Skull anatomy of the Oligocene toothed mysticete Aetioceus weltoni

Skull anatomy of the Oligocene toothed mysticeteAetioceus weltoni (Mammalia; Cetacea): implications formysticete evolution and functional anatomy

THOMAS A. DEMÉRÉ1* and ANNALISA BERTA2

1Department of Paleontology, San Diego Natural History Museum, PO Box 121390, San Diego, CA92112, USA2Department of Biology, San Diego State University, San Diego, CA 92182, USA

Received 20 September 2007; accepted for publication 25 September 2007

Toothed mysticetes of the family Aetiocetidae from Oligocene rocks of the North Pacific play a key role ininterpretations of cetacean evolution because they are transitional in grade between dorudontine archaeocetes andedentulous mysticetes. The holotype skull of Aetiocetus weltoni from the late Oligocene (28–24 Ma) of Oregon, USA,has been further prepared, revealing additional morphological features of the basicranium, rostrum and dentarythat have important implications for mysticete evolution and functional anatomy. The palate of Aetiocetus weltonipreserves diminutive lateral palatal foramina and associated delicate sulci which appear to be homologous with theprominent palatal foramina and sulci that occur along the lateral portion of the palate in extant mysticetes. Inmodern baleen whales these foramina allow passage of branches of the superior alveolar artery, which suppliesblood to the epithelia of the developing baleen racks. As homologous structures, the lateral palatal foramina of A.weltoni suggest that baleen was present in this Oligocene toothed mysticete. Cladistic analysis of 46 cranial anddental characters supports monophyly of the Aetiocetidae, with toothed mysticetes Janjucetus and Mammalodonpositioned as successive sister taxa. Morawanacetus is the earliest diverging aetiocetid with Chonecetus as sistertaxon to Aetiocetus species. © 2008 The Linnean Society of London, Zoological Journal of the Linnean Society,2008, 154, 308–352.

ADDITIONAL KEYWORDS: palaeontology – phylogeny – systematics.

INTRODUCTION

Aetiocetidae (Cetacea, Mysticeti) comprises a rela-tively diverse family of archaic, small, toothed mys-ticetes found in Oligocene sedimentary rocks on bothsides of the North Pacific (Barnes et al., 1995). Theyrepresent an important transitional group that pre-serves morphologies intermediate between basilo-saurid archaeocetes and edentulous mysticetes. Thetransitional status of aetiocetids should not be viewedas suggesting they were ancestral to all later diverg-ing edentulous mysticetes, but instead that aetio-cetids provide a glimpse of the morphological changesthat occurred in early members of the mysticete clade

as they evolved from bite-and-swallow predators tospecialized bulk, filter-feeders.

To date, aetiocetids are only known from the NorthPacific and can be considered an endemic lineage inthat ocean basin until remains are discovered in otherregions. The recent report of a new species of earlyOligocene cetacean, Willungacetus aldingensis, fromthe Port Willunga Formation in South Australia(Pledge, 2005) provisionally assigns this southernhemisphere taxon to the Aetiocetidae. This record,however, still needs to be confirmed. Two of the fournominal aetiocetid genera are monotypic, Ashorocetusand Morawanocetus, both from the Morawan Forma-tion near Ashoro, Hokkaido, Japan (Barnes et al.,1995). Chonecetus includes two species, C. sookensis(Russell, 1968) from the Hesquiat Formation on*Corresponding author. E-mail: [email protected]

Zoological Journal of the Linnean Society, 2008, 154, 308–352. With 11 figures

© 2008 The Linnean Society of London, Zoological Journal of the Linnean Society, 2008, 154, 308–352308

mailto:[email protected]

Vancouver Island, British Columbia, Canada, and C.goedertorum (Barnes & Furusawa in Barnes et al.,1995) from the Pysht Formation, Olympic Peninsula,Washington, USA. Aetiocetus, the most speciosegenus, is represented by A. cotylalveus Emlong, 1966and A. weltoni Barnes & Kimura in Barnes et al.,1995, both from the Yaquina Formation of coastalOregon, USA, and A. tomitai Barnes & Kimura inBarnes et al., 1995 and A. polydentatus Sawamura inBarnes et al., 1995 from the Morawan Formation nearAshoro, Hokkaido, Japan. The assignment of A. poly-dentatus to the genus Aetiocetus has been recentlyquestioned (Ichishima, 2005; Fitzgerald, 2006), aconclusion with which we disagree. Additional unde-scribed aetiocetids are reported from the upper Oli-gocene San Gregorio Formation near La Paz, BajaCalifornia Sur, Mexico (Gerardo-Gonzalez-Barba,pers. comm. cited in Barnes et al., 1995).

The presence of teeth and primitive cranial androstral features led to initial consideration of Aetioce-tus cotylalveus as an archaeocete (Emlong, 1966),although its many mysticete features were empha-sized in that report. Later, Van Valen (1968) assignedA. cotylalveus to the Mysticeti, noting that the ances-tors of mysticetes must have had teeth and thatAetiocetus in other respects was more similar to mys-ticetes than to archaeocetes. Although differing onhigher-level relationships, both Emlong and VanValen recognized the phylogenetically basal nature ofAetiocetus cotylalveus and its importance as a transi-tional cetacean taxon.

Since the 1960s a number of additional toothedmysticete fossils have been discovered and describedand the definition of the family Aetiocetidae as origi-nally proposed by Emlong (1966) has been modified toinclude up to eight species in possibly four genera(Barnes et al., 1995). These authors proposed a phy-logenetic hypothesis for the Aetiocetidae and namedthree subfamilies to contain this expanded diversity:Chonecetinae (Chonecetus spp.), Morawanocetinae(Morawanocetus yabukii) and Aetiocetinae (Ashoroce-tus eguchii and Aetiocetus spp.). In a later report,Sanders & Barnes (2002) established the superfamilyAetiocetoidea to include all then known toothedmysticetes: Aetiocetidae (Emlong, 1966), Llanocetidae(Mitchell, 1989), and Mammalodontidae (Mitchell,1989). As recently shown by Fitzgerald (2006),however, there is ample evidence that Aetiocetoidea isa grade taxon and that the many described species,genera and families of toothed mysticetes do notrepresent a monophyletic group. For example, posses-sion of teeth is clearly a primitive condition andamong toothed mysticetes there is considerable mor-phological variation in tooth shape, size, position andnumber. The recent mysticete phylogeny proposed byFitzgerald (2006) recognizes six toothed lineages: (1)

ChMTM (an undescribed purported toothed mysticetefrom the Oligocene of South Carolina, USA; Geisler &Sanders, 2003); (2) Llanocetidae (a large toothed mys-ticete from the latest Eocene or earliest Oligoceneof the Antarctic Peninsula, Llanocetus denticrenatuswas not included in Fitzgerald’s phylogenetic analysisbut was mentioned in the text); (3) Janjucetidae (ashort-faced toothed mysticete from the late Oligoceneof Victoria, Australia); (4) Mammalodontidae (a small-headed toothed mysticete from the late Oligocene ofVictoria, Australia); (5) a paraphyletic Aetiocetidaerepresented by a Chonecetus clade (only C. goeder-torum was considered); and (6) an Aetiocetus clade(only A. cotylalveus was considered). Although notincluded in the actual cladistic analysis, A. polyden-tatus is considered by Fitzgerald (2006) not to be amember of Aetiocetus because of possession of severalpurported derived features: (1) ascending process ofmaxilla terminating slightly posterior to antorbitalnotch, at a point level with the anterior third (antero-posteriorly) of the supraorbital process of frontal; (2)anteroposteriorly foreshortened supraorbital processof frontal; (3) concavity in lateral margin of supraor-bital process poorly developed with orbit opening lat-erally but not anterolaterally and dorsolaterally as inmore basal mysticetes; and (4) elongated intertempo-ral region with well-developed sagittal crest.

The controversial monophyly, definition and mem-bership of Aetiocetidae is not just a semantic nomen-clatural debate. Instead, it underlines the importanceof this taxon in the interpretation of basal mysticetephylogenetic relationships and functional hypothesesof character evolution related to the loss of an adultdentition and development of a palatal baleen filter.One of the obstacles to resolving the issue ofaetiocetid monophyly is the current paucity ofdetailed morphological descriptions and illustrationsof nominal taxa. The present report attempts par-tially to rectify this problem by presenting a thoroughredescription of the holotype skull and dentary of oneof the better-preserved aetiocetids, Aetiocetus weltoni.Insights gained through this analysis of key morpho-logical features and their taxonomic distributionamong stem mysticetes is then used to re-evaluatephylogenetic relationships of nominal aetiocetids.

MATERIAL AND METHODS

Measurements, to the nearest tenth of a millimetre,of crania, dentitions and dentaries were taken usingdigital and/or mechanical calipers; these measure-ments are provided in Tables 1–4.

The following aetiocetid fossils were studied for thisreport: (1) USNM 25210, holotype skull of Aetiocetuscotylalveus Emlong, 1966; (2) USNM 256593, partialskeleton of undescribed specimen here referred to

SKULL ANATOMY OF TOOTHED MYSTICETE 309


A. cotylalveus; (3) AMP 2, holotype cranium of Aetio-cetus tomitai Kimura & Barnes in Barnes et al., 1995;(4) AMP 12, holotype skeleton of Aetiocetus polyden-tatus Sawamura in Barnes et al., 1995; (5) NMC 6443,holotype cranium of Chonecetus sookensis Russell,1968; (6) LACM 131146, holotype cranium of Chonece-tus goedertorum Barnes & Furusawa in Barnes et al.,1995; (7) LACM 138027, paratype partial craniumand dentary of C. goedertorum; and (8) AMP 1, holo-type partial skull and dentition of Morawanocetusyabukii Kimura & Barnes in Barnes et al., 1995.The following specimens and/or published literaturewere used for outgroup comparison: (1) Tasmacetusshepardi (USNM 484878); (2) Agorophius sp. (ChMPV5852); (3) Agorophius pygmaeus (MCZ 8761; dis-cussed in Fordyce, 1981); (4) Dorudon atrox (numer-ous specimens discussed in Uhen, 2004); and (5)Zygorhiza kochii (USNM 11962). The following speci-mens were used for ingroup comparison: (1) Mammal-odon colliveri (cast of NMV P199986); (2) Janjucetushunderi (published description and illustrations of

NMV P216929); (3) Eomysticetus whitmorei (ChMPV4253); and (4) representative taxa of modern mys-ticetes (see Deméré, Berta & McGowen, 2005).

Acronyms for institutions housing specimens dis-cussed in this study are as follows: AMP, AshoroMuseum of Paleontology, Ashoro, Hokkaido, Japan;ChM, Charleston Museum of Natural History,Charleston, South Carolina, USA; LACM, NaturalHistory Museum of Los Angeles County, Los Angeles,California, USA; MCZ, Museum of ComparativeZoology, Harvard University, Boston, Massachusetts,USA; NMC, National Museum of Canada, Ottawa,Canada; NMV, National Museum of Victoria, Mel-bourne, Australia; UCMP, Museum of Paleontology,University of California, Berkeley, California, USA;UO, Museum of Natural History, University ofOregon, Eugene, Oregon, USA; USNM, NationalMuseum of Natural History, Smithsonian Institution,Washington, DC, USA.

The holotype skull and dentary of Aetiocetusweltoni was examined using computed tomography

Table 1. Measurements (cm) of the skull of Aetiocetus weltoni UCMP 12290 and Aetiocetus cotylalveus USNM 25210

Measurement UCMP 12290 USNM 25210

Condylobasal length 61.6 62.3Rostral length (functional) 33.0 34.0Braincase length (functional) 28.6 28.3Rostral width (infraorbital plate) 20.9 20.8Rostral width 2 (25%) 16.8 16.3Rostral width 3 (50%) 13.6 12.6Rostral width 4 (75%) 9.0 7.3Rostral width 5 (pmx/max suture) 6.8 6.4Premaxilla width 1 (25%) 15.0 2.3Premaxilla width 2 (50%) 2.4 3.0Premaxilla width 3 (75%) 2.8 2.7Nasal length (greatest) 13.8 11.8Nasal width (anterior width) 5.0e 3.7Maxilla width 1 (antorbital notch) 7.3 7.1Maxilla width 2 (25%) 5.7 4.9Maxilla width 3 (75%) 1.4 2.7Ascending process max. (width) 3.7 3.8Ascending process max. (length) 7.8 6.6Supraorbital process max. W (medial) – 11.1Suprorbital process max. W (lateral) 8.2 8.3Occipital shield width 24.1 25.2Occipital shield height – 10.0Occipital condyle width 9.7 9.0Occipital condyle height 6.1 5.7Zygomatic width 28.7 31.6Palate length (straight line) 45.4 52.8Palate length (curve) 46.1 –Palate chord 2.5 –Palatine length 11.8 15.5e

e, estimated measurement.

310 T. A. DEMÉRÉ and A. BERTA


(CT) techniques. The resulting scan images are storedin the Department of Paleontology, San Diego NaturalHistory Museum.

LOCATION, STRATIGRAPHYAND CORRELATION

The holotype of Aetiocetus cotylalveus was collectedfrom the Yaquina Formation as exposed intertidallyon the modern marine abrasion platform near themouth of Deer Creek between Seal Rock State Parkand the mouth of Beaver Creek, Lincoln County,Oregon, USA (UO Locality 2503). As reported byEmlong (1966) and Snaveley et al. (1976), the YaquinaFormation here consists of a north-west-dipping (19°)sequence of interbedded, grey, fine-grained sand-stones, medium-grained sandstones, and siltstonescontaining fossils of marine molluscs, as well as

carbonized wood. Emlong (1966: fig. 1) considered UOLocality 2503 to occur stratigraphically high in theYaquina Formation, not far below its depositionalcontact with the overlying Nye Mudstone. In fact,Emlong noted that the contact between the two rockunits ‘is visible west of the type locality at times ofextreme minus tide, but is normally below sea level.’The type locality of Aetiocetus weltoni (UCMP localityV-79013) occurs less than 1 km north of UO Locality2503 in the Yaquina Formation as also exposed inter-tidally on the modern abrasion platform at OnaBeach near the mouth of Beaver Creek, LincolnCounty, Oregon, USA. Although Barnes et al. (1995)asserted that UCMP Locality V-79013 is stratigraphi-cally higher in the Yaquina Formation than UO Local-ity 2503, given the general north-east strike (N20°E)of the strata at each locality, both sites may representthe same general stratigraphic horizon in the upper-most portion of this rock unit. Without more detailedfieldwork it is not certain how much, if any, strati-graphic thickness (and therefore time) separates thetwo fossil localities.

The Yaquina Formation was originally named byHarrison & Eaton (1920) for shallow-water marinedeposits exposed in the vicinity of the village ofYaquina, near Newport, Oregon. Addicott (1976) cor-related the Yaquina Formation in the Newport

Table 2. A, measurements (cm) of the left dentary ofAetiocetus weltoni UCMP 122900; B, measurements (mm)of cross-sectional dimensions measured at various pointsalong the left dentary

(A)Total length (straight line) 54.5Total length (outside curvature) 55.2Length of chord 1.8eNumber of mental foramina 7.0Condyle to coronoid process 6.3Coronoid crest length 17eCondyle to mandibular foramen 7.7Minimum height neck 8.2Mandibular condyle vertical diameter 5.3Mandibular condyle transverse diameter 3.5Mandibular foramen height 7.7eMandibular foramen width 2.5eMandibular symphysis length 3.2Symphyseal groove length 4.5

(B)Vertical/transverse diameter @ 12.5% of

dentary total length43.3/19.4

Vertical/transverse diameter @ 25% ofdentary total length

38.5/21.1

Vertical/transverse diameter @ 37.5% ofdentary total length

37.1/23.3


42.1/22.4

Vertical/transverse diameter @ 62.5% ofdentary total length

52.0/22.3


85.2/27.2

Vertical/transverse diameter @ coronoidprocess

161.4/31.7

Vertical/transverse diameter @ posterior end 86.7/52.4

e, estimated measurement.

Table 3. Measurements (mm) of upper dentition of Aetio-cetus weltoni UCMP 122900

Distance from anterior terminus of maxilla to centre oftooth alveolusI1 2.0I2 4.5I3 8.0C 11.2P1 14.2P2 17.7P3 21.0P4 24.5M1 27.9M2 30.8M3 33.2

I1, AP/T 6.3/4.9I2, AP/T 7.8/5.2/11.9I3, AP/T/H 8.8/5.8/11.0C, AP/T/H 8.3/4.9/11.4P1, AP/T 8.9/5.4/10.4P2, AP/T/H 10.6/6.2/12.2P3, AP/T 10.6/6.5d/9.6M2, AP/T 11.1/7.2/10.0eM3, AP/T 10.7/5.7/10.3e

AP, anteroposterior; d, distorted; e, estimated measure-ment; H, crown height, T, transverse.



Embayment with his Juanian provincial molluscanstage and with the Vaqueros provincial molluscanstage of California. Rau (1981) reported benthicforaminifers from the Yaquina Formation that werecorrelative with the Zemorrian provincial benthicforaminifer stage of Kleinpell (1938). Armentrout(1981) citing C. A. Repenning (pers. comm.) reportedthe co-occurrence of Zemorrian benthic foraminifers,Juanian marine molluscs and Arikareean marinemammals in the Yaquina Formation as exposed inthe Yaquina Bay section near Newport. Thesevarious biochronological correlations led Armentrout(1981) to suggest a late late Oligocene (late Chat-tian) age for the Yaquina Formation (approximately24–25 Ma). In a more recent palaeomagnetic analy-sis of Pacific Coast Palaeogene chronostratigraphy,Prothero et al. (2001) sampled the Yaquina Forma-tion and correlated the uppermost portion of therock unit with Chron C9r, suggesting an older agefor this stratigraphic interval of 28 Ma, early lateOligocene (early Chattian). Based on this correlationweb the late Oligocene age of the holotype of A.

weltoni is confirmed. However, its absolute age maybe older than previously thought, perhaps as old asearly late Oligocene (approximately 28 Ma).

SYSTEMATICSORDER CETACEA

SUBORDER MYSTICETI

FAMILY AETIOCETIDAE EMLONG, 1966

GENUS AETIOCETUS EMLONG, 1966

Type species: Aetiocetus cotylalveus Emlong, 1966

Distribution: Late Oligocene of Oregon and Japan

Included species: Aetiocetus cotylalveus Emlong, 1966,A. polydentatus Sawamura in Barnes et al., 1995,A. tomitai Kimura & Barnes in Barnes et al., 1995,A. weltoni Barnes & Kimura in Barnes et al., 1995.

Definition: The monophyletic group containing themost recent common ancestor of Aetiocetus cotyla-lveus and A. polydentatus and all of its descendants.

Emended diagnosis: Small toothed mysticetes withthree or fewer small and simple anterior and pos-terior denticles on posterior upper teeth, anteriorand postcanine teeth weakly heterodont (synapomor-phies of species of Aetiocetus); triangular anteriorextension of parietal–frontal suture, fine verticalenamel ridges only on lingual surface of postcanineteeth, wide diastemata between postcanine teeth,and rostral width at antorbital notch relative tooccipital condyle width > 170% (possible synapomor-phies of species of Aetiocetus); lacrimal with largeand lobate dorsal exposure, palate window exposingvomer present and formed by maxilla and premax-illa, and distinct internal narial notch formed bypalatine, pterygoid and vomer (synapomorphies ofAetiocetus + Chonecetus); zygomatic process of squa-mosal expanded near its anterior margin and at itsposterior end but narrow in its middle, shortoverlap of jugal by squamosal, and coronoid processof dentary well developed with concave posteriormargin (aetiocetid synapomorphies); mandibularsymphysis not sutured, descending process ofmaxilla developed as an edentulous infraorbitalplate, anterior and postcanine teeth moderately toweakly heterodont or absent (apomorphies sharedwith toothed and edentulous mysticetes); supraor-bital processes of frontal elevated and at same levelas cranial vertex, distinct intertemporal region,anteriorly placed external narial opening, and reten-tion of adult dentition (plesiomorphies shared withdorudontine archaeocetes).

Table 4. Measurements (mm) of lower dentition of Aetio-cetus weltoni UCMP 122900

Distance from anterior terminus of mandible to centre oftooth alveolusi1 8i2 25i3 57c 95p1 127p2 159p3 200p4 234m1 264m2 292m3 325em4 340

i1, AP/T 6.7/6.4i2, AP/T 8.3/6.9i3, AP/T/H 8.8/7.4/11.4c, AP/T/H –/–/11.3p1, AP/T 10.1/7.1p2, AP/T/H –/–/14.6p3, AP/T 8.2/5.1p4, AP/T 10.5/6.9m1, AP/T –/6.4m2, AP/T –/6.4m3, AP/T 11.1/4.9m4, AP/T 8a/–

a, alveolus; AP, anteroposterior; e, estimated measure-ment; H, crown height; T, transverse.



AETIOCETUS WELTONI BARNES &KIMURA IN BARNES ET AL., 1995

AETIOCETUS SP. NOV. KIMURA, BARNES &FURUSAWA, 1992

AETIOCETUS WELTONI BARNES & KIMURA,IN BARNES ET AL., 1995

Emended diagnosis: A species of Aetiocetus withdeeply concave orbital rim of supraorbital process offrontal and posteriormost edge of ascending process ofmaxilla terminating at posterior border of supraor-bital process of frontal (autapomorphies of A. weltoni);narrower transverse intertemporal width, morerobust zygomatic process of the squamosal, and lesslobate anterior parietal wings (symplesiomorphiesrelative to A. cotylalveus); non-polydont upper denti-tion, minor overhang of temporal wall by lambdoidalcrests, and narrower nasals (symplesiomorphies rela-tive to A. polydentatus).

Holotype: UCMP 122900, including associated ele-ments of one individual consisting of a nearly com-plete skull (missing most of the supraoccipital), rightand left dentaries, partial hyoid skeleton, cervical,thoracic, lumbar and caudal vertebrae, and some ribs.Only the skull and dentaries have been prepared todate.

Locality: UCMP Locality V-79013, Ona Beach, LincolnCounty, Oregon, USA.

Stratigraphic horizon and age: Yaquina Formation(upper part); late Oligocene, Chattian, ~24–28 Ma.

DESCRIPTIONGENERAL FEATURES OF SKULL

The holotype skull of Aetiocetus weltoni (UCMP122900) is nearly complete, lacking the right exoc-cipital (reconstructed with an epoxy prosthesis), theanterior 60% of the supraoccipital shield and thedorsalmost portions of the parietals at the anterome-dial corner of the temporal fossa. Although both tym-panic bullae were found in situ, breakage throughthe basicranium (which probably occurred at thetime of recovery) has resulted in damage to bothpetrosals. The right exoccipital is also missing. Theright and left dentaries were articulated with therostrum at the time of burial and through sedimentcompaction and diagenesis became solidly attached tothe alveolar margin of the upper jaw. Because of this,upper and lower teeth are in a forced ‘clenched’ posi-tion and a few teeth in each tooth row have beendriven into the over- or underlying bones (i.e. lowerteeth forced into the palatal surface of the maxillaand the upper teeth forced into the alveolar margin

of the dentary). Subsequent mechanical preparationwork has removed the entire left dentary and theposterior half of the right dentary from the skull. Thedegree of closure of cranial sutures (closed, but notfused) suggests that the skull belonged to a subadultindividual.

The functional rostrum (tip of rostrum to antorbitalnotch) forms 54% of the condylobasal length (CBL)and is broadly triangular in dorsal aspect (Table 1).The external narial fossa is anteriorly placed well infront of the antorbital notch, and its posterior border(defined by the anterior margin of the nasals) lies ata level midway between the midpoint of the rostrumand the back of the nasal bones. The ascendingprocess of the maxilla and premaxilla extend postero-medially across the anteromedial corner of the frontalto the level of the root of the postorbital process of thefrontal (Figs 1, 2). The functional cranium (antorbitalnotch to occipital condyles) forms 46% of CBL(Table 1). The supraorbital processes of the frontalsare elevated and lie at the same level as the cranialvertex (Fig. 1). The temporal fossae are large andovate and the intertemporal constriction is relativelylong and narrow. The zygomatic process of the squa-mosal reaches anteriorly to contact the postorbitalprocess of the frontal (Fig. 3). The dentition retains aminor degree of heterodonty and has a tooth count of11/12. Upper and lower teeth occlude in an alternat-ing, interlocking pattern (Fig. 4). The dentary has alarge mandibular foramen and a lobate coronoidprocess positioned within the temporal fossa. Thesymphysis is unfused with a distinct symphysealgroove.

PREMAXILLA

On the dorsal surface of the rostrum the right and leftpremaxillae are unfused but closely appressed at themidline between the tip of the rostrum and the levelof I3. Posterior to I3 the medial borders of the pre-maxillae diverge from one another to expose themesorostral groove (Fig. 2). This condition occurs inall species of Aetiocetus and is not unlike the condi-tion in later diverging edentulous mysticetes. Theanterior terminations of the premaxillae are narrowlytriangular in dorsal aspect and form a low crestwhere they meet at the midline. At the level of thecanine on the left side and the I3-C diastema on theright side is a single anterior premaxillary foramenopening anteriorly and positioned within 10 mm ofthe maxilla–premaxilla suture. A second foramenoccurs on the right side at the level of the I2-3diastema. There is no foramen in this position on theleft side. These appear to be the only foramina on thepremaxilla. There are no premaxillary foramina inthe area adjacent to the narial fossa, nor are there



any sulci (anteromedial, posteromedial, or posterolat-eral sulci) as occur in stem and crown odontocetes.Posterior to the canine the dorsal surface of thepremaxilla becomes distinctly elevated above thesurface of the adjacent maxilla. Although the twobones are closely appressed, the intervening sutureappears to have been unfused, a condition possiblyassociated with rostral kinesis in later diverging mys-ticetes. The condition in archaeocetes (e.g. Zygorhiza)has the premaxillae fused anteriorly at the level of I1and unfused posteriorly at the level of I3. In Janju-cetus and Mammalodon the condition at the tips ofthe premaxillae appears to be equivocal due to poorpreservation. However, in Janjucetus it is clear thatthe premaxilla in this region is more elevated abovethe adjacent maxilla than in species of Aetiocetus. InA. weltoni the elevation of the premaxilla increasesposteriorly and is most pronounced along the lateralhalf of the premaxilla, especially between P1 and M2.

In this region the lateral surface of the premaxilla isbroadly rounded and overhangs the medial margin ofthe adjacent maxilla. Medially, the surface of thepremaxilla in this region slopes inward toward themesorostral groove. The resulting embayment repre-sents the anterior portion of the external narial fossa.At about the level of P4 (roughly corresponding to themiddle of the narial fossa) the premaxilla narrowsabruptly as its medial margin becomes more steeplysloped. At the level of M1 (roughly corresponding tothe posterior part of the narial fossa) the lateralportion of the premaxilla ascends steeply to meet theadjacent nasal bone. The anterolateral corner of thenasal–premaxilla suture is marked by a rugoseprocess (the ‘crest-like eminence’ of Barnes et al.,1995) that is divided by this suture. Posterior to thisregion the premaxilla gradually ascends across thefrontal to the cranial vertex and is in tight suturalcontact with the nasal and maxilla until their mutual

Figure 1. Aetiocetus weltoni, UCMP 122900, holotype skull. A, dorsal view; B, ventral view; C, left lateral view.



posterior terminations. The sutures between thenasal, premaxilla and maxilla in this region appear tobe fused, suggesting a lack of kinesis in the posteriorportion of the rostrum. The ascending process of the

premaxilla is fairly uniform in transverse width andmeasures approximately 13–14 mm in that dimen-sion. The posterior termination of the premaxillaoccurs slightly (approximately 8 mm) in front of the

Figure 2. Aetiocetus weltoni, reconstructed skull in dorsal view showing anatomical features.



Figure 3. Aetiocetus weltoni, reconstructed skull in ventral view showing anatomical features.



termination of the nasals, which occurs at the level ofthe temporal crest on the posterior portion of thepostorbital process of the frontal. In Aetiocetuscotylalveus (USNM 25210), Chonecetus goedertorum(LACM 131146) and Janjucetus hunderi (NMVP216929) the rostral telescoping is less pronouncedand the nasal terminates more anteriorly and at thelevel of the middle of the orbit. The condition inAetiocetus polydentatus (AMP 12) is like that in A.weltoni. In Janjucetus the premaxilla does not over-hang the maxilla, tapers sharply between the nasaland the maxilla, and terminates posteriorly in front ofthe posterior terminations of both elements at thelevel of the preorbital process of the frontal.

On the palate, the premaxillae are only exposed atthe anterior end of the rostrum where they meet atthe midline. Posteriorly the premaxilla forms acomplex suture with the maxilla and vomer (Figs 1B,3). The latter is exposed in a narrow palatal windowformed posteriorly by the maxillae and anteriorly bythe palatine processes of the premaxillae. Medially,the palatine processes of the premaxillae have anarrow exposure between the anterior tips of themaxillae, while posteriorly they bifurcate and sur-round the anterior half of the narrow vomer exposure.The posterior terminations of the palatine processesextend to the level of P2. Anteriorly, the right and leftportions of the premaxilla meet at the midline at the

Figure 4. Aetiocetus weltoni, UCMP 122900, holotype skull. A, anterodorsal view; B, posterior view; C, lateral view ofright anterior upper and lower dentition.



level of the anterior border of Pl (at the middle of Cin A. cotylalveus). The intra-premaxillary suture atthe midline is open and unfused from the vomerexposure to the tip of the rostrum. Laterally, thepremaxilla contacts the maxilla along a posterolater-ally directed suture. This suture, although tight,appears to have also been unfused based on thepresence of a 1-mm-thick seam of siltstone matrixbetween the two bones. Anteriorly, this suture isorientated in the parasagittal plane beginning at thelevel of P2 and extending to the level of I3 at whichpoint the suture turns laterally then posterolaterallyas it ascends through the middle of the I3-C diastemato emerge on the dorsal surface of the rostrum.Within the diastema a distinct embrasure pit for thelower canine occurs on the anterior margin of thesuture.

MAXILLA

On the dorsal surface of the rostrum the maxillaforms an open suture with the adjacent premaxilla aspreviously described. The contact, however, betweenthe maxilla and the premaxilla is a tight articulationas in archaeocetes and unlike the much looser, kineticcondition in modern mysticetes. Posteriorly thesuture is recessed adjacent to the level of the externalnares (narial fossa) and is obscured from dorsal viewby the overhanging lateral wall of the premaxilla. Anacutely triangular ascending process of the maxillaextends onto the anteromedial portion of the frontalremaining in tight sutural contact with the ascendingprocess of the premaxilla. Both processes appear to befirmly sutured into the frontal unlike the relativelyloose articulation seen in later diverging mysticetes(e.g. Balaenoptera spp.). On the rostrum the thinlateral margin of the maxilla is slightly scallopedcorresponding to the interlocking occlusion of theupper and lower dentition. Embayed regions of thelateral margin correspond to placement of crowns ofthe lower teeth, while convex regions correspond tothe position of roots of the upper teeth. In the holo-types of A. cotylalveus (USNM 25210) and A. poly-dentatus (AMP 12) only the portion of the rostrumanterior to P3 appears to be scalloped, while posteriorto this the lateral margin is essentially straight. Theholotype of J. hunderi (NMV P216929) has a linear(unscalloped) lateral margin of the entire maxilla.

A distinct antorbital process marks the posterolat-eral corner of the maxilla of A. weltoni and is bor-dered anteriorly by a deep sulcus forming theantorbital notch. The antorbital process of the maxillais thin anteroposteriorly and thicker dorsoventrally.Posteriorly, the antorbital process overlaps and wrapsaround the large lacrimal, which is dorsally exposed.This appears to be the condition in all aetiocetids.

A series of dorsal infraorbital foramina are exposedon the medial half of the right maxilla above the levelof M2 (Fig. 1A). The surface of the maxilla is elevatedhere and faces anterolaterally relative to the morehorizontally orientated surface of the main rostralportion of the maxilla. A primary pair of infraorbitalforamina is dorsally placed and positioned behindand above two or possibly three additional, smallerforamina. Posterodorsal crushing, however, in thisregion makes unclear the exact number of foraminapresent. The primary dorsal infraorbital foramina areapproximately 4 mm in diameter and lie 8–12 mmbelow the premaxilla/maxilla suture on the elevatedmaxillary surface. CT imaging reveals a majorbranching of the infraorbital canal within the maxillaat this position. The infraorbital canal transmitscranial nerve V2 from the orbit to the face and lips. Asimilar pair of foramina occurs in this position on theleft and right maxillae in A. cotylalveus. In Chonece-tus goedertorum (LACM 131146) a single, large-diameter dorsal infraorbital foramen occurs here andis associated with three other foramina located ante-rior to and/or below the primary foramen. CT imagingof UCMP 122900 reveals a second branching of theinfraorbital canal at this same position (approxi-mately 25 mm anterior to the primary foramina),which is the location of at least two accessory infraor-bital foramina and their associated anteriorly directsulci. Internally, the infraorbital canal continuesanteriorly for another 20 mm where it crosses themaxilla–premaxilla suture and extends into thepremaxilla.

The posterior opening of the infraorbital canal (i.e.the internal maxillary foramen) is well exposed in themaxilla on the anterior wall of the right orbit inUCMP 129900 and occupies a position similar to thatnoted in Dorudon (Uhen, 2004). In Aetiocetus weltoni,however, the foramen is much larger in diameter andis bounded below by a broad, edentulous infraorbitalplate of the maxilla. As noted by Fordyce & Muizon(2001), the edentulous infraorbital plate is a diagnos-tic feature of mysticetes. The posterior margin ofthe infraorbital plate in Aetiocetus weltoni is evenlycurved and lacks the prominent posteromedialprocess seen in later diverging mysticetes (e.g. bal-aenopterids). An evenly curved posterior margin(posteriorly concave) is also seen in most ‘cetotheres’.

In Dorudon the horizontal portion of the maxillaforming the floor of the infraorbital canal is narrowand supports the M1-2 tooth row (Uhen, 2004). Inboth Dorudon and Aetiocetus spp. the posterior ter-minus of the maxilla reaches to the centre of the orbitwhere its lateral margin is first underlapped and thenoverlapped by the delicate jugal. The infraorbitalplate of the maxilla in species of Aetiocetus and Jan-jucetus is edentulous, in contrast to the toothed con-



dition in basilosaurid archaeocetes (e.g. Zygorhizaand Dorudon; Kellogg, 1936; Uhen, 2004).

On the palate the maxillae meet tightly at themidline (Figs 1B, 3) and appear to be fused for mostof their length in contrast to the condition inmodern mysticetes where the medial borders of themaxillae are distinctly separate and expose theunderside of the median rostral ‘gutter’ of the vomerfor their entire length. Anteriorly, the maxillae areseparated to expose the vomer in the palatalwindow described above. The anteromedial corner ofthe maxilla is developed as a narrow longitudinalridge closely appressed to a similar ridge on thepalatine process of the premaxilla. The anterior tipof the maxilla extends to the level of I3 (to the levelof the I3-I2 diastema in Aetiocetus cotylalveus).From this point the border of the maxilla extendsposterolaterally and passes through the middle ofthe I3-C diastema between and onto the dorsalsurface of the rostrum.

On the palate of A. weltoni both maxillae are gen-erally planar transversely and not divided into rightand left longitudinal ‘gutters’ as in adult extant mys-ticetes. Interestingly, the palate of fetal balaenopteridmysticetes is also flat and does not begin to acquirethe characteristic longitudinal ‘gutter’ until nearterm (Ridewood, 1923). In Aetiocetus weltoni a broad,shallow and generally oval fossa occurs on the poste-rolateral portion of each maxilla adjacent and medialto M1-3. This posterolateral fossa is bounded poste-riorly by the oblique maxilla–palatine suture andmedially by a broadly arched palatal convexitycentred on the midline. Anteriorly, the fossa mergesgradually with the planar surface of the palate.Whether or not the posterolateral fossa is homologousto the longitudinal ‘gutter’ of later diverging, baleen-bearing, edentulous mysticetes is unclear.

The arrangement of palatal foramina on themaxilla is complex (Fig. 3). The most posteriorforamina occur at the level of M3 and are locatedwithin 16 mm of the midline. These palatine foramina(one right and one left) open anteriorly and are ofsmall calibre (less than 2 mm diameter). Anterior tothese at the level of the M1-2 diastema is a pair oflarger palatine foramina (one right and one left); theright foramen is positioned more posteriorly than theleft and occurs 14 mm from the midline. The leftforamen occurs 12 mm anterior to the right and ispositioned approximately 11 mm from the midline.Both foramina open anteriorly into broad palatinesulci that extend anteriorly to at least the level of P1.Together these four foramina appear to be homolo-gous to typical mammalian major palatine foramina.If correct, this homology suggests that the foraminaprovided passage for the descending palatine arteryand greater palatine nerve.

In contrast to the medially positioned majorpalatine foramina are a series of delicate foraminaoccurring on the lateral half of the maxilla (Figs 3, 5).These lateral palatal foramina are closely associatedwith the maxillary dentition and may be homologouswith the foramina for transmission of blood (superioralveolar artery) and innervations (superior alveolarnerve) to the baleen racks in edentulous mysticetes.The series of eight lateral nutrient foramina occursslightly medial to the left toothrow between C and M2(the right dentary is still articulated with the skulland obscures the right side of the palate). The follow-ing description of foramina begins with the posteri-ormost foramen (#1) and proceeds to the anteriormostforamen (#8). Foramen #1 is small (~1 mm diameter),located approximately 10 mm from the medial alveo-lar margin of M2, and opens into a fine, anterolater-ally orientated sulcus ~11 mm in length that forms anangle of ~40° with the sagittal plane. Anterior toforamen #1, two foramina of the same small diameter(~1 mm) are positioned ~3 mm from M1. The posteriorforamen of this pair (#2) and its sulcus (~2.5 mm long)form an angle of ~60° with the sagittal plane, whilethe other foramen (#3) occurs 6 mm anterior to #2with an anteriorly directed sulcus (~6.5 mm long) thatis angled ~25°. Another pair of small (~1 mm diam-eter) foramina (#4 and #5) occur approximately 9 mmfrom the medial corner of P4. Sulci associated withthese delicate foramina form angles of ~20° andmeasure ~8.5 and 5.6 mm, respectively. A largerforamen (#6) ~1.5 mm in diameter is positioned~7 mm from the posteromedial corner of the exposedP3 root, and the associated sulcus (~15.5 mm long)forms an angle ~10° to the sagittal plane. Anotherlarge foramen (#7) is ~2 mm diameter and occurswithin 4 mm of the medial alveolar border adjacent toP2. The associated sulcus of this foramen is shorter(~7 mm long) and less distinct relative to those asso-ciated with the more posterior foramina and isroughly parallel to the sagittal plane. The anterior-most foramen (#8) occurs ~3 mm medial to thediastema between C and P1. This foramen is of smallcalibre (~1 mm diameter), and its associated sulcus ispoorly formed and orientated roughly parallel to thesagittal plane.

Delicate lateral nutrient foramina and sulci ofsimilar size and orientation were also observed intwo other Oligocene aetiocetids, Aetiocetus cotyla-lveus and Chonecetus goedertorum. Unfortunately,preservation in these specimens was not as good asin the holotype of A. weltoni. Only one obviousforamen is preserved on the palate of A. cotylalveus(USNM 25210) and occurs on the right maxilla adja-cent to the P4-M1 diastema and 3 mm from themedial margin of the M1 alveolus. The sulcus asso-ciated with this foramen forms an angle of ~5° with



the sagittal plane and measures ~12.8 mm in length.The holotype of Chonecetus goedertorum (LACM131146) preserves three obvious lateral palatalforamina. One on the right maxilla between the P2and P3 alveoli is matrix filled and opens anteriorlyinto a delicate sulcus orientated ~4° to the sagittalplane. This tiny (~1 mm diameter) foramen lies~2–3 mm from the medial margins of the adjacentalveoli. A second foramen occurs on the left maxillaadjacent to the P4 alveolus and ~1 mm from itsmedial margin. The associated matrix-filled sulcusmeasures ~7 mm in length and is orientated in theparasagittal plane. The third foramen is problematic,but occurs on the left maxilla adjacent to the M2alveolus. The matrix-filled sulcus is orientated ~40°to the sagittal plane and measures ~5 mm in length.Recently, Sawamura et al. (2006) noted the presenceof lateral nutrient foramina in close association withthe molar and premolar dentition in a new, unde-scribed species of Morawanocetus from the late Oli-gocene of Hokkaido, Japan.

In UCMP 122900 the alveolar region of the maxillais clearly defined and has distinct medial and lateralmargins (Fig. 5). The medial margin locally forms a

sharply elevated rim immediately adjacent to theroots of the canine and postcanine teeth. This sharprim diminishes anteriorly becoming indiscerniblewithin the intervening diastemata. Emlong (1966)mentions similar structures on the medial marginsof the alveoli in A. cotylalveus (USNM 25210). Thelateral margin of the alveolar region of the rostrum iscoincident with the lateral edge of the premaxilla andmaxilla. In fetal skulls of modern balaenopterids themedial and lateral margins of the ephemeral alveolargroove occupy similar and homologous positions onthe rostrum (Ridewood, 1923). In Aetiocetus weltoniconspicuous embrasure pits (not seen in USNM25210) occur between C and P1, P1 and P2, and P2and P3; the posteriormost embrasure pit is thedeepest and broadest. The diastema between P3 andP4 lacks an embrasure pit. The postcanine teeth areemergent, with the crowns well clear of their alveoli.Emlong (1966) noted the same condition in USNM25210 and speculated that the postcanine teeth of A.cotylalveus were less firmly planted in the maxillathan the anterior teeth. Rostral diastemata in UCMP122900 vary in length as follows: I1-2 ~13 mm, I2-3~21 mm, I3-C ~23 mm, C-P1 ~17 mm, P1-2 ~22 mm,

Figure 5. Aetiocetus weltoni, UCMP 122900, holotype; stereophotographs of portion of left palate showing location oflateral palatal foramina and sulci.



P2-3 ~ 22 mm, P3-4 ~17 mm, P4-M1 ~19 mm, M1-2~14 mm, M2-3 ~11 mm.

VOMER

On the anterior portion of the palate the vomer isexposed behind the palatine processes of the premax-illae in a small slit-like palatal window, as previouslydescribed (Fig. 3). This vomer exposure extends fromthe P2-3 diastema to the level of P1. Janjucetushunderi has a longer vomer exposure that extendsfrom the P2-3 diastema to the level of the canine. InA. weltoni (also A. cotylalveus) the premaxillae formthe anterior border of the palatal window, while in J.hunderi the vomer exposure apparently is formedentirely by the maxillae. Palatal exposures of thevomer have not been reported in any other knownfossil or living edentulous mysticete, nor in basilo-saurid archaeocetes. Fordyce (1994, 2002), however,reported probable homologous structures on thepalates of two Oligocene odontocetes. In Simocetusrayi the palatal window extends from approximatelythe P2-3 diastema to a point anterior to C, while inWaipatia maerewhenua the palatal window is moreposteriorly placed and reminiscent of the condition inextant odontocete cetaceans (e.g. delphinoids).

As revealed by CT imaging of UCMP 122900 thevomer in the floor of the mesorostral groove termi-nates anteriorly approximately 88 mm behind the tipof the rostrum. From here to a point approximately134 mm behind the rostral tip the vomer restsloosely on the premaxillae and is then exposed for60 mm in the palatal window. Over its anterior140 mm the vomer is a simple flattened trough.Posterior to this point the lateral margins of thevomer become progressively thickened and interlockwith the adjacent maxillae along a chevron-shapedcontact. The mesorostral groove extends posteriorlywell into the internal narial passage where it even-tually terminates, presumably against the meseth-moid. Within the internal nares the ventral portionof the vomer develops into a delicate vomerine crestthat descends to the level of the ventral surface ofthe palatines. More posteriorly the vomerine crestbecomes transversely expanded, with a maximumtransverse width of 4–5 mm. This expanded portionof the vomerine crest extends posteriorly to justbehind the posterior termination of the pterygoidhamuli. From this point the vomerine crest narrowsto a sharp keel and ascends posterodorsally to thebasicranium where it terminates within 2 mm ofthe vomer–basioccipital suture at a level between themidpoints of the tympanic bullae. The horizontallyorientated posterior nasal plate of the vomer broadlyunderlaps and obscures the fused basioccipital/basisphenoid suture.

PALATINE

Both palatines are preserved in UCMP 122900, theright more completely than the left. Although crushedslightly at the midline, the palatines appear to haveformed a relatively planar surface on the palate.Their contact in the sagittal plane is tight, completelyconcealing the vomer. Whether or not the inter-palatine suture was fused is unclear, although itappears to be fused in the holotype of A. cotylalveus(USNM 25210). The posterior border of the palatinehas a complex and unique morphology and isretracted medially to form the anterior border of theinternal nares (Figs 3, 6). A narrow sliver of thepalatine extends posteromedially along the deep, ven-trally flattened vomerine crest to create a W-shapedsuture between the palatine and vomer. The anteriormargin of the internal nares is thus formed by thepalatine, which is roughly orientated transverse atthis level and extends 15 mm laterally to a pointwhere the margin abruptly turns posteriorly to under-lap the anterior root of the pterygoid hamulus. Thepalatine here forms a lobate posterolateral projection,which covers the anterior portion of the pterygoid.The lateral margin of this projection forms thepalatine–pterygoid suture, which is orientated ante-rolaterally and crosses onto the temporal wall whereit terminates at the ventral margin of the largeorbital fissure. This complex contact between thepalatine and pterygoid is noteworthy as it involves asquamous suture medially on the palate and a ser-rated suture laterally on the temporal wall. A sharplateral crest on the palatine roughly parallels but ispositioned approximately 12–15 mm medial to theoblique pterygoid–palatine suture. The apex of thispalatal crest diminishes in size posteriorly and mergesalmost imperceptibly with the surface of the palatineat its termination on the posterolateral projection.Similar palatal crests occur in the same position on theholotype skulls of Aetiocetus cotylalveus, A. polydenta-tus and A. tomitai. Although the holotype of Chonece-tus goedertorum does not preserve lateral palatalcrests, it does possess posterolateral projections asnoted by Barnes et al. (1995). These projections aremore delicate and elongate, in contrast to the shorterand lobate projections in species of Aetiocetus.

The maxilla–palatine suture has been affected bycrushing in the holotype skull of Aetiocetus weltoni,but appears to have roughly the same orientation asnoted by Emlong (1966) in A. cotylalveus. Anteriorly,the suture is broadly convex, while laterally it curvesposteriorly to contact the posterior margin of theinfraorbital plate ~70 mm from the midline (60 mm inUSNM 25210). The palatines are exposed on theventral portion of the temporal wall and form thelower rim of the large orbital fissure (Fig. 7).



NASAL

The nasals are narrow, elongate and roughly rectan-gular elements (Figs 1A, 2), as noted by Barnes et al.

(1995), with tightly fused internasal and nasal–premaxilla sutures. Anteriorly, the combined nasalsmeasure ~40 mm transversely, compared with 37 mmin Aetiocetus cotylalveus (USNM 25210) and 53 mmin A. polydentatus (AMP12). Posteriorly, the nasalsgradually taper in width to a minimum of 25 mm asmeasured 25 mm anterior to their posterior termina-tion. In A. polydentatus the nasals are rather uni-formly broad for their anterior 66 mm, but abruptlydecrease in width to 37 mm behind this point. Theanterior border of the nasals is broken in the holotypeof A. weltoni, but appears to have been anteriorlyconcave. The dorsal surface of the nasals is fairlyuniform in profile for most of the combined length,but is abruptly deflected dorsally in its anterior20–25 mm. A similar dorsal deflection characterizesthe nasals of A. polydentatus and both may be theresult of post-depositional deformation. As previouslymentioned, the anterolateral corner of the nasal ismarked by a rugose process that is divided by thenasal–premaxilla suture. In the holotype of Janjuce-tus hunderi the nasals, although incomplete anteri-orly, appear to have been generally rectangular inoutline (Fitzgerald, 2006), but shorter relative tospecies of Aetiocetus.

FRONTAL

The laterally extended supraorbital process of thefrontal is not depressed below the vertex, but insteadlies in nearly the same plane as the cranial vertex(Fig. 1C). This is the primitive condition shared witharchaeocetes, other toothed mysticetes (e.g. Janjucetus

Figure 6. Aetiocetus weltoni, UCMP 122900, holotype; stereophotographs of left portion of basicranium and posteriorregion of palate.

Figure 7. Aetiocetus weltoni, UCMP 122900, sketch ofright temporal wall showing configuration of cranialelements. al = alisphenoid, fo = foramen pseudovale, fr =frontal, of = orbital fissure, pal = palatine, par = parietal,pt = pterygoid, sq = squamosal.



and Mammalodon) and stem edentulous mysticetes(e.g. Eomysticetus). Unlike archaeocetes, however, thesupraorbital process in species of Aetiocetus, Janjuce-tus, Mammalodon and Eomysticetus is broadly over-lapped by the posteromedial portions of the rostralbones (nasal, premaxilla and maxilla). In Aetiocetusweltoni and A. cotylalveus there are well-developedpreorbital and postorbital processes of the frontals thatare separated by a laterally concave (in dorsal aspect)orbital rim. In contrast, the orbital rim in A. polyden-tatus is only slightly concave. In lateral aspect theorbital rim is distinctly arched in all species of Aetio-cetus, and although the orbit diameters (OD) areimpressive in size; when compared with CBL they arerelatively smaller than in Janjucetus (OD/CBL = 12%in Aetiocetus weltoni compared with 24% in Janjucetushunderi; Fitzgerald, 2006). It should be noted thatcomparing OD to CBL in J. hunderi might be mis-leading because of this taxon’s uniquely shortenedrostrum. A more accurate metric might be to compareOD with occipital condyle width (ODW), but unfortu-nately the occipital condyles are not preserved in theholotype of J. hunderi. Another possible metric wouldbe comparison of OD with zygomatic width (ZW).Using this metric J. hunderi still has relatively largerorbits (OD/ZW = 27% in A. weltoni vs. 36% in J.hunderi), although not as markedly large as implied bythe OD/CBL metric.

A well-developed temporal crest occurs on the pos-teromedial border of each supraorbital process inAetiocetus weltoni (as noted by Barnes et al., 1995),but does not extend onto the postorbital process(Fig. 1A). The right and left temporal crests convergeposteromedially to merge with the sagittal ridge onthe parietal. Although the frontal–parietal (coronal)suture on the vertex is somewhat difficult to interpretbecause of minor deformation, the configuration ofthis suture appears to be broadly M-shaped asdescribed below for the parietal. Laterally, the coronalsuture ascends the temporal wall nearly verticallyfrom the alisphenoid until a level approximately25 mm below the cranial vertex at which point thesuture begins to slope anteriorly to intersect the tem-poral crest. From here the suture turns medially tofollow the ridge of a smaller coronal crest and thenturns and converges posteromedially at the midlinewith a similar coronal crest on the other side. Thesecoronal crests create a conspicuous V-shaped sulcus inthe frontal on the vertex of the skull (Fig. 2) and maybe age-related features.

The frontal exposure on the vertex is rather short,measuring ~20 mm in the anteroposterior dimension.In the holotype of Aetiocetus cotylalveus the frontalexposure is longer, measuring ~60 mm in the antero-posterior dimension. This same measurement in theholotype of A. polydentatus is only about 15 mm.

Although not clearly described by Fitzgerald (2006),the frontal exposure in Janjucetus appears to berelatively long (~100 mm).

In the holotype of Aetiocetus weltoni, several smallcalibre supraorbital foramina occur on the medialportion of the frontal. The posteriormost foramen islocated within 9 mm of the frontal–maxilla suture.Anterolateral to this single foramen is a closely posi-tioned pair of supraorbital foramina on the rightfrontal that lie within 15 mm of the frontal–maxillasuture. On the left frontal only a single anteriorforamen is preserved and is located approximately16 mm from the suture. Chonecetus goedertorum hasa greater number of supraorbital foramina, many ofwhich are associated with distinct, but delicate, elon-gate sulci.

On the ventral surface of the supraorbital processof the frontal the optic canal is broad and poorlydefined and occupies most of the exposed ventralsurface. The long axis of the optic canal forms ananterolaterally directed angle of approximately 45° tothe sagittal plane. The distinct dorsal and ventralorbital crests noted in Dorudon by Uhen (2004) areabsent in aetiocetids, although such crests do occur inmost fossil and living edentulous mysticetes (e.g.Aglaocetus, Diorocetus, Parietobalaena, Balaenopteraand Eschrichtius). A number of frontal foramina occuron the roof of the optic canal near its middle. Ante-riorly, the frontal forms a long transverse suture withthe lacrimal. Anteromedially, the frontal forms anobscure contact with the orbitosphenoid, although theconfiguration of this bone is poorly defined. Postero-medially, the frontal forms the roof and anteriormargin of the large orbital fissure.

PTERYGOID

The lateral pterygoid lamina in Aetiocetus weltoni(UCMP 122900) is exposed for 23 mm on the medialwall of the temporal fossa where it forms a complex,almost foliate suture with the falciform process of thesquamosal (Fig. 7). The foramen pseudovale does notinvolve the pterygoid, but is instead confined to thesquamosal. Dorsally, the lateral pterygoid laminaforms a relatively horizontal butt joint with thealisphenoid. The anterodorsal corner of the pterygoidforms the posteroventral margin of the large ovalorbital fissure (Fig. 7). Anteriorly, the lateral ptery-goid lamina has a foliate sutural contact with thepalatine.

On the palate the contact of the pterygoid with thepalatine is unusual as already discussed. The ptery-goid hamulus is distinctly elongate and developed asa transversely broad, but dorsoventrally thin, spatu-late process (Figs 3, 6). The posterior 60% of thehamulus is finely roughened on its ventral surface,



the pattern being reminiscent of scars left by Sharp-ey’s fibres at muscle insertions. Diagenetic distortionin this area has narrowed the internal narial openingand forced the hamuli to impinge medially on thevomer. Figure 3 depicts the reconstructed morphologyof the internal narial region, which consists of elon-gate slit-like choane divided medially by the ventrallyflared vomerine crest. The anterior margin of theinternal narial opening is formed by the palatine,while the lateral margin is formed by the pterygoidhamulus. The holotype of A. cotylalveus is also dis-torted in this region, but appears to have the samegeneral morphology. Emlong (1966) erroneously iden-tified the pterygoid hamuli of A. cotylalveus asthe ‘transversely narrowed posterior ends of thepalatines’ (p. 12). This error led Emlong and laterKellogg (1969) to incorrectly conclude that the inter-nal nares opened far posteriorly at the level of thepterygoid sinus fossa rather than at the level of thetemporal fossa. The holotype of A. polydentatus (AMP12) is essentially undistorted in this region and pre-serves well-defined, slit-like internal narial openingsbordered laterally by elongate pterygoid hamuli.Unfortunately, this morphology is not adequately rep-resented in the published illustrations and photo-graphs of AMP 12.

The deeply bifurcated internal narial opening foundin species of Aetiocetus represents a unique morpho-logical feature that is not shared with any otherknown toothed or edentulous mysticete. It should benoted, however, that the condition in species ofChonecetus is currently unknown due to the lack ofpreserved pterygoids.

The pterygoid sinus fossa (anterior pterygoid sinusof Luo & Gingerich, 1999), as prepared on the left sideof UCMP 122900, is deep and ovoid and is constructedas in later diverging mysticetes from the lateral,medial and superior pterygoid laminae (Fig. 6). Thisis similar to the condition in basilosaurids and differsslightly from the condition in stem odontocetes wherethe roof of the pterygoid sinus fossa is formed bythe superior pterygoid lamina plus portions of thealisphenoid and basisphenoid (Luo & Gingerich,1999). In Janjucetus the superior pterygoid laminadoes not completely cover the medial portion of thealisphenoid (Fitzgerald, 2006) and thus the roof of thepterygoid sinus fossa is formed from alisphenoid pos-teriorly and pterygoid anteriorly. Unfortunately, theconstruction of the pterygoid sinus fossa in eomystice-tids and the stem ‘cetothere’ Micromysticetus is cur-rently unknown due to poor preservation of thisregion in the respective holotype specimens.

In extant mysticetes the pterygoid sinus fossa isenlarged anteriorly within the pterygoid bone andis floored in part by the pterygoid hamulus. Thisadvanced condition is not seen in Aetiocetus weltoni

where the pterygoid sinus fossa is more open ven-trally and not floored by any part of the pterygoid.

In Aetiocetus weltoni the surface of the superiorpterygoid lamina forming the roof of the pterygoidsinus fossa has a rugose surface texture. A distinctsulcus for the mandibular branch of the trigeminalnerve obliquely crosses the posterolateral corner ofthe superior pterygoid lamina to exit through theforamen psuedovale. This condition is similar to thatdescribed in adult basilosaurids (Luo & Gingerich,1999). In Aetiocetus weltoni the posterior portion ofthe lateral pterygoid lamina is overlapped externallyby the falciform process of the squamosal. In addition,the posterolateral corner of the superior pterygoidlamina behind the trigeminal sulcus underlaps andobscures the anterior process of the petrosal. Medi-ally, the pterygoid sinus fossa is coincident with theanterior portion of the peribullar sinus, which isexcavated into the dorsolateral portion of the medialpterygoid lamina. The posterior portion of the perib-ullar sinus appears to extend slightly onto the lateraledge of the basioccipital above the basioccipital crest.The posterior pterygoid lamina (vaginal process) istransversely broad and extends posteriorly to coverthe basioccipital–basisphenoid suture. The posteriorpterygoid lamina thus terminates at the level of thepostglenoid process of the squamosal. The suturebetween the posterior pterygoid lamina and the nasalplate of the vomer is parasagittally orientated andoverrides the anterior edge of the basioccipital crest.

LACRIMAL

The lacrimals are well preserved in Aetiocetus weltoniand project laterally beyond the external surfaces ofthe preorbital process of the frontal and the antorbitalprocess of the maxilla. Unlike the lacrimal in extantmysticetes, which consists of a relatively small, flat-tened bone loosely situated between the preorbitalprocess of the frontal and the antorbital process of themaxilla, the lacrimal of A. weltoni is intimately joinedin articulation with the maxilla and frontal. In dorsalaspect the lacrimal has a large, narrowly oval expo-sure at the back of the rostrum and deeply invadesthe lateral margin of the base of the ascendingprocess of the maxilla (Figs 1A, 2). The lacrimal expo-sure on the left side measures ~72 mm in the trans-verse dimension and is slightly depressed below thelevel of the adjacent frontal and maxilla.

The rounded medial portion of the lacrimal inUCMP 122900 is ~17 mm in anteroposterior diameterand distinctly broader than the lateral portion, whichmeasures only 8.5 mm in the same dimension. Theanterior margin of the lacrimal is overlapped along atransversely elongate scarf joint by the prominentantorbital process of the maxilla. The maxilla forms



a sharp transverse crest in this region, whereasmedially the maxilla–lacrimal contact is flattenedand more subdued. Posteriorly, the maxilla–lacrimalcontact is only ~20 mm in transverse length before itterminates at the base of the ascending process of themaxilla. From this point the lacrimal continues lat-erally and wraps around the preorbital process of thefrontal. Here the lacrimal narrows as it is pinchedbetween the preorbital process of the frontal and theantorbital process of the maxilla. As mentioned, thelateral border of the lacrimal extends beyond both ofthese processes and is developed as a rugose projec-tion that is anteroposteriorly narrow and dorsoven-trally elongate. Within the orbit the lacrimal has arelatively large exposure where it forms a deeplyserrated suture with the frontal. There is no evidenceof a lacrimal canal, although this area is not wellpreserved in UCMP 122900.

A similar lacrimal with large dorsal exposure thatdeeply invades the maxilla occurs in Aetiocetus coty-lalveus (USNM 25210), A. polydentatus (AMP 12) andChonecetus goedertorum (LACM 131146). Emlong(1966) briefly described the large lacrimal of A. coty-lalveus and illustrated it (fig. 4). Barnes et al. (1995),however, did not mention the lacrimal in theirdescriptions of aetiocetids and omitted this bone fromtheir skull reconstructions. As reported by Fitzgerald(2006) the lacrimal of Janjucetus hunderi is enlargedboth posteromedially and anterolaterally and paral-lels the anterior edge of the supraorbital process ofthe frontal. Overall the lacrimal is thick and rodshaped and restricted to a level below the surfaceof the supraorbital process, anteroposteriorly. InDorudon atrox the lacrimal is small and largely con-cealed dorsally by the frontal. The large rostral expo-sure of the lacrimal appears to be a feature uniqueto aetiocetids and has not been reported in anypreviously described archaeocetes, odontocetes ormysticetes.

JUGAL

Portions of both jugals are preserved in UCMP122900. The right jugal is more complete anteriorlyand is a delicate strut-like element closely appressedto the infraorbital plate of the maxilla. Although theanterior contact of the jugal is obscured by crushingon this side, it appears that the jugal underlies theextreme posterolateral corner of the infraorbitalplate, but quickly crosses anteriorly over onto thedorsal side of the infraorbital plate to become sand-wiched between the maxilla and the lacrimal. On theleft side, only the posterior end of the jugal is pre-served. This narrow, spatulate remnant is broadlyoverlapped by the anterior tip of the zygomaticprocess of the squamosal (Fig. 6) to create a scarf joint

41 mm in length. The jugal here measures only10 mm in transverse width and 7–8 mm in dorsoven-tral thickness, and is lodged in a narrow facet onthe anteroventral corner of the zygomatic process.Overall, the jugal is a rather delicate, strap-likeelement with a relatively short articulation with theanteroposterior tip of the zygomatic process of thesquamosal. Emlong (1966) erroneously consideredthe jugal of A. cotylalveus to be sandwiched betweenthe zygomatic process of the squamosal and the pos-torbital process of the frontal. His error was probablydue to post-depositional distortion in USNM 25210.As reported by Fitzgerald (2006) Janjucetus hunderihas a thick and sturdy jugal that underlaps theextreme anterior tip of the zygomatic process.Although the anterior 60% of the jugal of Dourdonatrox is also strap-like, the posterior portion of thejugal has a much longer and more robust articulationwith the zygomatic process (Uhen, 2004).

PARIETAL

The parietal is incompletely preserved in UCMP122900 and is missing the posterodorsal portion onthe vertex. In lateral aspect the dorsal surface ofthe parietal on the vertex is at the same level as thefrontal and was presumably only slightly below thelevel of the apex of the supraoccipital as in the holo-type of Aetiocetus polydentatus (AMP 12). The dis-tinctly depressed position and dorsally concaveprofile of the parietal in the holotype of A. cotyla-lveus (USNM 25210) may be the result of post-depositional deformation. The undistorted parietal inJanjucetus hunderi appears to represent the gener-alized condition in toothed mysticetes (i.e. frontaland parietal at same level and close to level of apexof supraoccipital).

In UCMP 122900 the squamosal–parietal andalisphenoid–parietal sutures on the left temporal wallare distinct. The parietal–frontal (coronal) suture,however, is partially obscured. On the vertex, thecoronal suture is broadly M-shaped, with a narrowsliver of frontal extending posteriorly between thelobate anterior wings of the parietals (Fig. 2). Asimilar narrow frontal sliver occurs in A. polydentatus(AMP 12). In A. cotylalveus (USNM 25210) theposterior frontal extension is transversely broaderbetween the lobate anterior wings of the parietals.The coronal suture in Chonecetus goedertorum(LACM 131146) is roughly W-shaped with a bifidposterior frontal extension divided by a short anteriorparietal extension. In this taxon the anterior wings ofthe parietals are sharply angular rather than lobate(Barnes et al., 1995: fig. 10). The coronal suture inJanjucetus hunderi also has sharply angular anteriorwings of the parietals, but lacks the bifid posterior



frontal extension (Fitzgerald, 2006, electronic supple-ment, fig. S3).

Although most of the supraoccipital shield inUCMP 122900 is missing (and thus the occipital–parietal suture), the degree of parietal exposure onthe vertex can be estimated at 65 mm if the lobateanterior wings of the parietals are included. If onlythe medial posteriorly retracted region of the pari-etals at the midline is considered, the degree of pari-etal exposure is estimated at 40 mm. In Aetiocetuscotylalveus the maximum and minimum parietalexposures are 61.8 and 34.5 mm, respectively. Theminimum parietal length roughly corresponds to thepreserved length of the low sagittal ridge in UCMP122900. Both A. cotylalveus (USNM 25210) and A.tomitai (AMP 2) have a flattened cranial vertex withno sagittal ridge or crest, while A. polydentatus (AMP12) has a very short sagittal crest measuring ~10 mmin length that descends sharply from the apex of thesupraoccipital to the vertex. Although broken poste-riorly, a distinct sagittal ridge (measuring at least14 mm in length) does occur in UCMP 122900 and isbroadly triangular in cross-section in contrast to thenarrower (and shorter) sagittal crest in A. polydenta-tus. Janjucetus hunderi (NMV P216929) has a dis-tinct, low sagittal crest (Fitzgerald, 2006) thatextends nearly the full length of the parietal(~80 mm).

In A. weltoni the transverse width of the parietalacross the intertemporal region is relatively narrow(68 mm) and broadly rounded as in archaeocetes.In A. cotylalveus, A. tomitai and Janjucetus hunderithe cross-sectional shape of the intertemporal regionis similar, although distinctly broader (85, 86 and~80 mm, respectively). In A. polydentatus the pari-etals are marked by distinct laterally placed tempo-ral crests that diverge anterolaterally towards thesupraorbital processes of the frontals. These crestscome to within 20 mm of each other at the apex ofthe supraoccipital, but do not meet. Lateral to thesecrests, the parietal wall in A. polydentatus is nearlyvertical giving the intertemporal region a squared-off cross-section in contrast to the more ovoid cross-sectional shape seen in other species of Aetiocetus(e.g. A. weltoni, A. cotylalveus and A. tomitai). Jan-jucetus and Mammalodon appear also to possess theovoid cross-section (Fitzgerald, 2006). In this lightthe conclusion by Geisler & Sanders (2003) that anovoid cross-sectional shape of the parietals is anaetiocetid synapomorphy is in error. These authorsalso suggested that aetiocetids lacked a sagittalcrest.

SUPRAOCCIPITAL

The anterior 60% of the supraoccipital shield in Aetio-cetus weltoni (UCMP 122900) is missing and only the

region immediately dorsal to the foramen magnum ispreserved. The morphology of this region suggeststhat the shield was rather vertically orientated pos-teriorly and abruptly inclined anteriorly as alsoobserved in A. cotylalveus (USNM 25210) and A.polydentatus (AMP 12). In both of the latter speci-mens a distinct external occipital crest is preserved atthe midline on the dorsal region of the supraoccipital.On either side of this central occipital crest, near thedorsolateral margins of the shield, is an elevated,semicircular, rugose boss for attachment of the neckextensor musculature. Between the muscle attach-ment and the occipital crest the surface of the shieldis depressed to form right and left concave embay-ments. It is assumed that these features of thesupraoccipital shield as preserved in the holotypes ofA. cotylalveus and A. polydentatus were also presentin A. weltoni.

More posteriorly on the right side of the occipitalshield in UCMP 122900 is a distinct dorsal condy-loid fossa (in the exoccipital) just above the dorsalmargin of the occipital condyle (indistinct in USNM25210). Similar fossae appear in this same positionin Chonecetus goedertorum and later diverging mys-ticetes and have been reported in Janjucetushunderi (Fitzgerald, 2006). Dorsal to these fossae inUCMP 122900 the remnant of the supraoccipitalshield is nearly vertical, but becomes more anteri-orly inclined toward the broken vertex. The lamb-doidal crests are entirely missing except for a smallremnant at the posteromedial corner of the squamo-sal fossa. This remnant suggests that the lambdoi-dal crest was rather sharp edged and extended onlyslightly dorsolaterally as in the holotype of Aetioce-tus cotylalveus and unlike the condition in the holo-type of A. polydentatus (AMP 12) where thelambdoidal crest broadly overhangs the parietal wallin the temporal fossa. In the latter taxon the occipi-tal shield has an overall triangular shape in con-trast to the more broadly arcuate shield of A.cotylalveus (Barnes et al., 1995). Janjucetus andMammalodon have smaller and more vertically ori-entated occipital shields, which like A. polydentatushave lambdoidal crests that flare dorsally and ante-rolaterally to overhang the temporal wall (Fitzger-ald, 2006).

EXOCCIPITAL

Dorsally, the exoccipital–supraoccipital contact isobscure due to sutural fusion. The occipital condylesare nearly complete and have a generally reniformshape in posterior aspect. Their dorsal transversewidth is greater than the ventral width. The dorsalintercondylar notch is broad and shallow, while theventral intercondylar notch is narrower and more



recessed. The foramen magnum is broadly quadrate.The condyles occupy a short, elevated pedicle. Asprepared there are no obvious condyloid foramina.

The preserved left paroccipital process is veryrobust (i.e. anteroposteriorly thick) (Fig. 4A). In pos-terior aspect the planar surface of the paroccipitalprocess is roughly quadrate and extends ventrally toa level below the basioccipital crest. This appearsto be the condition in all species of Aetiocetus. InChonecetus goedertorum, instead, the irregular poste-rior surface of the paroccipital process is more deli-cate (i.e. anteropostiorly thin) and lobate (i.e. notquadrate) and does not extend ventral to the basioc-cipital crest. This appears also to be the condition inJanjucetus hunderi. In A. weltoni the ventral surfaceof the paroccipital process is rugose and has an ante-rior embayment adjacent to the well-defined sulcusfor the facial nerve (Fig. 6). This anterior embaymentmay be homologous with the paroccipital concavity ofodontocetes, which forms the posterior border of theposterior sinus as noted in archaeocetes (Oelschlager,1986). This feature appears to be a symplesio-morphy of Cetacea although it has been erroneouslyregarded as a synapomorphy of the Platanistoidea +Eurhinodelphoidea + Delphinida (Muizon, 1984,1991). The medial wall of the posterior sinus isformed by a curved portion of the exoccipital thatproduces a crescent-shaped margin ventrally (alsoobserved in the holotypes of A. cotylalveus, A. poly-dentatus and C. goedertorum). A poorly formed, smallbony cup near the roof of the conical posterior sinus isprobably the point of articulation of the stylohyalwith the exoccipital. The posterior process of thepetrosal is tightly fused to the exoccipital along theirmutual contact.

The portion of the exoccipital immediately lateral tothe basioccipital crest is marked by an elongatefinger-like projection 20 mm in length (broken inUSNM 25210). Lateral to this projection is thenarrow and well-defined jugular notch. Lateral to thejugular notch is the paroccipital process. There is noseparate hypoglossal foramen on the ventral surfaceof the exoccipital. Instead, the passage for the hypo-glossal cranial nerve (XII) is confluent with the pos-terior lacerate foramen (cranial hiatus). This is thecondition in basilosaurids, mysticetes and odontocetes(Luo & Gingerich, 1999).

BASIOCCIPITAL

The basioccipital is exposed for a short distancebehind the posterior terminations of the pterygoidand vomer and is marked laterally by prominent,dorsoventrally thickened basioccipital crests (Fig. 6),a mysticete synapomorphy (Fitzgerald, 2006). Theventral surface of each crest is ovoid and has a rugose

texture for attachment of the neck flexor muscula-ture. The posterior border of the basioccipital crest isenlarged and roughly pyramidal in shape and projectsposteroventrally beneath the posteromedial corner ofthe exoccipital to form a conspicuous notch betweenthese two bones. Similar basioccipital crests occur inAetiocetus cotylalveus (USNM 25210) and A. polyden-tatus (AMP 12). As noted by Fitzgerald (2006) thebasioccipital crest of Janjucetus hunderi is morebulbous and transversely elongated along an axisparallel with the external auditory meatus. A similarcondition occurs in Morawanocetus yabuki (AMP 1).Overall, the basioccipital in species of Aetiocetus ismore anteroposteriorly elongate and transverselynarrow.

SQUAMOSAL

A deep squamosal fossa extends from the posteriormargin of the temporal fossa to the juncture betweenthe zygomatic crest and the lambdoidal crest. Thisfossa is elongate in the parasagittal plane with asubtle anterolateral ‘pocket’ or second fossa (fideBarnes et al., 1995). The composite squamosal fossa isbounded laterally by a thick (not keeled) zygomaticcrest on the dorsal margin of the zygomatic process. Adistinct prominence occurs on the posterior portion ofthe zygomatic crest at the back of the anterolateralpocket of the squamosal fossa. Anteriorly, the zygo-matic crest aligns with the medial, rather thanlateral, edge of the zygomatic process.

The zygomatic process is elongate, narrow and ori-entated in the parasagittal plane. In lateral aspect(Fig. 1C) the zygomatic process is thinnest near itsmiddle and expanded both anteriorly and posteriorly.The zygomatic process is dorsoventrally thickest ante-riorly as in other species of Aetiocetus (Barnes et al.,1995). The lateral side of the zygomatic process isroughly planar, while the medial side is more convex.As mentioned previously, a distinct elongate facet ispresent on the ventral side of the anterolateral cornerof the zygomatic process. This facet houses thenarrow and spatulate posterior portion of the jugal.

The posteroexternal portion of the squamosallateral and anterior to the exoccipital forms a rugoseand posterolaterally orientated vertical platform forinsertion of neck extensor muscles. Muscle insertionsubdivisions are not obvious, but probably includedthe splenius, mastohumeralis, sternomastoideus andtrachelomastoideus (Schulte, 1916). There is no well-formed post-tympanic process of the squamosal andinstead this area is slightly concave and terminatesventrally against a laterally projecting posteriorprocess of the petrotympanic complex.

Medially, the squamosal contacts the parietal onthe temporal wall by a sinuous and slightly elevated



serrated suture that is dorsally arched. The suture isroughly parallel to the floor of the squamosal fossaanteriorly and rises rather abruptly posteriorly tomeet the lambdoidal crest (Fig. 2). Anteriorly, thesquamosal does not form a distinct pterygoid process,but instead makes a blunt, nearly vertical butt jointwith the pterygoid (Fig. 7). The foramen pseudovale isovoid and measures 13–14 mm in length and 6 mm inheight. The foramen is entirely contained within thesquamosal, with a delicate intra-squamosal sutureextending anteriorly to the squamosal/alisphenoidsuture. The falciform process broadly underlaps theforamen and terminates ~14 mm from its anteriorrim. A second, but smaller (12 mm by 3 mm), nar-rowly ovoid foramen occurs at the apex of the falci-form process in the region of the triple junctionformed by the squamosal, alisphenoid and pterygoidsutures. Both foramina align along a horizontalflexure that anteriorly follows the slightly deformedpterygoid/alisphenoid suture. The homology of thisforamen is uncertain.

The postglenoid process of the squamosal isdirected ventrally (not posteroventrally) and tapersdistally to a sharp edge (Fig. 1C). It is relatively thinanteroposteriorly, in contrast to the distally thickenedpostglenoid processes of more derived edentulousmysticetes. The postglenoid ridge is short and medi-ally ascends sharply to form a premeatal crest. Thereis no obvious spinous process of the squamosal, asthis area appears to be obscured on the left side bythe crushed posterior pedicle of the petrotympaniccomplex. A distinct sigmoid fossa occurs on theextreme medial portion of the postglenoid ridge justlateral to the epitympanic recess on the petrosal(Fig. 6). The sigmoid fossa received the dorsal marginof the sigmoid process of the tympanic as in proto-cetids (Geisler & Sanders, 2003) and basilosaurids(Luo & Gingerich, 1999) and indicates a morecomplex articulation between the tympanic bulla andbasicranium than occurs in later diverging edentu-lous mysticetes (i.e. three areas of articulation vs.only two). Sigmoid fossae of similar morphology occurin the holotypes of Chonecetus goedertorum, C.sookensis and J. hunderi. The holotype of Aetiocetuscotylalveus preserves the sigmoid process of the lefttympanic bulla still in articulation with the sigmoidfossa of the squamosal. The presence of a spinyprocess of the squamosal at the squamosal–petrosalsuture adjacent to the sigmoid fossa as seen inarchaeocetes and stem odontocetes cannot be con-firmed in A. weltoni due to poor preservation.

The glenoid fossa is distinctly concave and facesanteriorly, a feature consistent with that of basilosau-rids and other aetiocetids and unlike the more planarand even convex glenoid fossae of later divergingedentulous mysticetes. The curvilinear length of the

fossa indicates a broad articulation with the man-dibular condyle and suggests a synovial temporoman-dibular joint. Anteromedial to the glenoid fossa thepreglenoid part of the squamosal (sensu Luo & Gin-gerich, 1999) is broadly concave and extends only tothe level of the foramen pseudovale. The anterolateralmargin of the preglenoid region is extended into thetemporal fossa as a distinct, anteriorly directed sub-temporal crest (Fig. 6). This condition results in adistinct swelling (as viewed in dorsal aspect) of thesquamosal at the back of the temporal fossa and infront of the squamosal fossa. Homologous anteriorlydirected subtemporal crests occur in other aetiocetidsand appear to be absent in Janjucetus and Mammal-odon. Zygorhiza kochii (e.g. USNM 11962) has abroader, anteriorly orientated subtemporal crest,while in Dorudon atrox the large subtemporal crest ismore anterolaterally directed (Uhen, 2004).

The entoglenoid region of the squamosal (sensu Luo& Gingerich, 1999) is well separated (i.e. ventral)from the preglenoid region, is transversely narrowand does not have a facet for articulation with thetympanic bulla (Fig. 6). In basilosaurids the entogle-noid region is anteriorly broader where it contactedthe tympanic bulla (Luo & Gingerich, 1999). In Aetio-cetus weltoni the tympanic bulla contacted the ante-rior process of the petrosal anteriorly, the sigmoidfossa of the squamosal medially and the posteriorprocess of the tympanic posteriorly. This representsan intermediate condition between basilosaurids (andstem odontocetes) and later diverging mysticetes. Inthe latter taxa the tympanic bulla only articulateswith the anterior process of the petrosal and theposterior process of the tympanic via the anteriorand posterior pedicles, respectively. The entoglenoidmargin in A. weltoni is retracted dorsally to exposethe lateral portion of the anterior process of thepetrosal (tegmen tympani). Similar, dorsally retractedentoglenoid margins have been reported in Simocetusrayi (Fordyce, 2002) and occur in Eomysticetus whit-morei and Chonecetus goedertorum (our pers. observ.).

The external auditory meatus is deeply incisedbetween the postglenoid and posterior tympanic pro-cesses of the squamosal (Fig. 6). The latter is poorlydefined because instead of a distinct posterior meatalcrest there is a continuous near-vertical, planarsurface extending from the squamosal onto the pos-terior process of the tympanic. The infundibulum ofthe external auditory meatus measures ~41 mm intransverse length and varies in width from 10 mm(laterally) to 5 mm (medially). Extending dorsallyfrom the dorsolateral rim of the external auditorymeatus is a smooth-sided, vertical sulcus measuring~8 mm in width. This sulcus is deeply impressedventrally, but dorsally merges smoothly with thelateral surface of the zygomatic crest. A similar ver-



tically orientated sulcus occurs in other species ofAetiocetus, as well as in species of Chonecetus.

PETROSAL

The left petrosal is incompletely preserved and ismissing essentially the entire promontorium. Theanterior process is artificially bifurcated longitudi-nally by a fracture, but otherwise appears to be com-plete. It is tightly appressed to the squamosal andunderlapped by the posterolateral corner of the supe-rior pterygoid lamina at the back of the pterygoidsinus fossa. The anterior process is broad dorsoven-trally (~20 mm), thin transversely (~10 mm) and mea-sures ~23 mm in length (as preserved). A short(2 mm) remnant of the anterior pedicle occurs nearthe posteroventral corner of the anterior process andthere is no obvious lateral tuberosity. The medialsurface of the anterior process is generally smooth. Asharply defined embayment posterior to the anteriorpedicle represents the mallear fossa, which is par-tially blocked laterally by the dislodged incus.Further preparation and removal of the incus isneeded to facilitate its description. The same is truefor the stapes, which is still seated in the fenestraovalis. Anterior to the fenestra ovalis is a sharplydefined groove for the tensor tympani measuring10 mm in length. The facial sulcus is blocked anteri-orly by the stapes and incus, while posteriorly thesulcus is obscured by the crushed posterior pedicle ofthe tympanic. The posterolateral corner of the prom-ontorium is partially preserved including a bulbouscaudal tympanic process and the floor of a crushedfenestra rotunda. The stapedial muscle fossa isobscured by the caudal tympanic process. Posterior tothe caudal tympanic process the posterolaterally ori-entated facial sulcus is confluent dorsally with anarrow and elongate sylomastoid fossa, which doesnot extend onto the posterior process of the tympanic.The petrosal portion of the posterior process is diffi-cult to distinguish from the tympanic portion due toapparent fusion of these two elements. The remnantof the crushed posterior pedicle, however, does serveas a landmark to distinguish at least a portion of thetympanic contribution to the posterior process. It isalso clear that the nearly vertically orientated ante-rior portion of the posterior process is formed entirelyby the tympanic. This vertical surface is covered bythe thin posterior tympanic process of the squamosal,which defines the posterior margin of the externalauditory meatus. The petrosal portion of the posteriorprocess is marked by the shallow, smooth-walledfacial sulcus, which measures 37 mm in length andaverages approximately 6 mm in width. The compos-ite posterior process is laterally elongated and termi-nates in a dorsoventrally flared ‘mastoid’ process with

a rugose external surface texture. The apparentfusion of the petrosal and tympanic portions of theposterior process together with the transverse elon-gation of the process are viewed as derived features(see phylogenetic discussion).

As preserved, the left petrosal of Aetiocetus weltonidoes not retain enough morphology (Fig. 6) to providea complete understanding of the morphology of thisimportant region. However, some useful comparisonsare possible. The more complete left petrosal ofChonecetus goedertorum (LACM 131146) has asimilar dorsoventrally broad and transversely narrowanterior process, but a shorter composite posteriorprocess. In C. goedertorum the groove for the tensortympani is less distinct, as is the mallear fossa. Thepromontorium is complete and its dorsoventraldimension is nearly twice that of the transversedimension. The petrosal of Eomysticetus whitmorei(ChM PV4253) also has a similar shaped anteriorprocess, but a much shorter and more odontocete-likecomposite posterior process. In E. whitmorei the pet-rosal and tympanic portions of the posterior processmake contact along a spiraled suture and remainunfused. In addition the posterior process is not aselongate and is not fused with the adjacent exoccipitaland squamosal in E. whitmorei. The left petrosal ofJanjucetus hunderi is well illustrated, but largelyundescribed by Fitzgerald (2006). From the publishedfigures, however, it appears to have a similar shapedanterior process, but a relatively short posteriorprocess.

TYMPANIC BULLA

The left tympanic bulla of UCMP 122900 wasremoved from the skull for study (Fig. 8). Overall, thebulla is well preserved. The anterior pedicle is welldefined, while the posterior pedicle is crushed. Thedorsal edge of the lateral wall is broken and foldedinto the tympanic cavity and the sigmoid process isdeformed posteriorly and collapsed onto the conicalprocess (Fig. 8A).

The posterior rim of the sigmoid process is thickand broadly arched dorsally and embayed ventrally tosupport the tympanic membrane. The general mor-phology of this process closely resembles that of laterdiverging mysticetes. Thus, it seems that the mys-ticete sigmoid process is generalized and highly con-served. The left malleus is still connected to thesigmoid process by a distorted gonial process. Thesurfaces for articulation with the incus have beenartificially rotated ~90° to face posteriorly and lieclose to the sigmoid process. The conical process isrelatively small and domed. The ventral base ofthe conical process measures approximately 13 mmanteroposteriorly, while its dorsoventral dimension



measures only about 2 mm. Posterior to the conicalprocess is the elliptical foramen, which is obscuredby the deformed posterior pedicle. Anterior to thesigmoid process on the external surface of the lateralwall is a deep lateral furrow that extends nearly tothe ventral surface. The lateral wall of the bullaanterior to the lateral furrow is medially deflectedrelative to the lateral surface of the bulla posterior tothe lateral furrow (Fig. 8B). This condition gives thebulla a distinctly broader posterior transverse width(relative to anterior width) in ventral aspect. Asimilar condition occurs in Eomysticetus whitmorei(ChM PV4253). Both taxa also preserve a weak ante-rior spine at the anterior margin of the bulla beneaththe Eustachian notch (Fig. 8C).

The posterior margin of the bulla has a well-definedinterpromontorial notch between the inner and outerposterior prominences (Fig. 8C). The outer posteriorprominence is extended more posteriorly than the

inner posterior prominence. The interpromontorialnotch is continuous with a short, but broad medianfurrow, which extends midway onto the ventralsurface of the bulla. The involucral ridge medial tothis furrow is weakly developed (relative to the dis-tinct involucral ridges in later diverging mysticetes)and lies nearly at the same level as the ventral bullarsurface. Laterally, there is no obvious main ridgedeveloped on the ventral surface, although the ante-rior extension of the outer posterior prominence couldbe considered a rudimentary main ridge. This subtlefeature occurs near the ventrolateral border of thebulla. In posterior aspect the outer posterior promi-nence is smoothly globose. In contrast the innerposterior prominence is transversely broader andcharacterized by a distinct, transversely orientatedlinear ridge that extends across the interpromontorialnotch to contact the posteromedial corner of the outerposterior prominence (Fig. 8D). Similar transverse

Figure 8. Aetiocetus weltoni, UCMP 122900, holotype left tympanic bulla. A, medial view; B, lateral view; C, ventralview; D, dorsal view; E, posterior view.



ridges occur in all species of Aetiocetus examined, butdo not occur in Morawanocetus yabuki or the majorityof later diverging mysticetes. However, the basaledentulous mysticete Eomysticetus whitmorei has asharp transverse ridge on the inner posterior promi-nence as does the basilosaurid archaeocete Zygorhizakochii.

The medial surface of the involucrum is broadlycurved dorsoventrally and marked by poorly definedtransverse creases (Fig. 8A, D). There is no obviousdorsal posterior prominence. Although the dorsalmargin of the lateral wall is collapsed into the tym-panic cavity, the general outline of the large andanteriorly open Eustachian notch is preserved. Inter-nally, the tympanic cavity is divided into anterior andposterior portions by a broad, transverse median ridge.Posterior to this ridge is the involucral elevation.

The posterior process of the right bulla is solidlyfused to the posterior process of the petrosal andremains in the skull. As mentioned, the right sigmoidprocess is articulated with the sigmoid fossa of thesquamosal. Although the suture between the poste-rior process of the tympanic bulla and the petrosal isfused, it appears that the posterior process of thepetrosal extends laterally nearly the full length of thecomposite posterior process

ALISPHENOID

The alisphenoid on the ventral surface of the skullappears to be completely concealed by the superiorpterygoid lamina in the pterygoid sinus fossa. Incontrast, the alisphenoid is large and broadly exposedon the temporal wall and is not covered by the ptery-goid as in later diverging edentulous mysticetes(Fig. 7). Although the alisphenoid–frontal suture ismostly obscured, the sutural contacts with the pari-etal, squamosal, and pterygoid are fairly well definedand serrated. Overall, the alisphenoid is roughlyrhomboidal in shape with a vertical diagonal measur-ing ~60 mm and a horizontal diagonal measuring70 mm. The alisphenoid–frontal suture appears tobe relatively linear and extends anteroventrally for~45 mm before terminating at the posterodorsalcorner of the large orbital fissure. The generallylinear alisphenoid–parietal suture slopes posteroven-trally and measures ~55 mm. The alisphenoid–squamosal suture is more complex and consists of anirregular anteroventrally sloping segment (measuring~26 mm) dorsal to the foramen pseudovale and ashorter linear horizontal segment (measuring~15 mm) anteroventral to the foramen pseudovale(Fig. 7). This latter segment contacts the dorsalportion of the falciform process of the squamosal. Thesomewhat irregular alisphenoid–pterygoid suture isgenerally horizontal and extends anteriorly for~30 mm before terminating at the posteroventral

corner of the large orbital fissure. Several smallforamina are unevenly distributed along thealisphenoid–parietal suture. As in the basilosauridarchaeocetes Zygorhiza and Dorudon, the foramenpseudovale does not perforate the alisphenoid exter-nally, but is instead confined to the squamosal(Kellogg, 1936; Uhen, 2004).

Because the temporal walls of many fossil cetaceanspecimens are incompletely prepared, it is oftenimpossible to interpret the configuration of thealisphenoid and adjacent bones. This is unfortunategiven the potential value that clearly defined morpho-logical patterns for these bones would provide to ourunderstanding of cetacean cranial morphogenesis(Ridewood, 1923). The alisphenoid may be especiallyuseful in light of the apparent progressive covering ofthis bone by the pterygoid during mysticete evolution(our pers. observ.).

BASISPHENOID

The basisphenoid is completely covered by the vomerand pterygoid in Aetiocetus weltoni. The absence of aclear basisphenoid–basioccipital suture suggests thesuture is fused, a condition indicative of develop-mental maturity. The posterodorsal corner of thebasisphenoid is exposed behind the medial pterygoidlamina on the medial border of the peribullar sinus.The sinus is excavated slightly into the basisphenoid.

DENTARY

The right and left dentaries are nearly complete.Both, however, are missing the angular process. Thecoronoid process is very large and lobate as noted byBarnes et al. (1995), with a sharp-edged posteriormargin (Figs 9, 10). The coronoid process is posteri-orly inclined forming an angle of ~60° with the longaxis of the ramus. It is not laterally deflected. Aconspicuous strut forms the anterolateral corner ofthe process and defines the anterior border of themasseteric fossa. The straight posterior edge of thecoronoid process is inclined posterodorsally and over-hangs the neck of the dentary. As the posteriormargin descends towards the base of the coronoidprocess it curves posteriorly to merge with the neck ata point ~35 mm in front of the mandibular condyle(Table 2A). The coronoid process of Aetiocetus poly-dentatus more broadly overhangs the mandibularneck and is more fin-shaped than lobate. InMorawanocetus yabukii the coronoid process isrectangular in general outline with a long, straightanterior margin, a blunt dorsal margin and a broadlyconcave posterior margin. In Janjucetus and Mam-malodon the coronoid process has a broader baseand is more triangular, with the posterior margin



inclined posteroventrally and not overhanging theneck (Fitzgerald, 2006). In Eomysticetus the coronoidprocess also has a broad base, but is laterallydeflected at its dorsal tip. The posterior margin of thecoronoid process in this stem edentulous mysticetedoes not overhang the mandibular neck and in-stead descends posteroventrally towards the condyle(Sanders & Barnes, 2002). The coronoid processes ofbasilosaurids are also broad at their bases, but haveposterior margins that are nearly vertical to slightlyoverhanging (Kellogg, 1936; Uhen, 2004).

In A. weltoni the neck of the dentary (between thebase of the coronoid process and the mandibularcondyle) is very short anteroposteriorly (~15mm). Inlater diverging mysticetes the neck becomes moreelongated corresponding to the migration of the coro-noid process from a position within the temporal fossato a position anterior to the temporal fossa and belowthe orbit. The lateral portion of the neck is the site ofinsertion of the deep masseter (Schulte, 1916). Themandibular condyle of A. weltoni is similar in shapeto the condyles of modern odontocetes in being trans-versely broad dorsally. In posterior aspect the condyleis roughly triangular in shape, with the dorsal surfaceof the condyle corresponding to the base of the tri-angle. The lateral margin of the condyle extends~13 mm beyond the lateral surface of the dentary. Inlateral aspect the articular surface of the condyle is

broadly convex posteriorly. In medial aspect themargin of the condyle forms the posterior border ofthe large mandibular foramen (Figs 9, 10). The dorsalmargin of this foramen extends anteriorly in a broadarc and descends toward the ventral margin of thedentary. On the left dentary the foramen is 74 mm inlength and ~76 mm in maximum height. On the rightdentary the anterior margin of the mandibularforamen is complete and indicates that the margincontinued ventrally and curved posteriorly toward thebroken angular process. It appears that the ventralmargin of the mandibular foramen remained slightlyabove the ventral edge of the dentary and did notmerge with it as in some odontocetes. The large sizeof the mandibular foramen and mandibular canal,coupled with the thin bone on the lateral wall of thedentary adjacent to the foramen, is reminiscent of thecondition in basilosaurid archaeocetes and the panbone of odontocetes.

The horizontal ramus is elongate and narrow trans-versely and dorsoventrally. There are five distinctmental foramina on the lateral surface positioned 8,13, 16, 22.6 and 30 cm, respectively, behind the tip ofthe dentary. These foramina occur 25, 20, 17, 15 and17 mm, respectively, below the alveolar margin. Allforamina are anteroposteriorly elongate and openanteriorly. At the anterior tip of the dentary are twoclosely appressed circular foramina that open anteri-

Figure 9. Aetiocetus weltoni, UCMP 122900, holotype left dentary. A, lateral view; B, medial view.



orly. The anteriormost foramen is ~13 mm below thealveolar margin, while the other is approximately18 mm below the alveolar margin. The dorsomedialedge of the ramus lacks the distinctive gingivalforamina present in edentulous mysticetes. This isconsistent with the occurrence of an adult dentition inA. weltoni, as gingival foramina form in the region ofthe ossified mandibular alveolar groove during mys-ticete ontogeny (our pers. observ.).

There are portions of eight postcanine teeth pre-served on the left dentary, but only the pc3 crownremains. As noted by Barnes et al. (1995) there arealveoli for 12 teeth (eight postcanines), indicating arudimentary polydont condition. Diastemata alongthe toothrow are variable in length as follows: i1-2~8 mm, i2-3 ~20 mm, i3-c ~21 mm, c-pc1 ~19 mm,pc1-2 ~17 mm, pc2-3 ~25 mm, pc3-4 ~15 mm, pc4-5~18 mm, pc5-6 ~16 mm, pc6-7 ~24 mm, and pc7-8~5 mm. In occlusal aspect the alveolar margin of theramus is nearly straight lingually, but distinctly scal-loped labially at least to the level of p3. The anteriorscalloping is formed by elevated alveolar crests adja-cent to each tooth and by embayed embrasure pits or

sulci adjacent to each diastema. Strong posteroven-trally orientated embrasure pits/sulci occur betweeni2-3, i3-c, c-pc1 and pc1-2. The orientation of theseembrasure pits/sulci suggests that the crowns of theoccluding upper teeth were orientated in a similardirection. The best-developed embrasure pits occurbetween pc1-2 and pc2-3. Posterior to pc3 the labialportion of the alveolar margin in lateral aspect is acontinuous linear crest that extends to behind the lastpostcanine tooth. In occlusal aspect the alveolarregion between pc3 and pc8 has indistinct embrasurepits occurring between all preserved alveoli andlingual to the alveolar crest. These embrasure pitsreceived the crowns of the upper dentition. All man-dibular teeth appear to be single rooted, although theroots of pc4 and at least pc7 are more anteroposteri-orly elongate and appear to represent coalesced ante-rior and posterior roots. As reported by Fitzgerald(2006) the dentary of Janjucetus hunderi has seven,closely spaced, double-rooted postcanine teeth.

In transverse section the labial surface of the hori-zontal ramus is generally planar, while the lingualmargin is convex, at least to the level of pc5. In lateral

Figure 10. Aetiocetus weltoni. A, lateral view of reconstructed left dentary; B, medial view of reconstructed left dentary.



aspect the dorsal and ventral margins of the horizon-tal ramus are roughly parallel. The anterior marginof the dentary is inclined anterodorsally, forming anobtuse angle of approximately 145° with the ventralmargin. Posteriorly, the ventral margin is marked bya distinct genial tuberosity.

The anterior tip of the ramus on the lingual sidepreserves a distinct bony shelf and groove with acurvilinear length of ~40 mm anteroposteriorly(Figs 9, 10). Anterior to the bony shelf and groove is anarrow mandibular symphysis measuring 30 mm inanteroposterior length. The symphyseal surface issmooth and not pitted as in cetaceans with bony/sutural symphyses. (i.e. archaeocetes and odonto-cetes). Similar and, presumably, homologous featuresoccur on the dentaries of extant mysticetes and areassociated with a fibrocartilage skeleton that looselyconnects the right and left dentaries (Pivorunas,1977). This loose mandibular symphysis has beencorrelated with mandibular kinesis at the intra-mandibular joint (Lambertsen, 1983) in extant mys-ticetes and is a feature unique to extant mysticetecetaceans. Its occurrence in Aetiocetus weltoni indi-cates that mandibular kinesis evolved before the lossof teeth in later diverging edentulous mysticetes.Although the lower jaws of Mammalodon colliveri andJanjucetus hunderi also have non-sutured mandibularsymphyses, these toothed mysticetes apparently lacka distinct bony shelf as reported by Fitzgerald (2006).

In dorsal aspect the dentary is broadly convex lat-erally, a condition shared with all later divergingmysticetes and distinctly different from the laterallyconcave mandibular configuration of archaeocetes andstem odontocetes. In UCMP 122900 the alveolarmargins of the lower and upper jaws align with eachother rather precisely during occlusion. This contrastswith the condition in balaenids, neobalaenids andeven some balaenopterids in which the lower man-dibular arch is conspicuously bowed wider than therostrum.

UPPER DENTITION

The upper dentition of Aetiocetus weltoni is nearlycomplete and preserved on both the right and the leftsides of the rostrum in UCMP 122900. As noted byBarnes et al. (1995) there are 11 upper teeth on eachside identified as I1-3, C, P1-4 and M1-3. The leftdentary was removed from the skull to provide a closerview of the upper dentition. Unfortunately, there hasbeen post-depositional damage to the skull such thatthe anterior teeth (13-P3) have been distorted slightlywith their roots rotated laterally along a longitudinalaxis. This is not the case on the right side where thesesame teeth have a more vertical orientation in therostrum. Because the right dentary is still articulated

with the skull, it preserves the occlusal relationships ofthe upper and lower dentition, which consists of alter-nating, interlocking rows of teeth from the rostral tipto M3 (Fig. 4C). The general pattern of the upperdentition consists of procumbent anterior teeth (I1-3and C, P1) and less procumbent posterior teeth(P2-M3). Most of the procumbency in the anteriordentition is exhibited by the tooth roots, which basedon CT imaging are seen to be orientated nearly hori-zontally within their respective alveoli. The crowns ofthese same teeth exhibit the greatest amount of cur-vature. As previously described, broad diastemataoccur between all teeth. The left dentition has beenprepared and except where noted is the basis for thefollowing descriptions.

The left I1 is nearly complete except for the apex ofthe crown, which is broken; the right I1 is missing.The I1 crown is only slightly compressed transversely(Table 3) and compared with I2, 3 is nearly conical.Longitudinally, the crown is only slightly curved lin-gually. On the lingual surface the enamel is markedby fine longitudinal ridges, while the labial enamel issmooth. A distinct crista occurs on the anterolingualcorner of the crown. Breakage on the posterior marginof the crown makes it impossible to determine if therewas a similar crista in this area. There are no obviouswear facets preserved. The root of the tooth is emer-gent (i.e. 7–10 mm of the root is exposed outside of thealveolus) and has a larger diameter than the crown(Table 3). Internally (as revealed in CT imaging), theroot is orientated anteroposteriorly and extends hori-zontally ~16 mm into the premaxilla.

The left I2 is a larger tooth and has a more trans-versely compressed crown than I1 (Table 3). Longitu-dinally, the crown is slightly curved lingually. Intransverse section the lingual surface is nearly planarrelative to the more convex labial surface. Althoughthe crown is damaged by fractures, it is relativelycomplete and can be seen to terminate in a bluntlytriangular apex capped by a tiny obliquely orientated(dorsoventral) wear facet. On the lingual surface theenamel is marked by fine longitudinal ridges, whilethe labial enamel is smooth. A distinct anterior cristais preserved on the crown as in I1. The height of thecrown is ~13 mm. As was true for I1 the root of I2 isemergent and exposed for at least 10 mm outside ofthe alveolus. Internally (as revealed in CT imaging),the root is orientated anteroposteriorly and extends~14 mm into the premaxilla.

The crown of I3 is unbroken (Table 3) and termi-nates in a sharply pointed apex that lacks a wearfacet. Longitudinally, the crown is more stronglycurved lingually than I2. The crown is transversallycompressed (planoconvex) with distinct anterior andposterior cristae. The cristae are smooth and notcrenulated. On the lingual surface the enamel is



marked by fine longitudinal ridges, while the labialenamel is smooth. The height of the crown is approxi-mately 13 mm. The labial surface of the crown isbroadly convex both longitudinally and transversely.The exposed portion of the root is damaged, but likethe other incisors extends well outside the alveolus.Internally (as revealed in CT imaging), the root isorientated anteroposteriorly and extends horizontallyapproximately 7 mm into the premaxilla. The I3of Aetiocetus polydentatus is morphologically verysimilar.

The upper canine (Fig. 4C) closely resembles I3 insize and shape (Table 3). There are distinct anteriorand posterior cristae, a planoconvex cross-section,a sharply pointed apex, a lingual enamel surfacewith fine longitudinal ridges and a broadly convex(longitudinally and transversely) labial surface withsmooth enamel. The height of the crown is approxi-mately 12 mm. In lingual aspect the apex of the crownis placed slightly more posteriorly relative to the morecentrally placed apex in I3. This gives the posteriorcrista a steeper slope than the anterior crista. Inter-nally (as revealed in CT imaging), the root is orien-tated anteroposteriorly and extends horizontally~10 mm into the premaxilla. The canine of Aetiocetuspolydentatus is morphologically very similar.

The P1 closely resembles the canine in all featuresincluding the distinct anterior and posterior cristae,planoconvex cross-section, sharply pointed apex withasymmetric lingual profile, lingual enamel with finelongitudinal ridges and a broadly convex (longitudi-nally and transversely) labial surface with smoothenamel. An abraded wear facet occurs on the anteriorcrista near the base of the crown. The height of thecrown is ~12 mm. Internally (as revealed in CTimaging), the root is orientated more transverselythan in the anterior teeth and extends only about4 mm into the maxilla. The first postcanine of Aetio-cetus polydentatus is morphologically very similar.Emlong (1966) described an isolated tooth found withUSNM 25210 as P1. A tooth catalogued with USNM25210 and bearing the number 6 appears to bethis same tooth. The well-preserved crown closelyresembles the P1 of UCMP 122900 with the notableexception of a distinct denticle at the base of thesharp posterior crista and a much weaker denticle atthe base of the anterior crista. It is possible thatAetiocetus weltoni possessed similar denticles, whichhave been lost due to wear. The large wear facet onthe posterior crista of the UCMP 122900 P1 may havebeen the site of a posterior denticle. The root is wellpreserved in the USNM specimen and is relativelyshort with a posteriorly deflected bulbous termina-tion. There is only a single root.

The crown of P2 has a different morphology fromthe anterior teeth. Although the left P2 is damaged

and slightly corroded, it is clear that the anterior andposterior cristae possess numerous distinct triangu-lar denticles. On the posterior crista are three strongdenticles. Anteriorly, the denticles are three innumber but are smaller and less distinct. The crownapex is damaged. The lingual surface of the crownis sculpted with a series of fine longitudinal ridges,as noted for the anterior teeth. The labial enamelis smooth. In transverse cross-section the lingualsurface is generally planar in contrast to the convexlabial surface. The crown was broken at its root atone point and may not have been reattached cor-rectly. As currently positioned the crown is orientatedoblique to the sagittal plane and seems to be in anunnatural position. The root is broader than thecrown and is not as emergent as in the more anteriorteeth. The root is also more bulbous than the rootsof preceding teeth. Internally (as revealed in CTimaging), the root is orientated transversely andextends only about 4 mm into the maxilla. Thereappears to have only been a single root. The P2assigned by Emlong (1966) to A. cotylalveus is wellpreserved and has a lingually bowed crown withthree strong denticles on the sharp posterior cristaand two weaker denticles at the base of the anteriorcrista. The enamel on the lingual surface has 12–14fine longitudinal ridges, while the labial enamel issmooth. The root is single-rooted. The second postca-nine tooth of A. polydentatus has smooth anteriorand posterior cristae and lacks the distinct denticlesnoted in A. weltoni and A. cotylalveus.

The crown of P3 is poorly preserved, with all sur-faces exhibiting some degree of damage. This includesthe anterior and posterior margins, which do notpreserve the denticles seen on P2. It is likely thatdenticles occurred on these teeth, but this cannot beconfirmed. The crown of P3 was also broken at its rootand seems to have been reattached in an unnaturalposition. The root of P3 is quite bulbous and isdistinctly broader than the crown. Internally (asrevealed in CT imaging), the root is orientated trans-versely and extends only about 4 mm into themaxilla. There is only a single root. The isolated toothassigned by Emlong (1966) to this tooth position isbetter preserved and characterized by a series of foursmall worn denticles arranged along the anteriorcrista. A large wear facet on the ventral edge of theposterior crista removed any trace of denticles thatmay have been positioned here. The crown is plano-convex in cross-section, the lingual enamel is finelyridged and the labial enamel is smooth. Emlong(1966) describes a slight longitudinal median grooveon both the internal and the external sides of the rootthat he interprets as evidence for two coalesced roots.The third postcanine of A. polydentatus has a crownof similar size to the P3 of A. cotylalveus, but has only



two denticles, one each positioned near the base of theanterior and posterior cristae.

The crown of the left P4 of A. weltoni is damaged tothe point that little is left to describe. The isolatedtooth of A. cotylalveus assigned by Emlong (1966) toP4 has a crown with four small denticles arrangedalong the anterior crista and a large wear facet alongthe posterior crista. In transverse section both thelabial and the lingual surfaces are convex in contrastto the planoconvex cross-sections observed in I1-P3.The labial enamel is still smooth, but the lingualenamel consists of fine, longitudinal, wavy ridges.According to Emlong (1966) the root has a longitudi-nal median groove suggesting two coalesced roots.The fourth postcanine tooth of A. polydentatus has acrown with two small denticles near the base of theanterior crista and three stronger denticles on theposterior crista. In lingual aspect the crown is nearlysymmetrical with the posterior margin being onlyslightly steeper than the anterior margin. In trans-verse section both the labial and the lingual surfacesare convex.

The crown of the left M1 of A. weltoni is alsodamaged to the point that little is left to describe. Theisolated tooth of A. cotylalveus assigned by Emlong(1966) to M1 has a crown with three small denticlesarranged along the posterior crista and a small wearfacet along the anterior crista. A single tiny denticleremains at the base of the anterior crista. In trans-verse section both the labial and the lingual surfacesare convex. The labial enamel is still smooth, but thelingual enamel consists of fine wavy (not linear) lon-gitudinal ridges. According to Emlong (1966) the rootcondition is still single-rooted with no longitudinalmedian groove. The fifth postcanine tooth of A. poly-dentatus has a crown with three small denticles nearthe base of the anterior crista and two stronger den-ticles on the posterior crista. The crown is symmetricalin lingual aspect and in transverse section both thelabial and the lingual surfaces are convex. As in otherspecies of Aetiocetus, the labial enamel is smooth,while the lingual enamel has fine, longitudinal ridges.

The crown of the left M2 in UCMP 122900 has asymmetrical triangular shape in lateral aspect andsomewhat resembles the teeth of squalodontid odon-tocetes (Kellogg, 1923). The posterior margin of thecrown is slightly damaged, but preserves one acces-sory denticle and the broken base of another denticle.The apex of the crown (paracone) is sharply pointedand the anterior margin is slightly damaged but doesnot appear to have possessed any denticles. Overall,the crown is strongly convex lingually and weaklyconvex labially. Although abraded, the enamel on thelingual side of the crown bears faint remnants of finelongitudinal ridges near the cingulum. The anterol-ingual portion of the crown is convex both trans-

versely and longitudinally, while the posterolingualportion is more sharply angled at its juncture withthe posterior crista. The root of M2 is damaged. Thesixth postcanine tooth of A. polydentatus is toodamaged for comparisons to be made.

M3 resembles M2 in its overall triangular andsymmetrical shape, but is slightly smaller in size(Table 3). The posterior margin of the crown preservesthe worn bases of two closely appressed denticles. Thecrown apex (paracone) is sharply pointed and theanterior margin of the crown preserves a broad wearfacet that has obliterated any trace of denticles thatmay have been present. The lingual surface of thecrown is generally convex transversely with weaklydeveloped wavy longitudinal ridges adjacent to thecingulum. The labial margin is more broadly convexboth transversely and longitudinally and has smoothenamel. The root of this tooth is not well preserved.The seventh postcanine tooth of A. polydentatus has acrown with three small denticles evenly spaced alongthe anterior and posterior cristae. The crown is sym-metrical in lingual aspect and in transverse sectionand both the labial and the lingual surfaces areconvex. The crown of the eighth postcanine tooth of A.polydentatus preserves four strong denticles on theanterior crista and a large wear facet on the posteriormargin. Like PC7 both the labial and the lingualsurfaces of the crown are convex. The crown of theninth postcanine tooth preserves only a single den-ticle on the anterior margin, the remainder of whichis characterized by a large wear facet. A similar wearfacet has removed all traces of denticles from theposterior margin of the crown.

LOWER DENTITION

Portions of the lower dentition are preserved in bothdentaries (Figs 4C, 9; Table 2B). As noted by Barneset al. (1995) there are 12 lower teeth on each sideidentified as i1-3, c, pc1-8. The left dentary has beenremoved from the skull making it possible to examinethe left toothrow closely. Unfortunately, there arebroken crowns and missing teeth for most tooth posi-tions. The right dentary is still articulated with theskull and, although most of the right dentition ispreserved except for i1, the teeth posterior to pc1 arehidden from view by the overriding maxilla. Thegeneral pattern of the lower dentition consists ofprocumbent anterior teeth (i1-3, c and pc1) and lessprocumbent posterior teeth (pc2-pc8). All teeth areseparated by distinct diastemata, although the lengthof diastemata decreases from front to back.

The crown of the left i1 is broken and only the rootis preserved. The root is damaged but appears to havebeen circular in cross-section.

The left i2 preserves a short segment of the base ofthe crown. Very fine longitudinal enamel ridges are



preserved only on the lingual side of the crown, whichhas a smaller diameter than the more bulbous root.The labial surface of the right i2 is nearly completeand is broadly convex transversely (Table 5). Thecrown terminates in a sharply pointed apex that isbordered anteriorly by a sharp crista and posteriorlyby a more delicate crista. The crown height isapproximately 15 mm. As preserved, the right tooth isalmost out of its alveolus.

The crown of the left i3 is broken while that on theright side is complete. The crown is broadly convextransversely and longitudinally on the labial side.The enamel surface on the labial side is generallysmooth, although there are several broad longitudinalridges that converge toward the apex. The diameter ofthe crown is greater than that of i1, but is stillsmaller than the diameter of the more bulbous root(Table 5). Although closely appressed to the lateralmargin of the maxilla, it appears that the crown ofthe right i3 had anterior and posterior cristae. Thissuggests that the lingual margin was probably flat-tened and not convex.

The canine is absent from the left dentary and isincompletely preserved in the right dentary. Unfortu-nately, the apex of the crown is broken. The labialsurface is broadly convex transversely with smoothenamel and broad longitudinal ridges similar to thosenoted on i3. The ventral 5 mm of the anterior crista isclearly evident on the crown. The condition of theposterior margin of the crown, however, is obscureddue to its proximity to the maxilla. The diameter ofthe root is slightly greater than that of the crown.

Only the root of the left pc1 is preserved in thedentary while the right pc1 is slightly more completewith at least the base of the crown preserved. A cleanbreak reveals distinct concentric growth lines in thecentre of the tooth near the base of the crown. Thebase of the crown is swollen. The root is ovoid incross-section with the longest dimension orientatedanteroposteriorly. The root is broader and morebulbous than the base of the crown.

The left pc2 is not preserved and the right pc2 isincomplete. The apex of the crown protrudes into theoverlying maxilla and has been broken by post-depositional compaction. The anterior margin of thecrown preserves two distinct triangular denticlessimilar to those noted on the crown of P2. The pos-terior margin is broken. The root is ovoid in cross-section with the longest dimension orientatedanteroposteriorly. The root is broader and morebulbous than the base of the crown.

The tooth preserved in the alveolus for the left pc3is nearly complete although twisted 180° around itslong axis. Overall the crown is narrowly triangular inlateral aspect terminating in a sharply pointed apex.The enamel on the labial surface is generally smooth,

although there are two broad longitudinal ridges con-verging apically. The posterolabial corner of the crownpreserves a relatively large wear facet while theanterolabial corner of the crown is unworn. The dis-tinct anterior crista is smooth and lacks any evidenceof the denticles noted on p2. The enamel on thelingual surface of the crown is roughened and char-acterized by numerous fine longitudinal ridges. Theright pc3 is probably complete, but because of thetight occlusion of the upper and lower jaws much ofthe tooth is obscured. The dorsal surface of themaxilla immediately above the right pc3 is slightlyelevated probably due to protrusion of the pc3 crowninto the ventral surface of the maxilla. As preservedand prepared the labial portions of the crown of pc3are partially visible. The enamel is smooth and theanterior margin of the tooth bears at least twosharply pointed denticles as for pc2. The diameter ofthe crown appears to be nearly equal to that of theovoid root (Table 4).

The crown of the left pc4 is broken and only the rootis preserved. The root is apparently ovoid with adistinct longitudinal sulcus on the labial side suggest-ing fusion of anterior and posterior roots. The diam-eter of the crown is distinctly smaller than that of theroot. Although the right pc4 is probably preserved inthe skull, it has not been prepared and cannot bedescribed. The same can be said for the molar serieson the right side.

On the left side there are no crowns preserved forthe posterior teeth and only the roots of pc6 and pc7.The root of pc7 is bilobed with central sulci labiallyand lingually suggesting fusion of anterior and pos-terior roots. The open alveolus for the last postcaninetooth is shallow and ovoid. On the right side a portionof the crown of pc7 is partially exposed betweenthe closely appressed crowns of the adjacent teeth.Overall the crown is symmetrical and broadly trian-gular in lateral aspect terminating in a sharplypointed protoconid apex. The enamel on the labialsurface is smooth and the anterior margin of thecrown preserves three distinct accessory denticles.The denticle nearest the base of the crown is thesmallest and shows apical wear.

DISCUSSIONCOMPARISONS WITH OTHER AETIOCETIDS

Barnes et al. (1995) offered three characters to distin-guish Aetiocetus weltoni from A. cotylalveus, includ-ing: rostral bones extending posteriorly to a pointbetween the posterior margins of the supraorbitalprocess of the frontals, squamosal fossa more anter-orposteriorly elongate, and dental count 11/12. One ofthese features, the dental count, as noted by them, is



Tab

le5.

Ch

arac

ter–

taxo

nm

atri

xsh

owin

gdi

stri

buti

onof

cran

ial

and

den

tal

char

acte

rst

ates

amon

gse

lect

edto

oth

edm

ysti

cete

san

dre

late

dta

xa

Taxo

n

Ch

arac

ter

10

20

30

40

50

12345

67890

12345

67890

12345

67890

12345

67890

12345

6

Dor

ud

on00000

00000

00000

00000

00000

00000

00000

00000

00000

0Z

ygor

hiz

a00000

00000

00000

00000

00000

00000

00100

01000

00000

0A

goro

phiu

s00003

02120

10?0?

00?01

0?0??

0?221

00111

?1001

00000

0T

asm

acet

us

10203

03120

10100

00112

22000

02320

03002

21000

00000

1M

amm

alod

on200?1

1002?

?????

?0??1

11?1?

100?1

01010

??02?

00000

?J

anju

cetu

s20021

10100

101??

011?1

?1010

10010

01010

?002?

0?001

1A

etio

cetu

sto

mit

ai?00??

1?121

1????

?10??

?????

??2?1

0?100

102??

0000?

2A

etio

cetu

sw

elto

ni

20011

22011

11210

11012

10101

21211

02100

11121

00001

2A

etio

cetu

sco

tylv

aeu

s20011

20121

11210

11012

10101

??211

02100

11121

00001

2A

etio

cetu

spo

lyd

enta

tus

20012

10011

11210

?1012

20101

21211

02110

11221

00001

2M

oraw

anoc

etu

s200??

10121

1???0

10111

110?1

?1011

0110?

111??

00001

1C

hon

ecet

us

goed

erto

rum

20021

1012?

01210

11012

???11

21211

0?100

10210

00001

2E

omys

tice

tus

01000

11000

??00?

?000?

????1

202?1

0?010

2?2??

11111

2C

row

nM

ysti

cete

sa1/21/20/3a

0/1b120

10001/2

1000?

????1/2

23b11

1?00/10/1

20/1110/1

11111

b

Ch

arac

ter

stat

essc

ored

0re

pres

ent

ance

stra

lst

ate,

1–3

repr

esen

tde

rive

dst

ates

.a

=0/

2/3,

b=

0/1/

2.M

issi

ng

data

are

scor

edas

?.



probably the same in both species (the dentary isunknown for A. cotylalveus). The length of the squa-mosal fossa was not quantified by Barnes et al. (1995)and when compared by us for the two taxa seems tobe a subjective distinction. Rostral bones that extendfurther posteriorly in A. weltoni are indicative of thegreater degree of telescoping of the skull. We consid-ered the possibility that this difference was due toindividual variation within a single species. To testthis we made a limited comparison of the region ofrostral–cranial interdigitation in several specimensof a modern mysticete, Balaenoptera acutorostrata(minke whale), and in an aetiocetid, Chonecetus goed-ertorum, known from two skulls. In neither case wasthere significant variation in the extension of therostral bones. This result supports the recognition ofAetiocetus weltoni and A. cotylalveus as distinctspecies, as does the narrower transverse intertempo-ral width, more robust zygomatic process of thesquamosal and less lobate anterior parietal wings ofA. weltoni.

Aetiocetus polydentatus differs most strikingly fromA weltoni and A. cotylalveus in its greater degree ofpolydonty and its longer and broader nasals (Barneset al., 1995). Ichishima (2005) suggested that A. poly-dentatus was not a member of Aetiocetus noting thefollowing character states: (1) parasagittal crests onthe parietals in the intertemporal region; (2) lateralmargin of maxilla bending ventrally; and (3) rela-tively distinct temporal crests on posterior margin ofsupraoccipital process of frontal. Fitzgerald (2006)agreed that Aetiocetus was probably paraphyletic andthat A. polydentatus was a more derived species. Insupport of this hypothesis, Fitzgerald (2006: supple-ment) noted the following character states: (4)ascending process of maxilla terminates slightly pos-terior to antorbital notch, at a point level with theanterior third (anteroposteriorly) of the supraorbitalprocess of frontal; (5) anteroposteriorly foreshortenedsupraorbital process of frontal; (6) poorly developedconcavity in lateral margin of supraorbital processsuch that the orbit opens laterally but not anterolat-erally and dorsolaterally; and (7) elongated intertem-poral region with well-developed sagittal crest. Inevaluating these seven purported distinguishing char-acters we made the following conclusions: (1) theparasagittal crests in the intertemporal region are anautapomorphy of A. polydentatus; (2) the ventraldeflection of the lateral margin of the maxilla isprobably a diagenetic distortion; (3) the strong tem-poral crests are an extension of the distinct parasag-ittal crests; (4) the apparent anterior termination ofthe ascending process of the maxilla is a preserva-tional artefact; (5) the apparent anteroposteriorlyshortened supraorbital process of the frontal is also apreservational artefact; (6) although the orbital rim is

less embayed, poor preservation may be contributingto the effect; and (7) the intertemporal region is nomore elongate than in other species of Aetiocetus. Weconclude from this review that although A. polyden-tatus does possess distinct features, when viewed in aphylogenetic context (see below) these features arenot sufficient to remove this taxon from the Aetiocetusclade.

AETIOCETIDS AS TRANSITIONAL TAXA

The toothed aetiocetids from Oligocene rocks of theNorth Pacific can play a key role in interpretations ofcetacean evolution because they are transitional ingrade between archaeocetes and extant edentulousmysticetes. In turn, Aetiocetus weltoni has importantimplications for mysticete phylogeny and for refiningthe definition and diagnosis of Aetiocetidae. The skulland dentary of A. weltoni preserve a number of mor-phological features that underline this transitionalposition of aetiocetids. The anterior placement of theexternal nares is reminiscent of dorudontine archaeo-cetes, as are the elongate nasals, dorsally placedsupraorbital frontal processes, distinct intertemporalparietal exposure, flat palate with teeth, rigidrostrum, large and lobate coronoid process of dentarypositioned in temporal fossa, and large mandibularfossa. Except for the flat, toothed palate and rela-tively rigid rostrum, all of these plesiomorphicfeatures also occur to some extent in the stem eden-tulous mysticete Eomysticetus whitmorei. However,both Eomysticetus and Aetiocetus possess importantfeatures unique to mysticetes, including an edentu-lous and transversely expanded infraorbital maxillaryplate, robust basioccipital crests, outwardly boweddentary and mandibular symphyseal groove. In thislight, the retention of an adult heterodont dentition inspecies of Aetiocetus emphasizes the mosaic nature ofthis group of toothed mysticetes.

It seems clear that mysticete evolution has at itscore involved an expansion of the oral cavity volume.Among extant mysticetes this has been accomplishedin a variety of ways. Balaenopterids have increasedoral volume through broadening of the rostrum and afunctional expansion of the ventral portion of the oralcavity via a ventral throat pouch (Pivorunas, 1979;Lambertsen, 1983). Balaenids have increased oralvolume through dorsal arching of the rostrum andelongation of the baleen plates (Pivorunas, 1979).Eschrichtius has increased oral volume in a some-what similar fashion with functional expansion of thedorsal portion of the oral cavity (i.e. rostral arching),but also utilizes a minor degree of ventral throatexpansion coupled with oral suction (Pivorunas,1979).

All groups rely on cranial and mandibular kinesisto accommodate the increased oral cavity volume.



Cranial kinesis involves unfused and loosely joinedintra- and inter-rostral articulations (e.g. lack offusion between maxilla and premaxilla, betweenmaxilla and vomer, between maxilla and frontal, andbetween premaxilla and frontal) (Berta et al., 2006).In contrast, rostral sutures in archaeocetes and odon-tocetes are distinctly fused. Published accounts ofmysticete cranial kinesis, surprisingly, are largelyconfined to brief discussions of the loose nature ofmysticete rostral sutures (Miller, 1923; Kellogg,1928). Mandibular kinesis involves a loose mandibu-lar symphysis that allows a surprising degree of inde-pendent movement of the right and left dentaries atthe symphysis (Lambertsen, Ulrich & Straley, 1995).In both cases the kinetic anatomical conditionsdecrease torque-related strain during bulk feeding.Species of Aetiocetus possess some of the featurescorrelated with mandibular kinesis, but lack the fea-tures associated with rostral kinesis (i.e. they retainthe archaeocete condition of a rigid rostrum).

AETIOCETID DENTITION

Morphologically, the teeth of Aetiocetus weltoni aremarkedly similar to those of A. cotylalveus and A.polydentatus in size, enamel texture and overallshape, although not tooth count. All teeth have asmooth enamel surface on the lateral (labial) side ofthe crown, while the medial (lingual) side of thecrown has a roughened surface texture composed offine, generally longitudinal enamel ridges. Barneset al. (1995) considered species of Aetiocetus to behomodont; however, the dentitions are in fact weaklyheterodont. In all known species the crown of I1 isroughly conical, while those of I2, 3 are transverselycompressed, laterally and posteriorly curved, andhave sharp anterior and posterior cristae. The uppercanine and first postcanine closely resemble I3 withthe exception that they are less curved. In A. weltonithe P2 crown is planoconvex in transverse cross-section (planar lingually and convex labially) and hasthree small anterior and three posterior accessorydenticles on either side of the sharply pointed prin-cipal cusp. In A. polydentatus numerous accessorydenticles do not occur until the fourth postcaninetooth; the second postcanine lacks accessory denticlesand the third postcanine has only a single denticleanteriorly and posteriorly. In this taxon the crown ofthe second and third postcanines are planoconvex intransverse section, while that of the fourth postcanineis convex on both sides. In A. weltoni this change fromplanoconvex to convex–convex cross-sections probablyoccurs between P3, 4. In the latter taxon the crownsof M2, 3 are different from those of all preceding teethin being roughly triangular in lateral profile withsharply pointed principal cusps (paracones) flanked

posteriorly by two accessory denticles. The anteriorcristae of these teeth lack denticles. Thus, the denti-tion of A. weltoni consists of anterior caniniform teeth(I1-3, C, P1), central equally denticulate teeth (P2-4),and posterior asymmetrically denticulate teeth. Anisolated right M2? associated with the holotype ofMorowanocetus yabukii (AMP 1) preserves a verydifferent crown morphology with both roughenedlabial and lingual enamel surfaces consisting ofstrong irregular ridges converging on the apex, a lowparacone, and the worn bases of two anterior andthree posterior accessory denticles. The crown islarger than in any teeth of Aetiocetus spp. and mea-sures 16.3 mm in anteroposterior length, 7.96 mm intransverse width and 11.05 mm in crown height. Theaccessory denticles of M. yabukii are large and nearlythe same size as the paracone. The dentition of Jan-jucetus hunderi is also distinctly different from Aetio-cetus spp. and consists of heavily ridged enamel onboth the labial and the lingual surfaces of the crowns,large multiple accessory denticles on the posterioredge of P2-M2 and double rooted postcanines (Fitzger-ald, 2006). In Llanocetus only the morphology of thelower postcanine teeth has been described (Mitchell,1989), but these teeth are also quite different, withvery roughened labial and lingual enamel surfaces, arelatively small principal cusp (protoconid), andnumerous relatively large and palmate anterior andposterior accessory denticles.

THE PRESENCE OR ABSENCE OF BALEEN

IN AETIOCETIDS

The mosaic condition of aetiocetids with their mixtureof primitive dorudontine archaeocete characters, stemmysticete characters and adult dentition begs thequestion: ‘Did species of Aetiocetus possess baleen?’Previous workers think not. Barnes et al. (1995) wereequivocal in their treatment of this matter, althoughtheir stated dichotomy between aetiocetids and‘baleen-bearing mysticetes’ implies that they felt thataetiocetids lacked baleen. In a later report, Sanders &Barnes (2002) stated, ‘It seems most probable thatthe Aetiocetidae and the Mammalodontidae are sidebranches of Mysticeti in which baleen was neverdeveloped, the retention of teeth into adulthoodperhaps being a paedomorphic character that becamefirmly entrenched.’ In a more recent report, Geisler &Sanders (2003) suggest that baleen was absent inaetiocetids because these taxa lacked large vascularforamina on the palate. This latter point is discussedbelow.

Without preservation of actual baleen or impres-sions of baleen found with fossil aetiocetids this ques-tion cannot be unequivocally answered. The vascularsystem necessary to nourish baleen, however, does



leave bony landmarks on the palate of modern mys-ticetes. The clearest of these landmarks are thelateral foramina and associated sulci of the palate. Asdescribed recently by Deméré et al. (2008), the seriesof delicate lateral palatal foramina and sulci of Aetio-cetus weltoni are hypothesized as the landmarks of apalatal vascular system for rudimentary baleen.These authors contend that these lateral palatalforamina are homologous with the prominent baleennutrient foramina in edentulous mysticetes.

The palates of extant mysticetes are marked bynumerous foramina and associated sulci. A generaldistinction can be made between foramina positionednear and parallel to the midline and foramina morelaterally placed and often radially arranged. Thelatter foramina provide passage of the rich bloodsupply to the baleen racks (Walmsley, 1938). In fetalspecimens of modern mysticetes the lateral foraminahave not yet formed and instead there is a distinctopen alveolar groove running along the lateral marginof the flat palate (Ridewood, 1923). Embryologically,this alveolar groove is the site of the developingtemporary dentition, which passes through the bud,cap and bell stages of development before degradationand resorption of the deciduous tooth buds begins(Karlsen, 1962; Ishikawa & Amasaki, 1995; Ishikawaet al., 1999). Dermal papillae of the rudimentarybaleen plates begin to develop coincident with toothbud degradation. At the same time the open alveolargroove starts to fill with bone until finally the distinctlateral nutrient foramina begin to form. This pro-gressive ossification also involves a widening of thepalatal region above the developing baleen racks.

Given the close association between tooth andbaleen development observed in modern mysticetes itis plausible that we are seeing an ancient ontogenyexpressed in Aetiocetus. As described by Deméré et al.(2008), Aetiocetus weltoni possessed both baleen anda functional adult dentition and thus indicates thatthe evolution of mysticete bulk feeding involved astep-wise process from the ancestral condition ofteeth only, through an intermediate condition withboth teeth and baleen, to the derived condition withbaleen only. The form of this early baleen system ofAetiocetus is currently unknown, but perhaps con-sisted of small bundles of keratinized tubules similarto those occurring at the front and rear of the mainbaleen racks of modern balaenopterids. In Aetiocetussuch bundles could have formed a rudimentarybrush-like filter between the widely spaced teeth ofthe upper dentition. It is tempting to speculate thateven Llanocetus had rudimentary baleen, a hypoth-esis supported by the report of abundant fine groovesaround the tooth alveoli on the palate of thissouthern hemisphere Palaeogene toothed mysticete(Fordyce, 2003).

FEEDING APPARATUS OF AETIOCETIDS

The long rostrum and dentaries, the relatively poste-rior position of the coronoid process on the dentary,the large size of the coronoid process, the location ofthe coronoid process within the relatively large tem-poral fossa and restricted origins of the temporalmusculature (i.e. essential absence of the temporalmuscles from the supraorbital processes of the fron-tals) suggest that A. weltoni was capable of a snap-ping style jaw adduction, not unlike that proposed fordorudontine archaeocetes (Uhen, 2004) and stemodontocetes (Fordyce, 2002). This is the primitivecondition and suggests that A. weltoni did not utilizesustained mandibular abduction during filter feedingas in extant balaenids. These cranial and mandibularfeatures also seem counter to those necessary for theextreme depression of the dentary (Delta rotation)and lateral deflection at the tempormandibular joint(Omega rotation) reported in extant balaenopterids(Lambertsen et al., 1995). It is clear that species ofAetiocetus were capable of kinesis at the mandibularsymphysis (Alpha rotation), which was probablydeployed during expulsion of water from the enlargedoral cavity as the mouth was closing. Assuming thepresence of a rudimentary baleen apparatus, thewater flowing out of the closing mouth passedthrough the baleen trapping concentrations of school-ing prey species. In contrast, single prey were prob-ably captured by the functional adult dentition in theprimitive pierce feeding fashion employed by dorud-ontine archaeocetes. In both cases the size of preyspecies was limited by the spacing between baleenelements and/or teeth.

Questions of feeding strategy in toothed mysticeteshave recently been summarized by Fitzgerald (2006)who concludes that the crabeater seal analogue modelof tooth-aided filter feeding in these taxa is probablyincorrect. He cites the lack of closely appressedteeth with elaborately denticulated postcanine crownsas an obstacle to a filter-feeding interpretation.However, this argument assumes that the prey speciesare small and near the size of krill. The problem withthis approach is that it assumes filter feeding as astrategy and small size as a prey choice. Filter feedingat its core can be thought of as bulk feeding. Prey sizeis not part of this basic definition. In fact, contrary toconventional wisdom, not all living baleen whales eatkrill. Given the range of prey type and size seen inextant mysticetes, there is no reason to assume a priorithat species of Aetiocetus fed on small prey (i.e. krill)?Their bulk feeding behaviour could have focused onlarger prey (e.g. schooling sardines or squid). Theprincipal requirement for bulk feeding is enlargementof the oral cavity and aetiocetids had larger mouthsthan archaeocetes, a feature that can be viewed as an



adaptation for bulk feeding. In turn, their more widelyspaced and less lobate teeth could still be functional asa coarse sieve in filter feeding. The evolutionary devel-opment of baleen in aetiocetids in combination withtheir dentition would serve to expand the prey choicesfor these transitional mysticetes.

PHYLOGENETIC RELATIONSHIPSOF AETIOCETUS

Emlong (1966) in his original description of Aetiocetuscotylalveus made comparisons with basal odontocetes(Agorophius and Archaedelphis) and archaeocetes(Patriocetus, Microzeuglodon and Agriocetus). He con-cluded that ‘the degree and trend of telescoping aremore suggestive of the Mysticeti’ although the pres-ence of teeth prompted Emlong to exclude Aetiocetusfrom inclusion within this lineage. The presence ofseveral purported archaeocete features (e.g. narrowand very elongate intertemporal region and a func-tional adult dentition) supported its assignment tothis group.

The phylogenetic relationships of Aetiocetus toother aetiocetids has been discussed by Barnes et al.(1995) who recognized Chonecetus as the most basalaetiocetid with Aetiocetus positioned as a later diverg-ing taxon sharing a sister group relationship withMorawanocetus (Fig. 11). Barnes et al. (1995) wereunable to resolve relationships among species ofAetiocetus. Our hypothesis for the position of aetio-cetids and related species is shown in Figure 11.

Previous studies have identified different suites ofcharacters to diagnose the Aetiocetidae (charactersfrom prior studies are numbered sequentially below tofacilitate discussion). Barnes et al. (1995) originallyidentified five characters: (1) protuberance on premax-illa at anterolateral corner of nasals; (2) multiplemaxillary foramina; (3) groove in dentary for symphy-seal ligament; (4) zygomatic process long and slender;and (5) squamosal fossa short. This list was reduced tocharacters 1 and 3 by Sanders & Barnes (2002), whoconsidered only two characters as unique to aetio-cetids: (6) intertemporal region shortened anteropos-teriorly and widened transversely; and (7) elongate

Figure 11. Phylogenetic hypotheses of aetiocetid relationships. A, previous study based on Barnes et al. (1995); B, thisstudy.



notch present in posterior border of palatine at poste-rior edge of palate. Geisler & Sanders (2003) identifiedthe following three aetiocetid synapomorphies: (8)three or more infraorbital facial foramina; (9) premax-illae adjacent to and at posterior edge of the nasalopening overhang the maxillae; and (10) cross-sectionthrough intertemporal region, including parietalsovoid but sagittal crest absent. We re-evaluated thesecharacters and placed them into one or more of threecategories: (1) characters we use in the present analy-sis; (2) characters with an incorrect reported taxo-nomic distribution; and (3) characters minimallydescribed (e.g. lack of quantification). Our reallocationof these characters is summarized below. We found onecharacter to require no amendment and we used it inthe present study (#7). We found that six charactershave an incorrect distribution reported (#1, 2, 3, 4, 8,10). Finally, into the third category, we assigned twocharacters (#5, 6). Character #9 is equivalent to #1 asboth are describing the relative position of the premax-illa and maxilla adjacent to the external narialopening.

In our study we evaluated 46 craniodental char-acters (22 binary and 24 multistate; Table 5). Allcharacters and character-state transformations wereunweighted and all multistate characters were unor-dered. We considered 11 ingroup taxa: six aetiocetids(Aetiocetus tomitai, A. weltoni, A. cotylalveus, A. poly-dentatus, Morawanocetus yabukii and Chonecetusgoedertorum), two other toothed mysticetes (Janjuce-tus hunderi, Mammalodon colliveri), the basal eden-tulous mysticete Eomysticetus whitmorei, and allcrown mysticetes (i.e. Balaenidae, Balaenopteridae,Eschrichtiidae and Neobalaenidae). We excluded thepoorly preserved A. tomitai from later runs of thedata as its inclusion failed to resolve relationshipsamong other species of Aetiocetus. For outgroup taxawe employed the following: two dorudontine archaeo-cetes, Dorudon atrox and Zygorhiza kochii, and twobasal odontocetes, the extinct Agorophius sp. (ChM5852) and the extant ziphiid Tasmacetus shepardi.We used the branch and bound search option ofPAUP*vb10 (Swofford, 2001) with a maximum parsi-mony optimality criterion. All multistate characterswere treated as unordered. Character optimizationsexplored both ACCTRAN (which favours reversalsover parallelisms) and DELTRAN (which favoursparallelisms over reversals) although our preferredhypothesis of character evolution followed theDELTRAN optimization. Our rationale for this deci-sion is that for characters scored as unknown (?)for some taxa (e.g. Mammalodon) using ACCTRANwould result in hypothesizing the presence of certaincharacter states in taxa in which there is no evidencethat these states were ever present. The cladisticanalysis produced one most parsimonious tree with a

tree length of 106 steps and a rescaled consistencyindex of 0.4746 (Fig. 11). Based on our analysis werecognize the monophyletic groups/taxa discussedbelow. Bootstrap percentages based on 1000 replicateswere calculated and are shown in Figure 11. Mys-ticete monophyly was reasonably well supported(68%) as was monophyly of the edentulous mysticetes(96%). Mysticeti monophyly was supported by twounequivocal characters: mandibular symphysis notsutured (26) and anterior and postcanine teeth mod-erately heterodont, weakly heterodont, or absent (32).A clade containing Mammalodon and later divergingmysticetes is diagnosed by two unequivocal synapo-morphies: descending process developed as an infraor-bital plate (29; transformation from 2 to 1) and lateralmargins of maxillae thin (45). Mammalodon is distin-guished from other mysticetes by one equivocal char-acter: intertemporal region narrow (8; reversal 1 to 0).

AETIOCETIDAE

(AETIOCETUS + CHONECETUS + MORAWANOCETUS)

This monophyletic group is diagnosed by three un-equivocal synapomorphies (Fig. 11): zygomatic processof squamosal expanded near its anterior margin and atits posterior end but narrow in the middle (10), coro-noid process well developed with concave posteriormargin (27), and short overlap of jugal with squamosal(36). Three equivocal synapomorphies are potentiallydiagnostic of this clade: posterior sinus poorly devel-oped or absent (19), preglenoid portion of squamosalwith short anteriorly convex subtemporal crest extend-ing into temporal fossa (33) and lambdoidal crestsflush with temporal wall (34; reversal from 1 to 0).This clade is well supported (bootstrap value 70%).Morawanocetus, identified as the most basal aetio-cetid, is diagnosed by two equivocal characters: longbasicranial width (18) and posterior process of premax-illa contacting the frontal (46; transformation from 2 to1). Chonecetus and Aetiocetus spp. are united as sistertaxa by two unequivocal characters: lacrimal exposeddorsally (12) and posterior margin of palatine withnotch (14). Four additional equivocal characters arepotentially diagnostic of this clade: palate windowexposing vomer present and formed by maxilla andpremaxilla (13; transformation from 0 to 2), squamosalfossa with second fossa developed (17), posterior teethsingle rooted (20; transformation from 1 to 2, andposteriormost edge of nasals located on the posteriorhalf of the supraorbital process of frontal (28; trans-formation from 0 to 2). Chonecetus is diagnosed by fiveequivocal synapomorphies: triangular anterior exten-sion of parietal–frontal suture (4), zygomatic process ofsquamosal does not reach level of supraorbital processof frontal (11; reversal from 1 to 0), anterior position ofposterior end of ascending process of maxilla relative



to posterior extremity of nasal (37; reversal from 1 to0), posterior position of posterior extension of premax-illae relative to posterior extremity of maxillae (38;transformation from 1 to 2, and oblique groove onanterolateral face of zygomatic process of maxillapresent but low (39; transformation from 2 to 1).

AETIOCETUS

This clade is diagnosed by two unequivocal synapo-morphies: anterior and posterior denticles on poste-rior upper teeth three or fewer, small and simple (23)and anterior and postcanine teeth weakly heterodont(32; transformation from 1 to 2). An additional fourcharacters are potentially diagnostic of this clade:triangular anterior extension of parietal–frontalsuture (4), vertical enamel ridges only on lingualsurface of postcanine teeth (22; reversal from 1 to 0),teeth with wide diastemata (24; reversal from 1 to 0)and rostral width at antorbital notch (base) relative tooccipital condyle width > 170% (40). This clade is notwell supported (bootstrap value 58%). Aetiocetus coty-lalveus is identified as the most basal Aetiocetusspecies and is diagnosed by a single equivocal char-acter: supraorbital process of frontal with deeplyconcave orbital rim (6). The later diverging species ofAetiocetus (A. weltoni + A. polydentatus) are diag-nosed by two equivocal synapomorphies: intertempo-ral region narrow (8; reversal from 1 to 0) and sagittalcrest reduced (9). Aetiocetus weltoni is diagnosed bytwo equivocal characters: deeply concave orbital rim(6) and posteriormost edge of ascending process ofmaxilla terminates at the posterior border of thesupraorbital process of frontal (7; transformationfrom 0 to 2). Aetiocetus polydentatus is diagnosed byfour equivocal characters: broad nasals (5; transfor-mation from 1 to 2), a high tooth count (21; transfor-mation from 1 to 2), lambdoidal crests overhangingtemporal wall (34) and posterior extension of pre-maxilla posterior to posterior extremity of maxilla(38; transformation from 2 to 1).

CONCLUSIONS

Aetiocetids represent an important group of toothedmysticetes that document the morphological diversitythat existed among stem mysticetes during the Oli-gocene. The skull and dentary of Aetiocetus weltonipreserve a number of features that underline thetransitional nature of aetiocetids between dorudontinearchaeocetes and crown mysticetes. The possession ofan adult dentition in A. weltoni is seen as retention ofan ancestral condition. The lateral palatal foramina onthe palate of A. weltoni are interpreted as bony corre-lates for the presence of baleen. This heralds the majormacroevolutionary shift in mysticetes from pierce

feeding to bulk feeding using baleen. Concomitant withthis, aetiocetids exhibit an early stage in developmentof an expanded oral cavity, another evolutionarynovelty associated with bulk feeding.

Phylogenetic analysis confirms the monophyly ofAetiocetidae and recognizes Morawanocetus as theearliest diverging member of this clade. Species ofAetiocetus and Chonecetus share a more recentcommon ancestry. The speciose genus Aetiocetusappears to be monophyletic and its members docu-ment a limited range of morphological diversityincluding a trend toward polydonty.

Aetiocetids emerge as the most morphologically andtaxonomically diverse clade of toothed mysticetes andare phylogenetically positioned between the southernhemisphere stem toothed mysticetes (Janjucetusand Mammalodon) and the crownward edentulousmysticetes.

ACKNOWLEDGEMENTS

We thank L. G. Barnes (LACM), Pat Holyroyd (UCMP)and H. Sawamura (AMP) for allowing us to studyspecimens under their care. The late Robert L. Clarkskillfully prepared the holotype skull and dentaires ofAetiocetus weltoni. Glenn Daleo, formerly of San DiegoChildren’s Hospital, provided CT imaging of UCMP122900. Illustrations were produced by BarbaraMarrs, Tim Gunther, Kesler Randall and Ian Browne.Research was supported in part by NSF grant DEB-0212238 to T.A.D. and grant DEB 0212248 to A.B.

REFERENCES

Addicott W. 1976. Neogene molluscan stages of Oregon andWashington. In: Fritsche AE, Best HT Jr, Wornardt WW,eds. The Neogene Symposium. San Francisco, CA: Society ofEconomic Paleontologists and Mineralogists, 95–116.

Armentrout JM. 1981. Correlation and ages of Cenozoicstratigraphic units in Oregon and Washington. GeologicalSociety of America Special Paper 184: 137–148.

Barnes LG, Kimura M, Furusawa H, Sawamura H. 1995.Classification and distribution of Oligocene Aetiocetidae(Mammalia; Cetacea; Mysticeti) from western NorthAmerica and Japan. Island Arc 3: 392–431.

Berta A, McGowen M, Gatesy J, Demere T. 2006. Mys-ticete phylogeny: the role of stem taxa and character evo-lution in the transition to modern mysticetes. Journal ofVertebrate Paleontology 26: (Suppl. 3): 42A.

Bouetel V, de Muizon C. 2006. The anatomy and rela-tionships of Piscobalaena nana (Cetacea, Mysticeti), aCetotheriidae s.s. from the early Pliocene of Peru. Geodiver-sitas 28: 319–395.

Deméré TA. 1986. The fossil whale, Balaenoptera davidsonii(Cope 1872), with a review of other Neogene species ofBalaenoptera (Cetacea: Mysticeti). Marine Mammal Science2: 227–298.



Deméré TA, Berta A, McGowen MR. 2005. The taxonomicand evolutionary history of fossil and modern balaenopteroidmysticetes. Journal of Mammalian Evolution 12: 99–143.

Deméré TA, McGowen MR, Berta A, Gatesy JG. 2008.Morphological and molecular evidence for a stepwise evolu-tionary transition from teeth to baleen in mysticete whales.Systematic Biology 57: 15–37.

Emlong DR. 1966. A new archaic cetacean from the Oli-gocene of northwest Oregon. Bulletin of the Museum ofNatural History, University of Oregon 3: 1–51.

Fitzgerald EMG. 2006. A bizarre new toothed mysticete(Cetacea) from Australia and the early evolution of baleenwhales. Proceedings of the Royal Society B: Biological Sci-ences 273: 2955–2963.

Fordyce RE. 1981. Systematics of the odontocete whaleAgorophius pygmaeus and the Family Agorophiidae (Mam-malia: Cetacea). Journal of Paleontology 55: 1028–1045.

Fordyce RE. 1994. Waipatia maerewhenua, new genus andnew species (Waipatiidae, new family), an archaic late Oli-gocene dolphin (Cetacea: Odontoceti: Plantanistoidea) fromNew Zealand. Proceedings of the San Diego Society ofNatural History 29: 147–176.

Fordyce RE. 2002. Simocetus rayi (Odontoceti: Simocetidae,New Family): a bizarre new archaic Oligocene dolphin fromthe eastern North Pacific. Smithsonian Contributions toPaleobiology 93: 185–222.

Fordyce RE. 2003. Early crown-group Cetacea in theSouthern Ocean: the toothed archaic mysticete Llanocetus.Journal of Vertebrate Paleontology 23 (Suppl. 3): 50A.

Fordyce RE, Muizon C de. 2001. Evolutionary history ofcetaceans: a review. In: Mazin J-M, de Buffrenil V, eds.Secondary adaptations of tetrapods to life in water.Munchen: Verlag Dr. Friedrich Pfeil, 169–233.

Geisler J, Sanders AE. 2003. Morphological evidence for thephylogeny of the Cetacea. Journal of Mammalian Evolution10: 23–129.

Harrison and Eaton (firm). 1920. Report on investigation ofoil and gas possibilities of western Oregon. Oregon Bureauof Mines and Geology, Mineral Resources. Oregon 3: 3–37.

Ichishima H. 2005. Notes on the phyletic relationships of theAetiocetidae and the feeding ecology of toothed mysticetes.Bulletin of Ashoro Museum of Paleontology 3: 111–117.

Ishikawa H, Amasaki H. 1995. Development and physiologi-cal degradation of tooth buds and development of rudimentof baleen plate in southern minke whale, Balaenopteraacutorostrata. Journal of Veterinary Medical Science 57:665–670.

Ishikawa H, Amasaki H, Dohguchi H, Furuya A, SuzukiK. 1999. Immunohistological distributions of fibronectin,tenascin, type I, III and IV collagens, and laminin duringtooth development and degeneration in fetuses of minkewhale, Balaenoptera acutorostrata. Journal of VeterinaryMedical Science 61: 227–232.

Karlsen K. 1962. Development of tooth germs and adjacentstructures in the whalebone whale (Balaenoptera physalus(L.)). Hvalrådets Skrifter 45: 5–56.

Kellogg R. 1923. Descriptions of two squalodonts recentlydiscovered in the Calvert Cliffs, Maryland; and notes on the

sharktoothed cetaceans. Proceedings of the United StatesNational Museum 62: 1–69.

Kellogg R. 1928. The history of whales – their adaptation tolife in the water (concluded). Quarterly Review of Biology 3:174–208.

Kellogg R. 1936. A review of the Archaeoceti. Carnegie Insti-tute of Washington Special Publication 482: 1–366.

Kellogg R. 1969. Cetothere skeletons from the Miocene Chop-tank Formation of Maryland and Virginia. United StatesNational Museum Bulletin 294: 1–40.

Kimura M, Barnes LG, Furusawa H. 1992. Classificationand distribution of Oligocene Aetiocetidae (Mammalia, Mys-ticeti) from western North America and Japan. Abstracts,29th International Geological Congress, Kyoto, Japan 2: 348.

Kleinpell RM. 1938. Miocene stratigraphy of California.Tulsa, OK: American Association of Petroleum Geologists.

Lambertsen R, Ulrich N, Straley J. 1995. Frontomandibu-lar stay of Balaenopteridae: a mechanism for momentum re-capture during feeding. Journal of Mammalogy 76: 877–899.

Lambertsen RH. 1983. Internal mechanism of rorqualfeeding. Journal of Mammalogy 64: 76–88.

Luo Z, Gingerich PD. 1999. Terrestrial Mesonychia toaquatic Cetacea: transformation of the basicranium andevolution of hearing in whales. University of MichiganPapers on Paleontology 31: 1–98.

McLeod SA, Whitmore FC, Barnes LG. 1993. Evolutionaryrelationships and classification. In: Burns JJ, Montague JJ,Cowles CJ, eds. The bowhead whale. Lawrence, KS: TheSociety for Marine Mammalogy, 45–70.

Messenger SL, McGuire JA. 1998. Morphology, molecules,and the phylogenetics of cetaceans. Systematic Biology 47:90–124.

Miller GS. 1923. The telescoping of the cetacean skull.Smithsonian Miscellaneous Collections 75: 1–71.

Mitchell ED. 1989. A new cetacean from the late EoceneMeseta Formation, Seymour island, Antarctic Peninsula.Canadian Journal of Fisheries and Aquatic Sciences 46:2219–2235.

Muizon C de. 1984. Les Vertebres de al Formation Pisco(Perou). Deuxieme partie: Les Odontocetes (Cetacea, Mam-malia) du Pliocene inferieur de Sud-Sacaco. Travaux deInstitut Français d’Études Andines 27: 1–188.

Muizon C de. 1991. A new Ziphiidae (Cetacea) from the earlyMiocene of Washington State (USA) and phylogenetic analy-sis of the major groups of odontocetes. Bulletin du MuséumNational d’Historie Naturelle, Section C 4ème série 12: 279–326.

Oelschlager HA. 1986. Comparative morphology and evolu-tion of the otic region in toothed whales (Cetacea, Mamma-lia). American Journal of Anatomy 177: 353–368.

Pivorunas A. 1977. The fibrocartilage skeleton and relatedstructures of the ventral pouch of balaenopterid whales.Journal of Morphology 151: 299–314.

Pivorunas A. 1979. The feeding mechanisms of baleenwhales. American Scientist 67: 432–440.

Pledge NS. 2005. A new species of early Oligocene cetaceanfrom Port Willunga, South Australia. Memoirs of the Queen-sland Museum 51: 123–133.



Prothero DR, Bitboul CZ, Moore GW, Niem AR. 2001.Magnetic stratigraphy and tectonic rotation of the OligoceneAlsea, Yaquina, and Nye formations, Lincoln County,Oregon. In: Prothero DR, ed. Magnetic stratigraphy of thePacific coast Cenozoic. Pacific Section SEPM (Society forSedimentary Geology) Book, vol. 91, 184–194.

Rau WW. 1981. Pacific Northwest Tertiary benthic foramin-iferal biostratigraphic framework – An overview. GeologicalSociety of America Special Paper 184: 67–84.

Ridewood WG. 1923. Observations on the skull in foetalspecimens of whales of the genera Megaptera and Bal-aenoptera. Philosophical Transactions of the Royal Society ofLondon, Series B 211: 209–272.

Russell LS. 1968. A new cetacean from the Oligocene SookeFormation of Vancouver Island, British Columbia. Cana-dian Journal of Earth Sciences 5: 929–933.

Sanders AE, Barnes LG. 2002. Paleontology of the lateOligocene Ashley and Chandler Bridge Formations of SouthCarolina, 3: Eomysticetidae, a new family of primitive mys-ticetes. Smithsonian Contributions to Paleobiology 93: 313–356.

Sawamura H, Ichishima H, Ito H, Ishikawa H. 2006.Features implying the beginning of baleen growth in aetio-cetids. Journal of Vertebrate Paleontology 26 (Suppl. 3): 120A

Schulte HW. 1916. The sei whale (Balaenoptera borealisLesson). Anatomy of the foetus of Balaenoptera borealis.Monographs of the Pacific Cetacea. Memoirs of the AmericanMuseum of Natural History 1: 389–502.

Snaveley PD Jr, McLeod NS, Wagner HC, Rau WW. 1976.Geologic map of the Yaquina and Toledo quadrangles,Lincoln County, Oregon. U.S.G.S. Miscellaneous Investiga-tions Series Map, I-867.

Swofford DL. 2001. PAUP*’ Phylogenetic analysis using par-simony (* and other methods). Version 4.0b8. Sunderland,MA: Sinauer Associates.

Uhen MD. 1999. New species of protocetid archaeocete whale,Eocetus wardii (Mammalia, Cetacea), from the middleEocene of North Carolina. Journal of Paleontology 73: 512–528.

Uhen MD. 2004. Form, function, and anatomy of Dorudonatrox (Mammalia: Cetacea): an archaeocete from the middleto late Eocene of Egypt. University of Michigan Papers onPaleontology 34: 1–222.

Van Valen L. 1968. Monophyly or diphyly in the origin ofwhales. Evolution 22: 37–41.

Walmsley R. 1938. Some observations on the vascular systemof a female fetal finback. Carnegie Institution of Washing-ton, Monograph Series 496. Contributions to Embryology 27:109–178.

APPENDIX 1CHARACTERS AND CHARACTER STATES FOR

AETIOCETIDS AND RELATED TAXA

1. Rostrum, transverse width at midpoint relative tocondylobasal length (Uhen, 1999). 0 = narrow(15–22%), 1 = very narrow (5–12%), 2 = broad(24–31%), 3 = very broad (> 31%).

The primitive condition of a narrow rostrum occursin archaeocetes (e.g. Dorudon and Zygorhiza), thestem odontocete Agorophius and the stem edentulousmysticete Eomysticetus. A very narrow rostrum rep-resents derived condition 1 and is seen in some odon-tocetes (e.g. Tasmacetus). Mammalodon, Janjucetusand all aetiocetids are characterized by having abroad rostrum (derived condition 2). Certain bal-aenopterids are coded as having very broad rostra(derived condition 3). Crown balaenids show a rever-sal to a very narrow rostrum.

2. Frontal, supraorbital process (modified fromMiller, 1923). 0 = broad anteroposteriorly andshort transversely, 1 = moderately broad antero-posteriorly and moderately elongate transversely,2 = very narrow anteroposteriorly and very elon-gate transversely.

Archaeocete cetaceans such as Dorudon andZygorhiza, as well as aetiocetids, possess an antero-posteriorly broad, but transversely short supraorbitalprocess of the frontal; the primitive condition. Inderived condition 1 the anteroposterior diameter of thefrontals is reduced, while the transverse diameter isincreased. This is the condition seen in Eomysticetusand balaenopteroids. Pliocene through Holocene balae-nids possess derived condition 2, a frontal that is nar-row anteriorposteriorly, but elongated transversely.

3. Frontal, posteromedial corner. 0 = well exposed,1 = short overlap by maxilla, 2 = broadly over-lapped by maxilla.

The primitive condition of a broadly exposed frontaloccurs in Dorudon and Zygorhiza. Derived condition 1,posteromedial corner of frontal overlapped by maxilla,characterizes Janjucetus, Mammalodon, aetiocetidsand edentulous mysticetes. Derived condition 2,broad overlap of frontal by maxilla, distinguishesodontocetes.

4. Parietal–frontal suture, anterior extension.0 = none, 1 = lobate, 2 = triangular.

The primitive condition of a nearly transverseparietal–frontal (coronal) suture occurs in Dorudon,Zygorhiza and all odontocetes. Two derived conditionsare identified. A lobate anterior extension of the pari-etal frontal contact occurs in species of Aetiocetus(derived condition 1). A triangular anterior extensioncharacterizes Chonecetus (derived condition 2). Thecondition in Morawanocetus and Mammalodon is cur-rently unknown.

5. Nasals, width relative to length (Deméré et al.,2005). 0 = slender (15–25%), 1 = broad (26–45%),2 = very broad (46–70%), 3 = extremely broad(> 71%).

In Zygorhiza the width of the nasals represents only18% of the length. In Mammalodon, Janjucetus and



aetiocetids (except A. polydentatus) the nasals arebroader and nasal width relative to length is as fol-lows: Aetiocetus cotylalveus (31%), A. weltoni (31%),Chonecetus goedertorum (30%). As noted by Barneset al. (1995), Aetiocetus polydentatus is the only aetio-cetid with very broad nasals (47%), identified asderived state 2. The basal odontocete Agorophiusexhibits extremely broad nasals (derived state 3)approximately 50% as estimated from the line draw-ings in Fordyce (1981).

6. Frontal, orbital rim (Barnes et al., 1995).0 = straight, 1 = slightly concave, 2 = deeplyconcave.

The primitive condition of a straight orbital rim (indorsal aspect) occurs in Zygorhiza, Dorudon and odon-tocetes. An intermediate condition, derived condition1, in which the orbital rim is slightly concave, occurs inMammalodon, Janjucetus, Aetiocetus tomitai, A. poly-dentatus, Morawanocetus, Chonocetus, Eomysticetusand some other edentulous mysticetes. Derived condi-tion 2, a deeply concave orbital rim, is a synapomorphyfor A. cotylalveus and A. weltoni. Some crown mystice-tes (e.g. Balaenoptera acutorostrata) show a reversal tothe primitive condition.

7. Ascending process of maxilla, posterior termina-tion (Deméré et al., 2005). 0 = anterior half ofsupraorbital process of frontal, 1 = anterior borderof supraorbital process of frontal, 2 = posterior halfof supraorbital process of frontal, 3 = Posterior totemporal fossa.

The primitive condition, where the ascendingprocess of the maxilla terminates within the anteriorhalf of the supraorbital process of the frontal, charac-terizes Zygorhiza, Dorudon, Janjucetus, Mammalodonand all aetiocetids (except A. weltoni). In derivedcondition 1 seen in Eomysticetus, the posteriormostedge of the ascending process of the maxilla extendsonly to the anterior border of the supraorbital processof the frontal. In Aetiocetus weltoni, the posteriormostedge of the ascending process of the maxilla terminateswithin the posterior half of the supraorbital process ofthe frontal, derived condition 2. This is also the condi-tion in Agorophius. In Tasmacetus the posteriormostedge of the ascending process of the maxilla extendsposterior to the temporal fossa, derived condition 3.

8. Intertemporal region. 0 = narrow, 1 = broad.In the primitive condition, seen in Dorudon,

Zygorhiza, Aetiocetus polydentatus, A. weltoni, Mam-malodon and Eomysticetus, the intertemporal region isnarrow (< 70% width across occipital condyles). Allother aetiocetids, Janjucetus and later diverging eden-tulous mysticetes show the derived condition of a broadintertemporal region (> 80% width across occipitalcondyles).

9. Sagittal crest. 0 = present, 1 = reduced, 2 = absent.The presence of a well-developed sagittal crest is the

primitive condition seen in Dorudon, Zygorhiza, Jan-jucetus and Eomysticetus. Derived condition 1, reduc-tion of the sagittal crest, occurs in Mammalodon,Aetiocetus weltoni and A. polydentatus. In all otherdescribed aetiocetids, as well as edentulous mysticetesa sagittal crest is absent.

10. Squamosal, zygomatic process thickness. 0 =uniform thickness entire length, 1 = expandednear anterior margin and at the posterior end butnarrow in the middle.

In the primitive condition, seen in outgroup taxa andJanjucetus, Chonecetus and Eomysticetus, the zygo-matic process is of uniform width. The derived condi-tion, in which the zygomatic process is thin in themiddle and expanded anteriorly and posteriorly, rep-resents an aetiocetid synapomorphy. Edentulous mys-ticetes display a reversal to the primitive condition.

11. Squamosal, zygomatic process length. 0 = does notcontact postorbital process of frontal, 1 = contactspostorbital process of frontal.

In the primitive condition, seen in Dorudon,Zygorhiza, Agorophius and Eomysticetus, the zygo-matic process of the squamosal does not contact thepostorbital process of the frontal. The derived condi-tion, in which the zygomatic process contacts thepostorbital process of the frontal, characterizes Tas-macetus, Janjucetus, species of Aetiocetus, Morawano-cetus and modern mysticetes; this distributionindicates independent derivation among odontocetesand aetiocetids.

12. Lacrimal, external exposure. 0 = exposed laterally,1 = exposed dorsally.

In the primitive condition, seen in archaeocetes,odontocetes and Janjucetus the lacrimal is wellexposed laterally between the frontal and maxilla. Incrown mysticete taxa the lacrimal retains a lateralexposure, but is tightly sandwiched between thefrontal and antorbital process of the maxilla. Aetio-cetids display the derived condition in which thelacrimal is a broadly oval element exposed dorsally.

13. Palatal window, exposing vomer (Deméré et al.,2008). 0 = absent, 1 = present, delimited entirelyby maxilla, 2 = present, delimited by maxilla andpremaxilla.

In the primitive condition, seen in Zygorhiza andDorudon, the vomer is hidden from view on the palateby the maxilla. In Janjucetus the vomer is exposed ina narrow palatal window formed by the maxilla(derived condition 1). This appears also to be thecondition in Tasmacetus. In all aetiocetids with pre-served palates the vomer is exposed in a narrowpalatal window formed by bifurcation of the maxilla



and premaxilla (derived condition 2), a synapomorphyfor the group. The absence of a palatal window expos-ing the vomer in crown mysticetes appears to repre-sent retention of the primitive condition.

14. Palatine, posterior margin with notch. 0 = absent1 = present.

In the primitive condition, seen in Dorudon,Zygorhiza, Agorophius, Tasmacetus, Janjucetus andcrown mysticetes, the posterior margin of the palatinehas a smooth margin. The derived condition, in whichthe posterior margin of the palatine has a rectangularnotch, occurs in aetiocetids and represents a synapo-morphy for the group.

15. Palatines, posterior termination. (Deméré et al.,2005). 0 = extend to internal nares, 1 = extend toslightly overlap the pterygoids, 2 = long overlap ofpterygoids nearly reaching pterygoid fossa.

In the primitive condition, seen in Dorudon,Zygorhiza, Janjucetus, aetiocetids and odontocetes,the palatines are long and narrow posteriorly. Thederived condition, in which the palatines have a shortand broad extension that overlaps the pterygoid,occurs in modern mysticetes.

16. Lateral palatine foramina (modified from Geisler& Sanders, 2003). 0 = absent, 1 = present.

In the primitive condition, seen in Dorudon,Zygorhiza, odontocetes and Janjucetus, the lateralportion of the maxilla on the palate lacks nutrientforamina. The derived condition, in which foraminaand associated short sulci are positioned along thelateral portion of the maxilla, occurs in all edentulousmysticetes. Although more diminutive than in crownmysticetes, the tiny foramina and sulci found along-side and medial to the alveoli (see text) in aetiocetidsare considered homologous.

17. Squamosal fossa (Barnes et al., 1995). 0 = single,continuous fossa, 1 = second fossa.

In the primitive condition, seen in outgroup taxa,Mammalodon and Morawanocetus, a single, shallow,continuous squamosal fossa is developed. The derivedcondition, in which a second smaller fossa is developedposterior to the squamosal fossa creating a ‘stepped’profile, occurs in Janjucetus, Chonecetus and species ofAetiocetus.

18. Basicranial width. 0 = narrow, 1 = wide.The relative width of the anterior and posterior

borders of the basioccipital varies among mysticetes.In species of Aetiocetus and Chonocetus the two mea-surements are subequal (primitive condition). In thederived condition, seen in Morawanocetus and Janju-cetus, the posterior measurement is distinctly larger(> 1.5 times) than the anterior measurement. Thederived condition is also seen in odontocetes, a distri-

bution suggesting independent derivation amongodontocetes and aetiocetids.

19. Posterior sinus. 0 = poorly developed or absent,1 = present.

In the primitive condition, seen in Dorudon,Zygorhiza, Eomysticetus and modern mysticetes, theposterior sinus is poorly developed or absent. In aetio-cetids, the posterior sinus is well formed. The derivedcondition is also seen in certain odontocetes (e.g.Tasmacetus), a distribution suggesting independentderivation among odontocetes and aetiocetids.

20. Posterior teeth, root condition. 0 = two rooted withfused roots, 1 = double rooted, 2 = single rooted.

In the primitive condition, seen in Dorudon andZygorhiza, P4 and M1 are two-rooted with the pos-terior root developed from fusion of two separateroots. In Agorophius, Janjucetus and Morawanocetusthe posterior teeth are double rooted, derived condi-tion 1. In Mammalodon the posterior upper teethappear to have single fused roots. However, in thelower dentition the posterior teeth are clearly doublerooted. Because of this, Mammalodon is here consid-ered to possess condition 1. Species of Aetiocetus andChonecetus (based on alveoli only) are characterizedby upper and lower single rooted teeth, derived con-dition 2.

21. Number of teeth (Deméré et al., 2008). 0 = 10/11,1 = 11/12, 2 = 13-14/14-14.

In the primitive condition, seen in Dorudon,Zygorhiza, Agorophius and Janjucetus, the dentalformula is 10/11. In Mammalodon and aetiocetids(except for A. polydentatus) the dental formula is11/12, derived condition 1. Aetiocetus polydentatusdisplays derived condition 2, a high tooth count (poly-donty), 13-14/14-14. The high tooth count of Tasmace-tus is also scored as derived condition 2 and suggestsindependent evolution of polydonty in both odonto-cetes and mysticetes.

22. Postcanine tooth enamel (Deméré et al., 2008).0 = delicate vertical enamel ridges on lingualsurface only, 1 = very heavy vertical enamelridges on lingual and labial surfaces, 2 = enamelridges poorly developed or absent.

Zygorhiza has stronger enamel ridges on thelingual surfaces of P2-4 than on the labial sides ofthese teeth, while M1-2 have essentially smoothenamel surfaces on both the lingual and the labialsides (Kellogg, 1936). Species of Aetiocetus have deli-cate vertical enamel ridges on the lingual sides of thecheektooth crown and smooth enamel on the labialsides (primitive condition). Janjucetus, Mammalodonand Morawanocetus are characterized by derived con-dition 1, heavy enamel ridges on both the labial andthe lingual sides of the cheek tooth crowns. Derived



condition 2, in which striated enamel is poorly devel-oped or absent, characterizes Tasmacetus.

23. Posterior upper postcanine teeth (Deméré et al.,2008). 0 = five or more, large and well-developedanterior and posterior denticles, 1 = three orfewer, small and simple anterior and posteriordenticles.

Dorudon, Zygorhiza, Agorophius, Janjucetus andMorawanocetus show an increase in number andcomplexity of accessory denticles on the posteriorpostcanine teeth; the primitive condition. Species ofAetiocetus have fewer, simpler denticles on the post-canine teeth; derived condition 1. The condition inMammalodon is equivocal because of heavy apicalcrown wear in the holotype. Because of this Mam-malodon is coded as unknown. Chonecetus is alsoscored as unknown, because of the lack of preserveddentitions for this genus.

24. Upper teeth, diastemata (Deméré et al., 2008).0 = teeth with wide diastemata, 1 = teeth closelyappressed or with nearly equally narrowdiastemata.

The upper dentition of Zygorhiza and Dorudon con-sists of anterior teeth (I1-3, C, P1-2) with relativelywide diastemata and posterior teeth (P2-4, M1-2)more closely appressed. In species of Aetiocetus, theanterior teeth in the upper dentition also have widediastemata, while the more posterior teeth havenarrow diastemata. The derived condition, in whichthe alveoli of the upper dentition are of nearlyequally narrow width or are closely appressed, occursin Chonecetus and Janjucetus, and Mammalodon,respectively. The condition in Morawanocetus isunknown.

25. Dentary, outline in dorsal aspect (modified fromMiller, 1923; Deméré, 1986). 0 = laterallyconcave, 1 = straight, 2 = laterally convex.

In the primitive condition, seen in archaeocetes,odontocetes and possibly Janjucetus (see Fitzgerald,2006), the dentary is laterally concave in dorsalaspect. Two derived conditions are identified in mys-ticetes. Derived condition 1, a relatively straightdentary in dorsal aspect (e.g. the straight length ofthe dentary does not exceed 97% of the curvedlength), occurs in aetiocetids and Eomysticetus, aswell as Eschrichtius. In other crown mysticetes (e.g.balaenopterids and balaenids) the dentary is later-ally convex; derived condition 2 (Deméré et al., 2005).

26. Mandibular symphysis (Geisler & Sanders, 2003;Fitzgerald, 2006). 0 = sutured, but unfused,1 = unsutured, with shallow elongate fossa,2 = unsutured, with long groove.

Archaeocetes and odontocetes have a symphysisthat is sutured, but typically unfused; the primitive

condition. All mysticetes have a symphysis that isunfused and unsutured. Instead, the symphysis isheld together by soft tissue (fibrocartilage; Pivorunas,1977). Two derived conditions are identified in mys-ticetes. Derived condition 1, symphysis with ashallow, elongate fossa, is reported in Janjucetus andMammalodon (see Fitzgerald, 2006). Derived condi-tion 2, symphysis with a long groove and overlyingnarrow shelf, occurs in aetiocetids and all laterdiverging mysticetes.

27. Dentary, coronoid process. 0 = broad, with poste-rior margin nearly vertical and slightly over-hanging, 1 = well developed, with concaveposterior margin, 2 = reduced, 3 = finger-like orcrest-like

In the primitive condition, seen in Dorudon,Zygorhiza, Janjucetus and Mammalodon, the coro-noid process has a broad base and its posteriormargin is generally straight. In aetiocetids the coro-noid process is also well developed, but its posteriormargin is concave; derived condition 1. A reducedcoronoid process characterizes odontocetes andmodern mysticetes, which possess coronoid processesthat are either finger-like or crest-like; derived con-dition 2.

28. Nasal, posterior margin (modified from Geisler &Sanders, 2003). 0 = located at or just in front ofanterior edge of supraorbital process of frontal,1 = located in line with anterior half of supraor-bital process of frontal, 2 = located in line withposterior half of supraorbital process of frontal,3 = located posterior to temporal fossa.

Cranial telescoping plays a prominent role in ceta-cean evolution and involves the relocation of theexternal nares from an anterior position at the frontof the rostrum to a progressively more posterior posi-tion above or behind the orbits. One measure of thedegree of telescoping is the location of the nasal-frontal suture relative to the supraorbital process ofthe frontal. In the primitive condition the posteriorborder of the nasal lies at or just anterior to thesupraorbital process of the frontal. This conditionoccurs in archaeocetes, Janjucetus, Mammalodon andMorawanocetus. In derived state 1 the posteriormostedge of the nasal extends to a level above the centreof the orbit on the anterior border of the supraorbitalprocess of the frontal. This is the condition in somecrown mysticetes. In derived state 2 the posterior-most edge of the nasal extends to the posterior halfof the supraorbital process of the frontal. This is thecondition in Aetiocetus spp., Chonecetus and Eomys-ticetus, as well as in the basal odontocete Agorophius.In derived state 3, seen in Tasmacetus, the posteri-ormost edge of nasal extends to the posterior half ofthe temporal fossa.



29. Maxilla, descending process (modified fromMcLeod, Whitmore & Barnes, 1993). 0 = present,with teeth, 1 = present, as edentulous infraor-bital plate, 2 = absent.

The posterior portion of the maxilla is divided incetaceans into ascending and descending processes.In mysticetes the descending process becomes trans-versely expanded to form the infraorbital plate.Beginning with Miller (1923) the presence of aninfraorbital plate of the maxilla has been considereda diagnostic feature of mysticetes. In archaeocetesthe descending process houses the M1-3 toothrowand is not developed as an edentulous infraorbitalplate (primitive condition). In aetiocetids and othertoothed mysticetes, as well as all edentulous mystice-tes, the descending process is expanded into aninfraorbital plate (derived condition 1). Interestingly,even in the toothed aetiocetids the infraorbital plateis edentulous. In odontocetes there is no descendingprocess of the maxilla (derived condition 2).

30. Supraoccipital shield, orientation. 0 = verticallyorientated, 1 = anteriorly sloped dorsal portion.

In the primitive condition, seen in archaeocetes, Jan-jucetus and certain odontocetes (e.g. Tasmacetus), thesupraoccipital shield is orientated vertically to over-hanging. The derived condition, an anterodorsallysloped supraoccipital shield, occurs in all othertoothed and edentulous mysticetes. The presence ofan anterodorsally sloped supraoccipital shield in theoutgroup taxon Agorophius suggests independentevolution of this feature.

31. Frontal, supraorbital process. 0 = at level of ver-tex, 1 = gradually sloping or abruptly depressedbelow vertex.

In the primitive condition, seen in archaeocetes,Janjucetus, Mammalodon, aetiocetids, Eomysticetusand odontocetes, the supraorbital process of thefrontal is positioned at the level of the cranial vertex.In the derived condition, seen in later diverging mys-ticetes, the supraorbital process lies below the vertex.Variation of the derived state includes a graduallysloping supraorbital process (e.g. most ‘cetotheres’and crown balaenids) and an abruptly depressedsupraorbital process (e.g. balaenopteroids).

32. Teeth. 0 = anterior and postcanine teeth stronglyheterodont, 1 = anterior and postcanine teethmoderately heterodont, 2 = anterior and postca-nine teeth weakly heterodont, 3 = teeth homo-dont.

In the primitive condition, seen in archaeocetesand Agorophius, the postcanine teeth are stronglyheterodont (i.e. relatively large denticles). In Mam-malodon, Janjucetus and Morawanocetus postcanineteeth are moderately heterodont (i.e. distinct, but

small denticles); derived condition 1. In species ofAetiocetus the postcanine teeth have very small, deli-cate denticles; derived condition 2. In Tasmacetus thedentition is distinctly homodont; derived condition 3.

33. Squamosal, subtemporal crest (Fordyce, 2002).0 = absent, 1 = present.

In the primitive condition, seen in Dorudon, Tas-macetus, Mammalodon, Janjucetus, Eomysticetusand modern mysticetes, the posterior border of thetemporal fossa is evenly concave and there is nosubtemporal crest. The derived condition, a distinctsubtemporal crest extending into the temporal fossa,occurs in Zygorhiza, stem odontocetes (e.g. Agoroph-ius) and aetiocetids.

34. Lambdoidal crests. 0 = flush with temporal wall,1 = overhanging temporal wall.

In the primitive condition, seen in archaeocetes,Tasmacetus and aetiocetids (except A. polydentatus),the lambdoidal crests are low and subequal to thetemporal wall. The derived condition, lambdoidalcrests that overhang the temporal wall, occurs inAgorophius, Mammalodon, Janjucetus, Aetiocetuspolydentatus and Eomysticetus. The condition incrown mysticetes is polymorphic.

35. External narial fossa, position (Deméré et al.,2005). 0 = located between midpoint of rostrumand antorbital notch, 1 = located inline with orjust posterior to antorbital notch, 2 = located wellposterior to antorbital notch.

The position of the external narial fossa on therostrum tracks the posterior migration of the blow-hole in cetacean evolution. In the primitive condition,seen in Zygorhiza, Dorudon, Mammalodon, aetio-cetids and Eomysticetus, the narial fossa is anteriorlylocated between the midpoint of the rostrum and theantorbital notch. Two derived conditions are recog-nized. Derived condition 1, narial fossa positionedinline with or just posterior to the antorbital notch,occurs in Agorophius. Derived condition 2, narialfossa positioned well posterior to the antorbitalnotch, occurs in Tasmacetus and other extant odon-tocetes. Crown mysticetes are polymorphic for thischaracter and display both the primitive condition(e.g. balaenids and neobalaenids) and derived condi-tion 1 (e.g. balaenopteroids).

36. Jugal, contact with squamosal. 0 = broad overlap,1 = short overlap, 2 = no overlap, ligamentous.

Primitively, the contact between the jugal andzygomatic process of the squamosal is broadly over-lapping. The primitive condition occurs in archaeo-cetes (e.g. Dorudon and Zygorhiza). In aetiocetids thezygomatic process still overlaps the jugal, but thedegree of overlap is very reduced (derived state 1). Inderived state 2 the jugal is not overlapped by the



zygomatic process. This is the condition in Tasmace-tus, Eomysticetus and modern mysticetes.

37. Ascending process of maxilla, position of poste-rior end (Bouetel & de Muizon, 2006). 0 = locatedanterior to posterior margin of nasal; 1 = locatedapproximately inline with or slightly posterior toposterior margin of nasal.

In the primitive condition, seen in Dorudon, Jan-jucetus, Aetiocetus tomitai and Chonecetus, the pos-terior termination of the ascending process of themaxilla lies anterior to the posterior termination ofthe nasal. The derived condition, ascending process ofmaxilla terminating inline with or behind the nasal,occurs in the majority of aetiocetids and crown mys-ticetes (except balaenids).

38. Premaxilla, position of posterior end (Bouetel &de Muizon, 2006). 0 = located anterior to poste-rior extremity of maxilla; 1 = located inline withposterior extremity of maxilla; 2 = located poste-rior to posterior extremity of maxilla.

The relative position of the posterior termini of thepremaxilla and maxilla varies among cetaceans. Inthe primitive condition, seen in Zygorhiza, Dorudon,Janjucetus and Mammalodon, the premaxilla termi-nates anterior to the maxilla. Two derived conditionsare recognized. In derived condition 1, the premaxillaand maxilla terminate at approximately the samelevel. This is the condition in Morawanocetus, Aetio-cetus weltoni, A. cotylalveus and crown mysticetes. Inderived state 2, the premaxilla terminates behind theposterior extremity of the maxilla. This is the condi-tion in Eomysticetus, Aetiocetus tomitai, A. polyden-tatus and Chonecetus.

39. Maxilla, antorbital process (modified fromGeisler & Sanders, 2003). 0 = no groove on ante-rolateral face; 1 = oblique groove on anterolateralface present, but low; 2 = oblique groove on ante-rolateral face present and well developed.

The anterolateral face of the antorbital process ofthe maxilla merges smoothly with the rostral portionof the maxilla in archaeocetes and odontocetes(primitive condition). In mysticetes the two regionsare distinctly separated and the anterolateral face ofthe antorbital process is more steeply orientated thanthe rostral portion. In mysticetes this area is crossedobliquely (posteroventrolateral to anterodorsomedial)by a distinct groove for the facial nerve. Developmentof this groove is variable, being weakly developed inChonecetus and modern mysticetes (derived condi-tion1) and strongly developed in Janjucetus, Mam-malodon and species of Aetiocetus (derived condition2).

40. Rostrum, width at antorbital notch (modifiedfrom Geisler & Sanders, 2003). 0 = 130–165% of

occipital condyle width; 1 = > 170% of occipitalcondyle width.

In the primitive condition, seen in outgroup taxa(except Agorophius) and Chonecetus, the rostrum asmeasured at the antorbital notch is narrow. In thederived condition, seen in species of Aetiocetus andsome crown mysticetes, the rostrum is expanded, afeature correlated with enlargement of the oralcavity. The narrow rostra of crown balaenids repre-sents a reversal to the primitive condition.

41. Palate, shape (Messenger & McGuire, 1998).0 = flat with no median keel, 1 = median keeldividing palate into right and left concave sur-faces.

In the primitive condition, seen in outgroup taxaand toothed mysticetes, the palate is relatively flat.The derived condition, in which there is a well-developed median keel dividing the palate into rightand left halves, occurs in edentulous mysticetes.

42. Mineralized teeth in adults (Geisler & Sanders,2003). 0 = present, 1 = absent.

The presence of teeth is the primitive conditionseen among diverse archaic toothed taxa includingarchaeocetes, stem mysticetes and stem odontocetes.The derived condition, in which adult mineralizedteeth are absent, is a synapomorphy for all edentu-lous mysticetes. In several species of odontocetes (i.e.delphinapterines and ziphiids) a mineralized denti-tion has been secondarily lost in adult female indi-viduals.

43. Frontal, temporal crest (Geisler & Sanders,2003). 0 = does not extend far onto dorsal surfaceof supraorbital process of frontal, 1 = extends faronto dorsal surface of supraorbital process offrontal.

In the primitive condition, seen in outgroup taxaand toothed mysticetes, the temporal crest does notextend onto the dorsal surface of the supraorbitalprocess of the frontal. The derived condition, inwhich the temporal crest extends far onto thesupraorbital process of the frontal, characterizes alledentulous mysticetes and is correlated with theanterior expansion of the temporalis musculature.

44. Premaxilla–maxilla suture (Geisler & Sanders,2003). 0 = fused dorsally along rostrum,1 = unfused.

Rostral kinesis (i.e. loose articulations between pre-maxilla, maxilla and vomer) appears to be an impor-tant feature in bulk feeding. In the primitive condition,seen in outgroup taxa and toothed mysticetes, thepremaxilla–maxilla suture is fused along the dorsaland ventral portions of the rostrum. The derivedcondition, in which the premaxilla–maxilla suture isunfused, occurs in edentulous mysticetes.



45. Maxilla, lateral margin (modified from McLeodet al., 1993); 0 = thick, 1 = thin.

In the primitive condition, seen in outgroup taxaand the toothed mysticete Janjucetus, the lateralmargin of the maxilla is thick. The derived condi-tion, in which the lateral margin of the maxilla isthin, occurs in Mammalodon, aetiocetids and eden-tulous mysticetes.

46. Premaxilla, ascending process (modified fromMiller, 1923). 0 = no contact with frontals, 1 =

contacting the frontals, 2 = contacting frontalsand forming robust ascending process.

In the primitive condition, seen in archaeocetes andstem odontocetes, the ascending process of the premax-illa does not contact the frontals. Derived condition 1,in which the premaxilla weakly contacts the frontals,occurs in Janjucetus and Morawanocetus. Derivedcondition 2, in which the premaxilla contacts thefrontals and forms a robust ascending process, occursin aetiocetids (except Morawanocetus). Crown mystice-tes are polymorphic for this character.



Documents

Skull anatomy of the Oligocene toothed mysticete Aetioceus weltoni