF THE GREEN RIVER FORMATION OF WYOMING (EARLY EOCENE), WITH THE DESCRIPTION OF A NEW GENUS AND SPECIES AND AN ANALYSIS OF ITS PHYLOGENETIC RELATIONSHIPS (CHONDRICHTHYES: MYLIOBATIFORMES)

MARCELO R. DE CARVALHO Research Associate, Department of Ichthyology, Division of Vertebrate Zoology, American Museum of Natural History Departamento de Biologia Universidade de SaÄo Paulo Av. dos Bandeirante 3900, RibeiraÄo Preto SP, Brazil, 14040-901 ([email protected])

JOHN G. MAISEY Division of Paleontology, American Museum of Natural History ([email protected])

LANCE GRANDE Department of Geology, Field Museum of Natural History Roosevelt Road at Lake Shore Drive, Chicago, IL, 60605-2496 (grande@®eldmuseum.org)

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 284, 136 pp., 53 ®gures, 7 tables Issued June 9, 2004

Copyright ᭧ American Museum of Natural History 2004 ISSN 0003-0090 Frontispiece. Artistic rendering of the new genus and species of stingray described in the present paper in its natural environment. This stingray taxon occurred in Fossil Lake, an extinct tropical to subtropical freshwater intermontane lake that formed as a consequence of the orogeny of the Rocky mountains (®g. 1). Sediments from Fossil Lake are exposed as the Fossil Butte Member of the Green River Formation in present- day Wyoming (some 52 million years before present). The adult female is giving birth through viviparity; this scene is inspired by specimen AMNH 11557, in which an adult female stingray is preserved alongside two small stingrays that are inferred to be her aborted late-term fetuses (®g. 3; see text for details). Based on an original painting by David W. Miller and modi®ed from Maisey (1996: pl. 42). CONTENTS Abstract ...... 4 Introduction ...... 5 Materials and Methods ...... 7 Measurements and Terminology ...... 8 Comparative Stingray Material ...... 9 Abbreviations ...... 11 Geological Setting ...... 12 Systematic Paleontology ...... 17 ²Asterotrygon, new genus ...... 17 ²Asterotrygon maloneyi, new species ...... 24 Anatomical Description ...... 31 Neurocranium ...... 36 Jaws, Hyomandibulae and Visceral Arches ...... 44 Synarcual and Vertebral Skeleton ...... 50 Appendicular Skeleton ...... 52 Dermal Skeleton ...... 59 Dermal Denticles ...... 59 Caudal Stings ...... 61 Teeth ...... 63 ²Heliobatis Marsh, 1877 ...... 65 ²Heliobatis radians Marsh, 1877 ...... 72 Stingray Characters of ²Asterotrygon ...... 72 Phylogenetic Relationships of the Green River Stingrays ...... 73 Phylogenetic Procedures and Coding ...... 73 Character Descriptions ...... 77 Lateral-line Canals ...... 77 Skeleton ...... 78 Mandibular Muscle Plate ...... 93 Hypobranchial Muscle Plate ...... 93 Miscellaneous ...... 95 Results of Phylogenetic Analysis ...... 97 Discussion and Comparison to Previous Stingray Phylogenies ...... 102 Systematic Issues within Myliobatiformes ...... 105 Classi®cation of Myliobatiform Genera ...... 108 Biogeographical Implications ...... 108 Origin of the Green River Stingrays ...... 109 Origin of the South American Freshwater Stingrays ...... 112 Conclusions ...... 119 Acknowledgments ...... 122 References ...... 123 Appendix 1. Comparative Batoid (Nonstingray) Material Examined ...... 133 Appendix 2. Experimental Phylogenetic Analysis ...... 134 Index of Stingray Genera and Species ...... 136 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

ABSTRACT

Freshwater stingrays from the Fossil Butte Member of the late early Eocene Green River Formation of Wyoming are reviewed, and a new genus and species of fossil stingray is described. ²Asterotrygon maloneyi, n.gen., n.sp. is remarkably well preserved and is known from articulated skeletons of juveniles and adults, both males and females. It is distinguished from all Recent and fossil stingrays, including ²Heliobatis radians from the same formation, by the unique presence of a dorsal ®n covered with dermal denticles directly anterior to the caudal stings. Other characters that in combination distinguish the new fossil genus from all other stingrays include: retention of separate, individual vertebrae extending to the tail extremity instead of a cartilaginous rod posterior to caudal stings; dorsal surface of disc and tail covered by numerous, closely packed, minute denticles; tail relatively stout at base; and relative proportions of disc and tail. ²Asterotrygon, n.gen. shares with certain stingray genera postorbital processes of neurocranium separated from a supraorbital process by a small notch in the supraorbital shelf, presence of both dorsal and ventral tail-folds posterior to caudal stings (and internally supported by rudimentary radial elements), and hyomandibulae separated from lower jaws by a gap that originally contained the hyomandibular-Meckelian ligament. A calci®ed angular cartilage between the hyomandibula and Meckel's cartilage is tentatively identi®ed in ²Asterotrygon, n.gen. as well. ²Asterotrygon, n.gen. is unquestionably a stingray, presenting many myliobatiform synapomorphies including caudal stings on the dorsal aspect of tail, lack of jugal arches in neurocranium, a thoracolumbar synarcual cartilage posterior to scapulocoracoid, absence of thoracic ribs, and laterally expanded, shel¯ike postorbital processes. ²Aster- otrygon, n.gen. and ²Heliobatis primitively retain a narrow and slightly arched puboischiadic girdle and primitively lack calci®ed rostral elements in adults. A phylogenetic analysis of 23 stingray genera, two outgroups, and 44 informative morphological characters resulted in 35 equally most parsimonious trees. The strict consensus reveals the following hierarchical structure: Hexatrygon ϩ (²Asterotrygon, n.gen., Plesiobatis, Uro- lophidae ϩ (Urotrygonidae ϩ (²Heliobatis ϩ (Potamotrygonidae ϩ (amphi-American Himan- tura, Pteroplatytrygon, Himantura, Taeniura, Dasyatis ϩ (Gymnuridae ϩ Myliobatidae)))))). Our resulting tree has nodes in common with previous phylogenetic analyses of stingrays (e.g., Hexatrygon is the most basal stingray genus; gymnurids and myliobatids [pelagic stingrays] are well-supported sister-groups), but includes novel components, such as a clade that includes all dasyatid genera (as a polytomy) and the component Gymnuridae ϩ Myliobatidae. ``Das- yatidae'' is not monophyletic in any of the minimum-length trees obtained; Urolophidae (Uro- lophus and Trygonoptera) and Urotrygonidae (Urobatis and Urotrygon) are both monophyletic, but are not sister-groups. ²Asterotrygon, n.gen. forms a clade with urolophids in 21 of the 35 equally most parsimonious trees. Successive approximations weighting adds only one additional node in relation to the strict consensus, which unites Pteroplatytrygon, Dasyatis, and Himantura sensu stricto (in a polytomy) with Gymnuridae ϩ Myliobatidae. The resulting stingray phylogeny is at odds with previous phylogenies mostly regarding the af®nities of amphi-American Himantura and Taeniura, which do not form a monophyletic group with the South American freshwater stingrays (Potamotrygonidae) in any of the minimum-length trees obtained. Similar to most elasmobranch groups, stingrays display much character con¯ict, and cladogram topologies are very sensitive to changes in character coding. Due to a high degree of character variation present in certain generic-level terminal taxa, a more fully representative species-level phylogeny is necessary to clarify the systematic importance of tail-fold con®g- uration, ceratobranchial fusion patterns, and other characters discussed in our study. Three additional synapomorphies of stingrays were uncovered by our study, pertaining to the con- ®guration of the basihyal, ®rst pair of hypobranchial cartilages, and to the forward extension of the basibranchial copula. Our phylogenetic results imply the following biogeographic patterns: the relationships of ²Asterotrygon, n.gen. demonstrate a strong Indo-west Paci®c historical correlation, while ²Heliobatis displays an af®nity with the Americas; the node containing the greatest diversity of modern stingrays (``Dasyatidae'' ϩ (Gymnuridae ϩ Myliob- atidae)) evolved only after an American stingray lineage was established sometime earlier than the early Eocene; and potamotrygonids date at least from the late early Eocene, and not the Miocene, as previous studies have implied. The mechanism responsible for the invasion of the potamotrygonid ancestor into South America could indeed have been a marine transgres- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 5

sion as advocated by other authors, albeit a much earlier (pre-Miocene) one, during either the Late Cretaceous or the late Paleocene to early Eocene.

INTRODUCTION occurrences listed above, consist of isolated teeth, dermal denticles, and occasionally ser- Fossilized remains of stingrays (Mylioba- rated caudal spines (caudal ``stings''). tiformes) are not uncommon components of In contrast, more complete, articulated fos- Tertiary strata and are known from many sil stingray specimens are relatively rare in widespread localities, both freshwater and the fossil record and are presently known marine. These occurrences, summarized in from only two localities, both of early Eo- table 1, include the Paleocene of Africa, cene age: the Monte Bolca Formation of North America, and Europe (Arambourg, northeastern Italy and the Green River For- 1952; Estes, 1976; Halter, 1989); Eocene of mation of Wyoming. Both localities are al- Europe, Asia, and Africa (Casier, 1966; most contemporaneous but represent very Chang and Zhou, 1993; Cappetta, 1984); Ol- different paleoenvironments. The Monte Bol- igocene of Japan and Europe (Yabumoto and ca Formation was deposited in a shallow, Uyeno, 1994; Bor, 1990); Miocene of South tropical, marine coral back-reef lagoon (Lan- America, Cuba, Europe, and Asia (e.g., Hatai dini and Sorbini, 1996), while the Green Riv- and Kotaka, 1962; Sahni and Mehrotra, er Formation was deposited within a series 1981; Arratia and Cione, 1996; Schultz, of tropical to subtropical freshwater lakes 1998); and Pliocene of Europe, Japan, and (Schaeffer and Mangus, 1965; Grande, 1984, North America (Landini, 1977; Yabumoto 2001). The Monte Bolca deposits contain a and Uyeno, 1994; Purdy et al., 2001). Even more diverse stingray (and batoid) fauna, and more recent (Quaternary) stingray remains both formations have yielded many stingray have been found in Japan (reviewed in Ya- specimens represented by somewhat com- bumoto and Uyeno, 1994). The oldest sting- plete articulated skeletons. The batoid fauna ray fossils that have been reported to date, of Monte Bolca is presently under review; it however, are from the Early Cretaceous contains guitar®shes (Rhinobatidae), thorn- (Hauterivian) of northeastern England (Un- back rays (Platyrhinidae), and electric rays derwood et al., 1999). Other Mesozoic sting- (Torpediniformes) in addition to stingrays, ray records are from the early Late Creta- similar to many modern coral reef faunas. ceous (Cenomanian) of Texas (Meyer, 1974; The Green River Formation has yielded no Cappetta and Case, 1999) and Egypt (Strom- batoids other than stingrays, but to some ex- er, 1927; Werner, 1989), and upper Late Cre- tent this is to be expected, given that South taceous of Wyoming (Estes, 1964), Texas American potamotrygonid stingrays, and cer- (Maastrichtian: Welton and Farish, 1993; tain species of the stingray genera Dasyatis Case and Cappetta, 1997; see also Cappetta and Himantura, are the only obligate modern and Case, 1999), New Jersey (Cappetta and freshwater batoids. Case, 1975), South America (Bolivia: Stingrays were ®rst reported from the Schaeffer, 1963; Cappetta, 1975, 1992; Gay- Green River Formation by Marsh (1877), et et al., 1992; Arratia and Cione, 1996; who described ²Heliobatis radians on the Chile: Wetzel, 1930), Africa (Arambourg, basis of a single specimen from the Fossil 1952; Dartevelle and Casier, 1943, 1949, Butte Member of Fossil Lake. Marsh's de- 1959; Cappetta, 1987; Noubhani and Cap- scription is rather brief and his specimen was petta, 1997), Jordan (Zalmout and Mustafa, not illustrated. Cope may have overlooked 2001), and Europe (e.g., Albers and Weiler, Marsh's description, as shortly thereafter he 1964; Cappetta, 1987; Soler-GijoÂn and LoÂ- erected ²Xiphotrygon acutidens also from the pez-MartõÂnez, 1998), but Mesozoic stingray Fossil Butte Member of Fossil Lake (Cope, remains are not as abundant as those from 1879). Cope later provided a more complete the Tertiary (for a stratigraphical review, see description of his species accompanied by a Cappetta et al., 1993). The vast majority of remarkable illustration of a specimen which stingray fossils, however, including all of the is now lost (Cope, 1884). Since then many 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

TABLE 1 fossil stingray specimens have appeared from Some Tertiary Records of Stingrays Fossil Lake, compelling Fowler (1947) to de- (Myliobatiformes) scribe ²Palaeodasybatis discus to accom- All occurrences are of isolated teeth, dermal denticles or caudal stings. This compilation is not exhaustive, but modate what was thought to be a more di- additional records (for India and Africa in particular) are verse extinct stingray fauna. Fowler (1947) listed in the reference below (see also Cappetta, 1987). distinguished his new taxon from Cope's nominal species on the basis of minor differences in disc shape, apparently also unaware of Marsh's (1877) earlier account. Both nominal taxa of Cope (1879) and Fowl- er (1947) have been synonymized with ²He- liobatis radians (Grande, 1980, 1984), a de- cision we further corroborate in this study. The next indication of a more diverse Green River stingray fauna was provided by Grande (1980, 1984), who documented and illustrated an adult female specimen signi®- cantly distinct from typical ²Heliobatis radians. The limestone matrix on which this specimen is preserved (AMNH P 11557) is remarkable because it also contains two small specimens which we infer to be aborted late-term fetuses. Another large female specimen, which is also markedly different from ²Heliobatis radians, is just as notewor- thy due to the presence of a small embryo preserved inside its pleuroperitoneal cavity. This adult female, recently obtained by the Field Museum of Natural History (Chicago), is designated the holotype of a new genus and species of stingray from Fossil Lake, which is described below. A total of 15 specimens of this new form have now been collected and identi®ed, and form the basis of the present paper. Stingrays (order Myliobatiformes) collec- tively form a monophyletic group within batoids, sharing numerous synapomorphies including the presence of a caudal sting (serrated caudal spine), lack of thoracic ribs and presence of a second (thoracolumbar) synarcual cartilage, among other features. Re- cent stingrays presently include about 185 species in 24 genera, mostly inhabiting tropical to subtropical shallow marine areas. Phy- logenetic relationships among component stingray taxa have been the subject of a re- invigorated recent debate which has led to the understanding that some of the more common living stingray genera may not be monophyletic, most notably Dasyatis and Hi- mantura (Rosa, 1985; Miyake, 1988; Nishi- da, 1990; Lovejoy, 1996; McEachran et al. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 7

Fig. 1. A. Map of southwestern Wyoming, northeastern Utah, and northwestern Colorado showing the geographical extent of the intermontane lakes during their middle or late early Eocene phase. B. Fossil Lake during the deposition of the Fossil Butte Member of the Green River Formation, during the late early Eocene, showing a few prominent F-1 and F-2 localities where fossil stingrays have been found. From Grande and Buchheim (1994).

1996; see also Compagno and Roberts, 1982, is known from 15 specimens (in addition to 1984). In addition, these studies have corrob- two tentatively identi®ed specimens), some orated that at least two monophyletic sub- of which are well preserved and fairly com- groups exist within stingrays: the Neotropical plete (listed below is the systematic section), freshwater stingrays (Potamotrygonidae) and including juveniles and adults and specimens a clade containing pelagic stingray genera of both sexes. The only other described (Myliobatidae) plus the butter¯y rays (Gym- stingray from this formation (²Heliobatis ra- nuridae). The remaining stingray genera, dians Marsh, 1877) is known from hundreds which are primarily benthic in habitus, re- of specimens in public and private collec- main poorly resolved phylogenetically. Mor- tions, of which we were able to examine phological features of systematic signi®cance many (some of these are listed below in the are preserved in the new fossil stingray taxon systematic section). Both Green River sting- from the Green River Formation, and we ray taxa are dorsoventrally ¯attened and are have therefore included it in a matrix coding preserved in laminated limestone slabs of morphological characters of representative varying thickness and lithology. Specimens stingray genera. The evolutionary study pre- are readily prepared with the aid of a pine- sented here is not intended to resolve all per- vise and other matrix-removing tools with sisting problems in myliobatiform systemat- sharp tips and an abrasive machine. All prep- ics, although it is the most comprehensive aration of fossils was done with the aid of a morphological phylogenetic analysis of dissecting microscope. stingrays to date. Our systematic survey was Fossil stingrays, and fossil batoids in gen- conducted to reveal phylogenetic implica- eral, are usually dorsoventrally exposed and tions of salient anatomical features of the two-dimensional, a constraint imposed by Green River stingrays, allowing for their their enlarged pectoral ®ns that are continu- placement in a more inclusive framework of ous with the head (forming the disc). This stingray systematics and historical biogeog- renders working with fossil stingrays similar raphy. to working with radiographs of extant specimens, and many anatomical details (those MATERIALS AND METHODS observed in lateral view) are consequently The new genus and species of stingray re- obscured. This contrasts with three-dimen- ported here from the Green River Formation tional shark fossils embedded in matrices 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 amenable to acid or mechanical preparation VPS ϩ VST). The number of toothrows (Schaeffer, 1981; Maisey and Carvalho, could not be inferred with accuracy, as teeth 1997). It is therefore dif®cult to achieve the are not preserved in discrete rows but are same level of anatomical comprehension scattered in the vicinity of the jaws and neu- with fossil batoids (the only fully three-di- rocranium. Many measurements traditionally mensional fossil batoid is ²Iansan, a rhino- taken on extant specimens could not be made batoid from the Late Cretaceous of Brazil; on the fossils due to preservational imper- Maisey, 1991; Brito and SeÂret, 1996). fections (e.g., measurements involving the cloaca). The measurements above were taken MEASUREMENTS AND TERMINOLOGY to provide a general notion of size and proportions of the fossils, but because their ac- Measurements conducted on specimens curacy depends on the preservation of the were taken in a straight line, point-to-point, material, which varies signi®cantly, they modi®ed from Bigelow and Schroeder should not be accepted as strictly as if taken (1953), Rosa (1985), and Miyake (1988), and from extant specimens; the counts provided are as follows: TL, total length (snout tip to should be interpreted in a similar fashion. posterior tip of tail); DL, disc length (from Measurements and counts are summarized in anterior disc margin at snout region to pos- tables 2 and 3 for the new stingray taxon and terior disc margin at disc axil); DW, disc table 4 for ²Heliobatis. width (distance between lateral disc margins Anatomical terminology is according to at greatest width, i.e., at level of scapulocor- Daniel (1934), Miyake (1988), and Miyake acoid; width may be underestimated in some and McEachran (1991) unless otherwise not- specimens because of broken distal tips of ed. We follow Rosa (1985) in using ``sting'' pectoral radials); TAL, tail length (from pu- (as opposed to ``spine'') to refer to the elon- boischiadic bar to posterior tip of tail); TAW, gated and serrated dermal derivative on the tail width (greatest width of tail as exposed, dorsal surface of the tail (usually more than just posterior to pelvic ®ns); POL, preoral one sting is present, and sometimes more snout length (snout length from mouth open- than two). ``Sting'' and ``serrated caudal ing to anterior margin of disc); MSC, dis- sting'' are used interchangeably. The term tance between mouth and scapulocoracoid ``stingray'' is used here as equivalent to (from mouth opening to anterior margin of ``myliobatiform rays''. There are many dif- scapulocoracoid); SCW, scapular width ferent groups of myliobatiform rays, and (greatest width of scapulocoracoid, not in- most of these have a corresponding trivial cluding pectoral basal cartilages); PGW, pel- name (e.g., eagle, cownose, whiptail, manta vic girdle width (greatest width of pelvic gir- rays). ``Stingray'' is an all-encompassing dle); STL, sting length (from sting base to term used for the entire order (Myliobatifor- posterior sting tip); PGT, distance between mes) and does not include here platyrhinids pelvics and tip of tail; CL, clasper length (thornback rays) and Zanobatus, considered (from posterior margin of pelvic girdle and to be successive sister-groups to stingrays by posterior tip of clasper); NCL, neurocranial McEachran et al. (1996). In the descriptive length (from anterior contour of neurocrani- sections below the composition of the differ- um at nasal capsule con¯uence to occipital ent myliobatiform families is based on Com- region); NCW, neurocranial width (greatest pagno (1973, 1977, 1999) and on the phy- width at level of nasal capsules). The follow- logenetic analysis presented after the descrip- ing meristic counts were taken: PRO, prop- tive accounts. The terms ``benthic sting- terygial radials; MES, mesopterygial radials; rays'', ``non-myliobatid stingrays'', or MET, metapterygial radials; TPR, total pec- ``nonpelagic stingrays'' apply to Recent and/ toral radials (ϭ PRO ϩ MES ϩ MET); PVR, or fossil stingrays that have a mostly benthic pelvic radials; VSP, vertebral centra from niche, excluding the morphologically spe- scapulacoracoid to pelvic girdle; VPS, ver- cialized pelagic stingrays (which are here tebral centra from pelvic girdle to caudal lumped into a single family, Myliobatidae; sting; VST, vertebral centra posterior to the exception is the monotypic Pteroplaty- sting; TV, total vertebral centra (ϭ VSP ϩ trygon, which is a ``dasyatid'' but is also pe- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 9 lagic). Benthic stingrays include Hexatrygon, CSIRO); MNHN uncataloged (2 specimens, gymnurids, urolophids, Plesiobatis, urotry- from Fiji and New Caledonia). gonids, potamotrygonids, almost all dasyatids, and the fossil taxa from the Green Riv- Urolophidae er and Monte Bolca Formations (except Urolophus aurantiacus: AMNH 26690 (D, XR), ²Promyliobatis, a myliobatid from Monte FUMT P 10502 (XR; on ®le at CSIRO). Bolca). Consequently, we do not refer to a Urolophus cruciatus: AMNH 59890 (C&S), monophyletic group when making reference AMNH 98247 (C&S), CSIRO H 20 (XR), CSI- to ``benthic stingrays'', because they are not RO H 21 (XR), CSIRO CA 152 (XR), CSIRO monophyletic without myliobatids (see our CA 162 (XR). phylogenetic analysis below); these terms are Urolophus expansus: CSIRO T 293 (XR), CSIRO H 327 (XR), CSIRO H 343 (XR). used only for convenience. Synonymies are Urolophus ¯avomosaicus: CSIRO H 718±19 in abbreviated form, with the full citation in- (XR), CSIRO H 718.22 (XR). cluded only in the references. The synonymy Urolophus gigas: CSIRO H 50 (XR). for ²Heliobatis radians is not exhaustive. Urolophus lobatus: CSIRO P 8197 (XR), CSIRO The procedures employed in the phyloge- 28346±002 (XR). netic analyses are described in Phylogenetic Urolophus mitosis: CSIRO CA 2874 (XR), CSI- Procedures and Coding, after the morpholog- RO CA 2875 (XR), CSIRO CA 2877 (XR). ical characterization of ²Heliobatis. Urolophus orarius: AMS I 20194±043 (XR), SAM uncataloged (XR; on ®le at CSIRO). Urolophus paucimaculatus: CSIRO H 9 (XR), COMPARATIVE STINGRAY MATERIAL CSIRO C 4757 (XR), AMNH 99735 (SW), Recent stingray materials examined for AMNH 95342 (SW), AMNH 216702 (SW), AMNH 217079 (SW), AMNH 217080 (SW), this study are listed below, and include AMNH 217081 (SW), AMNH 217082 (SW) cleared-and-double-stained specimens AMNH 217083 (SW), AMNH 217084 (SW), (C&S), dry skeletons (SD), skeletons pre- AMNH 217085 (SW). served in alcohol (SW), X-ray radiographs Urolophus viridis: CSIRO CA 527 (XR), CSIRO (XR), and dissected specimens (D). Dry and H 779±01 (XR), CSIRO A 3355 (XR), USNM ``wet'' skeletons were prepared by subjecting 222681 (XR). dissected specimens to dermestid beetles Urolophus sp.: USNM 131105 (XR). (prepared by the staff of the Department of Urolophus sp.: AMNH 214469 (SW). Ichthyology, AMNH). Cleared-and-stained Trygonoptera mucosa: CSIRO H 898±02 (XR), specimens were prepared according to Din- CSIRO H 898±03 (XR), CSIRO H 898±06 (XR), CSIRO H 3499 (XR), CSIRO CA 3521 gerkus and Uhler (1977). Additional speci- (XR), CSIRO P 27217±001 (XR), CSIRO P mens of many stingray taxa were examined 27737±001 (XR). for external morphology, but only a few of Trygonoptera ovalis: CSIRO A 2817 (XR), CSI- these are included in the list below (belong- RO P 27958±001 (XR). ing to species in which available specimens Trygonoptera personata: CSIRO H 894±01 (XR), are more rare). Anatomical information ex- CSIRO H 901±01 (XR). tracted from the literature was con®rmed by Trygonoptera testacea: CSIRO H 33 (XR), CSI- examining specimens whenever possible. RO H 837±02 (XR), CSIRO H 837±06 (XR), Stingray classi®cation in this list follows CSIRO H 838±05 (XR), CSIRO H 874±07 Compagno (1999). (XR), CSIRO H 929±01 (XR), USNM 39993 (XR). Trygonoptera sp.: CSIRO H 43 (XR), CSIRO P Hexatrygonidae 14839 (XR), NMV A 1827 (XR), WAM P 14151 (XR). Hexatrygon bickelli: BPBM 27761 (XR), CSIRO H 2543±07 (XR), CSIRO H 2226±01 (XR); MNHN uncataloged (2 specimens from New Urotrygonidae Caledonia). Urobatis concentricus: AMNH 15692 (XR). Urobatis halleri: AMNH 15692 (XR), 44102 Plesiobatidae (XR), FMNH 42601 (C&S). Urobatis jamaicensis: AMNH 18195 (XR), Plesiobatis daviesi: RUSI 11157 (XR; on ®le at AMNH 25049 (XR), AMNH 28649 (D), 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

AMNH 44102 (D), AMNH 30385 (C&S), Pastinachus sephen: AMNH 44057 (XR), USNM USNM 144075 (XR). 39982, USNM 147420. Urobatis maculatus: AMNH 44144 (XR). Pteroplatytrygon violacea: MZUSP 49061 (D). Urobatis tumbesensis: AMNH 44021 (XR). Taeniura grabata: MNHN 1989±1793 (D). Urotrygon aspidura: CAS 48271 (XR). Taeniura lymma: AMNH 44076 (XR), AMNH Urotrygon chilensis: AMNH 44014 (XR), CAS 44076 (D), AMNH 44079 (C&S). 48271 (XR), FMNH 62371 (C&S), FMNH 93737 (C&S). Potamotrygonidae Urotrygon microphthalmum: NHM 1985.6.21.4± 7 (D, XR). Paratrygon aiereba: MZUSP 14772 (XR), Urotrygon nana: FMNH 72281 (C&S). AMNH uncataloged (D). Urotrygon venezuelae: AMNH 55623 (C&S). Plesiotrygon iwamae: MZUSP 42848 (D, XR), MNHN uncataloged (XR). Potamotrygon brachyura: MZUSP 14819 (XR). Dasyatidae Potamotrygon falkneri: UERJ 718.1 (D). Dasyatis akajei: AMNH 44065 (XR). Potamotrygon henlei: MZUSP 14768 (XR). Dasyatis americana: AMNH 30607 (C&S), Potamotrygon leopoldi: UERJ 719 (D, XR), AMNH P uncataloged (SD). MZUSP 35986 (XR). Dasyatis annotata: CSIRO T 449 (holotype; XR), Potamotrygon magdalenae: AMNH 55620 CSIRO T 694 (XR), CSIRO T 696 (XR), CSI- (C&S), MNHN 2368 (holotype; XR). RO CA 697 (paratype; XR). Potamotrygon motoro: AMNH 44034 (SW, C&S), Dasyatis geijskesi: NNM 20487 (holotype), MZUSP 19190 (D), FMNH 94503 (C&S). USNM 158726. Potamotrygon cf. motoro: AMNH 38138 (C&S, 4 Dasyatis kuhlii: AMNH 44080 (XR), CSIRO CA specimens), AMNH 44032 (XR), AMNH 4309 (XR), CSIRO CA 1241 (XR). 44034 (SW). Dasyatis leylandi: CSIRO CA 2806 (holotype; Potamotrygon cf. ocellata: MNRJ 10620 (XR). XR). Potamotrygon orbignyi: AMNH uncataloged (D, Dasyatis margarita: AMNH 41512 (XR), CU XR), MZUSP 14794 (XR). 53996 (XR), IRSNB 8497 (XR; on ®le at CSI- Potamotrygon signata: MCZ 600 (syntype; XR). RO). Potamotrygon sp. (Rio Negro): MNRJ 3532 (D). Dasyatis pastinaca: AMNH 1511 (XR), AMNH Potamotrygon sp. (Rio Taquari): MZUSP 25663 32796 (XR). (XR). Dasyatis sabina: AMNH 16356 (XR), AMNH Potamotrygon sp. nov. (Rio TapajoÂs): MZUSP 51483 (D), AMNH 211610 (SW). 25489 (XR), MZUSP 25580 (XR). Dasyatis thetidis: CSIRO CA 4125 (syntype; XR), Potamotrygon sp. nov. (RõÂo Corantijn): USNM CSIRO H 1036±15 (syntype; XR). 225574 (XR). Dasyatis ukpam: MNHN 1979±244 (XR). Dasyatis zugei: AMNH 44056 (XR), OSU 1512 Gymnuridae (XR, 4 specimens). Gymnura australis: CSIRO H 37 (XR), CSIRO H Dasyatis sp.: AMNH 41515 (XR). 323 (XR), CSIRO C 3584 (XR). Himantura chaophraya: QM I 11928 (XR). Gymnura japonica: AMNH 26691 (XR). Himantura gerrardi: NHM 1868.08.26.01.1 (XR). Gymnura marmorata: AMNH 18599 (XR). Himantura granulata: SMF 4747 (XR; on ®le at Gymnura micrura: FMNH 89990 (C&S). CSIRO). Himantura imbricata: AMNH 32501 (XR), CAS 41680 (XR), CSIRO I 1449 (XR), MNHN 2269 Myliobatidae (XR), MNHN 1985±211 (D, XR). Aetomylaeus maculatus: AMNH 32500 (XR). Himantura kremp®: AMNH 41567 (XR). Aetobatus narinari: AMNH 44142 (XR, dorsal ®n ``Himantura'' paci®ca: AMNH 15661, 15662, and caudal stings only), AMNH 53029 (C&S, 15663, 15710 (SD). gill arches only), AMNH 222833 (SW). ``Himantura'' schmardae: NHM 1908.5.28.2±3 Mobula kuhlii: AMNH 15319 (C&S). (D, XR), USNM 33719, USNM 86071, USNM Mobula sp.: AMNH P 123 (SD). 128388. Myliobatis californica: AMNH uncataloged (XR). Himantura toshi: CSIRO H 312 (XR), CSIRO H Myliobatis freminvillii: AMNH 15333 (XR). 959±01 (XR), CSIRO H 964±01 (XR), QM I Rhinoptera bonasus: AMNH 3728 (XR). 22355 (XR). Himantura uarnak: CSIRO CA 2405 (XR). Fossil stingrays examined for this study, Himantura walga: OSU 1506 (XR). in addition to specimens from the Green Riv- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 11 er Formation, are listed below. This material FMNH PF Department of Geology (Fossil Fish is from the Eocene Monte Bolca Formation Collection), Field Museum of Nat- of northeastern Italy; the entire Monte Bolca ural History (Chicago) batoid fauna is currently under review (Car- FUMT University Museum, University of valho, in prep.). Familial, generic, and some Tokyo (Tokyo) IRSNB Institut Royal des Sciences Natura- speci®c assignments below are regarded as les de Belgique (Brussels) provisional. MCSNV Museo Civico di Storia Naturale di Verona (Verona) Dasyatidae MNHN MuseÂum National d'Histoire Natu- ²``Dasyatis'' dezignoi: MCSNV VII B 86 and 87 relle (Paris) (part and counterpart); UP 150 Z and 151 Z MNRJ Museu Nacional (Rio de Janeiro) (part and counterpart). MZUSP Museu de Zoologia da Universidade ²``Dasyatis'' muricata: MCSNV VII B 92 and 93 de SaÄo Paulo (SaÄo Paulo) (part and counterpart); MCSNV T 1020 and NHM The Natural History Museum (Lon- 1021 (part and counterpart); MCSNV IG don) (formerly BMNH) 186653 and 186654 (part and counterpart); UP NMV National Museum of Victoria (Mel- 159 Z and 160 Z (part and counterpart). bourne) ²``Dasyatis'' sp.: MCSNV IG 23193 and 23194 NNM Nationaal Natuurhistorisch Museum (part and counterpart); MCSNV IG 129652; (Leiden) (formerly RMNH) MCSNV IG 129653. OSU Oregon State University (Corvallis) QM Queensland Museum (Brisbane) Urolophidae RUSI J.L.B. Smith Institute of Ichthyolo- gy, Rhodes University (Grahams- ²``Urolophus'' crassicaudatus: MCSNV VII B 82 town) (presently South African In- and 83 (part and counterpart); MCSNV VII B stitute of Aquatic Biodiversity, 84 and 85 (part and counterpart); MCSNV T SAIAB) 317 and 318 (part and counterpart); MCSNV SAM South African Museum (Cape VR 26607 and 26608 (part and counterpart). Town) ²``Urolophus'' sp.: MCSNV IG 174554; UP 8875 SMF Senckenberg Museum (Frankfurt) and 8876 (part and counterpart); UP 26227. SMMP Science Museum of Minnesota (St. Paul) Myliobatidae TCWC Texas A&M University, Department ²Promyliobatis gazolae: MCSNV VII B 90 and of Wildlife and Fisheries (College 91 (part and counterpart). Station) UERJ Universidade do Estado do Rio de ABBREVIATIONS Janeiro, Departamento de Biologia (Rio de Janeiro) Institutional USNM National Museum of Natural Histo- AMNH Department of Ichthyology (Divi- ry, Smithsonian Institution (Wash- sion of Vertebrate Zoology), Amer- ington, DC) ican Museum of Natural History UP UniversitaÂ di Padova (Padova) AMNH P Fossil Fish Collection (Division of UW Geological Museum, University of Paleontology), American Museum Wyoming (Laramie) of Natural History WAM Western Australian Museum (Perth) AMS Australian Museum (Sydney) YPM Peabody Museum of Natural Histo- ANSP Academy of Natural Sciences (Phil- ry, Yale University (New Haven) adelphia) ZMB Institut fuÈr Systematische Zoologie CAS California Academy of Sciences der Humboldt UniversitaÈt (Berlin) (San Francisco) ZMH Zoologisches Institut und Museum, CSIRO Commonwealth Scienti®c and Re- UniversitaÈt Hamburg (Hamburg) search Organization (Hobart) CU Cornell University (Ithaca) Anatomical DMNH Denver Museum of Natural History (Denver) aac anterior angular cartilage FMNH Department of Zoology (Division of ac angular cartilage Fishes), Field Museum of Natural avf anteroventral foramen History (Chicago) aoc antorbital cartilage 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 ax axial cartilage oc olfactory canal b beta cartilage of obturator foramen bb basibranchial copula (medial plate) on optic (II) nerve bh basihyal onc orbitonasal canal (posterior foramen) bp basipterygium pac posterior angular cartilage bpl basal (trabecular) plate pec r pectoral radial cal irregular calci®cation within hyoman- pel r pelvic radial dibular-Meckelian ligament paf parietal fossa cb1 ceratobranchial 1 pb1 pharyngobranchial 1 cb2 ceratobranchial 2 pb5 pharyngobranchial 5 cb5 ceratobranchial 5 pcf precerebral fontanelle (also pc or pf) cl clasper pib puboischiadic bar cr cartilaginous rod poc preorbital canal cst caudal sting pop postorbital process ct syn cervicothoracic synarcual pp (median) prepelvic process da denticle apex pq palatoquadrate db denticle base pro propterygium df dorsal ®n prp preorbital process dm dorsal marginal cartilage psc prespiracular cartilage dph dorsal pseudohyoid bar re rostral extremity dr dorsal radial element ro rostral cartilage (rostrum) dtf dorsal tail-fold sc scapulocoracoid e eye scp scapular process eb5 epibranchial 5 se sensory canal (junction of prenasal and efr enlarged ®rst radial element of pelvic subrostral canals) ®n sp supraorbital process epb epiphysial bar spo spiracular opening ®ca foramen for internal carotid artery ssc suprascapula fpf frontoparietal fontanelle t teeth ga gill arches (with asscociated gill rays) tc terminal cartilages of clasper gr gill rays tl syn thoracolumbar synarcual cartilage ha hemal arch vc vertebral centrum hb1 hypobranchial 1 vm ventral marginal cartilage hyo hyomandibula vph ventral pseudohyoid bar ilp iliac process vr ventral radial element is intermediate segment vt ventral terminal cartilage (ventral isp ischial process covering piece of other authors) lpp lateral prepelvic process vtf ventral tail-fold ls lateral stay of cervicothoracic synarcual GEOLOGICAL SETTING Mc Meckel's cartilage mdc medial crest of cervicothoracic synar- The Green River Formation represents an cual extinct great lake system (comprising Fossil mes mesopterygium Lake, Lake Uinta, and Lake Gosiute), that met metapterygium was formed as a result of the orogeny of the mt median (expanded) tooth and toothrow n neurocranium Rocky Mountains in what is now southwest- na nasal aperture ern Wyoming, northeastern Utah, and north- nc nasal capsule western Colorado (®g. 1). These intermon- ncr nasal cartilage(s) tane lakes were of extremely long duration, ns neural spine especially if compared to modern lake sys-

→ Fig. 2. Holotype of ²Asterotrygon maloneyi, n. gen., n.sp. (FMNH PF 15166), an exceptionally well-preserved adult female in ventral view (ca. 625 mm TL), with an unborn, late-term fetus visible in the pleuroperitoneal region (magni®ed in ®g. 13), indicated by arrowhead. Specimen is from fresh 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 13

water F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. The teleosts preserved at the top of the slab are of the clupeid ²Knightia eocaena. Anterior to top. 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 tems. For example, Lake Uinta is thought to Green River Formation may have been have lasted for approximately 15 million somewhat saline (review and references in years, ranging from late Paleocene to middle Grande, 1994). We follow Grande (1984, or late Eocene times, which is older than the 1989, 1994, 2001) and Grande and Buch- present-day east African great lakes (consid- heim (1994) in viewing the overwhelming ered to be among the oldest extant lakes amount of paleontological data as indicative known). Of the east African great lakes, Lake of a freshwater system, at least during de- Tanganyika is thought to have formed be- position of the main fossil-containing strata tween 9 and 12 million years ago, and Lake of the Fossil Butte Member. The climate of Malawi may be younger than 2 million years the lakes is estimated to have been from sub- old (Delvaux, 1995). Lake Victoria, the larg- tropical to tropical, perhaps similar to the cli- est of the African great lakes, may have been mate of the present-day Gulf Coast and dry as recently as 15,000 years ago (Johnson southern Atlantic regions of the United et al., 1996), but both lakes Malawi and Vic- States (Bradley, 1948; Grande, 1984). toria contain a much greater ichthyological McGrew and Casilliano (1975) provided species diversity compared to the older Lake thorough accounts of the geological history Tanganyika (Turner, 1999; Turner et al., and depositional environment of Fossil 2001; see also references in Grande, 1994; Butte, and McGrew (1975) discussed tapho- Grande and Buchheim, 1994). nomic implications of its fossil ®shes. Throughout their history, the extinct Green Both fossil stingray taxa have been ex- River lakes ¯uctuated in size as a result of tracted from two ``groups'' of localities with- both climactic and tectonic changes (Sulli- in the Fossil Butte Member of Fossil Lake, van, 1980; Grande, 1985, 1989, 2001, pre- both dated as late early Eocene. These sents rough outlines of the lakes throughout ``groups'' of localities are designated ``F-1'' their history). Furthermore, each of the three and ``F-2'' here, following the conventions lakes differed according to their duration, of Grande (1994) and Grande and Buchheim size, and sedimentology. The shortest-lived (1994). Both groups of localities were de- and smallest of the three lakes was Fossil posited almost contemporaneously, but differ Lake, which lasted for less than 5 million slightly with respect to their sedimentology years, all within the North American Was- and paleoenvironment, and also according to atchian stage. Within the Green River For- the fossil organisms they contain. mation, the Fossil Butte Member of Fossil The F-1 localities (also known as the ``18- Lake (exposed near Kemmerer, Wyoming), inch layer'') refer to a 30±40-cm-thick layer, of late early Eocene age (approximately 52 representing probably only a few hundred million years before present), has yielded the years of deposition. The lithofacies contain greatest diversity of fossils to date (Grande, ®nely laminated limestones representing the 1984, 2001), including all of the stingrays upper part of the Fossil Butte Member, and known. they are comprised of light-colored, kerogen- There is a vast amount of evidence indi- rich shales (Grande, 1989). The F-1 localities cating that Fossil Lake comprised a fresh- represent midlake deposits (®g. 1B) with water system for the period in which the deeper waters than F-2, and some 10 locali- highly fossiliferous layer of the Fossil Butte ties have been mined to date, all contempo- Member was deposited. This evidence de- raneous with each other (Grande and Buch- rives mainly from fossil organisms that have heim, 1994). The F-1 localities are thought been collected from localities within Fossil to represent a more anoxic paleoenvironment Lake: freshwater plants (²Ceratophyllum, with a slower rate of deposition, compared lily-pads, palms, cattails), insects and insect to the F-2 localities. larvae, freshwater molluscs and crustaceans The F-2 localities (sometimes referred to (e.g., cray®shes), freshwater-inhabiting tet- as ``split-®sh'') are also contemporaneous rapods, as well as an extensive freshwater with each other and represent near-shore pa- ®sh fauna (Grande, 1994). Recent work by leoenvironments located primarily on the sedimentologists, however, suggests that for north- and southeastern shores of Fossil Lake a good part of their history the lakes of the at the time of deposition (®g. 1B). The F-2 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 15

Fig. 3. Paratype of ²Asterotrygon maloneyi, n. gen., n.sp. (AMNH P 11557, ca. 378 mm TL adult female) in ventral view. Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Arrowheads indicate position of two small stingray specimens on limestone slab (these are inferred to be aborted late-term fetuses of the larger female; see ``Remarks'' under species account for further comment). Anterior to top. This specimen was generously donated by the late Thomas Maloney. 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 4. Paratype of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 12989, ca. 238 mm TL adult male) in ventral view. Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Note elongated and developed claspers overlapping base of tail 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 17 horizon is almost 4 m thick, and is the result over tail and part of dorsal disc surface); in- of greater sedimentation rates compared to dividual vertebral centra extending to distal the F-1 horizon. These deposits are inferred tip of tail, posterior to caudal stings (instead to represent a few hundred to a few thousand of an unsegmented cartilaginous or noto- years of deposition (Buchheim, 1994). The chordal rod extending to distal tip of tail, F-2 localities are less abundant but are also posterior to caudal stings); disc and tail heavily mined. It is thought that the F-2 lith- length almost equal; tail stout at base, taper- ofacies represented a more oxygen-rich hab- ing distally but not continuing caudally as a itat, with a higher dilution of organic mate- slender ``whip''; disc circular to oval in out- rials and higher bioturbation rates. The F-2 line (disc length and width almost equal, ex- deposits are of a slightly older age than those cept in FMNH 14567). of F-1. REMARKS: Fossil stingrays that are mor- Grande and Buchheim (1994) summarized phologically similar to ²Asterotrygon, n.gen. much data concerning the paleoenvironments and known from more-or-less complete skel- of both F-1 and F-2 localities. They present- etons include the monotypic ²Heliobatis ed stratigraphic sections and maps indicating Marsh, 1877, occurring in the same localities precise locations of mines and compared and horizon as ²Asterotrygon, n.gen., and both F-1 and F-2 horizons in detail from geo- ²``Dasyatis'' muricata and ²``Dasyatis'' de- logical and paleontological perspectives. zignoi, both from the Eocene (Lutetian) of Monte Bolca, Italy (Jaekel, 1894). The new SYSTEMATIC PALEONTOLOGY genus differs from all three taxa by the pres- CLASS CHONDRICHTHYES HUXLEY, 1880 ence of a conspicuous dorsal ®n (located just SUBCLASS ELASMOBRANCHII BONAPARTE, anterior to caudal stings, absent in all three 1832 taxa), a more rounded and smaller disc (disc trapezoidal or rhomboidal in the Monte Bol- DIVISION SQUALEA sensu SHIRAI, 1992 ca taxa and many specimens of ²Heliobatis), SUPERORDER HYPNOSQUALEA CARVALHO a shorter and much stouter tail at base (in all AND MAISEY, 1996 three taxa tail is slender at base, not tapering SERIES BATOIDEA sensu COMPAGNO, 1973 greatly, and is whiplike in ²``Dasyatis'' mur- ORDER MYLIOBATIFORMES sensu COMPAGNO, icata), and the presence of heavy denticula- 1973 tion over dorsal surface of disc, dorsal ®n, SUBORDER MYLIOBATOIDEI insertae sedis snout, and over almost entire tail region (sparse denticles, if present, occur as en- ²Asterotrygon, new genus larged spines in generally a single row over DIAGNOSIS: The presence of a dorsal ®n middisc and tail regions in all three taxa). covered by dermal denticles, just anterior to ²Asterotrygon, n.gen. further differs from the caudal stings, is autapomorphic for ²As- ²``Dasyatis'' muricata in having a much terotrygon, n.gen. The following unique shorter tail (less than or equal to disc length combination of characters further distin- in ²Asterotrygon, n.gen., much greater than guishes ²Asterotrygon, n.gen. from both fos- disc length in ²``D.'' muricata; Jaekel, 1894: sil and Recent stingray genera: dorsal surface 143, ®g. 32). of disc, snout, and tail, as well as base and ²Asterotrygon, n.gen. can be distinguished sides of tail, covered by closely packed den- from the Recent genera Potamotrygon Gar- ticulation; denticles minute, with posteriorly man, 1877, Paratrygon DumeÂril, 1852, and pointed hooklike crowns and stellate bases, Plesiotrygon Rosa, Castello, and Thorson, not forming discrete bucklers (also with dis- 1987 (Potamotrygonidae) by the absence of tinct series of enlarged spines forming rows the median prepelvic process. The combina-

← region (axial cartilage dark, impressions of terminal cartilages are indicated by arrowhead). Some sections of pectoral disc are missing. Anterior to top. 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 5. Relatively complete paratype of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 14069, ca. 149 mm TL, presumably an immature female) in ventral view. Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Arrowhead depicts well-exposed dorsal ®n at origin of caudal stings. Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 19

Fig. 6. ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 12914, ca. 205 mm TL, presumably adult female) in dorsal view. Specimen is from freshwater F-1 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Anterior left side of disc is missing, along with slab; arrowhead indicates posterior right portion of pectoral radials and metapterygium that have been disarticulated and overlap disc. Anterior to top. tion of external features used to separate the beyond caudal sting to posterior tip of tail new genus from fossil stingrays also sepa- (implying the absence of the cartilaginous rates it from all Recent myliobatiform genera notochordal rod at level of caudal stings, pre- as well. From potamotrygonids, Dasyatis Raf- sent in the above genera), circular to slightly inesque, 1810, Pastinachus RuÈppell, 1828, oval disc shape (as opposed to a more trap- Himantura MuÈller and Henle, 1837, Taeni- ezoidal or rhomboidal disc in most species, ura MuÈller and Henle, 1837, Pteroplatytry- except Urogymnus), and a conspicuously gon Fowler, 1910, and Urogymnus MuÈller thicker tail at base that is not whiplike (large and Henle, 1837 (Dasyatidae), ²Asterotry- specimens of Potamotrygon may have a gon, n.gen. is distinguished by the presence thick, not whiplike, tail as well). The ques- of the dorsal ®n (absent in all above genera), tionable dasyatid Urolophoides giganteus presence of individual vertebrae extending Lindberg, 1930 (Urolophoides Lindberg, 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

1930 is most likely a junior synonym of Das- cownose, and manta rays (Myliobatidae), in- yatis Ra®nesque, 1810) has a short, stout tail, cluding ²Promyliobatis gazolae from Monte but lacks the dorsal ®n and intense shagreen, Bolca, are easily separated from ²Asterotry- as well as having a strongly rhomboidal disc gon, n.gen. by disc shape (invariably broader (Lindberg and Legeza, 1959; Nishida and than long in these groups), head anterior to Nakaya, 1990). Urogymnus is further distin- and separated from disc (pectoral disc pro- guished from ²Asterotrygon, n.gen. by its jects anterior to head in ²Asterotrygon, lack of stings and proportionally shorter dis- n.gen.), less intense dorsal denticulation tance between eyes and anterior tip of disc (usually naked dorsal surface in pelagic (Compagno and Roberts, 1984; Last and Ste- stingrays), length of tail (usually long and vens, 1994). whiplike in myliobatids, but shorter and stout ²Asterotrygon, n.gen. differs from Ple- in ²Asterotrygon, n.gen.), naked dorsal ®n siobatis Nishida, 1990 (Plesiobatidae) in its (dorsal ®n covered in denticles in ²Astero- moderate snout length and anterior disc con- trygon, n.gen.), presence of cartilaginous rod tour (snout very long and anterior disc point- extending posteriorly from region of caudal ed in Plesiobatis), presence of dorsal ®n (ab- stings instead of individual vertebrae as in sent in Plesiobatis), and greater proximity of ²Asterotrygon, n.gen., presence of cephalic eyes to anterior margin of disc (eyes very extensions (``cephalic ®ns'') in Mobula Raf- reduced and located far from snout tip in Ple- inesque, 1810 and Manta Bancroft, 1828 (ab- siobatis; Last and Stevens, 1994). ²Astero- sent in ²Asterotrygon, n.gen.), and by differ- trygon, n.gen. is distinct from urotrygonids ences in dentition (teeth numerous, small, (Urobatis Garman, 1913 and Urotrygon Gill, and closely packed, with subtriangular cusps 1863) and urolophids (Urolophus MuÈller and in ²Asterotrygon, n.gen., as in most nonmy- Henle, 1837 and Trygonoptera MuÈller and liobatid stingrays; teeth in all myliobatids ex- Henle, 1841) by having a dorsal ®n covered cept manta rays are arranged in broad tooth- by denticles (dorsal ®n completely absent in plates). Anatomical features that further dis- the former two genera and present in some tinguish ²Asterotrygon, n.gen. from some or species of the latter two genera, but never all of the above genera are discussed in the coated with denticles), intense covering of skeletal description below. dermal denticles over dorsal surface (dorsal Many Recent genera assigned to the Das- disc and tail surface generally with sparse yatidae and Urolophidae are based on exter- denticulation in the former two genera, and nal characters not readily available in fossils mostly naked in the latter two genera), and (e.g., lack of tail-folds on both upper and lack of elongated caudal ®n (invariably pre- lower surfaces of tail in Himantura; lack of sent in all four genera, with conspicuous dor- dorsal tail-fold, but tall and long ventral tail- sal and ventral lobes that are internally sup- fold extending to distal tip of tail in Taeni- ported by radial cartilages). ura, etc.). Skeletal characters unique to most ²Asterotrygon, n.gen. is easily separated nonmyliobatid genera have not been found, from Hexatrygon Heemstra and Smith, 1980 and the skeleton is generally very conserva- (Hexatrygonidae) by the presence of dorsal tive in both Recent and fossil nonmyliobatid ®n (absent in Hexatrygon) and by snout and genera. The de®nition given here to the new disc shape (snout extremely elongated, tri- genus ²Asterotrygon, n.gen. is nevertheless angular, and somewhat demarked from disc consistent with generic diagnoses of Recent in Hexatrygon). From butter¯y rays (Gym- stingrays, allowing for a quick and straight- nuridae, Gymnura Kuhl, 1823, and Aetopla- forward identi®cation of all ²Asterotrygon, tea Valenciennes, 1841), ²Asterotrygon, n.gen. specimens examined. No other fossil n.gen. is distinguished by disc shape (much or Recent stingray genus has a dorsal ®n cov- broader than long in gymnurids, but disc ered with hooklike denticles and the unique width and length are about equal in ²Aster- combination of individual vertebrae extend- otrygon, n.gen.), stout tail that is about equal ing posteriorly to distal tip of tail, closely to disc length (tail slender and short in gym- packed denticulation over disc, snout and nurids), and covering of dermal denticles tail, and stout tail at base. (disc mostly naked in gymnurids). Eagle, ETYMOLOGY: The new generic name is de- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 21

Fig. 7. ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 14567, ca. 469 mm TL adult female), in dorsal view. Specimen is from freshwater F-1 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Some left pectoral radials and anteriormost tip of disc are missing. Anterior to top. 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 8. ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 14098, ca. 344 mm TL adult female) in dorsal view. Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation (currently on exhibit in the FMNH). Some pectoral radials and anterior region of neurocranium missing, gill arches disarticulated. Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 23

Fig. 9. ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 14097, ca. 402 mm TL adult female) in dorsal view. Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. Portions of neurocranium, jaws, visceral arches, pectoral basals and radials, and vertebrae are missing. Anterior to top. 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

TABLE 2 Measurements and Counts Conducted on Specimens of ²Asterotrygon maloneyi, n.gen., n.sp. Values are expressed as mm/percentage of disc width, except for total length (TL) and disc length (DL), which are shown in mm. See Measurements and Terminology for abbreviations of parameters. All specimens are female except FMNH PF 12989.

rived from the Greek asteros, meaning Buchheim, 1994 (pp. 42±43, 52±53; ®g. 9a; ``star'', and trygon, the Greek word for sting- photograph, brief mention). ray, in reference to the star-shaped bases of ²Heliobatis: Maisey, 1996 (pp. 108, 112±114, pls. the dermal denticles scattered over dorsal 42, 43; brief account) (not of Marsh, 1877; mis- disc and tail regions (see description of den- identi®cation). ``female stingray'': Grande, 1998 (p. 69; photo- ticles below; also ®g. 25). Gender feminine. graph of AMNH P 11557 only); Grande, 2002 TYPE-SPECIES:²Asterotrygon maloneyi, (p. 16; photograph of FMNH PF 15166 only). new species. `ùndescribed genus and species'': Grande, 2001 INCLUDED SPECIES: Presently considered to (pp. 6±8, ®g. 3b; table 1; brief mention). be monotypic. DIAGNOSIS: As for genus (see above). ²Asterotrygon maloneyi, new species MATERIALS: All specimens of the new ge- Figures 2±13, 18±26a, c, d, 27; tables 2, 3 nus and species of stingray are from the late early Eocene Fossil Butte Member of the `ùndescribed ray'': Grande, 1980, 1984 (pp. 23, Green River Formation. Specimens are either 25, 28±30; ®gs. II.6a, II.6b, II.7a, II.7b; partial description). from the nearshore deposits (F-2) or from the `ùndescribed [genus]'': Grande, 1989 (p. 25; midlake deposits (F-1). The quarries from brief mention). where specimens were excavated are given `ùndescribed species and genus'': Grande and here when available (refer to ®g. 1), along 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 25

TABLE 3 view, from the Hebdon quarry (F-2) (®g. 5). Measurements and Counts Conducted on Additional specimens: FMNH PF 12914, Juvenile Specimens of ²Asterotrygon maloneyi, female (205 mm TL), in dorsal view, from n.gen., n.sp. Tynsky quarry (F-1) (®g. 6); FMNH PF Values are expressed as mm/percentage of disc width, except for total length (TL) and disc length (DL), which 12990, adult (?) female (243 mm TL), in are shown in mm. See Measurements and Terminology ventral view, from the Tynsky quarry (F-2) for abbreviations of parameters. All specimens are (®g. 10A); FMNH PF 14097, adult female, female. preserved in part and counterpart (402 mm TL), from Tynsky quarry (F-2) (®g. 9); FMNH PF 14098, adult female (344 mm TL), in ventral view (F-2; currently on exhibit in the FMNH) (®g. 8); FMNH PF 14567, adult female (469 mm TL), in dorsal view (F-1) (®g. 7); FMNH PF 15180 (106 mm TL), juvenile female (F-2), in ventral view (®g. 12); FMNH PF 15181 (ca. 122 mm TL, 14 mm in height; specimen is tail only), exposed in lateral view (F-2) (®g. 24F); NHM P 61244, juvenile or neonate (86 mm TL), apparently female, probably exposed in dorsal view (F-2) (®g. 11A); SMMP 83.25, neonate or aborted fetus (80 mm TL), sex dif®cult to determine, counterpart of specimen near tail region of paratype AMNH P 11557 (F-2) (®g. 11D); USNM 2028, adult female (about 450 mm TL), in dorsal view (F-2) (®g. 10B). Tentatively identi®ed material: AMNH P 858, 2 small specimens, comparable to male paratype (FMNH PF 12989) in size, probably exposed in ventral view, with only anterior and lateral disc preserved (Twin Creek, probably F-1). STRATIGRAPHIC HORIZON: Fossil Butte Member of the Green River Formation, Yp- resian stage (corresponding to the North American Wasatchian stage), late early Eo- with information concerning how each spec- cene epoch (approximately 52 million years imen is exposed. Holotype: FMNH PF before present). 15166, a well-preserved adult female (ca. REMARKS: The ®rst mention of the new 625 mm TL), in ventral view, with an unborn stingray taxon described here from the Green late-term fetus in the abdominal region (from River Formation is in Grande (1980, 1984). an F-2 locality) (®g. 2); this slab has been Grande (1980, 1984: 23) provided a brief previously ®gured in Grande and Buchheim characterization of ²Asterotrygon, n.gen., (1994: 43, ®g. 9a) and Grande (2001: 10, ®g. noting that ``the specimens in ®gures II.6 and 3b). Paratypes: AMNH P 11557, adult fe- II.7 have thick tails covered with a dense se- male (378 mm TL), preserved along with ries of dermal denticles (placoid scales) bear- two small specimens (aborted late-term fe- ing curved hooks ....These may represent tuses) on limestone slab, in ventral view (F- a new species ...''.(Note that Grande's ®g- 2) (®g. 3); FMNH PF 12989, adult male (238 ure II.7c depicts the holotype of ²Heliobatis mm TL), in ventral view exposed, from the radians Marsh, 1877, and not a specimen of Tynsky quarry (F-2; collected by L. Grande, ²Asterotrygon maloneyi, n.sp. as implied). It 1984) (®g. 4); FMNH PF 14069, female (149 is dif®cult to precisely determine the number mm TL), presumably immature, in ventral of stingray specimens extracted from the 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 10. Preadult or small adult specimens of ²Asterotrygon maloneyi, n.gen., n.sp. Both specimens are from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. A. FMNH PF 12990, approximately 243 mm TL female, in ventral view. Note that anterior, central and much of lateral disc region, and distal tip of tail are missing. B. USNM 2028, approximately 450 mm TL female, dorsally exposed. Visceral arches, pectoral basals, and pelvic girdle are dislocated in this specimen. Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 27

Fig. 10. Continued. 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 11. Juveniles (and/or neonates) and fetuses of ²Asterotrygon maloneyi, n.gen., n.sp. All specimens are from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation. A. NHM P 61244, 86 mm TL juvenile, sex not determined, appears to be dorsally exposed. B. Late-term fetus preserved on left side of slab (adjacent to right side of disc) of paratype AMNH P 11557 (shown in ®g. 3); much of the specimen is not preserved, such as pectoral and pelvic radials and pelvic girdle; radial elements of pectoral disc of AMNH P 11557 visible on right side. C. Late-term fetus (some 80 mm TL, sex not determined) preserved on the right side of slab (to the left of posterior tail region) of paratype AMNH P 11557 (see ®g. 3); note that some features that are missing in this fossil are present in counterpart (D) to the right. D. SMMP 83.25, late-term fetus (counterpart of C). Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 29

Fig. 11. Continued.

Green River Formation, but of the approxi- spination that is concordant with that ob- mately 150 specimens of stingrays that we served among species of living stingray gen- have seen in a period of some 20 years, only era, for example, Dasyatis, Himantura, Uro- 15 are of ²Asterotrygon, n.gen. (in addition trygon, and Potamotrygon. Some specimens, to two more specimens that are only tenta- including the type series, have a prominent tively identi®edÐAMNH P 858). covering of denticles over all or most of the Presently, we describe only one species in dorsal disc region (e.g., ®gs. 3±6), while oth- the new genus ²Asterotrygon, n.gen. How- ers appear to lack intense denticulation over ever, the morphological variation observed outer disc margins (e.g., ®gs. 2, 7±9; AMNH among our specimens may indicate the pres- P 858). This does not appear to be due to ence of more than one species. ²Asterotry- sexual dimorphism, as FMNH PF 12989 is gon, n.gen. specimens exhibit variation in the only de®nite male specimen examined. 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Also, FMNH PF 12990 (®g. 10A) appears to n.gen. occurs in conjunction with other char- have a somewhat more rounded, urolophid- acters indicative of its generic separation like caudal ®n, slightly distinct from the tail- (e.g., covering of minute denticles over most folds present in all other specimens. How- of disc and tail, presence of individual ver- ever, this probably represents an elongated tebrae extending beyond caudal stings to dorsal caudal tail-fold supported internally posterior tail tip). Furthermore, the dorsal ®n by radials as in Potamotrygon (®g. 37D), and in ²Asterotrygon, n.gen. is unique among not a discrete and conspicuously shaped cau- stingrays in being thoroughly covered by dal ®n (see description of appendicular skel- denticles. eton below). The distal portion of the tail is The onset of sexual maturity for males missing in this specimen, so we cannot be must occur at a relatively small size, just be- certain of its original condition. Consequent- fore or around 120 mm in disc length, as the ly, we think that the Green River stingray male paratype has both claspers well pre- fauna may have been more diverse than pres- served and calci®ed (FMNH PF 12989, ca. ently reported, but we deem it premature to 240 mm in TL, 120 mm in DL, 115 mm in erect further taxa until more material is avail- DW; ®gs. 3, 23A). The size at which females able and details concerning the arrangement become sexually mature is more dif®cult to of denticles can be more fully understood. As discern in fossils, as no reproductive organs pointed out by Grande (2001), underestimat- are present. However, we can infer based on ing the species diversity in the extinct Green extant stingrays that female and male sizes River Lakes is mostly unavoidable, as im- at sexual maturity were probably similar to portant diagnostic features may not be pre- each other in ²Asterotrygon, n.gen. We know served in the fossils (such as color patterns, with certainty that the holotype (measuring crucial in identifying many species of living 625 mm in TL, 338 mm in DW) is an adult stingrays). A comparison between the ®sh as- female because it contains an unborn fetal semblages of the extinct Green River Lakes specimen (®gs. 2, 13), and that one female and the modern eastern African Rift lakes re- paratype (®g. 3) is in all likelihood sexually veals a much greater similarity in numbers mature as well (see following paragraph), as of families than in numbers of species (rang- it measures 378 mm in TL and 190 mm in ing from 10 families in Lake Gosiute to 14 DL and DW; females probably became sex- in Fossil Lake, compared to between 9 and ually mature at a much smaller size. 14 families in the African Rift lakes). This The two small specimens preserved along- may be an indication that the species diver- side one of the paratypes (AMNH P 11557; sity of many different groups of ®shes in the ®g. 3) are considered to be its aborted late- Green River Formation was originally great- term fetuses (which we believe were proba- er than our current estimates based on their bly very close to parturition) on the basis of fossil remains (Grande, 2001). the following evidence: they are unquestion- A dorsal ®n occurs in the following sting- ably identi®ed as ²Asterotrygon maloneyi,n. ray groups, besides ²Asterotrygon, n.gen.: sp.; their size in relation to their mother con- Myliobatidae, some gymnurids (Aetoplatea), forms very closely to maternal/late-term fetal and in selected species of Trygonoptera (in proportions in Recent species of stingrays; three of the six species) and Urolophus (in stingrays are a relatively rare component of about one-third of the species; 15 species are the Fossil Butte ®sh assemblage; and the recognized as valid by Last and Stevens, probability of having two small fetuses pre- 1994; Compagno, 1999). The occurrence of served alongside the mother after birth is un- the dorsal ®n in 2 of the approximately 12 likely given that there is generally no rearing species of gymnurids, and only in certain of the young or other forms of parental care species of Trygonoptera and Urolophus, may in chondrichthyans (Breder and Rosen, 1966; indicate that the mere presence of the ®n is Wourms, 1977), even though neonates of at not a reliable indicator of generic status, even least two species of Potamotrygon have been though it is the primary character for distin- observed to remain on their mother's disc for guishing between Aetoplatea and Gymnura. a period of a few days after birth (Achenbach However, the dorsal ®n in ²Asterotrygon, and Achenbach, 1976; ArauÂjo, 1998). The 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 31 fetuses contain small, scattered denticles over only one functional ovary, or one of the two the dorsal surface of the disc, albeit not as ovaries or uteri more developed than the oth- intensely as seen in the type series, but this er; Bigelow and Schroeder, 1953; CapapeÂet to be expected as fetuses generally have less al., 1992; Henningsen, 1996; W.D. Smith and intense denticulation compared to juveniles N.K. Dulvy, personal commun.). The sys- or adults. ²Heliobatis specimens lack dentic- tematic relevance of reproductive tract sym- ulation other than over the midline area close metry must await the availability of more to the tail. Preservation of the two fetuses is data for certain stingray genera (e.g., some not as complete as in the other specimens, pelagic stingrays, Hexatrygon, Plesiobatis). but again this is not unusual given that their However, the presence of symmetrical repro- skeletons are less calci®ed. The tail region of ductive tracts is probably the more general these specimens is poorly preserved, as are batoid condition (guitar®shes, electric rays, their small dorsal ®ns (NHM 61244, a small and pristids have both reproductive tracts neonate, nevertheless has a small dorsal ®n functional, as do pristiophorids), and asym- remnant clearly preserved). Previous ac- metry may have evolved as a stingray syn- counts of these specimens (AMNH P 11557) apomorphy (in this scenario, it would be re- reported three fetuses on the slab alongside versed at least in potamotrygonids, gymnur- the adult female (Grande 1980, 1984), but ids, and Urobatis). If this proves to be the one of these, situated close to the anterior case (i.e., asymmetry is derived for sting- disc margin, is actually a poorly preserved rays), then it is reasonable to assume that teleost (L. Meeker, personal obs.). ²Asterotrygon, n.gen. was asymmetrical as We have observed extant gravid female well, as ²Asterotrygon, n.gen. is resolved as stingrays of different genera and species a basal stingray (see phylogenetic analysis aborting fetuses upon capture. In particular, below; however, this still depends on the Urobatis jamaicensis from the Caribbean Sea condition in other basal stingrays such as (off Belize) agrees very closely in maternal/ Hexatrygon and Plesiobatis). fetal (late-term) proportions to ²Asterotry- ETYMOLOGY: The speci®c epithet, malo- gon, n.gen. as inferred from AMNH P neyi, is a patronym given in recognition of 11557. The relative rarity of ²Asterotrygon, the generous donator of the largest paratype n.gen. specimens compared to ²Heliobatis, (AMNH P 11557; ®g. 3), Thomas Maloney. coupled with the extreme rarity of fossilized neonates or small juveniles of stingrays in ANATOMICAL DESCRIPTION general, leads to the conclusion that paratype AMNH P 11557, an adult female, is mater- The following anatomical description of nally related to the two small specimens ly- ²Asterotrygon roughly applies to ²Heliobatis ing adjacently on the same slab. as well. Apart from diagnostic characters that The late-term embryo inside the holotype distinguish both taxa, their skeleton is very of ²Asterotrygon, n.gen. (®gs. 2, 13) appears generalized and similar, not differing greatly to be situated in the left uterus (the specimen from the skeleton of extant nonpelagic sting- is preserved in ventral view). This does not rays in most details that can be directly com- preclude the possibility that a fetal specimen pared. Extant dasyatid stingrays are also may have originally been present in the right morphologically similar in many respects uterus as well, which may have been aborted (Miyake, 1988; Nishida, 1990), a condition prior to death. Many extant stingrays have that has obscured their proper de®nition, only the left oviduct functional (such as das- which presently relies to a large degree on yatids; Bigelow and Schroeder, 1953; Snel- external characters. The fossil specimens son et al., 1988, 1989; H.F. Mollett and F.F. studied here are two-dimensionally pre- Snelson, Jr., personal commun.), while others served in limestone slabs that are not readily have both uteri functional and sometimes acid-prepared and do not lend themselves to synchronous (potamotrygonids, Urobatis, other matrix-removal methods (outside of Gymnura; Babel, 1967; Thorson et al., manual work with needles). Therefore, much 1983b; Carvalho, personal obs.), frequently anatomical detail that might be of use in de- with uneven development (such as having ®ning groups is not available (such as the 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 12. Juvenile specimen of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 15180, ca. 106 mm TL female, ventrally exposed). Specimen is from freshwater F-2 deposits of the early Eocene Fossil Butte Member of the Green River Formation; preserved alongside a teleost, the percopsid ²Amphiplaga brachyptera. Anterior to top. distribution of foramina within the orbit, ar- FMNH PF 15180 was particularly informa- rangement of condyles on scapulocoracoid). tive in relation to details of the ventral gill The descriptions below are based on many arches (®gs. 12, 20), including the only ba- of the fossil specimens, especially the type sihyal reported from a fossil stingray. series, as no single specimen provided all in- To serve as guides for the anatomical de- formative anatomical details. However, scription of the Green River stingrays, we 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 33

Fig. 13. Fetal specimen present in pleuroperitoneal cavity of holotype of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 15166; ®g. 2), enlarged to show caudal sting (indicated by arrowhead). provide images of entire cleared-and-stained cal structures given in the description below stingrays (of a round ray, Urotrygon chilen- are depicted and labeled in ®gs. 16±26, 31± sis in ®g. 14, and of a butter¯y ray, Gymnura 42). micrura, in ®g. 15), along with an illustration Prismatic super®cial calci®cation covers of a complete potamotrygonid stingray skel- much of the skeleton and is particularly in- eton (Potamotrygon sp. from the Rio Taqua- tense on the neurocranium, scapulocoracoid, ri, Rio Paraguai basin, Brazil; ®gs. 16, 17). pectoral basals, and puboischiadic bar. The Additional morphological details of these prisms are small (less than 1 mm in diame- two cleared-and-stained specimens, and of ter), generally polygonal but somewhat irreg- other extant stingray taxa, are given in ®gs. ular in shape, and vary slightly in size. A 31±42 (note that abbreviations for anatomi- single, thin layer of prismatic calci®cation is 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 14. Dorsal view of cleared-and-stained specimen of Urotrygon chilensis (FMNH 93737) showing articulated skeleton; included to serve as a guide for the anatomical descriptions contained in this paper (ventral view of same specimen on opposing page). This specimen is further depicted (as close- ups) in ®gures 33, and 36±39. Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 35

Fig. 14. Continued. Ventral view. 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 present, a widespread condition among neo- triangular rostral extremity (not the rostral selachians and most Mesozoic elasmo- appendix of Miyake et al., 1992b) is present branchs, but unlike xenacanths and a few between the pectoral ®n extremes (Holm- other Paleozoic sharks (Schaeffer, 1981; cf. gren, 1940; Rosa, 1985; Miyake, 1988; Mi- érvig, 1951). The juvenile specimens ex- yake et al., 1992b; McEachran et al., 1996). amined (e.g., NHM P 61244, SMMP 83.25) The subtriangular rostral extremity of sting- have less prismatic calci®cation compared to rays forms in connection with the anterior the larger fossils. medial outgrouth of the trabecula; it is carried forward by the expanding neurocranium NEUROCRANIUM in conjunction with pectoral ®n extension, but appears to be absent in pelagic stingrays The neurocranium (n) is preserved at least (Mobula kuhlii [AMNH 15319]; however, partially in almost all specimens, but FMNH more ontogenetic data are needed for pelagic PF 12989 (paratype; ®g. 4) was particularly stingrays). The rostral extremity is also ab- informative. Specimens FMNH PF 15166 sent in most small specimens of ²Asterotry- (holotype; ®g. 2), FMNH PF 15180 (®g. 12), gon (e.g., ®g. 11D), which is not surprising AMNH P 11557 (paratype; ®g. 3), FMNH given that in extant stingrays it is a hyaline, PF 14069 (paratype; ®g. 5), FMNH PF uncalci®ed structure; its absence in ²Astero- 12914 (®g. 6), FMNH PF 14567 (®g. 7), trygon may therefore be preservational. This FMNH PF 14098 (®g. 8), USNM 2028 (®g. conclusion is supported by specimen NHM 10B), and NHM P 61244 (®g. 11A) also pro- P 61244 (40 mm in DW and DL, 86 mm in vided structural details. In dorsal view, the TL; ®g. 11A), in which a small, wedgelike neurocranium of ²Asterotrygon has an out- structure (re?) occurs between both anterior line generally indistinct from extant nonmy- pectoral ®n extremes; this structure is par- liobatid stingrays such as potamotrygonids, tially covered by denticles and may represent Trygonoptera, urotrygonids, and most dasy- an impression of the rostrum. In juvenile atids. The neurocranium is widest anteriorly stingray specimens in which it is present, the at the level of the nasal capsules, where it is rostral extremity is usually not posteriorly almost in contact with the propterygium on elongated. This may be only an artifact of both sides. A keyhole-shaped dorsal fonta- preparation, however, as the rostral extremity nelle is present and can be seen in most spec- of Urolophus cruciatus (AMNH 98247) ex- imens, even those that are exposed ventrally tends farther posteriorly, comprising about (e.g., ®g. 4). Neurocranial features that can one-third of the distance between the anterior be observed are those that are prominent dor- disc margin and the anterior neurocranial soventrally. Prismatic calci®cation covers al- contour. Whether the rostral extremity con- most all of the neurocranium, including the tinues all the way to the anterior neurocranial nasal capsule area, and some specimens are margin in this specimen, perhaps as a very heavily calci®ed (®g. 7). The intense sha- slender and uncalci®ed rod, is not clear. In green of small to medium-sized denticles, addition, stingrays are reported to have a ves- covering much of the neurocranium, contrib- tigial medial outgrowth of the trabecula at utes to the dif®culty of observing certain fea- the junction of the nasal capsules (Holmgren, tures, such as the shape of the postorbital 1940; Miyake et al., 1992b), which eventu- processes in the male paratype (®g. 4). ally disappears in fully formed individuals A rostral cartilage (or rostrum; ro) is not (there is no indication of this structure in present in our adult fossil specimens, and Urolophus cruciatus [AMNH 98247]). This ²Asterotrygon lacks any calci®ed neurocra- outgrowth, supposedly the rudimentary ros- nial structure anterior to the nasal capsules, tral base, is lacking in Gymnura (®g. 31), as do adult stingrays in general. This con- Taeniura lymma (®g. 32), Urotrygon chilen- trasts with the condition observed in embry- sis (®g. 33), and potamotrygonids (®g. 34), onic or juvenile specimens of various extant but may be vestigially present in Urobatis stingray taxa (Urobatis, Urotrygon, Urolo- halleri (®g. 38A, ro?); its arrangement and phus, Trygonoptera, Potamotrygon, Plesi- distribution among stingrays is yet to be de- otrygon), in which a small, detached sub- termined. Among ²Asterotrygon specimens, 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 37 only FMNH PF 15180 appears to have a moidal wall) was presumably truncated, as in small, triangular outgrowth at the anterior modern stingrays and batoids in which the margin of the neurocranium between the na- rostral cartilages are reduced (Miyake, 1988). sal capsules (®g. 20). However, this structure As it has not been excavated and is partially may be part of the nasal capsules, the rostral crushed, the depth of the precerebral fossa is base per se, or the rostral extremity that has not known in our fossils. Nasal cartilages not been dislocated anteriorly with the ante- (ncr) are not preserved in ²Asterotrygon, rior extension of the pectoral ®n. even in FMNH PF 15180, a specimen with The nasal capsules (nc) in ²Asterotrygon preserved olfactory canals (®g. 20). are wider than long and are oval in general The neurocranium is markedly curved an- outline. There is a slight notch or concavity terolaterally at the level of the nasal capsules, anteriorly in between both nasal capsules, and appears to abut the internal surface of where they become con¯uent. This is more the propterygium. A preorbital process (prp) apparent in certain specimens (e.g., ®g. 18), was present on both sides of the neurocra- including the holotype, where the anterior nium, as in extant stingrays. In Recent ben- cranial contour is well preserved, but is only thic stingrays the preorbital process is trian- faintly visible in paratype AMNH P 11557. gular, directed posterolaterally and situated It is also not possible to precisely determine behind the nasal capsules on the dorsal sur- the extent of the dorsal internasal septum in face of the neurocranium. In our fossils, most specimens, but FMNH PF 12989 clear- however, this region is obscured by massive ly has a slender prismatically calci®ed area jaw cartilages and perhaps also by the ven- between the closely set olfactory canals (oc), trally situated antorbital cartilages. The an- somewhat similar to Urotrygon venezuelae torbital cartilages (aoc) are almost ventral to- (Miyake, 1988: 163, ®g. 35a). The dorsal in- pological counterparts of the dorsal preorbit- ternasal septum, considered to be the anterior al processes in many extant and fossil sting- segment of the trabecular cartilage (El-Toubi rays. They are situated ventrolaterally on the and Hamdy, 1959; Miyake et al., 1992b), nasal capsule just anterior to the level of the separates the nasal sacs and capsules; its ex- preorbital processes, and are generally more tent is determined by the trabecular contri- massive than the preorbital processes. Be- bution, nasal aperture size and position rel- cause fossil stingrays are dorsoventrally pre- ative to each other, the arrangement of the served, structures that are topologically on olfactory canals, and by the forward and lat- the same dorsoventral plane will usually be eral expansions of the central nucleus and obliterated. The antorbital cartilages are sub- lateral pallium of the telencephalon (North- triangular structures, usually curved and ta- cutt, 1978). The olfactory canals in FMNH pering caudally from the ventral posterolat- PF 12989 are visible through the precerebral eral surface of the nasal capsule with which fontanelle (seen in dorsoventral view), and they articulate. They are anteroposteriorly the dorsal internasal septum appears to be ¯attened in stingrays. In FMNH PF 12989 relatively slender (®g. 18). In the holotype, the antorbitals are about one-®fth as long as however, the internasal septum is wider, with the neurocranium (®g. 18), reaching to mid- a greater distance between the olfactory ca- neurocranial length, and are therefore pro- nals. The width of the dorsal internasal sep- portionally larger than in many extant uro- tum varies among stingray taxa, being rela- lophid and potamotrygonid species. There is tively wide in myliobatids, Hexatrygon, Ple- some variation in antorbital cartilage shape, siobatis, Trygonoptera, Urobatis, and some as in FMNH PF 14567; the left antorbital, species of Himantura and Dasyatis (accord- which remains articulated, is triangular but ing to material examined). Species of pota- straighter than in other fossils (®g. 19; this motrygonids also vary in this character, and structure is identi®ed as the antorbital, as it it is dif®cult at present to precisely quantify appears to be articulated and is too large to the width of the dorsal internasal septum in be the angular cartilage, judging from the lit- stingrays (more comparisons are presently tle space available for the hyomandibular- being undertaken). The precerebral fossa (a Meckelian ligament in this specimen). In perforation of the anterior cranial or eth- many specimens the antorbitals remain artic- 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 15. Dorsal aspect of cleared-and-stained butter¯y ray, Gymnura micrura (FMNH 89990) (ventral side of same specimen on opposing page). Anatomical details are further shown in ®gures 31, 35, 36, 38 and 39. Anterior to left. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 39

Fig. 15. Continued. Ventral view. 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 16. Skeleton of Potamotrygon sp., an extant freshwater stingray (Potamotrygonidae) from the Rio Taquari, Rio Paraguai basin, Brazil (MZUSP 25663, 336 mm TL, subadult male). This illustration is provided to serve as a comparative basis and guide for the anatomical description of the Green River stingrays. A. Entire skeleton. B. Detail of pelvic girdle (to scale) showing the elongated median prepelvic process, characteristic of potamotrygonids. Note that in panel A the gill arches and gill rays are sche- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 41 ulated in their original position, but in others frontoparietal fontanelle), but this may be an they are displaced to ®ll the space between artifact of preservation. The fontanelle is rel- the orbits and internal margins of the prop- atively wide in the holotype, where its inter- terygia. The left antorbital cartilage is par- nal borders are well delineated posteriorly. tially preserved in paratype AMNH P 11557, The posterior portion of the dorsal fontanelle and is relatively smaller than in FMNH PF displays no constriction in the fossils, and 12989, appearing to taper less while being therefore the tectum orbitale is not visible. more blunt at its posterior tip. The preorbital However, some stingrays may retain a pos- canal (poc), through which pass branches of terior constriction as well, faintly dividing the trigeminal (V) and other nerves from the the dorsal fontanelle farther posteriorly orbit to the ethmoid region, is not visible in (Meng, 1984). The frontoparietal fontanelle specimens of ²Asterotrygon, even though it tapers posteriorly, ending in an oval curve, is enlarged in Recent stingrays anterior to the which is clearly visible in many ²Asterotry- preorbital process on the dorsal surface of the gon specimens, including the holotype. nasal capsule (®gs. 31±34). The orbital region of ²Asterotrygon is lon- A single elongated dorsal fontanelle, rep- ger than wide and occupies less than two- resenting the unchondri®ed anterior roof of thirds of neurocranial length, being widest the cranial cavity (de Beer, 1937), is present anteriorly at the level of the preorbital pro- in stingrays and occupies over two-thirds of cesses. The orbits are therefore relatively neurocranial length. However, some modern elongate, but this does not mean that the eye stingray taxa exhibit a median constriction, diameter was also large. The eyes in ²Aster- representing a remnant of the epiphysial bar otrygon may have been large and bulging, as (epb), partially separating an anterior precer- in some Recent stingray genera (Potamotry- ebral fontanelle (pcf) from a posterior fron- gon, Dasyatis, Taeniura, Urobatis, Urolo- toparietal fontanelle (fpf) (Miyake, 1988). phus), but orbit length does not necessarily The epiphysial bar is more intact in juvenile correlate with eye diameter (Plesiotrygon has stingray specimens, but is especially well de- orbital proportions similar to Potamotrygon, veloped in Potamotrygon, where it can be even though it has signi®cantly smaller eyes almost complete (numerous specimens; ®g. in comparison). The orbits are slightly con- 34A); it appears as an anteromedian constriction in the dorsal fontanelle in Taeniura lym- cave in dorsal perspective. It is dif®cult to ma (separating the pcf and fpf in ®g. 32A; determine the width and extent of the supra- also Garman, 1913: pl. 71, ®g. 6). Two dor- orbital shelf in our two-dimensional fossils, sally exposed (FMNH PF 14098 and USNM but it appears to have been con¯uent anter- 2028) and one ventrally exposed specimen ioly with the preorbital process, which is (FMNH PF 12989) support the presence of largely obscured. The orbital region of ²As- a median constriction in ²Asterotrygon, and terotrygon was arranged as in most benthic the dorsal fontanelle is faintly divided into stingrays, and not shortened as in gymnurids anterior (precerebral fontanelle) and posterior (®g. 31) and myliobatids. (frontoparietal fontanelle) regions, as in spe- Whether the supraorbital shelf was contin- cies of Urolophus, potamotrygonids, Pastin- uous with the postorbital process (pop), or if achus sephen, Pteroplatytrygon violacea, a notch separated both structures, is of some Plesiobatis daviesi, and Taeniura lymma. systematic signi®cance. Species of Urolo- Most ventrally exposed ²Asterotrygon spec- phus, Plesiobatis daviesi, Pteroplatytrygon, imens (the majority of specimens, including and pelagic stingrays (with the exception of the holotype) display a single unconstricted Myliobatis and possibly Aetomylaeus) have dorsal fontanelle (precerebral fontanelle ϩ the supraorbital crest continuous with the

← matically depicted for clarity, but that pectoral radials, vertebral centra, and other features are accurately rendered; the iliac process on the left side and marginal cartilages of the clasper have been omitted; left pelvic ®n skeleton is depicted dorsal to pectoral disc (metapterygium) for clarity. 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 17. A. Enlarged neurocranial region of Potamotrygon sp. from ®gure 16A, in dorsal view. postorbital process, while dasyatids (except process of this specimen is slightly directed Pteroplatytrygon), potamotrygonids, Uroba- anteriorly, forming an angle in relation to the tis and Urotrygon, Trygonoptera, and Hex- long axis of the neurocranium. In FMNH PF atrygon present a groove or notch between 15180 the anterior margin of the left post- both structures. The postorbital groove al- orbital process is preserved, which is lows for the passage of the infraorbital lateral obliquely oriented, reaching to about one- line canal. Accordingly, the anterior margin third of the length of the hyomandibula (sit- of the postorbital process is at an oblique an- uated directly behind it). A small triangular gle to the supraorbital crest in the taxa pre- protuberance (the equivalent of the supraor- senting a groove. In Urolophus, Pteroplaty- bital process, sp; ®g. 17A) is apparent ante- trygon, Plesiobatis, and most pelagic sting- rior to the postorbital process of FMNH PF rays, however, the canal passes through a 14567 (®g. 19), separated from it by a slight dorsoventral opening in the postorbital pro- indentation of the supraorbital crest. (Holm- cess (Miyake, 1988), which also has a more gren [1942: 179] referred to this protuber- transversely directed anterior pro®le. In ²As- ance as the ``supraorbital process'', a term terotrygon, enlarged nuchal, orbital, and/or we adopt, and used it to strengthen his pres- spiracular dermal denticles, along with the ently unsupported view of the close relation- prominent hyomandibulae, obscure the post- ship between certain petalichthyid placod- orbital process (e.g., in the holotype and in erms, such as ²Macropetalichthys, and rays; FMNH PF 12989; ®g. 18), which is best pre- cf. Jarvik, 1980: 376±378). The supraorbital served in FMNH PF 14567. The postorbital process is visible on both sides of the neu- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 43

Fig. 17. Continued. B. Ventral view of same. Gill arches are somewhat diagramatically illustrated as in ®gure 16A. rocranium, and ²Asterotrygon therefore pre- margins. The hyomandibular facet is located sents the condition seen in dasyatids, pota- laterally on the ventral aspect of the neuro- motrygonids, Trygonoptera, and urotrygon- cranium, more or less between the fronto- ids among nonmyliobatid stingrays. The parietal fontanelle and parietal fossa, and postorbital process is broad and shel¯ike in posterior to the postorbital processes. In Re- ²Asterotrygon, as in all stingrays. Foramina cent stingrays there is also an articular facet through the supraorbital crest for rami of the for the anteroposteriorly compressed dorsal super®cial ophthalmic nerve are not discern- pseudohyoid element posterior to the hyo- ible in any specimen of ²Asterotrygon. mandibular articulation, but this cannot be The parietal fossa (paf) in stingrays may observed in the fossils. The otic capsules at contain both the endo- and perilymphatic for- the posterior corners of the neurocranium are mina, or only a single pair of endolymphatic tumid, but along with the occipital condyles openings, and it comprises the posterior one- they are largely obscured or crushed in ²As- ®fth of the dorsal surface of the neurocrani- terotrygon by elements of the pseudohyoid um. In ²Asterotrygon the fossa is ®lled with arch, gill arches, and possibly synarcual car- matrix, and is similar to Recent stingrays in tilage, and little anatomical detail remains. proportions, but whether both pairs of foram- As in all stingrays, a jugal arch is absent ina are con¯uent or separate cannot be de- from the posterolateral corners of the neu- termined. A crest outlines the fossa along its rocranium. 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 18. Neurocranium and visceral arches of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 12989, adult male paratype; entire specimen in ®g. 4). A. Ventrally exposed splanchnocranium. B. Schematic outline of neurocranium from same specimen. Note intense, almost uniform covering of closely packed and very small dermal denticles, which are more visible anterior to neurocranium between pectoral radials. Larger denticles are obliterating postorbital processes. Only impressions in the matrix of the putative angular cartilage(s) (aac?) remain, and they are not clearly visible in the photograph (A). Note that hyomandibula on right side lacks a large central section of prismatic calci®cation, and that portions of nasal capsules, jaws, neurocranium, and gill arches are missing. Anterior to top.

JAWS,HYOMANDIBULAE AND otrygon are robust, massive structures as in VISCERAL ARCHES most Recent stingrays (not slender as in Par- The jaws are well preserved in the female atrygon and gymnurids), extending laterally paratypes (®g. 21), but especially in the ju- to occupy almost the entire space between venile specimen FMNH PF 15180 (all of both propterygia. Both sets of jaws are which are ventrally exposed; ®g. 20). The strongly reinforced with prismatic calci®ca- jaws are dislocated, obscured, or even miss- tion, and both upper and lower jaws are com- ing in other specimens. The jaws of ²Aster- posed of mirror-image antimeres that meet 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 45

Fig. 19. Left aspect of neurocranium of ²Asterotrygon maloneyi, n.gen., n.sp. (FMNH PF 14567; entire specimen in ®g. 7), dorsally exposed, showing triangular shape of antorbital cartilage (indicated by arrowhead). Anterior to top. mesially. The upper jaws (palatoquadrates; liobatid stingrays), and are separated by a pq) are slightly thicker than the lower jaws small space that originally contained strong (mandibular or Meckel's cartilages; Mc) in horizontal ligaments attached to both anti- AMNH P 11557, but not so in FMNH PF meres. The arched appearance of the jaws 14069 or FMNH PF 15180, and this may be varies among individuals and is also proba- due to differences in the orientation of the bly strongly correlated with jaw position jaws when preserved. The mandibular carti- when mineralized. Both mandibular and pal- lages are more arched than the palatoquad- atoquadrate cartilages taper slightly toward rates and bear a widely rounded posterior the midline, and both have pronounced outer margin (®gs. 20, 21). Both antimeres of the corners that are curved in most specimens. mandibular cartilages and palatoquadrates The upper surface of the mandibular carti- are not fused symphysially (as in certain my- lage is relatively straight in AMNH P 11557 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 20. Enlarged view of FMNH PF 15180 from ®gure 12, depicting extremely well-preserved jaws, ventral gill arches, and other features of a juvenile specimen of ²Asterotrygon maloneyi, n.gen., n.sp. Note preservation of basihyal cartilage, ®rst pair of hypobranchials, and gill rays of ventral pseudohyoid bar preserved in original positions. and FMNH PF 14069, but in FMNH PF with the palatoquadrate in extant nonmyliob- 15180 there is a circular opening between the atid stingrays. There is also a pronounced lower surface of the palatoquadrate and the curvature lateroexternally for the stout hyo- upper aspect of the lower jaw, as in most mandibular ligament. Both of these curved Recent nonmyliobatid stingrays. The palato- surfaces appear to be present in our fossils, quadrate resembles the condition in pota- along with the large fossa for the adductor- motrygonids, Dasyatis, Himantura, Taeni- mandibulae musculature, but because of ura, and urolophids in being relatively overlying structures and partial crushing of straight on its dorsal ¯ange. The mandibular the jaws, the degree of development of these cartilage contains a strong concavity at its surfaces and the exact nature of the palato- laterointernal corner in which it articulates quadrate-mandibular cartiage articulation 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 47

Fig. 21. Jaws of ²Asterotrygon maloneyi, n.gen., n.sp., with scattered teeth visible between upper and lower jaws. A. FMNH PF 14069 (paratype; entire specimen in ®g. 5). B. AMNH P 11557 (paratype; entire specimen in ®g. 3). Jaw outlines have been reinforced. cannot be further discerned, even though acter are still to be determined. The ventro- they appear to resemble that of extant sting- lateral process of FMNH PF 15180 is more rays (Dasyatis, Potamotrygon). apparent on the right side, projecting past the The ventrolateral processes of the mandib- palatoquadrate. This feature, along with any ular cartilage are absent in all specimens of super®cial feature of the external aspect of ²Asterotrygon except FMNH PF 15180. The the lower jaw, was probably obliterated in ventrolateral process is a triangular, posterior other specimens. The winglike process (a projection from both outer corners of the stout, lateral or posterolateral projection of mandibular cartilage, and it is present in the lower jaw) appears to be restricted to pe- Taeniura, potamotrygonids, and many other lagic stingrays (Nishida, 1990) and is clearly nonmyliobatid stingrays. Its precise distri- absent in ²Asterotrygon. bution and importance as a systematic char- The hyomandibulae of stingrays are elon- 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 gate cartilages articulating with the braincase (Hexatrygon). A very small and inconspicu- at its posterolateral surfaces, from which they ous ligament is present in pelagic stingrays extend anteroventrally to the level of the out- and gymnurids, and this is considered to be er jaw corners. In ²Asterotrygon, the hy- a more direct connection as in Hexatrygon. omandibula (hyo) is strongly calci®ed and In ²Asterotrygon, there is a small, slightly therefore relatively well preserved in most transverse element situated between the hy- specimens. It projects not only anteroven- omandibula and lower jaw (more conspicu- trally but also slightly laterally, reaching the ous in FMNH PF 15166, AMNH P 11557, internal wall of the propterygium just pos- FMNH PF 15180, FMNH PF 12989, NHM terior to the lower jaw corner, and almost 61244). This element does not represent the reaching the jaw joint. The hyomandibula is recurved, medially directed distal tip of the stout when observed from a lateral perspec- hyomandibula. The distal (anterior) section tive in extant stingrays, but it is more slender of the hyomandibula can be oriented in this when observed in dorsoventral view. In ²As- fashion even though the stout hyomandibu- terotrygon, however, the hyomandibulae are lar-Meckelian ligament is present (e.g., Uro- relatively stout, especially in the holotype batis halleri [®g. 38A], Urotrygon chilensis (where they are not very elongated), even [®g. 33], Himantura kremp® [AMNH though they are dorsoventrally preserved (the 41567]). This in¯ection of the hyomandibu- hyomandibulae are also fairly wide in the la, however, is usually small and does not paratypes and in FMNH PF 12914 and PF extend to contact the lower jaw. The trans- 14567; ®g. 19). The hyomandibula is gen- verse element of ²Asterotrygon contacts the erally more slender in smaller specimens lower jaw and is heavily covered by pris- (NHM 61244, FMNH PF 14069), but FMNH matic calci®cation (especially in FMNH PF PF 15180 is an exception (®g. 20), as it pre- 15180; however, the individual prisms are sents relatively wide hyomandibulae that do mostly missing or scattered, and usually only not taper from their articulation toward the their impressions remain). This prismatic cal- jaws (this may be due to its orientation when ci®cation supports the notion that a separate preserved). In most specimens, the hyoman- cartilage, similar to the angular cartilages dibula is widest close to its attachment to the (ac) of potamotrygonids, was present in the ventral otic region of the neurocranium (from ligament uniting the hyomandibula and low- a dorsoventral perspective), as in most extant er jaw in ²Asterotrygon (best observed in benthic stingrays. The internal surface of the FMNH PF 15180 and in NHM 61244, both hyomandibula is slightly concave, especially juvenile specimens). Holmgren (1943: 74) close to the jaw corners. The proximal por- considered the angular cartilages (his ``man- tion of the hyomandibula is also partially ob- dibular rays'') to be of mandibular arch ori- scured by enlarged nuchal denticles in some gin. specimens (e.g., ®g. 18). The hyomandibula Even though the hyomandibular-Meckeli- is devoid of rays, as in extant stingrays. Pre- an ligament may become strengthened, and postspiracular cartilages and ``hyoman- slightly chondri®ed, and even reinforced dibular accessory cartilages'' (Nishida, 1990) with scattered internal calci®cation, actual are not preserved or cannot be identi®ed in prismatic calci®cation is absent from the lig- ²Asterotrygon. ament in specimens presenting these condi- In stingrays, the articulation between the tions (e.g., Taeniura lymma [®g. 32], Try- hyomandibular cartilage and the lower jaw gonoptera testacea [®g. 40B]). This condi- can be either through a strong, stout hyo- tion may also vary intraspeci®cally (such as mandibular-Meckelian ligament (Taeniura, in Trygonoptera testacea) and does not ap- Dasyatis, Urolophus, Trygonoptera, Himan- pear in most radiographs. The angular carti- tura), through the ligament reinforced by lages of Potamotrygon and Plesiotrygon, separate cartilages formed within it (angular however, are prismatically calci®ed and cartilages of Potamotrygon and Plesiotrygon; clearly visible in radiographs (®g. 40). These Garman, 1913), or through a more direct cartilages are stout, oval to rectangular in connection, without a stout ligament between shape, and embedded within the strong hyo- the hyomandibulae and Meckelian cartilages mandibular ligament. As such, they are 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 49 unique among Recent stingray taxa to pota- ments of the last two ceratobranchials (un- motrygonids except Paratrygon according to con®rmed in Hexatrygon), anteriorly and our specimens (see also Garman, 1913; cf. obliquely directed short, separate hypobran- Lovejoy, 1996). Himantura schmardae chial 1 elements (hb1), and a small transverse (NHM 1908.5.28:2±3) has very small calci- and separate basihyal cartilage (bh). Addi- ®ed elements that are barely visible in radio- tionally, the ceratobranchial and pseudohyoid graphs (Lovejoy, 1996); they may be easily cartilages are fused in stingrays (the number overlooked in dissections. The distal (ante- of fused ceratobranchial elements varies rior) extremity of the hyomandibula of po- among taxa). Most pelagic stingrays lack the tamotrygonids is not in¯ected medially to a ®rst hypobranchial along with the basihyal large extent, where the hyomandibular- (Nishida, 1990), but whether this is primitive Meckelian ligament is present. Although or secondary depends on the myliobatiform somewhat tentative because of preservational phylogeny adopted (derived in our phyloge- issues, we conclude that the condition in netic study below; cf. appendix 2). Primi- ²Asterotrygon resembles the discrete angular tively in stingrays, the last pair of cerato- cartilages of Potamotrygon and Plesiotrygon, branchial and epibranchial elements articu- because these appear to be prismatically cal- late with the anteromedial aspect of the cor- ci®ed, discrete cartilages. In turn, this sug- acoid bar, and the basihyal and the medial gests that the hyomandibular-Meckelian lig- plate are separated by a large gap (Nelson, ament was present in ²Asterotrygon as well, 1969). even though impressions of the ligament are The pseudohyoid and gill arches are poor- not clearly preserved in the fossils (which is ly preserved in ²Asterotrygon. The most in- to be expected). In FMNH PF 15180, the pu- formative specimen is FMNH PF 15180 tative angulars appear to be composed of a (®gs. 12, 20), which has certain elements of single, relatively elongate element on each the ventral gill skeleton intact. Five pairs of side, slightly directed posteriorly. A small gill arches are present in addition to the pseu- separate cartilage located along the internal dohyoid arch (also observed in FMNH PF margin of the hyomandibula, and posterior to 14567 and FMNH PF 12914). The basihyal the attachment of the hyomandibular liga- cartilage (bh) in ²Asterotrygon is a very ment, is present in numerous species of Uro- small and separate element, slightly wider lophus (observed in many specimens) and than long, and similar to the basihyal of Uro- pelagic stingrays (Garman, 1913). This car- batis jamaicensis (AMNH 30385), U. halleri tilage (termed here the secondary hyoman- (®g. 38B), and Dasyatis americana (AMNH dibular cartilage), however, is considered 30607), but stouter than the basihyal of gym- functionally isolated and not homologous to nurids, and not fragmented as in potamotry- the angular cartilages of potamotrygonids gonids. It is situated just posterior to the low- (Lovejoy, 1996; McEachran et al., 1996), but er jaw symphysis (®g. 20). Unfortunately, the this remains to be elucidated. No such ele- pseudohyoid arch is mostly obscured in ment is visible in specimens of ²Asterotry- FMNH PF 15180, but the right ventral pseu- gon. dohyoid (vph) is present as a very slender, Stingrays typically possess ®ve gill arches prismatically calci®ed element that still artic- (although there are six in Hexatrygon)inad- ulates with its associated pseudohyoid rays dition to the pseudohyoid arch located be- (which are posteriorly directed). Lateral to tween the hyomandibular and ®rst gill arch. the basihyal are the ®rst hypobranchials Elements of the ventral gill arch skeleton (hb1), which in FMNH PF 15180 are excep- have been studied in more detail (Miyake, tionally well preserved (®g. 20). The ®rst hy- 1988; Miyake and McEachran, 1991), and pobranchials are shaped as in potamotrygon- stingrays characteristically possess an en- ids (especially Potamotrygon magdalenae larged medial plate (comprising posteriorly [AMNH 55620]; see ®g. 34B), urolophids, the basibranchial copula, considered to be the urotrygonids (®g. 33B), and other benthic result of ontogenetic fusion of hypobranchial stingrays, that is, tapering posteriorly only and basibranchial components; bb), ankylosis slightly from their articulation with the ba- or slight overlapping of the proximal seg- sihyal, but not as tapered as in Gymnura (®g. 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

35B). Both hypobranchial 1 elements are al- between the pectoral propterygia in ²Aster- most as long as one of the mandibular car- otrygon. tilages from symphysis to jaw corner. The In specimens FMNH PF 15166 (®g. 2), hypobranchials are obliquely oriented and FMNH PF 15180 (®gs. 12, 20), FMNH PF contact posteriorly the ventral pseudohyoid, 12989 (®g. 4), FMNH PF 14069 (®g. 5), lateral to the anterior extension of the basi- FMNH PF 12914 (®g. 6) and FMNH PF branchial copula. The ®rst hypobranchial el- 14567 (®g. 7) numerous disarticulated gill ements in ²Asterotrygon were not as broad rays (gr) are scattered in the gill chamber, but anteriorly as in Taeniura, or even as much some are preserved in their original positions as in some species of Dasyatis. Posterior to in FMNH PF 15180 (®g. 20) and in FMNH the hypobranchials, and separated from them PF 15166. In Recent stingrays, gill rays sup- by a small gap, is the basibranchial copula porting the gill lamellae are present on the (bb; medial plate). The copula bears a poorly dorsal and ventral pseudohyoid arches and preserved anterior projection, very similar to on cerato- and epibranchials 1±4. In our fos- the condition seen in most benthic stingrays sils it is generally not possible to associate (Urobatis, Urolophus, Urotrygon, Taeniura, the slender and fragmented gill rays with any potamotrygonids), but absent in Plesiobatis particular arch, even though they occur and pelagic stingrays (Miyake and Mc- throughout the entire gill chamber (but in Eachran, 1991). This projection in ²Astero- FMNH PF 15180 the ventral pseudohyoid is trygon appears to be more anteriorly extend- associated with corresponding gill rays; ®g. ed than in Gymnura as well. The full poste- 20). The gill chamber is compacted antero- rior extent of the basibranchial copula is dif- posteriorly in FMNH PF 12989 and in ®cult to determine, but in stingrays it usually FMNH PF 14069. A large gap is visible be- extends as a very slender projection to the tween the basihyal and basibranchial anterior level of the scapulocoracoid. In ²Asterotry- extension in FMNH PF 14069, even though gon, the posterior extension of the copula is the basihyal is not well preserved and its precise shape cannot be discerned (our identi®- dif®cult to observe, but is present in FMNH cation of a patch of prismatic calci®cation PF 15180 as a sightly elevated, slender ridge just posterior to the mandibular cartilages as of prismatic cartilage, similar to the basi- the basihyal in this specimen is tentative). branchial copula present in Trygonoptera, There are no signs of extrabranchial cartilag- Urobatis, and potamotrygonids, and not as es in ²Asterotrygon (these are formed from short as in gymnurids. The extent to which the fused distal tips of gill rays). The gap that ceratobranchials are fused to each other (if at separates the mandibular cartilages from the all) and to the ventral pseudohyoid bar can- anterior limit of the basibranchial (which it- not be determined in the fossils. self is usually not visible) contains disartic- The articulation between the last pair of ulated elements of the visceral arches and ceratobranchials and the scapulocoracoid can synarcual cartilage in some of the fossils. be seen in FMNH PF 12989 and FMNH PF 15180, where the ceratobranchials appear SYNARCUAL AND VERTEBRAL SKELETON relatively straight and narrow. We consider these articulating elements to be ceratobran- Stingrays have two synarcual cartilages, chials and not epibranchials, as both speci- an anterior cervicothoracic synarcual (ct syn) mens are ventrally preserved and ceratobran- and a posterior thoracolumbar synarcual (tl chials are typically more slender and straight syn) (Compagno, 1973, 1977). The thoraco- compared to the slightly curved epibranchi- lumbar synarcual is unique to stingrays. The als of nonmyliobatid stingrays. The pre- anterior synarcual articulates with the occip- served posterior portion of both ceratobran- ital region of the neurocranium and extends chial 5 elements are prismatically calci®ed posteriorly to meet the scapular processes of and articulate with the coracoid lateral to its the shoulder girdle. It is prismatically calci- connection with the synarcual cartilage, at ®ed and consists of a dorsal and ventral com- about one-third of synarcual length. The gill ponent, both formed ontogenetically by fu- basket occupied almost the entire gill cavity sion of different vertebral (basidorsal and 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 51 basiventral) elements (Miyake, 1988). The well preserved, the thoracolumbar synarcual anterior synarcual is relatively well devel- is about ®ve times the width of individual oped and prominent in stingrays. In ²Aster- centra at its origin, where it attains its great- otrygon the cervicothoracic synarcual is pre- est lateral dimensions, and is prismatically served in many specimens (e.g., ®gs. 6, 7, calci®ed (®g. 22). Observation of the second 22). The medial crest (mdc) of the synarcual synarcual requires the fossil specimen to is a dorsal ridge that runs anteroposteriorly have a dorsoventrally preserved tail region along almost its entire length and is visible posterior to the shoulder girdle (e.g., FMNH in dorsally exposed specimens, especially PF 12989); in other words, the vertebral col- FMNH PF 14567 (®g. 22B) and FMNH PF umn cannot be dislocated and rest on its side, 12914, where the ridge is markedly elevated. as is commonly the case. FMNH PF 14069 One ventrally exposed specimen (FMNH PF has the column preserved laterally, and only 12989) clearly displays a string of vertebrae the very anterior portion of the second syn- running the entire length of the preserved arcual is present. synarcual (13 partially fused vertebrae along Vertebral centra do not articulate with the synarcual midline are present). In most pleural ribs in ²Asterotrygon, and ribs asso- stingrays, vertebral centra occupy only the ciated with basiventrals are absent, as in all posterior one-half to one-third of synarcual stingrays. Numbers of vertebral centra pos- length. Lateral stays (ls; supposedly formed terior to the ®rst synarcual are given in tables by fused pleural ribs and basiventrals) are not 2 and 3. It is more dif®cult to determine tran- visible in any specimen, even though they sitions between mono- and diplospondyly in are prominent components of cervicothoracic our fossils than in extant stingrays, because synarcuals of stingrays. Dorsally, the poste- myomeres are not preserved and individual rior segment of the ®rst synarcual is com- centra may be displaced. In stingrays, this posed of the secondarily fused suprascapulae transition usually occurs at the level of the that articulate with the ascending lateral pelvic girdle or just posterior to it, and that scapular processes of the shoulder girdle appears to be the case in NHM P 61244. In (FMNH PF 12989, FMNH PF 14567). The FMNH PF 14069, where the vertebral centra posterior, suprascapular portion of the syn- are exceptionally well preserved, the area of arcual appears rectangular in outline in al- transition is not entirely clear but appears to most every specimen, even those that are be posterior to the pelvic girdle, close to the ventrally preserved. The suprascapular por- posterior edge of the pelvic ®n. Specimens tion of the synarcual in ²Asterotrygon is ro- FMNH PF 15166 (holotype; ®g. 2), AMNH bust compared to extant nonmyliobatid sting- P 11557 (paratype; ®g. 3), FMNH PF 14069 rays, where it appears to be more slender (paratype; ®g. 5), FMNH PF 12914 (®g. 6), (Urotrygon, Urobatis, Potamotrygon). The and FMNH PF 14567 (®g. 7) have enlarged, overall width of the synarcual anterior to the laterally compressed neural spines (ns) or suprascapulae is similar to extant benthic arches from just posterior to the scapulocor- taxa as seen in NHM P 61244, where the acoid to close to the caudal stings. The neural synarcual width is just less than neurocranial arches are obliquely oriented posteriorly in width at the occipital region. Spinal nerve relation to the centra and are closely situated foramina cannot be identi®ed and counted in to each other. Ventral arches are also present our fossils. in extant stingrays, but are much smaller in The posterior or thoracolumbar synarcual comparison to the neural arches at least until is less pronounced than the cervicothoracic the level of caudal stings, and this was prob- synarcual, and is a relatively simple struc- ably the case in ²Asterotrygon as well. In- ture. This synarcual tapers posteriorly, end- dividual basidorsals or interdorsals cannot be ing at about midway between the scapulo- discerned in the fossils. Individual vertebrae coracoid and pelvic girdle, and has unfused are spool-shaped in lateral view and more individual vertebral centra along its entire elongated in the caudal region, even though length. The second synarcual in ²Asterotry- they are much smaller compared to centra gon is very similar to extant nonmyliobatid close to the second synarcual. Faint rings of stingrays. In FMNH PF 12989, where it is calci®cation are present peripherally within 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 22. Scapular region depicting scapulocoracoid and pectoral basals of ²Asterotrygon maloneyi, n.gen., n.sp. A. FMNH PF 12989, ventrally exposed (paratype; entire specimen in ®g. 4); note small dermal denticles distributed over entire surface. B. FMNH PF 14567, dorsally exposed (entire specimen in ®g. 7); mesopterygium missing from left side. Anterior to top. the centra and can be observed in specimens has discrete vertebrae extending to the tip of in which the centra are preserved, exposing the tail (we have con®rmed this in four spe- their anterior or posterior surfaces (FMNH cies of Gymnura so far), contrary to the ®nd- PF 14098, USNM 2028). ings of Lovejoy (1996) and McEachran et al. ²Asterotrygon, unlike almost all extant (1996). stingrays that lack a caudal ®n, has discrete vertebral centra continuing to the posterior APPENDICULAR SKELETON tip of the tail, which can be observed in juveniles, small specimens, and adults. The un- The entire shoulder girdle (scapulocora- segmented cartilaginous rod (cr), a noto- coid) consists of a single, somewhat ¯attened chordal extension that continues caudally be- U-shaped structure in ²Asterotrygon. The yond the caudal stings (instead of separate scapulocoracoid (sc) is composed of a trans- vertebral centra), is therefore absent in our verse ventral coracoid bar (co) and lateral, fossils (cf. ®gs. 24, 26 to ®gs. 37C, D, and dorsally projecting scapular processes (scp) 40E). This rod is present in most stingray that articulate medially with the suprascapu- genera and is absent in taxa that have a cau- lae (ssc) which are fused to the posterior seg- dal ®n (urotrygonids and urolophids), Ple- ment of the cervicothoracic synarcual. The siobatis (Nishida, 1990), Hexatrygon (Heem- coracoid bar is best preserved in ventrally stra and Smith, 1980), as well as in all other exposed specimens (FMNH PF 15180 [®g. nonmyliobatiform batoids. However, this 20], FMNH PF 14069), but the entire shoul- condition may vary slightly among species, der girdle is easily recognized as a prominent as Dasyatis annotatua has individual verte- transverse structure at mid disc in all speci- brae extending well beyond the caudal sting mens. The anteroposterior length of the cor- (CSIRO T 694, CSIRO T 697) and Gymnura acoid bar varies slightly among specimens, 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 53

Fig. 22. Continued. but how much of this variation is due to its The scapulocoracoid laterally joins the in- orientation during preservation is unknown. ternal skeleton of the pectoral ®ns or disc, In FMNH PF 14069 the coracoid resembles which is composed of basal cartilages (pro-, that of Recent Potamotrygon and other ben- meso-, and metapterygium) articulating with thic genera in being slender transversally, but numerous elongated and subdivided radial unlike Potamotrygon the anterior border of elements that extend to the outer margins of the coracoid is relatively straight across and the disc (plesodic condition). The proptery- not slightly concave. The scapular processes gium (pro) extends for slightly over one-half are best preserved in FMNH PF 14567 (®g. of disc length. In both ²Asterotrygon and ex- 22B) and FMNH PF 12914 (®g. 6), where tant stingrays, the anteriormost portion of the they appear as large subtriangular pieces ar- propterygium becomes subdivided into three ticulating with the synarcual by means of or more segments anterior to the level of the prominent articular surfaces. This articula- nasal capsules (even in Mobula [AMNH tion in stingrays is of the ball-and-socket 15319], which bears an elongated proptery- type (Compagno, 1973, 1977), and we can gium that is anteriorly expanded to support infer on the basis of its position and appear- the cephalic lobes, but in which there is also ance in dorsoventral view that this was the a small anterior segment). In some extant case in ²Asterotrygon as well. The scapular nonmyliobatid stingrays (Urolophus, Try- processes are obliquely positioned in relation gonoptera, Dasyatis, Himatura) the propter- to the synarcual cartilage, as in modern sting- ygium becomes subdivided at about midnas- rays (Urotrygon, Urolophus). The foramen al capsule length (cf. Lovejoy, 1996), but in present on the dorsal component of the scap- ²Asterotrygon this division appears to be in ular process of some stingray taxa (``af'' fe- front of the anterior cranial margin, even nestra of Miyake, 1988; ``foramen of the though this is dif®cult to determine in the scapular process'' of Lovejoy, 1996) cannot fossils. The anterior segments of the propter- be identi®ed in the fossil material. ygia bend medially toward the midline, leav- 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 ing a prominent gap that was occupied by FMNH PF 14567 (®g. 22B), and in the left the small rostral cartilage (a rostral piece was side of paratype AMNH P 11557, as in many presumably presentÐpossible vestiges of it nonmyliobatid stingrays, but in FMNH PF are found in NHM P 61244, see above). The 12989 (®g. 22A) the mesopterygium is more anteriormost segment of the propterygium elongate, projecting anteriorly somewhat (or the ®rst radial propterygial element an- obliquely to the propterygium. This discrep- teriorly) is bi®d, articulating with two rows ancy is probably due to incomplete preser- of radials as in extant nonmyliobatid sting- vation in the latter specimen. The scapular rays. The propterygium bears an elongated process is situated at close to the posterior ridge and sulcus for the insertion of the pec- two-thirds of its length. Approximately 13 toral ®n muscles along almost its entire mesopterygial radials are present in FMNH length. PF 12989 (counted on both sides; table 2). In ²Asterotrygon, the posterior section of The metapterygium (met) is elongated and the propterygium contacts the scapulocora- curved, tapering posteriorly and terminating coid, as in various extant benthic taxa (Das- close to the level of the pelvic girdle. In ²As- yatis, Urolophus, Potamotrygon), by means terotrygon, the metapterygium appears as a of an elongated articular surface with two single piece, unlike many nonmyliobatid distinct condyles (procondyles). The anterior stingrays where it can become subdivided procondyle is at the anterolateral corner of posteriorly into smaller segments that also the shoulder girdle, and the posterior one is articulate with radials. The metapterygium close to the mesocondyle (the articular sur- also bears a pronounced groove for the in- face of the mesopterygium). The pro- and sertion of the pectoral ®n musculature; it is mesocondyles are clearly separated from similar to the propterygium but is not as each other in stingrays (Miyake, 1988), and elongate or as wide. this is the case in ²Asterotrygon (®g. 22A). The radial elements articulating with basal The lateral aspect of the scapulocoracoid, cartilages are identical in structure (numbers however, is not visible in our two-dimen- of radials for each basal are shown in tables sional fossils, but the articulation between 2 and 3). All radials are segmented from the the propterygium and scapulocoracoid can be basal cartilages to the outer disc margin. In discerned from a dorsoventral perspective, FMNH PF 12989, FMNH PF 12914, and appearing very similar to Recent nonmyliob- FMNH PF 14567, where radial segments are atid stingrays. The medial or internal surface clearly preserved, there are at least 15 radial of the posterior tip of the propterygium is segments, but perhaps one or two additional slightly concave where it articulates with the segments were present distally. In the holo- scapular process (FMNH PF 12989), con- type (®g. 2), 17 radial segments are present tacting it in at least two different places (pre- at the level of greatest disc width. The indi- sumably at the procondyles) but possibly vidual segments of the radials are relatively throughout more of its length, as is common short and reduce slightly in length in the di- in many extant benthic stingrays. The lateral rection of the outer disc margin. In speci- aspect of the scapular process is relatively mens FMNH PF 15166 and FMNH PF straight, and consequently the propterygium 12914, radials of the ninth row (counting is not separated from the mesopterygium by from basals to outer margin of disc) are bi®d, a small process in ²Asterotrygon. but in FMNH PF 12989 and FMNH PF The mesopterygium (mes) is more dif®cult 14567 it is the eighth radial segment that bi- to observe in our fossils due to scattering of furcates. Therefore, there are always twice as its prismatic calci®cation. It is slightly many radials at the outer disc margin than wedge-shaped and subtriangular in the ho- articulating with the basal cartilages. The de- lotype FMNH PF 15166 (right side), in rived condition of Gymnura and myliobatids

→ Fig. 23. Pelvic girdle of ²Asterotrygon maloneyi, n.gen., n.sp. A. Male paratype (FMNH PF 12989) in ventral view (entire specimen in ®g. 4), depicting claspers situated dorsal to tail base; note axial 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 55

cartilage and impressions of terminal cartilages. B. Paratype (AMNH P 11557), also in ventral view (complete specimen in ®g. 3). Anterior to top. 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

(Nishida, 1990) in having laterally expanded 14567 and FMNH PF 14097, however, the radial elements that articulate with adjacent processes are more triangular and located radials (®g. 39A, B) is not present in ²Aster- where the posterior concavity of the pelvic otrygon. girdle curves inward, as in Urotrygon and The puboischiadic bar (pib; pelvic girdle), Urobatis. Other than the ischial processes, being broad and dorsoventrally ¯attened, is there are no distinct features of the posterior well preserved in most specimens, but more margin of the girdle, such as median post- so in FMNH PF 15166, FMNH PF 15180 pelvic processes described and ®gured by (right side only), PF 12989 (®g. 23A), and Nishida (1990: ®g. 36) for platyrhinids. PF 12990. The girdle is stoutest at its cor- (Nishida's ``post-pelvic processes'' [also ners, where the basipterygia (bp) originate McEachran et al., 1996] seem to amount to and extend posteriorly, tapering to a variable irregularities in the posterior surface of the extent. Its anterior surface is roughly straight puboischiadic bar, and not to distinct projec- in most specimens, and is much less arched tions. This is corroborated by the apparent in comparison to the slightly concave pos- absence of any distinct process from Platyr- terior surface (FMNH PF 12989). In other hina sinensis [AMNH 26413, AMNH 44055] specimens (FMNH PF 15166, FMNH PF and from Platyrhinoidis triseriata [OSU 40]; 14567) the pelvic girdle is slightly arched at Carvalho, in press.) The iliac process (ilp), both anterior and posterior margins. Minor which also projects posteriorly and medially distortion and preservational differences may from the corners of the puboischiadic bar in account for the observed variation among many stingrays and more basal batoids, may specimens. be present in our fossil specimens. However, The pelvic girdle resembles that of many because this process may overlie or be stingrays in general aspects, excluding pota- aligned with the basipterygium to some ex- motrygonids. It is arched anteriorly but not tent, its identi®cation is dif®cult in the fossils as much as in Gymnura (®g. 36C, D) or pe- (FMNH PF 14097). The iliac process is pre- lagic stingrays (Mobula). It is dif®cult to de- served in FMNH PF 12989 (®g. 23A), where termine if a median prepelvic projection is it projects posteriorly in an oblique fashion present due to crushing from vertebrae, but from the pelvic girdle. It is elongated, more if one was present it would have been a or less straight, and tapers posteriorly (ex- short, triangular process similar to extant tending to at least the sixth radial element in stingrays other than potamotrygonids (Uro- FMNH PF 12989). The iliac process of many trygon chilensis [®g. 36E], Taeniura lymma extant stingrays may be slightly curved to [®g. 36A, B], Urobatis jamaicensis [AMNH project medially (and dorsally), and is not as 30385], Urolophus cruciatus [AMNH tapered. The iliac process in extant stingrays 59890], species of Dasyatis). Also, no pro- is not as elongate as in FMNH PF 12989. nounced lateral prepelvic processes as in Po- The iliac process in ²Asterotrygon may also tamotrygon can be seen in ²Asterotrygon, but have a posterodorsal orientation, but this is the pelvic girdle projects anterolaterally at its not evident in the fossils. Obturator foramina corners in some specimens (e.g., FMNH PF (of), through which pass diazonal nerves, are 12914, FMNH PF 12990), similar to the con- not well preserved, but can be observed in dition in Urotrygon, Plesiobatis, and other FMNH PF 15180 (one small foramen on stingrays (Nishida, 1990). right side; ®g. 12). There is some evidence of small ischial The pelvic ®n is also plesodic; the ®rst processes (isp) (Miyake, 1988) on the inner pelvic radial element is characteristically en- corners of the posterior margin of the pelvic larged (efr), as in many stingray groups, and girdle in ²Asterotrygon (ischial processes are articulates with the lateral aspect of the pu- present in most stingrays, and are particularly boischiadic bar. Even though it has been well developed in potamotrygonids; ®g. called the pelvic propterygium (Nishida, 16B). In FMNH PF 12989 (right side; ®g. 1990), this element should not be considered 23A), FMNH PF 12990, FMNH PF 14069, the serial homolog of the pectoral proptery- and NHM P 61244, these processes appear gium; it is formed by ontogenetic fusion of as ill-de®ned protuberances. In FMNH PF radials. The numbers of pelvic radials are 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 57 given in tables 2 and 3, and they fall within slightly overlaps the caudal stings. Presum- the ranges observed in many extant sting- ably, the overlapping is due to the posterior rays, increasing slightly in relation to speci- lobe being free from the dorsal surface of the men size (cf. Nishida, 1990: 58, table 3). tail, with the ®n in¯ecting a short distance Each pelvic radial in FMNH PF 12989 has anteriorly before inserting onto the dorsal tail at least three segments (four in FMNH PF surface. The ®n is relatively small and round- 14567), the most basal segment being much ed in outline. Small hooklike denticles are more elongate than the more distal segments. present over most of the ®n in most speci- The most distal segment is bifurcated. mens (FMNH PF 15166, FMNH PF 12914, Claspers are present and well developed in AMNH P 11557), but in both FMNH PF only one specimen (®gs. 4, 23A). Prismatic 14069 and FMNH PF 15181 denticles are calci®cation is present over the entire axial more apparent on the anterior margin or cartilage (ax), further corroborating that this ridge of the ®n. The presence of hooklike specimen is sexually mature. The claspers denticles covering the dorsal ®n or at least are relatively long and straight, with pre- its dorsal ridge is a unique feature for ²As- served axial cartilages extending along most terotrygon, not occurring in other stingrays of its length. The entire length of the claspers where a dorsal ®n is present. is roughly just under one-third of tail length The dorsal ®n is clearly plesodic in FMNH as measured from posterior margins of pel- PF 14069 and FMNH PF 15181, with inter- vics. No beta cartilages can be seen, but this nal skeletal support primarily composed of may be preservational. On the right clasper, slender, elongated radial elements, which ra- running more or less parallel to the axial car- diate from the ®n base to the outer ®n margin tilage, there is a slightly curved piece that (less evident in FMNH PF 14567). In FMNH may be the beta cartilage (terminology from PF 15181 the radials bifurcate distally (less Miyake, 1988). The axial cartilage bends so than FMNH PF 14069). Basal compo- outward slightly at its posterior tip, and the nents of the dorsal ®n endoskeleton also ap- entire clasper is spoon-shaped. Discrete car- pear to be present in FMNH PF 14069, but tilages of the terminal clasper components these are not present in FMNH PF 15181. are not readily observed, but more than one These elements are slightly wider than the is present. The largest element apparently is more distal radials and are organized into a the ventral covering piece (ventral terminal poorly de®ned row from the ®n base to about cartilage; tc), but only its impression remains one-third of its height. The radial elements in the fossil. Anterior to the ventral terminal are very prominent in FMNH PF 14069, ex- cartilage are impressions of the dorsal and/or tending to the outer ®n margin. In this re- ventral marginal cartilages. spect the dorsal ®n of ²Asterotrygon resem- The dorsal ®n is present in most specimens bles the dorsal ®n of Urolophus ¯avomosai- of ²Asterotrygon in which the distal portion cus (CSIRO 718.22) and Aetoplatea zonura of the tail is laterally preserved to some de- (Nishida, 1990; ®g. 41A here), and even the gree (®g. 24). The dorsal ®n is more prom- dorsal ®ns of rajids (Nishida, 1990). The dor- inent and well preserved in one female para- sal ®ns of Recent myliobatids have enlarged type (FMNH PF 14069; ®g. 24B) and in basal elements and occupy a greater portion specimen FMNH PF 15181 (®g. 24F). NHM of the dorsal ®n base compared to FMNH PF 61244, a small neonate, also has a weak im- 14069 (Aetomylaeus maculatus [AMNH pression of the dorsal ®n present anterior to 32500], Aetobatus narinari [®g. 41E]; My- caudal stings. The holotype has a small dor- liobatis, Rhinoptera, Mobula, Manta; Nishi- sal ®n preserved, covered by denticles, but da, 1990). The ``basal elements'' in FMNH no internal structure is apparent (®g. 24E). In PF 14069 are probably not enlarged basal specimens FMNH PF 14069 and FMNH PF cartilages as in myliobatids, but just small 15181 the tail is markedly preserved on its radial segments horizontally arranged (as in side, accounting for the excellent exposure of Aetoplatea zonura; ®g. 41A). The height of the dorsal ®n (also FMNH PF 14567; ®g. the dorsal ®n is just greater than tail height 24D). The dorsal ®n is located just anterior at the level of the dorsal ®n in FMNH PF to the caudal stings, but its posterior lobe 14069 and FMNH PF 14567, but in AMNH 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 24. Dorsal ®n and caudal stings of ²Asterotrygon maloneyi, n.gen., n.sp. A. Paratype AMNH P 11557 (specimen in ®g. 3). B. Paratype FMNH PF 14069 (specimen in ®g. 5). C. FMNH PF 12914 (specimen in ®g. 6). D. FMNH PF 14567 (specimen in ®g. 7). E. Holotype FMNH PF 15166 (specimen in ®g. 2; arrowhead showing poorly preserved dorsal ®n). F. FMNH PF 15181 (specimen is tail only). Morphology of internal skeletal support of dorsal ®n is more clearly preserved in panels B and F. Anterior to left. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 59

Fig. 24. Continued.

P 11557 and FMNH PF 12914 the dorsal ®n mentary radials according to Nishida, 1990), is not as tall as the tail at its insertion (®g. reminiscent of dorsal and ventral tail-folds of 24). This may be due to different orientation some Potamotrygon species. Rudimentary of the ®n and tail in the specimens. Although radials, both dorsal and ventral, are also pre- the dorsal ®n is not apparent in all specimens sent in the holotype (®g. 24E), FMNH PF (usually not in those in which the tail is pre- 12989, and to a lesser extent in FMNH PF served dorsoventrally), a small cluster of 14069 (not visible in ®g. 24B). In FMNH PF denticles is sometimes observed anterior to 12989, rudimentary radials occur only on the caudal stings, indicating that the ®n was one side of the vertebral column, but it is not originally present (e.g., FMNH PF 12989). possible to discern which side, as the tail is Specimen FMNH PF 12990 is unique in preserved laterally. Low tail-folds were having an impression of a putative caudal ®n therefore present in ²Asterotrygon, and it posterior to the caudal sting (®g. 10A). The probably extended to the posterior tip of the ®n is incompletely preserved but was appar- tail and originated shortly after the caudal ently rounded dorsally and posteriorly, with stings. The height of the tail-fold both dor- a more prominent dorsal lobe and a smaller sally and ventrally was slightly greater than ventral lobe. If this interpretation is correct, the height of the tail where the tail-folds then the caudal ®n is similar to some species originated. of Urobatis, Urolophus, and Trygonoptera in being taller than long. Alternatively, the cau- DERMAL SKELETON dal ®n in FMNH PF 12990 may represent incompletely preserved tail-folds, as only a DERMAL DENTICLES:In²Asterotrygon, the small portion of the tail posterior to the cau- dermal skeleton is composed of numerous dal sting is present. We cannot be sure be- small denticles scattered over most of the cause the posterior extent of this ``®n'' is not dorsal surface of the disc, dorsal ®n, and tail known. The tail region posterior to the cau- region, as well as the lateral aspects of the dal stings in FMNH PF 14567 (®gs. 7, 24D) tail. These small denticles are interspersed has small dorsal and ventral elements (rudi- with larger denticles (or spines) primarily 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 over dorsal middisc and midtail regions. to ®ve or more corners, similar to many po- Denticles and spines are preserved in ventral, tamotrygonids (e.g. Castex, 1967), Taeniura dorsal, and lateral view, and both ventrally meyeni (Smith, 1952), and stingrays in gen- and dorsally preserved specimens display eral; the stellate denticle bases (db) are even denticles in lateral aspect. The smaller den- comparable to some of the denticles of the ticles and larger spines are similar in mor- astraspid Pycnaspis from the Ordovician phology, differing only in size. As in other Harding Sandstone of Colorado (érvig, myliobatiforms, the denticles lack peduncles 1958; Jarvik, 1980). In ²Asterotrygon, the between the crowns and basal plates. Ter- corners of the stellate denticle bases are basal minology for dermal denticles is mostly from ridges that are continuous with the longitu- Deynat and SeÂret (1996). dinal crown ridges that project toward the Specimens FMNH PF 12989 (®g. 4), apex of the denticle (da), similar to the den- FMNH PF 14069 (®g. 5), FMNH PF 12914 ticles of Manta described by Gohar and Bay- (®g. 6), and AMNH P 11557 (®gs. 3, 25A) oumi (1959: 198, text-®g. 4). The small den- have the densest shagreen. Small denticles ticles generally have very slender crowns are easily observed in most specimens, with- compared to the diameter of the basal plates. out the aid of magni®cation, in the region The larger spines are similar to the smaller lateral to the vertebral column between the denticles and vary in size as judged from the pectoral and pelvic girdles (offering a con- diameter of their basal plates and the height venient way to distinguish ²Asterotrygon of their crowns in laterally preserved denti- from ²Heliobatis). The small dorsal denticles cles. The basal plates of these spines can be are smaller than individual teeth, are present star- or asterisk-shaped, and many larger anterior to the snout, and cover the entire spines have irregularly shaped bases, but dorsal disc region almost to disc margins some of this variation is preservational, as (®gs. 4, 5). In FMNH PF 12989, an adult many of the spines are obliquely embedded male specimen (238 mm TL), the denticula- in the matrix. The longitudinal basal ridges tion is particularly dense, but this is not a determine the shapes of the spine bases, as secondary sexual feature, as FMNH PF in the smaller denticles. The bases are best 12914 (female, 204 mm TL) is also very observed in ventrally exposed specimens densely covered, particularly on disc margins (e.g., FMNH PF 15166, FMNH PF 12989, (cf. ®gs. 4 and 6). Paratype AMNH P 11557 FMNH PF 14069; see also ®g. 25). The has a particularly dense denticulation, espe- crowns are posteriorly curved, with relatively cially toward lateral disc surfaces (®g. 25A, sharp apexes on some denticles, while others B). Small denticles are also present anterior are more worn. In FMNH PF 12989, where to the neurocranium in a juvenile specimen, some spines are laterally preserved on the but these are barely exposed (FMNH PF tail region, the width of the crowns at mid- 15180; ®g. 20). The smaller, more numerous height is about equal to that of the basal denticles surround larger individual denticles plates, meaning that the enlarged spines did (spines) over mid disc and over much of the not have very thin crowns, at least on the dorsal tail region. The holotype (®g. 2) has dorsal tail surface, as did the smaller denti- small denticles scattered on its disc (not as cles. The adult male specimen (FMNH PF much as in the paratypes), as well as larger 12989) has enlarged nuchal or spiracular denticles over the neurocranium and tail re- spines partially obliterating the postorbital gion. Small denticles continue posteriorly to processes and hyomandibulae. This may be the very tip of the tail and cover the dorsal an indication of secondary sexual dimor- ®n in the holotype and other specimens (e.g., phism, as enlarged spines as such are not pre- FMNH PF 12914). The shape of the smaller sent in FMNH PF 14069 (female) or the larg- denticles varies according to how worn or er paratype (AMNH P 11557, also female), eroded they are, but generally the denticles but this is contradicted by FMNH PF 12914 are posteriorly curved, slender and hooklike, (female), which also has enlarged spines over and may be very apically pointed (cf. ®gs. lateral and posterior portions of the neuro- 25D and 38E). The basal plates of the den- cranium at the level of postorbital processes ticles are star-shaped (®g. 25C, D), with three (as does the holotype, FMNH PF 15166, a 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 61 large female). The enlarged spines or tuber- tremity (Bigelow and Schroeder, 1953). Cau- cles occur in more than one row over the tail, dal stings are hypertrophied dermal denticles as evidenced by FMNH PF 12989, FMNH (Daniel, 1934) and occur in all extant sting- PF 12914, FMNH PF 14069, FMNH PF ray genera except Urogymnus, and in almost 14567, and AMNH P 11557, but whether all stingray species; their presence is indica- these were organized in strict rows cannot be tive of stingray monophyly. The caudal determined. Enlarged spines are also present stings are enveloped in an epidermal sheath over the middisc region at least at the level in extant stingrays. This sheath contains of the pelvic girdle (FMNH PF 12989), but blood vessels ventrally, adjacent to the lon- possibly extending more anteriorly. These gitudinal ridges of the stings, and glandular spines are dif®cult to assess in many speci- epithelium towards its periphery, which is re- mens because they may be obliterated or dis- sponsible for the production of venom (Hal- located by the vertebral column over the stead and Modglin, 1950; Halstead, 1970). middisc region. FMNH PF 12914 has spines The caudal sting is already formed by par- over much of its midline, and these are turition in extant stingrays (although present, slightly smaller than the nuchal spines over it is slightly less acute in embryos), and this the neurocranium. was the case in ²Asterotrygon as the fetus Specimens FMNH PF 14567 (®g. 7) and preserved inside the holotype has a small (13 FMNH PF 14097 have less or no denticula- mm in length) sting that is fully formed (®g. tion over lateral disc margins and the anterior 13). snout region. Denticles are present over the The caudal stings of ²Asterotrygon are middisc region and posteriorly all over the easily observed in all specimens except tail as in all other specimens of ²Asterotry- USNM 2028; they occur at more or less the gon. These specimens are from both F-1 and posterior one-third of the tail on its dorsal F-2 localities, and both are adult females. surface (®gs. 24, 26). Usually more than one They may represent a distinct variety or spe- sting is present, but FMNH PF 15181 (®g. cies of stingray, but are here included in ²As- 24F), FMNH PF 15166 (®g. 24E), and terotrygon maloneyi (see above, p. 29). FMNH PF 12914 (®g. 24C) have four, which FMNH PF 14097 has small denticles scat- is uncommon even among extant stingrays. tered on the slab in which it is preserved, The lateral serrations along one side of the lateral to the tail, which may represent den- sting are separated by a space almost as wide ticles originally present on the dorsal disc or as the caudal sting itself in FMNH PF 12989, tail region that were dislodged postmortem. but in FMNH PF 14069 the lateral serrations There is no indication in any specimen of are more closely packed together. In FMNH ²Asterotrygon of denticles with enormously PF 14069, the lateral margins of the caudal enlarged basal plates (tubercles), or of der- sting super®cially appear to be smooth, but mal disc bucklers formed by fusion of en- under greater magni®cation (25ϫ) lateral ser- larged denticle bases, as in some extant spe- rations are visible at least posteriorly on the cies of Potamotrygon (P. constellata and P. caudal sting, and they probably occurred motoro) and isolated potamotrygonid fossils from at least midsting length. The serrations from South American Miocene deposits are very small, numerous, and directed al- (Deynat and Brito, 1994; Lundberg, 1998; most transversely to the main axis of the cau- Carvalho, Lundberg and Maisey, in prep.). dal sting in FMNH PF 14069 (®g. 24B), but Also absent are denticles with symmetrically in FMNH PF 12989 the serrations are more arranged crown dichotomies, as in Potamo- slanted posteriorly and are fewer in number trygon yepezi, P. motoro, P. falkneri, and P. (resembling the serrations of the caudal sting schuehmacheri according to our specimens of Plesiotrygon iwamae). In the holotype (see also Castex, 1967; Deynat and SeÂret, (FMNH PF 15166; ®g. 24E), the stings are 1996). somewhat slanted as well, but not as much CAUDAL STINGS: Caudal stings (cst) are as in the male paratype (FMNH PF 12989), elongated, dorsoventrally ¯attened, tapering and the tip of the dorsalmost sting is broken dermal structures, with posteriorly directed off and vertically lodged in the tail extremity. serrations on both sides and a very acute ex- The stings are best preserved in FMNH PF 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 25. Denticles of ²Asterotrygon maloneyi, n.gen., n.sp. A. Anterior snout region of paratype AMNH P 11557 in ventral view (specimen in ®g. 3). B. Base of tail region of paratype AMNH P 11557 (ventral view). C. Tail-base area of USNM 2028 (female), in dorsal view (specimen in ®g. 10B). D. Enlarged view of denticles at base of tail of USNM 2028 from panel C (also ®gured in Grande 1984: 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 63

rations of this specimen may be a secondary sexual character. In cross section, the stings are convex and broadly rounded dorsally, without a median longitudinal dorsal groove that is present in extant stingrays (e.g., Dasyatis, Myliobatis; Halstead and Modglin, 1950). Ventrally, the two longitudinal grooves extend almost the entire length of the sting on both sides of the elevated ventral median ridge (these are particularly pronounced in the holotype, where the stings are ventrally exposed). These grooves accommodated the elongated venom-producing glandular epithelium as in living stingrays (Halstead and Modglin, 1950; Halstead, 1970). The ventral median ridge is also rounded, but not as broadly rounded as the dorsal surface. The sting itself is made of bonelike dentine, covered by a thin layer of enamel, and penetrated by numerous canals forming a network within the dentine (va- sodentine). The canals are visible in cross section in FMNH PF 12989, and in this re- gard the caudal stings of ²Asterotrygon are similar to extant stingrays. In FMNH PF 12989, where three stings are present, the dorsalmost sting has a very worn outer surface, with blunt, barely noticeable serrations. The sting underlying the dorsalmost sting has more pronounced and less worn serrations, indicating that ²Asterotrygon replaced stings Fig. 25. Continued. as in extant stingrays, that is, from under- neath, by shedding the dorsalmost sting. The 14567 (®g. 24D), where two stings are pre- underlying stings appear longer because of sent and serrations are clearly visible begin- their more posterior attachment on the tail ning at about midsting and continuing cau- (also in the holotype, FMNH PF 15166). In dally. In FMNH PF 14098 the serrations are the fossils (such as the holotype), the basal shaped as posteriorly curved hooks (®g. end of the sting bears two small knobs that 26D), especially close to apex, and are more help to attach the sting in the dermis as in pronounced compared to other specimens, extant stingrays. resembling the stings of species of Dasyatis, TEETH: The teeth in ²Asterotrygon were Gymnura, and Myliobatis (Bigelow and originally in a quincunx arrangement but are Schroeder, 1953). All specimens appear to now usually scattered between upper and have more numerous, hooklike, and smaller lower jaws, being sometimes displaced as far serrations compared to FMNH PF 12989 (an as the pectoral propterygia. Numerous teeth adult male), and the low-angled, slanting ser- are present but their rows cannot be counted

← 29, ®g. II.7a). Note that in panels A and B only ventral aspect of denticle bases are exposed as specimen is ventrally preserved, while in panels C and D dorsal extremities (apexes) of denticles are visible (specimen is dorsally exposed). Anterior to top. 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 26. Posterior tail region and caudal stings of specimens of ²Asterotrygon, n. gen. (A, C, and D) and ²Heliobatis (B). Note individual vertebrae continuing posteriorly beyond caudal stings to distal extremity of tail. A. FMNH PF 14097 (entire specimen in ®g. 9). B. AMNH P 19665 (specimen in ®g. 28). C. FMNH PF 12914 (reversed; specimen in ®g. 6). D. FMNH PF 14098 (reversed; specimen in ®g. 8). Anterior to bottom. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 65

foramina. There are no foramina present directly on the root lobes. The crown overhangs the root labially and lingually in both sexes. The lingual surface of the crown continues basally to midroot level, while the labial crown surface projects posterodorsally from the top of the root, where a groove separates crown and root (seen in ventral view). It is not possible to establish if heterodonty (monognathic or dig- nathic) is present due to the scattering of teeth, but smaller teeth appear to be very similar morphologically to larger teeth in the same specimen, the main difference being tooth size and development of the cusp (especially in the male), as smaller teeth have less pronounced cusps than larger teeth. The tooth crown in the male specimen (®g. 27A±E) is very triangular, lingually oriented, and slightly concave on its lingual side, where the surface is smooth. In females (®g. 27F), the crown is more blunt, less pointed lingually, and may be subtriangular, rounded, or trapezoidal in some specimens (crown shape may also vary within specimens, e.g., FMNH PF 14069). In the male, the crown is more than twice the height of the root in the Fig. 27. Teeth of ²Asterotrygon maloneyi, larger teeth. The teeth are similar between n.gen., n.sp. Occlusal (dorsal) (A), basal (B), lin- males and females except in differences in gual (C), labial (D), and lateral (E, F) views. Pan- the development of the cusps (more pro- els A±E from FMNH PF 12989 (paratype; adult nounced and triangular in males). male); panel F from FMNH PF 14069 (paratype; female). Teeth are obliquely positioned in order ORDER MYLIOBATIFORMES SENSU to facilitate comparisons (magni®ed 60ϫ). COMPAGNO, 1973 SUPERFAMILY MYLIOBATOIDEA ²Heliobatis Marsh, 1877 in any specimen. Morphologically, the teeth in ²Asterotrygon are very similar to ²He- ²Heliobatis Marsh, 1877 (p. 256; original description, not illustrated; type-species: ²Heliobatis liobatis and other benthic stingrays (Dasy- radians Marsh, 1877, by monotypy; Fossil atis, potamotrygonids). Slight sexual dimor- Butte Member, Green River Formation); phism is evident in the teeth (®g. 27). The Grande, 1980, 1984 (pp. 23±24, ®gs. II.3±II.5, root is bilobed, wide, and not very elongated, II.7c; brief description, photographs). but not as wide as the root lobes of gymnur- ²Xiphotrygon Cope, 1879 (p. 333; original de- ids (where root lobes may be wider than the scription, not illustrated; type-species: ²Xipho- crown in an individual tooth; Cappetta, trygon acutidens Cope, 1879, by monotypy; 1984). There are two nutritive foramina of Fossil Butte Member, Green River Formation); slightly different sizes present in between McGrew and Casilliano, 1975 (pp. 21±22; list- both root lobes (one foramen adjacent to ed, with brief account and illustrated as ®g. 14). Dasyatis: Haseman, 1912 (pp. 98±99; apparently each root lobe; ®g. 27B). These may corre- unaware of the availability of ²Heliobatis spond to the paracentral foramina of Cap- Marsh, 1877); Fowler, 1941 (p. 402; listed both petta (1987). An enlarged central foramen ²Heliobatis and ²Xiphotrygon as synonyms of appears to have been absent, as there is no Dasyatis); Schaeffer and Mangus, 1965 (p. 17); prominent foramen in between both smaller Romer, 1966 (p. 351; ²Heliobatis Marsh, 1877 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

listed as a synonym of Dasyatis Ra®nesque, TABLE 4 1810, along with ²Xiphotrygus [sic], an unjus- Measurements and Counts Conducted on ti®ed emendation of ²Xiphotrygon Cope, 1879); Representative Specimens of Cappetta, 1987 (p. 163; synonymization, with- ²Heliobatis radians out comment, of both ²Heliobatis Marsh, 1877 Values are expressed as mm/percentage of disc width, and ²Xiphotrygon Cope, 1879 with Dasyatis except for total length (TL) and disc length (DL), which Ra®nesque, 1810). are shown in mm. See Measurements and Terminology ²Palaeodasybatis Fowler, 1947 (p. 14; original for abbreviations of parameters. Specimens are female, description, not illustrated; type-species: ²Pa- except FMNH PF 2020. laeodasybatis discus Fowler, 1947, by original designation and monotypy; Fossil Butte Mem- ber, Green River Formation).

DIAGNOSIS (emended): ²Heliobatis is distinguished from ²Asterotrygon by lacking intense denticulation composed of minute, closely packed and hooklike denticles over dorsal disc and tail regions; by lacking a dorsal ®n anterior to caudal stings; by having a more trapezoidal or subtrapezoidal disc (generally more or less rounded to oval in ²As- terotrygon); by a more slender tail at base; and tentatively by lacking angular cartilages between the hyomandibula and lower jar (see also diagnosis provided for ²Asterotrygon,p. 17). ²Heliobatis is distinguished from all other stingrays that lack a caudal ®n, except gymnurids, by presenting individual vertebrae extending beyond caudal stings to the distal tip of tail. From gymnurids, ²Heliob- atis is distinguished by usually having a trapezoidal to subtrapezoidal disc that is longer than wide in most specimens, but never close to twice as wide as long, as is autapomorphic for gymnurids. The lack of a caudal ®n dis- tinguishes ²Heliobatis from urolophids, urotrygonids, Plesiobatis, and Hexatrygon. DESCRIPTION (emended): Stingrays of moderate size (up to about 500 mm in TL, 240 mm in DL), with generally trapezoidal to subtrapezoidal discs, which are less frequent- mens (®g. 29); tail is relatively slender at ly rounded or oval in outline (slightly round- base compared to ²Asterotrygon, closer to ed in AMNH P 19665, and more oval in the condition present in some extant benthic FMNH PF 2020). Disc length is greater than genera (e.g., Urolophus, Urotrygon, Dasy- disc width in most specimens and is mark- atis). Individual vertebrae extend to posterior edly greater in others (FMNH PF 2020; table tip of tail, without a cartilaginous, unseg- 4); in some specimens disc length and width mented rod posterior to caudal stings, as in are about equal (AMNH P 19665). Disc par- extant stingrays that lack a caudal ®n (except tially covers pelvic ®ns, reaching posteriorly Gymnura). Dorsal surface of disc is mostly to about two-thirds of pelvic ®n length in naked, without intense shagreen of smaller most specimens. Greatest disc width is at denticles; smaller denticles over dorsal disc about middisc length, at level of scapulocor- region rarely, if ever, are present (AMNH P acoids. Tail length is about equal to disc 19665 has few very small denticles just an- length but is slightly greater in some speci- terior to puboischiadic bar; ®g. 28); enlarged 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 67

Fig. 28. ²Heliobatis radians, AMNH P 19665, approximately 255 mm TL subadult (?) female, dorsally exposed (from F-2 locality). Anterior to top. denticles or spines are present over midline tal anatomy to ²Asterotrygon, but differs in of dorsal disc surface (in some specimens) a few features. Much of the skeleton in ²He- and dorsally on tail region (in most speci- liobatis is also covered by prismatic calci®- mens); spines over midline of disc do not cation. The neurocranium is similar to that of reach scapular area anteriorly, and are con- ²Asterotrygon in general proportions. It also ®ned to region posterior to middisc; smaller lacks a calci®ed rostrum or rostral deriva- and larger denticles on dorsal tail alternate tives. There is a median constriction (rem- over tail midline in some specimens. Caudal nant of the epiphysial bar) at about the an- stings are located at midtail region in speci- terior one-third of the dorsal fontanelle, as in mens with intact posterior tail regions (®gs. ²Asterotrygon and many extant stingrays. 28, 29), but caudal stings are located farther The dorsal fontanelle is elongated, becoming posteriorly on tail, at about two-thirds of its slender posteriorly, and occupies a large ex- length, in most specimens. Dorsal and caudal tent of neurocranial length as in ²Asterotry- ®ns are lacking in all specimens. gon. In FMNH PF 2020 (®g. 29) the nasal ²Heliobatis presents a very similar skele- apertures are well preserved and demonstrate 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 29. ²Heliobatis radians, FMNH PF 2020, approximately 463 mm TL adult male (terminal cartilages of claspers indicated by arrowhead), ventrally exposed (from F-1 locality). Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 69 that the internasal septum was slender, but tact the scapulocoracoid posteriorly. Anterior probably not as much as in specimen FMNH epibranchial elements articulate with the lat- PF 12989 of ²Asterotrygon (®g. 18). The eral surface of the synarcual in FMNH PF neurocranium is widest at the level of nasal 2020. Branchial rays are numerous and ex- capsules, and these abut the inside margin of tend laterally, almost contacting the propter- the pectoral propterygium. A median inden- ygia. tation or notch is present anteriorly on the The propterygium contacts the pectoral neurocranium between the nasal capsules. A girdle by means of a double articulation, one preorbital process is present, but usually is anterior and one posterior. The propterygium obliterated by the jaws. The antorbital carti- is stout and contains a ridge along most of lages are slender and triangular, tapered posits length for dorsal constrictor muscles. It teriorly and contacting the inner margins of appears to become segmented at the midnas- the propterygia. The orbital regions are rel- al capsule level (FMNH PF 2020). The me- atively long, occupying about two-thirds of sopterygium is small and subtriangular, but neurocranial length. Postorbital processes are its precise morphology is dif®cult to discern. shel¯ike, projecting laterally and obliquely, It does not have a sinuous lateral margin ar- and separated from the supraorbital crest by ticulating directly with radials as in Urolo- a notch (very clear in AMNH P 19665 and phus and Trygonoptera. The metapterygium FMNH PF 2020; ®gs. 28, 29). Therefore the is almost as long as the propterygium, but is supraorbital crest bears a small triangular not as stout. The scapulocoracoid is similar protuberance (the supraorbital process) an- to ²Asterotrygon, with posteromedially di- terior to postorbital processes. The parietal rected scapular processes that articulate with fossa is not evident in most specimens, and the suprascapulae. There is no sign of a dor- the posterior corners of neurocranium are sal fenestra as in certain extant stingrays rectangular in outline. Jugal arches are ab- (e.g., Urotrygon, Taeniura). The coracoid sent. bar is transversely elongate and slender, with The palatoquadrate and Meckel's cartilage a concave anterior margin and straight pos- have rounded outlines in most specimens, es- terior aspect (best preserved in AMNH P pecially at their outer corners, and occupy 19665). Pectoral radials become bifurcated at almost the entire space between pectoral the level of their eighth or ninth segment, do propterygia. The jaw outline varies slightly not contact each other laterally as in Gym- among specimens, as AMNH P 19665 has nura and some myliobatids (Myliobatis and more angular lower jaw corners compared to Aetobatus), and extend posteriorly to almost FMNH PF 2020 (this variation is probably cover the pelvic ®n radials. The pelvic girdle preservational). Neither upper nor lower jaws is slightly arched and has stout corners. No are fused with their antimeres at the midline. prepelvic process is present, and the girdle is Hyomandibulae are well preserved in most similar to that of most benthic stingrays. It is specimens, appearing as relatively stout car- unclear whether ischial processes are present tilages attached to the neurocranium just pos- in ²Heliobatis. Posteriorly directed iliac pro- terior to the postorbital processes. Hyoman- cesses are present on both sides of the girdle dibulae project ventrally and anteriorly to but do not extend as far posteriorly as in ²As- contact jaws through the hyomandibular- terotrygon (FMNH PF 12989). The basipter- Meckelian ligament (AMNH P 856). There ygium is slightly curved posteromedially. is some evidence in FMNH PF 2020 of pris- Pelvic ®n radials become bi®d at their third matic calci®cation (angular cartilages?) be- or fourth segment. Claspers of adult males tween the hyomandibula and Meckel's car- range from about one-third to one-fourth of tilage, as observed in some specimens of tail length and are longer than pelvic girdle ²Asterotrygon, but more specimens are re- width. They have a calci®ed and elongated quired for con®rmation. The hyomandibulae axial cartilage, and at least three smaller ter- are devoid of hyoid rays. The pseudohyoid minal pieces, representing probably the dor- and gill arches are not clearly visible in any sal and ventral terminal cartilages and at least specimen, but the ®fth ceratobranchial (and one marginal cartilage (FMNH PF 2020). perhaps the ®fth epibranchial) elements con- Beta cartilages cannot be discerned. 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

The cervicothoracic synarcual is indistinct morphologically to that of ²Asterotrygon. Neural arches are relatively elongated and ¯attened laterally, somewhat spatulate (as in extant stingrays and ²Asterotrygon), with enlarged neural arches posterior to the thoracolumbar synarcual extending caudally to the midlength of the tail. Vertebral centra are morphologically similar to those of ²Aster- otrygon. Ribs are lacking altogether. The thoracolumbar synarcual extends posteriorly about one-third of distance between shoulder and pelvic girdles, and has individual centra throughout its length. In AMNH P 19665, the tail region is laterally exposed, as indicated by the caudal stings situated dorsal to the caudal vertebrae. If ²Heliobatis had a dorsal ®n, it would be clearly evident in this specimen, which lacks any indication that one was originally present. The enlarged dermal denticles over the midline of the tail and caudal stings are both morphologically similar to those in ²Aster- otrygon. The teeth are closely packed and are as numerous as in many extant stingrays. Morphologically, they are similar to teeth of ²Asterotrygon, with a single, elevated triangular crown in males and a more blunt, trapezoidal crown in females. In AMNH P 19665, remarkably, the teeth have mostly retained their original con®guration, forming partially discrete anteroposterior rows. REMARKS: Specimens of ²Heliobatis radians have been collected from both F-1 and F-2 in great numbers, making it the most abundant fossil stingray known from semi- Fig. 30. ²Heliobatis radians. A. Holotype of complete skeletons, having been featured in ²Heliobatis radians Marsh, 1877 (YPM 528); textbooks and popular accounts as represent- note that tail region posterior to disc is missing. ing a ``typical'' fossil stingray (e.g., Stevens B. Holotype of ²Palaeodasybatis discus Fowler, and Last, 1994; Frickhinger, 1995; Long, 1947 (ANSP 8344), length given as 345 mm TL 1995; note that Stevens and Last [1994: 61] in original description, but specimen missing dis- mistakenly credited ²Heliobatis as being tal tip of tail; note that radials of disc and pelvic from the Jurassic of Utah). Our smallest adult ®ns are painted over. This specimen has recently male specimen is about 330 mm in TL been reported as missing (Spamer et al., 1995: 83). Anterior to top. (around 190 mm in DW), but it is probable that the onset of sexual maturity occurred at an earlier stage. Therefore, it is dif®cult to determine whether There is some degree of variation in ²He- this ``variation'' is really intraspeci®c or liobatis radians, especially in relation to disc merely preservational, or if it is an indication shape (cf. ®gs. 28 and 29). However, the out- of greater species diversity within ²Heliob- ermost segments of the pectoral ®n radials atis. may not leave clear impressions, and the disc Examination of photographs and illustra- is not entirely preserved in many specimens. tions of the type-specimens of both ²Xipho- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 71

Fig. 30. Continued. trygon acutidens Cope, 1879 (Cope, 1884) disc shape and proportions, and lack the den- and ²Palaeodasybatis discus Fowler, 1947 ticulation and dorsal ®n of ²Asterotrygon (®g. 30B) support their synonymy with ²He- maloneyi. The tail region is not preserved in liobatis radians Marsh, 1877 (as previously the holotype of ²H. radians (®g. 30A), but concluded by Grande, 1980, 1984). All three because it lacks the minute denticles over nominal taxa are morphologically similar in disc, snout, and base of tail regions, we can 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 be certain that all three above-mentioned left side of disc and tail just posterior to base nominal species are synonymous and clearly not preserved; AMNH P 4345, adult male, distinct from ²Asterotrygon maloneyi. The exposed in ventral view (ca. 410 mm TL, original types of both Cope's and Fowler's 200 mm DL; Twin Creek); AMNH P 7828, nominal species are reported to be lost adult male (illustrated in Schaeffer and Man- (Grande, 1984; Spamer et al., 1995), but the gus, 1965), exposed in ventral (?) view; holotype of ²H. radians is available (YPM AMNH P 9873, adult male, with distal tip of 528) (®g. 30A). Bigelow and Schroeder tail missing, exposed in ventral view (ca. 400 (1953) doubted the synonymy of both ²He- mm TL, 230 mm DL); AMNH P 9874, teeth liobatis Marsh, 1877 and ²Xiphotrygon only; AMNH P 19665, subadult female, ex- Cope, 1879 with Dasyatis Ra®nesque, 1810, posed in dorsal (?) view, from F-2 (®g. 28); contrary to Fowler (1941). ANSP 8344 (holotype of ²Palaeodasybatis ETYMOLOGY: The generic name stems from discus Fowler, 1947, apparently lost, photo- helios, the Greek word for sun, and batis, graph in Grande, 1984: 27, ®g. II.5; repro- which is Greek for ray (the ``sun-ray''). Gen- duced here as ®g. 30B); DMNH 1530 (no der feminine. further data); FMNH PF 2020, adult male, INCLUDED SPECIES: Presently monotypic. exposed in ventral view, from F-1 (®g. 29); FMNH PF 6947, female, exposed in dorsal ²Heliobatis radians Marsh, 1877 view, from F-1; FMNH PF 10283 (tentative- Figures 26B, 28±30, table 4 ly identi®ed), adult male, exposed in dorsal ²Heliobatis radians Marsh, 1877 (p. 256; original view, from Thompson Ranch (F-2); FMNH description, not illustrated); Grande, 1980, 1984 PF 25009, adult male, exposed in ventral (pp. 24±30, ®gs. II.3±II.5, II.7c; description, view, from F-2 (specimen on exhibit); photographs); Stevens and Last, 1994 (p. 61; SMMP 77.27.1; SMMP 83.2.4; UW 11577; photograph); Frickhinger, 1995 (p. 212; photo- ZMB 1981.0 (adult male, F-2). Other re- graph, brief account); Long, 1995 (p. 87; pho- ferred material (not examined): UW 12309 tograph). (®gured in McGrew and Casilliano, 1975). ²Xiphotrygon acutidens Cope, 1879 (p. 333; orig- STRATIGRAPHIC HORIZON: Fossil Butte inal description, not illustrated); Cope (1884; Member of the Green River Formation, Yp- further description, illustrated). Dasyatis sp.: Haseman, 1912 (pp. 98±99). resian stage (corresponding to the North ²Palaeodasybatis discus Fowler, 1947 (pp. 14± American Wasatchian stage) of the late early 15; original description, not illustrated); Eocene epoch (some 52 million years before Grande, 1980, 1984 (p. 24, ®g, II.5; ®rst pub- present). lished photograph of type specimen, reported ETYMOLOGY: The trivial name radians is lost); Spamer et al., 1995 (p. 83; type-catalog, derived from radius, Latin for ``spoke'', re- specimen con®rmed as missing). ferring to the radial disposition of the pec- ²Dasyatis radians: Cappetta, 1987 (p. 163; trans- toral ®n rays. Gender feminine. feral without comment of ²H. radians Marsh, 1877 to Dasyatis Ra®nesque, 1810). STINGRAY CHARACTERS OF DESCRIPTION: As for genus. ²ASTEROTRYGON MATERIALS: Holotype: YPM 528, missing tail region (®gured in Grande, 1984: 30, ®g. ²Asterotrygon and ²Heliobatis are unques- II.7c; reproduced here as ®g. 30A). Other tionably stingrays, sharing with other my- specimens examined: AMNH P 856, male, liobatiforms the following synapomorphies: exposed in ventral view, with tip of tail miss- (1) caudal sting(s); (2) thoracolumbar (or ing (ca. 370 mm TL, 210 mm DL; Twin second) synarcual cartilage; (3) absence of Creek); AMNH P 857, female, exposed in ribs; (4) laterally expanded, shel¯ike post- ventral view (Newberry coll.); AMNH P orbital processes; (5) loss of the jugal arch 2474, probably female, exposed in ventral from the posteroventral corners of the neu- view, with tail tip posterior to stings missing rocranium; (6) articulation between scapular (ca. 210 mm DL); AMNH P 2475, appears process of shoulder girdle and cervicothorac- to be exposed in dorsal view; AMNH P ic synarcual cartilage of ball-and-socket type; 2985, female, exposed in ventral view, with (7) basihyal cartilage transverse, small and a 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 73 separate element, located between the ®rst cartilages, however, are absent from our dis- hypobranchials (see phylogenetic analysis sected material; cf. McEachran et al., 1996). below); (8) ®rst pair of hypobranchials rela- Another feature of ²Asterotrygon previously tively straight and obliquely oriented from thought to be unique to stingrays, but that the ventral pseudohyoid bar to articulate with also has a greater distribution, is the posterior the basihyal; (9) anterior expansion of the extension of the propterygium beyond the medial (basibranchial) plate (see phylogenet- procondyle to contact the scapulocoracoid ic analysis below). These characters are fur- between the pro- and mesocondyles. Mc- ther discussed in Compagno (1973, 1977), Eachran et al. (1996) used this character to Miyake (1988), Nishida (1990), McEachran further unite Zanobatus to stingrays, but it is et al. (1996), in the morphological descrip- also present in both species of Platyrhina tion of ²Asterotrygon above, and in the phy- (AMNH 44055, 26413, MNHN 1307, logenetic analysis below. Characters 7, 8, MNHN uncatalogued; Carvalho, in press). and 9 are considered synapomorphies of stingrays for the ®rst time in this study (see PHYLOGENETIC RELATIONSHIPS OF below), but are dependent on the topology THE GREEN RIVER STINGRAYS considered, as all three features are lacking The objective of this section is to place in pelagic (myliobatid) species; if this mono- ²Asterotrygon and ²Heliobatis in a larger phyletic group is basal to the remaining phylogenetic framework of extant stingrays. stingrays (e.g., appendix 2), then these fea- Fossil stingrays other than ²Heliobatis and tures are derived at a less inclusive node ²Asterotrygon are not included in the anal- within Myliobatiformes. ysis, but they will be in upcoming studies of Some characters currently accepted as syn- the Monte Bolca stingray fauna. The phylo- apomorphies of stingrays cannot be found in genetic analyses below are not intended to our fossil specimens due to incomplete pres- solve all problems regarding the relationships ervation, such as the absence of the abdom- among dasyatid genera, and do not address inal canal of the lateral-line system on the the larger issue of whether certain benthic coracoid bar (character 18 of McEachran et stingray genera are monophyletic (e.g., Taen- al., 1996), and the nasal curtain extending iura, Dasyatis; Himantura is unquestionably posteriorly to level of the mouth opening not monophyletic and is coded accordingly). (character 10 of McEachran et al., 1996; also The issues of whether the Daystidae and its present in Trygonorrhina, electric rays, and component genera are monophyletic, as well skates, but these are considered by Mc- as the immediate sister-group relationships Eachran et al. to have evolved independent- among ``dasyatid'' stingrays, are still terra in- ly). cognita and currently the most persisting ²Asterotrygon possesses other characters problems in stingray systematics. traditionally thought to be derived for stingrays, but that have recently been demonstrat- PHYLOGENETIC PROCEDURES AND CODING ed to occur in Zanobatus as well (Zanobatus was previously considered to be a platyrhin- Characters that do not vary within the in- id, but hypothesized as being the sister-group group (synapomorphies of Myliobatiformes) of stingrays by McEachran et al., 1996). were excluded from the matrix (characters 1± These are the reduction of the rostral carti- 6 and 8 listed above), as were characters re- lage in adults (even though it is frequently stricted to terminal taxa (autapomorphies). present as the rostral extremity in stingray However, autapomorphies were included if a juveniles, and Zanobatus retains a ®lamen- character is present in one terminal and tous, hyaline rostral cartilage), and the arched scored as uncertain or unavailable (?) in an- and relatively narrow puboischiadic bar. We other, or if the autapomorphy is part of a have dissected one specimen of Zanobatus multistate series. Dasyatis and Urobatis are and examined radiographs of others, and we each coded as a single terminal, contrary to con®rm the presence of the narrow and the coding of Lovejoy (1996), but similar to arched pelvic girdle, along with a stout hyo- that of McEachran et al. (1996). Lovejoy mandibular-Meckelian ligament (the angular coded Dasyatis as two separate terminals be- 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 31. Neurocranium of cleared-and-stained specimen of Gymnura micrura (FMNH 89990) in dorsal (A) and ventral (B) views (same specimen as in ®g. 15). Anterior to top. cause the species he examined of this genus (character 25 [ϭ character 19 of Lovejoy, differed in one character (character 3 below; 1996], coded here as polymorphic for Uro- however, it is not entirely clear what species batis). The other character used by Lovejoy, of Dasyatis are included in each of Lovejoy's degree of extension of the lateral stay of the terminals). We have coded Dasyatis as un- cervicothoracic synarcual cartilage, is not in- certain for this character. Urobatis was coded cluded because the degree of extension of the in Lovejoy (1996) as two terminals, one each lateral stay is dif®cult to properly divide into for Paci®c and Atlantic species, which dif- discrete categories (as also concluded by fered in two characters in his matrix (his McEachran et al., 1996). Lovejoy's (1996) characters 19 and 21). Urobatis is coded here matrix coded Plesiotrygon and Potamotry- as a single terminal because only one of gon identically, as well as for Indo-west Pa- these characters is included in our analyses ci®c Himantura and ``Dasyatis 1''. These 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 75

Fig. 32. Neurocranium of cleared-and-stained specimen of Taeniura lymma (AMNH 44079) in dorsal (A) and ventral (B) views. Note that the structure between the nasal capsules (se) is formed by chondri®ed sensory canals (probably at the junction of the prenasal and subrostral canals), and is not the rudimentary rostral base. Anterior to top.

Fig. 33. Neurocranium and visceral arches of cleared-and-stained specimen of Urotrygon chilensis (FMNH 93737) in dorsal (A) and ventral (B) views (same specimen as in ®g. 14). Synarcual has been removed, but gill arches remain in original position. Anterior to top. 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

TABLE 5 Matrix for the Phylogenetic Analysis of Stingrays Himantura represents Indo-west Paci®c species only, whereas ``Himantura'' represents amphi-American species (H. schmardae and H. paci®ca); ``P'' ϭ polymorphic (0, 1), ``-'' ϭ inapplicable, and ``?'' ϭ unknown (see character descriptions for discussion and justi®cations).

taxa are not identically coded in our matrix, family. The outgroups included are com- however, due to the inclusion of additional posed of two more distantly related batoids characters and differential coding of others. (a guitar®sh and a skate), but our morpho- Genera not available for study, or insuf®- logical comparisons included all batoid ciently described anatomically in the litera- groups. All multistate characters were run as ture, are not included in our analysis (e.g., unordered (characters 2, 3, 20, 22, 25, 28, Pastinachus and Urogymnus; the former is 30, 35, 37, and 43). The ®nal matrix (table monotypic and the latter has two species). 5) is composed of 44 characters, but about a Because specimens of Aetoplatea were un- dozen additional characters were considered available, only Gymnura is coded for the for inclusion and eventually discarded be- Gymnuridae. Himantura is coded as two ter- cause they either varied intraspeci®cally or minals, following Lovejoy (1996), because proved too dif®cult or arbitrary to divide into the amphi-American species (H. schmardae discrete states, and were therefore deemed and H. paci®ca) form a monophyletic group inappropriate for phylogenetic analysis in the (Carvalho and Lovejoy, in prep.) and are present context. Character coding should be morphologically very distinct from the other as precise and explicit as possible, involving species of Himantura, all of which are Indo- con®rmation in more than one specimen of west Paci®c in distribution. The genera of a given terminal when enough material is Potamotrygonidae are coded separately be- available, as accurate scoring is a fundamen- cause we provide additional observations for tal prerequisite of phylogenetic analysis (Pat- some of their characters, some of which may terson and Johnson, 1997; Grande and Be- have an effect on relationships within the mis, 1998), especially in groups that have 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 77

previously manifested high levels of homo- CHARACTER DESCRIPTIONS plasy. The numbers below correspond to char- Phylogenetic analyses were conducted us- acter numbers in the matrix (table 5). Ter- ing Hennig86 (Farris, 1988), and imple- minals in which the character is present are mented with the aid of the software Tree Gar- given in brackets after the respective char- diner (Ramos, 1997) and WinClada (Nixon, acter-state description (except for state 0). 1999). An exact parsimony strategy (*ie) was Literature citations containing more morpho- employed to search for all equally most par- logical information are included for charac- simonious trees. Nodes were diagnosed with ters when appropriate. Table 6 provides the the aid of the Dos Equis (xx) function of length (number of steps), consistency index Hennig86 and with tree diagnostics featured (CI), and retention index (RI) for all char- in Tree Gardiner and WinClada. The matrix acters, as well as the number of minimum- was also run in Nona (Goloboff, 1999), length trees in which each character is pre- which produced the same results as Hen- sent with these indices, including the strict nig86 (with fewer minimum-length trees, consensus tree (see Farris, 1989, for a dis- however). Characters used to diagnose nodes cussion of these indexes). For most charac- are only those that displayed the same opti- ters these indices are optimal for all mini- mization on all of the equally most parsi- mum-length trees. If more than one solution monious trees (i.e., only unambiguous char- is possible for a given character in a subset acters were used to diagnose monophyletic of the minimum-length trees, than variations groups; the mapping of con¯icting characters in length, CI, and RI are given for each sub- onto the strict consensus tree arti®cially in- set in table 6. creases its length due to its polytomies, and it is therefore less parsimonious than any of LATERAL-LINE CANALS the original trees). The strict consensus tree is depicted in ®gure 43. Characters that are 1. Tubules of the subpleural components ambiguous due to multiple optimizations that of the hyomandibular lateral-line canals: not require the same number of steps are re- branched at extremities (0); extremities di- solved by favoring reversals over indepen- chotomously branched (1) [Urobatis, Urotry- dent gains. This procedure was also used for gon]. This character is described and ®gured characters scored as unknown (?) in the for Urobatis jamaicensis and Urotrygon mi- Green River fossils but that could be pushed crophthalmum in Lovejoy (1996). Garman farther down the tree (which assumes, there- (1888) and Chu and Wen (1979) provided fore, that they were present in the fossils), as more detailed descriptions of the lateral-line this procedure saved a step for each character system in batoids. Both ²Asterotrygon and (characters 12, 22, and 44; all ambiguous ²Heliobatis are coded as unknown. characters are shown separately in ®g. 44). 2. Subpleural components of the hyoman- Accelerating the transformations of these dibular lateral-line canals: posterior branch characters ascribes a greater generality to extends caudally more or less parallel to lon- their distribution, which is considered to be gitudinal body axis (0); posterior branch in- a more coherent implementation of parsi- ¯ects towards midline to form a ``lateral mony (for a review, see discussion in Schuh, hook'' (1) [Indo-west Paci®c Himantura, 2000). However, ambiguous characters in Dasyatis, Pteroplatytrygon, Gymnura]; pos- which a step is not saved by accelerating terior branch in¯ects to continue anteriorly their transformations (characters scored with almost parallel to anterior branch, forming a a ``?'' in at least one terminal), are placed at large indentation (2) [Myliobatis, Aetobatus, the node where they are known to occur only Rhinoptera and Mobula]. The condition in with certainty (see ®g. 44; these features Hexatrygon is unknown. This character is il- were not pushed farther down the tree; in lustrated and described in Lovejoy (1996). other words, no assumptions were made Species of Dasyatis are slightly variable in about their presence in taxa scored as uncer- relation to the size and shape of the lateral tain when no steps could be saved; see leg- hook; however, they do not present the ex- end of ®g. 44 for more details). treme condition of pelagic stingrays. Both 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 34. Neurocranium and visceral arches of Potamotrygon motoro (FMNH 94503), stained only with alcian blue, in dorsal (A) and ventral (B) views (gill arches and synarcual in place). Asterisk denotes anterior extension of basibranchial plate (which is fragmented posteriorly). Anterior to top.

²Asterotrygon and ²Heliobatis are coded as ²Asterotrygon,²Heliobatis, and outgroups]. unknown. This character was compiled from Mc- 3. Suborbital components of the infraor- Eachran et al. (1996), and ®rst described by bital lateral-line canals: projecting posteri- Garman (1888). The condition in Hexatrygon orly lateral to mouth (0); projecting posteri- is coded as unknown, following McEachran orly lateral to mouth and anteriorly lateral to et al. (1996). This feature is not available in nasal openings (1) [Potamotrygon, Plesiotry- ²Asterotrygon and ²Heliobatis, which are gon]; forming a complex weblike pattern on also coded as unknown. lateral aspects of the anteroventral disc region (2) [Paratrygon]. According to Lovejoy SKELETON (1996), certain species of Dasyatis and Hi- 5. Anterior process of neurocranium: ab- mantura (Indo-west Paci®c species) present sent (0); present (1) [Rhinoptera and Mobu- another state of this character, ``extensive rela]. In these pelagic genera, the processes ticulation and looping'' (Lovejoy, 1996: (present on each side of the neurocranium, 216). Both taxa were scored here as uncer- representing forward extensions of the lam- tain, as one species, Himantura imbricata, ina orbitonasalis from the nasal capsules) shows yet another condition for this charac- provide support for the cephalic lobes (®ns); ter (Lovejoy, 1996; McEachran et al., 1996). the process is absent from both Green River Hexatrygon,²Asterotrygon, and ²Heliobatis stingrays. This character is also utilized in are coded as unknown. the analyses of Nishida (1990), Lovejoy 4. Scapular loops formed by scapular (1996), and McEachran et al. (1996). Figured components of the trunk lateral-line canals: in Nishida (1990: 28, ®g. 17). The anterior absence of loops (0); presence of scapular processes are clearly absent from both ²As- loops (1) [all terminals except Hexatrygon, terotrygon and ²Heliobatis. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 79

Fig. 35. Dorsal view of gill arches of (A) Taeniura lymma (AMNH 44079) and (B) Gymnura micrura (FMNH 89990). Dorsal elements pulled slightly to each side to reveal ventral gill arch structures with more clarity. Asterisk in panel A indicates expanded distal extension of left hyomandibula that articulates with Meckel's cartilage through the hyomandibular-Meckelian ligament. Anterior to top.

6. Preorbital process: present (0); absent lamina orbitonasalis according to Miyake et (1) [Rhinoptera and Mobula]. The preorbital al. (1992a), but by a different component of processes are widespread in batoids, but are the lamina, justifying its coding as a separate also absent in torpediniforms (which lack a character (also McEachran et al., 1996). The supraorbital crest altogether) and in Rhinop- preorbital process is present in both Green tera and Mobula among stingrays. This char- River stingrays (e.g., ®g. 18). acter, like the previous one, is formed by the 7. Preorbital canal for the passage of the 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 super®cial ophthalmic nerve: dorsally locat- while Nishida (1990) illustrated Aetomylaeus ed (0); anteriorly located (1) [Myliobatis, Ae- nichoftii as having the more restricted state tobatus, Rhinoptera and Mobula]. This fea- (specimens of Aetomylaeus were not avail- ture was described by Nishida (1990) and able for dissection). All other pelagic taxa utilized by McEachran et al. (1996; as the have the more restricted character state (Gar- ``anterior preorbital foramen'') in their phy- man, 1913; Nishida, 1990; McEachran et al., logeny. The condition in ²Asterotrygon and 1996), as observed in Aetobatus narinari ²Heliobatis is obscured, and they were there- (AMNH 222833) and Mobula kuhlii (AMNH fore coded as unknown for this character, but 15319). In Trygonoptera, the supraorbital it is reasonable to infer that the preorbital process is clearly present and is somewhat canal opened dorsally at the junction be- enlarged, with a deep groove for the infra- tween the nasal capsules and supraorbital orbital lateral-line canal, according to the crest, as both fossil taxa have the nasal cap- material we examined (contra Nishida, 1990: sules and anterior cranial roof con®gured as 34). ²Asterotrygon and ²Heliobatis both pre- in other benthic stingrays (changing their sent the supraorbital process (e.g., ®g. 18; but coding to state 0 does not affect the resulting see description of ²Asterotrygon above, p. phylogeny). 42). 8. Foramen for the optic (II) nerve: mod- 10. Extent of orbital region: orbital region erately sized (0); very enlarged (1) [Urolo- of neurocranium long (0); shortened orbital phus and Trygonoptera]. This condition was region with more anteriorly placed supraor- ®rst described by Miyake (1988) for ``Aus- bital and postorbital process (1) [Gymnura, tralian-western Paci®c Urolophus'', and it Myliobatis, Aetobatus, Rhinoptera, and Mob- was included by Lovejoy (1996) in his anal- ula]. ²Asterotrygon and ²Heliobatis have the ysis (as an autapomorphy of Urolophus). primitive condition, as do the remainder of Even though the different states of this fea- stingrays. This feature corresponds to char- ture may seem at ®rst inspection to be some- acter 28 of McEachran et al. (1996), in which what arbitrary and qualitative, the size of the it was coded as part of a multistate character foramen is very distinct between urolophids relating to the development of the postorbital and other stingrays according to our obser- process. This feature is less pronounced in vations (see also Miyake, 1988). We con®rm Myliobatis (Garman, 1913; Meng, 1984; its presence in Trygonoptera (Trygonoptera Nishida, 1990). mucosa [CSIRO H 898±06]; T. testacea 11. Postorbital process: without ventrolat- [CSIRO H 33, CSIRO H 838±05, CSIRO H eral projection (0); continuing ventrolaterally 837±06]; T. ovalis [CSIRO A 2817]). Both to form a cylindrical projection (1) [Myliob- ²Asterotrygon and ²Heliobatis are coded as atis, Aetobatus, Rhinoptera, and Mobula]. unknown. ²Asterotrygon and ²Heliobatis do not show 9. Postorbital process of neurocranium: any ventrolateral extensions of their postor- infraorbital lateral-line canal separates post- bital processes (e.g., ®gs. 18, 19), but this orbital process from small, anterior triangular character would hardly be preserved in fos- outgrowth (supraorbital process) of the su- sils (it is not preserved in the myliobatid praorbital crest (0); postorbital process with ²Promyliobatis from Monte Bolca), as the small foramen for passage of infraorbital lat- extensions are very slender, uncalci®ed ®la- eral-line canal (1) [Plesiobatis, Urolophus, ments. Described in Nishida (1990) and in- Pteroplatytrygon, Aetobatus, Rhinoptera, cluded in the phylogeny of McEachran et al. and Mobula]. Myliobatis has the infraorbital (1996). canal passing ventrally anterior to the post- 12. Ventrolateral expansion of nasal cap- orbital process (Meng, 1984; Nishida, 1990), sules: nasal capsules laterally expanded (0); while most pelagic taxa have the infraorbital nasal capsules ventrolaterally expanded (1) canal running through it. There is some [all terminals except Hexatrygon and out- doubt as to the condition in Aetomylaeus, groups]. Nishida (1990) described the nasal also a myliobatid. Meng (1984) described capsules of Plesiobatis as being only later- and depicted three species of Aetomylaeus as ally expanded, but McEachran et al. (1996) having the more general condition (state 0), coded Plesiobatis as having the more derived 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 81 condition, citing Miyake (1988). Miyake acter. Both ²Asterotrygon and ²Heliobatis (1988) and Lovejoy (1996) both examined are coded with the derived condition (see an- the same specimen and agreed that the nasal atomical description for details, p. 48). capsules are ventrolaterally expanded in Ple- 14. Angular cartilages: absence of angular siobatis, and Miyake's ®gure of its neurocra- cartilages within hyomandibular-Meckelian nium (1988: 166, ®g 36a) corroborates this. ligament (0); presence of angular cartilages Therefore, ventrolateral expansion of the na- within ligament (1) [Potamotrygon, Plesi- sal capsules is coded as derived for Plesiob- otrygon, amphi-American Himantura,²As- atis and all other myliobatiforms except Hex- terotrygon]. The presence of angular carti- atrygon. Note that electric rays also have (in- lages has been used previously as an indi- dependently) ventrolaterally expanded nasal cator of potamotrygonid monophyly, al- capsules (Carvalho, 1999). though Paratrygon lacks them altogether 13. Articulation between hyomandibula according to our radiographed and dissected and Meckel's cartilage: hyomandibulae di- specimens. There is variation in this charac- rectly attached to lower jaws (0); hyoman- ter within Potamotrygon (®g. 40), as some dibulae articulating with lower jaws through species have two cartilages lying in parallel strong, stout ligament (hyomandibular-Meck- in the ligament, while others have only a sin- elian ligament) at distal tip (1) [all terminals gle cartilage, as in Plesiotrygon (®g. 40F). except Hexatrygon, Gymnura, Mobula, and The morphology of the angulars may vary both outgroups]. This ligament, dubbed the from spool-shaped to very elongated and ``hyomandibulo-Meckelian tendon'' by concave in species of Potamotrygon with McEachran et al. (1996), is very obvious in single angulars. In Potamotrygon signata dissected and cleared-and-stained specimens (MCZ 600) the angulars are close to one-half as a stout and somewhat rigid structure. The the length of the hyomandibula (®g. 40E). thick ligament tightly connects the distal The ®rst angular is usually more robust than ends of the hyomandibulae to the posterior the second in species of Potamotrygon with aspect of Meckel's cartilage on each side. paired angulars (P. leopoldi, P. henlei, P. cf. Within the ligament some degree of calci®- ocellata, P. orbignyi), but there is variation cation may occur (see next character). Prim- here as well because in P. brachyura the pos- itively, the hyomandibulae are positioned terior angular element is more stout. The closer and more in parallel to the lower jaws, presence of angulars within the ligament has contacting the lower jaws throughout their been used as evidence of a monophyletic length by means of smaller ligaments (such group composed of amphi-American Himan- as in the electric ray genus Narke). This is tura and potamotrygonids (Lovejoy, 1996; the general condition in nonmyliobatiform McEachran et al., 1996). The angulars of Hi- batoids. The hyomandibular-Meckelian liga- mantura schmardae and H. paci®ca are, ment is widespread among stingrays (except however, somewhat distinct from the much Hexatrygon, mobulids, and gymnurids), but larger and discretely shaped cartilages of Po- is particularly well developed in Taeniura tamotrygon and Plesiotrygon, and they are lymma, Himantura schmardae and H. paci- only tentatively coded with state 1 here. ®ca, all examined species of Potamotrygon, Himatura schmardae has minute angulars, and Plesiotrygon iwamae. Paratrygon aier- dif®cult if not impossible to discern in radio- eba has a much shorter ligament in compar- graphs, and these may vary intraspeci®cally. ison to other potamotrygonids. Dissection of Some specimens of Himantura schmardae Dasyatis sabina, Urobatis jamaicensis, U. possess scattered, nonprismatic calci®cation halleri, Urolophus aurantiacus, Pteroplaty- within the ligament (which strengthens it trygon, and other taxa revealed a ligament considerably) in addition to, or in place of, not as robust in comparison. We con®rm that the small angular cartilages. This is very sim- Zanobatus also has a pronounced hyoman- ilar to the condition observed in one speci- dibular-Meckelian ligament (McEachran et men of Trygonoptera testacea (USNM al., 1996; these authors regarded Zanobatus 39993; ®g. 40B), but not in other specimens to be the sister-group of stingrays). Platyr- examined of the same species, and all other hinids, however, are primitive for this char- species of Trygonoptera lack discrete angu- 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 36. Pelvic girdles of cleared and stained stingrays. A. Dorsal view of Taeniura lymma (AMNH 44079). B. Ventral view of same. C. Dorsal view of Gymnura micrura (FMNH 89990). D. Ventral view of same. E. Ventral view of Urotrygon chilensis (FMNH 93737). The pelvic girdle of Potamotrygon sp. is shown in ®gure 16B. Anterior to top. lars. The occurrence of the small, discrete an- optera and in other putative stingrays with gulars in more than one specimen of H. scaterred, nonprismatic calci®cation. Mc- schmardae indicates that it is probably not Eachran et al. (1996) illustrated the presence simply a result of stress or any other ``exter- of an angular cartilage in Zanobatus, but dis- nal'' factor, as may be the case in Trygon- section of a specimen did not reveal it (al- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 83

Fig. 37. Caudal ®n extremities in representative stingray taxa (cleared and stained). A. Urotrygon chilensis (FMNH 93737); note dermal denticles on dorsal aspect. B. Urobatis jamaicensis (stained with alcian blue only; AMNH 30385). C. Taeniura lymma (AMNH 44079). D. Potamotrygon cf. motoro (AMNH 38138). Scale is same for panels C and D. Anterior to left. though it has a well-developed hyomandib- jaws: antimeres separate at symphysis (0); ular-Meckelian ligament). However, a small both antimeres of jaws symphysially fused ``angular'' does appear in a radiograph of an- (1) [Aetobatus, Rhinoptera, and Mobula]. other specimen of Zanobatus (ZMH 4617), ²Asterotrygon and ²Heliobatis clearly have but this appears more posterior in the liga- the antimeres of both the upper and lower ment and is here interpreted as the secondary jaws separate at symphysis (®g. 21), as do hyomandibular cartilage (see next character). all benthic stingrays and some pelagic taxa Angular cartilages are altogether lacking in (some species of Myliobatis and Aetomy- platyrhinids. ²Asterotrygon is coded here as laeus). Myliobatis is coded as polymorphic having angulars (see anatomical description for this character (as in Lovejoy, 1996; above, pp. 48, 49). McEachran et al., 1996). 15. Secondary hyomandibular cartilages: 17. Mandibular width at symphysis: lower absent (0); present (1) [Urolophus, Myliob- atis, Aetobatus, Rhinoptera, and Mobula]. jaws slender at symphysis (0); lower jaws Lovejoy (1996) and McEachran et al. (1996) symphysially thickened (1) [Myliobatis, Ae- described this character, which is ®gured for tobatus, and Rhinoptera]. ²Asterotrygon and pelagic stingrays in Garman (1913). We ob- ²Heliobatis have the more general condition, served the small cartilage in various species as does Mobula. This character appears to be of Urolophus. Both ²Asterotrygon and ²He- functionally correlated with the arrangement liobatis are coded as not having secondary of teeth in pelagic taxa (see character 19); hyomandibular cartilages, but their absence those with crushing, ¯attened teeth also have may be preservational, as the cartilages are much stronger lower jaws to accommodate generally small and poorly calci®ed. the greater pressures during feeding. These 16. Symphysial fusion of upper and lower features are considered separate here, but de- 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 activating one of them does not affect our [all taxa except outgroups]. Hexatrygon, results. ²Heliobatis, and ²Asterotrygon are coded as 18. Lateral projections of lower jaws unknown. The condition is simply not pre- (``wing-like processes''): absent (0); present served in both fossil taxa, and con¯icting ac- (1) [Myliobatis, Aetobatus, Rhinoptera, and counts regarding it in Hexatrygon have been Mobula]. These processes, which project lat- published. Nishida (1990) illustrated Hexa- erally from close to the symphysis in pelagic trygon as having both elements fused, but taxa, are absent in both ²Asterotrygon and Miyake and McEachran (1991) depicted ²Heliobatis (®g. 21). Described and illustrat- them as unfused, following Heemstra and ed in Nishida (1990). Smith (1980). As we have not been able to 19. Arrangement of teeth in both upper examine the gill arches of Hexatrygon,we and lower jaws: teeth with minute cusps and code it as unknown. Another related char- arranged in separate, individual rows (0); acter, fusion of anterior ceratobranchial ele- teeth ¯attened, and in a pavementlike ar- ments to each other, used by previous authors rangement (1) [Myliobatis, Aetobatus, and (Rosa, 1985; Dingerkus, 1995) to unite dif- Rhinoptera]. ²Asterotrygon (®g. 27) and ferent benthic stingray genera, must be used ²Heliobatis again have the more general con- with caution, as in many cases the elements dition present in all benthic taxa and mobu- are not fused but juxtaposed, super®cially lids, and not the horizontally expanded teeth appearing to be fused. This is the case in of most pelagic stingrays (depicted in ®g. 50; Taeniura lymma (®g. 35A; Carvalho, Garman, 1913; Notarbartolo-di Sciara, 1996b). Also, anterior ceratobranchial fusion 1987). is variable within certain genera (Potamotry- 20. Basihyal cartilage: basihyal laterally gon), and therefore can only be properly in- elongated, fused to ®rst hypobranchials (0); corporated in a species-level analysis. basihyal a single element, but separate from 22. Arrangement of posterior ceratobran- ®rst hypobranchials (1) [²Asterotrygon, Hex- chials: separate from each other (0); anky- atrygon, Plesiobatis, Urolophus, Trygonop- losis between fourth and ®fth ceratobranchi- tera, Gymnura]; basihyal separate from ®rst als (1) [all taxa except outgroups, Hexatry- hypobranchials but fragmented into more gon, and pelagic taxa]; fourth and ®fth cer- than one component (2) [Urobatis, Paratry- atobranchials fused to each other (2) gon, Potamotrygon, Plesiotrygon, amphi- [Myliobatis, Aetobatus, Rhinoptera, and American Himantura, Taeniura, Himantura, Mobula]. This character is described and il- Dasyatis, Pteroplatytrygon]; basihyal absent lustrated by Miyake and McEachran (1991). (3) [Urotrygon, Myliobatis, Aetobatus, Rhin- State 1 refers to the contact and partial over- optera, and Mobula]. Pelagic stingrays have lapping between the last two ceratobranchi- also lost the ®rst hypobranchial elements als, forming part of the insertion for the cor- (Miyake and McEachran, 1991). ²Heliobatis acobranchialis muscle (Miyake, 1988). In pe- is coded as unknown. The pattern of frag- lagic taxa, the ceratobranchials are fused me- mentation of the basihyal cannot be reliably dially, even though the limits of each coded into discrete conditions (cf. Rosenber- ceratobranchial element may be clearly de- ger, 2001b); the determination by Dingerkus marcated, as in Aetobatus). This character is (1995) that a basihyal divided into three sep- unknown in both ²Asterotrygon and ²He- arate pieces of equal size unites potamotryg- liobatis. nids with Taeniura is entirely without basis. 23. Median projection of the basibranchial McEachran et al. (1996) credited some sting- medial plate: absent (0); present (1) [all taxa rays (Urolophus, Gymnura; ®g. 35B) as hav- except outgroups, Plesiobatis, Myliobatis, ing fused basihyals and ®rst hypobranchials, Aetobatus, Rhinoptera, and Mobula]. The but this is not the same as in more general- basibranchial copula, or medial plate, pro- ized batoids, where these elements cannot be jects anteriorly in most nonmyliobatid sting- distinguished from each other and form a rays from the level of the ®rst ceratobran- continuous structure. chial (or ventral pseudohyoid) to occupy a 21. Fusion of ventral pseudohyoid and small space between both anterior hypobran- ®rst ceratobranchial: absent (0); present (1) chials. The medial plate is the result of fusion 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 85 of hypobranchial and basibranchial elements Rhinoptera, and Mobula lack a discrete me- (Miyake, 1988). ²Heliobatis is coded as un- sopterygium (see next feature), and therefore known. this character is coded as inapplicable, con- 24. Articulation between ®fth epi- and cer- trary to Lovejoy (1996), who coded them atobranchial elements to scapulocoracoid: with the more general state. close together (0); widely separated (1) [My- 28. Distinct components of the mesoptery- liobatis, Aetobatus, Rhinoptera, and Mobu- gium: mesopterygium single element (0); la]. Both Green River stingrays are coded as fragmented (1) [Gymnura, Myliobatis]; miss- unknown. This character is described and ®g- ing altogether (2) [Aetobatus, Rhinoptera, ured in Nishida (1990: 51, ®g. 33). and Mobula]. Dasyatis is coded with a poly- 25. Lateral stay of synarcual: originates morphism because both conditions occur in ventral to spinal nerve foramina (0); origi- component species (D. zugei has fragmented nates dorsal to spinal nerve foramina (1) [Po- mesopterygia, as does D. matsubarai accord- tamotrygon, Plesiotrygon]; contacting syn- ing to Nishida, 1990); coding Dasyatis as arcual both dorsally and ventrally to foram- having the more general condition (present ina (2) [Plesiobatis]. This character is com- in all other Dasyatis species examined), how- piled from Lovejoy (1996), but we added an ever, does not alter tree topology. Myliobatis additional character state for Plesiobatis be- has fragmented mesopterygia, but other pe- cause of the unique con®guration of its lat- lagic genera (including Aetomylaeus) are eral stay (®gured in Nishida, 1990: 60, ®g. lacking them altogether, with radials articu- 38a). We have con®rmed the distribution of lating directly to the scapulocoracoid. this feature in additional benthic stingrays (in 29. Lateral expansion of radials in pec- species of Urotrygon, Urobatis, and Taeni- toral region: absent (0); present (1) [Gym- ura). ²Asterotrygon and ²Heliobatis are cod- nura, Myliobatis, and Aetobatus]. Compiled ed as unknown. Aetobatus, Rhinoptera, and from Nishida (1990). The expanded pectoral Mobula are coded as inapplicable because radial segments are more pronounced ante- their lateral stay is extremely reduced or riorly but are present throughout most of the lacking altogether. disc (the segments appear enlarged in Gym- 26. Fossa on dorsal scapular region: ab- nura; ®g. 15). Both ²Asterotrygon and ²He- sent (0); present (1) [Trygonoptera, Urobatis, liobatis have the more general condition pre- Urotrygon, amphi-American Himantura, sent in most benthic stingrays (®g. 39C±F). Taeniura, Himantura, Pteroplatytrygon, This character (shown in ®g. 39A, B) is fur- Dasyatis, Myliobatis, Aetobatus, Rhinoptera, ther modi®ed (lost) in more derived myliob- and Mobula]. Miyake (1988), Nishida atids (Rhinoptera, Mobula, and Manta; Nish- (1990), Lovejoy (1996), and McEachran et ida, 1990). al. (1996) described this character, which is 30. External margin of mesopterygium: incorporated into their phylogenetic studies. more or less straight, not fused to radials (0); Both ²Asterotrygon and ²Heliobatis appear undulated, not fused to radials (1) [Gymnura; to lack the fossa on the scapular process (and ®g. 38C]; highly sinuous, appearing to be are coded as such), but this may be due to fused with articulating radial elements (2) poor preservation. [Urolophus, Trygonoptera]. The ``fused'' 27. Contact between pro- and mesoptery- condition of radials with the mesopterygium gium in the pectoral ®n: absent (0); present is unique to Urolophus and Trygonoptera, (1) [Potamotrygon and Plesiotrygon]. Love- which differ from each other in their nasal joy (1996) used this character as further ev- ¯ap arrangement (Trygonoptera has ¯eshy idence for monophyly of Potamotrygon plus external lobes, absent from Urolophus; Last Plesiotrygon, and we tentatively con®rm it and Stevens, 1994) and in other features here, although some stingray taxa appear to (such as the passage of the infraorbital lat- be somewhat intermediate (Paratrygon). eral-line canal through postorbital process in Both ²Asterotrygon and ²Heliobatis have the Urolophus). The mesopterygia of these gen- more general condition, but the interface be- era have markedly sinuous external margins tween the mesopterygium and propterygium (®g. 42), a condition not observed in any oth- is poorly preserved in the fossils. Aetobatus, er stingray. The pelagic stingrays Aetobatus, 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 38. Selected anatomical features of representative cleared-and-stained stingrays. A. Ventral aspect of anterior neurocranium of Urobatis halleri (FMNH 42601). B. Anterior portion of ventral gill arches (ventral view) of Urobatis halleri (FMNH 42601). C. Ventral aspect of pectoral ®n basal elements of Gymnura micrura (FMNH 89990), left side. D. Pectoral ®n basals of Urotrygon chilensis (FMNH 93737) in ventral view (left side). E. Denticles of Urotrygon chilensis (FMNH 93737) from anterior disc region between pectoral radials in dorsal view. Note that our identi®cation of chondri®ed sensory canals (se) in panel A is tentative (structure may represent the rostral base). Not to scale. Anterior to top.

Rhinoptera, and Mobula are coded as inap- gon]. This character has traditionally been plicable for this feature (see character 28 used as evidence of potamotrygonid mono- above). phyly (Garman, 1877, 1913; Rosa, 1985; 31. Median prepelvic process: absent or Rosa et al., 1987). Aetobatus, Rhinoptera and weakly developed (0); very elongated (1) Mobula, and, to a lesser extent, Myliobatis [Potamotrygon, Paratrygon, and Plesiotry- have a moderately developed median prepel- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 87

Fig. 39. Pectoral skeleton of representative stingray taxa (in left column), magni®ed (in right column, from box at left) to show relationship between adjacent radial elements (character 29 of phylogenetic analysis). A, B. Gymnura micrura (FMNH 89990), left side, ventral view (arrow indicates derived state of character 29). C, D. Urotrygon chilensis (FMNH 93737), left side, ventral view. E, F. Potamotrygon motoro (FMNH 94503), left side, ventral view. Not to scale. Anterior to top. 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 40. Enlarged dorsoventral view (positive prints from x-ray radiographs) of the articulation between the hyomandibula and jaw joint or lower jaw of some extant stingrays showing different states of characters 8 (presence of hyomandibular-Meckelian ligament) and 9 (referring to the arrangement of the angular cartilages) of phylogenetic analysis. A. Gymnura japonica (AMNH 26691), with hyomandibula articulating directly to jaws, that is, without developed hyomandibular-Meckelian ligament (angular cartilages absent; see also ®g. 15, ventral view). B. Trygonoptera testacea (USNM 39993), with scattered calci®cation present within well-developed hyomandibular-Meckelian ligament connecting hyomandibula to lower jaw and with lack of discrete angular elements. C. Potamotrygon leopoldi (UERJ 719), with two angular cartilages present within stout ligament (lower angular cartilage is roughly two- thirds the width of the upper angular element). D. Potamotrygon, sp. nov. (MZUSP 25580), with two angulars of more or less equal dimensions. E. Potamotrygon signata (MCZ 600) showing elongated and slender angular cartilage (roughly one-half length of hyomandibula) associated with much smaller angular closely contacting hyomandibula. F. Plesiotrygon iwamae (MZUSP 42848), depicting single angular cartilage (presumably anterior angular cartilage), positioned at a slightly oblique angle to hyomandibula (and not so much at a right angle to it, as seen in panels C±E). Figures are not to scale. Anterior to top. vic process, but it is not nearly as elongated natively be treated as a multistate character as that of potamotrygonids. Some benthic with the smaller condition of pelagic genera stingrays also have weakly developed medi- and some benthic species coded as an inter- an prepelvic processes (some species of Das- mediate step. We chose to include only the yatis, Taeniura, gymnurids; ®g. 36), but highly modi®ed potamotrygonid prepelvic these are much smaller than the condition in process, although not ideal, because accu- potamotrygonids. This feature could alter- rately coding the many different weakly de- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 89

Fig. 40. Continued. veloped prepelvic processes would require a small, indistinct dorsal ®n anterior to the cau- more inclusive species-level matrix. dal stings, but it is absent from most species. 32. Pelvic girdle shape: not arched or only The vast majority of butter¯y rays lack a dor- moderately so (0); greatly arched (1) [Gym- sal ®n (all species of Gymnura); coding this nura, Aetobatus, Rhinoptera, and Mobula]. feature as uncertain (taking into account that Both fossil Green River taxa have the more it is present in both species of Aetoplatea) general condition. Coding the relative degree would just add an extra step without any of arching among benthic stingrays proved change in topology. Lovejoy coded the pres- dif®cult and arbitrary (cf. Rosenberger, ence of the dorsal ®n as derived, because the 2001b), so only the extreme condition of outgoups he employed (Hexatrygon and Ple- Gymnura (®g. 36C, D) and pelagic genera, siobatis) lack this character. It is coded dif- in which the pelvic girdle is shaped like a ferently here because we did not assume that wishbone, was included (as in Lovejoy, these genera are basal to other stingrays, and 1996). more basal batoids were used as outgroups. 33. Dorsal ®n: present (0); absent (1) [all The dorsal ®n of stingrays, when present, is terminals except ²Asterotrygon, Trygonop- always single as opposed to the presence of tera, Myliobatis, Aetobatus, Rhinoptera, two dorsals in almost all other batoids. Cod- Mobula, and outgroups]. Dorsal ®ns are ing this feature to re¯ect the anatomical ar- primitively present in batoids. Urolophus is rangement of the radials and basals proved scored as uncertain, as some species have a very complex. Outgroups generally display 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 40. Continued. aplesodic dorsal ®ns internally supported by present (1) [Paratrygon, Potamotrygon, Ple- radials and ceratotrichia (guitar®shes), but siotrygon, Dasyatis, amphi-American Hi- skates are somewhat intermediate with radi- mantura, Taeniura, Himantura, Pteroplaty- als extending to almost the ®n margins. The trygon, Myliobatis, Aetobatus, Rhinoptera, internal structure of the dorsal ®n of Trygon- and Mobula]. The Green River stingrays optera needs to be investigated to ascertain have the more general batoid condition (®g. if it is similar to that of ²Asterotrygon, that 26), as do all taxa with a caudal ®n (urolo- is, plesodic without enlarged basals (as op- phids, urotrygonids, Plesiobatis, Hexatry- posed to pelagic genera, which have enlarged gon). All four species of Gymnura examined basal elements in the dorsal ®n; ®g. 41). also have discrete vertebrae extending to the Coding ²Asterotrygon and Trygonoptera distal tip of the tail (state 0). This contrasts identically to re¯ect a putative similarity in with the coding given to Gymnura by both dorsal ®n internal structure may support the Lovejoy (1996) and McEachran et al. (1996). placement of ²Asterotrygon as a basal uro- 35. Caudal ®n: present (0) [outgroups, lophid (a relationship present in 21 of the 35 Plesiobatis, Hexatrygon, Urolophus, Trygon- equally most parsimonious trees; see below), optera, Urotrygon, and Urobatis]; reduced to but more information is needed to properly tail-folds (1) [²Heliobatis,²Asterotrygon, code the dorsal ®n skeleton in benthic sting- Dasyatis, Potamotrygon, Plesiotrygon, Taen- rays. iura, Pteroplatytrygon]; absent (2) [Paratry- 34. Cartilaginous rod in tail: absent (0); gon, amphi-American Himantura, Himantu- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 91

Fig. 41. Dorsal ®n skeleton of extant stingrays. A. Aetoplatea zonura. B. Myliobatis tobijei. C. Rhinoptera javanica. D. Mobula japanica. E. Aetobatus narinari (from AMNH 44142 XR). Panels A± D modi®ed from Nishida (1990: ®g. 40); panel E is original. Note that caudal stings and vertebral elements are schematic. Anterior to left. ra, Gymnura, Myliobatis, Aetobatus, Rhin- atis have vertebrae extending posteriorly optera, and Mobula]. Because the caudal ®n well beyond caudal stings as far as can be of stingrays (when present) is different from observed and caudal ®ns reduced to tail- that of outgroups in being plesodic as op- folds (®gs. 24, 26). The tail-folds of numer- posed to aplesodic, we experimentally coded ous taxa (such as in species of Dasyatis and this character differently to re¯ect this, add- Potamotrygon) exhibit what has been termed ing an extra character state. This produced ``rudimentary radial elements'' (Nishida, no change in topology, and therefore the cau- 1990), justifying the coding of tail-folds as dal ®n was coded more simplistically. This part of a multistate character encompassing character is independent from character 34 the caudal ®n of stingrays in general. We above, because ²Asterotrygon and ²Heliob- have observed these rudimentary radials in 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 42. Articulation between mesopterygium and pectoral radials (character 20 in phylogenetic analysis) of representative extant stingray taxa in dorsoventral view. A. Potamotrygon leopoldi (UERJ 719), right side (inverted), ventral view. B. Dasyatis margarita (AMNH 41512), left side (inverted), dorsal view. C. Urolophus lobatus (CSIRO P8197), right side (inverted), ventral view. D. Trygonoptera testacea (USNM 39993), right side (inverted), ventral view. Arrows depicts sinuous margin of mesopterygium where it articulates to radials in panels C and D. Figures are not to scale. Anterior to top. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 93 countless specimens of Potamotrygon, Das- Potamotrygon (but more robust than in Par- yatis, and Taeniura (®g. 37). atrygon) and is more dif®cult to delimit due to an abundance of connective tissue. It is MANDIBULAR MUSCLE PLATE more strongly anastomosed with the depressor hyomandibularis (and also partly with the 36. Adductor mandibulae complex: with- coracohyoideus and coracohyomandibularis) out posteromedial extension (0); posterome- and is similar to the condition depicted for dial extension present (1) [Myliobatis, Aeto- ``Himantura'' paci®ca by Lovejoy (1996) batus, Rhinoptera, and Mobula]. This char- (state 2). Taeniura also has a very developed acter is unknown in both ²Asterotrygon and spiracularis that inserts medially, but dorsal ²Heliobatis and is described in Miyake to the coracomandibularis (state 1). The spi- (1988) and Nishida (1990; as the ``depressor racularis is absent in Urolophus aurantiacus mandibulae''). Our coding follows Mc- and is very slender and generally does not Eachran et al. (1996). continue posteriorly beyond the lower jaw 37. Spiracularis muscle: projecting ven- (inserting for the most part on lower jaw it- trally to insert on either palatoquadrate, self) in Urobatis (U. jamaicensis and U. hal- Meckel's cartilage, and/or hyomandibula (0); leri), Urotrygon microphthalmum, and Hi- projecting ventrally and posteriorly beyond mantura imbricata.InDasyatis sabina and hyomandibulae and both sets of jaws to in- Pteroplatytrygon, the spiracularis is also sert dorsal to coracomandibularis (1) [Taeni- slender but projects posterior to the lower ura]; projecting ventrally and posteriorly be- jaws to some degree, however, but not as yond jaws and hyomandibulae to insert ven- much as in Taeniura, Potamotrygon, Plesi- tral to coracomandibularis (2) [Potamotrygon, otrygon, and amphi-American Himantura. Plesiotrygon, amphi-American Himantura]. More comparative data for Dasyatis and Hi- This muscle originates on the otic region of mantura are needed, especially as Kesteven the neurocranium and projects ventrally to (1942) described the posterior extension of form part of the spiracular wall. Miyake the spiracularis in Dasyatis brevicaudata. (1988) and Miyake et al. (1992a) provided de- Myliobatids are coded as in Lovejoy (1996) scriptions of the different conditions of the and McEachran et al. (1996). In Zanobatus, spiracularis in batoids. This is a complex the spiracularis does not project posteroven- muscle with many variations in batoids trally to the same extent as in stingrays. Be- (more character states are described and uti- cause the muscle is very complex (Kesteven, lized by McEachran et al., 1996). Lovejoy 1942; Miyake, 1988), more descriptions are (1996) described this muscle for various needed to corroborate our use of it here. This stingrays, and our use of this character fol- character is unavailable in both ²Asterotry- lows his analysis to a large degree. gon and ²Heliobatis. We have con®rmed the different insertion 38. Depressor mandibularis muscle: pre- patterns of the spiracularis in many stingrays. sent (0); absent (1) [Myliobatis, Aetobatus, In Potamotrygon falkneri, P. motoro, P. or- Rhinoptera, and Mobula]. This character is bignyi, P. cf. schroederi, and ``Himantura'' unavailable in both ²Asterotrygon and ²He- schmardae, the spiracularis is thick, well de- liobatis. We code this character following veloped and very closely associated with the McEachran et al. (1996), and it is described depressor hyomandibularis posterior to the for stingrays in Miyake (1988) and Nishida lower jaw, inserting with its antimere ventral (1990; as the ``intermandibularis posterior''). to the coracomandibularis (state 2). In these taxa the spiracularis and the depressor hy- HYPOBRANCHIAL MUSCLE PLATE omandibularis are dif®cult to separate from one another, but the spiracularis is generally 39. Coracohyoideus muscle: not connected dorsal to the depressor hyomandibularis. In at midline (0); connected at midline (1) [My- Paratrygon, the spiracularis is very slender liobatis, Aetobatus, Rhinoptera, and Mobu- and is not continuous posteriorly with the de- la]. This character is unknown in both ²As- pressor hyomandibularis (state 0). In Plesi- terotrygon and ²Heliobatis. Coded as in otrygon, this muscle is more slender than in McEachran et al. (1996), and described in 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 43. Strict consensus tree (length ϭ 86 steps, CI ϭ 0.65, RI ϭ 0.79) obtained from 35 equally most parsimonious trees (length ϭ 82 steps, CI ϭ 0.68, RI ϭ 0.82), derived from the matrix in table 5. Characters are numbered as in the text and in table 5 (see text for character descriptions). Only unambiguous characters are shown; unique derivations depicted as closed circles, characters with homoplasy as open circles; character states in parentheses (®gure is modi®ed from output generated directly from WinClada). 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 95

Fig. 44. Tree topology taken from ®gure 43 (strict consensus tree) with ambiguous characters mapped (numbered as in text and matrix in table 5, unambiguous characters in ®g. 43). Some of the characters are ambiguous because they are scored as uncertain in Hexatrygon (characters 4 and 21) and in Myliobatis (character 16); these are displayed conservatively on the tree (i.e., it is not simply assumed that they will be found in these taxa). Other characters have more than one equally parsimonious optimization (characters 3, 20, 32, 43), whereas others are scored as uncertain in the Green River stingrays (characters 12, 22, and 44). The optimization chosen in both of these cases is accelerated transformation, favoring reversals over independent gains. Characters denoted with an asterisk (*) are part of multistate transformation series that have unambiguous character states in ®gure 43.

Miyake (1988; as the ``y'' muscle) and Nish- some physiological mechanism to eliminate ida (1990). The state in Hexatrygon is un- excess water from their metabolism. known. 41. Rectal gland: present (0); reduced (1) [Paratrygon, Potamotrygon, Plesiotrygon]. MISCELLANEOUS Coded as unavailable in ²Asterotrygon and ²Heliobatis. Inhabiting freshwater does not 40. Urea retention: urea retained in blood necessarily imply that the rectal gland is ab- (0); urea excreted in urine (1) [Potamotry- sent, as it is present in species of Dasyatis gon, Paratrygon, Plesiotrygon]. This char- that are strictly con®ned to freshwater (e.g., acter is unavailable in ²Asterotrygon and Dasyatis garouaensis; Thorson and Watson, ²Heliobatis, and both are coded as unknown. 1975), and the rectal gland of Potamotrygon Potamotrygonids produce copious, diluted is atrophied, not completely lacking (Thor- urine (Thorson, 1970; Thorson et al., 1967, son et al., 1978). Potamotrygonids are not 1983a). As the Green River stingrays were capable of retaining urea or secreting salt via freshwater inhabitants (see ``Geological Set- the rectal gland if placed in saline waters ting'' above), it is likely that they employed (references in Brooks et al., 1981). 96 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 45. Five subgroups of trees showing the different phylogenetic positions of ²Asterotrygon that result once uncertainty concerning the relationships among dasyatid genera is eliminated from the minimum-length trees obtained in this study (based on matrix in table 5; strict consensus in ®g. 43). ²Asterotrygon pairs with urolophids in the subgroups depicted in C±E, representing 21 of the 35 equally most parsimonious trees (see text for details).

42. Spiracular tentacle: absent (0); present optera and Mobula]. ²Asterotrygon and ²He- (1) [Urobatis, Urotrygon]. The structure of liobatis display the more general condition the spiracular tentacle or lobe was ®rst de- in benthic stingrays, that is, the disc contin- scribed in detail by LaMarca (1963) for Uro- ues anterior to the neurocranium without batis jamaicensis, but it is also present in forming a separate (cephalic) lobe. Our use Urotrygon (Bigelow and Schroeder, 1953; of this character strictly follows McEachran Miyake, 1988). The tentacle is present in em- et al. (1996). The cephalic lobes are inter- bryonic specimens and is resorbed shortly nally supported by radial elements of the before birth. It has been used previously in propterygia, which are interrupted at more or the phylogenetic analyses of Lovejoy (1996) less the level of the eyes and resume anterior and McEachran et al. (1996). This character to the neurocranium. The lobe is continuous is unknown in both ²Asterotrygon and ²He- anteriorly in Myliobatis, but a median inden- liobatis. tation divides the lobe into two smaller ones 43. Cephalic lobes: absent (0); single and in Aetobatus. Rhinoptera and Mobula have continuous (1) [Myliobatis]; single with an two prehensile lobes, one on each side, that indentation (2) [Aetobatus]; paired (3) [Rhin- are completely separated (further supported 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 97

Fig. 45. Continued. internally by the anterior process of the neu- variable. We coded this genus as having the rocranium). nasal curtain reaching the mouth opening (cf. 44. Nasal curtain: not reaching mouth re- McEachran et al., 1996), as observed on a gion (0); extending posteriorly as far as large specimen from New Caledonia and a mouth opening (1) [all taxa except outgroups smaller one from Fiji (both deposited in the and Hexatrygon]. This character is also un- MNHN). known in ²Asterotrygon and ²Heliobatis. The condition in the monotypic Hexatrygon RESULTS OF PHYLOGENETIC ANALYSIS is described by McEachran et al. (1996), and we follow their interpretation here, which is Phylogenetic analysis of the matrix pre- supported by the depictions of H. bickelli in sented in table 5, summarizing the anatomi- Heemstra and Smith (1980), Shen and Liu cal information described above, resulted in (1984), Shen (1986), and Last and Stevens 35 equally most parsimonious trees (length (1994), as well as by the specimens we ex- ϭ 82 steps, CI ϭ 0.68, RI ϭ 0.82), the strict amined. Even though the failure of the nasal consensus of which is shown in ®gure 43 curtain to reach the mouth opening may be (ambiguous characters are shown separately in¯uenced by postmortem distortion due to in ®g. 44). The strict consensus tree (length poor preservation, the condition in Hexatry- ϭ 86 steps, CI ϭ 0.65, RI ϭ 0.79) is not gon is extreme compared to other stingrays. entirely dichotomous, but contains various The nasal curtain in Plesiobatis may be more monophyletic components and only two un- 98 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 45. Continued. resolved nodes (the monophyletic groups ence of tail-folds has traditionally been used discovered, some of which are novel, are de- to diagnose the family (Bigelow and Schroe- scribed below). The unresolved relationships der, 1953; Compagno and Roberts, 1982, mostly concern the various dasyatid genera, 1984; Nishida, 1990), but these are absent in which are placed in a large polytomy with amphi-American Himantura and Himantura the node Gymnuridae ϩ Myliobatidae in the sensu stricto, and are also present in pota- strict consensus, and the af®nities of ²Aster- motrygonids and ²Heliobatis and are there- otrygon and Plesiobatis. The 35 minimum- fore plesiomorphic at this level. length trees can be reduced to ®ve subgroups In relation to the Green River stingrays, of trees once the uncertainty regarding the our analysis indicates that both genera do not relationships of dasyatid genera are removed; form a monophyletic unit. But this is not sur- the ®ve subgroups of trees vary only in the prising given that no putative homologies positions of ²Asterotrygon and Plesiobatis were discovered and included in the matrix. (summarized in ®g. 45). The nonmonophyly The phylogenetic placement of ²Asterotry- of the whiptailed stingrays (Dasyatidae) in gon varied in the minimum-length trees; it the strict consensus is not surprising, given was resolved as either the next most basal that there are no unequivocal characters stingray genus after Hexatrygon (in 7 of the known to support it (a monophyletic Dasy- 35 minimum-length trees), as the next most atidae does not occur in any of the equally basal stingray genus after Hexatrygon but in most parsimonious trees either). The pres- a polytomy with urolophids (also in 7 of 35 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 99

Fig. 46. Previous morphological phylogenies of stingrays (Myliobatiformes). A. Nishida (1990). B. Lovejoy (1996) and McEachran et al. (1996). Nishida's analysis contained many species, and what is shown is only a summary of his scheme (e.g., Trygonoptera was shown nested within Urolophus). The trees of Lovejoy and McEachran et al. originally differed in relation to the taxa included (see text) but are summarized here identically for simplicity without altering the pattern of relationships that they obtained. trees), or as the sister-group to the Urolophi- tomy with ²Asterotrygon ϩ Urolophidae dae (in 21 of the 35 trees; ®g. 45C±E). ²He- (again in 7 of 35 trees), or (in 21 of 35 trees) liobatis is resolved as the most basal genus sister-group of the large node Urotrygonidae of the Myliobatoidea, sister-group to the ϩ Myliobatoidea (where Myliobatoidea ϭ clade Potamotrygonidae ϩ (``Dasyatidae'' ϩ ²Heliobatis ϩ (Potamotrygonidae ϩ (``Das- (Gymnuridae ϩ Myliobatidae)) in all equally yatidae'' ϩ (Gymnuridae ϩ Myliobatidae)))). most parsimonious trees. Therefore, ²Aster- The latter relationship is a departure from otrygon and ²Heliobatis (or their direct an- previous studies, in which Plesiobatis has cestors) probably independently invaded the ®gured as either the most basal stingrays ge- freshwater system of Fossil Lake and were nus or the next most basal genus after Hex- phylogenetically divergent well before this atrygon (Nishida, 1990; Lovejoy, 1996; invasion occurred (further discussed below McEachran et al., 1996). under ``Biogeographical Implications''). Our phylogenetic analysis uncovered two The relationships of Plesiobatis varied additional stingray synapomorphies that are considerably in our minimum-length trees present in all minimum-length trees: charac- (®g. 45); it was either sister-group to the re- ters 20 (state 1) and 23. The latter feature, maining stingrays except Hexatrygon (in 7 anterior expansion of the medial (basibran- out of 35 trees), sister-group to the remaining chial) plate, requires homoplasy as it is re- stingrays except Hexatrygon but in a poly- versed in myliobatids, while Plesiobatis has 100 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

TABLE 6 independently lost the basibranchial anterior expansion (a further autapomorphy for this Length (number of steps), Consistency (CI) and genus). Character 20, regarding the basihyal Retention (RI) Indices for Each Character, and the Number (N) of Trees in Which Each cartilage, is highly modi®ed in stingray evo- Optimization Occurs lution: it is lost in Urotrygon, becomes frag- Based on the phylogenetic analysis of matrix in table 5 mented for the the large node Urotrygonidae (strict consensus in ®g. 43); MPTs ϭ equally most- ϩ (²Heliobatis ϩ (Potamotrygonidae ϩ parsimonious trees. (``Dasyatidae'' ϩ (Gymnuridae ϩ Mylioba- tidae)))), but is further lost for myliobatids (note that there is more than one possible optimization of this latter transformation as shown in ®g. 44). The loss of the dorsal ®n (character 33) is a putative stingray synapomorphy but is not optimized in this way in all resulting trees (it is in some, however; see ®g. 44). Myliobatids, Trygonoptera, Aetopla- tea, and ²Asterotrygon have all regained the dorsal ®n independently. In trees that support a monophyletic ²Asterotrygon ϩ Urolophi- dae, it is possible that the dorsal ®n evolved only once and was subsequently lost in Uro- lophus, which would render the mere presence of the dorsal ®n plesiomorphic for ²As- terotrygon; our use of this character to diagnose this fossil taxon is not compromised in this scenario, however, as the ®n is uniquely covered in denticles and has a well-de®ned plesodic internal structure (whereas the internal structure of the dorsal ®n of Trygon- optera, if present, is unknown; it may not be arranged as in ²Asterotrygon). The node comprising all stingrays with the exception of Hexatrygon is unequivocally supported by a single character: presence of the hyomandibular-Meckelian ligament (character 13). Additional features that are placed at this node in the tree in ®gure 44 are the ventrolaterally projecting nasal capsules (character 12), ankylosis of fourth and ®fth ceratobranchial cartilages (character 22; further modi®ed for Myliobatidae), and the nasal curtain reaching posteriorly to level of mouth opening (character 44). These three characters are unknown in the Green River stingrays, precluding their inclusion in the strict consensus of ®gure 43. There are two additional characters featured at this node, but both are scored as unknown for Hexatry- gon: the scapular loop of the trunk lateral- line canal (character 4) and the fusion of the ventral pseudohyoid with the ®rst ceratobranchial (character 21). The former character is unknown in Hexatrygon (and we 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 101 were unable to examine prepared material), lateral-line canal (character 2) are highly and the latter has been con¯ictingly de- modi®ed in dasyatid genera and in the node scribed in the literature (see character de- Gymnuridae ϩ Myliobatidae, and were scription above). If both characters prove scored with different conditions of a multi- upon further examination to be absent from state character; however, because dasyatid Hexatrygon they will represent additional genera do not form a monophyletic unit, synapomorphies for this node (as in ®g. 44), placing state 1 of this feature as a synapo- but if present they will be synapomorphic for morphy of ``Dasyatidae'' ϩ (Gymnuridae ϩ stingrays. Myliobatidae) is equivocal (see ®g. 44). The node uniting Urotrygonidae and the There are important biogeographic implica- large group composed of ²Heliobatis ϩ Po- tions derived from this relationship, and tamotrygonidae ϩ ``Dasyatidae'' ϩ Gymnur- these are further discussed below. idae ϩ Myliobatidae is supported by the Gymnuridae forms a monophyletic group fragmented basihyal cartilage (character 20, with pelagic stingrays (Myliobatidae), shar- state 2), which further reverts to an unfrag- ing a shortened orbital region (character 10), mented (complete) state in Gymnura, and is fragmented mesopterygium (character 28; independently lost in Urotrygon and pelagic further lost in higher myliobatids), and lateral stingrays (®gs. 43, 44). This character re- extension of radial segments in the pectoral quires ®ve steps in all equally most parsi- ®n (character 29). Loss of the tail-folds monious trees (as well as in the consensus). (character 35, state 2) is ambiguous at this The component ²Heliobatis ϩ Potamotry- node (®g. 44). Other characters putatively gonidae ϩ ``Dasyatidae'' ϩ Myliobatidae is synapomorphic at this level are the presence monophyletic, as evidenced by the reduction of the lateral hook of the subpleural com- of the caudal ®n to tail-folds (character 35; ponent of lateral-line canal (state 1 of char- but homoplasy is required for this character acter 2), and the highly arched pelvic girdle as Paratrygon, amphi-American Himantura, (character 32) (see ®g. 44). Character 2 has Himantura, and the node Gymnura ϩ My- multiple optimizations (either states 1 or 2 liobatidae have all lost the tail-folds indepen- can be equally interpreted as a de®ning fea- dently [state 2], and ²Asterotrygon has ac- ture for this node), while character 32 is ab- quired them independently). The clade Po- sent in Myliobatis, allowing for its indepen- tamotrygonidae ϩ ``Dasyatidae'' ϩ Myliob- dent acquisition in Gymnura and the node atidae is supported by the cartilaginous Aetobatus ϩ (Rhinoptera ϩ Mobula). The unsegmented rod in the tail (character 34, de- transverse, unfragmented state of the basi- void of homolasy). hyal (character 20, state 1) can also be placed The most innovative result of our phylo- at this node (®g. 44; further modi®ed for my- genetic study is the relationship between das- liobatids). yatid genera (not as a monophyletic unit, The Australian stingarees, Trygonoptera however) and the group Gymnuridae ϩ My- and Urolophus, are united as a monophyletic liobatidae. This relationship, not present in Urolophidae based on the enlarged optic the phylogenies of Nishida (1990), Lovejoy nerve foramen (character 8) and the greatly (1996), and McEachran et al. (1996), is sup- sinuous external margin of the mesoptery- ported by the dorsal fossa on the scapular gium (apparently fused to radials, character process (character 26), which occurs inde- 30). Urotrygonidae, containing amphi-Amer- pendently in Urotrygonidae and Trygonop- ican stingrays with a caudal ®n (Urobatis and tera, and is secondarily lost in Gymnura Urotrygon), is monophyletic based on the (note that this character was also included in branched extremities of the subpleural canal the analyses of both Lovejoy and McEachran (character 1), dorsal fossa on scapular pro- et al.). Even though there is much homoplasy cess (character 26, independently acquired in this character, it supports a more recent elsewhere), and spiracular tentacle (character relationship between dasyatid genera (not as 42). South American freshwater stingrays a clade however) with gymnurids and my- (Potamotrygonidae) share the following un- liobatids in all minimum-length trees. The ambiguous evolutionary novelties: a greatly subpleural components of the hyomandibular extended median prepelvic process (charac- 102 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 ter 31), loss of urea retention (40), and reappear highly controversial (such as Gym- duction of the rectal gland (41) (note that nura forming a monophyletic group with Po- modi®cations of the suborbital components tamotrygon; Dunn et al., 2003). of the infraorbital lateral-line canal [character Our strict consensus tree has more com- 3] may be optimized at this node as well; ®g. ponents in common with the stingray rela- 44). tionships presented by Lovejoy (1996) and McEachran et al. (1996) than with the phy- DISCUSSION AND COMPARISON TO PREVIOUS logeny of Nishida (1990) (cf. ®gs. 43 and STINGRAY PHYLOGENIES 46). Comparison with Nishida's phylogeny is dif®cult because his tree was not generated Some of our monophyletic components by computerized phylogenetic procedures, are present in previous phylogenies of sting- and when implemented by Hennig86 the re- rays, including Gymnuridae ϩ Myliobatidae sults are entirely at odds with the phylogeny (and two additional clades within Mylioba- he advocates (Carvalho, 1996a; McEachran tidae: Aetobatus ϩ (Rhinoptera ϩ Mobula)), et al, 1996). Nonetheless, the tree he favors South American freshwater stingrays (Pota- shares just one node in common with our motrygonidae), Potamotrygon ϩ Plesiotry- phylogeny, Gymnuridae ϩ Myliobatidae, gon, Urotrygonidae (Urotrygon ϩ Urobatis), which is supported by similar features (see and a large clade that includes all stingrays above). None of these authors supports a except Hexatrygon. The relationships within monophyletic Dasyatidae (neither do we, but the large monophyletic group comprising po- see appendix 2). However, our resulting phy- tamotrygonids, ``dasyatids,'' gymnurids, and logeny also differs in many respects from the myliobatids, which contains most extant spe- cladograms of Lovejoy (1996) and Mc- cies of stingrays, differ substantially from the Eachran et al. (1996), which are very similar relationships within this clade advocated by to each other. Although we included char- Nishida (1990), Lovejoy (1996), and Mc- acters discussed by these workers, some of Eachran et al. (1996) (their phylogenetic these are coded differently (e.g., characters schemes are shown in ®g. 46). The clade 13, 34, 35 and 44) and contribute to differ- ²Heliobatis ϩ (Potamotrygonidae ϩ (``Das- ences in tree topology. Other characters in- yatidae'' ϩ (Gymnuridae ϩ Myliobatidae))) cluded in the analyses of Lovejoy (1996) and the group comprising dasyatid genera ϩ were excluded from consideration. These in- (Gymnuridae ϩ Myliobatidae) are novel to clude the degree of lateral projection of the our analysis, not being present in previous lateral stays of the ®rst synarcual (Lovejoy's stingray phylogenies. The node Urolophus ϩ character 21) and the level of ®rst segmen- Trygonoptera (Urolophidae) is also unique, tation of the propterygium (his character 25). as Trygonoptera was not included as a sep- We found it dif®cult to quantify both char- arate terminal by Lovejoy (1996) or Mc- acters, which varied intragenerically to a Eachran et al. (1996), and Nishida (1990) in- large degree according to our observations cluded only Trygonoptera testacea, the type- (especially in Dasyatis and Urolophus). Oth- species of Trygonoptera, in his study (as er excluded characters are autapomorphic for Urolophus testacea). Our phylogenetic anal- individual genera (5 of Lovejoy's 39 char- ysis is also the ®rst to include both fossil and acters). We also excluded from consideration extant stingrays, and includes more infor- the direct insertion of the depressor rostri mative characters (44 with potential for muscle on the pectoral propterygium (i.e., in- grouping within stingrays; Lovejoy's matrix serting without an aponeurosis; Nishida, contains 33 characters that vary within sting- 1990), as it only occurs in gymnurids and rays, whereas McEachran et al.'s phylogeny Manta, and the latter genus was not included includes 35 with potential to unite stingray in our matrix. genera; note that McEachran et al.'s analysis Relationships among myliobatid genera in was broader in scope, designed to clarify re- our phylogeny are in accordance with the relationships among extant batoids). Molecular sults of Lovejoy (1996) and McEachran et approaches to the phylogeny of stingrays are al. (1996), but contrary to the topology ad- still in their infancy, and some recent results vocated by Nishida (1990), who favored a 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 103 restricted Myliobatidae including only My- somewhat more varied, and amphi-American liobatis ϩ (Aetobatus ϩ Aetomylaeus), which Himantura is no longer unequivocally sup- in turn is sister-group to Rhinoptera ϩ Mob- ported as the sister-group to potamotrygon- ulidae. Most of the characters used by Nish- ids. We have included the characters sup- ida for this arrangement are optimized dif- porting Lovejoy's clade Taeniura ϩ (amphi- ferently on our trees (we did not include the American Himantura ϩ Potamotrygonidae) ``hyomandibular accessory cartilage 1'' be- in our matrix (see characters 14 and 37 cause it is highly variable within genera), above), but these were not suf®cient to sup- supporting the group Myliobatis ϩ (Aetoba- port their monophyly in any of our equally tus ϩ (Rhinoptera ϩ Mobula)). Furthermore, most parsimonius trees. This holds true even the characters that support Aetobatus ϩ Ae- when character 37 (describing the different tomylaeus in Nishida's phylogeny also occur insertion patterns of the spiracularis muscle) for Rhinoptera ϩ Mobulidae, so that on is run as an ordered multistate feature. Nishida's tree more than one optimization Relationships within Potamotrygonidae, exists for these features (they could also be however, are resolved identically as in the interpreted as synapomorphies for all pelagic analysis of Lovejoy (1996), with Potamotry- stingrays, lost in Myliobatis). gon and Plesiotrygon as sister-groups, con- The major difference between our phylog- trary to the hypothesis of Rosa (1985) and eny and the trees of Lovejoy (1996) and Rosa et al. (1987). We concur with Lovejoy's McEachran et al. (1996) is the recognition of reasoning for excluding two of the characters a large component uniting all dasyatid genera used by Rosa to unite Potamotrygon and and Gymnuridae ϩ Myliobatidae. The posi- Paratrygon (modal numbers of pectoral ®n tion of the clade Gymnuridae ϩ Myliobati- radials and pelvic ®ns covered by pectoral dae as a monophyletic group nested within disc). These characters are dif®cult to code, successive higher level taxa of benthic sting- and more comparative data are needed before rays is common to all stingray phylogenies numbers of pectoral ®n radials can be con- generated to date (cf. appendix 2). Differ- ®dently used. The proximal fusion of the ®rst ences between these phylogenies and ours and second ceratobranchials also used by are the result of the inclusion of different Rosa (1985) and Rosa et al. (1987) to unite taxa and characters, as well as alternative in- Potamotrygon to Paratrygon occurs in other terpretations for certain features. Lovejoy taxa, such as gymnurids and many myliob- (1996) and McEachran et al. (1996) placed atids, and in at least one species of Urotry- Dasyatis (including Indo-west Paci®c Hi- gon and Urolophus (Miyake and McEachran, mantura in Lovejoy, 1996) as sister-group to 1991). However, this character is a putative Gymnuridae ϩ Myliobatidae, a grouping not synapomorphy of both genera (requiring ho- at odds with our phylogeny. These authors, moplasy), but it must be included in a more however, also support a group comprising comprehensive species-level phylogeny of Taeniura, amphi-American Himantura, and stingrays. the Potamotrygonidae, a relationship contra- Our phylogenetic analysis reinforces that dicted by our results. The characters ad- tree topology can be dramatically altered vanced by these authors supporting this re- with only small modi®cations in character lationship were also included in our analyses coding, or with the addition of characters or (relating to the angular cartilages and pat- terminals, when taxa exhibit such high levels terns of insertion of the spiracularis muscle). of homoplasy. For instance, scoring character The results of Lovejoy (1996) concerning 44 (nasal curtain) identically to the coding potamotrygonid relationships are not veri®ed given in McEachran et al. (1996; with Ple- here not only because of differential coding siobatis displaying the primitive condition of of certain features, but also because we im- having a shortened nasal curtain) resolves plemented our analyses without ordering Plesiobatis as sister-group to Urolophus ϩ multistate characters, a prerequisite to Trygonoptera and alters resolution elsewhere achieve the topology advocated by Lovejoy. on our phylogeny (²Asterotrygon is then When his analysis is run without ordering placed in a polytomy at the node with ²He- multistate characters, the results obtained are liobatis, potamotrygonids, dasyatids, gym- 104 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 nurids, and myliobatids). We have coded Ple- topology presented in appendix 2. To date, siobatis differently but accurately in our as- however, there have been no unambiguous sessment (with a nasal curtain reaching very characters discovered and correctly included close to the mouth; also J.D. McEachran, in a phylogenetic analysis to support dasyatid personal commun.). Another example of this monophyly (but see appendix 2). sensitivity emerges if pelagic stingrays are Removing the Green River fossils from scored as a single terminal (as Myliobatidae), our matrix reduces resolution considerably. as opposed to the inclusion of various my- Parsimony analysis of the matrix in table 5 liobatid genera in the matrix (as in our ®nal without ²Asterotrygon and ²Heliobatis re- matrix of table 5; this experimental analysis sulted in 273 trees (length ϭ 80 steps, CI ϭ is presented in appendix 2). This form of 0.70, RI ϭ 0.82), and the strict consensus ``groundplan'' (as opposed to `èxemplar'') (length ϭ 86 steps, CI ϭ 0.65, RI ϭ 0.78) coding seriously alters the relationships does not support a monophyletic Urolophi- among benthic genera that are more distantly dae or the large clade Potamotrygonidae ϩ related to pelagic stingrays (i.e., that are not (``Dasyatidae'' ϩ (Gymnuridae and Myliob- their direct sister-groups), perhaps unexpect- atidae)), the major contribution to stingray edly (for theoretical considerations, see Bin- phylogeny presented in this study; dasyatid inda-Emonds et al., 1998; Wiens, 1998, genera are placed in a polytomy with pota- 2000; Kornet and Turner, 1999; Prendini, motrygonids. The other components of our 2001; Simmons and Geisler, 2002). Elasmo- phylogeny in ®gure 43 remain unaltered. branchs are notorious for elevated levels of Successive approximations weighting of this character con¯ict (see, for example, the phy- modi®ed matrix resulted in 123 trees (length logenies of Naylor, 1992; Shirai, 1992; ϭ 474 steps, CI ϭ 0. 88, RI ϭ 0.93), and the McEachran et al., 1996; McEachran and strict consensus of these (length ϭ 478 steps, Dunn, 1998; cf. Carvalho, 1996a), and sting- other tree indices remain unchanged) adds rays are no exception. Coding must be done two clades: (1) Urolophidae ϩ Pteroplatytry- cautiously if taxa are to be combined into a gon; (2) Dasyatis, Himantura sensu stricto ϩ single family-level entry (which should be (Gymnuridae ϩ Myliobatidae). The former avoided if possible), either because of inad- clade (1) is highly counterintuitive; amphi- equate sampling or for the sake of simplify- American Himantura, Taeniura and pota- ing tree construction, as this single entry motrygonids remain in a polytomy. The im- will, undoubtedly, contain a high degree of pact of ²Heliobatis on the relationships character variation (taxonomic polymor- among extant stingrays is greater than that of phism in the sense of Nixon and Davis, ²Asterotrygon; when only ²Asterotrygon was 1991). This is precisely the case of pelagic removed, relationships were identical to stingrays in the analysis of appendix 2 re- those presented in ®gure 43 (derived from garding the hyomandibular-Meckelian liga- the matrix in table 5, with both fossil taxa ment. When the single terminal Myliobatidae included). Implementing the analysis solely was coded as having the ligament in that without ²Heliobatis produced the highly analysis (appendix 2), on the basis of this modi®ed and more ambiguous results de- being the inferred ancestral state (`ÌAS cod- scribed above. The in¯uence on extant sting- ing'' of Rice et al., 1997), resolution was ray phylogeny of the Green River stingrays substantially decreased in the strict consen- is therefore substantial, but differs for both sus, and the number of equally most parsi- fossil genera. The more dramatic changes monious trees increased considerably. Simi- caused by the inclusion of ²Heliobatis stems larly, adding to our matrix (table 5) only two from its additional missing entries in the ma- characters with potential to unite dasyatid trix (²Heliobatis has three more unknown genera would be enough to support a mono- entries than ²Asterotrygon); both taxa differ phyletic Dasyatidae in all minimum-length otherwise in only one additional feature (dor- trees obtained. A monophyletic Dasyatidae, sal ®n, character 33). The different in¯uence in turn, dramatically affects the relationships of ²Asterotrygon and ²Heliobatis on stingray of other stingray genera in the analysis, and phylogeny is therefore determined by this the outcome would be more similar to the particular combination. This is further evi- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 105 dence that the impact a fossil taxon may have yake, 1988; see character description above). on a phylogeny of living forms depends Dissection of additional taxa is still needed. more on its speci®c combination of features Deactivating this character in our matrix than on any other factor (Gauthier et al., does not alter the strict consensus, but it re- 1998; Donoghue et al., 1989; Huelsenbeck, duces the number of equally most parsimo- 1991; Novacek, 1992; Graybeal, 1998; nious trees to 28 (length ϭ 79 steps, CI ϭ O'Leary and Geisler, 1999 [cf. Naylor and 0.68, RI ϭ 0.82; strict consensus tree: length Adams, 2001]; Kearney, 2002; contra Patter- ϭ 79 steps, CI ϭ 0.65, RI ϭ 0.79). son, 1981; Rosen et al., 1981). Previous authors have advocated that Das- yatis and Himantura may not be monophy- SYSTEMATIC ISSUES WITHIN letic genera, a proposition that we have only MYLIOBATIFORMES partially tested in our analysis (Compagno and Roberts, 1982; Last and Stevens, 1994; The hyomandibular-Meckelian ligament Lovejoy, 1996; Rosenberger, 2001b). Himan- requires further comparisons within the My- tura, on morphological grounds alone, liobatidae. The ligament is present in three should be divided into separate genera, one of the seven genera of pelagic stingrays (My- for amphi-American species (H. schmardae liobatis, Aetobatus, and Rhinoptera), but its and H. paci®ca) and one for Himantura sen- presence in Myliobatis is based on con¯ict- su stricto for Indo-west Paci®c species ing evidence (Garman, 1913; Nishida, 1990). (Lovejoy, 1996). Both amphi-American spe- The condition in Pteromylaeus is unknown. cies share unique specializations of the der- Lovejoy (1996) excluded this character from mal skeleton, disc proportions, and shape, his analysis, but he was in¯uenced by the and perhaps even of the caudal sting (Car- ambiguous account given by Nishida (1990) valho and Lovejoy, in prep.). They were con- concerning the condition in Plesiobatis. The sequently coded as two separate terminals in hyomandibular-Meckelian ligament was in- our matrix (following Lovejoy, 1996), and cluded in the matrix of McEachran et al. they do not form a monophyletic Himantura (1996), who correctly coded Plesiobatis as on any of the minimum-length trees ob- having the ligament, the condition present in tained. There are characters that may poten- benthic stingrays except Hexatrygon and tially unite species of Himantura sensu stric- gymnurids. to (Indo-west Paci®c species) as a monophy- Dissection of various stingrays revealed letic unit, such as arrangement of their nasal that the spiracularis muscle is more complex capsules (which are anteroposteriorly abbre- than was indicated in previous interpretations viated), anterior projection of the scapular (Lovejoy, 1996; McEachran et al., 1996), in- process (e.g., Nishida, 1990: ®g. 30f), and a cluding our own observations summarized in narrow, anteriorly rounded pelvic girdle. We the matrix (table 5), which largely follows have observed these features in various spe- Lovejoy. We included the character some- cies of Himantura, but they vary among spe- what tentatively to see how it would affect cies to a certain degree, and in order to verify the relationships of Taeniura, amphi-Ameri- this claim they must be included in a more can Himantura, and potamotrygonids, hy- comprehensive species-level matrix. pothesized by Lovejoy as forming a mono- The situation concerning Dasyatis is more phyletic group (it does not support this node complex. This genus presently includes from in our analysis, even when run as ordered; 35 to 40 species (Compagno, 1999), many of see above). The spiracularis of these taxa which are poorly known morphologically. (except Paratrygon) is indeed robust and Rosenberger (2001b) attempted the ®rst cla- projects posteroventrally, more so than in distic test of its monophyly. Her phylogenet- most other stingrays examined. However, ic analysis included 13 Dasyatis species (15 Pteroplatytrygon and some species of Dasy- if both Pastinachus sephen and Pteroplaty- atis may have the spiracularis inserting pos- trygon violacea are considered in Dasyatis, terior to the lower jaws and somewhat con- as she did), in addition to ®ve species be- tinuous with the depressor hyomandibularis longing to the genera Gymnura, Taeniura, muscle as well (also Kesteven, 1942; Mi- amphi-American Himantura, Himantura sen- 106 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 su stricto, and Urobatis. Her resulting phy- closely related to the clade Gymnuridae ϩ logeny nests H. gerrardi, Gymnura micrura, Myliobatidae (see comments on swimming Pteroplatytrygon, and Pastinachus within patterns below). Rosenberger's analysis is a Dasyatis. Gymnura micrura clearly does not step in the right direction, but in addition to belong in Dasyatis, as would have been con- complementing her matrix with more spe- cluded had more species of Gymnura been cies, morphological characters that are more coded and their many generic synapomor- reliable at the species level must be included, phies included in the matrix. We included and these have been highly elusive so far Pteroplatytrygon as a separate entry in our (i.e., characters that are constant within spe- analysis because it differed in relation to the cies but that vary among them). postorbital process of the neurocranium Subdividing the larger benthic genera into (character 9), a feature that did not vary geographical components, if these are dem- among species of Dasyatis examined (Pter- onstrated to be monophyletic (as is the case oplatytrygon is also considered valid in re- with amphi-American Himantura), may al- cent compilations, e.g., Compagno, 1999; leviate dif®culties in coding generic-level McEachran and Carvalho, 2002). Rosenber- matrices, as an all-inclusive species-level ger's analysis was partially based on features matrix is still too prohibitive at present. This that are dif®cult, if not impossible, to divide is precisely what has transpired in the sys- into discrete states because their variation is tematics of skates, in which very large, tra- either too subtle or continuous (cf. Rae, ditional taxonomic units (e.g., Raja) have 1998) (such as her character 1, snout length been subdivided into putative monophyletic less than 25% of disc length [state 0], greater groups that were previously given subgeneric than 25% [1]; character 14, midregion of ranking. These ``subgenera'' are now recog- frontoparietal fontanelle greatly constricted nized as separate genera that are geographi- and narrow [0], moderately constricted [1], cally more restricted (McEachran and Mi- wide [2]; character 15, anterior margin of yake, 1990; McEachran and Dunn, 1998). frontoparietal fontanelle rounded [0], slightly Geographically restricted monophyletic sub- rounded [1], straight [2]; character 24, me- units may exist within Dasyatis, but this hy- sopterygium wide [0], narrow [1]). The fron- pothesis, contradicted by Rosenberger toparietal fontanelle (her character 14), for (2001b), remains to be fully explored. example, may be greatly constricted and Our observations indicate that Taeniura is wide (e.g., in Potamotrygon; ®g. 34A), and probably monophyletic (note that our matrix the difference between a wide and narrow did not test Taeniura monophyly), as all mesopterygium in many species of Dasyatis three species have the ®rst (or anterior) pair may be trivial or somewhat arbitrary (the of hypobranchials conspicuously cleaver- or mesopterygium can even vary slightly in club-shaped. Species of other stingray genera length and width on both sides of the same may have hypobranchials in which the an- specimen, e.g., Dasyatis pastinaca AMNH terior portion is wider than the posterior half, 32796). Other characters vary among speci- but in Taeniura the anterior portion is con- mens within certain species according to our spicuously broad, with the broad segment ex- data (her character 27, number of metapter- tending posteriorly to at least one-half of hy- ygial segments; e.g., in Dasyatis margarita, pobranchial length (therefore extending pos- D. zugei, Urobatis jamaicensis; of course, teriorly farther in comparison to other genera this does not necessarily imply that they vary in which the ®rst hypobranchial has a broad among Rosenberger's terminals) or they are anterior segment). The anterior hypobranchi- simply not very clear (her character 31, als also have slightly concave internal (me- ``posterior portion of pelvic gridle arch''). dial) margins in Taeniura. These elements Rosenberger's conclusions regarding the va- are particularly similar in Taeniura lymma lidity of Pteroplatytrygon and Pastinachus (AMNH 44077) and T. grabata (MNHN may eventually prove to be correct, but the 1989±1793). The anterior hypobranchials of former genus did not pair with Dasyatis in T. meyeni are also markedly broad (Nishida, any of our minimum-length trees, and there 1990) and resemble the hypobranchials of is even a hint of evidence that it may be more the other species of Taeniura more than 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 107 those of other stingrays. More specimens of Trygonoptera examined (in ®ve of the six T. meyeni are needed for further compari- species recognized by Last and Stevens, sons. Other features that may be indicative 1994) and accept it as evidence of its mono- of Taeniura monophyly are the shape of the phyly along with the ¯eshy lateral nasal basihyal (uniquely anteroposteriorly extend- lobes. Yearsley (1988) arrived at similar con- ed, about two-thirds as long as wide), the an- clusions and presented more (and different) terior projection of the hyomandibulae (this skeletal evidence distinguishing both genera. process is set at an angle to the main ramus The degree of development of the tail- of the hyomandibula, accommodating the folds may still be of some value in delimiting hyomandibular-Meckelian ligament; marked groups within dasyatids, but a comprehen- with an asterisk in ®g. 35A), and the arrange- sive species-level phylogeny is necessary to ment of the ceratobranchials (in which the evaluate this character more thoroughly, and forward extensions that contact the anterior more information is needed for many sting- ceratobranchial element are particularly slen- ray species before this can be attempted. Our der and elongated). These features require analysis demonstrates that lack of tail-folds con®rmation on additional specimens; they occurs independently for at least four sting- may be autapomorphies of Taeniura lymma, ray groups (Gymnuridae ϩ Myliobatidae, synapomorphies of Taeniura, or prove un- Paratrygon, amphi-American Himantura, reliable due to intraspeci®c variation. and Himantura), and that possession of only Trygonoptera MuÈller and Henle, 1841 has ventral tail-folds appears, when mapped on been considered a junior synonym of Uro- our trees, independently at least twice (for lophus MuÈller and Henle, 1831 by some pre- Taeniura and Plesiotrygon) but probably vious authors (e.g., Whitley, 1940; Mc- more times because certain species of Das- Eachran, 1982; Nishida, 1990), but others yatis also lack dorsal tail-folds (presenting have recognized it as a valid genus (Bigelow only a dorsal keel in its place; Garman, 1913; and Schroeder, 1953; Last and Gomon, 1987; Bigelow and Schroder, 1953; Nishida and Last and Stevens, 1994; Compagno, 1999; Nakaya, 1990). As there is much variation in Last and Compagno, 2000). Although they the pattern of tail-folds within Dasyatis can be distinguished on the basis of external (some authors have used Amphotistius Gar- morphology (primarily by the ¯eshy nasal man, 1913 for species with both dorsal and lobe in Trygonoptera), skeletal features fur- ventral tail-folds; e.g., Paxton et al., 1989), ther separate both genera, particularly those this character cannot be fully evaluated until from the neurocranium. Trygonoptera has a species-level phylogeny is available. In the infraorbital lateral-line canal passing be- fact, because many characters appear, disap- tween the postorbital process and the more pear, and reappear in stingray evolution, anterior lateral projection (supraorbital pro- amounting to high levels of character varia- cess), while Urolophus has the canal passing tion within genera, a species-level phylogeny through a large foramen in the postorbital is necessary to fully evaluate their potential process per se as in most myliobatids (char- for grouping (e.g., arrangement of orbital fo- acter 9 in our analysis). The dorsal internasal ramina, degree of development of lateral septum is also somewhat distinct in both gen- stays of ®rst synarcual, extent of internasal era, being relatively wider in Trygonoptera septum, patterns of subdivision of the prop- (similar to Plesiobatis; however, the phylo- terygium). genetic signi®cance of the dorsal internasal The phylogenetic analysis presented here septum has yet to be fully explored). Both allows for some consideration of swimming genera also differ in character 26 (fossa on in stingrays, which is accomplished by the dorsal scapular region, present in Trygonop- following patterns (or modes, Rosenberger, tera but absent in Urolophus). Furthermore, 2001a): undulatory (wavelike, with anterior in Trygonoptera the supraorbital process is to posterior motions of the disc), oscillatory particularly elongated compared to other (birdlike, with ¯apping of the disc), or a genera in which it is present (genera with combination of these (the `ìntermediate'' state 0 for this character, listed above). We condition of Rosenberger, 2001a). Even have observed this feature in all species of though the extreme manifestations of ``wave- 108 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 like'' and ``birdlike'' swimming are quite lytomies of the strict consensus, resulting in distinct (such as between a skate and a manta 20 equally most parsimonious trees (strict ray, respectively), there is a large gray zone consensus in ®g. 47). The relationships of of in-between conditions (the undulation/os- ²Asterotrygon are not altered, but three cillation continuum of Rosenberger, 2001a), (Pteroplatytrygon, Dasyatis, Himantura)of rendering the divisions into discrete states the ®ve dasyatid genera form an additional somewhat arbitrary. Nonetheless, adding this node with the Gymnuridae ϩ Myliobatidae character to our matrix (i.e., coding swim- clade, which conforms to the results of Love- ming patterns) resulted in 107 equally most joy (1996) and McEachran et al. (1996). Im- parsimonious trees (length ϭ 85 steps, CI ϭ plementing the weighting scheme of Golo- 0.68, RI ϭ 0.82) and a strict consensus boff (1993; using implied weights, with his (length ϭ 89 steps, CI ϭ 0.65, RI ϭ 0.79) program Piwe) produces results identical to identical to the tree obtained without this those obtained from successive approxima- character (®g. 43, from the matrix in table tions weighting. The following sequenced 5). However, after successive approximations classi®cation of stingrays is derived from our weighting was applied to this dataset (with study of stingray phylogeny: swimming as a character, resulting in 33 trees: length ϭ 489 steps, CI ϭ 0.89, RI ϭ Order Myliobatiformes 0.93), the resulting consensus tree (length ϭ Suborder Hexatrygonoidei 491 steps, CI ϭ 0.88, RI ϭ 0.93) supported Family Hexatrygonidae (Hexatrygon Heem- the placement of Pteroplatytrygon as sister- stra and Smith) group to the clade containing Gymnuridae Suborder Myliobatoidei and Myliobatidae. We obtained this result Suborder Myliobatoidei incertae sedis (²Aster- with weighting when coding Pteroplatytry- otrygon, n.gen.) gon identical to Gymnura (i.e., with state 1, Superfamily Urolophoidea, new combination intermediate) or to myliobatids (state 2, os- Family Urolophidae (Urolophus MuÈller cillatory). Even though optimization-depen- and Henle, Trygonoptera MuÈller and dent, this indicates that the oscillatory swim- Henle) ming mode indicative of a pelagic lifestyle Superfamily Plesiobatoidea Family Plesiobatidae (Plesiobatis Nishida) may have arisen only once within stingrays Superfamily Urotrygonoidea, new combina- (with gymnurids subsequently settling back tion into a benthic niche), and not twice as ad- Family Urotrygonidae (Urotrygon Gill, vocated by Rosenberger (2001a). Rosenber- Urobatis Garman) ger's (2001a) study of swimming patterns Superfamily Myliobatoidea, new combina- within stingrays revealed that there may be tion additional characters for Pteroplatytrygon Family Heliobatidae (²Heliobatis Marsh) and myliobatids (and gymnurids), but these Family Potamotrygonidae (Potamotrygon were not included in our modi®ed matrix be- Garman, Paratrygon DumeÂril, Plesi- cause they either appear to be correlated to otrygon Rosa, Castello, and Thorson) the oscillatory swimming mode (and there- Genera incertae sedis (Dasyatis Ra®n- fore not independent) or we could not yet esque [including Urolophoides Lind- con®rm them in dissections (e.g., greater pro- berg], Himantura MuÈller and Henle portion of red muscle ®bers in the disc as [Indo-west Paci®c species], Pastinachus opposed to white [red ®bers are more im- RuÈppell, Pteroplatytrygon Fowler, Uro- gymnus MuÈller and Henle, Taeniura portant for slow, steady swimming in pelagic MuÈller and Henle, `àmphi-American forms], angle of articulation between scapu- Himantura'') lar processes and suprascapulae). Family Gymnuridae (Gymnura Kuhl, Ae- toplatea Valenciennes) CLASSIFICATION OF MYLIOBATIFORM GENERA Family Myliobatidae (Myliobatis Cuvier, Aetomylaeus Garman, Pteromylaeus Successive approximations weighting Garman, Aetobatus Blainville, Rhinop- (Farris, 1969; Carpenter, 1988, 1994; Fitz- tera Kuhl, Mobula Ra®nesque, Manta hugh, 1989) further resolves one of the po- Bancroft) 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 109

overwhelming majority of these are too fragmentary to be reliably included in a phylogenetic analysis, a more complete historical appreciation of their distribution may be beyond our comprehension.

ORIGIN OF THE GREEN RIVER STINGRAYS None of the equally most parsimonious trees recovered in our phylogenetic study allows for a single invasion of Fossil Lake by an ancestral stingray form which subsequently underwent speciation to result in the two known stingray genera (see area cladogram in ®g. 48). In other words, ²Asterotrygon and ²Heliobatis do not from a monophyletic unit in any of the resulting trees (®gs. 43, 45), nor do they form successive sister-groups to the node Potamotrygonidae ϩ (``Dasyati- dae'' ϩ (Gymnuridae ϩ Myliobatidae)), a topology that would still allow for a single invasion of Fossil Lake by an ancestral stingray taxon. Quite the contrary, the Green Riv- er stingrays are phylogenetically placed far apart from each other in all of the minimum- length trees recovered. It may seem counterintuitive that more than one stingray lineage entered and successfully adapted to the freshwater environment of Fossil Lake, especially given that the lake is hypothesized to have Fig. 47. Strict consensus tree (length ϭ 477 existed for ``only'' some 5 million years steps, CI ϭ 0.87, RI ϭ 0.92) of the 20 equally (which is not insigni®cant if compared to the most parsimonious trees (length ϭ 472 steps, CI much more species-diverse and younger ϭ 0.88, RI ϭ 0.93) that resulted after successive Lake Victoria in eastern Africa, however; cf. approximations weighting. The marked node indicates the extent of further resolution obtained Verheyen et al., 2003), and that Fossil Lake with weighting (compare with phylogeny in ®g. is considered to have been of modest size 43). An identical tree was produced with implied throughout its history by most estimates weights (see text for discussion). compared to the other extinct Green River lakes (even though its size ¯uctuated through time, Fossil Lake was never much greater BIOGEOGRAPHICAL IMPLICATIONS than perhaps 70 km north±south by 35 km The myliobatiform phylogeny presented in west±east; Grande and Buchheim, 1994). But this study (®g. 43) allows for some consid- the extensive actinopterygian fauna of Fossil eration of stingray distribution through time, Lake is by no means monophyletic either, nor particularly in relation to the origin of the is it expected to be. ²Asterotrygon and ²He- freshwater stingrays of the Green River For- liobatis are presently unknown from the oth- mation and the origin of South American po- er extinct lakes of the Green River lake com- tamotrygonids, events that are completely in- plex (Lake Uinta and Lake Gosiute), so their dependent. A more comprehensive historical presence in Fossil Lake cannot be considered biogeographic analysis of Myliobatiformes a remnant of a previously more widespread would have to take into account the myriad continental distribution. The presence of widespread fossil stingray taxa that have stingrays in the Green River Formation is been described (see Introduction). As the therefore more similar to the separate fresh- 110 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 48. Area cladogram derived from phylogeny (®g. 43) of stingrays (Myliobatiformes) presented in this paper (see Biogeographical Implications for discussion). The sequence of branches (mostly family-level terminal taxa) from left to right is as follows: Hexatrygonidae, ²Asterotrygon, Plesiobatidae, Urolophidae, Urotrygonidae, ²Heliobatis, Potamotrygonidae, Dasyatidae, Gymnuridae, Myliobatidae. Asterisks (*) indicate phylogenetic position and distribution of Green River stingrays. M ϭ marine; FW ϭ freshwater; B ϭ brackish water. water colonizations by endemic living spe- nuridae ϩ Myliobatidae; all other stingray cies of Dasyatis and Himantura in Africa, genera form a large polytomy together with Australia, and southeast Asia, and it is not all of these clades. The only character that is analogous to the single (monophyletic) po- optimized as uniting ²Asterotrygon and ²He- tamotrygonid radiation in South American liobatis in the strict consensus of this con- freshwaters. strained analysis is the absence of the dorsal A phylogenetic analysis of our matrix (ta- fossa of the scapular process, which is hy- ble 5) with ²Asterotrygon and ²Heliobatis pothesized to have been lost in their common constrained as a monophyletic unit resulted ancestor (i.e., its presence united all stingrays in 58 minimum-length trees that are only a except Hexatrygon in the constrained analy- single step longer than our original trees sis). Evidence for the monophyly of the summarized in ®gures 43±45, 47 (this anal- Green River stingrays is consequently mea- ysis was run in Nona; length ϭ 83 steps; CI ger (individual vertebrae extending to the ϭ 0.67, RI ϭ 0.81). However, resolution in distant tip of the tail is another putative syn- the strict consensus is dramatically decreased apomophy, but is not optimized in this fash- in this constrained analysis, with only a few ion in all minimum-length trees obtained in nodes present: all stingrays except Hexatry- this analysis). We are therefore con®dent, gon, Urotrygonidae, Green River stingrays, given our present understanding of their mor- Potamotrygonidae, and a node uniting Gym- phology, that the genera of Green River 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 111 stingrays are not sister-groups, which un- chthyidae (ϭ ²Paraclupeidae; Chang and equivocally discards the possibility that they Grande, 1997), Catostomidae, and, to a lesser evolved within Fossil Lake from a common extent, the Percopsidae (²Amphiplaga; Pat- ancestor. Note that their phylogenetic posi- terson and Rosen, 1989; cf. Murray and Wil- tions (®gs. 43, 48) still allow for a single son, 1999). This trans-Paci®c component has invasion of Fossil Lake (such as through a been demonstrated, therefore, for groups that common biogeographical event), but both are strictly freshwater (e.g., paddle®shes, Po- stingray genera (or their direct ancestors) lyodontidae, known only from freshwater or were phylogenetically separated when the freshwater deposits in China and North colonization of Fossil Lake occurred. America; Grande and Bemis, 1991) as well The phylogenetic placement of ²Asterotry- as from those that are primarily marine (such gon close to the base of our strict consensus as stingrays). The area relationships imposed tree (®g. 43), or its placement as the sister- by the putative ²Asterotrygon ϩ Urolophidae group to urolophids (Urolophus ϩ Trygon- clade recovered in most of our minimum- optera) in 21 of the original 35 minimum- length trees is congruent with area relation- length trees (®g. 45C±E), is an indication of ships among some species of ²Phareodus,in a strong Indo-west Paci®c association. Both which the Australian ²P. queenslandicus is Hexatrygon and Plesiobatis are widely (but considered to be the sister-group to the Green sporadically) distributed in the Indian and River ²P. encaustus (Li, 1994; Li et al., Paci®c Oceans (Last and Stevens, 1994). 1997). Urolophids (Urolophus, Trygonoptera) are The phylogenetic position of ²Heliobatis, somewhat more restricted, occurring off on the other hand, is indicative of a more southern Japan and Korea (East China Sea) recent historical af®nity with the Americas and in the South China Sea, but they are (®g. 48), possibly (but not necessarily, see more diverse and common around Australia above) indicating an independent stingray in- (Last and Stevens, 1994). Because the most vasion of Fossil Lake. The clade basal to basal living stingrays are all Indo-west Pa- ²Heliobatis in our stingray phylogeny (®g. ci®c (with more or less restricted distribu- 43) is Urobatis ϩ Urotrygon, and both gen- tions), and ²Asterotrygon is nested among era are distributed in North and Central them in our phylogeny (®gs. 43, 45; area America with a few species reaching as far cladogram in ®g. 48), it is reasonable to infer south as northern South America (Meek and that an Indo-west Paci®c relationship is im- Hildebrand, 1923; Bigelow and Schroeder, plied by ²Asterotrygon. This historical rela- 1953; Miyake and McEachran, 1986; Mc- tionship is even more geographically restrict- Eachran, 1995). The node above ²Heliobatis ed in the trees in which ²Asterotrygon forms in our tree is divided into two clades, one a clade with urolophids (the majority of our containing species distributed in South equally most parsimonius trees). This hy- American freshwaters (Potamotrygonidae), pothesis would not be contradicted if sting- and another, the most species-diverse of all rays even more basal to Hexatrygon are dis- stingray lineages (``Dasyatidae'' ϩ (Gym- covered. nuridae ϩ Myliobatidae)), with species pres- A transoceanic Indo-Paci®c historical as- ently occurring in most temperate and trop- sociation has also been recognized for other ical marine regions, as well as some eight Green River Formation ®shes (Grande, 1985, exclusively freshwater species (in the genera 1989, 1994; Li et al., 1997), as well as for Dasyatis and Himantura in Africa, southeast its fossil plants (MacGinitie, 1969). In fact, Asia, and Australia). Our phylogeny there- most Green River ®shes that have been in- fore is congruent with the hypothesis that the cluded in phylogenetic studies indicate a largest radiation of species of stingrays (the trans-Paci®c relationship (Grande, 1985, clade ``Dasyatidae'' ϩ (Gymnuridae ϩ My- 1994), as corroborated by independent stud- liobatidae)) originated only after an Ameri- ies of the Polyodontidae (²Crossopholis; can stingray lineage was established (®g. 48). Grande and Bemis, 1991) and the teleosts This early American stingray component Hiodontidae, Osteoglossidae (²Phareodus; may have shared an ancestry with the species Li et al., 1997), Pellonulinae, ²Ellimmi- of stingrays described from Cretaceous and 112 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Paleocene formations from central and east- the subject of much debate and has conse- ern North America. Many of these stingrays quently become a benchmark problem in his- inhabited the Late Cretaceous Seaway, a torical biogeography (Brooks et al., 1981; large inland sea that occupied an immense Rosa, 1985; Brooks and McLennan, 1993; trough in western North America that at one Brooks, 1995; Lovejoy, 1996, 1997; Hoberg point extended from present-day Gulf of et al., 1998; Marques, 2000). The discussions Mexico to Alaska, with a maximum width of concern primarily the sister-group of the Po- some 2000 km (Smith et al., 1994; cf. Fun- tamotrygonidae and the nature of the data nell, 1990; see ®g. 52). These stingray spe- used to infer potamotrygonid origin or af®n- cies are known from isolated teeth, including ity (for a more detailed summary of the dif- forms from the early Late Cretaceous (Cen- ferent points of view, see Lovejoy, 1996, omanian) of Texas (Meyer, 1974) and Late 1997). The theory advanced by Lovejoy Cretaceous of Wyoming (Estes, 1964) and (1996, 1997) and Lovejoy et al. (1998), that Texas (Welton and Farrish, 1993). The sting- potamotrygonids originated from a common rays described from the Late Cretaceous of ancestor with amphi-American Himantura New Jersey (Monmouth group; Cappetta and from the Caribbean Sea by way of an early Case, 1975) occurred at the opposite (east- Miocene epicontinental seaway (the ``proto- ern) margin of the North American continent Caribbean/Himantura'' hypothesis), was re- during the Maastrichtian (®g. 52), but rep- cently given support by a molecular phylo- resent a very similar if not identical stingray genetic analysis of parasites and their pota- fauna, as this coastline was probably contig- motrygonid hosts by Marques (2000). This uous with the Late Cretaceous Seaway over contrasts with the earlier theory of Paci®c central North America. The Paleocene sting- origin of the family by means of a phylo- rays from the Cannonball Formation, which genetic relationship with ``urolophids'' (ϭ inhabited the last marine incursion (the Can- Urotrygonidae), with the potamotrygonid nonball Sea) into central North America common ancestor having been con®ned to (Cvancara and Hoganson, 1993), may also be freshwater as a result of the orogeny of the related to this early American stingray line- Andean Cordilleras, an ongoing process that age. These stingrays (known only from iso- started in the Late Cretaceous some 90 mil- lated teeth) include two species of Dasyatis lion years ago (Brooks et al., 1981; Rosa, (note that one tooth referred by Cvancara and 1985; Brooks and McLennan, 1993; Brooks, Hoganson [1993: ®g. 3, v±x] to their new 1995). The parasite data from which much of species ²D. concavifoveus probably repre- the latter theory is founded are beyond our sents a gymnurid), ²Hypolophodon sylves- concern but have been extensively criticized tris, and a myliobatid (see also Best, 1987). by Straney (1982), Caira (1990), and more Unfortunately, it is not possible to compare recently by Lovejoy (1997; cf. Hoberg et al., any of the above taxa in a meaningful way 1998), who provided an arrant discussion of to the Green River stingrays (and more pre- problems pertaining to Brooks et al.'s (1981) cisely to ²Heliobatis) to determine if they data, host/parasite coevolution in general, share a more recent common ancestry. How- and the implicit assumptions of their ``Pacif- ever, it is certainly conceivable that such a ic/urolophid'' hypothesis. This hypothesis is relationship existed, which would indicate completely contradicted by our phylogenetic that an ancestral stingray taxon originally results (and previously by those of Lovejoy, present in the Late Cretaceous Seaway, or in 1996) and will not be further explored here some other marine incursion into North (cf. Rosa, 1985). All of the above authors, America, is historically correlated with ²He- however, are in agreement that the Potamo- liobatis of Fossil Lake. trygonidae is monophyletic (even though their different parasite lineages may not be), ORIGIN OF THE SOUTH AMERICAN a crucial point that we fully corroborate and FRESHWATER STINGRAYS which is congruent with the theory of a single freshwater invasion into South America The origin of the South American fresh- by the potamotrygonid ancestor. Our focus water stingrays (Potamotrygonidae) has been on the problem here stems solely from infer- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 113 ring potamotrygonid biogeography on the curred as a result of the formation of the Pan- basis of their phylogenetic relationships, amanian Isthmus in the Pliocene, after the which in our view takes precedence over data initial divergence between amphi-American concerning the af®nities of their parasites Himantura and the Potamotrygonidae. Love- (also Straney, 1982; Lovejoy, 1997). Our joy et al. (1998) further supported this spe- stingray phylogeny (®g. 43) has prompted us ci®c marine-incursion hypothesis and gave it to reevaluate the origin of the potamotrygon- a more precise Miocene timeline, by analyz- ids, coupled with a radically different assess- ing rates of divergence of the gene-encoding ment of their minimum age. mitochondrial cytochrome b, DNA sequenc- The phylogenetic relationships of pota- es from which were used to corroborate the motrygonids advocated by Lovejoy (1996) amphi-American Himantura ϩ Potamotry- have been reviewed above and compared to gonidae clade. The divergence rate inferred our results (cf. ®gs. 43 and 46B). His novel for the cytochrome b gene was calibrated clade Taeniura ϩ (amphi-American Himan- based on the orogeny of both the eastern cor- tura ϩ Potamotrygonidae), the basis of his dillera of the Andes and the Merida Andes biogeographic hypothesis for potamotrygon- (which took place some 8 million years ago ids, was not present in any of our equally [mya]; Lovejoy et al., 1998). The resulting most parsimonius trees (even when the mul- molecular clock estimate placed the origin of tistate character pertaining to the insertion of the Potamotrygonidae at between 15 and 23 the spiracularis muscle was run as ordered, a mya, in early Miocene to late Oligocene constraint necessary to obtain this clade in times (Lovejoy et al., 1998). These results his analysis). Instead, we found support for were further supported by the molecular phy- a clade comprising Potamotrygonidae ϩ logenetic analysis of Marques (2000), based (Taeniura, amphi-American Himantura, mostly on mitochondrial rDNA sequences Pteroplatytrygon, Dasyatis, Himantura ϩ (12S, 16S, CO I, and again cyt-b). Marques' (Gymnuridae ϩ Myliobatidae)), with some molecular clock estimates placed the origin evidence (implementing the weighting of this clade at approximately 19 mya, well schemes of Farris, 1969 and Goloboff, 1993) within the mostly early Miocene range of for an additional node uniting Pteroplatytry- Lovejoy et al., but his consideration of mo- gon, Dasyatis, Himantura ϩ (Gymnuridae ϩ lecular clock con®dence limits allowed for Myliobatidae)). the potamotrygonid original divergence to The amphi-American Himantura ϩ Pota- have occurred from between 6 and 38 mya, motrygonidae clade permitted Lovejoy as early as the late Eocene (Marques, 2000). (1996) to postulate that potamotrygonids The application of molecular clock esti- were derived from a freshwater-invading an- mates as a method to establish dates of evo- cestor that was distributed along the northern lutionary events is highly controversial, and coast of South America (similar to the pre- many researchers consider molecular clocks sent-day ranges of H. paci®ca and H. to be mostly unreliable (review in Hillis et schmardae). The timing of this derivation al., 1996). Molecular clocks must be cali- was established by reference to the earliest brated by benchmark dates from geology, pa- putative potamotrygonid fossils known (from leontology, or biogeography (Smith, 1992; the late Miocene of Peru; e.g., Frailey, 1986). Lundberg, 1998), and the rate of genetic di- This, in turn, led to the inference that an epi- vergence in a given gene may vary signi®- continental (probably Miocene) seaway was cantly even within a single taxonomic line- the route of the South American invasion by age (Knowleton et al., 1993; Hillis et al., the potamotrygonid ancestor, which subse- 1996). Furthermore, there is much discrep- quently underwent extensive speciation ancy among methods used to infer actual (some 20 species are presently recognized in amounts of genetic divergence (see referenc- the family, with as many as six known un- es in Lundberg, 1998), and the con®dence described forms; Carvalho et al., 2003). limits allowed by rate calibrations may, in Lovejoy's theory also allowed for a vicariant many cases, be as great as the age of the explanation for the origin of the two species actual events being dated (Hillis et al., 1996; of amphi-American Himantura, which oc- precisely the situation concerning potamotry- 114 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 gonids described above). Hillis et al. (1996) cartilaginous rod in the tail posterior to the and Lundberg (1998) provided cogent re- caudal stings in place of individual vertebrae marks on the use and accuracy of molecular (contrary to the Green River stingrays). The clocks, which need not be repeated here. It caudal stings in ²Promyliobatis are located is our view that the fossil record, even farther posteriorly than in any other stingray, though frequently imprecise, will provide fossil or extant, and because its tail was very better estimates of divergence times even if long (which is mostly preserved), this sting- these are underestimates (minimum diver- ray had enormous reach with its caudal gence dates), as also argued by Lundberg stings. On our tree, ²Promyliobatis may be (1998). There are two prerequisites to em- more closely related to Myliobatis or simply ploy fossils as benchmarks for dating: they to an unresolved myliobatid (the most con- must unequivocally possess derived features servative hypothesis of relationship); in any that permit them to be placed with con®- case, the placement of ²Promyliobatis at the dence in a phylogeny, and their geological node with pelagic stingrays is very well sup- dating must be relatively accurate. ported (®g. 51). Fossil stingrays that are pertinent for es- ²``Dasyatis'' muricata (®g. 49B) is also tablishing a minimum age of the potamotry- unquestionably a stingray, but its position on gonid lineage (or for the invasion of the po- the tree in ®gure 51 is not as well resolved. tamotrygonid common ancestor into South Because it presents the cartilaginous rod in American freshwaters) were added to our the tail, it is phylogenetically positioned at strict consensus tree, allowing for a more least at the node uniting potamotrygonids, precise consideration of the relative ages of ``dasyatids,'' gymnurids, and myliobatids. certain stingray lineages. These taxa stem The dorsal fossa on the scapular process is from the Monte Bolca deposits of northeast- dif®cult to discern in this fossil, but it ap- ern Italy and are represented by well-pre- pears to be present according to our obser- served skeletons (®gs. 49, 50) that retain vations, and consequently ²``Dasyatis'' mur- characters indicative of their placement in icata belongs at the next more derived node our stingray phylogeny (®g. 51). These sting- (uniting ``dasyatids,'' gymnurids, and my- ray taxa have not been incorporated into our liobatids; ®g. 51). This species clearly does phylogenetic analysis because the Monte not present any of the derived characters that Bolca batoid fauna is still under investiga- unite other monophyletic components position, but their phylogenetic positions in ®g- tioned at lower nodes in our phylogeny (uro- ure 51 are well supported by uniquely de- lophids, urobatids, potamotrygonids); its rived features. phylogenetic position is therefore strongly One of the most remarkable stingrays from corroborated. Monte Bolca is ²Promyliobatis gazolae (®gs. The Monte Bolca Formation of northeast- 49A, 50). This stingray is unquestionably a ern Italy represents an extinct coral reef em- myliobatid, as evidenced by its teeth that are bayment of the tropical Tethys Sea that was coalesced into a horizontally expanded, frequently exposed to the vagaries of ¯uvial grinding pavement, as in pelagic stingrays systems and open lagoons (Sorbini, 1983; except mobulids (which have reversed to Landini and Sorbini, 1996). The deposits of having small, individual teeth). The dentition the two main fossil-bearing sites, Pesciara of ²Promyliobatis is more similar to the den- and Monte Postale (the third site, Purga di tition of Myliobatis or Aetomylaeus, with a Bolca has not yielded fossil ®shes), are ®rm- much wider central tooth articulating later- ly dated as early Eocene (Lutetian, NP 14, ally with smaller, hexagonal pavementlike Discoaster sublodoensis nanoplankton zone; teeth (De Zigno, 1885; ®g. 50 here). This Barbieri and Medizza, 1969; Patterson, 1993; contrasts with the dentition of Rhinoptera, Landini and Sorbini, 1996), even though pre- where the central tooth is only slightly wider vious authors have given a slightly older age than the lateral ones, which are also hexag- to these beds (as topmost Ypresian; Blot, onal (or with Aetobatus, in which a single, 1969, 1980). These discrepancies in dating greatly expanded tooth is present in each hor- are minor, however, and the 50 million-year- izontal row). ²Promyliobatis presents the old early Eocene age of the Monte Bolca de- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 115 posits is presently not disputed (note that 27 million years older than Lovejoy et al.'s some authors have interpreted the lowermost (1998) hypothesis. A 50 million-year-old Po- Lutetian as middle Eocene; Tyler and Santi- tamotrygonidae is clearly feasible when the ni, 2002). morphological similarities between fossil and Our phylogeny (®g. 51) indicates that po- living elasmobranchs, including stingrays, tamotrygonids must be at least as old as are taken into account. A conservative mor- ²Promyliobatis gazolae and ²``Dasyatis'' phology per se is not necessarily evidence muricata, taking into account their relative that a particular lineage is very ancient, but phylogenetic positions (and using a stepwise given that the oldest stingrays known from correlation between nodes, moving ``down'' skeletal remains (52 mya) are morphologi- the tree as outlined by Lundberg, 1993, cally very similar to modern ones (a circum- 1998). The Monte Bolca stingrays provide stance present in many elasmobranch groups benchmarks by which the origin of the Po- known from even older fossils, e.g., ²Pseu- tamotrygonidae can be dated, imposing a dorhina from Late Jurassic of Solnhofen and minimum age of at least 50 mya to this lin- modern Squatina;²Asterodermus from Sol- eage (early Eocene). This is much older than nhofen and modern rhinobatids; Woodward, their Miocene origin proposed by Lovejoy 1889; Saint-Seine, 1949), a signi®cant cor- (1996, 1997) and Lovejoy et al. (1998). This relation between age and degree of morpho- speci®c age correlation with the Monte Bolca logical conservativeness in stingrays, as ev- stingrays is not possible in Lovejoy's phy- idenced by their fossil record, cannot be dis- logeny (or in the tree of McEachran et al., carded. 1996), due to existence of the node uniting Certain components of the Neotropical Taeniura, amphi-American Himantura, and ichthyofauna have pre-Miocene remains in Potamotrygonidae, which could have origi- South America, and it is believed that by nated well after the divergence of their com- middle or late Miocene the Neotropical ®sh mon ancestor from the main stem of stingray fauna was essentially modern (Lundberg, evolution (®g. 46B); moreover, the clade am- 1998). Some of the more ancient groups phi-American Himantura ϩ Potamotrygoni- mentioned below have a historical associa- dae could only have come into existence at tion with Africa (predating its Late Jurassic an even later time. Conversely, this inference to Early Cretaceous separation from South is not possible in our phylogeny, as pota- America, some 150 to 110 mya), and there- motrygonids branch out directly from the fore present a pattern of area relationships main stingray evolutionary branch, without different from that of potamotrygonids, any intervening node (®g. 51; note that this which are clearly not more closely related to is also the case in the analysis presented in a clade of African stingrays. Late Cretaceous appendix 2, in which potamotrygonids would (Maastrichtian) Neotropical fossils, mostly be at least as old as ²``Dasyatis'' muricata, isolated fragments, include relatives of Lep- but not necessarily as old as ²Promyliobatis). idosiren (Schultze, 1991; Gayet and Meunier, Potamotrygonids are consequently as old as 1998), osteoglossids (possibly Aptian; Lund- the main evolutionary stem of stingrays, berg, 1998; Gayet and Meunier, 1998), char- which is at least 50 million years old as in- aciforms (Erythrinidae, Serrasalminae; Rob- ferred by the Monte Bolca stingrays. erts, 1975; Gayet, 1991; Vari, 1995; Lund- In his overview of the temporal context for berg, 1998; Gayet et al., 2003), and siluri- the origin and diversi®cation of Neotropical forms (overview in Lundberg, 1998). ®shes, Lundberg (1998: 55) posed the ques- Callichthyid cat®shes (Cockerell, 1925) and tion, ``[c]ould the Potamotrygonidae be that a variety of different characiforms (Gayet [Late Cretaceous] old?'' According to our and Meunier, 1998) from the late Paleocene, study, this certainly seems possible. Our phy- and fossils related to Lepidosiren and to the logeny supports a divergence age for pota- Percichthyidae (a percomorph family) from motrygonids that is considerably older than the Eocene, are also reported from South previous estimatesÐat least 12 million years America (Arratia and Cione, 1996; Lund- older than the most generous timeframe al- berg, 1998; cf. Patterson, 1993). These lowed by Marques (2000), and at least some groups are therefore older than 50 million 116 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 49. Fossil stingrays from the marine early Eocene Monte Bolca deposits of northeastern Italy. A. ²Promyliobatis gazolae (MCSNV VII B 90), a myliobatid. B. ²``Dasyatis'' muricata (MCSNV VII B 92), a whiptailed stingray. The phylogenetic positions of both taxa are shown in ®gure 51. years. The lack of a pre-Miocene fossil re- es; see summary in Lundberg, 1998). The cord for potamotrygonids should not be in- Cretaceous stingray fossils that have been terpreted too decisively, as other Neotropical described from South America may even groups that are inferred to have originated share a common ancestry with potamotry- much earlier than the Miocene, such as those gonids, but this position is presently unten- with an historical African association, also able as these fossils consist of very gener- lack a robust pre-Miocene fossil record alized isolated teeth (e.g., Schaeffer, 1963; (many characiforms, e.g., Roberts, 1975; see references in the Introduction), and the Vari, 1995; doradoid and loricarioid cat®sh- search for shared, derived dental features is 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 117

stingray phylogeny is a case in point). Of course, if the postulated phylogenetic relationships are inaccurate, then a speci®c area relationship is not more probable just because its biogeographic implications appear to be more alluring. Our phylogeny (®g. 43), in fact, is not at odds with the mechanism proposed by Lovejoy (1996) and Lovejoy et al. (1998) to explain the presence of the potamotrygonid common ancestor in South America (through a marine introgression). Although we refute that amphi-American Hi- mantura is the sister-group to potamotrygonids, the common ancestor of the clade Pota- motrygonidae ϩ (``Dasyatidae'' ϩ (Gymnur- idae ϩ Myliobatidae)) could also have inhabited the proto-Caribbean region (or the eastern Paci®c realm, which was continuous with the proto-Caribbean until the comple- tion of the Panamanian Isthmus in the Plio- cene). Therefore, the ancestor of the Pota- Fig. 50. Enlarged view of the dentition of motrygonidae could still have invaded South ²Promyliobatis gazolae (MCSNV VII B 90); entire specimen depicted in ®gure 49A. Individual America through a marine introgression from lateral teeth are visible, scattered at both sides of that area. This invasion is possible because, the enlarged median tooth row (mt). according to our phylogeny, there was already a large stingray lineage present in the Americas which eventually gave rise to the still not a standard employed by fossil teeth Urotrygonidae, ²Heliobatis, and to the com- specialists (e.g., Cappetta, 1975, 1987). mon ancestor of the large clade above (this Our stingray phylogeny (®gs. 43, 51) and component originally had a strong historical its implied area cladogram (®g. 48) do not association with the Indo-Paci®c region; see permit the formulation of such a clearcut bio- discussion above concerning the origin of the geographic scenario as that presented by Green River stingrays). Lovejoy (1996) and Lovejoy et al. (1998). We disagree, however, with Lovejoy et Because potamotrygonids are the sister- al.'s (1998) hypothesis in relation to the tim- group of a large clade that contains most of ing of the vicariant event that allowed for the the modern stingray diversity, it is more dif- introduction of the potamotrygonid ancestor ®cult to accurately pinpoint or restrict to a into South America. A Miocene epicontinen- small area the distribution of their common tal seaway (such as the Pebasian Seaway of ancestor. This contrasts with Lovejoy's hy- Lundberg et al., 1998) occurred far too re- pothesis (and to a lesser extent with the ``Pa- cently according to our phylogeny, but ear- ci®c/urolophid'' theory forwarded by Brooks lier marine incursions are also known from et al., 1981), in which the common ancestor the proto-Caribbean region (i.e., from the of the amphi-American Himantura ϩ Pota- north; Hoorn, 1993; Hoorn et al., 1995; motrygonidae clade was candidly inferred to Smith et al., 1994; Lundberg et al., 1998). have been distributed in the proto-Caribbean According to Smith et al. (1994) and Lund- Sea. Biogeographic hypotheses that deal with berg et al. (1998), extensive marine trans- ``deeper'' phylogenetic divergence events, gressions over northern South America oc- that is, those hypotheses based on relation- curred frequently (at least four times) from ships among higher level taxa or more inclu- the Late Cretaceous (Maastrichtian) to the sive monophyletic groups, will usually en- late Miocene, and the duration of each in- counter this dif®culty, especially regarding cursion and its resulting seaway lasted for relatively widespread marine forms (our millions of years. Two epicontinental incur- 118 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 51. Tree resulting from our phylogenetic analysis of myliobatiforms, showing placement of two early Eocene stingray taxa from the Monte Bolca Formation of northeastern Italy (fossils presented in ®g. 49). These taxa allow for the timing of divergence events between the node that gives rise to ²Heliobatis and the node that contains gymnurids and myliobatids; the groups derived directly from the main stem of the phylogeny between these nodes must have originated at least by the early Eocene (roughly 50 mya). See ®gure 48 for geographical distribution of terminal taxa. sion episodes in particular are congruent with bean/northern West Atlantic)). We simply the minimum age of the potamotrygonid lin- think that a further area relationship (the eage (50 mya) derived from our phylogeny: American component) should be added after a Late Cretaceous seaway, and another one the Indo-west Paci®c region (the basalmost that lasted from the late Paleocene to the area). These area relationships largely apply very early Eocene (®g. 52). Either incursion to other groups in which there is a basal Neo- may have trapped the potamotrygonid ances- tropical freshwater species that is the sister- tor in the neotropics, where it slowly adapted group to an eastern Paci®c/Caribbean species to a progressively more freshwater environ- pair (such as anchovies of the genera Ceten- ment. graulis and Anchovia; Nelson, 1984) or other The biogeographic hypothesis presented taxa that may have undergone vicariance and here for the origin of the Potamotrygonidae subsequent diversi®cation due to the forma- (a pre-early Eocene marine incursion from tion of the Panamanian Isthmus (such as am- the north) does not refute the area relation- phi-American Himantura). These historical ships of other taxa mentioned by Lovejoy associations are still possible under our hy- (1996) in support of his theory (the ``proto- pothesis; the only adjustment is that their an- Caribbean/Himantura'' hypothesis). Our hy- cestors may have invaded the Neotropical pothesis (cf. ®g. 48) is congruent with the freshwater system much earlier than the Mio- area cladogram presented by Lovejoy (1997: cene (but not necessarily, at least regarding ®g. 4a): Indo-west Paci®c ϩ (South Ameri- the engraulids). Without robust dating (i.e., can freshwater ϩ (eastern Paci®c ϩ Carib- strong fossil evidence), it is not possible to 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 119 precisely determine which marine transgres- terotrygon and ²Heliobatis) retain a keyhole- sion (or other mechanism) was responsible shaped dorsal fontanelle, posteriorly directed for the invasion of the ancestral taxon; it antorbital cartilages, elongated iliac process- could have been the same epicontinental sea- es on the pelvic girdle, and many other fea- way that carried the potamotrygonid ancestor tures present in most benthic stingrays. The (or other primitively marine groups), but this new genus is morphologically very similar to event could also have occurred much later. many extant nongymnurid and nonmylioba- Although amphi-American Himantura is tid stingrays. placed in a polytomy with other ``dasyatid'' (2) ²Asterotrygon and ²Heliobatis present genera and Gymnuridae ϩ Myliobatidae, it many of the characters unique to stingrays may eventually prove to be basal to these (Order Myliobatiformes), such as caudal taxa (as indicated when analyzing our matrix stings, laterally expanded shel¯ike postorbit- with the weighting schemes of Farris, 1969 al processes, second synarcual cartilage, ab- and Goloboff, 1993; see above and ®g. 47). sence of jugal arches, and absence of thoracic In addition, our biogeographic theory does ribs. ²Asterotrygon also presents an anterior not contradict the northwest±southeast deri- expansion of the medial (basibranchial) vation sequence or vicariant pattern that has plate, a small transverse basihyal element, been reported for some Neotropical ®shes and the ®rst pair of hypobranchials relatively distributed in northern South America (e.g., straight and obliquely oriented from the ven- the curimatid Potamorhina, Vari, 1984; to a tral pseudohyoid bar to articulate with the ba- lesser extent Steindachnerina and Creagutus, sihyal; these features are considered herein Vari, 1991; Vari and Harold, 2001; for sum- to be additional stingray synapomorphies. maries, see Vari, 1988; Vari and Weitzman, (3) Our phylogenetic analysis resulted in 1990; Lundberg and Chernoff, 1992). In 35 equally most parsimonious trees, which these cases, according to the above authors, are reduced to ®ve subgroups of topologies basal species are distributed in northern once the uncertainty concerning the af®nities South America and, through vicariance, more of dasyatid genera are removed. Both Green derived species have been pushed progres- River stingray genera are systematically po- sively farther south or southeast from the an- sitioned far apart in all minimum-length trees cestral area. obtained. ²Asterotrygon is resolved as a basal stingray either in a polytomy with Ple- CONCLUSIONS siobatis and urolophids (as in the strict consensus tree), or as sister-group to urolophids (1) The new monotypic genus ²Asterotry- (in 21 of the 35 minimum-length trees ob- gon is distinguished from all stingrays, both tained). ²Heliobatis is unequivocally placed fossil and extant, on the basis of its unique as sister-group to a large clade that includes plesodic dorsal ®n covered by denticles, and potamotrygonids, dasyatids, gymnurids, and by a combination of features including an in- myliobatids. tense covering of minute hooklike denticles (4) Results from our phylogenetic study over most of dorsal disc and tail regions and are as follows: Hexatrygon is basal to all oth- stout tail at base. The new genus has many er stingray taxa, urolophids (Trygonoptera ϩ characters present in selected Recent stingray Urolophus) are monophyletic, urotrygonids genera, such as well-developed hyomandi- (Urotrygon ϩ Urobatis) are monophyletic, bulae connected to Meckel's cartilage by ro- potamotryonids are monophyletic and Para- bust ligaments containing an angular carti- trygon is the most basal genus within the lage; separate vertebral elements extending family, potamotrygonids are sister-group to a to tip of tail (instead of a continuous carti- large node comprising dasyatid genera (in a laginous rod); postorbital processes separated polytomy) and Gymnuridae ϩ Myliobatidae, from triangular outgrowth (the supraorbital and the Dasyatidae is not monophyletic on process) of the supraorbital crest by a notch; any of the equally most parsimonious trees and both dorsal and ventral tail-folds sup- obtained. Himantura should be divided into ported by rudimentary radial elements. Prim- two genera, one for amphi-American species itively, both Green River stingray taxa (²As- and Himantura sensu stricto exclusive for 120 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fig. 52. Maps showing the distribution of the continents and the extent of their coastlines (in darker line) from Late Cretaceous (Maastrichtian) to middle Eocene, spanning some 25 million years (modi®ed from Smith et al., 1994). Arrowheads indicate the extent of marine incursions (epicontinental seaways) into South America from the north (the proto-Caribbean/eastern Paci®c region) during these intervals. Mya ϭ million years ago. west Indo-west Paci®c species. Taeniura is the true systematic relevance of many fea- putatively monophyletic based on the unique tures, such as the con®guration of tail-folds, morphology of their ®rst hypobranchial (and fusion of ceratobranchial elements, and even possibly the anterior expansion of the hy- the hyomandibular-Meckelian ligament. omandibulae). Morphological variation among component (5) Phylogenetic relationships among cer- species of familial-level stingray terminal tain stingray genera are problematic due to taxa may lead to elevated polymorphisms, lack of information concerning many of their but more data for most stingray species are component species. A species-level phylog- still needed before a comprehensive species- eny is necessary to adequately understand level phylogeny can be accomplished. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 121

Fig. 52. Continued.

(6) Our phylogenetic study led to the fol- have entered South America through an epi- lowing biogeographic implications: ²Astero- continental seaway from the north (from the trygon displays a strong historical association proto-Caribbean/eastern Paci®c region), dur- with the Indo-west Paci®c region (and pos- ing the period ranging from the Late Creta- sibly with Australia); a large American sting- ceous or Paleocene to the early Eocene. ray component was established before the early Eocene, which eventually gave rise to ACKNOWLEDGMENTS the Urotrygonidae and ²Heliobatis; the origin of the Neotropical freshwater stingrays Numerous collection managers and cura- (Potamotrygonidae), which is completely in- tors have provided invaluable assistance in dependent of the origin of the Green River furnishing stingray specimens for dissection stingrays, occurred before the early Eocene, and/or work space in their collections: G. much earlier than previous Miocene esti- Nelson (retired), Scott A. Schaefer, Melanie mates; the potamotrygonid ancestor could L. J. Stiassny, Barbara A. Brown, Norma R. 122 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Feinberg, and Radford Arrindell (AMNH); specimens. We thank our two reviewers for Peter R. Last, John D. Stevens, Alastair Gra- pointing out mistakes and omissions, for ham, and Gordon Yearsley (CSIRO); Barry their careful reading of the manuscript, and Chernoff, Mary-Anne Rogers, and Kevin for their insightful comments. Preliminary Swagel (FMNH); the late Lorenzo Sorbini, presentations on different parts of the re- Alessandra Aspes, Roberto Zorzin, Anna search presented in this paper were given at Vaccari, and Francesca Buttorini (MCSNV); ASIH 96 (New Orleans) and at the Sympo- Karsten Hartel and Karel Liem (MCZ); Fran- sium on the Phylogeny and Classi®cation of cËoise Meunier, Bernard SeÂret, Guy Duhamel, Neotropical Fishes in 1997 (Porto Alegre); Patrice Pruvost, and Pascal Deynat (MNHN); funding for the ®rst author to attend those JoseÂ Lima de Figueiredo, Osvaldo Oyakawa, meetings was provided by the Donn Rosen and Alberto Akama (MZUSP); Peter Forey, Fund of the Department of Ichthyology Anthony C. Gill, Darrel J. Siebert, Oliver (AMNH) and by Roberto Reis, Luiz Mala- Crimmen, and Sean Davidson (NHM); Bruce barba, and other symposium organizers, re- Erickson (SMMP); Ulisses L. Gomes spectively. The ®nal version of this paper (UERJ); Luca Altichieri (UP); Lynne R. Par- was presented at ASIH 2003 (Manaus) enti, Susan Jewett, Leslie Knapp, and Jerry through the ®nancial support provided by the Finan (USNM); Hans Joachim-Paepke and FundacËaÄo de Amparo aÁ Pesquisa do Estado Gloria Arratia (ZMB). Most of the Green de SaÄo Paulo (Fapesp proc. no. 02/06459±0) River stingrays examined for this paper were to M.R.C. Major funding to the ®rst author prepared by the staff of the Department of in relation to this project was largely provid- Geology of the FMNH and the Division of ed by the American Museum of Natural His- Paleontology of the AMNH. Special thanks tory (as part of an Axelrod Postdoctoral Fel- are due to Peter Last, Gus Yearsley, and Al lowship in the Division of Paleontology) and Graham of CSIRO in Hobart who were very Field Museum of Natural History, and both generous in allowing the ®rst author to ex- institutions are sincerely thanked. We also amine numerous radiographs and slides of gratefully acknowledge the generosity of stingrays in their collections, and for discus- Thomas (deceased) and Hilda Maloney for sions concerning stingaree systematics. The the donation of the AMNH paratype of As- ®rst author has bene®ted a great deal from terotrygon, and of Dr. Herbert R. and Evelyn discussions with Nathan R. Lovejoy and es- Axelrod for their constant and invaluable pecially John D. McEachran concerning support of paleoichthyological research at the stingray evolution. For a productive ex- AMNH. change of ideas regarding reproductive sym- metry in stingrays, we sincerely thank Nick REFERENCES K. Dulvy, Buck Snelson, Wade D. Smith, and Henry Mollet. For assistance with tech- Achenbach, G.M., and S.V.M. Achenbach. 1976. nical matters and discussions regarding sys- Notas acerca de la raya ¯uvial (Batoidei, Pota- motrygonidae), que frecuentam el sistema hid- tematics and biogeography in general, thanks rograÂ®co del RõÂo ParanaÂ Medio en el Departa- are due to Robert C. Schelly and W. Leo mento La Capital. Comunicaciones del Museo Smith (AMNH). Carl Mehling (AMNH) pro- Provincial de Ciencias Naturales 8: 3±34. vided assistance with PhotoShop. The superb Albers, H., and W. Weiler. 1964. Eine Fischfauna artistic rendering of the skeleton of the po- aus der oberen Kreide von Aachen und neuere tamotrygonid stingray in ®gure 16 was illus- Funde von Fischresten aus dem Maestricht des trated by Lorraine Meeker (AMNH), who angrenzenden belgisch-hollaÈndischen Raumes. also mounted the ®gures of the Green River Neues Jahrbuch fuÈr Geologie und PalaÈontologie stingrays and many others in this paper; Abhandlungen 120: 1±33. along with Chester Tarka (AMNH), Lorraine Arambourg, C. 1952. Les verteÂbreÂs fossiles des gisements de phosphates (Maroc-AlgeÂrie-Tuni- is sincerely thanked for her efforts and pa- sie). Proctectorat de la ReÂpublique FrancËaise au tience. Most of the excellent photographs of Maroc, Direction de la Production Industrielle the cleared-and-stained stingrays depicted in et des Mines, Division des Mines et de la GeÂo- the color plates were taken by John Wein- logie, Service GeÂologique Notes et MeÂmoires stein (FMNH), including those of the full 92: 1±372. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 123

ArauÂjo, M.L.G. de. 1998. Biologia reprodutiva e san (Chondrichthyes, Rhinobatoidea) from the pesca de Potamotrygon sp C (Chondrichthyes- Early Cretaceous of Brazil and its phylogenetic Potamotrygonidae), no meÂdio Rio Negro, Ama- relationships. In G. Arratia and G. Viohl (edi- zonas. Unpublished MS dissertation. Universi- tors), Mesozoic ®shes, systematics and paleo- dade do Amazonas, Manaus, 171 pp. ecology: 47±62. Munich: Verlag F. Pfeil. Arratia, G., and A. Cione. 1996. The record of Brooks, D.R. 1995. Neotropical freshwater sting- fossil ®shes of southern South America. In G. rays and their parasites: a tale of an ocean and Arratia (editor), Contributions of southern a river long ago. In M.I. Oetinger and G.D. South America to vertebrate paleontology: 9± Zorzi (editors), The biology of freshwater elas- 72. Munich: Verlag F. Pfeil. mobranches: 52±61. Journal of Aquariculture Babel, J.S. 1967. Reproduction, life history, and and Aquatic Sciences 7: 1±161. ecology of the round stingray, Urolophus hal- Brooks, D.R., and D.A. McLennan. 1993. Para- leri, Cooper. California Fish and Game Bulletin script: parasotes and the language of evolution. 137: 1±104. Washington, DC: Smithsonian Institution Press. Barbieri, G., and F. Medizza. 1969. Contributo Brooks, D.R., T.B. Thorson, and M.A. Mayes. alla conoscenza geologica della regione di Bol- 1981. Freshwater stingrays (Potamotrygonidae) ca (Monti Lessini). Memoire degli Istituti di and their helminth parasites: testing hypotheses Geologia e Mineralogia dell'UniversitaÁ di Pa- of evolution and coevolution. In V.A. Funk and dova 27: 1±36. D.R. Brooks (editors), Advances in cladistics, Best, W. 1987. A sedimentologic and stratigraphic proceedings of the ®rst meeting of the Willi study of the Paleocene Fort Union Formation Hennig Society: 147±175. New York: New in the South Dakota Cave Hills of Harding York Botanical Society. County, South Dakota. Unpublished MSc dis- Buchheim, H.P. 1994. Paleoenvironments, litho- sertation. South Dakota School of Mines and facies and ``varves'' of the Fossil Butte Mem- Technology, Rapid City, 124 pp. ber of the Green River Formation, southwestern Bigelow, H.B., and W.C. Schroeder. 1953. Fishes Wyoming. Contributions to Geology 30(1): 3± of the western north Atlantic. Part II. Saw®sh- 14. es, guitar®shes, skates, rays and chimaeroids. Caira, J.N. 1990. Metazoan parasites as indicators Memoirs of the Sears Foundation for Marine of elasmobranch biology. In H.L. Pratt, S.H. Research 2, xv ϩ 588 pp. Gruber, and T. Taniuchi (editors), Elasmo- Bininda-Emonds, O.R., H. Bryant, and A.P. Rus- branchs as living resources: advance in the bi- sell. 1998. Supraspeci®c taxa as terminals in ology, ecology, systematics, and behavior, and cladistic analysis: implicit assumptions of status of the ®sheries. NOAA (National Oce- monophyly and a comparison of methods. Bi- anic and Atmospheric Administration) Techni- ological Journal of the Linnaean Society 64: cal Report NMFS (National Marine Fisheries 101±133. Service) 90: 71±96. Blot, J. 1969. Les poissons fossiles du Monte Bol- CapapeÂ, C., J. Zaouali, J.A. Tomasini, and J.L. ca, classeÂs jusqu'ici dans les familles des Car- Bouchereau. 1992. Reproductive biology of the angidae, Menidae, Ephippidae, Scatophagidae. spiny butter¯y ray, Gymnura altavela (Linnaeus, Studi e Ricerche sui Giacimenti Terziari di Bol- 1758) (Pisces: Gymnuridae) from off the Tuni- ca, vol. 1(text): 1±525. sian coasts. Scientia Marina 56(4): 347±355. Blot, J. 1980. La faune ichtyologique des gise- Cappetta, H. 1972. Les poissons creÂtaceÂs et terti- ments du Monte Bolca (Province de VeÂrone, aires du Bassin des Lullemmeden (Rep. Du Ni- Italie). Catalogue systeÂmatique preÂsentant l'eÃtat ger). Palaeovertebrata 5(5): 179±251. actuel des recherches concernant cette faune. Cappetta, H. 1975. Sur quelques seÂlaciens nou- Bulletin du MuseÂum National d'Histoire Natu- veaux du CreÂtaceÂ supeÂrior de Bolivie. Geobios relle, ser. 4, sect. C, 2(4): 339±396. 8(1): 5±24. Bor, T.J. 1990. A new species of mobulid ray Cappetta, H. 1984. DeÂcouverte du genre Gymnura (Elasmobranchii, Mobulidae) from the Oligo- (Batomorphii, Myliobatiformes) dans le Tha- cene of Belgium. Contributions to Tertiary and neÂtien des Ouled Abdoun, Maroc. Observations Quartenary Geology 27(2, 3): 93±97. sur le denture de quelques espeÁces actuelles. Bradley, W.W. 1948. Limnology and the Eocene Geobios 17(5): 631±635. lakes of the Rocky Mountain region. Geologi- Cappetta, H. 1987. Chondrichthyes II. Mesozoic cal Society of America Bulletin 59: 635±648. and Cenozoic Elasmobranchii. In H.-P. Schultze Breder, C.M., and D.E. Rosen. 1966. Modes of (editor), Handbook of paleoichthyology, vol. reproduction in ®shes. Garden City, NY: The 3B: 1±193. Stuttgart: Gustav Fisher Verlag. Natural History Press. Cappetta, H. 1992. Late Cretaceous selachian fau- Brito, P., and B. SeÂret. 1996. The new genus Ian- nas from Bolivia: new data and summary. In R. 124 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Suarez-Soruco (editor), FoÂssiles y facies de Bo- Abteilung A Palaeozoologie-Stratigraphie 171: livia, vol. 1. Revista TeÂcnica de Yacimentos Pe- 75±103. trolõÂferos Fiscales Bolivianos 12(3, 4): 435±440. Case, G.R. 1981. Late Eocene selachians from Cappetta, H., and G.R. Case. 1975. Contribuition south-central Georgia. Palaeontographica Ab- aÁ l'eÂtude des seÂlaciens du groupe Monmouth teilung A Palaeozoologie-Stratigraphie 176(1± (Campanien-Maestrichtien) du New Jersey. Pa- 3): 52±79. laeontographica Abteilung A 151(1±3): 1±46. Case, G.R. 1996. A new selachian fauna from the Cappetta, H., and G.R. Case. 1999. Additions aux lower Hornerstown Formation (early Paleo- faunes de seÂlaciens du CreÂtaceÂ du Texas (Al- cene/Montian) of Monmouth County, New Jer- bien supeÂrieur-Campanien). PalaeoIchthyologica sey. Palaeontographica Abteilung A Palaeo- 9: 5±111. zoologie-Stratigraphie 242(1±3): 1±14. Cappetta, H., and D. Nolf. 1981. Les seÂlaciens de Case, G.R., and H. Cappetta. 1997. A new sela- l'auversian de Ronquerolles (EoceÁne SupeÂrieur chian fauna from the Late Maastrichtian of Tex- du Bassin de Paris). Mededelingen van de as (Upper Cretaceous/Navarroan; Kemp For- Werkgroep voor Tertiare en Kwartiare Geologie mation). MuÈnchner Geowissenschaften Abtei- 18(3): 87±107. lung A Geologie und PalaÈontologie 34: 131± Cappetta, H., C. Duf®n, and J. Zidek. 1993. Chon- 189. drichthyes. In M.J. Benton (editor), The fossil Casier, E. 1946. La faune ichthyologique de record, vol. 2: 593±609. London: Chapman and l'YpreÂsien de la Belgique. MeÂmoires du MuseÂe Hall. d'Histoire Naturelle de Belgique 104: 3±267. Carpenter, J.M. 1988. Choosing among multiple Casier, E. 1966. Faune ichthyologique du London equally parsimonious cladograms. Cladistics 4: Clay. London: British Museum (Natural His- 291±296. tory), 496 pp. Carpenter, J.M. 1994. Successive weighting, reli- Castex, M.N. 1967. Observaciones en torno a las ability and evidence. Caldistics 10: 215±220. formaciones estelares que recubren el dorso de Carvalho, M.R. de. 1996a. Higher-level elasmo- algunas especies de rayas de agua dulce (Chon- branch phylogeny, basal squaleans and para- drichthyes, Potamotrygonidae). Physis 73: phyly. In M.L.J. Stiassny, L.R. Parenti, and 485±491. G.D. Johnson (editors), Interrelationships of Chang, M.-M., and L. Grande. 1997. Redescrip- ®shes: 35±62. San Diego: Academic Press. tion of ²Paraclupea chetungensis, an early clu- Carvalho, M.R. de. 1996b. Review of: ``Fresh- peomorph from the Lower Cretaceous of south- water Elasmobranchs, a Symposium to Honor eastern China. Fieldiana Geology new. ser. 37: Thomas B. Thorson.'' Copeia 1996(4): 1047± 1±19. 1050. Chang, M.-M., and J. Zhou. 1993. A brief survey Carvalho, M.R. de. 1999. A systematic revision of the Chinese Eocene ichthyofauna. Kaupia 2: of the electric ray genus Narcine Henle, 1834 157±162. (Chondrichthyes: Torpediniformes: Narcini- Chu, Y.T., and M.C. Wen. 1979. A study of the dae), and the higher-level phylogenetic relation- lateral-line canal systems and that of lorenzini ships of the orders of elasmobranch ®shes ampullae and tubules of elasmobranchiate ®sh- (Chondrichthyes). Unpublished PhD disserta- es of China. Monograph of Fishes of China 2. tion The City University of New York/Ameri- Shanghai: Shanghai Science and Technology can Museum of Natural History, New York, Press. 735 pp. Cione, A.L. 1978. Aported paleontoloÂgicos al Carvalho, M.R. de. In press. A Late Cretaceous conocimiento de la evolucioÂn de las paleotem- thornback ray from southern Italy, with a phy- peraturas en la aÂrea austral de AmeÂrica del Sur logenetic reappraisal of the Platyrhinidae durante el Cenozoico. Ameghiniana 15(1, 2): (Chondrichthyes: Batoidea). In G. Arratia and 183±208. A. Tintori (editors), Mesozoic ®shes 3. Munich: Cione, A.L., and S. ExpoÂsito. 1980. Chondrichthy- Verlag Dr. F. Pfeil. es del ``Patagoniano'' s.l. de Astra, Golfo de San Carvalho, M.R. de, N.R. Lovejoy, and R.S. Rosa. Jorge, provincia de Chubut, Argentina. Su sig- 2003. Potamotrygonidae. In R.E. Reis, C.J. Fer- ni®cado paleoclimaÂtico y paleobiogeograÂ®co. In raris, Jr., and S.O. Kullander (editors), Check- Congreso Argentino de PaleontologõÂa y Bioes- list of freshwater ®shes of South and Central tratigrafõÂa No. 2 y Congreso Latinoamericano de America: 22±29. Porto Alegre: Editora da Pon- PaleontologõÂa No. 1, Actas 2: 275±290. tifõÂcia Universidade CatoÂlica. Cockerell, T.D. 1925. A fossil ®sh of the family Case, G.R. 1980. A selachian fauna from the Callichthyidae. Science 62: 317±322. Trent Formation, Lower Miocene (Aquitanian) Compagno, L.J.V. 1973. Interrelationships of liv- of eastern North Carolina. Palaeontographica ing elasmobranch ®shes. In P.H. Greenwood, C. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 125

Patterson, and R. Miles (editors), Interrelation- Deynat, P.P., and P.M. Brito. 1994. ReÂvision des ships of ®shes: 15±61. London: Academic tubercles cutaneÂs des raies (Chondrichthyes, Press. Batoidea) du basin du ParanaÂ, Tertiaire Compagno, L.J.V. 1977. Phyletic relationships of d'Amerique du sud. Annales de PaleÂontologie living sharks and rays. American Zoologist 80: 237±251. 17(2): 303±322. Deynat, P.P., and B. SeÂret. 1996. Le reveÃtement Compagno, L.J.V. 1999. Checklist of living elas- cutaneÂ des raies (Chondrichtyes, Elasmobran- mobranches. In W.C. Hamlett (editor), Sharks, chii, Batoidea). I. Morphologie et arrangement skates, and rays, the biology of elasmobranch des denticules cutaneÂs. Annales des Sciences ®shes: 471±498. Baltimore: John Hopkins Uni- Naturelles Zoologie 17(2): 65±83. versity Press, 515 pp. De Zigno, A. 1885. Sopra uno scheletro fossile di Compagno, L.J.V., and T. Roberts. 1982. Fresh- Myliobates esistente nel museo Gazola in Ve- water stingrays (Dasyatidae) of southeast Asia rona. Memoria del Reale Istituto Veneto di and New Guinea, with description of a new Scienze Lettere ed Arti 22: 1±13. species of Himantura and reports of unidenti- Dingerkus, G. 1995. Relationships of the pota- ®ed species. Environmental Biology of Fishes motrygonin stingrays (Chondrichthyes: Batifor- 7(4): 321±339. mes: Myliobatidae). In M.I. Oetinger and G.D. Compagno, L.J.V., and T. Roberts. 1984. Marine Zorzi (editors), The biology of freshwater elas- and freshwater stingrays (Dasyatidae) of west mobranches: 32±37. Journal of Aquariculture Africa, with description of a new species. Pro- and Aquatic Sciences 7: 1±161. ceedings of the California Academy of Scienc- Dingerkus, G., and L.D. Uhler. 1977. Enzyme es 43(18): 283±300. clearing of alcian blue stained whole smalle Cope, E.D. 1879. A stingray from the Green River vertebrates for demonstration of cartilage. Stain shales of Wyoming. American Naturalist 13: Technology 52(4): 229±232. 333. Donoghue, M.J., J.A. Doyle, J. Gauthier, A.G. Cope, E.D. 1884. The Vertebrata of the Tertiary Kluge, and T. Rowe. 1989. The importance of formations of the West. Report of the United fossils in phylogeny reconstruction. Annual Re- States Geological and Geographical Survey view of Ecology and Systematics 20: 431±460. Territories 3: 1±1009. Dunn, K.A., J.D. McEachran, and R.L. Honeycutt. Cvancara, A.M., and J.W. Hoganson. 1993. Ver- 2003. Molecular phylogenetics of myliobatiform tebrates of the Cannonball Formation (Paleo- ®shes (Chondrichthyes: Myliobatiformes), with cene) in North and South Dakota. Journal of comments on the effects of missing data on par- Vertebrate Paleontology 13(1): 1±23. simony and likelihood. Molecular Phylogenetics Daniel, J.F. 1934. The elasmobranch ®shes, 3rd and Evolution 27(2003): 259±270. ed. Berkeley: University of California Press. El-Toubi, M.R., and A.R. Hamdy. 1959. Studies Dartevelle, E., and E. Casier. 1943. Les poissons on the head skeleton of Rhinobatus halavi, fossiles du Bas-Congo et des reÂgions voisines. Rhynchobatus djiddensis and Trygon kuhlii. Annales du MuseÂe du Congo Belge seÂrie A Publications of the Marine Biological Station, (MineÂralogy, GeÂology, PaleÂontology) 2(1): 1± Al-Ghardaqa, Red Sea 10: 3±40. 200. Estes, R. 1964. Fossil vertebrates from the Late Dartevelle, E., and E. Casier. 1949. Les poissons Cretaceous Lance Formation, eastern Wyo- fossiles du Bas-Congo et des reÂgions voisines. ming. University California Publications in Annales du MuseÂe du Congo Belge seÂrie A Geological Sciences 49: 1±180. (MineÂralogie GeÂologie PaleÂontologie) 2(2): Estes, R. 1976. Middle Paleocene lower verte- 205±255. brates from the Tongue River Formation, south- Dartevelle, E., and E. Casier. 1959. Les poissons eastern Montana. Journal of Paleontology fossiles du Bas-Congo et des reÂgions voisines 50(3): 500±520. (troisieme partie). Annales du MuseÂe du Congo Farris, J.S. 1969. A successive approximations ap- Belge seÂrie A (MineÂralogie GeÂologie PaleÂon- proach to character weighting. Systematic Zo- tologie) 2(3): 257±568. ology 18(4): 374±385. de Beer, G.R. 1937. The development of the ver- Farris, J.S. 1988. Hennig86. Program and docu- tebrate skull. Chicago: University of Chicago mentation. Port Washington, NY: Published by Press. the author. Delvaux, D. 1995. Age of Lake Malawi (Nyasa) Farris, J.S. 1989. The retention index and rescaled and water level ¯uctuations. Rapport Annuel du consistency index. Cladistics 5(4): 417±419. DeÂpartement de GeÂologie et de MineÂralogie du Fitzhugh, K. 1989. Cladistics in the fast lane. MuseÂe Royal de l'Afrique Centrale 1993 and Journal of the New York Entomological Soci- 1994: 99±108. ety 97: 234±241. 126 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Fowler, H.W. 1941. The ®shes of the groups Elas- and paleobiogeographic implications. Journal mobranchii, Holocephali, Isospondyli and Os- of Vertebrate Paleontology 23(1): 28±46. tariophysi obtained by the United States Bureau Gohar, H.A.F., and A.R. Bayoumi. 1959. On the of Fisheries steamer ``Albatross'' in 1907 to anatomy of Manta ehrenbergi, with notes on 1910, chie¯y in the Philippine Islands and ad- Mobula kuhlii. Publications of the Marine Bi- jacent seas. Bulletin of the United States Na- ological Station, Al-Ghardaqa, Red Sea 10: tional Museum 100(13): x, 1±879. 191±238. Fowler, H.W. 1947. New taxonomic names of Goloboff, P.A. 1993. Estimating character weights ®sh-like vertebrates. Notulae Naturae 187. during tree search. Cladistics 9(1): 83±91. Frailey, C. 1986. Late Miocene and Holocene Goloboff, P. 1999. Nona (no name) vers. 2. Tu- mammals exclusive of the Notoungulata, of the cumaÂn: Published by the author. rio Acre region, western Amazonia. Natural Grande, L. 1980. Paleontology of the Green River History Museum of Los Angeles County Con- Formation, with a review of the ®sh fauna. tributions in Science 347: 1±46. Geological Survey of Wyoming Bulletin 63, 1st Frickhinger, K. 1995. Fossil atlas. Fishes. Stutt- ed., xviii ϩ 333 pp. gart: Baensch. Grande, L. 1984. Paleontology of the Green River Funnel, B.M. 1990. Global and European shore- Formation, with a review of the ®sh fauna. lines, stage by stage. In R.N. Ginsberg and B. Geological Survey of Wyoming Bulletin 63, Beaudoin (editors), Cretaceous resources, 2nd ed., xviii ϩ 333 pp. events and rhythms: 221±235. Amsterdam: Grande, L. 1985. The use of paleontology in sys- Kluwer Academic Publishers. tematics and biogeography, and a time control Garman, S. 1877. On the pelvis and external sex- re®nement for historical biogeography. Paleo- ual organs of selachiens, with special reference biology 11(2): 234±243. to the new genera Potamotrygon and Disceus. Grande, L. 1989. The Eocene Green River lake Proceedings of the Boston Society of Natural system, Fossils Lake and the history of the History 19: 197±215. North American ®sh fauna. In J. Flynn (editor), Garman, S. 1888. On the lateral-line canal system Mesozoic/Cenozoic vertebrate paleontology: of the Selachia and Holocephala. Bulletin of the classic localities, contemporary approaches: Museum of Comparative Zoology of Harvard 18±28. Washington DC: American Geophysical University 17(2): 57±119. Union. Garman, S. 1913. The Palgiostomia. Memoirs of Grande, L. 1994. Studies of paleoenvironments the Museum of Comparative Zoology of Har- and historical biogeography in the Fossil Butte vard College 36(1): 1±515, 75 pls. and Laney members of the Green River For- Gauthier, J., A.G. Kluge, and T. Rowe. 1988. Am- mation. Contributions to Geology 30(1): 15±32. niote phylogeny and the importance of fossils. Grande, L. 1998. Fossil Lake. Natural History Cladistics 4(2): 105±210. 107(6) [July/August 1998]: 66±69. Gayet, M. 1991. ``Holostean'' and teleostean ®shes Grande, L. 2001. An updated review of the ®sh of Bolivia. In R. Suarez (editor), Fosiles y facies faunas from the Green River Formation, the de Bolivia. Revista Tecnica de Yacimientos Pe- world's most productive freshwater LagerstaÈt- troliferos Fiscales Bolivianos 12: 453±494. ten. In G.F. Gunnell (editor), Eocene biodiver- Gayet, M., L.G. Marshall, and T. Sempere. 1992. sity: unusual occurrences and rarely sampled The Mesozoic and Paleogene vertebrates of Bo- habitats: 1±38. New York: Kluwer Academic/ livia and their stratigraphic context: a review. Plenum Publishers. In R. SuaÂrez-Soruco (editor), FoÂsiles y facies Grande, L. 2002. Portraits of nature's past. In the de Bolivia, vol. 1. Revista TeÂcnica de Yaci- Field, Spring 2002: 16±17. mientos PetrolõÂferos Fiscales Bolivianos 12(3, Grande, L., and W. Bemis. 1991. Osteology and 4): 393±433. Santa Cruz. phylogenetic relationships of fossil and Recent Gayet, M., and F.J. Meunier. 1998. Maastrichtian paddle®shes (Polyodontidae) with comments to early late Paleocene freshwater Osteichthyes on the interrelationships of Acipenseriformes. of Bolivia: additions and comments. In L.R. Society of Vertebrate Paleontology Memoir 1: Malabarba, R.E. Reis, R.P. Vari, Z.M.S. Lu- 1±121. cena, and C.A.S. Lucena (editors), Phylogeny Grande, L., and W. Bemis. 1998. A comprehen- and classi®cation of Neotropical ®shes: 85± sive phylogenetic study of amiid ®shes (Ami- 110. Porto Alegre: EDIPUCRS. idae) based on comparative skeletal anatomy. Gayet, M., M. JeÂgu, J. Bocquentin, and F.R. Ne- An empirical search for interconnected patterns gri. 2003. New characoids from the Upper Cre- of natural history. Journal of Vertebrate Pale- taceous and Paleocene of Bolivia and the Mio- ontology (suppl.) 18(1): 1±690. Pliocene of Brazil: phylogenetic position and Grande, L., and H.P. Buchheim. 1994. Paleonto- 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 127

logical and sedimentological variation in early Part IV. General morphology of the head in Eocene Fossil Lake. Contributions to Geology ®sh. Acta Zoologica 24: 1±188. 30(1): 33±56. Hoorn, C. 1993. Marine incursions and the in¯u- Graybeal, A. 1998. Is it better to add taxa or char- ence of Andean tectonics on the Miocene de- acters to a dif®cult phylogenetic problem? Sys- positional history of northwestern Amazonia: tematic Biology 47(1): 9±17. results of a palynostratigraphic study. Palaeo- Halstead, B.W. 1970. Poisonous and venomous geography Palaeoclimatilogy Palaeoecology marine animals of the world, vol. 3. Washing- 105: 267±309. ton DC: U.S. Government Printing Of®ce. Hoorn, C., J. Guerrero, G.A. Sarmiento, and M.A. Halstead, B.W., and F.R. Modglin. 1950. A pre- Lorente. 1995. Andean tectonics as a cause for liminary report on the venom apparatus of the changing drainage patterns in Miocene northern bat-ray, Holorhinus californicus. Copeia South America. Geology 23: 237±240. 1950(3): 165±175. Huelsenbeck, J.P. 1991. When are fossils better Halter, M.C. 1989. Additions to the ®sh fauna of than extant taxa in phylogenetic analysis? Sys- N.W. Europe. A new dasyatid genus from the tematic Zoology 40(4): 458±469. early Paleocene (Danian) of the Limburg area, Itoigawa, J., H. Nishimoto, H. Karasawa, and Y. Belgium. Tertiary Research 10(4): 179±191. Okumura. 1985. Miocene fossils of the Mizun- Haseman, J.D. 1912. The relationship of the ge- ami group, central Japan. 3. Elasmobranchs. nus Priscacara. Bulletin of the American Mu- Monographs of the Mizunami Fossil Museum seum of Natural History 31(3): 97±101. 5: 1±99. Hatai, K., and T. Kotaka. 1962. Dasybatus from Jaekel, O.M.J. 1894. Die eocaÈnen Selachier vom the Japanese Miocene. Transactions and Pro- Monte Bolca. Berlin: Springer. ceedings of the Palaeontological Society of Ja- Jarvik, E. 1980. Basic structure and evolution of pan n. ser. 45: 201±205. vertebrates, vol. 1. London: Academic Press. Heemstra, P.C., and M.M. Smith. 1980. Hexatry- Johnson, T.C., C.A. Scholz, M.R. Talbot, K. Kelts, gonidae, a new family of stingrays (Mylioba- R.D. Ricketts, G. Ngobi, K. Beuning, I. Ssem- tiformes: Batoidea) from South Africa, with manda, and J.W. McGill. 1996. Late Pleistocene comments on the classi®cation of batoid ®shes. desiccation of Lake Victoria and rapid evolution Ichthyological Bulletin of the J.L.B. Smith In- of cichlid ®shes. Science 273: 1091±1093. stitute of Ichthyology 43: 1±17. Kearney, M. 2002. Fragmentary taxa, missing data, Henningsen, A.D. 1996. Captive husbandry and and ambiguity: mistaken assumptions and con- bioenergetics of the spiny butter¯y ray Gym- clusions. Systematic Biology 51(2): 369±381. nura altavela (Linnaeus). Zoo Biology 15: Kesteven, H.L. 1942. The evolution of the skull 135±142. and cephalic muscles: a comparative study of Herman, J., M. Hovestadt-Euler, and D.C. Hoves- development and adult morphology. Part I. tadt. 1989. Additions to the Eocene ®sh fauna of Fishes. Memoirs of the Australian Museum Belgium. 9. Discovery of Eomobula gen. et 8(1): 1±63. spec. nov. (Mobulidae, Chondrichthyes) from Knowleton, N., L.A. Weigt, L.A. Soloranzo, D.K. the Ypresian. Tertiary Research 10(4): 175±178. Mills, and E. Birmingham. 1993. Divergence in Hillis, D.M., B.K. Mable, and C. Moritz. 1996. proteins, mitochondrial DNA and reproductive Applications of molecular systematics. In D.M. compatibility across the isthmus of Panama. Hillis, B.K. Mable, and C. Moritz (editors), Science 260: 1629±1632. Molecular systematics: 515±543. Sunderland: Kornet, D.J., and H. Turner. 1999. Coding poly- Sinauer. morphism for phylogeny reconstruction. Sys- Hoberg, E.P., D.R. Brooks, H. Molina-UrenÄa, and tematic Biology 48(2): 365±379. E. Erbe. 1998. Echinocephalus janzeni n. sp. LaMarca, M.J. 1963. The embryonic spiracular (Nematods: Gnathostomatidae) in Himantura fold of the yellow stingray Urolophus jamai- paci®ca (Chondrichthyes: Myliobatiformes) censis. Copeia 1963(3): 593±595. from the Paci®c coast of Costa Rica and Mexico, Landini, W. 1977. Revisione degli ittiodontoliti with a historical biogeographical analysis of the pliocenici della collezione Lawley. Paleonto- genus. Journal of Parasitology 84: 571±581. graphia Italica 70(n. ser. 40): 92±134, tav. V. Holmgren, N. 1940. Studies on the head in ®shes. Landini, W., and L. Sorbini. 1996. Ecological and Part I. Development of the skull in sharks and trophic relationships of Eocene Monte Bolca rays. Acta Zoologica 21: 51±257. (Pesciara) ®sh fauna. In A. Cherchi (editor), Holmgren, N. 1942. Studies on the head in ®shes. Autoecology of selected fossil organisms: Part III. The phylogeny of elasmobranch ®shes. achievements and problems. Bollettino della Acta Zoologica 23: 129±262. SocietaÁ Paleontologica Italiana, special vol. 3: Holmgren, N. 1943. Studies on the head in ®shes. 105±112. Modena. 128 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

Larrazet, M. 1886. Des pieÁces de la peau des qu- R.L. Cifelli, and J.J. Flynn (eds.), Vertebrate elques seÂlaciens fossiles. Bulletin SocieteÂ Geo- paleontology in the Neotropics. The Miocene logique d'France 14(3): 255±277. fauna of La Venta Colombia: 67±91. Washing- Last, P.R., and L.J.V. Compagno. 2000. Family ton, DC: Smithsonian Institution Press. Urolophidae. In K.E. Carpenter and V. Niem Lundberg, J.G. 1998. The temporal context for the (editors), FAO species identi®cation guide for diversi®cation of Neotropical ®shes. In L.R. ®sheries purposes. The living marine resources Malabarba, R.E. Reis, R.P. Vari, Z.M.S. Lu- of the western Central Paci®c Ocean: 1468± cena, and C.A.S. Lucena (editors), Phylogeny 1476. Rome: FAO. and classi®cation of Neotropical ®shes: 49±68. Last, P.R., and M. Gomon. 1987. New Australian Porto Alegre: EDIPUCRS. ®shes. Part 15. New species of Trygonoptera Lundberg, J.G., and B. Chernoff. 1992. A Mio- and Urolophus. Memoirs of the Museum of cene fossil of the Amazonian ®sh Arapaima Victoria 48(1): 63±72. (Teleostei: Arapaimidae), from the Magdalena Last, P.R., and J.D. Stevens. 1994. Sharks and Riverregion of ColombiaÐbiogeographic and rays of Australia. Melbourne: CSIRO. evolutionary implications. Biotropica 24: 2±14. Li, G.-q. 1994. Systematic position of the Austra- Lundberg, J.G., L.G. Marshall, J. Guerrero, B. lian fossil osteoglossid ®sh ²Phareodus (ϭ Horton, M.C.S.L. Malabarba, and F. Wesse- Phareoides) queenslandicus Hills. Memoirs of lingh. 1998. The stage for Neotropical ®sh di- the Queensland Museum 37: 287±300. versi®cation: a history of tropical South Amer- Li, G.-q., L. Grande, and M.V.H. Wilson. 1997. ican rivers. In L.R. Malabarba, R.E. Reis, R.P. The species of ²Phareodus (Telestei: Osteo- Vari, Z.M.S. Lucena, and C.A.S. Lucena (edi- glossidae) from the Eocene of North America tors), Phylogeny and classi®cation of Neotrop- and their phylogenetic relationships. Journal of ical ®shes: 13±48. Porto Alegre: EDIPUCRS. Vertebrate Paleontology 17(3): 487±505. MacGinitie, H.D. 1969. The Eocene Green River Lindberg, G.U., and M.I. Legeza. 1959. Fishes of ¯ora of northwestern Colorado and northeast- the Sea of Japan and the adjacent areas of the ern Utah. University of California Publications Sea of Okhotsk and the Yellow Sea. Part 1: Am- in Geological Sciences 83: 1±140. phioxi, Petromyzones, Myxini, Elasmobranchii, Maisey, J.G. 1991. Rhinobatos Link, 1790. In J.G. Holocephali. Keys to the Fauna of the USSR Maisey (editor), Sanatana fossils: an illustrated No. 68: i±iv ϩ 1±198. Moscow: Zoological In- atlas: 110±116. Neptune: T.F.H. Publications. stitute of the Academy of Sciences of the USSR. Maisey, J.G. 1996. Discovering fossil ®shes. New [in Russian; translated by the Israel Program for York: Henry Holt. Scienti®c Translations, Jerusalem, 1967] Maisey, J.G., and M.R. de Carvalho. 1997. A new Long, J.A. 1995. The rise of ®shes. Baltimore: look at old sharks. Nature 385: 779±780. Johns Hopkins University Press. Marques, F.P. de. 2000. Evolution of Neotropical Lovejoy, N.R. 1996. Systematics of myliobatoid freshwater stingrays and their parasites: taking elasmobranchs: with emphasis on the phyloge- into account space and time. Unpublished PhD ny and historical biogeography of neotropical diss., University of Toronto, Toronto. freshwater stingrays (Potamotrygonidae: Raji- Marsh, O.C. 1877. New vertebrate fossils. Amer- formes). Zoological Journal of the Linnaen So- ican Journal of Science 14: 249±256. ciety 117: 207±257. McEachran, J.D. 1982. Chondrichthyes. In Syn- Lovejoy, N.R. 1997. Stingrays, parasites, and opsis and classi®cation of living organisms, Neotropical biogeography: a closer look at vol. 2: 831±858. New York: McGraw-Hill. Brooks et al.'s hypothesis concerning the ori- McEachran, J.D. 1995. Peces batoideos. In W. Fi- gins of Neotropical freshwater rays. Systematic scher, F. Krupp, W. Schneider, C. Sommer, K.E. Biology 46(1): 218±230. Carpenter, and V.E. Niem (editors), Paci®co Lovejoy, N.R., E. Bermingham, and A.P. Martin. Centro Oriental, vol. II: 746±792. Rome: FAO. 1998. Marine incursion into South America. McEachran, J.D., and M.R. de Carvalho. 2002. Nature 396: 421±422. Dasyatidae. In K.E. Carpenter (editor), The liv- Lundberg, J.G. 1993. African-South American ing marine resources of the western Central At- freshwater ®sh clades and continental drift: lantic. Vol. 1: Introduction, molluscs, crusta- problems with a paradigm. In P. Goldblatt (ed- ceans, hag®shes, sharks, batoid ®shes and chi- itor), Biological relationships between Africa maeras. FAO Species Identi®cation Guide for and South America: 156±199. New Haven: Fisheries Purposes and American Society of Yale University Press. Ichthyologists and Herpetologists Special Pub- Lundberg, J.G. 1997. Fishes of the La venta fau- lication 5: 562±571. Rome: FAO. na: additional taxa, biotic and paleoenviron- McEachran, J.D., and K.A. Dunn. 1998. Phylo- mental implications. In R.F. Kay, R.H. Madden, genetic analysis of skates, a morphologically 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 129

conservative group of elasmobranchs. Copeia Miyake, T., J.D. McEachran, P.J. Walton, and B. 1998(3): 271±293. Hall. 1992b. Development and morphology of McEachran, J.D., and T. Miyake. 1990. Zooge- rostral cartilages in batoid ®shes (Chondrich- ography and bathymetry of skates (Chondrich- thyes: Batoidea), with comments on homology thyes, Rajoidei). In H.L. Pratt, S.H. Gruber, and within vertebrates. Biological Journal of the T. Taniuchi (editors), Elasmobranchs as living Linnaen Society 46: 259±298. resources: advance in the biology, ecology, sys- Murray, A.M., and M.V.H. Wilson. 1999. Contri- tematics, and behavior, and status of the ®sh- butions of fossils to the phylogenetic relation- eries: 304±326. NOAA (National Oceanic and ships of the percopsiform ®shes (Teleostei: Par- Atmospheric Administration) Technical Report acanthopterygii): order restored. In G. Arratia NMFS (National Marine Fisheries Service) 90. and H.-P. Scultze (editors), Mesozoic ®shes 2, McEachran, J.D., K. Dunn, and T. Miyake. 1996. systematics and fossil record: 397±411. Mu- Interrelationships of batoid ®shes. In M.L.J. nich: Dr. Friedrich Pfeil. Stiassny, G.D. Johnson, and L. Parenti (edi- Mustafa, H., and I. Zalmout. 2002. Elasmo- tors), Interrelationships of ®shes: 63±84. San branchs from the late Eocene Wadi Esh-Shal- Diego: Academic Press. lala Formation of Qa'Faydat ad Dahikiya, east McGrew, P.O. 1975. Taphonomy of Eocene ®sh Jordan. Tertiary Research 21(1±4): 77±94. from Fossil Basin, Wyoming. Fieldiana Geol- Naylor, G.J.P. 1992. The phylogenetic relation- ogy 33(14): 257±270. ships among requiem and hammerhead sharks: McGrew, P.O., and M. Casilliano. 1975. The geo- inferring phylogeny when thousands of equally logic history of Fossil Butte National Monu- most parsimonious trees result. Cladistics ment and Fossil Basin. National Park Service 1992(4): 295±318. Occasional Paper 3. Naylor, G.J.P., and D.C. Adams. 2001. Are the Meek, S.E., and S.F. Hildebrand. 1923. The ma- fossil data really at odds with the molecular rine ®shes of Panama. Part I. Field Museum of data? Morphological evidence for Cetartiodac- Natural History Zoological Series 15(publica- tyla phylogeny reexamined. Systematic Biolo- tion no. 215): 1±330. gy 50(3): 444±453. Meng, Q. 1984. Study of the neurocranium of Ba- Nelson, G.J. 1969. Gill arches and the phylogeny tomorpha of China. Acta Zoologica Sinica of ®shes, with notes on the classi®cation of ver- 30(2): 181±187. [In Chinese] Meyer, R.L. 1974. Late Cretaceous elasmobranches tebrates. Bulletin of the American Museum of from the Mississippi east Texas embayments of Natural History 141: 475±552. the gulf coastal plain. Unpublished PhD thesis, Nelson, G.J. 1984. Identity of the anchovy Hil- Southern Methodist University, Dallas, 419 pp. debrandichthys setiger with notes on relation- Miyake, T. 1988. The systematics of the stingray ships and biogeography of the genera Engraulis genus Urotrygon, with comments on the inter- and Cetengraulis. Copeia 1984(2): 422±427. relationships within Urolophidae (Chondrich- Nishida, K. 1990. Phylogeny of the suborder My- thyes, Myliobatiformes). Unpublished PhD liobatidoidei. Memoirs of the Faculty of Fish- diss., Texas A&M University, College Station. eries Hokkaido University 37(1/2): 1±108. Miyake, T., and J.D. McEachran. 1986. Taxonomy Nishida, K., and K. Nakaya. 1990. Taxonomy of of the stingray genus Urotrygon (Myliobatifor- the genus Dasyatis (Elasmobranchii, Dasyati- mes: Urolophidae): preliminary results based didae) from the North Paci®c. In H.L. Pratt, on external morphology. In T. Uyeno, R. Arai, S.H. Gruber, and T. Taniuchi (editors), Elas- T. Taniuchi, and K. Matsuura (editors), Indo- mobranchs as living resources: advance in the Paci®c ®sh biology: proceedings of the second biology, ecology, systematics, and behavior, international conference on Indo-Paci®c ®shes: and status of the ®sheries. NOAA (National 291±302. Tokyo: Ichthyological Society of Japan. Oceanic and Atmospheric Administration) Miyake, T., and J.D. McEachran. 1991. The mor- Technical Report NMFS (National Marine phology and evolution of the ventral gill arch Fisheries Service), no. 90. skeleton in batoid ®shes (Chondrichthyes: Ba- Nixon, K.C. 1999. WinClada ver. 1.0. Ithaca, NY: toidea). Zoological Journal of the Linnaen So- Published by the author. ciety 102: 75±100. Nixon, K.C., and J.I. Davis. 1991. Polymorphic Miyake, T., J.D. McEachran, and B. Hall. 1992a. taxa, missing values, and cladistic analysis. Edgeworth's legacy of cranial muscle devel- Cladistics 7(3): 233±241. opment with an analysis of muscles in the ven- Northcutt, R.G. 1978. Brain organization in the tral gill arch region of batoid ®shes (Chondrich- cartilaginous ®shes. In E.S. Hodgson and R.F. thyes: Batoidea). Journal of Morphology 212: Mattewson (editors), Sensory biology of 213±256. sharks, skates and rays: 117±193. Arlington, 130 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

VA: Of®ce of Naval Research, Department of es from Lee Creek Mine, Aurora, North Caro- the Navy. lina. In C.E. Ray and D.J. Bohaska (editors), Notarbartolo-di-Sciara, G. 1987. A revisionary Geology and paleontology of the Lee Creek study of the genus Mobula Ra®nesque, 1810 Mine, North Carolina, III. Smithsonian Contri- (Chondrichthyes: Mobulidae), with the descrip- butions to Paleontology 9: 71±202. tion of a new species. Zoological Journal of the Rae, T.C. 1998. The logical basis for the use of Linnaean Society 91: 1±91. continuous characters in phylogenetic system- Noubhani, A., and H. Cappetta. 1997. Les Orec- atics. Cladistics 14(3): 221±228. tolobiformes, Carcharhiniformes et Myliobati- Ramos, T.C. 1997. Tree gardener 2.2. Program formes (Elasmobranchii, Neoselachii) des bas- and documentation. SaÄo Paulo, Brazil. sins aÁ phosphate du Maroc (Maastrichtien-Lu- Rice, K.A., M.J. Donoghue, and R.G. Olmstead. teÂtien basal). Palaeo Ichthyologica 8: 1±327. 1997. Analyzing large data sets: rbcL 500 re- Novacek, M.J. 1992. Fossils, topology, missing visited. Systematic Biology 46(3): 554±563. data, and the higher level phylogeny of eutherian Richter, M. 1984. Dental histology of a characid mammals. Systematic Biology 41(1): 58±73. ®sh from the Plio-Pleistocene of Acre, Brazil. O'Leary, M.A., and J.H. Geisler. 1999. The po- Zoologica Scripta 13: 69±79. sition of Cetacea within Mammalia: phyloge- Roberts, T.R. 1975. Characoid ®sh teeth from de- netic analysis of morphological data from ex- posits in the Cuenca basin, Ecuador. Journal of tinct and extant taxa. Systematic Biology 48(3): Zoology (London) 175: 259±271. 455±490. Romer, A.S. 1966. Vertebrate paleontology. Chi- érvig, T. 1951. Histologic studies of placoderms cago: University of Chicago Press. and fossil elasmobranchs 1: The endoskeleton, Rosa, R.S. 1985. A systematic revision of the with remarks on the hard tissues of lower ver- South American freshwater stingrays (Chon- tebrates in general. Arkiv foÈr Zoologi, ser. 2, drichthyes, Potamotrygonidae). Unpublished band 2(2): 321±454, 8 pls. PhD diss., College of William and Mary, Wil- érvig, T. 1958. Pycnaspis splendens, new genus, liamsburg, xvi ϩ 523 pp. new species, a new ostracoderm from the Upper Rosa, R.S., H. Castello, and T.B. Thorson. 1987. Ordovician of North America. Proceedings of Plesiotrygon iwamae, a new genus and species the United States National Museum 108: 1±23. of neotropical freshwater stingray. Copeia Patterson, C. 1981. Signi®cance of fossils in deter- 1987(2): 447±458. mining evolutionary relationships. Annual Re- Rosen, D.E., P.L. Forey, B.G. Gardiner, and C. view of Ecology and Systematics 12: 195±223. Patterson. 1981. Lung®shes, tetrapods, paleon- Patterson, C. 1993. Osteichthyes: Teleostei. In tology, and plesiomorphy. Bulletin of the M.J. Benton (editor), The fossil record 2: 621± American Museum of Natural History 167(4): 656. London: Chapman and Hall. 163±275. Patterson, C., and G.D. Johnson. 1997. The data, Rosenberger, L.J. 2001a. Pectoral ®n locomotion the matrix, and the message: comments on Be- in batoid ®shes: undulation vs.oscillation. Jour- gle's ``Relationships of the Osmeroid Fishes.'' nal of Experimental Biology 204: 379±394. Systematic Biology 46(2): 358±365. Rosenberger, L.J. 2001b. Phylogenetic relation- Patterson, C., and D.E. Rosen. 1989. The Para- ships within the stingray genus Dasyatis (chon- canthopterygii revisited: order and disorder. In drichthyes: Dasyatidae). Copeia 2001(3): 615± D. Cohen (editor), Papers on the systematics of 627. gadiform ®shes. Natural History Museum of Sahni, A., and D.K. Mehrotra. 1981. The elas- Los Angeles County Science Series 32: 5±36. mobranch fauna of coastal Miocene sediments Paxton, J.R., D.F. Hoese, G.R. Allen, and J.E. Han- of peninsular India. Biological Memoirs Luck- ley. 1989. Zoological catalogue of Australia. now 5: 83±121. Vol. 7. Pisces. Petromyzontidae to Carangidae. Saint-Seine, P. 1949. Les poissons des calcaires Zoological Catalogue of Australia 7: 1±665. lithographiques de CeÂrin (Ain). Nouvelles Ar- Perea, D., and M. Ubilla. 1989. Selacifauna del chives du MuseÂum d'Histoire Naturelle de Mioceno superior del uruguay. BoletõÂn de la Lyon 2: 1±357, 25 pls. Sociedad ZooloÂgica del Uruguay 5: 11±12. Schaeffer, B. 1963. Cretaceous ®shes from Boliv- Prendini, L. 2001. Species or supraspeci®c taxa as ia with comments on pristid evolution. Ameri- terminals in cladistic analysis? Groundplans can Museum Novitates 2159: 1±20. versus exeplars revisited. Systematic Biology Schaeffer, B. 1981. The xenacanth shark neuro- 50(2): 290±300. cranium, with notes on elasmobranch mono- Purdy, R.W., V.P. Schneider, S.P. Applegate, J.H. phyly. Bulletin of the American Museum of McLellan, R.L. Meyer, and B.H. Slaughter. Natural History 169(1): 1±66. 2001. The Neogene sharks, rays, and bony ®sh- Schaeffer, B., and M. Mangus. 1965. Fossil lakes 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 131

from the Eocene Green River Formation dis- Philadelphia. Taxonomic, systematic, and his- closes its 10-million year history. Natural His- torical perspectives. Academy of Natural Sci- tory 74: 10±21. ences Special Publication 15: 1±434. Schuh, R.T. 2000. Biological systematics. Princi- Stevens, J.D., and P.R. Last. 1994. Sharks, rays ples and applications. Ithaca, NY: Cornell Uni- and chimaeras. In J.R. Paxton and W.N. Es- versity Press. chmeyer (editors), Encyclopaedia of ®shes: 60± Schultze, H.-P. 1991. Lung®sh from the El Molino 70. San Diego: Academic Press. (Late Cretaceous) and Santa Lucia (early Pa- Straney, D.O. 1982. Review of ``Advances in cla- leocene) formations in southcentral Bolivia. In distics: proceedings of the ®rst meeting of the R. Suarez (editor), Fosiles y facies de Bolivia. Willi Hennig society,'' edited by V.A. Funk and Revista Tecnica de Yacimientos Petroliferos D.R. Brooks. Systematic Zoology 31: 337±341. Fiscales Bolivianos 12: 441±448. Stromer, E. 1904. Myliobatiden aus dem Mitte- Schultz, O. 1998. Die Knorpel- und Knochen- leocaÈn der bayerischen Alpen. Zeitschrift der ®schfauna (excl. Otolithen) aus dem Karpat des Deutschen geologischen Gesellschaft 56: 249± Korneuburger Beckens (NiederoÈsterreich). Bei- 265, 2 taf. traÈge PalaÈontologie 23: 295±323. Stromer, E. 1905. Die ®schreste des Mittleren und Shen, S.-C. 1986. A new species of stingray Hex- oberen eocaÈns von AÈ gypten. I. Teil: die Sela- atrygon taiwanensis from Taiwan Strait. Jour- chier. BeitraÈge zur PalaÈontologieund Geologie nal of Taiwan Museum 39(1): 175±180. 18: 37±58, 2 taf. Shen, S.-C., and C.-S. Liu. 1984. A new stingray Stromer, E. 1927. Ergebnisse der Forschungsre- of the genus Hexatrygon from Taiwan. Acta isen Prof. E. Stromers in den WuÈsten AÈ gyptens. Oceanographic Taiwanica 15: 201±206. II. Wirbeltier-Reste der Baharije-Stufe (unter- Shirai, S. 1992. Squalean phylogeny: a new stes cenoman). 9. Die Plagiostomen, mit einem framework for of squaloid sharks and related Anhang uÈber kaÈno- und mesozoische RuÈcken- taxa. Sapporo: Hokkaido University Press. ¯ossenstacheln von Elasmobranchiern. Abhan- Simmons, N.B., and J.A. Geisler. 2002. Sensitivity dlungen der Bayerischen Akademie der Wis- analysis of different methods of coding taxo- senschaften Mathematische-Naturwissenschaf- nomic polymorphism: an example from higher- tiche Abteilung 31(5): 1±64, 3 taf. level bat phylogeny. Cladistics 18(6): 571±584. Sullivan, R. 1980. A stratigraphical evaluation of Smith, A.G., D.G. Smith, and B.M. Funnell. 1994. the Eocene rocks of southwestern Wyoming. Atlas of Mesozoic and Cenozoic coastlines. Geological Survey of Wyoming Investigative Cambridge: Cambridge University Press. Reports 20: 1±48. Smith, G.R. 1992. Introgression in ®shes: signif- Thorson, T.B. 1970. Freshwater stingrays, Pota- icance for paleontology, cladistics and evolu- motrygon spp: failure to concentrate urea when tionary rates. Systematic Biology 41(1): 41±57. exposed to saline medium. Life Sciences 9: Smith, J.L.B. 1952. Tropical ®shes recently found 893±900. in South Africa. Annals and Magazine of Nat- Thorson, T.B., D.R. Brooks, and M.A. Mayes. ural History 12(v): 1020±1025. 1983a. The evolution of freshwater adaptation Snelson, F.F., Jr., S.E. Williams-Hooper, and T.H. in stingrays. National Geographic Society Re- Schmidt. 1988. Reproduction and ecology of search Reports 15: 663±694. the Atlantic stingray, Dasyatis sabina, in Flor- Thorson, T.B., C.M. Cowan, and D.E. Watson. ida coastal lagoons. Copeia 1988(3): 729±739. 1967. Potamotrygon spp.: elasmobranchs with Snelson, F.F., Jr., S.E. Williams-Hooper, and T.H. low urea content. Science 158: 375±377. Schmidt. 1989. Biology of the bluntnose sting- Thorson, T.B., J.K. Langhammer, and M.I. Oetin- ray, Dasyatis sayi, in Florida coastal lagoons. ger. 1983b. Reproduction and development of Bulletin of Marine Science 45: 15±25. the South American freshwater stingrays, Po- Soler-GijoÂn, R., and N. LoÂpez-MartõÂnez. 1998. tamotrygon circularis and P. motoro. Environ- Sharks and rays (Chondrichthyes) from the up- mental Biology of Fishes 9(1): 3±24. per Cretaceous red beds of the south-central Thorson, T.B., and D.E. Watson. 1975. Reassign- Pyrenees (Lleida, Spain): indices of an India- ment of the African freshwater stingray, Pota- Eurasia connection. Palaeogeography Palaeo- motrygon garouaensis, to the genus Dasyatis, climatology Palaeoecology 141(1998): 1±12. on physiologic and morphologic grounds. Cop- Sorbini, L. 1983. La collezione Baja di pesci fos- eia 1975(4): 701±712. sili di Bolca. Verona: Museo Civico di Storia Thorson, T.B., R.M. Wooton, and T.D. Georgi. Naturale. 1978. Rectal gland of freshwater stingrays, Po- Spamer, E.E., E. Daeschler, and L.G. Vostreys- tamotrygon spp. (Chondrichthyes: Potamotry- Shapiro. 1995. A study of fossil vertebrate gonidae). Biological Bulletin 154(3): 508±516. types in the Academy of Natural Sciences of Turner, G.F. 1999. Explosive speciation of African 132 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

cichlid ®shes. In A.E. Magurran and R.M. May guide to fossil sharks and rays from the Cre- (editors), Evolution of biological diversity: taceous of Texas. Lewisville: Before Time. 113±129. New York: Oxford University Press. Werner, C. 1989. Die Elasmobranchier-Fauna des Turner, G.F., O. Seehausen, M.E. Knight, C.J. Al- Gebel Dist Member des Bahariya Formation lender, and R.L. Robinson. 2001. How many (Obercenoman) des Oase Bahariya, Aegypten. species of cichlid ®shes are there in African Paleoichthyologica 5: 1±112. lakes? Molecular Ecology 10: 793±806. Wetzel, W. 1930. Die Quiriquina-Schichten als Tyler, J.C., and F. Santini. 2002. Review and re- Sediment und palaÈontologisches Archiv. Pa- constructions of the tetraodontiform ®shes from laeontolographica 73: 49±106. the Eocene of Monte Bolca, Italy, with com- White, E.I. 1931. The vertebrate faunas of the En- ments on related Tertiary taxa. Museo Civico glish Eocene. Vol. I. From the Thanet sands to di Storia Naturale di Verona Studi e Ricerche the Basement bed of the London Clay. London: sui Giacimenti Terziari di Bolca 9: 47±119. British Museum (Natural History), 123 pp. Underwood, C.J., S.F. Mitchell, and K.J. Velt- Whitley, G.P. 1940. The ®shes of Australia. Part kamp. 1999. Shark and ray teeth from the Hau- 1. The sharks, rays, devil ®shes and other prim- terivian (Lower Cretaceous) of north-east Eng- itive ®shes of Australia and New Zealand. Syd- land. Palaeontology 42(2): 287±302. ney: Royal Zoological Society of New South Vari, R.P. 1984. Systematics of the Neotropical Wales, 230 pp. characiform genus Potamorhina (Pisces: Char- Wiens, J.J. 1998. The accuracy of methods for aciformes). Smithsonian Contribution to Zool- coding and sampling higher-level taxa for phy- ogy 400: 1±36. logenetic analysis: a simulation study. System- Vari, R.P. 1988. The Curimatidae, a lowland Neo- atic Biology 47(3): 397±413. tropical ®sh family (Pisces: Characiformes); Wiens, J.J. 2000. Coding morphological variation distribution, endemism, and phylogenetic bio- within species and higher taxa for phylogenetic geography. In W.R. Heyer and P.E. Vanzolini analysis. In J.J. Wiens (editor), Phylogenetic (editors), Proceedings of a workshop on Neo- analysis of morphological data: 115±145. tropical distribution patterns: 343±377. Rio de Washington, DC: Simthsonian Institution Press. Janeiro: Academia Brasileira de CieÃncias. Woodward, A.S. 1889. Catalog of the fossil ®shes Vari, R.P. 1991. Systematics of the Neotropical in the British Museum. Part I. London: British characiform genus Steindachnerina Fowler (Pi- Museum, 474 pp. sces: Ostariophysi). Smithsonian Contribution Wourms, J.P. 1977. Reproduction and develop- to Zoology 507: 1±118. ment in chondrichthyan ®shes. American Zo- Vari, R.P. 1995. The Neotropical ®sh family Cten- ologist 17: 379±410. oluciidae (Teleostei: Ostariophysi: Characifor- Yabumoto, Y., and T. Uyeno. 1994. Late Mesozoic mes): supra and intrafamilial phylogenetic re- and Cenozoic ®sh faunas of Japan. The Island lationships, with a revisionary study. Smithson- Arc (1994) 3: 255±269. ian Contributions to Zoology 564: 1±97. Yearsley, G.K. 1988. Aspects of the taxonomy Vari, R.P., and A.S. Harold. 2001. Phylogenetic and phylogeny of Australian stingarees (Uro- study of the Neotropical ®sh genera Creagutus lophidae: Batoidea). Unpublished MSc diss., GuÈnther and Piabina Reinhardt (Teleostei: Os- University of Tasmania, Hobart. tariophysi: Characiformes), with a revision of Zalmout, I., and H. Mustafa. 2001. A selachian the cis-Andean species. Smithsonian Contribu- fauna from the Late Cretaceous of Jordan. Ab- tion to Zoology 613: 1±239. hath Al-Yarmouk, Basic Sciences and Engi- Vari, R.P., and S.H. Weitzman. 1990. A review of neering 10(2b): 377±434. the phylogenetic biogeography of the freshwater ®shes of South America. In G. Peters and R. Hutterer (editors), Vertebrates in the tropics: APPENDIX 1 343±377. Bonn: Museum Alexander KoÈnig. Verheyen, E., W. Salzburger, J. Snoeks, and A. COMPARATIVE BATOID (NONSTINGRAY)MATERIAL Meyer. 2003. Origin of the super¯ock of cichlid EXAMINED ®shes from Lake Victoria, East Africa. Science 300: 325±329. Abbreviations: C&S, cleared and stained; XR, Ward, D.J. 1979. Additionas to the ®sh fauna of x-ray radiograph; D, dissected; SW, wet skeleton the English paleogene. 2. A new species of (maintained in diluted ethanol); SD, dry skeleton Dasyatis (sting ray) from the London Clay (Eo- (prepared with dermestid beetles). cene) of Essex, England. Tertiary Research 2(2): 75±81. Pristidae Welton, B.J., and R.F. Farrish. 1993. Collector's Anoxypristis cuspidata: AMNH 3268 (C&S). 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 133

Pristis pectinata: AMNH 44010 (XR), AMNH 2540 (XR). Pristis perotteti: AMNH 55624 (C&S). Pristis pristis: AMNH 44011 (C&S, gill arches only). Pristis zijsron: AMNH 44048 (C&S, gill arches only). Rhinidae Rhina ancylostoma: AMNH 44015 (XR). Rhynchobatidae Rhynchobatus djiddensis: AMNH 44066 (XR). Rhinobatidae Rhinobatos albomaculatus: AMNH 44068 (XR). Rhinobatos annulatus: AMNH 53081 (XR). Rhinobatos irvinei: AMNH 53080 (XR). Fig. 53. Strict consensus tree (length ϭ 57 Rhinobatos lentiginosus: AMNH 8913 (C&S), steps, CI ϭ 0.63, RI ϭ 0.69) of 12 equally most AMNH 44043 (XR), AMNH 44044(XR), parsimonious trees (length ϭ 53 steps, CI ϭ 0.65, AMNH 44045 (XR), AMNH 44046 (XR). RI ϭ 0.72) obtained with phylogenetic analysis Rhinobatos percellens: AMNH 3938 (XR), performed with matrix in table 7 (with pelagic AMNH 55621 (C&S), AMNH 55622 (C&S). stingrays experimentally coded as a single termi- Rhinobatos rhinobatos: AMNH 40994 (XR). nal, Myliobatidae). See appendix 2 for further de- Trygonorhina fasciata: AMNH 44104 (XR). tails. Zapteryx exasperata: OSU 3954 (XR). Zanobatus schoenleinii: MNHN 1989±1528 (D; XR), MNHN uncataloged (XR). MCZ 37130 (XR), TCWC 1903.01 (SW), Rajidae USNM 157958 (C&S). Bathyraja albomaculata: AMNH 44109 (XR). Diplobatis ommata: AMNH 44100 (XR). Bathyraja kincaidii: AMNH uncataloged (XR). Diplobatis pictus: TCWC 1900.01 (SW). Leucoraja erinacea: AMNH 43158 (XR). Narcine bancroftii: AMNH 2488 (C&S); Leucoraja ocellata: AMNH 43158 (XR). AMNH 218276 (SD). Okamejei kenojei: AMNH 44058 (XR). Narcine brasiliensis: UERJ 776.3 (C&S), Psammobatis bergi: AMNH 44020 (XR). UERJ 1157 (C&S), USNM 216829 (C&S). Psammobatis lentiginosa: AMNH 44019 (XR). Narcine entemedor: AMNH 15695 (XR), Psammobatis cf. scobina: AMNH 44044 (XR). AMNH 11800 (XR). Raja binoculata: AMNH 38156 (C&S). Narcine leoparda: USNM 222198 (C&S), Raja clavata: AMNH 1510 (XR). 222198 (D). Raja texana: AMNH 16350 (C&S). Narcine oculifera: CAS 66838 (C&S). Rajella fyllae: AMNH 49510 (XR). Narcine rierai: RUSI 48256 (C&S). Narcine tasmaniensis: AMNH uncataloged Platyrhinidae (XR). Platyrhina limboonkengi: MNHN uncataloged (XR). Narkidae Platyrhina sinensis: AMNH 44055 (C&S, XR), Narke capensis: AMNH 44031 (XR). AMNH 26413 (XR), MNHN 1307 (holo- Narke impennis: AMNH 44054 (XR). type, XR). Temera hardwickii: RMNH 7431 (C&S). Platyrhinoidis triseriata: OSU 40 (XR). Typhlonarke aysoni: AMNH 44047 (XR). Torpedinidae Torpedo fuscomaculata: AMNH uncataloged APPENDIX 2 (D). Torpedo torpedo: AMNH 1509 (XR), AMNH EXPERIMENTAL PHYLOGENETIC ANALYSIS 4128 (C&S). Torpedo sp.: OSU 37 (XR). Pelagic stingrays are coded as a single terminal (Myliobatidae). Matrix is presented in table 7 Narcinidae (note that the basihyal is coded more simplisti- Benthobatis marcida: AMNH 56011 (XR), cally in this matrix and that a single outgroup is 134 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284 used). This analysis is not a substitute for the phy- is resolved as the sister-group of the node uniting logenetic study of the Green River stingrays pre- Potamotrygonidae and Dasyatidae in all mini- sented in this paper, which is based on the matrix mum-length trees. The monophyly of the clade in table 5 (with pelagic stingray genera coded in- Dasyatidae ϩ Potamotrygonidae in the strict con- dividually) and summarized in the cladogram in sensus is supported by a single homoplastic char- ®gure 43 (see text for further discussion). This acter (number 34 in table 7): cartilaginous rod in phylogeny is included here to highlight how sen- the tail as opposed to complete, individual verte- sitive stingray topologies may be to the differen- bral centra (occurs independently for Myliobati- tial treatment of characters and/or the inclusion of dae). The Dasyatidae is supported as monophy- taxa, a re¯ection of their high levels of character letic due to the dorsal fossa on the scapular pro- con¯ict. cess (character 26), which is homoplastic (but un- Parsimony analysis of the matrix in table 7 with ambiguous). The resulting 12 minimum-length Hennig86 (using an exact algorithm, *ie) resulted trees vary only in the relationships within Dasy- in 12 equally most parsimonious trees (length ϭ atidae and whether ²Asterotrygon pairs with Uro- 55 steps, CI ϭ 0.55, RI ϭ 0.72); the strict con- lophidae or is placed in a basal polytomy with it. sensus (length ϭ 57 steps, CI ϭ 0.63, RI ϭ 0.69) Plesiobatis is nested well within the cladogram, is shown in ®gure 53. The consensus includes a as opposed to being basal to all or almost all few original clades, such as the component in- stingrays (Nishida, 1990; Lovejoy, 1996; Mc- cluding all stingrays except Hexatrygon ϩ (Gym- Eachran et al., 1996), due in part to our treatment nuridae ϩ Myliobatidae) and, somewhat supris- of characters 33 (dorsal ®n) and 13 (hyomandib- ingly, the monophyly of the Dasyatidae (i.e., with- ular-Meckelian ligament), and because it has a na- out including potamotrygonids). Both clades are sal curtain extending posteriorly to the mouth not present in the phylogenetic analyses of Nish- opening (character 44). ida (1990), Lovejoy (1996), McEachran et al. Overall, these results are signi®cantly at odds (1996), or our own analysis presented above (ma- with the phylogenetic results derived from the trix in table 5, phylogeny in ®g. 43). ²Heliobatis more comprehensive matrix presented in this pa-

TABLE 7 Matrix Summarizing Character-States Used for Phylogenetic Analysis with Myliobatidae and Gymnuridae Each Scored as Single Terminal (cf. table 5) Characters numbered as in matrix in table 5 and text. Characters that varied solely within Myliobatidae were excluded. Himantura represent Indo-west Paci®c species (Himantura sensu stricto); ``Himantura'' represent amphi-American species (H. schmardae and H. paci®ca). See text for further description and information for each character. 2004 CARVALHO ET AL.: GREEN RIVER STINGRAYS 135 per (table 5), but the resulting systematic positions RI ϭ 0.89), the strict consensus of which has one of both Green River stingrays are similar to that additional node in relation to the strict consensus analysis. Successive approximations weighting re- obtained without weighting: Pteroplatytrygon ϩ sulted in six trees (length ϭ 279 steps, CI ϭ 0.88, Himantura sensu stricto ϩ Dasyatis.

INDEX OF STINGRAY GENERA AND SPECIES (f ϭ ®gure, t ϭ table) Aetobatus, 69, 77, 79, 80, 83±86, 89±91, 93, 96, ²Heliobatis, 8, 9, 17, 24, 31, 60, 63, 64f, 65±67, 101±103, 105, 108, 114 69, 70, 72, 73, 77, 78, 80, 81, 83±85, 90, 91, A. narinari, 10, 57, 80, 91f 93, 95±104, 108±112, 117±119, 121, 122, 134 Aetomylaeus, 41, 80, 83, 85, 103, 108, 114 ²H. radians, 5±7, 9, 25, 65, 66t, 67f, 68f, 70, 70f, A. maculatus, 10, 57 71, 72 A. nichoftii, 80 Hexatrygon, 9, 20, 31, 37, 42, 48, 49, 52, 66, 77, Amphi-American Himantura, 76, 81, 85, 90, 93, 78, 80, 81, 84, 89, 90, 95, 95f, 97±102, 108, 98, 101, 103, 105±108, 112, 115, 117, 119 110, 111, 121, 134 Amphotistius, 107 H. bickelli, 9, 97 Himantura, 5, 6, 17, 20, 29, 37, 46, 48, 73, 74, ²Asterotrygon, 8, 17, 19, 20, 25, 29±31, 36, 37, 76±78, 84, 85, 90, 93, 98, 101, 103±105, 107± 41±44, 47±54, 56, 57, 59±61, 63, 64f, 65, 66, 109, 111, 113, 121, 135 67, 69, 70, 72, 73, 78, 80, 81, 83±85, 89±91, H. chaophraya, 10 93, 95±101, 103, 104, 108±111, 118±120, 122, H. gerrardi, 10, 106 134 H. granulata, 10 ²A. maloneyi, 13f, 15f, 16f, 18f, 19f, 21f±23f, 24, H. imbricata, 10, 78, 93 24t, 25, 25t, 26f±29f, 30, 32f, 33f, 44f±47f, 52f, H. kremp®, 10, 48 53f, 55f, 58f, 59f, 61, 62f, 63f, 65f, 71, 72 H. toshi, 10 Dasyatis, 5, 6, 17, 19, 29, 37, 41, 46, 47, 48, 50, H. uarnak, 10 53, 54, 56, 63, 65, 66, 72±74, 77, 78, 84±86, H. walga, 10 88, 90, 91, 93, 95, 102±109, 111±113, 135 ``Himantura'', 10, 76t, 84, 85 D. akajei, 10 ``H.'' paci®ca, 10, 76t, 81, 93, 105, 113 D. americana, 10, 49 ``H.`` schmardae, 10, 49, 76t, 82, 93, 105, 113 D. annotata, 10, 52 D. brevicaudata, 93 Manta, 20, 57, 60, 85, 102, 108 D. garouaensis, 95 Mobula, 20, 53, 56, 57, 77±81, 83±86, 89±91, 93, D. geijskesi, 10 96, 101±103, 108 D. kuhlii, 10 M. kuhlii, 10, 36, 80 D. leylandi, 10 M. japanica, 91f D. margarita, 10, 92f, 106 M. sp., 10 D. pastinaca, 10, 106 D. sabina, 10, 81, 93 Myliobatis, 40, 41, 57, 63, 69, 77, 80, 83±86, 89, D. sp., 10 90, 91, 93, 95f, 96, 101, 103, 105, 108, 114 D. thetidis, 10 M. californica, 10 D. ukpam, 10 M. freminvillii, 10 D. zugei, 10, 85, 106 M. tobijei, 91f ²``Dasyatis'' dezignoi, 11, 17 ²Palaeodasybatis, 66 ²``D.'' muricata, 11, 17, 114±116f ²P. discus, 6, 71f, 72 ²``D.'' sp., 11 Paratrygon, 17, 44, 49, 78, 81, 84, 85, 86, 90, 93, Gymnura, 20, 30, 31, 36, 49, 50, 52, 54, 56, 63, 95, 101, 103, 105, 107, 108, 121 66, 69, 76, 77, 80, 81, 84, 85, 89±91, 101, 102, P. aiereba, 10, 81 105, 106, 108 G. australis, 10 Pastinachus, 19, 76, 106, 108 G. japonica, 10, 88f P. sephen, 10, 41, 105 G. marmorata, 10 Plesiobatis, 9, 20, 31, 37, 42, 50, 52, 56, 66, 80, G. micrura, 10, 33, 38f, 39f, 74f, 79f, 82f, 86f, 81, 84, 85, 89, 90, 97±99, 103±105, 107, 108, 87f, 106 111, 120, 134 136 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 284

(Plesiobatis) daviesi, 9, 41 T. mucosa, 9, 80 Plesiotrygon, 17, 36, 41, 48, 49, 74, 78, 81, 84, T. ovalis, 9, 80 85, 86, 90, 93, 95, 102, 103, 107, 108 T. personata, 9 P. iwamae, 10, 61, 81, 88f T. testacea, 9, 48, 80, 81, 88f, 92f, 102 T. sp., 9 Potamotrygon, 17, 19, 29, 30, 33, 36, 41, 47, 48, 49, 51, 53, 54, 56, 59, 61, 74, 78, 81, 84, 85, Urobatis, 20, 31, 36, 37, 42, 50, 51, 56, 59, 72± 86, 90, 91, 93, 95, 102±104, 106, 109 74, 77, 84, 85, 90, 93, 96, 101, 102, 106, 108, P. brachyura, 10, 81 111, 121 P. constellata, 61 U. concentricus, 9 P. falkneri, 10, 61, 93 U. halleri, 9, 36, 49, 81, 86f, 93 P. henlei, 10, 81 U. jamaicensis, 9, 31, 49, 56, 77, 81, 83f, 93, 96, P. leopoldi, 10, 81, 88f, 92f 106 P. magdalenae, 10, 49 U. maculates, 10 P. motoro, 10, 61, 78f, 87f, 93 U. tumbesensis, 10 P. cf. motoro, 10, 83f Urolophus, 20, 30, 36, 41, 42, 48±50, 53, 54, 59, P. cf. ocellata, 10, 81 66, 69, 80, 83±85, 89, 90, 99f, 100, 101±103, P. orbignyi, 10, 81, 93 107, 108, 111, 121 P. cf. schroederi, 93 U. aurantiacus, 9, 81, 93 P. schuehmacheri, 61 U. cruciatus, 9, 36, 56 P. signata, 10, 81, 88f U. expansus, 9 P. sp., 10, 40f, 42f, 43f, 82f U. ¯avomosaicus, 9, 57 P. sp. nov., 10, 88f U. gigas, 9 P. yepezi, 61 U. lobatus, 9, 92f ²Promyliobatis, 9, 80, 114, 115 U. mitosis, 9 ²P. gazolae, 11, 20, 114, 115, 116f, 117f U. orarius, 9 U. paucimaculatus, 9 Pteroplatytrygon, 8, 41, 42, 77, 80, 81, 84, 85, U. sp., 9 90, 93, 104±106, 108, 113, 135 U. viridis, 9 P. violacea, 10, 41, 105 Rhinoptera, 57, 77±80, 83±86, 89±91, 93, 96, ²``Urolophus''crassicaudatus, 11 101±103, 105, 108, 114 ²``U.'' sp., 11 R. bonasus, 11 Urotrygon, 20, 29, 36, 42, 50, 51, 53, 56, 66, 69, R. javanica, 91f 77, 84, 85, 90, 96, 100, 101±103, 108, 110, Taeniura, 19, 20, 41, 46±48, 50, 69, 73, 84±86, 111, 121 88, 90, 93, 103±108, 113, 115, 121 U. aspidura, 10 T. grabata, 10, 106 U. chilensis, 10, 33, 34f, 35f, 36, 48, 56, 75f, 82f, T. lymma, 10, 36, 41, 48, 56, 75f, 79f, 81, 82f, 83f, 86f, 87f 83f, 84, 106, 107 U. microphthalmum, 10, 77, 93 T. meyeni, 60, 106 U. nana, 10 U. venezuelae, 10, 37 Trygonoptera, 20, 30, 36, 37, 42, 43, 48, 50, 53, 59, 69, 80±82, 84, 85, 89, 90, 99f, 100±103, ²Xiphotrygon, 5, 65, 72 107, 108, 111, 121 ²X. acutidens, 65, 70, 72