PaleoBios 19(2): 1-18, August 20, 1999 © 1999 University of California Museum of Paleontology New Pterodontinae (Creodonta: Hyaenodontidae) from the late Eocene-early Oligocene Jebel Qatrani Formation, Fayum province, Egypt PATRICIA A. HOI.ROYD Museum of Paleontology, University of California, Berkeley, California 94720; Phone: (303) 642-3733, Pax: (303) 642-1822, e-mail: pholroydfaucmpl .berkeley.edu Rcanaly.sis of existing pterodontine creodont collections and study of new finds from the late Eocene to early Oligocene Jebel Qatrani Formation, Fayum Province, Egypt, reveals greater diversity than previously recognized and provides the first opportunity in nearly 80 years to revise these taxa. All the taxa previously considered members of the subfamily 1 [yaenodontinae are formally recognized as belonging to the subfamily Pterodontinae. The genus Pterodon is revised and restricted to one European and three African species, one new. A new genus is erected to accommodate material previously referred to /'. leptognathus. The genus Hyacnodon is no longer recognized in the Fayum sequence. Material previously referred to that genus is transferred to Metttpterodon, and two new species are described. The presence of an additional, diminutive pterodontine and at least three medium- to large sized taxa are noted from fragmentary material. INTRODUCTION In the Paleogene of the Afro-arabian continent Creodonta were the dominant carnivorous mammals. Given the geographic isolation of Africa at this time and the absence of the Carnivora, a radiation of crcodonts is not surprising. However, few taxa have been described. Andrews (1906) described Pterodon africanus from the Fayum Province, distinguishing it from all other Pterodon de Blainvillc 1839 by its massive size, although finding it "closely similar" morphologically to the European Eocene /'. dasyuroidesde Blainvillc 1839. He also tentatively attrib- uted a single molar and an associated mandibular fragment to Hyacnodon Lazier et Parieu 1838, noting that the molar differs from that of other Hyacnodon species in its antero- posterior shortness and overall height of the posterior cusp, the protoconid. Osborn (1909) referred three new specimens of the lower dentition from Fayum Quarry A, to P. africanus, described two new species—medium-sized P. leptognathns and the larger P. phioniensis—and illustrated a still smaller pterodontine as Pterodon sp., that was not discussed in the text. Osborn confirmed the presence of an advanced hyaenodontid in the Fayum and described a new species, Hyacnodon brachycephalus, noting that it differed from typical Hyacnodon in the presence of a vestigial talonid on the third molar. Based on a well-preserved dentary, he also described a new genus and species—Mctasinopa fraasi— related to the proviverrine hyacnodontids "Sinopa" and Tritcmnodon, and considered M. fraasi was likely to be congeneric with Andrews' "Sinopa" ethiopica. To this new genus, he also attributed with some doubt a maxilla that he considered similar to Hyacnodon in the loss of M3 but differed in the presence of a vestigial protocone on Ml and M2. Schlosser (1911) attributed additional specimens from the Fayum sequence to Pterodon and Hyacnodon. These initial descriptions of Fayum crcodonts, particu- larly those of Andrews (1906), emphasized their close similarity to the crcodonts of the F.uropean Eocene. Nei- ther Andrews nor Osborn attempted species-level compari- sons of the Fayum and European crcodonts beyond noting simple diagnostic differences sufficient for identification. These studies leave an impression that the Fayum forms differ little from their F.uropean counterparts. Following these early studies, the Fayum pterodontines went unmentioned until the mid-60's when Van Valen (1965, 1966) published his wide-ranging revisions of se- lected insectivorous and carnivorous mammals and Savage (1965) revised East African carnivores. Van Valen (1965) removed from M. fraasi the maxilla that Osborn (1909) attributed with doubt to that genus and species, consider- ing it to represent a small species of Hyacnodon (Van Valen 1965, 1966). Savage (1965) referred an upper dentition from the F.ast African Miocene to Pterodon africanus. He devoted greater attention to the status of the genus Hyacnodon, particularly noting the need for revision. As a compromise between a complete revision of that geo- graphically widespread genus and simply lumping every- thing into one genus, Savage named a new subgenus, Isobyacnodon, for all African Hyacnodon, differentiating it from all other Hyacnodon by the presence of subequal paraconid and protoconid on M,. He also removed the Fayum Hyacnodon material in the British collections from Osborn\s H. brachycephalus and placed it in his new species H. andrewsi (named on material from the Miocene of East Africa). More recently, Polly (1996) demonstrated that the tra- ditional grouping I lyaenodontinae, uniting Hyacnodon and Pterodon, was not monophyletic and erected a new subfam- ily Pterodontinae Polly 1996 for Pterodon and its closest relatives. Holroyd et al. (1996) reported the first new pterodontine from the Fayum in 85 years and the first specimen from the late Eocene Qasr el Sagha Formation by referring an eroded left dentary fragment to cf. Hyacnodon brachycephalus. 2 PAI.EOBIOS, VOL. 19, NUMBER 2, 1999 Additional material collected by Duke University Pri- mate Center expeditions in conjunction with the Egyptian Geological Survey and Mining Authority have resulted in the discovery of several new taxa that show a greater diversity obtained among early African pterodontines. Afri- can Ptcrodon is revised and one new species is described. A new genus is erected for material formerly placed in Ptcrodon kptognathus. The genus Metapterodon is recog- nized for the first time in the African Paieogenc, and two new species arc recognized. Additionally, the presence of four more pterodontines is acknowledged. Unfortunately, these four taxa are known from remains too fragmentary to form the basis for new species or genera. GEOLOGIC SETTING The fossils described here come from a series of quarries within the Dir Abu I.ifa Member of the Qasr el Sagha Formation and upper and lower sequences of the Jcbcl Qatrani Formation as exposed along escarpments north of lake Birket Qarun, Fayum Province, Egypt (Fig. 1). The rocks of the 77-meter thick Dir Abu I.ifa Member of the Qasr el Sagha Formation are characterized by the presence of large or giant-scale cross-stratified sandstone and/or mudrock deposits indicative of deposition on immense side-attached bars in stream channels (Bown and Kraus 1988). The overlying, 360 meter thick Jebel Qatrani For- mation comprises two sequences: (1) a lower sequence dominated by gravelly sandstones representing coalesced channel deposits in which most of the productive fossil quarries are found and localized occurrences of variegated mudrock, in one of which Duke Primate Center Ix>cality 41 occurs; and (2) an upper sequence formed by variegated beds and a sheet gravel sequence. I Iighly productive quar- ries occur at two levels within the upper sequence: Quarries G and V within the coarse to medium sandstones; and Quarries I and M in the sheet gravel sequence (Bown and Kraus 1988). Detailed data on the stratigraphic position and lithology of these quarries is presented in Bown and Kraus (1988). Magnetostratigraphic analysis of these formations (Kappclman 1992, Kappelman et al. 1992), constrained by radiometric dates obtained from the overlying Widan el Faras Basalt, indicated that most of the vertebrate fossil localities occur in Chrons C13-C16, spanning the Eocene- Oligocene boundary. The majority of the localities of the Dir Abu Life Member of the Qasr el Sagha Formation lie within Chrons C16.1r and CI6.2, giving them the approxi- mate age of 36.15 to 36.5 Ma. Estimated ages for the principal localities in the lower sequence of the Jebel Qatrani are: less than 35.12 Ma for Quarry E; 35.40-35.56 Ma for Quarries A and B; and approximately 35.7 Ma for Locality 41. Estimated ages for the productive quarries of the upper sequence. Quarries I and M are 33.14-33.32 Ma for Quar- ries I, M, P, and J and approximately 33.37 Ma for quarries G, V, and L-12, rendering all these early Oligocene in age. The Eocene-Oligoccne boundary occurs within an unfossiliferous level but is approximately coincident with the lithologic transition between the upper and lower se- quences of the Jebel Qatrani Formation. During the time represented by the upper part of the Qasr el Sagha Formation and the overlying Jebel Qatrani Formation, the Fayum was a warm, seasonally wet tropical forest bordering freshwater swamplands as indicated by the presence of freshwater birds, fossil termitaria, accumula- tions of fossilized logs, ostracodes, megaflora similar to present-day tropical Indomalaysian forms, and paleosols suggesting seasonally wet conditions (Bown 1982, Bown et al.^1982, Bown and Kraus 1988). TERMINOLOGY AND ABBREVIATIONS Dental nomenclature used here and measurements taken arc shown in Figure 2. Tooth positions of the upper dentition are designated by upper case letters (e.g., M2 for the upper second molar) and those of the lower dentition by lower case letters (e.g., m2 for the lower second molar). Abbreviations used are as follows: AMNH, American Mu- seum of Natural History, New York; BMNH, The Natural History Museum (formerly British Museum of Natural History), Ix»ndon; CGM, Cairo Geological Museum, Cairo; DPC, Duke University Primate Center, Durham, North Carolina; KNM, Kenyan National Museum, Nairobi; and UCMP, University of California Museum of Paleontol- ogy, Berkeley, California. Locality descriptors shown in quotations are transcribed from museum specimen tags. SYSTEMATIC PALEONTOLOGY Order Creodonta Family Hyaenodontidae Leidy 1869 Subfamily Pterodontinae Polly 1996 Emended diagnosis-Hyaenodontids without meta- conid, with distinct talonid on lower molars, connate meta- cone and paracolic present on Ml -M2, a weak to absent P3 lingual cingulum, P4 lacking continuous lingual cingu- lum, relatively large anterior keels on the lower molars, m3 talonid reduced relative to that of ml-m2, and lower molar protoconids and paraconids subequal in length, circular subarcuate fossa present on petrosal, and nuchal crest not extending laterally to mastoid processes. Discussion-Polly's (1996) original diagnosis of the Pterodontinae was based almost exclusively on characters of the upper dentition and cranium, precluding assignment of taxa known exclusively from lower dentitions to the sub- family based on diagnostic criteria. The sole character of the lower dentition was loss of the metaconid, but this condition is one known to have evolved independently in other hyaenodontid lineages (Tilden et al. 1990, Polly 1996) and is therefore of little diagnostic value. The ex- panded diagnosis given above is consistent with the defini- tion of the group as hyaenodontids lacking a metaconid that are more closely related to Ptcrodon than Hyacnodon, HOLROTD-NEW PTERODONTINAE FROM EGYPT Widan el Faras El Ekhwat el Talata Tel Taleb EL E^ U Tel Akgrab Tel Homar L"41 Tel Markgraf Etnain Efreet Qasr el Sagha Temple FAYUM DEPRESSION 0 1 2 3 4 5 km N A B Elwaht Hialla Naqb el Fig. 1. A. Simplified map of northern Egypt showing major land forms, bodies of water, and position ot Fayum Depression. Topographic depressions outlined by dashed lines; water shown shaded. B. Simplified topographic map of main collecting area showing major escarpment lines and landmarks. The area between the Qasr el Sagha and El Ekwhat el Talata escarpments comprises the exposures of the lower sequence of the Jebel Qatrani Formation; that between the Jebel Qatrani and El Ekhwat el Talata escarpments comprises the upper sequence of the Jebel Qatrani Formation. Important vertebrate localities shown by boldface letters that indicate their name (e.g.. Quarry M is indicated by a boldface M). Base map modified from Bown and Kraus (1988) and after Gagnon (1993). 4 PALEOBIOS, VOL. 19, NUMBER 2, I999 length width metastylar length B trigonid width trigonid length talonid length talonid width Fig. 2. Dental nomenclature and measurements taken. A. Schematic left upper molar: 1, parastyle; 2, ectoflexus; 3, postmetacrista; 4, metastyle; 5, postprorocrista; 6, protocone; 7, trigon basin; 8, paracolic; 9, metacone. B. Schematic left lower molar: 1, paraconid; 2, postparaeristid; 3, preprotocristid; 4, postprotocristid; 5, cristid obliqua; 6, hypoconid; 7, entoeristid; 8, lingual flexid; 9, anterior keel; 10, protoconid; 11, trigonid; 12, talonid. but permits a greater range of included taxa to be identified based on diagnostic characters. Pterodon&c Blainville 1839 Synonymy— Ptcrodon Andrews 1903, p. 342, 1906, p. 219; Osborn 1909, p. 419 (part); Schlosser 1911, p. 34; Savage 1965, p. 272; Lange-Badre 1979, p. 65. Type species—Ptcrodon dasyuroides de Blainville 1839, p. 23. Included species— Ptcrodon dasyuroides de Blainville 1839, P. aj'ricanus Andrews 1903, P. phiomensis Osborn 1909, P. nyanzae Savage 1965, /'. syrtos Holroyd n.sp. Emended diagnosis—Differs from Mctaptcrodon in its larger size and in having: m3 talonid present; relatively larger ml-m2 talonids; more lingually placed and less sa- lient anterior accessor)' cuspulids on p2-p3; narrower upper molars; relatively shorter upper molar metastyles; incom- plete fusion of paracone and metacone; a single, pro- nounced upper molar ectoflexus. Differs from Akbnatcnavus Holroyd n. gen. in having: molar talonids as wide as trigonids; molar protoconid and paraconid subequal in size; anterior "bulge" on molars lacking; unreduced premolars; premolar diastcmata lacking. Differs from and Hyainolouros in smaller size and in having: relatively larger talonids and larger anterior keels on lower molars. Discussion—The genus Ptcrodon is here restricted from its traditional usage. Ptcrodon has long served as a "waste- basket taxon" for middle to late l'aleogene and Miocene pterodontines that possess neither the advanced characters of Hyacnodon (e.g., the development of elongate paracristids and loss of protocone) nor the unusual charac- teristics unique to that taxon. I have restricted the usage of this genus to species that appear to show a close relation ship to the type species, /'. dasyuroides, based on shared derived characters. PiKROixm ai-ricanus Andrews 1903 Fig. 3A Table 1 Other illustrations—Andrews 1903, fig. 3; Andrews 1906, plate XIX, fig. 3, text-figs. 69-71; Osborn 1909, fig. 9d; Schlosser taf. XIV, fig. 2. Synonymy— Pterodon africanus Andrews 1903, p. 342, 1906, p. 220; Osborn 1909, p. 419; Schlosser 1911, p. 34; Simons 1968, p. 18; non Ptcrodon african us Swage 1965, p. 272. Holotype-BMNH M8503, right dentary with p2-m3. HOLROTD-NEW PTERODONTINAE FROM EGYPT 5 1 cm D i ¦ Fig. 3. Pterodon spp. lower dentitions. A. P. phiumaisis, AMNH 13253, holotype, right p2-m3 (reversed); B. P. africanns, AMKH 14549, right e, p2-p4, m3 (reversed); C. /'. dasynroidcs(from Qucrcy, France), AMNH 11048, right p4-m3; D. /'. pbiometisis, AMNH 13254, left p2-m3. 6 PALEOBIOS, VOL. 19, NUMBER 2, 1999 Type locality— "Fluvio-marine beds (Upper Eocene): north of Birket-cl-Qurun." Referred specimens—AMNH nos. 13256, 13257, 13258 (Quarry A); 13251 ("west of Quarry A"); AMNH 13252 ("near Quarry A"); AMNH 14549, CGM 10192 (locality unknown); Stuttgart skull and associated dentary. Questionably referred specimens—UCMP 41475 (UCMPV4756). Distribution—Quarry A, lower sequence, Jebel Qatrani Fm. Revised diagnosis—Differs from all Pterodon in its much larger size and in having: relatively longer/narrower premolars; anteroposteriorly splayed lower premolars (i.e., p2 is canted anteriorly, and p4 posteriorly); strong, lin- gually-oriented postprotocristid; more pronounced lingual flexid on lower molars; relatively wider talonid. Differs from Miocene Pterodon (P. nyanzae and P. "africanus") in having: a higher P4 crown; P4 posterior accessory cusp higher relative to paracone; less bulbous P4 posterior acces- sory cusp; an unconstricted P4 talon; the posterior, rather than the anterior, root the smallest of the three P4 roots; almost complete fusion of molar paracone and metacone with only a slight sulcus present. Differs from P. plriomensis in having: pi absent or represented by only a shallow alveolus; protoconid larger and taller relative to paraconid; molar postprotocristids lingually oriented. Differs from P. dasyuroides in having: shorter and unfused symphysis; p3 with tall and well-developed posterior accessor)' cusp; stron- ger lower premolar cristids. Differs from P. syrtos in having upper molar parastyles present. Description Upper dentition: Hour specimens are attributable to P. africanus based on their large size: an isolated P4 (AMNH 13252), a damaged maxillary fragment preserving P2, P3, and the broken bases of P4-M2 (AMNH 13251), a crushed facial fragment with right P2-M1 and left P3-P4, M2 (CGM 10192), and the Stuttgart skull. These are attrib- uted to this species, rather than P. phiomensis, based on the large size of the premolars. The P2 is a long narrow tooth with slightly developed pre- and postparacristae; the crown tapers and is recurved slightly posteriorly. The P3 is more robust and bears a large basal swelling that projects lingually from the tooth, giving it a subtriangular appearance. The P4 is broken off in AMNH 13251. However, it is possible to see that P3 and P4 did not abut and that P4 was rather mesiodistally short. P4 has three roots with the lingual root larger than the mesial root, that in turn, is larger than the distal root. Partial crown morphology of P4 is preserved in AMNH 13252. The paracone is large and slightly flattened labiolingually. The posterior accessory cusp is large and quite tall. Its posterior face bears an interstitial facet where it abutts M1. There is a smaller anterior accessory cusp with a rounded wear facet exposing dentine. The lingual face of the tooth slopes steeply to a broad, broken area that possibly bore a protocone. The maxillary molars are known only from CGM 10192. The bases of these teeth and some enamel are preserved in AMNH 13251 and are sufficient to permit the following observations: Ml possessed a small, anteriorly-placed pro- tocone that overlapped with the most posterior portion of P4. The talon basin would have been small and narrow. Paracone and metacone seem to have been vcrv close together and possibly fused. The metastyle was long and at a low angle. A short parastyle was present. The P4 fit into the Ml parastyle notch. The M2 was larger than Ml and had a marked ectoflexus. The parastyle was slightly larger than on Ml. The metastyle is separated from the paraconc- metacone area by a deep notch. The metastyle was long and tall. There is no evidence of an M3. A fifth maxillary fragment, UCMP 41475, may also be referred to this species. However, as premolars (whose differing sizes are the only means currently available for proper allocation of uppers to the type lowers) are not presence!, it cannot be attributed with confidence to either species. UCMP 41475 is a left maxillary fragment with a partial M2 and the alveolus for M3. The posterior half of the fused paracone/metacone and part of the metastyle are broken. There is a slight groove on the lingual surface of the fused paracone/metacone as well as a bulge at the base of the cusp where it joins the talon. The bulge slightly overhangs the talon and a small sulcus lies beneath it. M3 is missing, but its alveolus is large and open posteriorly. A slight ridge runs from the alveolar border toward the apex of the alveolus, indicating that M3 possessed a root composed of two compressed halves. A large fossa lies lingual to the M2 metastyle and would have received the m3 protoconid. The presence of M3 in UCMP 41475 contrasts with its absence in the specimens attributed with confidence to Pterodon africanus. However, the presence of a distinct wear facet on the posterior trigonid wall of m3 in AMNH 13257 suggests that M3 was present in at least some individuals. Furthermore, the variable presence of M3 in P. rffl.$VM7wVfe(Langc-Badrc 1979) indicates that M3 presence or absence is not a reliable criterion for distinguishing between species of Pterodon. As there is no evidence of a wear facet on the posterior trigonid wall of m3 in the two known specimens of P. plriomensis (and hence no evidence of M3), it is more conservative to assign this specimen to P. africanus. Lower dentition: Andrews (1906) provided an accurate description of the lower dentition for this species, but additional observations are relevant to a consideration of interspecific variability in dental characters. The first pre- molar was only variably present. Only a small, shallow alveolus is present in AMNH 13258, and there is no evidence for pi in AMNH 13257. The second premolar is a simple, triangular tooth with a posterior face that is HOLROTD-NEW PTERODONTINAE FROM EGYPT ~ longer than the anterior. The third premolar is the longest and bears both a small anterior cuspulid and a short talonid. The p4 is the tallest, and its anterior surface consists of a long wear facet extending onto the root. The crown is tall and triangular; there is a well-developed talonid that is as wide as the protoconid. The talonid probably bore two cusps—one large central cusp and a smaller lingual cusp. There is no cingulum on either p3 or p4. The molars ate composed of an enlarged protoconid and paraconid that are, respectively, slightly labial and lingual to the midline of the tooth. A strong paracristid joins the two cusps and forms a trenchant blade on the labial surface. A notch separates the anterior (paraconid) and posterior (pro- toconid) parts of the paracristid. There is a lingual flexid and shallow groove from the paracristid notch to the base of the tooth between the protoconid and paraconid. The preparacristid runs down the mesiolingual side of the paraconid, ending approximately one-half of the way down the face of the tooth. The prominent postprotocristid passes lingually from the tip of the protoconid. A slight groove is formed anterior to the postprotocristid, which does not descend to the base of the tooth but ends just above the level of the talonid. In AMNH 13256, at least, the end of this crest is slightly rugose and could be inter- preted as a vestigial metaconid. A large wear facet is devel- oped on the posterior trigonid wall, indicating the presence of an occluding M3. All the molars retain a trenchant talonid composed of a single cusp. Discussion—Savage (1965) attributed Miocene speci- mens to /'. afrkamts as well as a related species P. nyanzae. The specimens of both P. nyanzae and /'. "afrkamts" arc consistent in size and morphology with remains of the better known Miocene genus Hyainolouros from the Mi- ocene Dera Bugti beds of Pakistan. Hyainolouros and Pterodon, along with the middle Miocene North African genus Metjistotberiurn, clearly form a closely related group of dog to bear sized carnivores. Traditionally, where we have chosen to draw distinctions between these three gen- era has been based largely on size and temporal and geo- graphic occurrence. The only comparable portion of the dentition between the Favum /'. afrkamts and the Miocene East African species of Pterodon (P. nyanzae and /'. "afrkamis") is the P4. Although AMNH 13252 and the holotype of P. nyanzae are virtually identical in length and width, the two specimens differ in several ways: the crown of AMNH 13252 is much higher, especially in the height of its poste- rior accessory cusps; the posterior accessor.' cusps are less bulbous than in /'. nyanzae; the talon of P. afrkamts is not pinched-in as it is in /'. nyanzae., and the anterior root of P. nyanzae is the smallest, rather than the posterior one as in /'. afrkamts. In comparison with Savage's Miocene Pterodon "afrkamts" AMNII 13252 is larger, higher-crowned, lacks an ectoflexus, has a larger posterior accessor)' cusp, and has a longer lingual shelf that is more posteriorly expanded. Savage (1965) differentiated both the Miocene and Paleo- gene specimens he attributed to P. afrkamts from other species in the genus based on their large size and lumped them together because he found the differences between the two trivial. He differentiated P. nyanzae from P. afrkamts on the basis of P. nyanzae1* larger size and the presence of an anterior keel on P4 and Ml. However, the type, and only specimen of P. nyanzae, is not appreciably larger than any of the Miocene specimens he attributed to P. afrkanus nor docs the anterior keel in the type of P. nyanzae differ from that in Miocene P. afrkamts. Conse- quently, I regard the Miocene P. afrkamts material (BMNH Ml9090 and KNM-CMF 4024) to be specifically distinct from the Fayum P. afrkamts and better attributed to P. nyanzae. Pterodon phiomensis Osbom 1909 Figs. 3A, D Table 1 Other illustrations-Osbom 1909 fig. 5; Szalay 1967, figs. 9a, 10a. Synonymy— Pterodon phwmensis Osbom 1909, p. 421; Simons 1968, p. 18. Holotypc-AMNH 13253, a left dentary with c-pl alveoli, p2-m3. Type locality—Quarry A, Jebel Qatrani Fm., Fayum Depression, Egypt. Referred specimens—AMNH 13254 ("lower beds"). Emended diagnosis—Differs from other Pterodon in having: greater size differential between molars; strong and distally oriented postprotocristid that is confluent with the trenchant talonid. Differs from P. afrkanus in its smaller size and in having: p2 more upright; p3 with stronger anterior accessory cuspid; relatively narrower trigonid rela- tive to its length; relatively shorter and broader premolars; molar lingual flexid less marked or absent. Differs from P. dasynroides in its larger size and in having: well-developed posterior accessory cusps on p3; shorter and unfused sym- physis, reaching only as far posteriorly as p3. Differs from P. syrtos I lolroyd n. sp. in its larger size. Description—The incisors, canines, and pi are un- known. A shallow, single pi alveolus is present in AMNH 13254. p2 is a simple, single cusped tooth, with a tall and slightly recurved crown. A preprotocristid courses down the anterolingual face of the tooth; there is no cingulum. A weak postprotocristid lies in the midline of the posterior face. Both cristids are stronger at the base and become indistinct towards the apex of the cusp. There is a small posterior basal cuspid, and the anterior accessory cuspid is variably present. The p3 is similar to p2, but more robust in size and development of crests. A large distinct posterobasal cusp is present, and the posterior crest is particularly well defined. The anterior crest is more lingually placed, and the PALEOBIOS, VOL. 19, NUMBER 2, 1999 anterior accessory cuspid is consistently present. The p4 is the tallest premolar, and its anterior crest is well-developed, curving lingually from the protocone. The p4 talonid is well developed, with both a large hypoconid and a smaller, distolingually placed entoconid. The molars are similar to P. africanus in size and general morphology, but do differ in several ways: the trigonid is slightly narrower relative to its length; the preparacristid does not occur near the tip of the paracristid but only on the lower half of the crown; the postprotocristid runs distally (rather than lingually) and meets the trenchant talonid crest; the lingual flexid is less well-developed; the talonid (rather than the trigonid) is the widest part of m2. Pterodon srivros Holroyd, new species Fig. 4 Holotype—CGM FN 84-244, right maxillary fragment containing P3-M2, roots of P2. Type locality—Quarry M, upper sequence, Jebel Qatrani Formation, Fayum Depression, Egypt. Etymology—syrtos, Greek, meaning "swept or carried along by a stream"; in reference to the deposition of this specimen along the point-bar deposits of Quarry M. Diagnosis—Differs from other species of Pterodon in its smaller size and in lacking molar parastyles. Differs from P. pbiomensis in having: weak molar cctoflcxus; relatively smaller protocone; a straight (non-bulging) lingual face on the fused paracone-metaconc; more anteriorly placed and smaller protocone. Differs from P. dasyuroides in having: more anteriorly-placed protocone; larger P4 protocone; P3 with a parastvle. Description—CGM FN 84-244 is a right maxillary frag- ment with roots of P2 and P3-M2. The P2 was a small, double-rooted tooth. The P3 is three-rooted and has a small parastvle. The distolingual portion of the P3 is dam- aged, and a broken segment has migrated anteriorly and lingually. The P4 is dominated by a large paracolic that is somewhat labiolingually compressed and whose lingual sur- face is fractured by a midline crack. The P4 posterior accessory cusp is moderate in size, and the centrally posi tioned P4 protocone equals that of the molars in size. The molars are similar to those of P. pbiomensis and /'. africanus but lack a parastvle and have a less pronounced cctoflcxus. Pterodon sp. DPC 5036, a broken right canine from Quarry E, pos- sesses an appropriate morphology to be referred to Pterodon. It is smaller than P. africanus and could possibly pertain to /'. pbiomensis or a similar-sized form. Akiinatenavus Hoi.royi), new genus Synonymy— Pterodon Osborn 1909, p. 419 (in part). Type species—A. leptojjnathus (Osborn 1909). Fig. 4. Pterodon syrtos, type and only specimen, right P4-M2 in occlusal and lingual views. Scale bar 1 cm. HOI.ROTD-NPW PTERODONTINAE FROM EGYFI 9 Etymology— For the Egyptian pharaoh, Akhnaten; name coined by Thomas M. Bovvn. Diagnosis—Differs from Ptcrodon and Mctapterodon in its smaller size and in having: more gracilc and relatively smaller premolars; presence of diastemata between pi and p2, p2 and p3; smaller and shorter talonids on ml and m2; anterior cuspulids lacking on p2-p4; narrower molars; no lingual bulge on p3; slender canines; paraconid smaller relative to the protoconid. Differs from Mctapterodon in retaining a m3 talonid. A. leptognathvs (Osbom 1909) Holroyd, n. comb. Fig. 5 Table 1 Other illustrations-Osborn 1909, figs. 4a,b; 9c. Synonymy— Ptcrodon leptojjnatbusOsborn 1909, p. 419; Simons 1968, p. 18. Holotype—AMNH 13263, a right dentary with p3-m3 and alveoli for c, pi and p2. Type locality—Quany A, lower sequence, Jebel Qatrani Formation, Fayum Depression, Egypt. Emended diagnosis—As for genus. Description Lower dentition: The canine alveolus is broken medi- Fig. 5. Akhnatenavusleptognathus, AMNH 13263, holotype, left p3 ni3, in lateral and lingual views. ally, is mediolaterally narrow, and does not extend beyond pi. Based on the size of the alveolus, the canine was not as large or stout as in Ptcrodon. A very short diastema sepa- rates the canine from pi. The pi has a single root and the alveolus is small. A slightly longer diastema separates pi and p2. The p2 was two rooted and approximately equal in size to p3 based on alveolar sizes. A long diastema separates p2 from p3, and this area also shows alveolar bone loss as indicated by a depression between the teeth. The p3 is small, low-crowned, with a single central cusp and two roots. The posterior root is the larger. There is only a single cusp located over the center of the two roots. The pre- and postprotocristids are very weak. The distal margin of the tooth is slightly wider than the mesial margin. The fourth premolar is approximately equal to m2 in height. Like p3 it is narrow, lacks marked accessory cusps, and has only weak crests. The p4 crown is steep-sided and bears a small anterior "extension''' and very small talonid. The ml trigonid is heavily worn. The small, short tal- onid bears a lingual wear facet for the protocone. There is a distinct wear facet where the trigonid and talonid meet on the labial side. This facet is also present on m2. On m2 the protoconid is broken. However, based on the size of its preserved portion, it would have been approximately twice the size of the paraconid. The talonid is short and tren- chant. The blade is slightly labiad of center and drops oil sharply labiad with a small wear facet as on m 1. The lingual side of the talonid is much Hatter. An anterior accessory crest is present but not high, only approximately one-third the height of the tooth. 1'he third molar is slightly longer and taller than m2. The talonid is relatively smaller and has two cuspulids. The protoconid is taller than the paraconid and labiad of the midline. The paraconid is lingual. An anterior keel extends apically from the base along the lower third of the tooth. The postprotocristid is not well-developed. On both m.2 and m3 a small weak crest on the paraconid's mesiolingual face extends approximately halfway up the cusp from the base of tooth. Dentary: The horizontal ramus is shallow, and the symphysis extends to beneath p3. There is a small mandibu- lar foramen low on the ascending ramus. There are three mental foramina: beneath pi, at the posterior limit of the p2/p3 diastema and beneath p3/p4. Akhnatenavus aff. A. leptojjnatbus Table 1 DPC 7765, a right dentary from Locality 41(L-41), closely resembles that of A leptognathus. It differs from the type of A leptognathus primarily in having a narrower p4 with smaller anterior shelf; somewhat more reduced p2, a vestigial m3 talonid. These differences may be sufficient to recognize this specimen as a species distinct from A Icptofjnathus. However, premolar size and minor details of morphology are somewhat variable within pterodontine Table 1. Measurements in mm (rounded to nearest tenth) of pterodontine lower dentitions. 1 = length; w = width; Tal = talonid; Tri = trigonid. specimen p2 1 p2 w P3 1 p3 w p41 p4 w ml 1 ml w ml TalL ml TriL m2l m2 w ii-2 TalL m 2 TriL m3l m 3 w m3 TriL m3 TalL Pterodon phiomensis AMNH 13254 13.0 AMNH 13253 14.0 8.0 8.7 14.0 14.0 9.5 9.2 18.0 17.0 11.0 17.0 14.0 8.2 12.0 9.7 3.4 3.5 26.0 24.0 13.0 12.0 20.0 18.0 5.4 5.6 32.0 28.0 16.0 14.0 28.0 22.4 4.6 3.6 Pterodon africanus AMNH 14549 19.0 AMNH 13258 AMNH 13256 10.0 21.0 11.0 23.0 14.0 31.0 31.0 29.0 21.0 16.0 26.5 25.4 4.0 5.0 Indeterminate pterodontincs, close to Pterodon DPC 6555 20.0 12.0 DPC 10315 25.0 19.8 5.7 Metaptcrodon bmcbycephalus AMNH 13264 7.6 6.9 19.0 12.0 17.9 1.5 Metaptcrodon schlosseri AMNH 13262 6.4 4.3 9.0 5.1 10.0 9.2 5.3 13.0 7.3 CGM 88123 BMNH 15048 DPC 4877 7.7 4.4 DPC 10793 DPC 5664 Akhnatenavus leptognathns AMNH 13263 DPC 7765 7.4 3.1 13.0 7.0 13.0 0 8.9 4.1 12.0 6.0 15.0 7.5 15.4 0 10.0 6.(1 11.0 5.0 ".6 3.1 13.0 7.1 9.7 3.2 13.0 7.8 12.6 0 7.0 5.4 7.0 5.3 8.0 5.5 8.2 4.4 6.6 1.6 13.0 6.6 11.0 2.2 14.0 7.7 14.0 0 9.0 3.7 12.0 5.7 8.4 2.9 14.0 7.2 12.0 2.7 18.0 9.5 15.9 2.2 10 3.9 11.0 5.8 13.0 3.4 8.1 13.0 9.3 15.7 HOLROTD-NEW FPERODONTINAE FROM EGTP'I 11 species. Furthermore, nothing is known of population varia- tion in the L-41 or Quarry A samples, so it is not possible at this time to determine if these are distinct taxa or if the differences between the two specimens is simply an expres- sion of individual variation within a single lineage. Conse- quently, an assignment to A. aff. A. leptojjnathus is provisional until more is known of this genus in the lower sequence of the Jebel Qatrani Formation. MiiTAWHRonoN Stromer 1926 Synonymy—Hyaenodon Andrews 1906, p. 218; Osborn 1909, p. 423; Savage 1965, p. 268 (part); Hyaenodon (Isohyaenodon) Savage 1965, p. 280 (part); Pterodon Osborn 1909, p. 419 (part); Metapterodon Stromer 1926, p. 110; Savage 1965, p. 268 (part); Metasinopa (?) Osborn 1909, p. 423; Savage 1965, p. 264. Type species— Metapterodon kaiseri Stromer 1926. Included species -M. kaiseri, M. zadoki Savage 1965, M. markjjrafi I lolroyd n. sp.; M. schlosseri Holroyd n. sp.; M. brachycephalus (Osborn 1909). Revised diagnosis— Differs from Hyaenodon, Pterodon, and Akhnatcnavus in having: P4 postparacrista developed into a shearing crest; narrower lower molars; narrower upper molar metastyles; undulant labial face on upper molars; p2 with anterior accessory cuspid; blade-like P4 metastyle; double-rooted P3. Differs from Pterodon in smaller size and in having: longer P4 metastyles; smaller protocones; narrower upper molars; more labially placed and more salient anterior accessory cuspids on p2-p3; pi lacking; fully compressed fused paracone/metacone with- out trace of groove; more sharply angled posterior trigonid wall; much shorter ml-m2 talonids; m3 talonid lacking. Differs from Hyaenodon in having: upper molar protocones primitively present; lower molar posterior paracristid not markedly longer than anterior paracristid. Discussion —The species here united under Metapterodon have been some of the most vexing for students of African creodonts, with individual specimens and/or species having been assigned both to proviverrine genera and other pterodontinc genera at one time or another. Osborn (1909) and later Van Valen (1967) con- fused some AMNH specimens with the proviverrine Metasinopa. Metapterodon has not always been recognized as distinct from other pterodontines. Van Valen (1967), followed by Savage (1978), considered the types species of Metapterodon (M. kaiseri) and Pterodon (P. dasyuroides) to be closely related and recognized only Pterodon. Metapterodon has also frequently been confused with Hyaenodon. Osborn (1909) described the species H. brachycephalus from the Fayum and assigned it to Hyaenodon because it lacked a talonid on the third molar. Savage (1965) described H. (Isohyaenodon) andrewsi as the type of the new subgenus, based on five mandibular speci- mens: four from sites in the Kavirondo Gulf, Kenya and one from the Fayum. As discussed below, the Fayum specimen is attributable to Metapterodon and, based on size and morphology of the Kenyan specimens, aH." andrewsi probably represents the previously unknown lower denti- tions of M. kaiseri or a closely related species. It is probable that the Miocene and Eocene to Oligocene forms arc generically distinct. However, due to problems in the attri- bution of certain specimens included in "//." andrewsi (see Savage 1965), there is not adequate material to differentiate these species at the generic level. Mellert (1977) regarded all the African material placed in Hyaenodon by Savage (1965) to be better allied with Pterodon, and that they might represent a subgenus of Pterodon. He observed that the obliquity of the main shearing surfaces was like that in Pterodon and in contrast to the almost longitudinally parallel surface in Hyaenodon. 1 lowevcr, based on the characters enumerated above, these species can also be excluded from Pterodon. Most recently, Polly (1996:316) mentioned that Metapterodon (as well as several other genera lacking a metaconid) might be assign- able to the subfamily Pterodontinae, but lacked "referred material allowing placement." The expanded diagnosis of Pterodontinae presented above permits this genus to be included within the subfamily. Although the Eocene-Oligocene species referred here to Metapterodon may ultimately prove to be a seperate genus from M. kaiseri, I have here united them in an attempt to demonstrate their distinctiveness from both Hyaenodon and Pterodon, and their probable closer relationship to one another than to contemporaneous species of Pterodon. M. MARKGRAFi Holroyd, new species Fig. 8C Other illustrations—Osborn 1909, fig. 7. Synonymy—^Metasinopa Osborn 1909, p. 423; Metasinopa (?) sp. Savage 1965, p. 264. Type and only specimen: AMNH 14452, left maxilla containing alveoli forC-Pl, roots of P2-P3, partial P4, Ml, and partial M2. Type locality—unknown. Etymology—for Richard Markgraf, collector of the type and in recognition of his many years of collecting in the Fayum region. Diagnosis—Differs from both M. schlosseri Holroyd n. sp. and M. kaiseri in having: greater reduction of upper molar protocones; anreroexternal cingula on P4-M2, shorter and broader P2; single-rooted PI. Differs from M. schlosseri Holroyd n. sp. in having: parastyle present on P4 but lacking on molars, relatively longer metastyle. Differs from M. brachycephalus in smaller size. Description Upper dentition: AMNH 14452 is a left maxillary fragment preserving the palate from the incisor alveoli to M2. The facial region is short, and diastemata are absent. Three small grooves indicate the presence of incisor alveoli. The canine alveolus is largely broken. The PI was single- 12 PALEOBIOS, VOL. 19, NUMBER 2, 1999 Fig. 6. Metapterodon spp. A. M. schlosseri, AMNII 13262, left p2- m3, lateral view. B. M. brachycephalus, AMNH 13264, holotypc, left e, p2, p4, m2-m3, medial view. C. M. schlosseri, DPC 4877, left p2-m3, lateral view. rooted and lay completely lingual to the posterior part of the canine. The P2-P3 were short, double-rooted teeth. The P4 is a triangular tooth bearing a small parastyle. The posterior paracrista is well-developed and with the metastyle, now broken, would have formed a shearing blade as on the molars. The molars are highly developed for shearing with the protocone absent and the fused paracone-metacone and metastyle aligned to form a single shearing surface. There is no hint of a paracone-metacone separation. The fused paracone-metacone is compressed labiolingually into a tri- angular shape whose posterolingual surface is continuous with the metastyle. The metastyle is long, greater in length than the fused paracone/metacone, and narrow. No proto- cone is developed, but a remnant cuspule is present in this position on M1 and slight eristae or cingula are present in the position of the pre- and postparacristae. A slight ante- rior cingulum is present from the protocone remnant to the anterolabial surface of the fused paracone/metacone. A 1 cm I CHI Fig. 7. Metapterodon schlosseri, DPC 5664, holotype, left p3-m3- very weak labial cingulum is present between the fused paracone/metacone and the metastyle. The labial surface of the tooth is undulant, with two labial flexures. Measurements in mm for AMNH 14452 are: P4 width- 6.4; Ml length-12.3, width-6.4, metastylar length-7.2; M2 width-7.8. M. schlosseri Holroyd, new species Figs. 6A, C, 7, 8A, B Tabic 1 Other illustrations-Osborn 1909, fig. 9D. Synonymy— Hyacnodon sp. Andrews 1906, p. 20; Pterodon sp. Osborn 1909, p. 422; Hyacnodon (Isohyaenodon) andrewsi Savage 1965, p. 281, 1978, p. 252." Holotype—DPC 4877, left dentary containing canine alveolus, p2-m3. HOLROYD-NEW PTERODONTINAE FROM EGYPT 13 -ft. " B 1 cm 1 cm 0— Fig. 8. Metapterodon spp. A. M. schlosseri, DPC 10791, right lower canine, lateral view. 15. M. schlosseri, DPC 5431, left P2, P4-M2, occlusal view. C. M. markgrafi, AMNH 14452, left P4-M2. 14 PALEOHIOS, VOL. 19, NUMBER 2, 1999 Etymology—In honor of Max Schlosscr, in recognition of his contributions to the early descriptions of Fayum mammals. Type locality—Quarry V, upper sequence of Jebel Qatrani Formation, Fayum Province, Egypt. Distribution—Type locality and, tentatively. Quarry B, lower sequence of Jebel Qatrani Formation, Fayum Prov- ince, Egypt. Late Eocene to early Oligocene. Hypodigm-DPC nos. 5431, 5664, 10791, 10793 (Quarry V). Tentatively referred specimens—AM NH 13262 (Quarry B), BMNH C8812-13 (lower sequence of the Jebel Qatrani Formation?). Diagnosis—Differs from M. tnarkgmfi and M. zadoki in having: more salient upper molar protocone; P4 parastyle lacking; M1-M2 parastyle present. Differs from M. zadoki in larger size. Differs from M. bracbycephalus'm smaller size. Description Upper dentition: DPC 5431 is the only maxilla referrable to this taxon. It is a left maxillary fragment retaining the alveoli of the canine, P3 and M3, a single root of PI, a broken P2, and P4-M2. The entire tooth row has been warped by depositional processes into a labially-arcing semi-circle that has distorted the relationships among teeth. However, most morphological aspects of the teeth them- selves are undistorted. The canine alveolus is large, and no diastema is present between it and the PI alveolus. The PI has a single root, but is relatively larger than in M. markjjrafi. The P2 is a simple, oval, double-rooted tooth with a small posterior accessory cuspule. The P3 was a longer tooth, approxi- mately equal in length to P4, based on the size of its alveoli. The P4 is a triangular tooth with a well-developed protocone and short metastyle separated from the paracolic by a notch. A parastyle is absent, but a slight cingulum is present in its position. The protocone is too damaged to determine its size and the presence or absence of cristae. The tip of the paracolic is broken, but a weak preparacrista is present below the break. The postparacrista and metastyle are only moderately developed. The molars are composed of a small protocone, a large fused paracone-metacone, and an elongate metastyle. A very small parastyle is present on Ml (this region is missing on M2). On M2, the labial cusp and trigon basin are separated by a short groove at the base of the tooth. No labial cingulum is present. The orientation of the proto- cone relative to the rest of tooth cannot be determined at this time; on both molars, damage along the midline pre- cludes accurate reconstruction. Measurements for DPC 5431 are: P2 length-8.7, width- 4.2; P4 length-11.7 (approx.), width-9.2 (approx.); Ml length-12.5, width-8.9 (approx.), metastylar lcngth-6.3; M2 length-13.3; width-10.5 (approx.); metastylar length- 8.2. Lower dentition: The lower dentition is represented by the type, DPC nos. 5664, 10793, and tentatively 10791, a right canine. AMNH 13262 from Quarry B and BMNH C8812-13 (probably from the lower sequence of the Jebel Qatrani Formation) have also been tentatively assigned to this species based on their size, but may represent M. markjjrafi or another species of this genus. Unfortunately, neither specimen is sufficiently preserved to assign them with confidence to any particular species. Description of the lower dentition of this species is based solely on the specimens from Quarry V. DPC 10791, a pterodontine right canine, is of an appro- priate size to fit in the alveolus preserved in the holotype left dentary. It is a stout, slightly-recurved, conical tooth whose crown is splayed laterally from its root. The crown is worn both on the distal surface of the tip and on its lingual surface. The root is large and has a shallow groove coursing down from the crown along the middle of its lingual surface. The first premolar is not present, and p2 is an ovoid, single-cusped tooth whose postparacristid is much longer than the preparacristid. A very small cuspulid lies on its anteroiingual face. The p3 is similar to p2 in construction, but is anteroposteriorly shorter and has a small, well developed talonid and faint lingual cingulum. The p4 is a taller tooth with a talonid composed of a high hypoconid and a much smaller, vestigial hypoconulid lingual to and separated from it by a faint groove. A slight lingual cingu- lum is present. The p4 is shorter than ml, a condition in contrast to that in other pterodontines in which p4 is always longer than m 1. The molars are narrow teeth composed of enlarged paraconid and protoconid and a variably developed tren- chant talonid. There is no metaconid. The protoconid is the tallest of the cusps, and the paracristid notch is narrow and deep. The paraconid lies anteroiingual to the proto- conid, but the difference in location is less than in Pterodon. The preparacrista is absent, and the anterior keel is well developed on the anterolabial face of the paraconid. The postprotocrista is strong, directed distally, and meets the cristid obliqua at the base of the protoconid. In the type m.3, however, this crest has a distolingual course and does not meet the cristid obliqua. The talonid is absent on m3, but present on both ml and m2. On m3 only a vestigial cuspid remains. Based upon the two specimens currently available, the length of the talonid varies considerably between individuals. In the type, the ml and m2 talonids are short but wide, and m 1 bears a hypoconid, vestigial hypoconulid, and slight entocristid. In DPC 5664, the talonids of ml and m2 are much reduced in length and width and lack entocristids and hypoconulids. M. brachtcephaws Osborn 1909 Fig. 6B Table 1 Other illustrations Osbom 1909, figs. 8, 9g. Synonymy— Hyaenodon bracbyccphalus Osborn 1909; Simons 1968, p. 18. HOI.ROTD-NEW PTERODONTINAE FROM EGYPT IS Fig. 9. Indeterminate pterodontines, new unnamed genera and species. Pterodontine (', new genus and species indeterminate. A. DPC 9243, right M. fragment, lingual (above) and labial (below) views. 15. DPC 10315, right M,fragment, lingual (above) and labial (below) views. Pterodontinae D, new genus and species indetermi- nate. C. DPC 6545, right mandibular fragment in labial (left) and lingual (right) views. Holotype—AMNH 13264, left dentary with broken canine, p2, p3 alveolus, broken and worn p4, ml alveolus, m2 m3. Type locality—The specimen label reads "lower beds, west of Quarry A." This locality presumably lies within the lower sequence of the Jebel Qatrani Formation. Diagnosis—Differs from other species of Mctapterodon in its larger si/e and in having: protoconid and paraconid on m3 equal in size and height; m2 talonid reduced in size; two enspulids present on vestigial m3 talonid. Mctapterodon brachycephalusdiffers from the Miocene "H. (Isohyacnodon) andrcwsF (probably the same as M. zadoki) and in its much larger size and in having only a moderately developed, rather than large, anterior lingual cingulum crest on the face of paraconid. Description—The dentary is anteroposteriorly short given the size of the third molar. There are two large foramina present, one each beneath p2 and p4. The stout canine is round in cross-section. A large anteromedial wear facet on the canine, which must have occluded with 13. The canine is very near the midline, indicating that the lower incisors were small or reduced in number. The second premolar is simple and worn both anteriorly and posteri- orly. Neither p3 nor ml are preserved., and p4 and m2 are heavily worn. However, on m2 a very small talonid can be discerned. The third molar is the best preserved tooth. It is dominated by the enlarged and subequal paraconid and protoconid. A vestigial talonid bearing two small cuspids remains. Measurements in mm: Canine length-10.3, width-14.1; p2 length-7.6, width-6.9; ml alveolus-11.9; m2 talonid length-3.6; m3 length-19.1, width-12.0, paraconid length- 10.06, protoconid length-11.63, talonid length-1.5. Last two molars length-33.75; premolar series length 32.5; ca- nine and premolars length-54.4; length of last three alveoli- 45.5; length of first four alvcoli-45.5; length of symphysis parallel to tooth row-40.9; mandibular depth at m3-34.1; c-m3-97.9; p2-m3-76.5. Osborn's (1909) table of mea- surements is incorrect for the "space occupied by three inferior molars." His measurement is actually for the last two molars only. Pterodontinae A DPC 6555, from Quarry I, is a large right p2 or p3, measuring 19.5 mm in length and 11.7 mm in width. It is a simple, robust, oval tooth with crenulate enamel. The pre- and postcristids are well-developed, and the precristid is lingually deflected. It is morphologically similar to but slightly wider than the largest specimens of Ptcrodon africanus. DPC 8345, the tip of a left canine from Quarry M, may represent the same taxon as DPC] 6555. Like other pterodontine canines, it is conical, slightly recurved, and shows a strong internal ridge. Pterodontinae B DPC" 2541 is a very large left canine fragment from Locality 30. Only the root and the base of the crown are preserved. In size and morphology, it is most similar to Ptcrodon, but is larger than P. africanus. Ptkrooontinak C, unnamed new genus and species DPC 9243 and DPC] 10315 from Quarry V (Figs. 9A, B) represent a different pterodontine taxon, morphologi- cally near Pterodon or Akhnatenavus. DPC 9243 is a partial right lower molar, probably ml, and DPC" 10315 is a partial right m3. These teeth are larger than contemporane- ous Mctapterodon and smaller than Ptcrodon from lower in the sequence. This taxon differs morphologically from Akhnatenavus, Pterodon, and Mctapterodon in having a much shorter and smaller paraconid relative to the proto- conid, a strong preparacristid running down the anterior (rather than anterolateral) face of the paraconid, and a strong anterior protrusion overhanging the root. There 16 PALEOBIOS, VOL. 19, NUMBER 2, 1999 Widan El Faras Basalt Pterodon syrtos, pterodontine A Metapterodon schlosseri, pterodontine C Pterodon sp. en A 5 D Pterodon africanus, P. phiomensis, 0)1 oU fTl A & D Akhnatenavus leptognathus, Metapterodon schlosseri ^h 46m L-41 Akhnatenavus aff. leptognathus, pterodontine D Metapterodon sp. Fig. 10. Diagram showing the stratigraphic distribution and patterns of co-occurrence of pterodontines in the Fayum sequence. The position of Metapterodon markgrafi'is unknown; M. bracbyccpbalus, from west of Quarry A, may come from approximately the same stratigraphic level. Stratigraphic data for localities from Bown and Kraus (1988). fore, it probably represents a new genus and species of pterodontine. Pi•krooontinak D, unnamed new genus and species DPC 6545 (Fig. 9C) is a small right dentary from Locality 41 bearing broken p4 and m3, broken roots of ml-m2 and the alveoli of p3. This specimen appears to represent a diminutive pterodontine. The p3 alveolus indicates that this was a double-rooted tooth, similar in length to p4. The p4 is broken but sufficiently preserved to show it was a narrow, oval tooth with a strong postprotocristid, a fairly high trenchant tal- onid, and an anterior accessory cusp. A slight lingual cingulid extends from the base of the talonid to the middle of the protoconid. The alveolus for ml indicates that this was a slightly shorter tooth than p4 and much smaller than m2 or m3. The second and third molars, based upon alveolar dimensions, were subequal in size as in pterodontines and unlike any F.uropcan or Fayum proviverrines in which m2 is always smaller than m.3. The trigonid of m3 is considerably damaged with only the lower portion of the tooth remaining. The protoconid is centrally placed and the paraconid is large, nearly equal in size to the protoconid. The paracristid plane is obliquely oriented. Among Fayum creodonts, similar orientation of the paracristid is only seen in Mctasinopa (an advanced proviverrine) and pterodontines. The protoconid and paraconid are separated lingually by a slight sulcus that does not extend to the base of the tooth. A strong anterior keel is present. Like pterodontines and unlike Fayum proviverrines, the m3 has a reduced talonid with indistinct cusps and no entoeristid. The tooth is also narrower relative to the trigonid length than in proviverrines and is more similar to pterodontines in this respect. Further, the broken crown shows an oval root canal for the protoconid near the midline of the tooth. In proviverrines this canal is typically circular and lies well labial to the midline. The root canal typically occurs directly beneath the apex of the cusp. This feature suggests that no metaconid was present, and the HOI.ROTD-NEW FEERODONTINAE FROM EGTPl 17 protoconid was centrally placed. The mandibular ramus is rather shallow. Three small mental foramina are present, one each beneath p3-ml. The posterior portion of the symphysis is present and extends posteriorly, ending be- neath the posterior root of p3. Its much smaller size, coupled with the relatively large anterior accessory cusp on p4, indicates that is a new genus and species distinct from oilier Fayum pterodontines. DISCUSSION New discoveries and reanalysis of older collections have revealed a more diverse pterodontine fauna in the late Eocene and early Oligocene of Africa than was previously recognized. Minimally, three genera and six species are now known, and another four species and possibly two genera are likely present. Two of these genera, Pterodon and Mctaptcrodon, span the Eocene-Oligocene boundary; only Akimatcnavus is known exclusively from the late Eocene. The stratigraphic distribution of pterodontines in the Fayum sequence is shown in Fig. 10. Discerning the relationships among the early African pterodontines, their Paleogene counterparts in Europe and Asia, or their possible descendents in the Neogene is prob- lematic. Among the African pterodontines only Pterodon africanus is adequately represented by dental and cranial remains, and the small sample sizes for all taxa permit no grasp of intraspecific variation. Further, the reduced nature of the dental morphology (known to be convergent on that of hyaenodontines, see Polly 1996) make character-based analyses fraught with possible sources of interpretational error. Holroyd (1995) attempted such an analysis of pterodontine relationships based on dental characters and confirmed the monophyly of Paleogene pterodontines com- pared to selected European and Asian proviverrines and hyaenodontines (sensu stricto). However, branching pat- terns among Pterodontinae were unstable due to the large amount of missing data. At this time, the best conjecture for the phyletic rela- tionships among Paleogene pterodontines is that African pterodontines form a clade that also includes European Pterodon dasyuroides. This clade could be united by a weak to absent P3 lingual cingulum, relatively large anterior keels on the lower molars, m3 talonid reduction, and lower molar protoconids and paraconids subequal in length. The taxa comprising this group, and the characters that unite them, correspond to the Pterodon grade noted by Mellett (1977). Larger Neogene pterodontines—Meqistotherium, Hyainohmros, P. nyanzae— may be most closely related to the Fayum Pterodon species based on general morphology. Smaller Neogene pterodontines may be more closely re- lated to Mctaptcrodon or the diminutive pterodontine from Locality 41. However, the morphological similarities ob- served among these taxa could also reflect convergences due to dental allometry as pterodontines exploit different size niches. Also, the absence of a recent comprehensive review of African Neogene creodonts has left a very poor understanding of the alpha-level taxonomy of these forms, complicating any comparisons. Only much better material will permit relationships among the diverse African pterodontines to be clarified. However, it is clear that Africa hosted a tremendous diversification in this group, and greater attention should be directed toward expanding the geographic and temporal sampling of this continent's still poorly-known Paleogene record. ACKNOWLE DGMENTS I am deeply grateful to E.L. Simons for the opportunity to study the Fayum creodonts, P.S. Chatrath for help in the lab and field, and to the members of my committee (T.iYL Bown, R.F. Kay, M.C. Maas, E.L. Simons, and F.J. White) for their insights while preparing this work as part of my doctoral dissertation at Duke University. J.H. Hutchison (UCMP), J. Alexander, M.C. McKenna and R.H. Tedford (AMNH), and A. Gentry (The Natural History Museum, formerly BMNH) kindly provided access to and assistance with the collections in their care. This work benefitted from many discussions with T.M. Bown, M. Gagnon, and P.D. Polly, and insightful and constructive reviews by H. Bryant, G. de luliis, and P.D. Polly. K.C. McKinney provided invaluable darkroom help. Travel funds were provided by a grant from the national Sigma Xi organization and from the Sally Hughes-Schrader Travel Award of the Duke Univer- sity chapter of Sigma Xi. This is UCMP publication no. 1695 and DPC publication no. 685. LITERATURE CITED Andrews, C. W. 1903. Notes on an expedition to the Fayum, Egypt, with descriptions of some new mammals. Geological Magazine 4:337-43. Andrews, C. W. 1904a. Further notes on the mammals of the Eocene of Egypt. Geological Magazine, Decade V 1:109-15 and 157-62. Andrews, C. W. 1906. Catalogue of the Tertiary Vertebrata of the Fayum, F.gypt. Taylor and Francis, London. 324 pp. Bown, T. M. 1982. Ichnofossils and rhizoliths of the nearshorc fluvial Jebel Qatrani Formation (Oligocene), Fayum Province, Egypt. Palacoijeojjrapby, Palacoclimatolojiy, Palaeoccology 40:255-309. Bown, T. M., .VI. J. Kraus, S. L. Wing, J. G. Fleagle, B. II. Tiffhey, E. L. Simons, and C. F. Vondra. 1982. The Fayum primate forest revisited. Journal of Human Evolution 11:603- 32. Bown, T. M., and M. J. Kraus. 1988. Geology and paleoenvironment of the Oligocene Jebel Qatrani Formation and adjacent rocks, Fayum Depression, Egypt. U.S. Geologi- cal Survey Professional Paper 1452, United States Govern- ment Printing Office, Washington. 64 pp. Gagnon, M. 1993. Mammalian paleoecology of the Fayum of Egypt. Ph.D. diss. Duke University, Durham, NO. Holroyd, P. A. 1995. An examination of dispersal origins for 18 PALEOBIOS, VOL. 19, NUMBER 2, 1999 Fayum Mammalia. University Microforms International, Ann Arbor, Michigan, p. i-ix, 1-328. Holroyd, P. A., E. L. Simons, T. M. Bown, P. D. P. Polly, and M. J. Kraus. 1996. New records of terrestrial mammals from the upper Kocene Qasr el Sagha Formation, Fayum Depres- sion, Egypt. Palacovertcbrata, Vol. jtibil. D. P.. Russell 25(2- 4): 175-92. Kappelman, J. 1992. The age of the Fayum primates as deter- mined by palcomagnetic reversal stratigraphy. Journal of Hu- man Evolution 22(6):495-503. Kappelman, J., F. I,. Simons, and C. C. Swisher. 1992. New age determinations for the Eocene-Oligocene sediments in the Fayum Depression, Northern Egypt. Journal of Gcolo/jy 100(6):647-54. Lange-Badre, B. 1979. Les creodontcs (Mammalia) d'Europe occidentals' de l'Eoccne supcricur a l'Oligoccnc supcricur. Mcmoires du Museum National d'Histoirc Naturcllc, nouvelle serie, Serie. C, t. 42:1-249. Mellett, J. S. 1977. Paleobiology of North American Hyaenodon (Mammalia, Crcodonta). Contributions in Vertebrate Evolu- tion 1:1-133. Osborn, H. F. 1909. New carnivorous mammals from the Fayum Oligocene, Egypt. Bulletin of the American Museum of Natu- ral History 26:415-24. Polly, P. D. 1996. The skeleton of Gazinocyon vulpcculus gen. et comb. nov. and the cladistic relationships of Hyaenodontidae (Eiuhcria, Mammalia). Journal of Vertebrate Paleontology L6(2):303-19. Savage, R. J. G. 1965. Fossil Mammals of Africa: 19. The Miocene Carnivora of East Africa, Bulletin of the British Mu- seum (Natural History), Geology 10:242-316. Savage, R. I. G. 1978. Carnivora. pp. 299-330 in V. J. Maglio and II. 15. S. Cooke (cds.). Evolution of African Mammals. Harvard University Press, Cambridge, MA. Schlosser, M. 1910. Uber einige fossil Saugetierc aus dem Oligocan von Agyptcn. Zoologische Anzcigcr 35:500 508. Schlosser, M. 1911. Bcitragc zur Kenntnis der Oligozanen Landsaugetiere dem Fayum (Agyptcn). Bcitragc zur Paldontologie und Geologic von Ostcrrcicb-Uunarns and des Orients 24:51-167. Simons, E. I,. 1968. Early Cenozoic mammalian faunas, Fayum Province, Egypt, Part I: African Oligocene Mammals: Intro duction, history of study, and fauna] succession. Peabody Museum Bulletin 28:1-22. Tilden, C. D., P. A. Holroyd, and E. I.. Simons. 1990. Phylctic aHinities of Aptcrodon (Hyaenodontidae, Crcodonta). Journal of Vertebrate Paleontology I (((supplement ):46A. Van Valen, I.. 1965. Some European Provivcrrinac (Mammalia, Deltatheridia). Paleontology 8:638-65. Van Valen, I.. 1966. Deltatheridia, a new order of mammals. Bulletin of the American Museum of Natural History 132:1- 126. Van Valen, L. 1967. New Paleocene inscctivores and insectivore classification. Bulletin of the American Museum of Natural History 135:217-84.