T

Odobenocetops peruvianus, the Walrus-Convergent Delphinoid (Mammalia: Cetacea) from the Early Pliocene ofPeru

Christian de Muizon, Daryl P. Domning, andDarlene R. Ketten

etops and the tusks ofOdobenus was as orientation guides in feed­ ABSTRACT ing. This reopens the question of whether the tusks of walruses playa role in feeding, as it seems that these also may be useful as Odobenocetops peruvianus Muizon, 1993 (early Pliocene, orientation guides for the mouth and vibrissal array. southern Peru), is a bizarre cetacean that is convergent in its skull, general aspect, and presumably feeding habits with the modem walrus Odobenus rosmarus (Linnaeus). Its cranial specializations are unique among cetaceans and include loss of the elongated ros­ Introduction trum, development of large premaxillary processes housing asym­ '.ctrical tusks, forward migration ofthe bony nares, reversal of the Abundant remains of fossil odontocete cetaceans have been typical cetacean telescoping of the skull, dorsal binocular vision, found in the rocks of the Pisco Formation near the southern large vaulted palate, and an inferred upper lip. The structure of the coast of Peru. Although the occurrence of cetacean bones in basicranium (possession of palatine expansions of the pterygoid this area has been known for more than one hundred years (Lis­ sinus and presence of a large cranial hiatus) and face (possession of a medial portion of the maxillae at the anterior border of the son, 1890), the first odontocete described from this formation nares) indicates that it belongs to the odontocete infraorder Del­ was Incacetl/s brogii Colbert, 1944. Subsequently, Hoffstetter phinida and to the superfamily Delphinoidea. Within this group (1968) was the first to show the importance of the faunal as­ Odobenocetops is related to the Monodontidae because of the lat­ semblage (fish, reptiles, birds, and mammals) of the locality of eral lamina of its palatine flooring the optic groove, the anteropos­ terior elongation of the temporal fossa, and the thickness of the Sacaco, in the southern outcrops of the Pisco Formation, 540 'isphenoid and squamosal in the region of the foramen ovale. We km south of Lima. Further studies by Muizon (1981, 1983a, hypothesize that Odobenocetops, like the walrus, fed upon shal­ 1983b, 1983c, 1984, 1986, 1988), Pilleri (1989, 1990), and low-water benthic invertebrates and probably used its tongue and Cheneval (1992) have described part of the vert,ebrate fauna of upper lip jointly in extracting the, soft parts of bivalves or other the Pisco Formation, but abundant material still remains un­ invertebrates by suction. The highly modified morphology of the rostrum indicates that there was no melon as in all other odonto­ studied (work in progress includes that ofC. de Muizon and G. cetes, and therefore that Odobenocetops was probably unable to McDonald on mammals, and 1. Cheneval on birds). As estab­ echolocate; binocular vision could have compensated for this lished by Marocco and Muizon (1988), the Sacaco vertebrate inability. The most probable function of the tusks themselves was fauna was deposited under shallow waters in a littoral and social, as ir the living walrus, but we suggest that the historically ',rimary function ofboth the premaxillary processes ofOdobenoc- beach environment. The preservation of the fossils is excep­ tional, and associated skeletons are not rare; both characteris­ tics indicate calm waters. Christian de Muizon, UMR 8569 CNRS, Laboratoire de Paleontolo­ In 1990, the skull of an unexpected walrus-like cetacean gie, Museum National d'Histoire Naturel/e, 75005 Paris, France. (Muizon, 1993a) was found in the locality called "Sud-Sacaco" Email: mui::[email protected]. Daryl P. Domning,Research Associ­ by Muizon (198 I). In the Sacaco area, Muizon (1981, 1984, ale, Department ofPaleobiology, National Museum ofNatural His­ 1988a), Muizon and DeVries (1985), and Muizon and Bellon lory, SWithsonial/ II/stitutiol/, Washingtol/, D.C. 20560-0121, and LaboratoJ)' o/Evolutionary Biology, Department ofAnatomy, College (1986) recognized five vertebrate horizons, which range from of}vledicine, Howard University, Washington, D.C. 20059, United the late Miocene to the late Pliocene (approximately 9 Ma to 3 Slates. En/ail: [email protected]. Darlene R. Ketten, Massa- Ma). The specimen came from the Sud~Sacaco (SAS) horizon, husetts Eye and Ear Infirmary, Harvard Medical School, Boston, earliest Pliocene, which has yielded an abundant vertebrate .>fassachusetts 02114, United States. Email: [email protected]. fauna (Muizon, 1981, 1984). Cetaceans are represented there

223 224 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

by balaenopterids, cetotheriids, pontoporiids (Pliopontos lit­ Genus Odobenocetops Muizon (1993a) toratis), ziphiids (Ninoziphius platirostris), and phocoenids TYPE SPECIES.-Odobenocetops peruvianus Muizon (1993). (Piscolithax longirostris). The walrus-like cetacean, named DIAGNOSls.-As for the species. Odobenocetops peruvianus, was briefly described by Muizon (1993a, 1993b), who concluded that its feeding adaptations Odobenocetops peruvianus Muizon (1993a) were convergent with those of the walrus. Herein we describe that specimen more thoroughly and present a more detailed FIGURES 1-18, TABLE I study of its features as a foundation for observations on the pa­ TYPE SPECIMEN.-USNM 488252 (originally numbered leobiology of this unique cetacean. Its phylogenetic relation­ USNM 460306), an incomplete skull lacking much of the left ships with delphinoid odontocetes are confirmed. side and all ofthe auditory bones. INSTITUTIONAL ABBREVIATIONS.-The following abbrevia­ REFERRED SPECIMENS.-MNHN SAS 1613, a left periotic tions for institutions have been used: and partial tympanic; and MNHN SAS 1614, a left periotic. MNHN Museum National d'Histoire NatureIle, Paris One other, recently discovered specimen (MNHN SAO 202) NMNH National Museum of Natural History, Smithsonian Institution, is relevant to this study but is not described herein in detail; it Washington, D.C. will be described in a separate report (Muizon andDomning, SMNK Stnatliches Museum fUr Naturkunde, Karlsruhe, Gcnnany USNM Collections of the NMNH, which include those of the former 2002). It is a badly weathered partial skeleton with a partial United States National Museum skull, including the left tympanic, periotic and asymmetrical tusk; the distal half of a humerus; a complete radius; several ACKNOWLEDGMENTS.-This research was supported by dorsal and caudal vertebrae; and rib fragments. It was found in funds from the Centre National de la Recherche Scientifique the SAO horizon of the Pisco Formation (ca. 3.5 Ma), which is (CNRS), Paris, in 1991; by two short-term visitor awards from younger than the SAS horizon (ca. 5 Ma), and is provisionally the Smithsonian Institution, Washington D.C., in 1992 and referred herein to Odobenocetops sp. (see "Addendum," be­ 1993; and by funds from the Institut Franyais d'Etudes Andines low; Muizon et aI., 1999). (Lima, Peru) in 1994 (to C. de Muizon). CT scanning and peri­ ETYMOLOGY.-From odon, tooth, and baino, walk (Greek); otic analyseswere supported by Office ofNaval Research con­ and from cetus (Latin, from Greek ketos), whale, and ops tracts NOOOI4-2-J-4000 and NOOOI4-93-1-0940 (to D.R. (Greek, masculine), like: "the cetacean that seems to walk on \ Ketten) and by the Three-Dimensional Head and Neck Imaging its teeth," to refer to its feeding position (see below) and also to n Service, Department of Otolaryngology, Massachusetts Eye its similarity to the walrus (Odobenus). Species: peruvianus, s and Ear Infirmary. Beluga periotics were provided by S. Ridg­ from Peru. r way, Naval Research and Development (NRD, San Diego. We DIAGNOsls.-Delphinoid cetacean characterized by loss of s' thank C. Martin, who found the specimen, and P.L. Larson, elongate cetacean rostrum and concomitant probable absence of melon; development of large, ventrally directed premaxil­ who collected it and donated it to the Smithsonian Institution. lary alveolar processes housing straight tusks; right tusk longer P. Kroehler prepared the specimen. Illustrations, including the J than 55 cm, cylindrical, with oval section, with long, open pulp life reconstruction (Figure 20), are by Mary Parrish (NMNH); s cavity 23 cm deep; left tusk probably unerupted and probably s photographs are by Vic Krantz and Chip Clark (both NMNH), no longer than 20 cm, with short conical pulp cavity 1.5 cm a' and D. Serrette (MNHN). Irina Koretsky helped greatly with deep; important muscular insertion and numerous neurovascu­ d the illustrations. Valuable insights through discussion were pro­ lar foraminaon anterior side ofpremaxillae and on anteriorex­ r vided by L.G. Barnes, T.A. Demere, R.E. Fordyce, J.G. Mead, tremity of palate, suggesting presence of strong upper lip and c C.E. Ray, J.S. Reidenberg, and A. Werth. Ray and Werth also possible vibrissae; regression in telescoping of skull by for­ provided detailed formal reviews of the manuscript. ward migration of nares and anterior withdrawal of maxillae

We are pleased to dedicate this paper to Clayton E. Ray, who and frontals; nasal on vertex of skull, contacting occipital and l invited us to study the specimen under his care andwho greatly right maxilla, lying upon frontals but separated from meseth­ I helped and encouraged us in our work. moid, which followed bony nares in their forward migration; t large temporal fossa open dorsally; large dorsal exposure ofpa­ Systematic Paleontology rietals and concomitant development of temporalis muscle ori­ \ gin; large orbits facing dorsally and not laterally as in other od~ Superfamily DELPHINOIDEA ontocetes; large pre- and postorbital processes expanded' Family ODOBENOCETOPSIDAE Muizon (1993a) anterolaterally; maxillae articulating behind nares; large, deep and vaulted palate; maxillae excluded from palate but forming TYPE GENUS.-Odobenocetops Muizon (1993a). part oflateral wall ofskull; no maxillary teeth; no jugal; vomer a DIAGNOSIS.-As for the single known species, O. peruvi­ very large and lanceolate; lateral lamina ofpalatine flooring the n anus Muizon (1993a). anteroposteriorly oriented optic gutter; pterygoid with flattened il .JMBER 93 225

subhorizontal hamular process and large rounded lateral crest reduction of the rostrum and the development oflarge premax­ for origin of pterygoid muscle; ventral surface of zygomatic illary alveolar processes housing the tusks that give the skull a process ofsquamosal forming large and deep anteroposteriorly strong, although superficial, resemblance to that of the modern elongated, trough-like glenoid fossa; large cranial hiatus; al­ walrus. As preserved, the skull is large, with an anteroposterior isphenoid and squamosal strongly thickened on anterolateral length of approximately 46 cm (the posterior part of the only region of cranial hiatus. Periotic with relatively small pars co- condyle preserved is missing) and an estimated bizygomatic 'dearis and long and thickened anterior process with small width of 18 x 2=36 cm (Figure I). 'tillar facet; ventral process with relatively flattened ventral Premaxilla: The premaxillae are certainly the most amaz­ rim and large ventral tuberosity; very large dorsal aperture of ing bones of the skull of O. perllvianlls. Compared with other aqueductus vestibuli; periotic tympanic with clearly sigmoid­ odontocetes, the elongate rostrum has disappeared, and the pre­ shaped involucrum. maxillae show long ventral processes just anterior to the preor­ GEOLOGICAL FORMATION AND AGE.-Pisco Formation, bital notch that form an angle of approximately 60° with the SAS (Sud-Sacaco) level as defined by Muizon (1981) and horizontal plane of the skull (Figures 2, 3). This condition de­ Muizon and DeVries (1985). As stated by these authors, the fines a small dorsal horizontal portion ofthe premaxillae, situ­ age of the Sud-Sacaco beds is early Pliocene, probably basal ated between the anterior edge of the naris and the apex of the "iocene, approximately 5 Ma. snout, and a large subvertical portion, the alveolar process, lo­ TYPE LOCALITY.-Sud-Sacaco, southern coast of Peru, on cated almost entirely on the ventral part ofthe skull (i.e., below the west side ofthe Panamerican Highway at km 542. the supraorbital process of the frontal). The ventral extremity of the alveolar process is posterior to its dorsal extremity and DESCRIPTION below the postorbital process of the frontal. It is hollowed by large alveoli housing the tusks (see below). The right premax­ GENERAL FEATURES.-The general morphology of the skull illa is almost complete, but the left is broken and lacks most of of Odobenocetops perllvianlls is far from that ofa typical ceta­ the alveolar process. cean. The hyperspecialization of this species (within the ceta­ On the dorsal side of the skull (Figures I, 2), the premaxillae ···~ans) has resulted in the modification of the characteristic form the anterolateral edges of the nares; in that region, the .iGscoping ofthe cetacean skull. The skull is bilobate in dorsal bones have a conspicuous posterolateral process that runs view, with a large braincase posteriorly and an anterior part along the anterior half of the lateral border of the nares on the made up of the narial, orbital, and rostral regions. These are left side, and along nine-tenths of its length on the right side. separated by a well-marked constriction situated behind the su­ .The anterior border ofeach naris is formed medially by a small praorbital processes ofthe frontals. Such a condition is not ob­ medial portion of the maxilla and laterally by the premaxilla. served in the other Neogene odontocetes, where the posterior Anterior to the nares the premaxillae become narrower, being expansions ofthe frontals and maxillae cover the braincase and narrowest just behind the fairly large premaxillary foramina. In the temporal fossa. In some primitive odontocetes (Xenoro­ that region they are thick and each bears a dorsolateral rounded ·"!lIs. Ardweodelphis. and Agorophills) and in archaeocetes, a keel that runs anteriorly from the posterolateral process, passes ·milar condition can be observed because of the slight tele­ lateral to the premaxillary foramen (forming its lateral border), scoping ofthe skull; in Odobenocetops. the condition described and converges with the other keel in the anteriomlOst region of above is the result of a character reversal that considerably re­ the snout. duces the telescoping. Related to this reversal is the anterior These keels mark the limits ofan elongate triangle just ante­ position of the nares and of the orbits. The nares are very large rior to the bony nares; in this area, in other odontocetes, are lo­ compared with those in the Holocene Delphinoidea. They are cated the premaxillary sacs (posterior to the premaxillary fo­ anteroposteriorly elongate, and their length is more than double ramina) and the origins of the nasal plug muscles (medial and their width. The left naris is slightly larger than the right. The anterior to the premaxillary foramina). The premaxillary sacs '(J[sal openings of the nares are horizontal (contrary to other . of the other Delphinoidea are generally situated upon the part ·dphinoidea, where they are anterodorsally oriented), and of the premaxilla posterior to the premaxillary foramina and they are not partially overhung in their anterior part by the anterior and lateral to the nares; in Odobenocetops there was maxillae and the premaxillae. In Odobenocetops the anterior very little space for such air sacs and, ifpresent, they must have walls of the narial passages face posterodorsally (posteroven­ been extremely reduced. The origin ofthe nasal plug muscle on trally in other delphinoids) and their contact with the dorsal the premaxillae of other delphinoids is generally a rugose loz­ surface ofthe premaxillae is a gently rounded surface, whereas enge-shaped or triangular area located medial and mostly ante­ it is a sharp crest in the other delphinoids. In lateral view, the rior to the premaxillary foramina; in Odobenocetops, this part dorsal profile ofthe skull is markedly concave; the nuchal crest of the premaxilla is small, suggesting that the nasal plug mus­ .,od the rounded anterior part ofthe snout clearly overhang the cle, ifpresent, would have been fairly weak. .:mch lower narial fossae and the posterior portion of the max­ Demarcating the lateral edges of the triangular surface, and illae. Another obvious characteristic of Odobenocetops is the anterior to the premaxillary foramina, the rounded keels men- 226 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

\ \ \ \ \ \ \ \ \ ( \ \ \ /'----. , ( / I i \ I I \ \ I \ \ I \ I \ \ I I \ I I \ / I ,_/ I / F \: / , ./ I \ I \ I \ \ I ) \ / "- ,,/ "\ \- ~ --- \ A "- B E

NA

, "\ I I (~ ...... / ,"- \ \ I ...... , I \'" ) J \ \\ ,\ \ ,\ '\ ,\ \ ,\ ' \.\ \ c '~, \. , L--J 5CM

FIGURE I.-Odobenocetops pel:ul'ionus, hQlotype (USNM 488252): skull in dorsal view (0), ventral v.iew (b), lateral view (c), and anterior view (d); right pn,maxillary tusk (e) in lateral view (bottom) and proximal view (top); left premaxillary tusk (f) in lateral view (bottom) and proximal view (top). tioned above are very thickened and most probably represent eral to the origin of the nasal plug muscle receive fibers of the strong muscular attachments. These strong attachments are not medial portion of the rostral muscle (Lawren.ce and Schevill, in the position of the nasal plug muscle of other delphinoids, . 1956; Mead, 1975). In Odobenocetops, they probably represent however, and consequently are not related to it. In fact, in. the part of the rostral muscle modified for moVements of a very other delphinoids the two rounded keels of the premaxillae lat- strong upper lip (see below and "Discussion").

I .1 NUMBER ')3 227

riar sides of the premaxillae show a very spongy structure and several large and small foramina (also seen on the anterior part of the palate), which indicate extensive vascularization and in­ nervation. The anterior edge of the alveolar process of the premaxilla presents a strong crest along its whole length (Figure 4). This crest is a long and irregular surface of bone, concave medially, 1-2 em wide, which might have carried a long and narrow horny pad or, more likely, strong connective tissue for attach­ ment of the upper lip. Several large vascular foramina are ob­ servable on the dorsomedial portion ofthis attachment area. Its morphology is very similar to what is observed on the dorsal side ofthe walrus mandible, between the first mandibular tooth and the apex ofthe symphysis. Fay (1982: 167) mentioned that, in the walrus, "the incisive area is covered by an extraordinar­ ily tough firm gingiva, unlike that on any other part of the mouth but closely resembling the skin on the palms and soles of the flippers." Furthermore, Howell (1927:21) mentioned the presence of a very hard horny tissue surrounding the lips of Neophocaena phocaenoides, and Kleinenberg et al. (1969:80) noted in the lips of Delphinapterus leucas a "many-layered keratinized epithelium" indicating that such structures are not uncommon in other delphinoids. Given the important modifica­ tion of the superior edge of its mouth, however, it is likely that the keratinization was much more pronounced in Odobenoce­ tops than in the other delphinoids. The condition of the anterior face of the premaxilla suggests the presence ofa well-developed upper lip, and given the vas­ cularization of the bone, it is possible that it possessed strong vibrissae, as observed in the walrus. Hair is known to occur on the snouts ofvarious fetal and newborn odontocetes (Yablokov and Klevezal, 1962; Tomilin, 1967; Brownell, 1989) and on the apex of the rostrum of the living Amazon dolphin Inia geoff­ rem'is (Best and da Silva, 1989). The rough and vascularized area mentioned above is re­ ,,,,(j'RE 2,-Odo!Jclloce{OpS pel'/I\'il/II/1s. holotype (USNM 488252): dorsal stricted to the upper one-third ofthe alveolar process, except view ofthe skull. for the anterior crest, and may mark the limit of muscle origin and vibrissae. Anterior to the premaxillary foramina, the premaxillae are In anterior view, the upper part of the mouth opening had a strongly expanded transversely; at the level of the preorbital wide and elevated parabolic shape (Figure 5). In lateral view, process they are approximately twice as wide as at the level of the alveolar process ofthe right premaxilla is in sutural contact the premaxillary foramina. On the dorsal portion, the right pre­ along the proximal one-half of its length with the maxilla, f!'1Xilla is slightly wider and larger than the left, resulting in whereas its distal half is free. It is slightly concave posteriorly 11l11etry of the skull. The two premaxillae are separated by a and widened at its apex; the posterior border of its free portion prominent premaxillary groove, which is partially filled in its is rounded in cross section, differing therefore from the anterior posterior part by the mesethmoid. The anterior region of the crest, which received muscular and/or tendinous attachment of snout is formed by the premaxillae only; it is very wide, mas­ the upper lip. The lateral surface of the alveolar process is sive, and rou11lkd. In anterior view, the apex of the snout is a clearly convex and gently rounded, whereas its medial side is high and nearly \'ertical wall, fissured vertically by the premax­ slightly convex at the apex and distinctly concave in its proxi­ illary groove (Figures 4, 5). From that massive apex, two large mal portion, where it forms most of the anterior bony palate. alveolar processes are directed ventrolaterally and diverge at an The section ofthe alveolar process on its proximal two-thirds is 'angle of approximately 60°. In anterior view, the rounded apex teardrop-shaped, having a wide, convex, and rounded posterior ( ,:le snout shows an irregular and spongy surface that seems edge and a thin, narrow, angular anterior edge. On the posterior tv indicate strong muscular attachments (Figure 4). The ante- side of the alveolar process, a small ridge runs from the poster- 228 SMITHSOl':IAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 3.-0dobenocerops penlVianlls, holotype (USNM 488252): lateral view ofthe skull.

omedial point ofjunction of the alveolar process with the pal­ ate (exactly at the anteroventral angle of the maxilla) to the posterolateral extremity of the process. On the palate, the premaxilla is in contact with the vomer medially and with the palatine posteriorly. The maxilla does not participate in formation of the bony palate. Along its me­ dial suture with the vomer, the premaxilla shows a deep, an­ teroposteriorly elongated fossa with a very large foramen (7-8 mm in diameter) posteriorly; this foramen, which opens in the palatine, is the anterior opening ofa canal originating in the in­ fraorbital canal and extending into the premaxillary fossa via a well-marked gutter. This greatly enlarged foramen corresponds to the palatine foramen that is commonly observed, but very re­ duced and sometimes nearly closed, in the other Delphinoidea . (Figure 6). On. the lateral edge of the premaxillary fossa and lateral to it, several other smaller (but still 2-4 mm in diameter) foramina perforate the premaxilla. There is little doubt that these particularly large foramina contained blood vessels for the abundant vascularization of the upper lip. The fossa is lim­ ited medially by the vomer, which forms its medial wall. The significance ofthis fossa is not apparent, but in many other mammals, it is the lbcation ofa chemical sense organ (the vomeronasal organ). Chemoreception organs in living dol­ phins, however, are located on the tongue (Fried et aI., 1990; Kuznetzov, 1990). Nothing definite can be said about the possi­ bility ofhomology ofthe premaxillary fossae with the incisive

'foramina ofterrestrial mammals (these foramina are absent in FIGURE 4.-0dobenocerops perllvianlls. holotype (USNM 488252): anterior cetaceans), but the various pieces ofevidence given below that view of the skull approximately perpendicular to a horizontal swimming posi­ Odobenocetops is a cetacean \vould favor the hypothesis that tion.

1 ~.. ~.v1l3ER 93 229

FIGURE5.-0doh<.'lloc<.'(O!IS p<.'rl/viaill/s. holotypc (USNM 488252): anterior view of the skull approximately parallel to the plane made by the alveolar pro­ cesses (i.e., to the plane parallel to the bottom in a feeding position).

':ese structures are not homologous. FIGURE 6.-0dob<.'IIOC<.'(OPS perl/ViClIll/S. holotype (USNM 488252): ventral The premaxilla extends onto the palate in a triangular poste­ view of the skull. rior expansion that is in contact with the palatine; the contact between the two bones is flat and nearly horizontal, and the su­ continuous growth; at its base the tooth itself is only a thin ture on the palate forms a large V-shape, opening anteriorly. The alveolus of the right premaxilla is straight, transversely layer of dentine surrounding the pulp cavity, whose diameter flattened (i.e., oval in section), and 30 cm deep; the left alveo­ decreases toward the apex. A tentative reconstruction of the lus is not completely preserved, but it is clear that it had a much tooth shows that it could not have been shorter than 45 cm and smaller diameter than the right. possibly was no longer than 55 cm, 15-25 cm of the tusk thus Tusks: The right tusk (Figure 7) as preserved is not com- being external to the alveolus. ,ete; however, a 39 cm long portion ofitindicates that it was a The left tusk (Figure 8) is known from a small, 8 cm long straight cylindrical tooth \-vith a transversely flattened section fragment. It has a much smaller diameter than the right tusk (at 10 cm ii'om the base of the tusk, the two diameters are 38 (approximately 20% smaller); its section also is flattened trans­ mm and 30.5 mm). Its cross section is constant along the length versely, and at its base the section is 30 mm long and 24 mm of the whole fragment available, although it is possible that the wide. Like the right tusk, it has no enamel and its surface section of the (badly crushed) anterior extremity of the fossil is shows very fine longitudinal striations. Unlike the right, its sec­ slightly smaller than the intra-alveolar portion. The tusk is tion is not constant and at its apex, as preserved, the diameters made of dentine only and does not show any enamel; its sur­ are 27 mm and 25 mm; in the middle part of the fragment the face is regular and only shows fine and regular longitudinal tooth is slightly inflated, and its diameters are 32 mm and 27 triation. A 23 cm long pulp cavity indicates that the tooth had mm. Consequently, the tooth shows a slight but distinct taper- T 230 SMITHSONIA;\ CONTRIBUTIO,\S T()\UU1310LOGY I I

b

2cm

FIGURE 7 (left).-Odobenocetops peruvianus, holo­ type (USNM 488252), right tusk: a, lateral view; b. proximolateral view, showing the deep pulp cavity.

FIGURE 8 (right).-Odobenocetops peruvianus. holotype (USNM 488252), left tusk: a, lateral view; b. proximolateral view, showing the small pulp cav­ ity.

a ing toward its apex. By projecting apically the edges of the fragment, it is possible to estimate the original length of the tusk to be between 15 cm and 20 cm. A wide-open but very short pulp cavity is present at the base of the tooth; it has a small conical opening, 15 mm deep, 30 mm long, and 24 mm wide at its base. This condition indicates that the tooth was still growing, although obviously much more slowly than the right tusk. Evidence of continued growth does not prove that the tooth .was erupted, as indicated by the presence of a deeper pulp cav­ ity on an unerupted incisor of a 17-year-old female Dugong dugon (Marsh, 1980). The condition ofthe unerupted teeth of 2cm the narwhal, however, favors the hypothesis ofan erupted tooth (Hay, 1980). In the newborn narwhal, the unerupted teeth (the sexual maturity. With the closure of the cavity, the root of the right in the male and both in the female) have a long pulp cav­ tooth develops a knot made ofdentine and irregu Jar cementun' ity that occupies almost the entire length ofthe tooth; the cavity Consequently, the narwhal's unerupted tusk stops growing be­ closes after deposition of approximately eight to nine dentine fore the animal has attained its full size and sexual maturity. In layers, which generally corresponds to an age younger than the skull of Odobenocetops, the pulp cavity is small but clearly ~UMBER 93 231 open, the knot at the root of the tooth is lacking, and the indi­ a deep groove in the maxilla, communicating with the premax­ vidual described herein is a relatively old adult, considering the illary foramen; and (3) a third small foramen anteromedial to state of fusion of its cranial bones. If the tusk was erupted, the antorbital notch and distinctly pinched in the "folding" of therefore, the premaxillae of Odobenocetops could have been the rostrum. On the left maxilla, the dorsal opening of the in­ strongly asymmetrical and the alveolar process of the left pre­ fraorbital canal is divided into four foramina, and through a maxilla should not have been longer than 18 cm, considering broken portion of the lateral wall of the alveolus it is possible 'lat a small portion of the tusk had to be external to the pre­ to see that one or two of these foramina gave passage to blood maxilla; the alveolar process ofthe left premaxilla would there­ vessels for irrigation of the tusk pulp cavity. fore have been, at a minimum, almost 10 cm shorter than the The portion of the maxilla ventral to the supraorbital process right. Such asymmetry, which strongly modifies the external of the frontal is a small triangle of bone bordered anteriorly by aspect of the head, is unknown in mammals, however; conse­ the premaxilla, posteriorly by the palatine, and dorsally by the quently we conclude that the left tusk of Odobenocetops was frontal. A large posterior opening of the infraorbital canal is unerupted and that, as in the female dugong, it was still grow­ present in this ventral portion of the maxilla, just below the su­ ing very slowly in the closed alveolus. The alveolar process of ture with the frontal. As mentioned above, the maxillae ofOdo­ the left premaxilla cO,uld, therefore, have had a length similar to benocetops have been excluded from the palate and no maxil­ 'lat of the right side (or only slightly smaller) and the - 20 cm lary teeth are present. The ventral portions of the maxillae are long left tusk could still have grown for a long time in its 30 cm tightly fused to the premaxillae, thus indicating that the indi­ long alveolus. vidual was not a young adult. Furthermore, it is noteworthy that the right tusk of Odobeno­ Mesethmoid: The mesethmoid is a large, relatively thick cetops differs from all other tusks found among mammals in blade of bone separating the nares. In dorsal view its posterior being straight and cylindrical. extremity presents a Y-shaped relief that overhangs a lower Maxilla: The maxillae of Odobenocetops peruvianus are portion, posterior to the nares and partially overlapped by the considerably reduced in comparison with those ofother delphi­ maxillae. This condition is unique among odontocetes and is noids; they do not cover the braincase posteriorly or the fron- related to important modifications of the skull, such as the long -Is laterally, and they do not participate in the construction of contact of the maxillae behind the nares, among others. The ,he bony palate. They form the posterolateral borders of the nares are large and anteroposteriorly elongated (almost three narial openings, and they are in contact behind the nares; be­ times longer than wide); the left naris is wider than the right tween that region and the occipital crest they are reduced to a and is located a little more anteriorly. The right naris is -28 narrow strip of bone in the median area of the skull. The right mm wide and -62 mm long, and the left naris is -31 mm wide maxilla is two to three times as wide as the left. They lie upon and -62 mm long. the frontals, and at their posterior ends they contact the nasals. On the dorsal face of the nasal passage, at the same level as On the specimen described herein, they are only preserved for the anteriormost point ofthe brain but slightly more lateral, is a two-thirds of the distance between the nares and the nasals, but group of small foramina that are directed posteromedially. ~eir sutures with the frontals are clearly visible. Lateral to the Those foramina are most probably related to the small olfac­ dares, the maxilla forms a very narrow flange, no more than 2 tory lobes of the brain (see below). cm wide, that is strongly withdrawn medially from the lateral Frontal: The frontals have the typical cetacean trait of en­ edge of the frontal. Its lateral border is sigmoid and more or larged supraorbital processes, composed of the preorbital pro­ less follows the lateral profi Ie of the frontal, being wider above cess and the postorbital process separated by a deep orbital the pre- and postorbital processes and narrower above the deep notch. The preorbital process is an elongate, flat, and horizontal orbital notch. Atthe level of the preorbital process of the fron­ wing; its anterior extremity is square, and its lateral angle is tal, the maxilla shows a small lateral spine, which corresponds distinctly twisted dorsally. It has a marked anterior orientation to the preorbital process ofthe maxilla in the other delphinoids, and runs along the posterolateral edge of the base of the alveo­ '.d which overhangs the jugolacrimal and is lateral to the an­ lar process of the premaxilla. The antorbital notch is long (40 (orbital notch. In Odobenocetops the jugolacrimal has disap­ mm) and narrow (8 mm) and does not widen anteriorly as it peared. The maxillary rim, which forms the posterolateral bor­ does in most delphinoids. The postorbital process is large and der of the nares, continues posteriorly (4 cm behind the nares much stouter than the preorbital process. It is a large triangular on the right side and 3 on the left side) and overhangs the flat plate whose apex faces posterolaterally and not laterally as in lateral portion of the maxillae that overlaps the base of the su­ other delphinoids. Like the preorbital process, it is markedly praorbital process of the frontal. stretched anterolaterally. It is improbable that the masseter Three maxillary foramina are found on the right maxilla: (I) muscle, which arises partly from the postorbital process in the a small foramen at the level of the posterior edge of the nares, other odontocetes (Howell, 1927), originated on that process in cated between the maxilla and the frontal and opening poste­ Odobenocetops (see below for discussion). These two pro­ liOrly; (2) the anterodorsal opening of the infraorbital canal at cesses are separated by a deep and wide orbital notch that faces the contact of the premaxilla and maxilla, which is preceded by dorsally and not laterally as in other delphinoids, therefore in- 232 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

dicating very good dorsal binocular vision. The eyes were very of that morphology would therefore be that the right elongated large, probably the size of a tennis ball, and abutted the poster­ fossa and the crest serve as muscular attachments for part ofthe olateral side of the base of the alveolar process of the premax­ maxillonasolabialis. In odontocetes the pars nasalis of this illa. muscle is the blowhole muscle, which arises from the maxilla In dorsal view, the posterior edge of the supraorbital process and part of the frontal; the withdrawal of these bones from the is almost straight and oblique, and it forms the anterior border roof of the braincase would have forced the fibers to concen­ of the temporal fossa. The supraorbital process does not cover trate in that fossa and on that crest. It would, however, be sur­ the temporal fossa as it does (at least partially) in all Neogene prising ifthis phenomenon occurred only on the right side. Un­ and Holocene odontocetes. In those forms its posterior edge is til new material is discovered, the first interpretation seems anteroposteriorly oriented and not oblique as in Odobenoce­ more reasonable. Consequently, it is likely that the maxillona­ tops. As the supraorbital process is not (or only slightly) cov­ solabialis was relatively reduced in Odobenocetops. This mus­ ered by the maxilla, it is widely exposed dorsally on the antero­ cle is related to the movements ofthe blowhole and ofthe nasal lateral region of the braincase. As a consequence of the air sac complex, two structures related to ultrasonic sound pro­ reduction of the telescoping of the skull, the frontal of Odo­ duction, so it is probable that Odobenocetops had a relatively benocetops does not cover the parietal posteriorly and does not reduced air sac system (already observed in the reduction ofthe participate in the formation ofthe roof ofthe temporal fossa as premaxillary sacs) and consequently reduced ultrasonic sound it does in all the other delphinoids. Consequently, this bone production consistent with loss ofecholocation. The posterior also is widely exposed on the posterolateral region ofthe brain­ extremities ofthe frontals articulate with the nasals. case. The anterolateral portion of the suture with the parietal is Ventrolaterally, the frontals contact the alisphenoid and the transverse, but it shows two big indentations, which are clearly palatine posteriorly. In ventral view, the supraorbital process of visible on the right side where it is completely preserved. Me­ the frontal is in a subhorizontal plane that meets the lateral wall dially, each frontal is exposed, like the maxilla, in a long me­ ofthe palate at a right angle; along this angle, the frontals artic­ dian strip that apparently joins the occipital posteriorly (sutures ulate successively with the orbitosphenoid (optic gutter), with in the vertex are not easy to determine as the bones are badly the palatine, and anteriorly with the maxilla. encrusted with iron oxide). In that region, therefore, the classi­ Nasal: The nasals are small plates of bone that have been cal relationships of the telescoped odontocete skull are pre­ pushed posteriorly; they articulate with the frontals anteriorly served, as the maxillae overlap the frontals, which overlap the and with the occipital posteriorly. The nasals apparently lie parietals. upon the parietals, whereas in other odontocetes they always lie On the right frontal, the narrow posterior portion is elevated upon the frontals. The lateral halfofthe right nasal is preserved medially and presents a strong keel that runs as far as the occip­ in situ, but its medial half, as well as the left nasal, have been ital; the strip fomled by the right maxilla lies on the medial side lost during fossilization. The sutures, however, are clearly visi­ of this frontal keel and reaches the nasals posteriorly. In fact, ble in numerous small anteroposteriorly directed grooves that lateral to the maxilla-frontal suture there is a deep, elongate mark the skull in that region. The nasals were more wide than fossa located medial to the bottom of the temporal fossa and long, and apparently the left nasal was larger than the right. running from the posterior edge of the supraorbital process to Parietal: The parietal forms most of the dorsal face of the approximately the middle ofthe braincase. This fossa is absent braincase as a smooth bony plate that is covered only by the on the left side of the skull and could be interpreted as a patho­ frontal medially. It contacts the squamosal by a very strong in­ logical deformation. The keel mentioned above also could be a terdigitated suture, which differs from the squamous suture pathological structure related to the first one; in the middle re­ generally found in mammals. As seen below, this condition is gion of the braincase, the right frontal appears to have been probably related to the feeding adaptations of Odobenocetops. pinched transversely and thus elevated. It is possible that the Ventrally, the parietal shows a small expansion in the anterior animal suffered a minor skull trauma when young, the traces of part ofthe glenoid cavity. which are still observable on the skull. Because of that defor.; Palatine: The palatine forms an importantpart ofthe pal­ matiori, the posterior median strip ofthe frontal clearly appears ate; it is located posterior to the premaxilla, medial to the pter:­ to be, in dorsal view, narrower than the left frontal at that level. goid, and lateral to the vomer. As mentioned above, a very In fact, however, if one "unfolds" the right frontal, both fron­ large palatine foramen opens anteriorly in a deep elongate tals have approximately the same width. fossa on the anteromedial border ofthe premaxilla. The palate Another interpretation would be that the asymmetry of these is very large and deeply arched transversely, as well as longitu­ structures is related to the asymmetry that characterizes the del­ dinally. Very wide anteriorly, at the level of the premaxillae, it phinoids. If this were the case, however, the asymmetry pattern narrows posteriorly at the level of the palatines and widens observed in Odobenocetops would be unique among cetaceans. again at its posterior extremity at the level of the pterygoids. In delphinoids, the asymmetry is always manifested by larger The anterior portion ofthe palatines is located in the narrowest size of the right maxilla and premaxilla and a displacement of part of the palate, and consequently the anterior part of the pa­ the vertex to the left side of the skull. A possible interpretation latine (on the palate) is much narrower than the posterior. On "UMBER 93 233 the lateral side of the skull, below the supraorbital process, the notch excavated in a very thickened and dense bony wall. This palatine forms a subvertical wall with a deeply concave ventral structure is in the location of the passage of the eustachian tube border. The suture with the maxilla is markedly concave pos­ observed in the other odontocetes, although it is never as con­ teroventrally. Above the lateral edge of the palate and just ante­ spicuous in the latter. The pterygoid forms the dorsal wall of rior to the suture with the pterygoid, the palatine presents an the choanae and partially overlaps the vomer medially, contrary elongate fossa that also extends onto the pterygoid. That fossa to the condition in other odontocetes, where the vomer gener­ :5 deeper on the palatine and gradually shallows and disappears ally overlaps the medial border of the medial lamina of the . n the pterygoid posteriorly. It foIIows the lateral edge of the pterygoid. The lateral lamina of the pterygoid contacts the pa­ palate and consequently is oriented anterodorsally-posteroven­ latine anteriorly and the alisphenoid posteriorly. It is large and trally. Boenninghaus (1904), referring to Phocoena, and Fraser smooth and participates in the formation of a continuous bony and Purves (1960), referring to Delphinus, mentioned that the bridge between the palate and the frontal, a condition that origin ofthe internal pterygoid muscle was on the lateral edge strongly recalls that observed in the Monodontidae. of the maxiIIa, palatine, pterygoid, and (in the case of Phoc­ Vomer: The vomer is very large, and its participation in the oena) basioccipital. The elongated fossa observed on the lateral palate is much more extensive than in any other odontocete. On edge of the palatine, therefore, probably corresponds to a the palate the vomer has the characteristic lanceolate shape ob­ .trong attachment of the internal pterygoid muscle, which also served in those odontocetes in which the vomer participates in ltaches on the pterygoid hamulus and on the medial lamina of the formation of the palate. It is long and occupies approxi­ the pterygoid (see below). mately two-thirds of the midline of the palate. Its maximum Although the palatine is partially broken in its dorsal portion, width is at the level of the anteriormost point of the palatine­ it clearly had an important contact with the frontal. Ventral to premaxillary suture; its anterior half contacts the premaxilla, the optic gutter of the orbitosphenoid, the dorsal edge ofthe pa­ and its posterior half contacts the palatine. The posterior part of latine shows a small curvature (concave dorsally), which we re­ the bone, in the basicranial basin, is relatively narrow but much gard herein as the medial part of the lateral lamina that was thicker than in the other delphinoids. covering the optic gutter and contacting the frontal dorsally. Orbitosphenoid: The orbitosphenoid has a long and wide T'osterodorsally, the palatine also contacts the anterior edge of optic canal, whose anterior opening is situated below the poste­ ,:e. alisphenoid. rior edge of the supraorbital process and above the inflexion Ptel}'goid: The pterygoid is excavated bya relatively small point of the sigmoid lateral border of the palate. It has an al­ pterygoid sinus and, therefore, is divided into lateral and me­ most anteroposterior orientation and faces laterally, whereas in dial laminae. On the palate, its suture with the palatine is an an­ the other odontocetes it faces ventrally and forms an angle of teroposterior line observable at the posterolateral corner ofthe approximately 45° with the anteroposterior axis of the skull. A palate. The suture between the two bones on the palate is actu­ large optic foramen opens in the optic canal just below the nar~ ally a horizontal plane, as the large backward expansion of the rowest part of the temporal fossa; lateral to it, the large sphe­ palatine covers the pterygoid posteriorly. Because of this mor­ norbital fissure (anterior lacerate foramen) is walled lateraIIy '1hology of the palatine, the pterygoids are laterally displaced by the lateral laminae of the palatine and pterygoid. The mor­ ld widely separated, and the palatine must have formed the major part of the posterior edge ofthe bony palate. Widely sep­ phology of this region of the skull is related to the strong ante­ arated pterygoids are known in the Holocene Monodontidae rior displacement of the nares and the orbits, which have and Phocoenidae and in some living Delphinidae. dragged the optic canal anteriorly (above the palatine-ptery­ The pterygoid hamulus is only partially excavated by the goid suture); in most delphinids and phocoenids, the optic ca­ pterygoid sinus and possesses a large ventrolateral crest, proba­ nal lies posterior to the lateral lamina of the pterygoid. It is bly for the insertion of the pars lateralis of the pterygoid mus­ noteworthy, however, that a condition intermediate between cle. The outline of the crest is strongly convex in ventral view, that of Odobenocetops and that of most of the other Delphi­ :n contrast to the markedly concave lateral border of the pa- noidea is observed in Delphinapterus, in which the orbits are tine anteriorly. The anteriormost extremity of the palatine­ located fairly far anteriorly. pterygoid suture is on the lateral edge of the palate at the in­ Alisphenoid: The alisphenoid is a very thick bone located flexion point of the curve. just anterior to the large squamosal gutter. It contacts the fron­ Although the apex ofthe pterygoid hamulus is broken on the tal dorsally, the squamosal and parietal posteriorly, and the specimen, it is probable that only a small part is missing. The pterygoid ventrally. On its ventral edge can be observed the medial lamina of the pterygoid, posterior to the hamulus, is dorsal border ofan enormous foramen ovale. This foramen was strongly thickened relative to other odontocetes. Its lateral edge not closed and was confluent with the cranial hiatus of the au­ shows a sort of semicylindrical crest, which delimits a semicir­ ditory region, contrary to what is observed in the other delphi­ cular cupule. Probably an extremely strong pterygoid muscle noids. Apart from this feature, the major characteristic of the 'iginated partly on this structure. On its medial side and ven­ alisphenoid is its thickness, which, among the delphinoids, also tral to this muscle attachment is an anteroposteriorly directed is found in the Monodontidae (Muizon, 1988b). 234 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Basioccipital: The basioccipital is almost totally broken well developed and may have extended onto most ofthe medial away, and only the anterior portion of the right side is pre­ wall of the squamosal gutter. At the posteromedial corner of served. On its ventrolateral side, a small and shallow gutter the bone, on the suture with the occipital, there is no articula­ very probably represents the passage for the internal carotid. tion area for the periotic but rather some squamous and spongy The carotid foramen was, therefore, confluent with the foramen bone for attachment of the ligament that, as in the other Del­ ovale and with the cranial hiatus. On the dorsal side of the ba­ phinida, held the periotic in place. sioccipital, the carotid gutter reaches the lateral edge of the Behind the saddle-shaped posterior extremity ofthe squamo­ sella turcica, as in the other delphinoids. The carotid gutter has sal gutter, and squeezed between it and the exoccipital, the pas­ a strong anteroposterior orientation, however, and the internal sage for the external auditory meatus is long but very reduced; carotid entered the skull at a point well posterior to the sella it opens dorsolaterally in the ventral part of the very small ster­ turcica. In the other delphinoids the carotid foramen is located nomastoid fossa. The latter is reduced and occupies the poster­ approximately at the same level as the sella turcica. odorsal corner of the zygomatic process (or of the lateral wall Squamosal: The squamosal is a relatively small bone of the squamosal gutter). It is limited dorsally by a small crest, whose entire morphology is conditioned by a wide, deep, and which is much more reduced than its homologue in other del­ anteroposteriorly elongate gutter, which is formed mainly by phinoids. The lateral side ofthe zygomatic process is wide and the highly modified zygomatic process. The gutter is open an­ elongated anteroposteriorly. In other delphinoids this surface teriorly and posteriorly and is approximately cylindrical. It cor­ gives rise to the masseter muscle; therefore, the condition ob­ responds in other delphinoids to the depression observed be­ served in Odobenocetops suggests a strong masseter (for a ce­ tween the lateral wall of the zygomatic process and the crest tacean). that joins the falcate process to the posteromedial origin of the On the lateral wall of the braincase, the suture of the squa­ fossa for the middle sinus. The lateral edge of the squamosal mosal with the parietal is very unusual. In most other mam­ gutter of Odobenocetops is very thick, and its anterior extrem­ mals, and certainly in all the other odontocetes, the squamosal­ ity represents the greatly shortened zygomatic process. It is parietal suture is mainly a subvertical plane facing laterally, likely that the glenoid cavity did not occupy the whole volume and the part of the squamosal contacting the parietal is a squa­ of the gutter. A comparison with the position of the glenoid mous lamina, which gives its name to the bone. In Odobenoce­ fossa in a beluga or narwhal shows that in Odobenocetops it tops, the articulation is a rough zigzag line, and the joint sur­ probably occupied the posterolateral quarter of the squamosal face faces dorsally. This articulation, together with the general gutter. The articulation ofthe condyle was therefore on the pos­ stoutness of the squamosal, suggests that large muscular ten­ terior half of the medial side of the lateral wall of the gutter. sions were applied to the bone. In the posterolateral region of The posterior region ofthe gutter bears a posterodorsally extro­ the squamosal, just anterior to the paroccipital process, the en­ verted saddle-like crest, which strongly suggests the presence, tire region of the external auditory meatus is tremendously anterior to it, of a very mobile articulation. The prominence of thickened, further exemplifying the exceptional stoutness of this saddle-shaped crest also indicates the mature development this bone in Odobenocetops peruvianus. of the articulation, in agreement with the extensive fusion of Periotic: One isolated left periotic (MNHN SAS 1614) and the bones ofthe skull, and indicates a relatively old animal. It is one partial tympanic associated with another left periot;:: therefore likely that the condyle articulated only with the poste­ (MNHN SAS 1613) are referred to Odobenocetops peruvianus rolateral extremity of the squamosal gutter. (Figures 9, 10). The taxonomic assignment ofthese isolated el­ The squamosal gutter (and consequently the glenoid fossa) is ements is clearly confirmed by the recent discovery of a partial located in a much higher position than in the other delphinoids. skeleton of Odobenocetops sp. (MNHN SAO 202) that in­ It is well above the lateral wall ofthe cranial hiatus, and its bot­ cludes, associated in situ with the skull, a tympanic and periotic tom is approximately at the same level as the posterior edge of like those described herein. In the following description the the postorbital process of the frontal. In the other delphinoids, tympanic side ofthe periotic will be called ventral; the cerebral the glenoid cavity is at the same level as the lateral wall of the side, dorsal; the cochlear promontory side, medial; and the side cranial hiatus and well below the posterior edge ofthe postor­ opposite to the cochlear side, lateral. The nomenclature of the bital process of the frontal. external anatomy of odontocete periotics follows Fordyce The ventral edge ofthe medial wall of the gutter is rounded, (1994). Both periotics are morphologically as unusual as the very thick, and dense, and it fonns the lateral edge ofthe enor­ skull and do not closely resemble those of any other cetacean. mous cranial hiatus; it is related anteriorly to the extremely The periotic of Odobenocetops is a large and stout bone whose thickened medial edge of the alisphenoid. Immediately behind· proportions are in agreement with the very large cranial hiatus the suture \vith the alisphenoid, on the thickened medial wall of of the skull. It has a large, long, and robust anterior process the squamosal gutter, a small crest probably represents the very with a blunt rounded apex. On the ventral side of the anterior reduced falciform process of the squamosal. At the posterior process, a large lateral tuberosity is located lateral to the m:+ extremity ofthe medial wall ofthe squamosal is the medial ex­ lear fossa. From the tuberosity, a wide and flat ventral rim c.' tremity ofthe middle sinus fossa; the sinus was probably fairly tends anteriorly as far as .the apex of the anterior process. It is F

",UMBER lJ3 235

b 1 em

FIGURE I O.-OdobelloCl'lloc

tion is almost medial. The epitubarian fossa receives the pro­ wider and flatter than in other delphinoids, especially in its an- cessus tubarius of the tympanic, also called the accessory ossi­ 'ior portion. This rim bears numerous fine parallel wrinkles, cle, unciform process, or uncinate process. When compared which are slightly concave anteriorly and oblique to the axis of with that in other delphinoids, the epitubarian fossa is small the bone. This ~ti'llcture, although flattened, cOITesponds to the relative to the size ofthe bone, but it is more concave. The third "bourrelet ventr,l1" defined by Muizon (l988b), which is a syn­ and most anterior fossa is the bullar facet, as defined by apomorphy of th,' Delphinida. Fordyce (1994), and erroneously termed epitubarianfossa by Three small fL"'';IC' are observed on the ventral face ofthe an­ Muizon (1987, 1988a, 1988b, 1988c, 1991). The bullar facet is terior process, I, .. iill to the ventral rim. They are, from rear to a shallow fossa, slightly longer than it is wide, which occupies froll" 'he malle,,' ;~)ssa (or fossa capitis mallei), the epitubarian approximately the anterior one-third of the ventral face of the f ,:;;t, and the DU i!ilr facet. The mallear fossa receives the head process. ,11'.; malleus: it is oriented more ventrally than in the other The bullar facet is surprisingly well developed for a delphi­ Delphinoidea, especially the Monodontidae where the orienta- noid; however, it is noteworthy that a small bullar facet also is 236 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

Jbserved in some Kentriodontidae (Muizon, 1988b: 172), such common in Squalodon and in the Eurhinodelphidae (Muizon, as Liolithax. The bullar facet is a common structure in several 1988c, 1991). In MNHN SAS 1614, the fenestra rotunda is groups of odontocetes (Squalodontidae, Squalodelphidae, Pla­ wider than in MNHN SAS 1613, a condition that could be tanistidae, Waipatiidae, Ziphiidae, Physeteridae, and Li­ pathologic or related to erosion. potidae). The loss ofthe bullar facet (=epitubarian fossa sensu The pars cochlearis is extremely small considering the large Muizon, 1987, 1988a, 1988b, 1988c, 1991) has been consid­ size of the periotic. It is very low and somewhat recalls that of ered a synapomorphy of the monophyletic clade Delphinoidea Squalodon, although it is not as compressed dorsoventrally in + Inioidea (Muizon, 1988b). This clade is defined by three medial view, is not ventrally shifted, and has a rounded overall other synapomorphies (Muizon, 1988b: 165), so it is possible shape in ventral view. It differs strongly from the much higher that loss ofthe bullar facet occurred several times in the evolu­ profile (in dorsal or ventral view) pars cochlearis of other Del­ tion of the Delphinida. It also is possible that the presence ofa phinoidea and especially that of the Monodontidae. The poste­ ·bullar facet in Odobenocetops is related to the increase in rior process is relatively slender and bears a narrow, anteropos­ length and thickness of the anterior process of the periotic, teriorly elongate articular facet for the tympanic. Lateral to the which is as highly specialized as the rest ofthe skull. tympanic facet is a conspicuous anteroposteriorly oriented In dorsal view, the anterior process shows several slits, fo­ crest that bordered the external auditory meatus dorsally. ramina, and fissures, as is often the case in the Monodontidae. On the dorsal face of the bone is the internal auditory win­ The fossa incudis (or fossa crus breve incudis) is located just dow, composed ofthe tractus spiralis foraminosus (or fundus of posterior to the mallear fossa and anterior to the anterior ex­ the internal auditory meatus) (posteromedially), the foramen tremity ofthe tympanic articular facet of the posterior process. singulare, and the dorsal aperture of the facial canal for the fa­ It is small and shallow, and it faces more anteroventrally than cial nerve (anterolaterally). The window is large but relatively in other delphinoids. smaller than in the other Delphinoidea. It is approximately ~he Anterolateral to the fossa incudis is the ventral (or tympanic) size of that of Delphinapterus. although the periotic of Odo­ opening of the facial canal, which transmits the facial nerve benocetops is twice as large. In particular, the tractus spiralis (VII). This foramen is small, approximating the size ofthat ob­ foraminosus is smaller than that ofmost large delphinoids. The served in Delphinapterus, although the periotic of Odbbenoce­ window in Odobenocetops. however, differs from that of the tops is almost twice the size of that of Delphinapterus. Be­ other delphinoids in being much deeper, probably because of tween the mallear fossa, the fossa incudis, and the ventral the thicker (in medial view) pars cochlearis. The same can be opening ofthe facial canal is a small elongate fossa that faces said concerning the aperture for the endolymphatic duct (dorsal posteriorly. Such a fossa is not present in all Delphinoidea; opening ofthe vestibular aqueduct), which is a large conical pit among the Holocene species, we have observed it in Pseud­ posterolateral to the window. It is larger than that of Delphi­ orca. In some periotics ofan undescribed fossil delphinapterid napterus but resembles it in being widely open. The aperture and of a fossil kogiine (Lua and Marsh, 1996) from Lee Creek for the cochlear aqueduct is a small pit facing dorsally, antero­ Mine (North Carolina), a very small pit is observed ventrolat­ medial to the aperture for the endolymphatic duct. The dorsal eral to the ventral opening ofthe facial canal. face ofthe periotic has an irregular aspect and presents numer­ . The fossa for the stapedial muscle is large but shallow and ous exostoses made ofdense bone; it does not show the smooth faces almost ventrally. It differs in this respect from the condi­ and regular morphology generally observed in the other Dei­ tion in Delphinapterus and most other odontocetes, where the phinoidea. In this respect, it resembles the periotics of Physe­ fossa faces laterally or ventrolaterally. The groove for the facial teridae. In lateral view, the periotic of Odobenocetops shows a nerve is not well separated from the stapedial muscle fossa as is very large and relatively flat surface. generally observed in other Delphinoidea. Posteromedial to the ventral opening of the facial canal is the fenestra ovalis. Con­ CT SCANS OF THE PERIOTICS trary to what is observed in Delphinapterus. Monodon, and generally in the Delphinoidea, where the fenestra ovalis is METHODS.-The two periotics of Odobenocetops peruvi­ oval-shaped, in Odobenocetops it is almost circular, and the anus were scanned with an ultra-high resolution protocol de­ stapes (preserved in MNHN SAS 1613) is a small conical yeloped for analysis of fossil material. The technique provides bone, inflated at its ventral extremity for articulation with the fine-level differentiation of mineralization characteristics, incus (Figure 12). It is not flattened as in most other odonto­ which in tum provides good delineation of inner-ear structures cetes. inCT images. Simple surface reconstruction algorithms were Posteromedial to the fenestra ovalis is the fenestra rotunda, used to obtain three-dimensional reconstructions ofthe periotic which is clearly reniform in MNHN SAS 1613. The posterolat­ shell. Segmentation algorithms were used to visualize particu­ eral wall ofthe fenestra rotunda is thickened and shows a con­ lar features, e.g., the cochlear canal, vestibular aqueduct, facial spicuous elevation responsible for its oval shape. Thatmor­ nerve, and auditory nerve. In live ears, these algorithms utilize phology is not observed in the other Delphinoidea, but it is the attenuation coefficients for cochlear fluids and neural tis ,UMBER 93 237

FIGURE II.-Odobellocelops peruviullus, left periotic (referred specimen, MNHN SAS 1613): reconstruction of inner ear spaces from CT data. (Scalc bar= I cm; a=apex of¢o¢hlea; an=auditory nerve canal; b",basal tum; ca= cochlcar aqueduct; fn=facial ncrve canal; oosl=outer osseollsspirallaminar ridge; va=vestibu1ar aqueduct.) sues to produce three-dimensional representations of inner-ear by species at the lower end of the type II group of odontocete anatomy; in fossils, the differences in X-ray attenuations of ears, which have a peak frequency of35-50 kHz. ."diments or mineralization of bone versus neural and inner ear Although Odobenocetops, judging from its cochlear configu­ areas are used for the reconstructions. ration, could (like most mammals) perceive frequencies above RESULTS.-Inner Ear: Both specimens are left periotics 20 kHz, it is important to note that having ultrasonic hearing with well-preserved inner ears. Small di fferences in cochlear abilities is not synonymous with echolocation.Echolocation dimensions and turn number (Table 2) between these speci­ per se requires not only perception but also synchronized pro­ mens are within the range of normal interindividual variability duction ofultrasonic signals coupled with the ability to analyze for modem odontocete inner ears (Kellen and Wartzok, 1990). target features from the retuming echoes. The cochleae have approximately 2.5 turns with the three clas­ A notable deviation from typical extant odontocete inner ear configuration is the presence of well-defined semicircular ca­ sic hallmarks ofodontocete cochleae: a ventrolateral apical ori­ nals (Figure 12), a large, reniform vestibular aqueduct, and, .ntation, a substantial outer osseous lamina in the basal turn, judging from the diameter of the residual VIIIth nerve canal and an exceptionally large cochlear aqueduct (Figure II). The (Table 3), relatively large vestibular and facial nerve fiber cochlear canal follows a conventional odontocete type II for­ counts (Figure 13). Extant Odontoceti have vestibular volumes mat (Ketten, 1984; Ketten and Wartzok, 1990), with closest that are substantially less than their cochlear canal volumes. In morphometric affinities to De!phinapterus teucas and Mon­ odon monoceros. The three-dimensional cochlear canal length for each specimenis approximately 44.4 mm, implying an ani­ TABLE I.-Measurements (in cm) of holotype of Odobellocelopsperuviallus mal length of 360-400 em, assuming that the ratio of canal (USNM 488252).

'::ngthtobody mass follows the same allometry as extant odon­ Total length of skull from anterioml0st extremity ofpre­ 45.7 IOcetes (Table I). A substantial outer osseous lamina (as indi­ maxillae to posterior border ofoccipital condyle cated by the laminar indentation in Figure II) is present Width of postorbital constriction 11.6 throughout the first 16 mm of the lower basal tum in both spec­ Width oforbit from apex ofantorbital process to apex of 10.5 postorbital process imens and is.a ,clear indication of some ultrasonic hearing in Length of right premaxillary sheath from its apex to an- 31.7 Odobenocetops. The outer lamina covers approximately 35% tero-dorsal hump ofpremaxi lIa of the (:ochlear length. again consistent with a type II odonto­ Depth of right ah colus 30.7 Bizygomatic width lestimaled) 15.4 x 2=30.8 cete ce"chlear format. Cochleae of this type have maximum Anteroposterior length of braincase 18.3 peak ultrasonic sensiti\ities below 80 kHz. In fact. the cochlear Width of braincase (estimated) 10.2 x 2=20.4 .)rofile of these Odohenocetops cochleae is best approximated Length ofmandibular gutter 10.2 238 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE I 2.~Odvh"nvce/()ps peruvill/llIs, left periotic (referred spccimcn. MNHN SAS 16 (3): rcconstruction of inllcr ear spaces from CT data. (Scalc bar= 1 cm; a=apcx ofcochIca; an=auditory Ilcrve canal; b=basal tum; s= hcad ofstapcs; ssc=scmicircular canal; va=vcstibular aqucduct: yn=vestibular nerve bundlc.) many species, the semicircular canals are sufficiently small and \'elopment is underscored by the tiber distributions of cetacean thread-like that the complete canal system cannot be traced in acoustovestibular nerves. Recent data show that the majority of periodic histologic sections, much less detected by high-resolu­ odontocetes have 3%-5'10 of the total of VIIIth nerve fibers de" tion CT (Gray, 1951; Ketten, 1992). This contrasts sharply with \'oted to vestibular components, as compared with an average the condition in land mammals, in which vestibular canal vol­ of 45% in land mammals-percentage ranges from about 20% umes rival cochlear volumes and are readily imaged by even in bats to about 60% in brachiating primates (Gao and Zhou, conventional CT (Gray, 1951; Spoor, 1993). Although size is 1995). not a criterion for \'estibular function, the majority ofcetaceans The three readily imaged semicircular canals and the well­ appear to have uniquely small semicircular canals that are sig­ defined vestibule of Odobenocelops are similar in dimensions nificantly shorter and narrower than their cochlear canals (Gray to those of the beluga (Delphinaplerlls leucas) (Figure 12). 1951; Jansen and Jansen, 1969). This anomalous vestibular de- Delphinaplerlls leucas has the highest semicircular canal ratio ofany extant odontocete studied. In all mammals, cochlear size is best correlated with body mass, but the exact correlates of TABLE 2.~Cochlear morphometry of cetaceans. Measurement are in millime- the semicircular-canal size in most mammals remain unclear. ters. (e=estimated.) For primates, a positive relationship to locomotory behavior Scalae Basal Axial Body Species appears relatively robust (Spoor, 1993). It has been suggested Tums length diameter height length that fusion of the cervical vertebrae in Cetacea resuitedinllm­ Delphinapterlls lellcas 1.75 4\.99 10 4.21 325 ited head movements and substantially fewer inputs to the ves­ Grampus grisells 2.5 40.5 8.73 5.35 228 tibular system, leading to a loss of related receptors (Ketten, Lagenorhynchlls albirvslris 2.5 34.9 8.74 5.28 207 l\-fonodon monoceros 2.5 45.12 I\.6 5.3 425 1992). Better-developed semicircular canals in D. teucas are Phocoena phocoena \.5 25.93 5.25 \.47 133 consistent with this species ha\'ing a well-defined neck and Physeter calOdon 1.75 72.21 14.3 3.12 1361 greater rotational and lateral head movements (a full right an­ Stenella allenllala 2.5 36.9 8.61 4.36 185 gie) than other living odontocetes. The diameters of the audi­ Tlirsiops trllncalus 2.25 40.65 9.45 5.03 259 tory and vestibular canals in Delphinapterus, Monodon. and Odobenocetops peru\'iallus 2.5 44.37 10.1 4.31 380e (MNHN SAS 1613) Odobenocelops are very close but not substantially different Odobenocetops peru\'ill/Ills 2.25 44.41 10 4.5 380e from those of other odontocetes. Cross~sectional areas of the (MNHN SAS 1614) canals for the vestibular component versus the auditory compo- NUMBER 93 239

FIGURE 13.-0dohellocetol's perul'ialllls. left periotic (referred specimen, MNHN SAS 1613): reconstruction of inner ear spaces frolll CT data. (Scale bar= I em; abbreviations as in Figure II.) nentof the VIWhnerve are somewhat higher for both De/phi­ tops also may have had substantially greater head motions than naptr;:rus.(7%) and Odobenocctops (6%-9%) than for most del­ most modern cetaceans. This motility could be related to the phinids. Assuming that the diameters of neural fibers benthic feeding adaptations proposed herein. occupying those cross sections were equivalent in extinct and The significance of an enlarged vestibular aqueduct and pre­ extant species, we speculate that OdohenoCl!tops had a vestibu­ sumably equally large endolymphatic duct is unclear. Large lar-ganglion cell population between 4500 and 7000 neurons, renifonn aqueducts are found in some baleen whales, e.g., Ba/­ with an average auditory ganglion population ofapproximately aena mysticetlls. No detailed comparative studies of the aque­ 75,000 to 80,000 There are, however, a great many caveats to duct across a variety of species are currently available for this speculation, chief among them being that the available da­ aquatic mammals, however. tabase is too small for explicit conclusions about exact neu­ Equally striking is the relative size of the facial nerve canal ronal levels. At this point, the most appropriate interpretation is in Odobenocetops compared with that of most odontocetes. that the exceptionally close morphometric resemblance be­ With the exception of Monorion monoceros, the facial nerve is tween the De/phinaptcrlls and Odobenocetops vestibular sys­ typically less than 1.5 mm in diameter. The facial nerve of the tems is striking and consistent with the idea that Odobenoce- specimen of M monoceros examined for this study, an adult

TABLE 3.-Neural canal morphometry ofcetaceans. Measurements are in millimeters.

Vestibular Facial Internal auditory Auditory Species nerve diameter nerve diameter canal diameter nerve diameter Delplzillapterus lellcas 1.20 1.47 5.6 4.4 Gramplls grisells 4.7 J.7 LagellOrlz.\·llclzlls albirostris 4.6 3.6 ;\tollodoll mOlloceros 1.50 3.04 6.22 4.92 Pizocoella plzocoellG 0.52 1.09 3.7 2.78 Plzyseter calOdoll 11.3 Stelle/la allellllata 4.9 Tursiops /rlIllCaiIiS 1.00 1.41 5.6 4.5 Odobellocetops perm'imws 1.15 2.7 5 3.85 (i\lNHN SAS 161:l) Odohcllocet0l's peru\'iolllls 1.00 2.5 5 4.0 (i\l'\HN SAS 1614) 240 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ri stapes footplate and is more ventrally directed in this specimen than in other odontocetes. Tympanic: The tympanic is incomplete, as only the involu­ crum and a small portion ofthe ventral side are preserved (Fig­ ure 14). It shows, however, a typical feature of the Delphinida (Muizon, 1988b), namely the sigmoid morphology of the in­ volucrum in medial view. This feature is very clear in the Li­ potidae, Inioidea, and Delphinoidea. It is absent in the Platanis­ toidea (sensu Muizon 1987, 1991) and in the Physeterida. The involucrum is much more robust and stout in Odobenocetops than in any other Delphinoidea. It is not dorsoventrally flat­ tened as in the Holocene Delphinoidea, even in the most robust FIGURE 14.-0dobenoce/ops perul'ianll.l'. left tympanic (referred specimen, forms such as Orcinus. In some Kentriodontidae, such as Ato­ MNHN 1613): a, medial view; b, ventral view. cetus, the involucrum is relatively robust but much smaller. Occipital: The occipital is relatively lower than in other male, measured 3.1 mm, or approximately double the odonto­ Delphinoidea but is very wide andconvex (Figure 15). It has cete average. The Odobenocetops facial nerve canals averaged undergone some slight postmortem deformation, but this ap­ 2.6 mm at mid-periotic level. Once again, it is not possible to parently had little effect on the overall shape of the bone. The draw definitive neural conclusions from the anatomy ofone or portion of the bone lateral to the occipital condyle is much two animals; however, it is noteworthy that in each of these wider and lower than in the Delphinidaeand Phocoenidae and species the tusks coincide with an unusually large volume of is strongly expanded laterally. It somewhat resembles the con­ facial nerves. dition in the Monodontidae, where the posterior crest of the Middle Ear: Little can be said abollt the middle ear be­ temporal fossa (the lambdoid crest) is oblique and not vertical. cause ofthe lack ofcomplete tympanies. One stapes remains in In Odobenocetops the lambdoid crest is assimilated into the situ. It is monocrural with a large head and footplate, a configu­ nuchal crest as a consequence ofthe strong modification of the ration common among both Odontoceti and Sirenia. The distal temporal fossa. The paroccipital process, although partially facet has a surface area that is large in comparison with the broken, seems to have been very well developed, robust, and

FIGURE 15.-0dohenocelops penll'ianlls, holotype (USNM 488252): occipital view ofthe skull. NUMBER 93 241 markedly expanded laterally. Its anterior face very probably re­ napterlls the distance between the exits of the optic nerves is ceived a posterior sinus. It is thickened and fused to the poste­ generally two to three times greater in absolute value than in rior thickened region of the squamosal. The very stout mor­ Odobenocetops. phology of this squamoso-paroccipital angle of the skull, as Posterolateral to the optic canal are the sphenorbital fissure well as the robustness of the zygomatic process ofthe squamo­ and the foramen rotundum. They are large and the latter is su­ ,ai, seems to indicate very strong musculature inserted on this perposed to the former. They are separated by a bony wall (par­ cgion ofthe skull: the scalenus ventralis and scalenus dorsalis, tially broken) where they exit the skull. Among the Delphi­ which insert on the anteroventral and posteroventral regions of noidea this morphology is present in Monodon. In the the paroccipital process respectively, and possibly the costo­ Delphinidae and Phocoenidae, the sphenorbital fissure and the humeralis and sternomastoideus, which insert in part on the foramen rotundum are most commonly confluent with the optic posterolateral border ofthe process. In its dorsal region, the oc­ foramen. In Delphinapterlls the optic foramen is isolated, but cipital crest does not seem to have been more prominent than in the sphenorbital fissure and foramen rotundum are confluent. other delphinoids, and the muscle insertions are not as marked The sphenorbital fissure of Odobenocetops is much larger as in some Monodontidae (in some males of Monodon). The than in Delphinapterlls and approaches the size observed in the occipitil1 shield is strongly convex, much more so than in other much larger Monodon, although in that genus the foramen'ro­ Delphinoidea. It differs from the somewhat flattened condition tundum is smaller than in Odobenocetops. The sphenorbital fis­ observed in the Monodontidae and is closer to the more convex sure is the passage for the nerves and blood vessels that provide condition ofthe Delphinidae and Phocoenidae. innervation and blood supply to the anterior part ofthe skull, in­ The right condyle only is preserved, and that not totally. It is cluding the oculomotor nerve (III), trochlear nerve (IV), oph­ much more convex and salient than in any Holocene delphi­ thalmic branch of the trigeminal nerve (V1), abducens nerve noidand adjoins a very deep supracondylar fossa. These struc­ (VI), anastomotic artery, and cavemous sinus. The foramen ro­ tures are much more developed than in Delphinapterlls. which tundum is the passage for the maxillary branch ofthe trigeminal is well known for having a very well-defined and flexible neck nerve (V2). The large size ofthese apertures in the skull ofOdo­ for a cetacean. Consequently, the morphology of the occipital benocetops is certainly related to the major modifications ofthe ,,:ondyle of Odobenocetops obviously indicates that it was ca­ rostrum (large mobile eyes, premaxillary tusks, and inferred pable of very ample movements of its head, which was cer­ large upper lip), which would have required extensive blood tainly much more mobile and flexible than in Delphinapterlls. and nerve supplies. The very large sphenorbital fissure in the Endocranial Cavity: The endocranial cavity is widely walrus and the narwhal is very probably related to the growth of opened, as major parts of the posterolateral and ventrolateral the tusks (and to the large upper lip, in the case ofthe walrus). regions ofthe left side of the skull are missing, allowing obser­ The remainder of the internal view of the braincase shows vation of the intemal morphology of the braincase (Figure 16). mainly the cavities for the cerebral hemispher~s, which appear Its major characteristic is the presence ofethmoidal fossae for a to be proportionally shorter and wider than in the other Delphi­ pair of small olfactory bulbs, separated by a small bony wall. noidea. Immediately dorsal to the lateral border of the foramen The fossae are horizontal, slightly higher than they are wide, ovale, on the internal side of the braincase, is a longitudinal and approximately 4 mm wide, 6 mm high, and 8 mm deep. groove that seems to reach the foramen rotundum; it is absent Such structures are very uncommon in the living odontocetes from the other Delphinoidea, and its function is not clear. but have been mentioned in the Delphinidae (Sinclair, 1966) and Eoplatanistidae (sensu Pilleri and Ghir, 1981, 1982; COMPARISONS AND AFFINITIES Muizon, 1988c), and Muizon observed them in the Eurhinodel­ phidae, in the Platanistidae (cf. Pomatodelphis sp., USNM Comparing Odobenocetops perllvianlls with other taxa is not 214759), and in the Squalodontidae (Sqllalodon tiedemani). easy, as most of the typical odontocete characters have been Kellogg (1926) observed, in the posterior wall of the bony strongly modified. As already stated (Muizon, 1993a, 1993b), nares of the platanistid Zarhachis flagellator. two crescentic fo­ Odobenocetops is a delphinoid odontocete cetacean. Within the ramina that represent the exits of the olfactory nerves. Further­ Delphinoidea, comparisons are not very productive because more, Oelschlager and Buhl (l985a, 1985b) noted, in early on­ Odobenocetops differs from the other Delphinoidea in almost togenetic stages of Phocoena phocoena. an olfactory bulb that all the features observed. The exceptional morphology of the becomes reduced in later growth stages. cetacean described herein, however, requires a statement of The groove for the optic chiasma is located just below the which characters allow a precise taxonomic definition of Odo­ ethmoidal fossae, and the optic nerves exited the skull at the benocetops. Five major features allow assignment ofOdobeno­ very front of the brain, as in the living Monodontidae but un­ cetops to the Cetacea: like the condition in the other odontocetes. The exits of the op­ 1. The presence of large air sinuses in the auditory region, tic nerves are located more medially (much closer to the mid­ .connected to well-developed pterygoid sinuses. All cetaceans sagittal plane of the skull) and the chiasmatic groove is much have a peribullary sinus, but pterygoid sinuses are not found in narrower than in the other delphinoids, however. In Delphi- all members ofthe order. Although they ar~ absent in Pakicetlls 242 S\lITHSO\IA\ CO'iTRI13UTIO\S TO PALEOBIOLOCY

FIGURE 16.-Endocranial cast ofthe holotype of Od"hL'lIoCL'lOpS l'L'rul"iWllIS (USNM 488252): a. dorsal vic\\'; h. anterior yic\\': c. posterior yiew; d. lateral yiew. and in Protocetlls, they are present in all other cetaceans. Air condition of the bony nares is significantly different from that sinuses surrounding the auditory region and invading the ptery­ in the cetaceans; the nares open anteriorly in a wide narial ba­ goid are not known in any other mammals. A pterygoid dupli­ sin that opens dorsally and is located in the anterodorsal region cated into two laminae also is found in the Erinaceidae, the of the skull. In the proboscideans, the narial fossae are not lo­ Macroscelidae, and the Tupaiidae. In these families, however, cated at the anterior end of the skull but they open anteriorly. A the fossa is always opened anteriorly, unlike what is observed condition convergent with that of the cetaceans is found. how­ in all cetaceans. and there are no peribullary sinuses. ever, in Macrauchenia, a Pleistocene South American lito­ 2. The large supraorbital process of the frontal, which ptern. The cetacean condition is not found in any carnivores. widely overhangs the orbital region. This feature is present in 4. The absence of a true cribriform plate. This structure is all adeq\wrdy known cetaceans; hO\vever, it has not been ob­ located in the mesethmoid, at the anterior region of the cranial served in ['ukicetlls because that part of the skull is unknown in cavity, and gives passage to the olfactory nerves. The olfactory this genus. nerves have not totally disappeared in the cetaceans, however: 3. The narial fossae of Odobenocetops, which open dorsally small olfactory bulbs have been observed in several cetacean:>, and are not located at the apex of the skull. In the Sirenia the and Odobenocetops still retains small ethmoidal fossae (see NUMBER 93 243 above). The loss of the cribriform plate is obviously related to the aquatic life habits of the cetaceans. A well-developed crib­ riform plate is present in all the aquatic carnivores, but it is re­ duced in the Sirenia. 5. An immobile elbow. Although archaeocetes do not have an immobile elbow, this feature is constant in mysticetes and odontocetes. The partial forelimb illustrated in Figure 17 was found associated with a partial skull of Odobenocetops (as mentioned above) and shows the characteristic immobility of the elbow seen in the modern cetacean forelimb. Several other features allow us to classify Odobenocetops among the odontocetes, although the hyperspecialization ofthe skull partially hides the key character of the suborder. It has been noted elsewhere (Muizon, 1994) that the odontocetes are diagnosed by a posterior projection of the maxilla that covers the supraorbital process of the frontal totally or partially. In ()dobenocetops, the maxillae have withdrawn medially and the supraorbital process is almost totally uncovered dorsally. Nonetheless, this key character of the odontocetes is still ob­ servable in Odobenocetops. The following features allow the assignment ofOdobenocetops to the odontocetes. I. Although strongly withdrawn, the maxillae still cover part of the medial part of the supraorbital processes medially, are expanded far backwards behind the nares,and are in contact with each other medially in that region. This condition is found .:either in the Archaeoceti nor in the Mysticeti. 2. The fossa for the pterygoid sinus of Odobenocetops is greatly expanded dorsoventrally, and its dorsal limit is dorsal to the floor of the braincase. This condition is never found in any Archaeoceti or Mysticeti but is common in the odontocetes. In some very early taxa, such as agorophiids and Waipatia. the pterygoid fossa is not as developed as in the Holocene forms, but it is still clearly more developed than in the archaeocetes '!ld mysticetes. 3. The premaxillary foramina are large., The premaxillary foramen of odontocetes is a supplementary passage for branches of the external carotid artery and for the maxillary di­ vision of the trigeminal nerve. This need for supplementary blood supply and innervation is related to the major specializa­ tions of the odontocete head, namely the melon and the air sac system. Enlarged premaxillary foramina are observed neither in the archaeocetes nor in the mysticetes. 4. The bones of the skull in the facial region are asymmetri- especially the premaxillae and the maxillae; the right bone is always more developed than the left. Although this asymme­ try is lacking or extremely reduced in several fossil groups FIGURE 17.-0dob

the Delphinida is the protrusion of the pterygoid sinus within and the temporal fossa is greatly elongated anteroposteriorly. the palatine, which is thereby divided into a medial and a lat­ The morphology of Odobenocetops is obviously a much more eral lamina. Such a structure is present in all eight families of derived stage of the condition observed in the Monodontidae. the group (Lipotidae, Iniidae, Pontoporiidae, Kentriodontidae, The Monodontidae also were diagnosed by two additional Albireonidae, Monodontidae, Phocoenidae, and Delphinidae). features: the extension of the maxilla-premaxilla suture on the The Delphinida also share the presence of a cranial hiatus that lateral side of the nares (in the other Delphinoidea, the suture greatly enlarges the posterior lacerate foramen, which in turn is always remains on the anterior edge of the nares) and the re­ coalescent with the median lacerate foramen; in the Delphinida duction of the lateral lamina of the hamular process of the the periotic hangs in the hiatus, because it is connected to the 'pterygoid (all the other delphinoids have well-developed lateral skull by ligaments only. As mentioned above, Odobenocetops laminae) (Muizon, pel's. obs.). These characters are not ob­ has a very large cranial hiatus. Furthermore, ifthe tympanic de­ served in Odobenocetops, which is therefore less derived than scribed above actually belongs to Odobenocetops, the sigmoid the living Monodontidae for those features. morphology of its involucrum in medial view leaves little The Monodontidae are further characterized by widely sepa­ doubt that the walrus-like dolphin belongs to the Delphinida rated pterygoid hamuli. This monodontid condition is greatly (see Muizon, 1988c:163, fig. 3). The sigmoid involucrum of accentuated in Odobenocetops, where the pterygoids are very the tympanic of the Delphinida contrasts with the olive shape widely separated. This feature, however, also is known in some observed in the Platanistoidea (sensu Muizon 1987, 1991) and Delphinidae (generally as an individual variation) and in the with the indented involucrum of the Eurhinodelphoidea and Holocene Phocoenidae (it is absent in Piscolithax, a Tertiary Ziphiidae. phocoenid). Odobenocetops perllvianlls is a delphinoid because it bears a In addition to the above evidence, the discovery of a partial medial maxilla-premaxilla suture in the anterolateral edge of skeleton (including skull and forelimb) of Odobenocetops sp. each narial fossa, which is formed by the maxilla. This charac­ (MNHN SAO 202) in a slightly younger level at Sacaco con­ ter, which was analyzed by Muizon (l988c: 199), is present in firms both that 0. peruvianus is a cetacean and that its affinities all the living Delphinoidea (with some individual variation), lie with the Monodontidae. The forelimb shows the typical ce­ . but it is not constant in the fossil groups, such as the Kentri­ tacean modifications (immobility of the elbow), and its overall odontidae. This feature is found only in the Delphinoidea, how­ morphology recalls that of the Monodontidae (Muizon and ever, so its presence in Odobenocetops is evidence for its clas­ Domning, 2002) (Figure 17). The monodontid affinities of sification in that superfamily. Odobenocetops are clearly reinforced by the morphology ofthe Among the Delphinoidea, Odobenocetops is related to the inner ear, as shown above. Monodontidae by three derived features that have been re­ In view ofthe exceptional specializations of Odobenocetops garded as synapomorphies of the Monodontidae (Muizon, peruvianus, it was referred (Muizon, 1993a) to a new family, 1988b:191): the Odobenocetopsidae, regarded as the sister group of the I. A ht-:rallamina of the palatine that passes below the op­ Monodontidae. Odobenocetops peruvianus shows obvious au­ tic gutter and joins the frontal laterally. tapomorphies, the most important of which are (I) loss of the 2. An alisphenoid and adjacent portions of the squamosal typical cetacean rostrum and enormous thickening of the pre­ that are very thickened lateral to the foramen ovale and medial maxillae at the anterior region of the skull, (2) development on to the zygomatic process. the premaxillae of large, downturned alveolar processes hous­ 3. The presence ofa very long and low temporal fossa (con­ ing one large erupted right tusk and a small, probably trary to what Muizon (l988b:207) stated, it is clearly present in unerupted left tusk, (3) extreme anterior position of the en­ Denebola (Barnes, 1984, fig. 6». This morphology corre­ larged nares and the dorsally facing orbit, (4) withdrawal ofthe sponds to an anteroposterior elongation ofthe middle region of frontal and maxilla from the posterodorsal angle of the skull, a the skull. Related to this modification are the anterior stretch­ condition that opens the temporal fossa dorsally, (5) reduction ing ofthe supraorbital process ofthe frontal, the less transverse of the maxillae, which are excluded from the bony palate and orientation ofthe optic gutter than in the other delphinoids, and only form part of the lateral wall of the skull, (6) extensive the position of the exits of the optic nerves at the most anterior modification of the zygomatic process of the squamosal into region of the brain and not posteroventral to the apex of the the thick lateral wall of a large, anteroposteriorly elongated brain as in the other delphinoids. In fact, the transformation in squamosal gutter, part of which houses the glenoid cavity and the Monodontidae seems to be the consequence ofthe anterior the middle sinus, (7) great reduction or absence ofthe premax­ migration ofthe eyes. That original modification is present to illary sacs, and (8) a strongly thickened posteroventral region an extreme degree in Odobenocetops, where the supraorbital of the squamosal and a very solid, interdigitated parietal-squa­ process of the frental is tremendously stretched anteriorly, the mosal suture. oplic gutter is oric'nted almost anteroposteriorly, the.. exits of the The occurrence of tusks in Odobenocetops is a convergen,'c optic nerves arc located at the anterior extremity of the brain, with Monodon; in the latter genus the large tllsk ofthe male .is NUMBER 93 245 implanted in the left maxilla, whereas in Odobenocetops the terfaces between bone and soft tissues (which behave like liq­ large tusk is implanted in the right premaxilla. Consequently, uid in sound transmission) and the foamy filling ofthe air sinus the tusk ofMonodon and that of Odobenocetops are not homol­ (which behaves like air in sound transmission). The sound ogous. waves are therefore forced to enter the cochlea through the ex­ Enlarged apical teeth or tusks are known in some other ceta­ ternal auditory meatus, thus providing good directional hear­ ceans. Enlargement of apical teeth is common in squalodonts, ing. McCormick et al. (1970) rejected this idea, however, stat­ where the anteriormost tooth of the premaxilla and of the den­ ing that the acoustic vibrations reach the ear through the tissues tary are enlarged and protrude anteriorly, being almost horizon­ of the head, and Norris (1964, 1968, 1969) and Brill et al. tal (Dal Piaz, 1916; Kellogg, 1923). A similar condition is (1988) proposed that the echolocation sounds may return to the found in Kentriodon pernix from the middle Miocene of the body "by the way ofthe intramandibular fat body which acts as Calvert Formation of Maryland. These teeth, in squalodonts a passive wave guide and enter the middle ear via the tympanic and Kentriodon, do not show asymmetry. Kharthlidelphis bulla which transmits sounds directly to the ossicular chain and dieeros from the Oligocene of Georgian Republic (Mchedlidze cochlea through the processus gracilis," which attaches the and Pilleri, 1988) also possesses one pair ofenlarged apical up­ malleus to the tympanic. per teeth that protrude anteriorly and horizontally. In this case The melon is a fatty organ located on the dorsal side the ros­ the left tooth has a diameter approximately twice that of the trum anterior to the nasal complex. In some species, the melon right one, and the teeth are implanted in the maxillae. This con­ reaches the apex of the rostrum; this is generally true, for ex­ dition is basically. similar to that of the narwhal, but the size ample, in the Globicephalinae. The melon is embraced by the difference between the teeth is much less and the right tusk is medial portion of the rostral muscle (Mead, 1975), also called erupted. Kharthlidelphis is therefore very different from Odo­ the pars labialis of the maxillonasolabialis (Lawrence and benoeetops, where the tusks are implanted in the premaxillae Schevill, 1956) and the nasolabialis profundus pars lateralis and the right tusk is much larger than the left one. Furthermore, (Rodinov and Markov, 1992). Below the posterior part of the the downturned tusks of Odobenoeetops are unique among ce­ melon are the nasal plug muscle and the premaxillary sacs taceans. (Mead, 1975), in which it is partially imbricated. The most commonly suggested function of the melon is as an acoustic lens (Lilly, 1961; Norris, 1964, 1968, 1969; Wood, 1964). Al­ Functional Anatomy though it se.ems clear that the melon is involved in sound recep­ When compared with the other delphinoids, Odobenoeetops tion and transmission (Norris and Harvey, 1974; Mead, 1975), shows drastic modifications of the skull that have been men­ its function remains unclear. tioned in the description above. These can be classified in three On the skull of Odobenoeetops, three features can be ob­ groups: (1) those related to the nasal sacs, basicranial sinuses, served that denote lesser ability in sound production and/or and auditory region, and consequently to sound production and transmission than in the other delphinoids: (1) the premaxillary reception; (2) those related to feeding; and (3) the tusks. In the sacs and nasal plug muscles were either absent or very reduced; ;;,lIowing sections we analyze each of these groups ofspecial- (2) the melon was vestigial or absent; and (3) apparently the 'ations. ,nasal diverticula ofthe nasal passages were very reduced or ab­ sent. The latter statement requires explanation. The nasal diverticula of the Delphinidae have been well de­ NASAL SACS, BASICRANIAL SINUSES, AND scribed by Lawrence and Schevill (1956) and Mead (1975). AUDITORY REGION The nasal sacs and the communication between them are con­ Several authors have stated that the nasal sacs are at least trolled by the various layers of the pars nasalis of the maxillo­ partially implicated in sound production (Lawrence and Schev­ nasolabialis muscle. Most of that muscle is attached to the dor­ ill, 1956; Lilly, 1961; Lilly and Miller, 1961; Evans and Pres­ sal side of the skull, lateral to the nasal opening and to the cott, 1962; Norris, 1964, 1968, 1969; Norris and Evans, 1967). premaxilla, and posterior to the antorbital notch and the nuchal '~ad (1975) critically reviewed the literature on the relations crest. It is divided into six layers according to Lawrence and between the nasal diverticula and sound production and Schevill (1956), and five layers according to Mead (1975). reached the conclusion that "the structures more likely to be in­ Rodinov and Markov (1992) recognized nine muscles, four in volved in sound production are those in the vicinity ofthe nasal their nasolabialis group (which includes the nasal plug muscle, plugs." Mead (1975) also stated that the premaxillary sacs their nasolabialis profundus pars anterior medialis) and five in could be used as sources for air during phonation. their maxillonasalis group. None of these layers or muscles is The role of the basicranial air sinus has been debated. Fraser attached to the parietal. In Odobenoeetops, the anterior with­ and Purves (1960) stated that their function is the phonic isola­ drawal of the frontal and maxilla and the loss of the roof of the tion of the periotic from bone-conducted sounds. The sound temporal fossa are supposed to have considerably reduced the \ ,ves transmitted by bone conduction are reflected on the irt- attachment area of the pars nasalis of the maxillonasolabialis. 246 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

This statement assumes that very few or no muscle fibers suggested that both the nasal sac system and the larynx are im­ would have migrated onto the broadly uncovered parietal, pos­ plicated in sound production (Lilly and Miller, 1961; Evans teriorly. The five layers recognized by Mead (1975) are the and Prescott, 1962; Schevill, 1964; Evans, 1967). Ridgway et pars posteroextemus (PE), the pars intermedius (1), the pars an­ al. (1980), however, clearly rejected the possibility of any teroexternus (AE), the pars posterointernus (PI), and the pars sound production in the larynx and stated that sounds are pro­ anterointernus (Al). The first four layers (PE, I, AE, PI) attach duced by the nasal system only. to the lateral edge of the supraorbital process and to the tempo­ Consequently, Odobenoeetops probably had no (or vestigial) ral crest. The pars AI attaches mainly to the ascending process nasal plugs, no (or vestigial) melon, and very reduced premax­ of the maxilla and to the lateral border of the premaxilla. illary sacs, if any; and, as the maxillonasolabialis was appar­ Consequently, we can assume that the great reduction of the ently very reduced in comparison to the other delphinoids, it maxilla would have resulted in an extreme reduction (or loss) also is likely that the other nasal diverticula were either re­ ofAI. Furthermore, the disappearance ofthe temporal crest and duced or totally absent. Ifso, then Odobenoeetops had little (or the reduction of the lateral edge of the supraorbital process no) ability to produce sounds (in the usual cetacean way) and (which is very thin and strongly notched in dorsal view) also therefore to echolocate. suggest a great reduction of the four layers PE, I, AE, and PI. As indicated earlier, the inner-ear structure implies that Odo­ Even ifall the nasal diverticula totally disappeared, however, it benoeetops, like most mammals, was capable of perceiving ul­ is obvious that at least part of the maxillonasolabialis (pars an­ trasonic sounds (frequencies >20 kHz), but whether Odobeno­ teroextemus) must have been retained, as this layer is involved cetops could echolocate cannot be determined from ears alone. in opening the blowhole. High-frequency facility is not synonymous with echolocation. Furthermore, Odobenoeetops has large nasal apertures, and The ability to echolocate implies the production of self-gener­ the problem is to determine whether it had nasal plugs or not. ated, beamed, gated signals and analysis of the corresponding Given the size of the bony nasal passages, if nasal plugs were echoes. Analysis of ambient sound is passive listening and is present, they would have been very large and consequently common to all mammals. Passive listening clearly provides would have been moved by equally developed muscles. We ob­ considerable information about the immediate environment served, however, that the nasal plug muscle, if present, was (e.g., directionality or the relative distance of two sound very small. Furthermore, Mead (1975:53) stated that the "PI sources), but it does not provide spatial or textural information and AI also serve to seat the nasal plugs in the orifice of the with the level of resolution commonly obtained through bioso­ bony nasal passage." The AI originates on the posterior border nar. Biosonar is an extraordinarily sophisticated acoustic imag­ ofthe supraorbital process and on the temporal crest, and the PI ing system that conventionally uses very high frequencies be­ originates on the ascending process of the maxilla. As stated cause the level ofdetail that can be transduced is related to the above, these layers were probably very reduced in Odobenoee­ wavelength of the ensonifying signal. The stronger the evi­ tops, a condition that would be in agreement with the absence dence against the presence of a melon, which is directly impli­ ofnasal plugs. Furthermore, Lawrence and Schevill (1956) and cated in outgoing signal generation and control, the less sup­ Mead (1975) have shown that the nasal plugs are tightly imbri­ port there is for functional echolocation by Odobenoeetop.\ cated with the dorsal surface of the premaxillary sacs and with The absence of nasal plugs would be critical evidence against the melon and that there is a sort of histological continuity echolation, as that structure and the diagonal membrane seem among these organs. Consequently, the absence or great reduc­ to represent the major sound producers in odontocetes, al­ tion ofthe premaxillary sacs and the absence ofa melon would though it is not possible to refute definitively the hypothesis be consistent with the absence or reduction of the nasal plugs. that Odobenoeetops had a reduced air sac system and nasal Ifthe nasal plugs were actually present and well developed, plugs. there would be an inconsistency between their large size on the Whatever the morphology was, it is very probable that Odo­ one hand and the reduction ofthe nasal plug muscle, premaxil­ benoeetops had little echolocational ability, ifany-a condition Imy sacs, and melon on the other. From this discussion it there­ that may have been compensated for by good binocular vision. fore follows that the nasal plugs were very probably absent or The orbit of Odobenoeetops is proportionally much larger than vestigial in Odobenoeetops. in any other cetacean. The distance between the apices of the There seems to be a consensus that the nasal plugs and the postorbital and antorbital processes is close to that in Monodon nasal diverticula are related to sound production (Lilly, 1961;· even though Odobenoeetops is an animal approximately 15% Lilly and Miller, 1961; Norris, 1964, 1968, 1969; Norris and smaller. Furthermore, the dorsal shifting of the process and the Evans, 1967; Kleinenberg et aI., 1969; Diercks et aI., 1971; depth ofthe orbital notch clearly indicate that dorsal or anterior Norris et aI., 1971; Evans and Maderson, 1973; Mead 1975; binocular vision was possible. Furthermore, good dorsal or an­ Dormer, 1979; Ridgway et aI., 1980). It also is possible that terior binocular vision is in agreement with the probable feed· sound could be produced by the larynx (Lawrence and Schev­ ing posture of Odobenocetops (see "Scenario for the Feeding ill, 1956; Purves, 1967; Schenkkan, 1973). Some authors have Mode ofOdobenoeetops peruvianus," below). ,..

l' 8ER 93 247

FEEDING ADAPTATIONS medialis) originate on the lateral side of the skull, on the ptery­ goid but also on the orbitosphenoid. In Neophocaena, Howell The modifications of the skull of Odobenocetops are corre­ (1927) observed a very reduced muscle originating on the lated with major modifications of its musculature. Four aspects pterygoid membrane and on the lateral edge of the pterygoid should be considered: mandibular movements, head move­ bone adjacent to it. The "insertion is not upon the mandible but ments, morphology of the palate, and inferred presence of an on the tough tissue near the ear bone" (Howell, 1927:23). In upper lip and vibrissae. \ 'ANDIBULAR MOYEMENTS.-In Odobenocetops, the with­ Kogia, however, Schulte and Forest Smith (1918) observed two (\al of the maxillae and frontals from the dorsal side of the relatively strong pterygoid muscles (internus and externus) that braincase has opened dorsally the temporal fossa, which in insert upon the medial side of the dentary. Seagars (1982) most odontocetes is roofed. The parietals are widely uncovered found that in long-jawed delphinids with numerous teeth, the dorsally and secondarily increased in size so that they occupy pterygoideus internus (=medialis) is in fact the dominant ad­ most of the dorsal surface of the braincase. In the other delphi­ ductor muscle, followed in importance by the temporalis, mas­ noids, the temporal fossa is roofed by a lateral expansion of the seter, and pterygoideus externus (=Iateralis). In species with frontal and the maxilla. The parietals are small bones restricted fewer teeth and broader jaws (Globicephala, Grampus), the to the lateral sides of the skull and do not contact each other. temporalis is more powerful relative to the pterygoideus inter­ F,~ temporalis is a small weak muscle whose origin is located nus, and in Orcinus the temporalis is actually the dominant (. :he parietal, on the lateral side ofthe skull, and whose inser­ muscle (Murie, 1870, 1973). In Odobenocetops, the pterygoid tion is on the coronoid process of the dentary. Its action is to shows a very strong laterally convex crest in the position ofthe raise the mandible to close the mouth. If muscle insertions and pterygoideus medialis of the dog and Kogia, and ofpart of the origins are at least partially related to particular bones (which is undivided pterygoid ofNeophocaena. This crest is more devel­ known to be not always true), the great development of the pa­ oped than any structure observed in the other odontocetes and rietal of Odobenocetops might have led to a great extension of indicates, in Odobenocetops, a pterygoid muscle much stronger the temporalis on the dorsal side of the skull. Consequently, than in the other delphinoids. Together with the evidence of a Odobenocetops probably was capable ofmuch stronger adduc­ large temporalis cited above, this seems consistent with the pat­ tion ofthe mandible than are other odontocetes. tern documented by Seagars (1982) in delphinids, which would \nother important elevator of the mandible is the masseter. lead us to expect a large temporalis and internal pterygoid in In the dog, the masseter is divided into three layers. The super­ connection with the presumably short, toothless mandible of ficial and middle layers originate on the lateral side of the zy­ Odobenocetops. gomaticarch (on the jugal and squamosal), and the deep layer The main mandibular depressor muscle in terrestrial mam­ originates on its medial side (Miller et aI., 1964). As for the od­ mals is the digastricus. In the dog it originates on the anterior ontocetes, Howell (1927) described two layers in Neopho­ side ofthe paroccipital process of the occipital and inserts upon caena. superficialis and profundus, but only the former had a the ventral border of the mandible (Miller et aI., 1964). In the bony origin, on the zygomatic process of the squamosal and on odontocetes, Howell (1927, 1930) termed it monogastricus, the postorbital process of the frontal. The superficial layer in- and the origin is on the anterolateral border of the thyrohyal ts on the posteroventral angle of the mandible and the deep and/or basihyal. Apparently there is no attachment to the skull. layer on the dorsal edge of the dentary in front of the insertion The tremendous thickening of the squamosal and, to a degree, of the temporalis. in Odobenocetops. the large distance exist­ of the occipital at the posterolateral angle of the skull in Odo­ ing between the postorbital process of the frontal and the zygo­ benocetops, however, seems to indicate considerable muscular matic process might indicate that the masseter was divided into stress in that region. Although it is therefore possible that part two clearly defined elements. Several features suggest, how­ or all of the digastricus muscle was attached to the posterolat­ ever, that the masseter did not originate on the postorbital pro­ cess, butTather that it was attached only on the strong zygo­ eral angle of the skull, a condition that would have to be re­ matic process of the squamosal. This structure is the very stout garded as a reversal, the neck muscles (see below) are more , eral wall of the squamosal gutter described above. It is a probably responsible (at least partially) for the strong develop­ tdick bony wall strongly attached to the skull all along its ment ofthat region ofthe skull. length. Furthermore, the squamosal-parietal contact is not The masticatory musculature of Odobenocetops thus seems squamous and vertical as in most other mammals; it is a very to have been much more powerful than in any other odonto­ solid, transversely oriented, interdigitated suture apparently ca­ cetes. Considering the lack of maxillary teeth, it indicates a pable of resistance to strong anteroposterior stresses. Conse­ very peculiar mode of feeding involving powerful movements quently, the architecture of the areas of origin of the masseter of the lower jaw and perhaps of the tongue, gular, and hyoid indicates a muscle much stronger than in the other delphinoids. musculature. Although there is no direct evidence of these lat­ On the ventral side of the skull the pterygoid muscles also ter muscle groups, the very deep and arched palate suggests a ..;: mandibular adductors. In a terrestrial mammal (the dog, cf. very large tongue and a consequently strong throat musculature \liller et aI., 1964), both pterygoideus muscles (Iateralis and (see below). 248 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

HEAD MOVEMENTS.-The muscles involved in head move­ is much narrower, as the lateral plane of the palate is oblique in rostJ ments are numerous and strong. In odontocetes, the most im­ relation to a line joining the squamosal gutter to the apex of the phin portant are the semispinalis, rectus capitis, splenius, multifidus, rostrum, the supposed axis of the dentary. If the dentary of part stemomastoideus, longissimus, and scaleni. The first four mus­ Odobenocetops was straight as it is in all the other odontocetes, part cles insert on the occipital shield, and not much can be said then the dentary in its middle portion was only slightly thicker the J about their morphology in Odobenocetops. The three latter than in the other delphinoids; however, it was probably very obs< muscles insert on the posterior border of the paroccipital pro­ high, as it needed to be strong to bear the stresses of the power­ the j cess of the occipital and/or on the zygomatic process of the ful musculature. The anterior part of the mandible probably top~ squamosal. The sternomastoideus and the longissimus insert on was greatly thickened to partially fill the anterior region of the I the posterior region of the sternomastoid fossa and the lateral palate. Another possibility is that at their anterior ends, the den­ sist< edge of the paroccipital process. In Odobenocetops this region taries were as high as the concave portion of the medial side of obs is greatly expanded ventrally and forms the tremendously the premaxillary process (i.e., its proximal two-thirds, approxi­ pro· thickened posterolateral angle of the squamosal. The enlarge­ mately 15 em) and formed two divergent processes or expan­ mel ment of the area of insertion of these muscles and the stoutness sions complementary to those ofthe premaxillae. This mandib­ sup ofthe bone in this region suggest that the stemomastoideus and ular morphology would somewhat resemble that observed in lip longissimus of OdobcllocL'topS were very powerful muscles. the borhyaenoid marsupial Thylacosmilus from the Pliocene of (an The same could be said about the scaleni, which insert on the Argentina. Such a morphology would allow the apex of the ~o t posterior surface of the paroccipital process and basioccipital lower jaw to fit perfectly in the anterior region of the palate, j crests. The bony origin of the stemomastoideus is on the ante­ thereby allowing a perfect closing of the mouth. car rolateral edge of the sternum; the bony origin of the longissi­ . An even slightly deep, wide, and/or vaulted palate is un­ (Ta mus is on the transverse process of the anterior thoracic verte­ known among other cetaceans, so the morphology of Odobeno­ riot brae (Pabst, 1990); origin of the scaleni (dorsalis and ventralis) cetops is unprecedented. In non-cetacean mammals, a deep, vib is on the first ribs. These three muscles, ifacting bilaterally, are arched palate is known in the Odobenidae (Odobenus ros­ ~ depressors of the head. and if acting unilaterally, are lateral ro­ maniS, Valenictus chulavistensis) and in the Otariidae (Otaria bel tators ofthe head. The region ofthe imchal crest located behind flavescens and to a lesser extent Phocarctos hookeri). As men­ \vil the sternomastoid fossa also is extremely thickened. In tioned above, the deep palate of Odobenocetops is consistent Neophocaena and MOllodoll the splenius and semispinalis capi­ with a large tongue and consequently strongtongue muscles. tis insert in this area. The lateral extremity of the insertion of Also, the gular musculature must have been fairly powerful, as tur these muscles also is on the posterior part of the upper part of it is related to the tongue and mandible movements. vit the sternomastoid fossa. 1f acting bilaterally these muscles are UPPER LIp AND VIBRISSAE.-Muizon (I993a, I 993b) has Ini levators of the head. suggested the presence in Odobenocetops of a strong upper lip. ph Consequently, the assumed strength ofthe stemomastoideus, That hypothesis is maintained herein. Lips are known in several Yr longissimus, and scalenus in Odobenocetops would indicate other odontocetes and are especially well developed in the bel­ Si strong and active YCrtical and transverse movements of the uga (Kleinenberg et aI., 1969:80). Although these authors state tn head or at least a need for a strong control ofthese movements. that the lips of the beluga are devoid of musculature and nut This conclusion is in agreement with the morphology of the movable, figure 5 of Brodie (1989: 132) seems to demonstrate condyles and of the supracondylar fossa, which would allow considerable mobility of the beluga's lips. Furthermore, Mead bl significant vertical movements of the head (see above). It is (1975:39) mentioned that part ofthe lateral rostral muscle ofthe Ti further corroborated by the relatively large size of the semicir­ delphinoids is associated with the upper lip and stated that tu cular and vestibular canals, in Odobenocetops as in Delphi­ "[t]he lips of Cetacea are generally considered to be immobile, v{ napterlls, which are apparently related to the mobility of the but in view ofthe great expanse oflip surface in animals such as (L neck (see above). Globicephala and Grampus and the relatively large mass of tl PALATE MORPHOLOGY.-The palate of Odobenocetops is muscle inserting into the connective tissue of the lips in these t1 very wide, very deep. and arched. The morphology of its ante­ forms, a certain amount of mobility may be present and may be rior part is dominated by the exceptionally developed premax­ important in feeding." A strongly developed, mobile upper lip illae and the large vomer. The palate is very wide posteriorly in Odobenocetops could therefore result from the enlargement because of the lateral expansion of the pterygoid, which forms ofa structure already existing in the other delphinoids. a ventrolaterally oriented wing. The palatines also are very en­ We mentioned above the strong muscle attachments on the larged in the posterior region of the palate. Between the alveo­ dorsal and anterior face of the apex of the rostrum. Given that c lar process of the premaxilla and the pterygoid wing, a deep, the medial rostral muscle is inserted into the melon (Mead, rounded notch fom1s the narrowest (middle) portion of the pal­ 1975:10), which Odobenocetops lacks, both the medial and lat­ ate. Ihis notch lies exactly in the path of a prolongation of the eral portions ofthe rostral muscle might be implicated in upp,:r squamosal gutter anJ was obviously for the passage of the lip movements in Odobenocetops. Consequently, the muscle mandible. Seen ventrally, the notch is wide and almost semicir­ arising from the subapical rims described above (passing later­ cular; viewed along the axis ofthe squamosal gutter, though, it ally to the premaxillary foramina, converging at the apex ofthe NUMBER 93 249 rostrum, and enclosing the triangular area where, in other del­ etops are premaxillary and strongly asymmetrical, whereas phinoids, the nasal plug muscle is attached) would represent they are maxillary and symmetrical in Odobenus. The poster­ part of the medial portion of the rostral muscle. The remaining oventral angle of the skull is made up of the paroccipital pro­ part of the medial portion of the rostral muscle, together with cess and the squamosal, whereas in Odobenus it is formed by the lateral portion, would create the strong muscle attachments the tremendously enlarged mastoid process ofthe periotic. The observed on the anterodorsal sides of the alveolar processes of attachment area of tough tissue is located on the anterior edge the premaxillae. The whole rostral musculature of Odobenoce­ of the premaxilla; in the walrus it is on the dorsal side of the tops would therefore be related to upper lip movements. mandibular symphysis. The occurrence ofa large upper lip in Odobenocetops is con­ Muizon (1993a, 1993b) regarded Odobenocetops peruvianlls sistent with the presence of the large neurovascular foramina as a walrus-convergent bottom feeder. We have noted above observed on the anterior and ventral borders of the alveolar that Odobenocetops and Odobenlls share several strong simi­ process and with the presence of the very large palatine fora­ larities in their skull morphology. Functional comparisons will men, which could have provided extensive blood and nerve help determine whether the adaptations of Odobenocetops supply to the ventral region of the lip. The dorsal region of the could fit with the walrus's mode offeeding. lip was irrigated and innervated via the very large maxillary The living walrus feeds mainly upon benthic invertebrates, (and probably premaxillary) foramina, which also were related such as thin-shelled bivalves (Mya, other clams, and mussels), ;0 the blood and nerve supply of the pulp cavities of the tusks. gastropods, molting crustaceans, holothurians, and tunicates, As noted above, the facial nerve of Odobenocetops is signifi­ but rarely upon cephalopods, fish, or other vertebrates. Prey cantly larger than in the other delphinoids except Monodon buried in the substrate, such as clams, are excavated by jetting (Table 3). The large diameter ofthe facial nerve canal in the pe­ water from the mouth (Kastelein and Mosterct, 1989; Kastelein riotic is consistent with the presence of a large upper lip and et aI., 1991). As suggested by Vibe (1950) and clearly ex­ vibrissae. pressed by Fay (1982), the walrus does not crush the shells of Muizon (1993a, 1993b) suggested that the upper lip of Odo­ the mollusks it ingests; rather, it holds them between its lips benocetops could have carried vibrissae, evolved convergently and jaws at the front of the mouth and sucks out their siphons, \vith those of the walrus. The extensive vascularization and in­ using its mouth as a vacuum pump. Fay (1982:171) described lervation of the upper lip and the very spongy nature of the the walrus mouth as follows: "The extraordinary vacuum pump bone of the anterior side ofthe rostrum suggest that such struc­ ofthe walrus ... is powered by very large lingual retractors and tures could possibly have been present. Furthermore, vestigial depressors (M. styloglossus, hyoglossus, and genioglossus), vibrissae are well known in some adult cetaceans (mysticetes, complemented by the highly vaulted palate, the long firmly Inia) and have been observed in several newborn or fetal del­ ankylosed mandibular symphysis, and the unusually powerful phinoids and Pontoporia (BourdelJe and Grasse, 1955; m. tensores veli palatini and m. buccinatorius, which provide Yablokov and Klcvezal, 1962; Tomilin, 1967; Best and da rigidity to the walls ofthe 'cylinder.' The tongue is the 'piston.' Silva, 1989). The vascularization of the anterior side ofthe ros­ The small rigid oral aperture, complemented by ample muscu­ trum of Odobenocetops also could be related to the lip only, lar lips, insure that the full effect ofthe vacuum is exerted only owever, so the presence ofvibrissae remains undemonstrated. on objects held in the incisive area at the front of the mouth." COt\\PARISON WITH Odobenus.-Several features of Odo­ Gordon (1984) confirmed that the vacuum is created by move­ benocetops are convergent with the morphology of the walrus. ment of the tongue directly rearward. Fay (1982: 167) also They are (1) the large, deep, vaulted palate; (2) the presence of stated, concerning the tough tissue covering the dorsal surface tusks; (3) the \vide and low occipital, in posterior view, and the of the mandibular symphysis: "The rough cornified surface ventrolateral expansion of the ventrolateral angle of the skull; seems admirably suited for grasping and holding slippery prey; (4) the thickness and the extreme stoutness ofthe bone forming I believe that it functions also to hold molluscan shells while the ventrolateral angle of the skull and of the alisphenoid; (5) their contents are removed by suction." the well-developed upper lip and the presence of vibrissae (not When searching for food, the walrus moves along the bottom :rtain in Odobenocetops); and (6) the presence of a rough, ir­ with its head down and its body at an angle to the bottom of regular, and expanded area for attachment of tough or horny about 10°-45° (Fay, 1982: 164). The vibrissae are the sensitive tissue on the edges ofthe mouth (inferred for Odobenocetops). organs that are constantly in contact with the bottom and trans­ The palate ofOdobenocetops, however, differs strongly from mit information to the animal. As mentioned by Fay (1982), the that of Odobenus. It is much wider and deeper. It is composed walrus feeds in darkness and "the eyes are comparatively of the premaxillae, palatines, vomer, and pterygoid, but the smaller than those of other pinnipeds;" "they may be the least maxillae do not crop out on the palate as in most other mam­ important as sensors" and "the animals probably locate their mals. There are no maxillary teeth in Odobenocetops, although prey by vibrissal contact." Fay (1982:164) also mentioned that i l is known that some fossil odobenine walruses (Valenictus excavation of buried animals may be achieved by "rooting" '/lIlavistensis) also lacked maxillary teeth (Demere, 1994). with the snout: "the pattern of vibrissal abrasion, the greater The vomer is very large and lozenge-shaped, as (but much cornification of the upper edge of the snout..., and the powerful larger than) in most other odontocetes. The tusks of Odobenoc- cephalo-cervical musculature of the walrus are unlike those of 250 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 18 (Ieft).-Odobenocetops peruviallus. holotype (USNM 488252): anterolateral view of the skull in a feeding position.

5 ern

any other pinnipeds. Each points to frequent use of the snout as rus, it was a sucking feeder that did not have the masticatory a digging organ." This mode of feeding creates long, sinuous apparatus necessary to crush any hard exoskeleton of its prey. I furrows in the seafloor (Nelson and Johnson, 1987), each pre­ Fay (1982: 166) has shown that the teeth of the living walrus I sumably representing a single bout of feeding by a walrus on a are not used for crushing, and it is noteworthy that some fossil single dive. odobenids (Demere, 1994) had lost the postcanine teeth alto­ In Odobenocetops. although the mandible is unknown, sev­ gether. eral aspects of the skull morphology are in perfect agreement Another noteworthy feature of Odobenocetops is the with suction bottom feeding (see Figures 18, 20): The very strol1gly modified zygomatic process with a large squamosal large vaulted palate (even larger than in the walrus); the in­ gutter and a tremendously resistant parieto-squamosal suture. ferred cornified (or strongly fibrous) edge of the upper lip; the As it has no equivalent in any living mammal, this structure is fairly well-developed masticatory musculature (for a cetacean), difficult to interpret. If, however, this structure is related to the close, in strength, to that of the walrus; the powerful occipital very strong masticatory mus9ulature mentioned above, it likely musculature (at least the scalenus, sternomastoideus, and long­ permitted very powerful (thqugh probably not very extensive) issimus); and the significant atlanto-occipital movements de­ movements of the lower Jaw.· The exceptional strength of the noted by the very convex and salient occipital condyles and the mouth of Odobenocetops;would thus have allowed it to main­ large supracondylar fossae. Contrary to walrus morphology, tain its prey extremely firmly in its mouth. The extreme devel­ though, the eyes of Odobenocetops were large. and good binoc­ opment of the squamosal gutter and its high position are diffi­ ular vision compensated for the probably poor echolocational cult to explain because the mandible is unknown. ability. Even walruses have good anterodorsal binocular vision Odobenocetops also shares with Odobenus the possession of and seem to forage visually when water clarity permits, how­ tusks, but the large premaxillary alveolar processes of Odo­ ever (Kastelein and Wiepkema. 1989; Kastelein et a!., 1993). benocetops are absent in the walrus. Fay (1982:136) stated that Odobenocetops had no maxillary teeth; like the living wal- the tusks of the walrus are not related to feeding but instead ';UMBER 93 251 playa social role. The same has been stated for the narwhal crushing teeth, which were used to break the hard shells of (Nishiwaki, 1972; Hay and Mansfield, 1989). Consequently, benthic invertebrates (Tedford et al., 1994). we assume that the tusks of Odabenacetaps were used more as Desmostylians and Sirenians: Like Odobenocetaps, most social organs than as feeding implements (see below for further marine mammals that are believed to have dug for food in the discussion of the tusks). The deep premaxillary fossae border­ seafloor possess large ventrolateral expansions of the poster­ ing the anterior extremity ofthe vomer on the palate are absent oventral angle of the skull, which presumably reinforced the in all the other odontocetes, and nothing similar is observed in neck when the animal was searching for its food, with its head ;he walrus. Their function has not been elucidated. down and the body in an oblique position in the water. Desmo­ COMPARISONS WITH MARINE MAMMALS OTHER THAN Oda­ stylians provide another example of this; they were evidently benus.-Delphinapterus leucas: Kleinenberg et aL (1969: substrate-feeding herbivores (Domning et aL, 1986) and all 89-90) presented an interesting interpretation ofthe functions shared an enlarged paroccipital process (Ray et aL, 1994). Fur­ ofthe tongue ofthe beluga for feeding. According to these au­ thermore, the peculiar sirenian Miosiren, which has been sus­ thors, the tongue is a very mobile and strong organ that has pected of molluscivory on the basis of its dentition and rein­ three functions. First the tongue orients the pr~y in the mouth. forced palatal region, also has enlarged post-tympanic Second, the animal "draws its tongue back, pressing the prey to processes (Sickenberg, 1934). Other tusked sirenians, however, 'he palate and forcing it to the entrance of the pharynx." Third, do not, and at least the most specialized ofthem may have used ·the tongue forces the water out of the mouth, preventing its their jaw muscles instead of their neck muscles for digging entrance into the intestinal tract." The tongue thus appears to be (Domning, 1989). a powerful and extremely specialized organ for food ingestion. The Holocene sirenian Dugang dugon (Muller) superficially It is therefore very probable that Odabenacetaps, like all living resembles Odabenacetaps in having a pair ofslender, relatively straight, and open-rooted upper incisor tusks that are largely odontocetes, had a powerful, enlarged (given the size of the enclosed in prominent, downward-directed premaxillae. In the palate) tongue, and that it could have been involved in a pump­ dugong, however, these tusks are symmetrical, are not strongly ing feeding action as is observed in the walrus. divergent, project at most a few centimeters outside the gum, Bottom feeding is well established for belugas; worms and and are directed anteroventrad rather than posteroventrad. Al­ ',enthic animals are often found in their stomachs (Matthews, though tusks of fossil dugongids were surely used in feeding i 978). Belugas also are capable ofpowerful water-jetting simi­ (probably to excavate sea grass rhizomes; Domning, 1989), lar to that observed in the walrus (D.P. Domning, pers. obs., those of the living species seem to have lost this function and and J.G. Mead, pers. comm., 1992). Bottom feeding also is have become relegated to purely social uses, such as fighting known in Glabicephala, and water-jetting in Orcinus and Gla­ between males (Preen, 1989; Domning, unpubL data). Evi­ bicephala (Werth, 1992). dently as a result of this restriction in function, tusks of D. Kalpanamas clallamensis and K. newpartensis: Tedford et dugon (uniquely among sirenians) have become sexually di­ al. (1994) described the skull of Kalpanamas, an aquatic ursid morphic, normally erupting only in the males (Marsh, 1980). It from the early Miocene of Washington and Oregon. Among the thus appears that the tusks of the herbivorous dugong can shed 'ost peculiar modifications of the skull ofKalpanamas are the no additional useful light on the biology ofthe presumably car­ -:xtremely enlarged mastoid processes, which are oriented ven­ nivorous Odobenocetaps. trolaterally as is observed in the walrus. According to Tedford Scenario for the Feeding Made ofOdobenocetaps peru­ et aL, Kalpanomos could have fed upon benthic hard-shelled vianus: Odobenocetaps lived in the early Pliocene on the invertebrates, mainly mollusks but also echinoids. Conse­ coast of Peru, in the shallow waters ofthe bay of Sud-Sacaco, quently, the substantial and strong movements ofthe head must probably close to the shore (Marocco and Muizon, 1988). It haw necessitated strong neck musculature. was a sucking feeder preying upon benthic invertebrates such Furthermore, Kalpanamas had anterior binocular vision, as thin-shelled mollusks and/or molting crustaceans, which which assisted its search for food, as in Odabenacetaps and were abundant in the Pisco Formation (Muizon 1981; Muizon dabenus (Kastelein et aL, 1993). Tedford et aL (1994) also and DeVries, 1985; Carriol et aL, 1986). It searched for food suggested that Kolpanamas could have had enhanced tactile using a position similar to that of the walrus. The head main­ sensitivity of the lips and the muzzle, possibly possessing a tained contact with the bottom while the body was at an angle large upper lip and tactile vibrissae like the walrus. If that was to the bottom, probably close. to the angle of the premaxillary actually the case, then Odobenacetaps also would resemble process with the anteroposterior axis of the skull (-45°). The Kalpanomas, in having both good binocular vision and an en­ tail helped in keeping that position and, as in other cetaceans, larged, highly tactile upper lip. That bottom feeders could rely provided propulsion (as mentioned by Fay (1982: 164), the wal­ upon binocular vision and/or tactile sensibility of the upper lip rus also uses the posterior part of the body to move forward, ~nd vibrissae is in agreement with the apparent lack (or reduc- but by means of the hind limbs). The forelimbs probably had a )n) of echolocational ability in Odabenacetaps. Kalpanamas, stabilizing role, as in the walrus, and would have helped to however. was very different from Odobenacetaps in its mode maintain the upper lip and vibrissae in contact with the bottom. ofingestion ofprey. Kalponamos had verystrong sea-otter-Iike It used its very good dorsal binocular vision and its large and 252 SMITHSONIA\ CONTRIBUTIONS TO PALEOBIOLOGY

FIGURE 20.-Artis't's reconstruction of OdobenocelOps peruvial1l1s in feeding position (original painting by Mary Parrish, Department of Paleobiology, NMNH). FIGURE 19.-0dobenocelops peruvianus. two possible reconstructions of the head: a. hypothesis without vibrissae, a "cetacean-like" interpretation; b. hypothesis with vibrissae, a "walrus-like" and more speculative intcrpretation 1982; Martin, 1990; Heyning and Mead, 1996). An excellent (sculptures by Mary Parrish, Department of Paleobiology, NMNH). review of the topic was presented by Werth (1992). According to Werth, it is in fact the absence of suction feeding that is un­ highly sensitive upper lip, possibly assisted by strong vibrissae, usual among odontocetes. Werth reported suction feeding in a to search for food (F igures 18-20). Most probably, little or no number of species, both through direct observation: Delphi­ echolocation was involved. Once found, buried prey, such as napterus lel/cas and Globicephala melaena; and inferred from clams, was probably excavated by water-jetting, as in the wal­ anatomical or ecological evidence: Berardius bairdi, Mesopl­ rus, then caught in the anterior part of the mouth and strongly odon layardi, Monodon monocdos, Phocoena phocoena, Pho­ held in this position by the powerful lips and jaws. Then dors­ coenoides dalli, Physeter macrocephalus, and Tursiops trunca­ oventral and/or anteroposterior movements of the tongue tus. Werth (1992:37) also reported a personal communication of W. Walker describing "a Tursiops stomach full of fresh, un­ would have transformed the mouth into a sort of vacuum digested siphons removed from clams; this type ofdiet (and in­ pump, and part of the prey was sucked out of its shell and in­ gested. ferred feeding method) is strikingly similar to that of suction­ Consequently, Odobenocetops was a highly specialized ceta­ feeding walruses." The observations made by Werth and others c.ean that com~ined adaptations of both Odobenus (water-jet­ make the feeding adaptation inferred for Odobenocetops peru­ vianus much more likely than could be suspected at first tmg and suckmg) and Kolponomos (good binocular vision), and therefore was probably the best-adapted known bottom­ glance. feeding marine carnivore. I Possible Functions of the Tusks and Alveolar Processes This proposed feeding scenario is not unexpected for an od­ \, ontocete. Suction feeding has been reported many times in od­ The most striking and unexpected features ofOdobenocetops l' ontocetes (Kleinenberg, 1938; Wilke et aI., 1953; Ray, 1966; are its tusks and the massive alveolar processes of the premax­ I Rae, 1973; Matthews, 1978; Crawford, 1981; Seaman et aI., illae that support them~features paralleled in no other ceta- 1 ~,J NUMBER 93 253 cean. These structures, in our opinion, embody the twofold im­ enlarged as in Odobenus (Horikawa, 1995). At least the begin­ portance of Odobenocetops: Not only does the discovery of nings of the evolution of suction feeding thus apparently pre­ this new taxon dramatically extend the known morphological ceded tusk enlargement. The tusk of Protodobenus is open­ and ecological diversity of the Cetacea, but it also reopens the rooted, however, and apparently ever-growing, so it could be question ofthe function of tusks in true walruses. argued that this taxon also fulfills the prediction of Fay's hy­ After centuries ofspeculation and study, a consensus appears pothesis that tusk enlargement should have followed immedi­ to have emerged that walruses use their tusks primarily in so­ ately upon a shift away from piscivory. cial (mostly agonistic) interactions, and secondarily for a vari­ In Odobenocetops, tusk enlargement appears to have been ety of other purposes, but not to any significant extent in the correlated with the adoption ofspecifically Odobenus-like suc­ course of their normal bottom-feeding (Fay, 1982; Nelson and tion feeding, i.e., benthic suction feeding that involved continu­ Johnson, 1987; Kastelein and Mosterd, 1989; Kastelein et aI., ous direct contact of the snout with the substrate, possibly with 1991). In other words, according to this view, walrus tusks are major dependence on tactile vibrissae. If, as believed by stu­ not feeding adaptations and do not form an integral part of the dents of living walruses, this mode of feeding does not involve functional complex offeeding adaptations. any use of, nor necessitate any enlargement of, the tusks, then All who have examined the skull of Odobenocetops have the co-occurrence of benthic suction feeding and large tusks in been immediately struck by its gross resemblance to that of a both walruses and Odobenocetops is purely coincidental. If, on walrus, most obviously in regard to the tusks. Moreover, this the other hand, this "explanation" is deemed unparsimonious, study has confirmed that the feeding behavior of Odobenoce­ then we should look more closely for a functional connection tops was walrus-like in that it consisted mainly ofwater-jetting between feeding and tusk enlargement in the modern walrus. and suction. This, however, raises a paradox: If walrus tusks There is a wide range of possible functions for the tusks of are not feeding adaptations, then why should an animal whose Odobenocetops. Because the alveolar processes that support feeding adaptations converge on those ofa walrus be expected them are even more prominent than the tusks themselves, and to have walrus-like tusks? differ from themin probably having been more nearly symmet­ Fay (1982: 137-138) proposed a plausible explanation of rical, it is worthwhile to consider their possible functions sepa­ how walruses evolved large tusks. He hypothesized that "posi­ rately. tive selective pressures and the potential for tusk development Use ofTusks in Feeding: PRO: Tusks are teeth and primi­ probably have existed in all polygynous pinnipeds from the be­ tively serve for food-gathering. Even a single, asymmetrical ginning," but that the functional demands of pelagic piscivory, tusk directed downward and backward could be used for stab­ which required an unobstructed gape, precluded enlargement bing prey, and walruses may kill seals on occasion (Fay, of the tusks beyond a certain point until odobenids took up 1982: 153). On the other hand, most, if not all, dental special­ feeding on benthic mollusks. Freed from the opposing selective izations for feeding are symmetrical, but the tusks are asym­ pressure for a large gape, the animals were then able to respond to the preexisting social selection for hypertrophied tusks. metrical. Asymmetry of feeding structures may occur in ceta­ This explanation, however, would probably not apply to ceans. Fin whales have asymmetrical color patterns around the Odobenocetops. Early odontocete cetaceans, unlike pinnipeds, mouth that are possibly associated with feeding (Gamble, did not have enlarged caniniform teeth with obvious "potential 1985). Tursiops preferentially bottom-feed on the right side and for tusk development." Moreover, although some odontocetes have an asymmetrical larynx (with the piriform recess wider on that have become specialized for suction feeding have devel­ the right) that may make it easier to swallow with that side oped tusk-like upper or lower teeth (narwhals, ziphiids), others down (Joy S. Reidenberg, pers. comm., 1995). have developed similar specializations without any enlarge­ CON: The tusks are asymmetrical; most, ifnot all, dental spe­ ment of teeth (belugas); "selective pressures for tusk develop­ cializations for feeding are symmetrical. Asymmetry offeeding ment" thus seem not to have existed in all odontocetes. Fur­ structures in cetaceans is not yet conclusively demonstrated thermore, if any of the suction-feeding specializations in and is not known to include the dentition. belugas, narwhals, and Odobenocetops were inherited from a Tusks as Ballast: PRO: The relatively dense tusks, by add­ common ancestor, they obviously did not coincide in time of ing extra weight to the front ofthe head, may have helped keep evolution with any enlargements of teeth, because tusks are ab­ the snout against the bottom during feeding. Walrus tusks and sent in the first and nonhomologous in the other two. It ap­ their supporting bone also may serve this function. pears, then, that any net social selection for tusk enlargement in CON: If this was the primary function of the tusks, it is hard odobenocetopsids did not result directly from, nor was it corre­ to explain why they were not equal in size. Also, it seems un­ lated with, a shift from primitive piscivory to suction feeding. likely that tusks as small as those of Odobenocetops would Th is also may have been true in walruses. Protodobenus have had a significant ballast effect in an animal that could japonicus, a new odobenine from the early Pliocene ofJapan, have approached the mass ofa narwhal (between 800 and 1000 seems to have been principally a piscivore with incipient adap­ kg). The much larger tusk ofa narwhal is not known to produce tations for suction feeding, but its upper canine is not nearly so (either facultatively or obligately) a down-by-the-head attitude 254 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY in that animal, even though it has a much greater moment for Alveolar Processes as Hydrofoils or Diving Planes: PRO: doing so than would have been the case in Odobenocetops. The processes were prominent features of the head, at the very Use ofTusks for Hauling Out: PRO: This is an important front end ofthe body where control surfaces would be most ef­ function oftusks in walruses. fective. Their oblique orientation resembles that of cetacean CON: Post-Eocene cetaceans do not haul out. Tusks used in flippers. The bottom-feeding behavior ofOdobenocetops might hauling out would be most useful ifthey were symmetrical and well have benefited from hydrofoils to keep the snout pressed stronger than observed in Odobenocetops. against the substrate. Use of Tusks in Piercing, Abrading, or Anchoring to Ice: CON: The processes may have been too small to have much PRO: Both walruses and narwhals use their tusks to make effect as normal hydrofoils, especially if Odobenocetops was a breathing holes, and walruses use them to anchor to ice (Fay, relatively slow-swimming cetacean. Their angle ofattack was 1982:137). not easily adjustable, so they would not have been as effective CON: Sea ice did not exist in Peru in the early Pliocene. Al­ as the flippers. Other cetaceans manage without such control though Odobenocetops may have lived in the Antarctic, it has surfaces on the head. In cross section, the processes are thinner not yet been found there. in front and thicker behind, hence not like that ofan airfoil, so Social Role ofTusks: PRO: In many species tusks play so­ they would not have generated significant lift, even if they cial roles, such as in combat, in courting or mating, and in vi­ could have been held in a transverserather than an oblique po­ sual display. This is true both in walruses and in narwhals, the sition. In bottom feeding, the alveolar processes would have sister group of Odobenocetops. in which they also are asym­ lain against and parallel to the bottom (and hence in the same metrical. In walruses, they are used for ritualized dominance­ plane as the direction ofadvance) and would have had minimal threat displays by both males and females (Fay, 1982: effectiveness as hydrofoils. Moreover, their cross-sectional 135-136). Use for combat would be consistent with the tusks' shape would, in that position, have tended to lift the head off orientation, which is suitable for a slashing attack to an oppo­ the bottom rather than the contrary. If muscles or skin inserted nent's flank, using powerful muscles attached to the paroccipi­ on their posterior edges (see below), the processes plus at­ tal processes. Ifthe opponent were always approached head-on .tached soft tissues might have had the size and shape necessary (right side to right side), this also would be consistent with the to function as hydrofoils, but the other objections would still tusks' asymmetry. Use for social roles also would predict that apply. the tusks would be larger in some individuals than in others. Alveolar Processes as Ballast: PRO: The extra mass of CON: The sex ofthe available specimen and the degree of in­ bone in the process, by adding weight to the front of the head, trapopulational variation in the size of tusks are unknown. The may have helped keep the snout against the bottom during tusks may have had a role in actual contact with other animals, feeding. Walrus tusks and their supporting bones also may but they seem rather slender to have been used for extremely serve this function. forceful contact. Neither do their downward-and-backward di­ CON: The bone of the processes is not denser than the rest of rection, unilateral enlargement, and slenderness seem opti­ the skull, unlike the bone in the rostrum ofthe walrus. Further­ mized for an impressive visual display, especially under water. more, as is true for the tusks, the mass ofthe alveolar processes Tusks as Primitive Retentions: PRO: The tusks may not was not great when compared with that ofthe animal. have been adaptive in Odobenocetops itself, but merely re­ Use ofAlveolar Processes for Sediment Displacement: PRO: tained from an earlier evolutionary stage. The backward divergence of the processes would tend to pro­ CON: The weight, location, and external form of the tusks duce a plowshare effect during feeding, shoving sediment to would probably have had deleterious hydrostatic and hydrody­ the sides and increasing the effective width of the path namic effects on the animal's behavior and locomotion. Pre­ searched for prey. Cornified skin on the leading edge of the sumably, they would therefore have been eliminated quickly by processes would be consistent with such use. selection in the absence ofsome positive selective value. CON: The dorsolaterad slope (in head-down position) of the Alveolar Processes as Supportfor the Tusks: PRO: Any use lateral surface ofthe processes is unlike the ventrolaterad slope ofthe tusks would likely result in bending stresses, which sup­ of a plowshare, and would press sediment downward rather porting sheaths·would help resist. than digging in and lifting it. Increasing by this means the area CON: The alveolar processes were probably relatively sym­ searched for prey would presuppose the presence of vibrissae metrical, whereas the tusks are grossly asymmetrical. An along the whole length ofeach process, which is questionable unerupted tusk could not be used for anything and therefore (see below). would require no extra support. Walruses use their tusks very Use ofAlveolar Processes for Muscle Attachment: PRO: forcefully in a variety of ways and do not have such elongate The processes would have provided highly effective lever arms bony sheaths to support them. Rather, they have been strength­ for muscles (such as a modified platysma and/or cutaneus ened by being made thicker. In Odobenocetops they are rela­ trunci) that flexed the neck and/or turned the head, both impor­ tively slender, implying relatively little selection pressure for tant actions in bottom feeding and in most of the conceivable resistance to bending. uses of the tusks. The faint ridge on the posterior edge of the JMBER 93 255 process might represent the insertion of a muscular aponeuro­ nids with "walrus-like" morphotypes have analogous struc­ sis. tures, i.e., long, symmetrical tusks with oval cross sections. CON: The large ventrolateral expansions of the occipital re­ CON: No contrary evidence is known. gion would seem adequate for neck-muscle attachments, and Use 0/Alveolar Processes in Combat: PRO: The large alve­ such anterior muscle insertions are absent in walruses. olar processes might have served a function in combat between Use 0/Alveolar Processes for Skin Attachment: PRO: The males. This would be consistent with a social role for the tusks. faint ridge on the posterior side of each alveolar process could The downward, backward, and lateral slope of the processes yen more plausibly mark the attachment ofa sheet ofskin ex­ would serve, in a head-on collision, to guide an opponent's al­ lending backward from the process. Such a sheet might have veolar process and tusk ventrolaterad and away from the eye served as a barrier to keep churned-up mud and turbid water and flank. Cornified skin on the leading edge of the process away from the eye during head-down bottom feeding. Al­ would be consistent with such use. though Odobenus lacks such a sheet ofskin (and a bony strut to CON: The sex ofthe available specimen is unknown. support it), it depends less upon vision during feeding, as sug­ Use 0/Alveolar Processes in Visual Display: PRO: This, gested by its more lateral placement of the eyes compared with too, would be consistent with a social role for the tusks. The Odobenocetops. Walruses do sometimes use their vision in for­ processes are bulkier than the tusks themselves, and they would aging, but they do not always do so (Kastelein et aI., 1993). have accentuated and called attention to the presence and ori­ CON: No contrary evidence is known. entation of the tusks. In head-on view, they also would have Alveolar Processes as Restricting Size 0/Mouth Opening/or made the entire animal look larger. Use for display would pre­ Suction Feeding: PRO: In walruses, known suction feeders, dict that the processes would be larger in some individuals than the short bony sheaths of the tusks form rigid sides for the in others. mouth opening and help concentrate the suctional force (see CON: The sex of the available specimen and the degree ofin­ Fay, 1982: 171, fig. 106). trapopulational variation in the size of the processes are un­ CON: This also would be true for the most proximal parts of known. the alveolar processes of Odobenocetops, but not the distal Alveolar Processes as Primitive Retentions: PRO: Enlarged parts, and so would not explain the great elongation of the pro­ alveolar processes of the tusk-bearing bones are a necessity in ·~sses. Both tusks of Odobenocetops seem large enough to animals with enlarged tusks. If tusks (however oriented) were have served this function even ifsupported only by short, wal­ present in the ancestors of Odobenocetops, enlarged alveolar rus-like processes. processes also would have been present (as in narwhals); Alv(?olar Processes as Increasing Area and/or Breadth 0/ hence, enlargement of the processes may call for no special Vibrissal Array: PRO: The symmetry and backward diver­ adaptive explanation. gence of the processes increase the total width of the snout, CON: For the reasons stated above, support for the tusks which could, therefore, support an increased number of vibris­ alone seems inadequate to account for the size ofthe processes. sae. Cornitied skin on the leading edge ofthe process would be In any case, the enlarged processes of Odobenocetops relative consist<:11l with such use. to the rest ofthe skull are derived with respect to all other ceta­ CON: Large nutrient foramina are found only on the proxi­ ceans, and this enlargement calls for explanation. mal, not the distal, portions of the alveolar processes (however, We conclude that the most plausible and important uses of these foramina were not necessarily coextensive with the the tusks ofOdobenocetops were social ones. The alveolar pro­ vibrissae). Moreover, it isnot certain that vibrissae were in fact cesses probably also played social roles, incidentally as a sup­ present. port for the erupted tusk and perhaps more directly as weapons Alveolar Processes as Orientation Guides and Stabilizers/or and/or shields in agonistic encounters, or in visual displays. the Mystacium: PRO: More-or-less symmetrical tusk-like Other functions, such as restricting the mouth opening for suc­ bony processes could serve as guides, somewhat in the manner tion feeding or supporting an array of vibrissae, cannot be en­ ofsled runners, to maintain the proper orientation of the myst- tirely ruled out and may even have been important in the early cium and vibrissal array to the substrate while the animal stages of evolutionary enlargement of the processes. Likewise, searched for food and to reduce the exertion of neck muscles. attached sheets of skin might have been useful in keeping sus­ In Odobenocetops. the symmetry and backward divergence of pended sediment out of the field of vision in later stages of the the processes would have given them more leverage in counter­ processes' enlargement. At all stages of evolution, the in­ ing roll and yaw, as well as pitch. The right tusk, at least, would creased skeletal mass represented by both tusks and processes have enhanced this leverage (but the advantage of this must also would have had an incidental value in keeping the snout hl1\"e been small, or else both tusks would have been ofsimilar against the substrate during feeding. length). The cross-sectional shape ofeach process would have We suggest, however, that the single most important function caused it to ride over sediment rather than digging in, likewise of the alveolar processes, and the one that may have controlled ;tabilizing the head against roll. Cornified skin on the leading their evolution, was that of orientation guides for the myst­ edge of the process would be consistent with such use. Odobe- acium and vibrissal array. Tbis is the only hypothesis that 256 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY seems to offer an explanation for the co-occurrence of benthic role. In fact, it is difficult to explain them merely as sheaths for suction feeding and tusk-like structures in both Odobenocetops the tusks, because walruses manage to wield much larger tusks and walruses. Indeed, one of the most significant implications very forcefully without the support of such elongated sheaths, ofthe discovery ofOdobenocetops is the fact that it reopens the and relatively symmetrical sheaths should not be needed for the question ofwhether the tusks ofwalruses are important in their highly asymmetrical (and, in the case of the unerupted tusk, feeding strategy. functionless) teeth of Odobenocetops. We therefore posit a As noted above, the work of Fay (1982) and others seemed functional role for the alveolar processes independent of the to have answered this question in the negative, and the consen­ tusks themselves. sus today is that walrus tusks serve primarily in social behav­ This still leaves unexplained the evolution of tusks in Odo­ ior. In addition to their subsidiary functions, such as hauling benocetops. Given their asymmetry and the unlikelihood of out and opening holes in ice (Fay, 1982), perhaps walrus tusks functions such as hauling-out or ice-breaking in the case of a also serve as orientation guides for the mystacium and vibris­ cetacean living in Peru, it is probably safe to make an analogy sae. An analogy can be drawn with sled-mounted undersea with narwhal tusks and attribute to them a primarily social cameras that are designed to be dragged along the seafloor; the function. It would not be surprising if sister groups such as sled runners stabilize the sled in roll, pitch, and yaw, keeping monodontids and odobenocetopsids evolved such analogous the lights and cameras pointed in the right direction. Similarly, structures in parallel, albeit developing the tusks from canines tusk-like structures would stabilize an animal's head and help and incisors, respectively. maintain the mouth and vibrissal array in a fixed attitude rela­ tive to the bottom, increasing the efficiency of search by the Conclusions sense organs and, possibly, the efficiency and accuracy of wa­ ter-jetting and/or suction feeding. It appears, then, that the resemblance between Odobenoce­ If such orientation guides are important in the odobenine tops and walruses is not merely superficial, but a genuine case style of bottom feeding (and it should be possible to test this offunctional convergence in feeding adaptations-specifically, experimentally in living walruses), then we would expect both adaptations for suction feeding on infaunal benthic prey. Fur­ sexes to have them (as is the case with the tusks of Odobenus). thermore, from the fact that structures resembling walrus tusks The external form of such guides rather than their internal evolved in an animal clearly convergent on walruses in other structure would be functionally important, however, so they characters that are unquestionably feeding adaptations, we can would not necessarily have to be genuine tusks; for example, surmise that the tusk-like structures are probably associated tusk-like bony structures would serve just as well, provided with feeding in both animals. Otherwise, the correlation of they were reasonably symmetrical and sufficiently long. Such benthic suction feeding with tusk-like structures in both ani­ structures would make real tusks redundant, and vice versa, so mals is merely a striking coincidence-a conclusion we regard we would not expect to find both long, symmetrical tusks and as unparsimonious. bony equivalents ofthem in the same species. This is borne out Apart from the tusks and alveolar processes, the features in by fossil odobenines such as Vcilenictus (Demere, 1994), which which Odobenocetops resembles a walrus are its hourglass­ resemble modern Odobenus in this regard-although this shaped skull, vaulted palate, enlarged paroccipital processes, would of course be expected from their close relationship possible vibrissae and sensitive upper lip, and possible horny alone. covering ofpart ofthe upper lip. These features form a charac­ This hypothesis also is consistent with the ontogeny ofwal­ ter complex apparently unique to odobenids and odobenoce­ rus tusks. Weaning usually occurs in Odobenus at 14-27 topsids, and in the former are functionally related to a style of months, and benthic feeding begins at 6-24 months (Fay, feeding documented in no other animals: powerful suction 1982: 138-141). Tusk eruption also begins in the first year of feeding on infaunal benthic prey. This character complex is life; the tusk begins to show signs ofdiscoloration and wear by wholly or partly absent in animals that prey on benthic inverte­ the age of two years, and at this time the tip of the crown ex­ brates but are not suction feeders (e.g., sea otters), as well as in tends below the ventral side of the mandible and usually 2-4 ones that are suction feeders but prey on animafs that are free­ cm below the edge of the upper lip (Fay, 1982:105-107, fig. swimming or merely rest upon the bottom (e.g., many odonto­ 71). It therefore seems possible that the tusks could be starting cetes; A. Werth, pers. comm., 1992). It is thus only in associa­ to function as guides even at this age. tion with this "benthic suction feeding" character complex that What we find in Odobenocetops also is consistent with this structures resembling walrus tusks are found-specifically in interpretation. Although tusks are present, they are not sym­ derived odobenines and in Odobenocetops. metrical; indeed, only one of the pair may have erupted. In­ If, instead, we attempt to explain the tusk-like structures in stead we find elongated, and probably much more symmetrical, each case as having evolved mainly for social functions, Vvc alveolar processes of the premaxillae, which might very well face an obvious problem. There is no apparent reason why have served as orientation guides for the snout. Their posterior fights or displays using such structures should occur always divergence would have enhanced their effectiveness in this and only in benthic suction feeders. Moreover, it would be very --;UMBER 93 257

surprising if both an odontocete and a pinniped were to evolve however, and the difficulty of applying the same social expla­ closely similar social displays, rituals, or modes offighting that nation to this otherwise very different animal, we propose that were used in and out of water, respectively, by animals of such the utility ofthe tusk-like structures in both cases was primarily very different body form. in feeding (as orientation guides for the mystacium) and only If Fay (1982: 137) is correct in attributing a social selective secondarily in social interactions. In other words, we consider pressure for tusk enlargement to all polygynous pinnipeds (and the tusk-like structures to be an integral part of the "benthic this may well be true), then we might expect to find conver­ suction-feeding" character complex, because we see no other ;ences on the walrus morphotype among other pinnipeds if we plausible explanation for this striking correlation ofcharacters. found them anywhere-yet this is not the case. We think Fay is Students ofmodern Odobenus should therefore reexamine its correct in arguing that benthic suction feeding was the key feeding behavior with the possibility in mind that the tusks may adaptive shift that led to the enlargement of tusks in walruses. serve as orientation guides, as suggested herein. This hypothe­ In view of the similar morphology seen in Odobenocetops, sis should be amenable to experimental tests.

Addendum

This paper was already in the editorial process when three The morphology ofthe snout of0. leptodon differs from that new skulls of Odobenocetops were collected from the Pisco of o. peruvianus in being more rounded and wider in dorsal Formation, which justified the publication of a preliminary view and in lacking large premaxillary foramina. Furthermore, note (Muizon et aI., 1999). the dorsal face of the premaxilla bears a fossa for a premaxil­ One of the skulls (SMNK 2491), referred to O. peruvianus, lary sac, and a pair ofsupplementary rostral bones is present at is from the Sud-Sacaco locality and from the SAS horizon (as the anterodorsal apex ofthe snout. The anterodorsal edge ofthe was the holotype of O. peruvianus), which is earliest Pliocene orbit is only slightly concave in O. leptodon. whereas it is in age. It is almost symmetrical and bears two small tusks; al­ deeply notched in o. peruvianus. The presence ofpremaxillary though the right is slightly larger than the left, it is still drasti­ sacs (which were probably absent in O. peruvianus) is probably cally smaller than the large tusk of the holotype. The alveolar an indication of the presence of a melon in o. leptodon. This sheaths are relatively small and ofthe same size. The two tusks organ, related to echolocation, was probably absent or vestigial are incomplete and their apices are missing. The right sheath is in 0. peruvianus. As stated above in the text, the inferred ab­ partly damaged at its apex, but because the preserved portion of sence of a melon in O. peruvianus was probably compensated the tusk is distinctly longer than the sheath as preserved, it is for by good anterodorsal binocular vision. Muizon et al. (1999) clear that the tusk was erupted. The left sheath and tusk also are concluded that binocular vision was either reduced Or absent in incomplete, and the tusk is broken in the alveolus. It is there­ 0. leptodon and that this was compensated for by echolocation fore not possible to determine whether the left tusk was abilities inferred from the probable presence ofa melon. erupted. In the narwhal the unerupted tusks of the female are The three new skulls have at least one tympanic and one pe­ generally similar in size and definitely smaller than the large riotic in situ. The characteristic morphology ofthese bones (es­ left tusk of the male, so this new skull of Odobenocetops peru­ pecially the massiveness of the anterior process ofthe periotic) vianus (SMNK 2491) was identified as a female. definitely confirms the referral of the periotics and tympanic The other two skulls (SMNK 2492 and MNHN SAO 202) described above to Odobenocetops peruvianus. come from the locality of Sacaco in the SAO horizon, which is The holotype skull of Odobenocetops leptodon was associ­ slightly younger than the SAS horizon, in which the O. peruvi­ ated with its atlas. The only occipital ~ondyle preserved on the anus specimens were found. They were referred by Muizon et skull is damaged, which makes it difficult (but not impossible) al. (1999) to a different and new species, O. leptodon. The ho­ to evaluate the position of the head relative to the axis of the :otype (SMNK 2492) is a partially damaged skull that retains body. An extrapolation could be done using the new skull of O. both tusks in situ. The right tusk is more than 1.35 m long as peruvianus (SMNK 2491), which has well-preserved condyles. preserved (the apex was broken and worn during life), and its The position ofthe atlanto-occipital articulation ofOdobenoce­ erupted portion measures 1.07 m. The left tusk is small and tops indicates that in swimming position the head was bent slender and was approximately 25 cm long. Its apex bears a ventrally, bringing the dorsal plane ofthe skull into an antero­ distinct wear facet, which indicates that the tooth was erupted. dorsal orientation. With the head in this position, the long tusk The presence of a very long right tusk and of an erupted left was almost parallel to the long axis ofthe body. This interpreta­ Tusk in 0. leptodon is an indication (not a proof) that this con­ tion is compatible with the length of the large tusk, as it is un­ dition also could be present in 0. peruvianus. This has to be likely that such a long appendage could be at an angle ofabout I. I ·;onfirmed by the discovery of new specimens of O. peruvi­ 45° to the body during swimming (Muizon et aI., 1999, fig. 2). I anus, however. The head of Odobenocetops peruvianus in Figure 20 therefore L T 258 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY I

should probably be bent slightly ventrally to make the long right being only slightly larger than the left. This condition is tusk parallel to the axis ofthe body. likely to have also occurred in O. leptodon, although this has to I To conclude, the new specimens of Odobenocetops indicate be confirmed by additional specimens. the following: 4. The new specimens confirm referral to O. peruvianus of I. The long right tusk of0. peruvianus was probably longer the periotics and tympanic described above. than initially thought, although this needs to be confirmed by 5. During swimming, the head ofboth species was bent ven­ the discovery ofnew specimens ofthis species. trally and, in such a position, the long tusk was almost parallel 2. The small left tusk was probably erupted, but this also has to the axis of the body. Odobenocetops peruvianus (which has to be confirmed by the discovery ofnew specimens. a deeply notched anterodorsal edge of the orbit) therefore had 3. The female of O. peruvianus had two small tusks, the good dorsal (anterior) binocular vision when swimming.

Literature Cited

Barnes, L.G. Dal Piaz, G. 1984. Fossil Odontocetes (Mammalia, Cetacea) from the Almejas Fonna­ 1916. Gli odontoceti del Miocene Bellunense; Squalodon. Memorie deli 'l,'­ tion, Isla Cedros, Mexico. PaleoBios. 42: 1-46. tituti Geologico delia (R) Universita di Padova, part 2, 4: 1-79. Best, R.C., and Y.M.F. da Silva Demere, T.A. 1989. Amazon RiverDolphin, Boto: Inia geoffrensis (de Blainville, 1817). 1994. Two New Species of Fossil Walruses (Pinnipedia: Odobenidae) In S.H. Ridgway and RJ. Harrison, editors, Handbook ofMarine from the Upper Pliocene San Diego Fonnation, California. Proceed­ Mammals. 4: River Dolphins and Larger Toothed Whales. pages ings ofthe San Diego Society ofNatural History, 29:77-98. 1-23. London: Academic Press. Diercks, KJ., R.T. Trochta, C.F. Greenlaw, and W.E. Evans Boenninghaus, Georg 1971. Recording and Analysis of Dolphin Echolocation Signals. Journal 1904. Das Ohr des Zahnwales, zugleich ein Beitrag zur Theorie der Schal­ ofthe Acoustical Society ofAmerica, 49: 1732-1732. leitung: Eine biologische Studie. Zoologische Jahrbiicher; Abteilung Domning, D.P. fiir Anatomie und Ontogenie del' Tiere. 19(2): 189-360. 1989. Fossil Sirenia ofthe West Atlantic and Caribbean Region, III: Xeno­ Bourdelle, E., and P.-P. Grasse siren yucateca, gen. et sp.nov. Journal ofVertebrate Paleontolog:.; 1955. Ordre des Cetaces (Cetacea Brisson 1762; Cete Linne 1758). In P.-P. 9(4):429-437. Grasse, editor, Traite de Zoologie. Anatomie, Systematique. Biolo­ Domning, D.P., C.E. Ray, and M.C. McKenna gie, 17(1):341-450. Paris: Masson. 1986. Two New Oligocene Desmostylians and a Discussion of Tethythe­ Brill, R.L, M.L. Sevenich, TJ. Sullivan, J.D. Sustman, and R.E. Witt rian Systematics. Smithsonian Contributions to Paleobiology. 59: 1988. Behavioral Evidence for Hearing through the Lower Jaw by an I-56. Echolocating Dolphin (Tursiops truncatus). Marine Mammal Sci­ Donner, KJ. ence. 4(3):223-230. 1979. Mechanism of Sound Production and Air Recycling.in Delphinids: Brodie, P.F. Cineradiographic Evidence. Journal ofthe Acoustical Society of 1989. The White Whale: Delphinapterus leucas (Pallas, 1776). In S.H. America. 65:229-239. Ridgway and RJ. Harrison, editors, Handbpok ofMarine Mammals, Evans, W.E. 4: River Dolphins.and Larger Toothed Whales, pages 119-144. Lon­ 1967. Special Acoustic Problems in Cetacean Research, General Discus­ don: Academic Press. sion. In W.N. Tavolga, editor, Marine Bio-Acoustics, 2:395-398. Bro\\·nell, R.L. New York: Pergamon Press. 1989. Franciscana: Pontoporia blainvillei (Gervais et d'Orbigny, 1844). In Evans, W.E., and P.F.A. Maderson S.H. Ridgway and RJ. Harrison, editors, Handbook ofMarine 1973. Mechanisms of Sound Production in Delphinid Cetaceans: A Re­ Mammals. 4: River Dolphins and Larger Toothed Whales, pages view of Some Anatomical Considerations. American Zoologist. 45-67. London: Academic Press. 13:1205-1213. Carriol. R.P., C. de Muizon, and S. Secretan Evans, W.E., andJ.H. Prescott 1986. Les cnistaces (Cirripedia,becapoda) du Neogene de la fonnation 1962. Observation ofthe Sound Capabilities ofthe Bottlenose Porpoise: A Pisco (Perou). Annales de Paleontologie. 73(3):137-164. Study of the Whistles and Clicks. Zoologica, 47(3):121-126. Cheneval, J. Fay, F.H. 1993. L'avifaune Mio-Pliocene de 1a fonnation Pisco (Perou): Etude 1982. Ecology and Biology ofthe Pacific Walrus, Odobenus rosmarus di­ pniliminaire. Documents des Laboratoires de Geologie de la Fac­ vergens Illiger. North American Fauna. 74: 1-279. Washington, ulte des Sciences de Lyon. 125:85-95. D.C.: U,S. Department of Interior, Fish and WildlifeService. Colbert, E.H. Fordyce, R.E. 1944. A New Fossil Whale from the Miocene of Peru. Bulletin of the 1994. Waipatia maerewhenua. New Genus and New Species (Waipati'idae, American Museum ofNall/ral History, 83(3): '195-216. New Family), an Archaic Late Oligocene Dolphin (Cetacea: Odon­ Crawford, T.W. toceti: P1atanistoidea) from New Zealand. Proceedings ofthe San 1981. Vertebrate Prey of Phocoenoides dalli (ball's Porpoise) Associated Diego Society ofNatural History. 29: 147-176. with the Japanese High Sea Salmon Fishery in the Northern Pacific Fraser, F.C., and P.E. Purves Ocean. Master's thesis, University of WaShington, College of Fish­ 1960. Hearing in Cetaceans: Evolution of the Accessory Air Sacs and ....• eries, Seattle. Structure and Function ofthe Outer and Middle Ear in Recent Cet~- JMBER 93 259

ceans. Bulle/in of/he British Museum (Na/ural Histol)'), Zoology, Kastelein, R.A., R.C.V.J. Zweypfenning, H. Spekreijse, J.L. Dubbeldam, and 7(1):1-140. E.W. Born Fried, W.A., P.E. Nachtigall, and P. W.B. Moore 1993. The Anatomy ofthe Walrus Head (Odobenus rosmarus), Part 3: The 1990. Taste Reception in the Pacific Bottlenose Dolphin (Tursiops trunca­ Eyes and Their Function in Walrus Ecology. Aqua/ic Mammals. Ius gilli) and the California Sea Lion (Zalophus californianus). In 19(2):61-92. J.A. Thomas and R.A. Kastelein, editors, Sensory Abilities ofCe/a­ Kellogg, R. ceans: Labora/ory and Field Evidence, pages 447-454. New York: 1923. Ken/riodon pernix. a Miocene Porpoise from Maryland. Proceed­ Plenum Press. ings of/he Uniled States Na/ional Museum. 69(19):1-55. ·-c:lmble,R. 1926. Supplementary Observations on the Skull of Zarhachis flagella/or 1985. Fin Whale Balaenoptera physalus (Linnaeus, 1758). In S.H. Ridg­ Cope. Proceedings ofthe Uniled States Na/ional Museum. 67(28): way and R.J. Harrison, editors, Handbook ofMarine Mammals, 3: 1-18. Sirenians and Baleen Whales, pages 171-192. New York: Academic Ketten, D.R. Press. 1984. Correlations of Morphology with Frequency for Odontocete Co­ Gao, G., and K. Zhou chlea: Systematics and Topology. 335 pages. Doctoral dissertation, 1995. Fiber Analysis of Vestibular Nerves in Small Cetaceans. In R.A. Johns Hopkins University, Baltimore, Maryland. Kastelein, J.A. Thomas, and P.E. Nachtigall, editors, Sensory Sys­ 1992. The Marine Mammal Ear: Specializations for Aquatic Audition and tems ofAquatic Mammals, pages 447-453. Woerden, The Nether­ Echolocation. In D. Webster, R. Fay, and A. Popper, editors, The lands: De Spil Publishers. Evolu/ionary Biology ofHearing. pages 717-754. New York: Gordon, K.R. Springer-Verlag. 1984. Models of Tongue Movement in the Walrus (Odobenus rosmarus). Ketten, D.R., and D. Wartzok Journal ofMorphology, 182: 179-196. 1990. Three-dimensional Reconstructions ofthe Dolphin Ear. In J.A. Tho­ Gray, O. mas and R.A. Kastelein, editors, Sensory Abililies ofCe/aceans: 1951. An Introduction to the Study of the Comparative Anatomy of the LaboralOry and Field Evidence. pages 81-105. New York: Plenum Labyrinth. Journal ofLGlyngology and O/ology, 65:681-703. Press. Hay, KA Kleinenberg, S.E. 1980. Age Deternlination ofthe Narwhal, Monodon monoceros. L. In W.F. 1938. [Some Data on the Feeding of Tursiops IUrsio Fabr. in the Black Perrin and e.A. Myrick, editors, Age Determination of Toothed Sea.] Byulleten Moskovskogo Obshches/Va Ispy/atelei Prirody. 0/­ Whales and Sirenians. Repor/s of/he IllIernational Whaling Com­ del Biologicheskii [Bulletin of the Moscow Society of Researchers mission, Special Issue. 3: 119-132. on Nature, Section of Biology], 47:406-413. [In Russian.] Hay, K.A., and A.W. Mansfield Kleinenberg, S.E., AV. Yablokov, B.M. Bel'kovich, and M.N. Tarasevich 1989. Narwhal Monodon monoceros Linnaeus, 1758. In S.H. Ridgway and 1969. [Beluga (Delphinapterus leucas): Inves/igation on the Species.] 376 R.J. Harrison, editors, Handbook ofMarine Mammals, 4: River Dol­ pages. Academy of Science of the USSR, A.N. Severtsov Institute phins and Larger Too/hed Whales, pages 145-176. London: Aca­ of Animal Morphology. [In Russian. English translation by Israel demic Press. Program for Scientific Translations, Jerusalem.] Heyning, J.E., and J.G. Mead 1996. Suction Feeding in Beaked Whales: Morphological and Observa­ Kuznetzov, V.B. 1990. Chemical Sense of Dolphins: Quasi-Olfaction. In J.A. Thomas and tional Evidence. Contribu/ions in Science. Na/ural HistOlY Museum R.A. Kastelein, editors, Sensory Abililies ofCe/aceans: Labora/ory I!(Los Angeles Coun/y. 464: 1-12. Horikawa. H. and Field Evidence. pages 481-503. New York: Plenum Press. 1995. A Primitive Odobenine Walrus of Early Pliocene Age from Japan. Lawrence, B., and W.E. Schevill The Island Arc. 3:309-328. 1956. The Functional Anatomy ofthe Delphinid Nose. Bulletin ofthe Mu­ 'lowell, A.B. seum ofCompara/ive Zoology. Harvard College. 114(4): 103-151. 1927. Contribution to the Anatomy ofthe Chinese Finless Porpoise Neom­ Lilly, J.e. eris phocaenoides. Proceedings of/he Uniled Slates National Mu­ 1961. Man and Dolphin. 312 pages. Garden City, New Jersey: Doubleday sewlI.70(13):1-43. &Co. 1930. Myology of the Narwhal (Monodon monoceros). American Journal Lilly, J.c., and A.M. Miller ofAna/om;; 46(2): 187-215. 1961. Sounds Emitted by the Bottlenose Dolphin. Science. 5:1689-1693. Jansen, J., and J.K.S. Jansen Lisson, e.1. 1969. The Nervous System ofCetacea. In H.T. Anderson, editor, The Biol­ 1898. Los fosfatos de Ocucaje. Boletin de Minas. Industria y Construc- ogy ofMarine Mammals. pages 175-252. New York: Academic ciones. 14(5):45-47. Press. Luo, Z., and K. Marsh 1996. Petrosal (Periotic) and Inner Ear of a Pliocene Kogiine Whale Kastelein, R.A., N.M. Gerrits, and J.L. Dubbeldam 1991. The Anatomy ofthe Walrus Head (Odobenus rosmarus), Part 2: De­ (Kogiinae, Odontoceti): Implications on Relationships and Hearing scription of the Muscles and ofTheir Role in Feeding and Haul-Out Evolution of Toothed Whales. Journal of Verlebra/e Paleon/ology, Behaviour. Aqua/ic J\;fammals, 17(3):156-180. 16(2):328-348. Kastelein, R.A., and P. Mosterd Marocco, R., and C. de Muizon 1989. The Excavation Technique for Molluscs of Pacific Walruses (Odo­ 1988. Los vertebrados del Neogeno de Ia costa Sur del Peru: ambiente y bem,s rosmarus divergens) under Controlled Conditions. Aqua/ic condiciones de fosilizaci6n. Bulletin de l'Instilu/ Franryais d'Etudes J[ammals, 15(1):3-5. Andines.27(2):I05-117. Kastelein, R.A.. and P.R. Wiepkema Marsh, H. 1989. A Digging Trough as Occupational Therapy for Pacific Walruses 1980. Age Determination of the Dugong (Dugong dugon (Mliller)) in (OJohenus rosmar;ls divergens) in Human Care. Aqua/ic Mammals, Northern Australia and Its Biological Implications. Repor/s ofthe IS( 1):9-17. In/ernalional Whaling Commission, Special Issue. 3: 181-201. 4p ~1

260 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY !'II

Martin, A.R. 1994. Are the Squalodonts Related to the Platanistoids? Proceedings oJth 1990. The Illustrated Encyclopedia of Whales and Dolphins. 192 pages. San Diego Society ofNatural History, 29: 135-146. e New York: Portland House. Muizon, C. de, and H. Bellon Pi Matthews, L.H. 1986. Nouvelles donnees sur l'age de la formation Pisco (Perou) Co~ .. ,... ,,'ptes 1978. The Natural History ofthe Whale. 219 pages. New York: Columbia Rendus H~bdomadalres des Seances de I Academle des SCiences University Press. (Pans), senes 2, 303(5):1401-1404. McCormick, J.G., E.G., Wever, and J. Palin Muizon, C. de, and T.J. DeVries 1970. Sound Conduction in the Dolphin Ear. Journal ofthe Acoustical So­ 1985. Geology and Paleontology ofLate Cenozoic Marine Deposits in the ciety ofAmerica, 48(6): 1418-1428. Sacaco Area (Peru). Geologisches Rundschau, 74(3):547-563. Mchedlidze, GA, and G. Pilleri Muizon, C. de, and D.P. Domning pr 1988. Kharthlidelphis diceros, a New Genus and Species from the Oli­ 2002. The Anatomy of Odobenocetops (Delphinoidea, Mammalia), the gocene of Georgia (Caucasus, USSR). Investigations on Cetacea, Walrus-like Dolphin from the Pliocene ofPeru and Its Paleobiologi_ 21:9-15. cal Implications. Zoological Journal ofthe Linnean Society, pu Mead,J.G. 134(4):423-452. 1975. Anatomy of the External Nasal Passages and Facial Complex in the Muizon, C. de, D.P. Domning, and M. Parrish Delphinidae (Mammalia, Cetacea). Smithsonian Contributions to 1999. Dimorphic Tusks and Adaptive Strategies in a New Species ofWal_ Zoology, 207:1-72. rus-like Dolphin (Odobenocetopsidae) from the Pliocene of Peru. Miller, M.E., G.C. Christensen, and H.E. Evans Comptes Rendus de I'Academie des Sciences, series 2 (Sciences de Ra 1964. Anatomy ofthe Dog. 941 pages. Philadelphia: W.B. Saunders. la Terre et des Planetes), 329:449-455. Muizon, C. de Murie, J. 1981, Les vertebres fossiles de la formation Pisco (Perou), premiere partie: 1870. On Risso's Grampus, G. rissoanus (Desm.). Journal of Anatomy Ra Deux nouveaux Monachinae (Phocidae, Mammalia) du Pliocene de and Physiology, 5: 118-138. Sud-Sacaco. Travaux de /'Institut Frant;:ais d'Etudes Andines, 22: 1873. On the Organization of the Caaing Whale, Globicephala melas. 160 pages. [Recherches sur les Grandes Civilisations, Memoire 6. Transactions ofthe Zoological Society ofLondon. 8(4):235-301. Ra Paris: A.D.P.F.) Nelson, C.H., and K.R. Johnson 1983a. Pliopontos littoralis, un nouveau Platanistidae (Cetacea) du Plio­ 1987. Whales and Walruses as Tillers of the Sea Floor. Scientific Ameri­ cene de la cote peruvienne. Comptes Rendus Hebdomadaires des can, 256(2):112-117. Seances de I 'Academie des Sciences (Paris), series 2, 296: Nishiwaki, M. 1101-1104. 1972. General Biology. In S.H. Ridgway, editor, Mammals ofthe Sea: Biol­ 1983b. Un nouveau Phocoenidae (Cetacea) du Pliocene inferieur du Perou. ogy and Medicine, pages 1-204. Springfield, Illinois: c.c. Thomas. Comptes Rendus Hebdomadaires des Seances de I'Academie des Norris, K.S. Sciences (Paris), series 2, 296: 1203-1206. 1964. Some Problems in Echolocation in Cetaceans. In W.N. Tavolga, edi­ 1983c. Un Ziphiidae (Cetacea) nouveau du Pliocene inferieur du Perou. tor, Marine Bio-Acoustics, I :317-355. New York: Pergamon Press. Comptes Rendus Hebdomadaires des Seances de I'Academie des 1968. The Evolution ofAcoustic Mechanisms in Cetaceans. In E.T. Drake, Sciences (Paris), series 2, 297:85-88. editor, Evolution and Environment, pages 297-324. New Haven: Rc 1984. Les vertebres fossiles de la formation Pisco (Perou), deuxieme par­ Yale University Press. tie: Les odontocetes (Cetacea, Mammalia) du Pliocene inferieur de 1969. The Echolocation of Marine Mammals. In H.T. Andersen, editor, Sud-Sacaco. Travaux de I'lnstitut Fran,ais d'Etudes Andines, 25: The Biology ofMarine Mammals. pages 391-423. New York: Aca­ 188 pages. [Recherches sur les Grandes Civilisations, Memoire 50. demic Press. Sc Paris: A.D.P.F.) Norris, K.S., KJ. Dormer, J. Pegg, and GJ. Liese 1986. Un nouveau Phocoenidae(Odontoceti, Mammalia) du Miocene 1971. The Mechanism of Sound Production and Air Recycling in Por­ superieur de la formation Pisco (perou). Comptes Rendus Hebdomac poises: A Preliminary Report. Proceedings ofthe Eighth Annual daires des Seances de I'Academie des Sciences (Paris), series 2, Conference on Biological Sonar and Diving Mammals, pages Sc 303(16): 1509-1512. 113-129. Menlo Park, California: Stanford Research Institute. 1987. The Affinities of Notocetus vanbenedeni, an Early Miocene Pla­ Norris, K.S., and W.E. Evans tanistoid (Cetacea, Mammalia) frorn Patagonia, Southern Argentina. 1967. Directionality of Echolocation Clicks in the Rough-tooth Porpoise, Sc American Museum Novitates, 2904: 27 pages. Steno bredanensis (Lesson): In W.N. Tavolga, editor, Marine Bio­ 1988a. Les vertebres fossiles de la formation Pisco (Perou), troisieme par· Acoustics, 2:305-316. New York: Pergamon Press. tie: Les odontocetes (Cetacea, Mammalia) Miocenes. Travau.. de Norris, K.S., and G.W. Harvey /'Institut Fran,ais d'Etudes Andines, 42: 244 pages. [Recherches 1974. Sound Transmission in the Porpoise Head. Journal ofthe Acoustical sur les Grandes Civilisations, Memoire 78. Paris: A.D.P.F.) Society ofAmerica, 56(2):659-:-664. 1988b. Le polyphyletisme des Acrodelphidae, odontocetes longirostres du Oelschlager, H.A., and E.H. Buhl Miocene europeen. Bulletin du Museum National d'Histoire Na­ 1985a. Development and Rudimentation of the Peripheral Olfactory Sys­ turelle, series 4, section C, IO( 1):31-88. tem in the Harbor Porpoise Phocoena phocoena (Mammalia: Ceta­ 19S5c. Les relations phylogenetiques des Delphinida. Annales de Pateon­ cea). Journal ofMorphology, 184:351-360. tologie, 74(4): 159-227. 1985b. Occurrence of an Olfactory Bulb in the Early Development of the 1991. A New Ziphiid (Odontoceti, Mammalia) from the Early Miocene of Harbor Porpoise (Phocoena phocoena L.). Fortschritte del' Zoolo­ Washington State (USA) and a Phylogenetical Analysis of the Ma­ gie, 30:695-698. jor Groups ofOdontocetes. Bulletin du Museum National d'Histoire Pabst, D.A. Naturelle. series 4, section c, 12:279-326. 1990. Axial Muscles and Connective Tissues ofthe Bottlenose Dolphin. In 1993a. Walrus-Like Feeding Adaptation in a New Cetacean from the S. Leatherwood and R.R. Reeves, editors, The Bottlenose Dolphin, Pliocene of Peru. Nature, 365:745-748. pages 51-67. San Diego: Academic Press. 1993b. Odobenocetops peruvianus, una remarcable convergencia de adapt­ Pilleri, G., editor aci6n alimentaria entre morsa y delfin. Bulletin de f'lnstitut 1989. Beitrage sur 'Palaontologie del' Cetaceen Penis. Volume 1. 233 Fran,ais d'Etudes Andines, 22(3):671-683. pages. Ostermundingen [Bern): Hirnanatomisches Institut. r tvlBER 93 261

i 1990. Beitrage sur Palaontologie der Cetaceen Perus. Volume 2, 248 ofKogia breviceps (Blainville). Bulletin o/the American Museum 0/ pages. Ostermundingen [Bern]: Hirnanatomisches Institut. Natural History, 38(2):7-72. pilleri, G., and M. Ghir Seagars, DJ. 1981. The Brain (Endocranial Cast) of Schizodelphis sulcatus and the 1982. Jaw Structure and Functional Mechanics of Six Delphinids (Ceta­ I Cephalization ofEoplatanista italica (Cetacea): Paleoneurological cea: Odontoceti). 179 pages. Master's thesis, San Diego State Uni­ I and Paleoecological Considerations. Memorie deg/i Istituti di Geo­ versity, San Diego. logia e Mineralogia dell 'Universita di Padova, 34:387-440. Seaman, G.A., L.E Lowry, and KJ. Frost I 1982. The Brain and Cephalization of Schizodelphis sulcatus Gervais, 1982. Foods of the Belukha Whales (Delphinapterus leucas) in Western 1861. Investigations on Cetacea, 13: 13-25. Alaska. Cetology, 44(Sept.):1-19. I ~,c~n, A.R. Sickenberg, O. 1989. Observations of Mating Behavior in Dugongs (Dugong dugon). Ma­ 1934. Beitrage zur Kenntnis Tertiarer Sirenen. Memoires du Musee Royal rine Mammal Science, 5:382-387. d'Histoire Naturelle de Belgique, 63: 1-352. Purves, P.E. Sinclair, J.G. 1967. Anatomical and Experimental Observations on the Cetacean Sonar 1966. The Olfactory Complex ofDolphin Embryos. Texas Reports 0/Bio­ System. In R.-G. Busnel, editor, Animal Sonar Systems. Biology and logical Medicine, 24:426-431. Bionics, I: 197-270. Jouy-en-Josas: Laboratoire de Physiologie Spoor, E acoustique, INRA-CNRZ. 1993. The Comparative Morphology and Phylogeny of the Human Bony Rae, B.B. Labyrinth. Doctoral dissertation, Universiteit Utrecht, Utrecht. 1973. Additional Notes on the Food of the Common Porpoise (Phocoena Tedford, R.H., L.G. Barnes, and C.E. Ray phocoena). Journal o/Zoology. 169:127-131. 1994. The Early Miocene Littoral UrsidCarnivoran, Kolponomos: Sys­ ly,C. tematics and Mode ofLife. Proceedings 0/the San Diego Society 0/ 1966. Round Table: Practical Problems. In K.S. Norris, editor, Whales, Natural History. 29:11-32. Dolphins and Porpoises. page 671. Berkeley: University ofCalifor­ Tomilin, A.G. nia Press. 1967. Cetacea. In Mammals o/the USSR and Adjacent Countries, 9: 717 Ray, C.E., D.P. Domning, and M.C. McKenna pages. [Original in Russian, published 1957. Translated by Israel 1994. A New Specimen ofBehemotops proteus (Mammalia: Desmostylia) Program for Scientific Translations, Jerusalem.] from the Marine Oligocene of Washington. Proceedings o/the San Vibe, C. Diego Society 0/Natural History, 29:205-222. 1950. The Marine Mammals and the Marine Fauna in the Thule District Ridgway, S.H., D.A. Carder, R.F. Green, A.S. Gaunt, L.L. Gaunt, and WE. (North-West Greenland) with Observations on Ice Conditions in Evans 1939-41. Meddelelserom Gronland, 150(6):1-115. 1980. Electromyographic and Pressure Events in the Nasolaryngeal Sys­ Werth, AJ. tem of Dolphins during Sound Production. In R.-G. Busnel and J.E 1992. Anatomy and Evolution ofOdontocete Suction Feeding. xiii+313 Fish, editors, Animal Sonar Systems [NATO Advanced Science In­ stitute, series A (Life Science), Volume 28], pages 239-249. New pages. Doctoral dissertation, Harvard University, Cambridge, Mas­ York: Plenum Press. sachusetts. Rodinov, V.A., and V.I. Markov Wilke, E, T. Taniwaki, and N. Kuroda 1992. Functional Anatomy of the Nasal System in the Bottlenose Dolphin. 1953. Phocoenoides and Lagenorhynchus in Japan, with Notes on Hunt­ In J.A. Thomas, R.A. Kastelein, and A.Ya. Supin, editors, Marine ing. Journal 0/Mammalogy, 34(4):488-497. Mammal Sensory Systems, pages 147-177. New York: Plenum Press. Wood, EG. Schenkkan, EJ. 1964. General Discussion. In WN. Tavolga, editor, Marine Bio-Acoustics, 1973. On the Comparative Anatomy and Function of the Nasal Tracts in I:395-398. New York: Pergamon Press. Odontocetes (Mammalia, Cetacea). Bijdragen tot de Dierkunde, Yablokov, A.V., and G.A. Klevezal 43(2):127-159. 1962. [Whiskers of Whales and Seals and Their Distribution, Structure Schevill, WE. and Significance.] In S.E. Kleinenberg, editor, Mor/ologicheskie Os­ 1964. General Discussion. In W.N. Tavolga, editor, Marine Bio-Acoustics, obennosti Vodnikh Mlekopitayushchikh [Morphological Characteris­ 1:395-398. New York: Pergamon Press. tics of Aquatic Mammals], pages 48-81. Moscow: Izdatel'stvo Schulte, H. von W., and M. de Forest Smith "Nauka." [Translated from Russian by the Fisheries Research Board 1918. The External Characters, Skeletal Muscles, and Peripheral Nerves ofCanada, 1969.]