J Mammal Evol DOI 10.1007/s10914-014-9254-9 ORIGINAL PAPER Comparative Anatomy of the Petrosal Bone of Dichobunoids, Early Members of Artiodactylamorpha (Mammalia) M. J. Orliac & M. A. O’Leary # Springer Science+Business Media New York 2014 Abstract Among Artiodactylamorpha, dichobunoids are some of the oldest fossil species that have been associated with Artiodactyla, the crown clade that includes hippopotamids, camelids, suoids, ruminants, and cetaceans. These important fossil species are known from early Eocene rocks of North America, Europe, and Asia, but their phylogenetic position has yet to be well resolved. Before generating such a phylogeny, it is first critical to document all of the anatomy of known dichobunoid fossils. Here we use CT scans to describe previously undescribed anatomy of the petrosal bone, a complex part of the mammalian skull that contains many variable and phylogenetically informative features. Results show that these extinct species share a number of features that are not documented in modern species including a lateral process of the epitympanic wing constituting the medial border of the piriform fenestra, and a tegmen tympani foramen that may have given passage to the ramus superior of the stapedial artery. Future comprehensive phylogenetic studies may show that many of these characters are plesiomophic for Artiodactylamopha. Some species (Diacodexis, Homacodon and ?Helohyus) exhibit a dorsolateral exposure of the mastoid region of the petrosal on the temporal part of the cranium. This uncommon feature has, to our knowledge, not been reported in another euungulate group. Electronic supplementary material The online version of this article (doi:10.1007/s10914-014-9254-9) contains supplementary material, which is available to authorized users. M. J. Orliac (*) ISE-M, Université Montpellier2, Place Eugène Bataillon, 34095 Montpellier cedex 05, France e-mail: maeva.orliac@univ-montp2.fr M. J. Orliac : M. A. O’Leary Department of Anatomical Sciences, HSC-T-8 (040), Stony Brook University, Stony Brook, NY 11794-8081, USA Keywords Middle ear . Eocene . Euungulata . micro CTscan Introduction Dichobunoidea is a paraphyletic assemblage of Eocene fossil species that has been linked taxonomically to Artiodactyla (McKenna and Bell 1997; Theodor et al. 2007). Spaulding et al. (2009) recently defined Artiodactyla as a crown clade and named Artiodactylamorpha as its associated total clade (fossil species more closely related to Artiodactyla than to any other living species, following de Queiroz 2007). Dichobunoidea is composed of seven families: Diacodexeidae, Homacodontidae, Dichobunidae, Helohyidae, Leptochoeridae, Cebochoeridae, and Raoellidae (Theodor et al. 2007). The relationships of most of these families to modern artiodactyl clades are still being investigated and several interesting phylogenetic questions remain to be tested. These include whether Diacodexis, one of the oldest artiodactyl species documented in the fossil record (Rose 2006; Theodor et al. 2007), is also one of the most basal branches within Artiodactylamorpha (e.g., Geisler et al. 2007; Thewissen et al. 2007, 2009; Geisler and Theodor 2009), or whether it is a more highly nested species (Spaulding et al. 2009; Orliac and Ducrocq 2011; Gatesy et al. 2012), and whether dichobunoid taxa are stem taxa to different extant artiodactyl clades (e.g., Gatesy et al. 2012). Petrosal bones of the ear region have long been viewed as highly relevant to mammalian systematics because these complex bones preserve important anatomical variation for comparison (e.g., Mammalia: MacIntyre 1972; Wible et al. 1995, 2001; Metatheria: Ladevèze 2007; Carnivora: Hunt 1974, 2001; primates and insectivorans: MacPhee 1981; Asher et al. 2002; “condylarths”: Cifelli 1982; and more recently, Artiodactyla: Geisler and Luo 1998; Theodor 2010; O’Leary 20 10). Yet for many foss il species assigned to J Mammal Evol Artiodactylamorpha the petrosal bone has remained undocumented, or only partially described, because few skulls are known and, of those, the petrosal is preserved in situ in the skull such that most views are obscured. For example, the petrosal bone of Diacodexis, and several other dichobunoid species (Homacodon, ?Helohyus, Gobiohyus), has been described only from its ventrolateral surface (Coombs and Coombs 1982; Russell et al. 1983; Geisler and Luo 1998). Here we describe the entire petrosal of representatives of five of seven dichobunoid families: Diacodexis (Diacodexeidae), Homacodon (Homacodontidae), Dichobune (Dichobunidae), ?Helohyus (Helohyidae), Gobiohyus (Helohyidae?), and Acotherulum (Cebochoeridae) using CT scans of in situ petrosals that permit visualization of surfaces that could not be seen previously. Our survey of petrosal morphology from this broad sample of putative basal artiodactylamorphans provides important primary data for future phylogenetic analyses that are even more taxonomically comprehensive than those mentioned above. We propose some hypotheses of character evolution based on the patterns we observe. Abbreviations AMNH – VP MNHN Qu UM2 USNM American Museum of Natural History, Vertebrate Paleontology collection Museum Nationale d’Histoire Naturelle, Paris, Quercy collection Université Montpellier2, France United States Museum of Natural History, Vertebrate Paleontology collection. Material and Methods In situ petrosals of Diacodexis ilicis (Diacodexeidae), Homacodon vagans (Homacodontidae), Dichobune leporina (Dichobunidae), ?Helohyus plicodon (Helohyidae), Gobiohyus orientalis (Helohyidae?), and Acotherulum saturninum (Cebochoeridae) were documented using virtual 3D extraction. The suprageneric systematics follows that of Theodor et al. (2007) and Erfurt and Métais (2007). The familial attribution of Gobiohyus is still problematic (Foss 2007), and this taxon is here attributed to “Helohyidae?.” The specimens were first scanned using the high resolution GE phoenix|vtome|x s240 industrial CT scanner at the AMNH or the Skyscan/1076/in vivo CT scanner at the ISE-M (UM2). The number, dimensions, and thickness of CT slices varied depending on the size of the specimen and which machine was used. The list of specimens and the resolution of their corresponding scans are provided in Appendix 1. We reconstructed two-dimensional slices from X-rays using VGStudioMax® (version 1.2; VolumeGraphics GmbH 2004). We extracted digital endocasts using the segmentation tools of AVIZO 6.3 (Visualization Sciences Group). Paired stereo images of the petrosals are provided in Electronic Supplementary Material (Figs. S1-S10, MorphoBank Project 946). Descriptions Because the petrosal anatomy of these species is largely unknown, we first present five new full petrosal descriptions followed by a comparative discussion. Figure 1 shows the general orientation of the petrosal bone within the skulls of these artiodactylamorphans in medial, anterior, and dorsal views. Note that the bone is tightly wedged into the skull. Figures 2, 3, 4, 5, 6, 7, 8, and 9 consist of comparative illustrations showing the petrosals of the different species in ventrolateral (Figs. 2, 6, and 8), dorsomedial (Figs. 3, 7, and 9), dorsolateral (Fig. 4), and ventromedial (Fig. 5) views. Figure 10 is a comparative plate showing the external exposure of the mastoid region on the skull. Stereo images of the petrosals are provided in Electronic Supplementary Material (Figs. S1-S10, MorphoBank Project 946) as well as illustrations of the ventral views of the basicranium of Homacodon vegans (Fig. S11), ?Helohyus plicodon (Fig. S12), and Gobiohyus orientalis (Fig. S13). The tympanic bulla is absent in all of the specimens described here except Acotherulum saturninum, where it is only partially preserved. Terminology (Appendix 2) follows that of Wible (2003), Giannini et al. (2006), and Luo and Gingerich (1999), which was applied to, and expanded upon, for Artiodactyla and relatives in O’Leary (2010:Table 2). Four new terms are defined here: i) the lateral process of the epitympanic wing, a prong of bone of variable size protruding laterally and anteriorly from the epitympanic wing and constituting the lateral border of the piriform fenestra, it typically terminates in a point anteriorly (Fig. 1c); ii) the tegmen tympani canal, a channel that traverses the tegmen tympani vertically creating a passage between the middle ear and the cranial cavity, its tympanic opening lies lateral to the fenestra vestibuli and its endocranial opening on the dorsal aspect of the tegmen (see Fig. 4); iii) the tegmen tympani fossa, a depression located on the tegmen tympani, dorsal to the hiatus Fallopii that is directed anteriorly toward the cerebral cavity; and iv) the anteromedial tuberosity of the petrosal, a structure located on the lateral edge of the dorsomedial surface of the petrosal lateral to the foramen acusticum superius (= “knob” of O’Leary 2010). Inferences about soft tissues hypothesized to be associated with the osteological structures described here are briefly discussed in the section “Anatomical Comparisons and Character States.” In Fig. 1 we identify an opening in the skull as the piriform fenestra using the topological definition from O’Leary et al. (2013) that the piriform fenestra is a “gap between petrosal and sphenoid on basicranium, anterolateral to the J Mammal Evol Fig. 1 Orientation of left petrosal in skull, CT scan of Homacodon vagans (USNM 482369) showing petrosal in situ in, with the skull cut virtually through a), sagittal b) coronal, and c) and d) transverse sections; a) shows the dorsomedial petrosal surface, which faces the cerebellum, b) shows the petrosal in anterior view with the outline of the dorsolateral surface that abuts other skull bones (e.g., the squamosal), c) and d) show the relationships of the petrosal to the surrounding bones, in dorsal and ventral views of the skull, respectively. The view shown in (d) indicates that the medial edge of the petrosal approximates the basioccipital/basisphenoid. Scale=10 mm promontorium; typically medial or posteromedial to foramen ovale (…) also distinguished positionally from carotid foramen, which is anteromedial to promontorium versus anterolateral for piriform fenestra” (Character 564, MorphoBank project 773). Our definition of the promontorium includes all flanges extending from it combined with its central, bulging portion. Dichobunoid artiodactyls described in this work all present one single gap anterolateral to the promontorium, identified here as the piriform fenestra. There is no separate carotid foramen, and soft tissue structures such as the internal carotid artery and nerve may be reconstructed as passing through the piriform fenestra (see prior work on artiodactyl basicranial anatomy by Geisler and Luo 1998). Diacodexeidae: Diacodexis ilicis (Figs. 2a, 3a, 4a, 5a, and 10c; Fig. S1) The specimen consists of the right petrosal of a nearly complete skull and partial dentition of Diacodexis ilicis (AMNH 16141; Willwood Formation (“Sand Coulee beds, Clarks Fork Basin, Wyoming, expedition 1912). Based on subsequent biostratigraphic work in this area, the locality yielding the specimen is likely of the earliest Wasatchian age (approximately 55 Ma; Coombs and Coombs 1982; Gingerich 1989), making it one of the oldest fossils ever attributed to Artiodactyla. Only the ventrolateral view of the petrosal of this rare specimen has been described previously (Coombs and Coombs 1982; Geisler and Luo 1998). The reconstructed 3D model shows that the bone suffered little postmortem damage other than the minor abrasions visible on the ventral surface, such as breakage on the promontorium (not previously figured) anterior to the fenestra cochleae and on its medial border. On the ventrolateral surface (Fig. 2), the promontorium is elongate and ellipse-shaped but relatively flat (Fig. 2a). It is defined medially by a subtle groove that distinguishes the flanges of the pars cochlearis from the promontorium itself. There is a distinct epitympanic wing and adjacent to it a narrow posteromedial flange. These two flanges are contiguous around the promontorium from laterally, anteriorly, and medially. The epitympanic wing terminates anteriorly as a point and also has a small lateral process that is pointed and distinct. The fenestra vestibuli is oval and appears to be smaller than the oval-shaped fenestra cochleae; however, the J Mammal Evol Fig. 2 Ventrolateral views of the petrosals of (a) Diacodexis ilicis (AMNH 16141; reversed from right); and (b) Acotherulum saturninum (MNHM Qu 16366). Grey surfaces indicate broken parts of the specimens. Scale=2 mm latter is broken making it possible to only estimate its full size. These fenestrae are separated by a wide crista interfenestralis. Grooves on the promontorium are very subtle but there appears to be a single transpromontorial sulcus that is broad and shallow, and a short and shallow sulcus for the stapedial artery that runs toward the fenestra vestibuli (also noted by Geisler and Luo 1998). We interpret the fossa for the tensor tympani muscle to be wider than that figured by Geisler and Luo (1998) and this depression does not excavate the tegmen tympani. The secondary facial foramen is directly lateral to the fenestra vestibuli occupying a relatively anterior position. It opens onto a distinct and wide facial sulcus that passes lateral to the stapedial muscle fossa. The latter is a broad, poorly-defined depression directly posterior to the fenestra cochleae. Despite breakage to the fenestra cochleae (noted above) it can be ascertained that that it was directed posteriorly. As described by Coombs and Coombs (1982:225), the medial part of the mastoid region directly adjacent to the squamosal bears a small process identified as the tympanohyal. The stylomastoid notch is a broad trough. J Mammal Evol Fig. 3 Dorsomedial views of the petrosals of (a) Diacodexis ilicis (AMNH 16141; reversed from right); and (b) Acotherulum saturninum (MNHM Qu 16366). Scale=2 mm On the pars canalicularis, the tegmen tympani is narrow, only moderately inflated, and lacks an anterior process. The hiatus Fallopii is a long, narrow slit that lies lateral to the anterior margin of the fossa for the tensor tympani muscle. The tegmen tympani is perforated by a clear canal that opens ventrally, directly lateral to the secondary facial foramen, and posterodorsally on the dorsal surface of the tegmen tympani (Figs. 2a and 4a). The epitympanic recess is a shallow depression that lacks a distinct fossa for the head of the malleus; the fossa incudis cannot be identified. The petrosal contribution to the external acoustic meatus is wide and deep, and the ventrolateral tuberosity is a low bump. The tympanohyal is preserved on the medial extremity of the crista parotica, which is a very clear ridge extending posterior to the facial sulcus and stapedial muscle fossa. There is a small caudal tympanic process directly posterior to the fenestra J Mammal Evol Fig. 4 Anterior views of the petrosals of (a) Diacodexis ilicis (AMNH 16141; reversed from right); (b) Acotherulum saturninum (MNHM Qu 16366); (c) Dichobune leporina (MNHN Qu 16586); (d) Homacodon vagans (USNM 482369; reversed from right); (e) ?Helohyus plicodon (USNM 13079; reversed from right); and (f) Gobiohyus orientalis (USNM 26277; reversed from right). Scale=2 mm for a, b, c, f and scale=5 mm for d and e cochleae. The mastoid region is a wedge of bone whose ventral surface is relatively smooth. The dorsomedial surface of the petrosal (Fig. 3) is smooth and the mastoid region comprises approximately half the total size of the bone. The internal acoustic meatus lacks a distinct border and both the foramina acusticum inferius and superius are close to the surface of the bone and clearly visible. A wide crista transversa separates these two openings. The tegmen tympani meets the dorsomedial surface at a right angle and there is no prefacial commissure fossa (a fossa located on the medial surface of the tegmen tympani (O’Leary 2010)= suprameatal fossa of Luo and Gingerich 1999). The basicapsular groove is very distinctive and extends continuously from the anterior extreme of the bone to the mastoid region. Posteriorly, a sharp ridge separates the internal acoustic meatus from the subarcuate fossa. This fossa is extremely deep and large, almost equal in size to the internal acoustic meatus. The opening into the fossa is oval with the long axis oriented mediolaterally. The petromastoid canal is absent and the subarcuate fossa is distinguished by the presence of a large, wide groove on its anterior border. A sharp and long crista petrosa is present anterior to the subarcuate fossa (Fig. 3a). It delineates a wide fossa on the tegmen tympani, dorsal to the hiatus Fallopii (Fig. 4a). The roughened bone of the mastoid region and some of the cancellous bone internal to it are visible in this view, as well as in ventromedial view J Mammal Evol Fig. 5 Ventromedial views of the petrosals of (a) Diacodexis ilicis (AMNH 16141 reversed from right); (b) Acotherulum saturninum (MNHM Qu 16366); (c) Dichobune leporina (MNHN Qu 16586); (d) Homacodon vagans (USNM 482369; reversed from right); (e) ?Helohyus plicodon (USNM 13079; reversed from right); and (f) Gobiohyus orientalis (USNM 26277; reversed from right). Scale=2 mm for a, b, c, f and scale=5 mm for d and e (Fig. 2a). There is no mastoid plate, a thin ledge of bone that may overhang the subarcuate fossa (see O’Leary 2010; Orliac 2012: fig. 3). This structure, so far identified only in suoids (Orliac 2012), is not observed in D. ilicis. The cochlear aqueduct is situated at the posterior end of the foramen acusticum inferius, on the medial surface of the petrosal. It lies in a deep concavity outlined by a bony ridge. The vestibular aqueduct is a small aperture, opening in a posterior position, located medial and posterior to the subarcuate fossa. The mastoid region consists of two main areas: a ventrally protruding part that bears a wide vascular groove and a flat wedge-shaped dorsolateral part (Fig. 2a). The external acoustic meatus lies between these two regions. The ventrally protruding part of the mastoid region is widely exposed on the ventral part of the skull, where it is intercalated between the base of the paracondylar process of the exoccipital and the posttympanic process of the squamosal. Our in situ reconstruction reveals that part of the flattened, wedge-shaped dorsal part of the mastoid region is also exposed on the temporal part of the skull (Fig. 10a). We cannot, however, eliminate the possibility that this unexpected exposure may be due to deformation and breakage of the specimen. J Mammal Evol Fig. 6 Ventrolateral views of the petrosals of (a) Dichobune leporina (MNHN Qu 16586); and (b) Homacodon vagans (USNM 482369; reversed from right). Grey surfaces indicate broken parts of the specimens. Scale=2 mm for a, and scale=5 mm for b Cebochoeridae: Acotherulum saturninum (Figs. 2b, 3b, 4b, and 5b; Fig. S2) The virtual reconstruction of the left petrosal was derived from the 3D reconstruction of the in situ petrosal of the skull of Acotherulum saturninum (MNHN Qu 16366) from the Quercy fissure fillings (S-W France, Paleogene, unknown locality, BioChroM’97 1997). Right and left petrosals of the specimen had been previously exposed by removal of the tympanic bullae (completely removed on the right side, but half remains on the left). On the ventrolateral surface (Fig. 2b), the promontorium is an anteroposteriorly elongate ellipsoid that reveals little evidence of the underlying coiling of the cochleae. The promontorium gives rise to a large epitympanic wing, which is fully contiguous with the posteromedial flange on the promontorium. The anterior extreme of the epitympanic wing comes to a gentle point. There is a pointed lateral process of the epitympanic wing that defines the posterolateral edge of the piriform fenestra. The fenestra vestibuli is an elongate oval and is larger than the fenestra cochleae. The fenestra cochleae is narrow and opens in a posterior direction. The promontorium is distinguished by two very distinct grooves: a single transpromontorial sulcus that extends from the posteromedial to the anterior aspect of the promontorium, and a sulcus for the stapedial artery branching from it that terminates at the medial edge of the fenestra vestibuli. The transpromontorial sulcus creates a pronounced furrow on the J Mammal Evol Fig. 7 Dorsomedial views of the petrosals of (a) Dichobune leporina (MNHN Qu 16586); and (b) Homacodon vagans (USNM 482369; reversed from right). Scale=2 mm for a, and scale=5 mm for b medial side of the lateral process of the epitympanic wing. The sulcus for the stapedial artery is on the promontorium anterior to the broad crista interfenestralis. The fossa for the tensor tympani is a broad, shallow depression with indistinct boundaries; it very gently excavates the tegmen tympani. On the pars canalicularis, the tegmen tympani is a narrow shelf that is distinctly separated from the promontorium; it is only somewhat inflated, occupying approximately one-fourth of the width of the ventrolateral surface of the petrosal. There is a very large tegmen tympani canal, located just posterior to the opening of the hiatus Fallopii (Fig. 4b). The hiatus Fallopii is a wide opening on the dorsolateral edge of the petrosal, halfway along the fossa for the tensor tympani. The anterior process, which terminates in a point, does not extend anterior to the epitympanic wing. The secondary facial foramen is directly lateral to the fenestra vestibuli. The epitympanic recess is an indistinct depression lateral to the secondary facial foramen, and it lacks a distinct fossa for the head of the malleus. The fossa incudis cannot be identified. Anterior to this is the ventrolateral tuberosity, which is a small process. The petrosal contribution to the external acoustic meatus forms a shallow, wide trough. The facial sulcus is a distinct and deep channel extending to the stylomastoid notch; the stapedial muscle fossa, is not visible, covered by the crista parotica. The tympanohyal is a small knob located just posterior to the fenestra cochleae. The crista parotica extends from the posterior aspect of the stylomastoid notch and runs anteriorly between the secondary facial foramen and the J Mammal Evol Fig. 8 Ventrolateral views of the petrosals of (a) ?Helohyus plicodon (USNM 13079; reversed from right); and (b) Gobiohyus orientalis (USNM 26277; reversed from right). Grey surfaces indicate broken parts of the specimens. Scale=5 mm for a, and scale=2 mm for b epitympanic recess. The caudal tympanic process is an indistinct bump posterior to the fenestra cochleae that is separated from the posteromedial flange by a notch. The pars cochlearis does not protrude medially relative to the mastoid region. The dorsomedial surface of the petrosal (Fig. 3b) is smooth. The internal acoustic meatus is shallow with the foramen acusticum superius relatively round and the foramen acusticum inferius more elongate and broad. These openings are separated by a distinct, ridge-shaped crista transversa. There is no prefacial commissure fossa and the tegmen tympani meets the dorsomedial surface of the petrosal at a right angle. The basicapsular groove is present and visible on the dorsomedial surface extending just anterior to the minute opening for the cochlear aqueduct. The cochlear aqueduct is situated ventromedial to the foramen acusticum inferius. The small opening of the vestibular aqueduct is situated in a dorsal depression, located medial and posterior to the subarcuate fossa. A distinct ridge defines the ventromedial border of the deep and oval-shaped subarcuate fossa. The subarcuate fossa shelters a petromastoid canal (out of view in Fig. 3b). There is no crista petrosa and no pronounced tegmen tympani fossa. The mastoid region tapers posteriorly and then fans into an irregular, roughened area. A mastoid plate is absent. The J Mammal Evol Fig. 9 Dorsomedial views of the petrosals of (a) ?Helohyus plicodon (USNM 13079; reversed from right); and (b) Gobiohyus orientalis (USNM 26277; reversed from right). Scale=5 mm for a, and scale=2 mm for b mastoid region represents approximately half of the total size of the petrosal. The mastoid region is also visible on the lateral aspect of the skull ventrally near the base of the paracondylar process of the exoccipital, just posterior to the opening of the external auditory meatus. The flattened, wedged-shaped dorsal part of the mastoid region is not exposed on the temporal part of the skull. Dichobunidae: Dichobune leporina (Figs. 4c, 5c, 6a, and 7a; Fig. S3) The petrosal of a representative species of Dichobunidae is described here from the virtual reconstruction of the right in situ petrosal of Dichobune leporina (MNHN Qu 16366) from the Quercy fissure fillings (S-W France, Paleogene, unknown locality). The tympanic bullae are missing on both sides. J Mammal Evol The promontorium (Fig. 6a) is oval in outline shape and is relatively flat except immediately anterior to the fenestra cochleae where it is convex. The promontorium lacks transpromontorial and stapedial sulci. At the anterior aspect of the promontorium, between the epitympanic wing and the lateral process of the epitympanic wing, are two depressions of relatively equal size. Posteriorly, the fenestra vestibuli is oval and separated by a wide crista interfenestralis from the larger fenestra cochleae, which has a round outline. The outline of the fenestra cochleae is weakly defined at the posteromedial margin. The fossa for the tensor tympani is a large, wide trough that has a poorly defined anterior margin. The hiatus Fallopii is a very small pit situated just anterior to the anterior margin of the fossa for the tensor tympani on the anterolateral edge of the petrosal. There is a large epitympanic wing extending anteriorly from the fossa for the tensor tympani; this wing comes to a sharp point anteriorly. The epitympanic wing also has a sharp lateral process that defines the posterolateral edge of the piriform fenestra. The posterior border of the piriform fenestra is a distinct concavity in the petrosal. The epitympanic wing is fully continuous with the posteromedial flange. The caudal tympanic process is an unremarkable knob posterior to the fenestra cochleae. A sharp crista parotica runs from the posteromedial extreme of the bone along the medial edge of the tegmen tympani as a distinct, unbroken ridge. The tegmen tympani is a modest shelf of bone with a small, knob-shaped ventrolateral tuberosity. The anterior process of the tegmen tympani is small and blunt. The tegmen tympani canal is a small but distinct opening. The contribution of the petrosal to the external acoustic meatus is limited to a shallow and wide trough; the epitympanic recess is very indistinct. The secondary facial foramen is a slit that is anterolateral to the fenestra cochleae (and thus relatively anterior in position). It opens onto a wide and distinctive facial sulcus that is continuous with the posterior margin of the fenestra cochleae. The tympanohyal cannot be identified on the mastoid region and is either broken or absent. The stapedial muscle fossa is wide and deep, its medial margin is not defined. The mastoid region is wedge-shaped and roughened. The epitympanic recess is an unremarkable depression that lacks a distinct fossa for the head of the malleus. The fossa incudis cannot be identified. The smooth dorsomedial surface (Fig. 7a) has a rather small internal acoustic meatus with relatively poorly defined borders. The foramina acusticum superius and inferius are small, oval, and of approximately equal size. They are distinctly separated by a crista transversa. The basicapsular groove is very conspicuous in this view and runs from the anterior extreme of the bone to the mastoid region. The cochlear and vestibular aqueducts are located within very small pits, and are positioned more posteriorly than is the internal acoustic meatus. The vestibular aqueduct is located in a relatively dorsal position, in a small depression, superior to the posterior margin of the subarcuate fossa. The internal acoustic meatus is well-separated from the deep and wide subarcuate fossa, a depression that is larger and deeper than the internal acoustic meatus. The posterolateral edge of the subarcuate fossa is marked by a clear groove, and a petromastoid canal is present. The tegmen tympani meets the dorsolateral surface at a right angle (Fig. 4c) and the prefacial commissure fossa is absent. A short crista petrosa is present anterior to the subarcuate fossa (Fig. 7a). It delineates a small fossa, dorsal to the hiatus Fallopii (Fig. 4c). The mastoid region occupies almost half of the volume of the petrosal and the cancellous bone that comprises it is partly visible (Figs. 5c and 7a). A mastoid plate is absent. The mastoid region is visible on the ventral surface of the skull anterior to the base of the paracondylar process of the exoccipital. The mastoid region is not exposed on the temporal surface of the cranium. Homacodontidae: Homacodon vagans (Figs. 1, 4d, 5d, 6b, 7b, and 10a; Figs. S4, S11) The petrosal of a representative member of the Homacodontidae is described here from the virtual reconstruction of the left in situ petrosal of Homacodon vagans (USNM 482369). It was collected by Frank L. Pearce in 1941 from the Bridger D Formation, locality “Bridger Basin, 100–150 ft above white layer of S Twin Buttes, Wy [sic].” It has some breakage just anterior to the fenestrae cochleae and vestibuli, visible in the reconstructed specimen. The tympanic bullae are not preserved. The promontorium is mostly flat and is oval in outline. We interpret this specimen to have both transpromontorial and stapedial sulci as these sulci are observable on the in situ left petrosal (not reconstructed). The fenestra cochleae is much larger than the fenestra vestibuli. The crista interfenestralis created a substantial separation between these openings; however, due to breakage the full size of this crest cannot be observed. The fossa for the tensor tympani is a large, wide, and well-defined ovoid depression. There is a large epitympanic wing that comes to a point anteriorly. The epitympanic wing is fully connected to the posteromedial flange forming a complete lip around the promontorium. The specimen has a large and pointed lateral process of the epitympanic wing that defines a distinct piriform fenestra. The hiatus Fallopii is oval and opens dorsally, slightly posterior to the anterior margin of the fossa for the tensor tympani. The tegmen tympani is mediolaterally narrow, terminating in a small, spike-shaped anterior process. The tegmen tympani canal is a narrow passage with a small opening that is closer to the ventral than to the dorsal surface of the bone (Fig. 4d). The ventrolateral tuberosity is a small knob. The contribution of the petrosal to the external acoustic meatus is a shallow, narrow but deep trough. The epitympanic recess is an indistinct structure that is offset ventrally from the level of the facial sulcus; a very subtle depression could correspond to the fossa incudis. The secondary facial foramen is a small slit that is anterolateral to the fenestra vestibuli, and that opens onto a wide and shallow facial sulcus. The crista parotica is a J Mammal Evol distinctive, continuous ridge from the anterior process of the tegmen to the stylomastoid notch. There is a small caudal tympanic process. The tympanohyal is broken on the reconstructed petrosal but can easily be seen on the other side of the skull where it consists of a small knob located just posterior to the fenestra cochleae. It almost closes off the anterior margin of the stylomastoid foramen as on the specimen illustrated by Coombs and Coombs (1982:fig. 1). The mastoid region is large and wedge-shaped comprising over half of the total size of the bone (Figs. 4, 5, 6b, and 7); its cancellous structure is visible in several figures. The dorsomedial surface of the petrosal (Fig. 7b) is smooth. The internal acoustic meatus is poorly defined; both the foramina acusticum superius and inferius are oval in outline and separated by only a narrow crista transversa. The basicapsular groove is very wide and distinct, running from the anterior extreme of the bone to the mastoid region. The cochlear and vestibular aqueducts are small oval pits, both hidden under Fig. 10 Coronal (left column) and lateral (right column) views of the right ear region and skull showing dorsal exposure of the mastoid region: (a) Diacodexis ilicis (AMNH 16141), reversed from left, (b) Homacodon vagans (USNM 482369), reversed from left; and (c) ?Helohyus plicodon (USNM 13079), reversed from left. Double-headed arrows indicate the location of the coronal section in the 3D reconstruction. Scale=10 mm J Mammal Evol flanges of bone. The cochlear aqueduct is situated posterior to the posterior margin of the foramen acusticus inferius, and the vestibular aqueduct is more posterior, as posterior as the posterior margin of the subarcuate fossa. The internal acoustic meatus is separated from the subarcuate fossa by a wide bar of bone. The subarcuate fossa is deep and wide; it is broadly open on the posterodorsal extreme and contains a wide petromastoid canal (out of view in Fig. 7b). The tegmen tympani is perpendicular to the dorsomedial surface of the petrosal and there is no prefacial commissure fossa. A crista petrosa is present anterior to the subarcuate fossa (Fig. 7b) and delineates a wide fossa on the tegmen tympani, dorsal to the hiatus Fallopii (Fig. 4d). A mastoid plate is absent. The mastoid region is widely exposed on the external surface of the skull (Fig. 10a), particularly on the ventral surface, where it intercalates between the base of the paracondylar process of the exoccipital and the posttympanic process of the squamosal (Fig. 1d). The mastoid region is also visible in lateral view, and its dorsal part is also widely exposed on the temporal surface of the skull situated between the parietal of the occipital bones (Fig. 10a). Helohyidae: ?Helohyus plicodon (Figs. 4e, 5e, 8a, 9a, 10b; Figs. S5, S12) The petrosal of ?Helohyus plicodon is described here from the virtual reconstruction of the right in situ petrosal of the specimen AMNH 13079 from the Bridger Formation locality B5, Sweetwater County, Wyoming. Only the ventrolateral surface of this specimen was previously described by Coombs and Coombs (1982). Sinclair (1914) tentatively referred this partial skull to Helohyus. The specimen has a broken posterior tooth row, and this taxonomic assignment is based on the size of the specimen and on an associated P3. Coombs and Coombs (1982) noted the specimen’s potential affinities with Lophiohyus, another helohyid, and thus referred to this specimen to as ?Helohyus. In the absence of any new information, we follow Coombs and Coombs (1982) here. The promontorium is circular in outline and slightly bulging. Sulci on the promontorium are faint but can be seen in the CT scan. The transpromontorial sulcus makes an almost right angle bend anterior to the fenestra cochleae, where it gives rise to a sulcus for the stapedial artery that is equally wide. The fenestra cochleae is much larger and more oval than is the fenestra vestibuli. The fossa for the tensor tympani is a very shallow and low depression. The epitympanic wing and posteromedial flange form the anterior and medial periphery of the promontorium but their separations from it are indistinct. The tegmen tympani is mediolaterally broad (about half the width of the promontorium); the anterior process of the tegmen tympani is blunt and short. There is a small tegmen tympani canal, opening lateral to the fenestra vestibuli and visible in lateral view (Fig. 4e). A very modest lateral process of the epitympanic wing lies anteromedial to the tegmen tympani. The ventrolateral process of the tegmen tympani is absent. The hiatus Fallopii is a small hole oriented dorsally (Fig. 4). The location of the secondary facial foramen is difficult to determine because the floor of the fossa for the tensor tympani muscle is partly broken. The foramen appears to be positioned relatively far anteriorly, but this may be due to it being broken through to the dorsal side. The facial sulcus is a deep and distinct channel. The stapedial muscle fossa is not well preserved and has a crack running through it. The caudal tympanic process is a small knob posterior to the fenestra cochleae. The tympanohyal cannot be identified on the mastoid region and may either be broken or absent. The mastoid region is large and wedge-shaped (Figs. 4 and 5) and exhibits cancellous bone (Fig. 5). The epitympanic recess is a wide indistinct depression that lacks a distinct fossa for the head of the malleus. The fossa incudis cannot be identified. In dorsomedial view (Fig. 9a) the surface of the bone is completely smooth; the internal acoustic meatus is roughly oval in outline shape. The foramen acusticum superius is oval and narrow and the foramen acusticum inferius is larger and round. The crista transversa forms a wide separation between these two openings. The basicapsular groove is continuous from anteromedial to posteromedial and does not extend posterior to the internal acoustic meatus. The subarcuate fossa is set off from the internal acoustic meatus by a wide bar of bone; the fossa itself is extremely shallow and wide, and it is relatively expanded anteriorly. A petromastoid canal occurs at the deepest part of the fossa. The tegmen tympani is perpendicular to the dorsomedial surface and a prefacial commissure fossa is absent. A very short crista petrosa is present anterior to the subarcuate fossa and joins the anteromedial tuberosity of the petrosal anteriorly (Fig. 9a). There is no indication of a tegmen tympani fossa. The cochlear aqueduct is a minute opening within a rounded pit that is situated directly posterior to the foramen acusticum inferius. The vestibular aqueduct lies more posteriorly and dorsally, nearer the posterior margin of the subarcuate fossa. A mastoid plate is absent. The mastoid region is widely exposed on the external surface of the skull (Fig. 10a). This exposure consists of two parts: a ventral protrusion that bears a wide sulcus, and a flat, wedgeshaped dorsolateral part. The ventrally protruding part of the mastoid region is the widest area of external exposure and it intercalates between the paracondylar process of the exoccipital and the posttympanic process of the squamosal. The more dorsal part of the mastoid region that is exposed on the temporal part of the skull intercalates between the parietal and occipital bones. Helohyidae?: Gobiohyus orientalis (Figs. 4f, 5f, 8b, and 9b; Figs. S6, S13) The petrosal of Gobiohyus orientalis is described here from the virtual reconstruction of the right in situ petrosal of the specimen AMNH 26277 from Inner Mongolia (Ulan Shireh Formation, Chimney Butte North Mesa, Shara Murun region). J Mammal Evol The specimen consists of a skull (the only one that has ever been reported) and dentition. The ventrolateral surface of the right petrosal is exposed and the dorsomedial surface has been prepared and is partially exposed (the right petrosal is only partially preserved). This view of the petrosal has been described previously in Coombs and Coombs (1982), and we include here new details visible from the CT scans. The tympanic bullae are not preserved. There are a few breaks in the promontorium anterior to the fenestra cochleae but the bone is generally intact. The promontorium is almost circular to square in outline shape. It is surrounded by an epitympanic wing and a posteromedial flange that are fully contiguous. The lateral process of the epitympanic wing is a short prong that marks the lateral edge of the piriform fenestra. The transpromontorial sulcus is very shallow and stapedial sulus is not conspicuous. The fenestra vestibuli is round and only slightly smaller than the oval fenestra cochleae. The two are separated by a crista interfenestralis that was as wide as the fenestra cochleae. The fossa for the tensor tympani muscle shows some breakage but is preserved adequately to show that it was a distinct, large, and oval depression lateral to the secondary facial foramen. This fossa extensively excavates the tegmen tympani. The hiatus Fallopii is very wide and is located anterior to the anterior margin of the fossa for the tensor tympani. This margin is broken, creating an artificial union between the hiatus Fallopii and the facial sulcus. The secondary facial foramen appears to have been located in a relatively anterior position in comparison to the promontorium. The facial sulcus is a long trough that passes lateral to the fenestra vestibuli and medial and posterior to the stapedial muscle fossa. The tegmen tympani is a narrow bar of bone that terminates in a blunt anterior process. The external acoustic meatus is deep and very wide and the epitympanic recess is indistinct. The fossa incudis cannot be identified with certainty, but a very subtle depression could correspond to this fossa. The ventrolateral tuberosity is an indistinct knob anterior to the external acoustic meatus. A very small tegmen tympani canal appears approximately halfway between the ventral and dorsal surfaces of the petrosal; it perforates the tegmen tympani lateral to the fenestra vestibuli and opens on its lateral surface through a small foramen (Fig. 4e). Coombs and Coombs (1982:225) described a slight development of a tympanohyal on the mastoid region of Gobiohyus. We could not identify this structure; it may have been broken subsequent to their paper or is absent. The crista parotica is a sharp ridge that extends fully medial to the external acoustic meatus and terminates posterior to the estimated location of the stylomastoid notch. The mastoid region is large and wedge-shaped. The caudal tympanic process is fully continuous with the posteromedial flange and forms a low shelf posterior to the fenestra cochleae. The dorsomedial surface of the petrosal is completely smooth with a deep internal acoustic meatus that has an irregular margin. The prefacial commissure fossa is absent. The crista transversa is broken making the distinction between the foramen acusticum superius and inferius unclear. The basicapsular groove is most clearly defined at its anterior extreme. It is identifiable, but not well defined, posteriorly. The cochlear aqueduct is a small opening in a slit posteromedial to the internal acoustic meatus. The vestibular aqueduct is situated more dorsally and is posterior to both the internal acoustic meatus and the subarcuate fossa. Separating the internal acoustic meatus from the subarcuate fossa is a wide ridge of bone. The subarcuate fossa is deep; however, it is poorly defined on its dorsolateral margin where it becomes shallower. It is separated from a thin flange of bone by a groove. The subarcuate fossa shelters a clear petromastoid canal (out of view in Fig. 9b). There is no crista petrosa and no contact with the cerebrum on the tegmen tympani. A mastoid plate is absent. The mastoid region consists of two main areas: a ventrally protruding part that bears a wide sulcus and a dorsolateral part that is a flat wedge. These are separated by the external acoustic meatus. The ventrally protruding part of the mastoid region is widely exposed on the ventral part of the skull, where it is intercalated between the base of the paracondylar process of the exoccipital (structure not preserved on specimen) and the posttympanic process of the squamosal. Anatomical Comparison and Character States Below we compare the anatomy of the petrosal bone among dichobunoids and other mammals and make brief remarks about the hypothesized soft tissue contents of certain bony features. We also review relevant character states previously published for basal Artiodactyla based on data available from the literature. Remarkable characters and their hypothesized polarities are here discussed in the current context of an unstable phylogeny for dichobunoids (e.g., Spaulding et al. 2009; Geisler and Theodor 2009). The distribution of the discrete characters discussed bellow among dichobunoid taxa is summarized in Table 1. Promontorium Shape and Vascular Grooves Each of the dichobunoids studied here has an elongate promontorium that is round-oval in outline, like that of many terrestrial artiodactyls (O’Leary 2010). The promontoria of the dichobunids are relatively flat whereas those of cetaceans are relatively bulbous (Luo and Gingerich 1999; O’Leary 2010). Several of the promontoria described here exhibit both transpromontorial and stapedial sulci arranged at a right angle. The petrosal of Dichobune is devoid of such sulci (Fig. 6). The grooves are particularly pronounced in the cebochoerid Acotherulum (Fig. 2b) where they deeply excavate the promontorium. Within crown Artiodactyla, a transpromontorial groove has been reported in each of its major extant subclades (suines, hippopotamids, ruminants, and camelids), either within the subclades themselves, or on their stems, or both (O’Leary J Mammal Evol Table 1 Comparative morphological features of the dichobunoid species in this study Diacodexis Dichobune Acotherulum Homacodon Gobiohyus ?Helohyus tps lpew ftt tts aptt cp amt ttc ttf hfs demr sf + + ? + + + + + + + small + + + + + + + + + + + + + - + + + + + + + + + + + - + - + + ? + + + + + + - Abbreviations: amt anteromedial tuberosity; apt anterior process of tegmen tympani; cp crista petrosa; demr dorsal exposure of mastoid region; ftt fossa for tensor tympani excavating the tegmen tympani; hfs hiatus Fallopii size; lpew lateral process of the epitympanic wing; sf subarcuate fossa; tps transpromontorial sulcus; ttc tegmen tympani canal; ttf tegmen tympani fossa; tts tegmen tympani size 2010). Unambiguous transpromontorial grooves in Diacodexis and Acotherulum shown here indicate that this bony feature also characterized the oldest fossils to have been referred to Artiodactyla. Future comprehensive phylogenetic work testing the positions of these taxa will be crucial for determining whether these ancient taxa are also the most basal, and whether, as hypothesized by Wible (1986), the presence of these grooves is the primitive condition for both Artiodactyla and Artiodactylamorpha. Whether the presence of promontorial grooves necessarily implies the presence of a transpromontorial internal carotid artery and/or nerve, or the stapedial artery requires further description of soft tissue data from dissections and the incorporation of these data directly into the phylogenetic analysis for optimization of both types of characters (Bryant and Russell 1992; Witmer 1995; O’Leary et al. 2013). This work will be particularly important given that Artiodactyla is a clade for which loss or reduction of the internal carotid artery has been reported (reviewed in Geisler and Luo 1998; O’Leary 2010), making the clade an interesting test case for how osteological and soft anatomy co-evolve. Flanges Extending from the Promontorium All of the species examined here have flanges extending from the promontorium as reported in many other terrestrial artiodactyls (O’Leary 2010). These include both the epitympanic wing and the posteromedial flange. These structures are continuous in all of these species but vary in width from relatively narrow in Diacodexis (Fig. 2a) and Dichobune (Fig. 6a) to wider in Gobiohyus (Fig. 8b). The epitympanic wing typically forms a sharp point anteriorly in Dichobune, Homacodon, Acotherulum, and particularly in Diacodexis. Lateral to the point is the lateral process of the epitympanic wing. These taxa also lack a separate carotid foramen. This process, which is present in all of the taxa examined here, and is particularly sharp in Dichobune, Homacodon, and Acotherulum, has not previously been identified in extant or fossil artiodactyls (e.g., Geisler and Luo 1998; Theodor 2010; O’Leary 2010), except in the basal suoids Perchoerus and Palaeochoerus (= lateral anterior process, Orliac 2012). As noted above, the lateral process of the epitympanic wing delineates the posterior and lateral edges of the piriform fenestra (this fenestra is clearly separated from the foramen ovale in dichobunoids; Dechaseaux 1969; Coombs and Coombs 1982; Russell et al. 1983). In ?Helohyus, the posterior margin of the piriform fenestra indents only slightly into the pars cochlearis, and, as such, the lateral process of the epitympanic wing is limited to a small blunt spike. Overall the morphology of this region differs most substantially in ?Helohyus, in which the anterior edge of the epitympanic wing is rounded, without a pointed process. The lateral process of the epitympanic wing has not been reported in “condylarthran” ungulates, and is absent in Protungulatum (MacIntyre 1972; O’Leary 2010), Meniscotherium (Gazin 1965), Hyopsodus (Cifelli 1982), Pleuraspidotherium and Orthaspidotherium (Ladevèze et al. 2010). Among euungulates (Asher et al. 2009), a lateral process of the epitympanic wing has so far only been observed in Artiodactyla. In the fetal stage of living eutherians, an alicochlear commissure separates the carotid foramen from the piriform fenestra (MacPhee 1981: 21 and fig. 1). This commissure can be replaced in the adult by an expansion of the petrosal, which ensures the complete separation between the carotid foramen and the piriform fenestra (e.g., in Carnivora, Wang and Tedford 1994). In the species described in this paper there is no separate carotid foramen, a condition that has been described in other extinct artiodactyls and relatives (see Geisler and Luo 1998), and the transpromontorial branch of the internal carotid artery may have entered the cranial cavity via the piriform fenestra (Geisler and Luo 1998). Fossa for the Tensor Tympani All the dichobunoids described here present a clearly delimited fossa for the tensor tympani on the ventromedial surface of the petrosal, anterior to the fenestra vestibuli. Diacodexis shows a wide and shallow fossa for the tensor tympani muscle; J Mammal Evol however, this fossa does not form an excavation into the tegmen tympani as reported in some artiodatyls (O’Leary 2010). The fossa is particularly wide in ?Helohyus and Gobiohyus (Fig. 9) and deeply excavates the tegmen tympani. sensu Billet and Muizon (2013) and most probably received part of the temporal lobe of the cerebrum and the trigeminal ganglion of the trigeminal nerve (V). Tegmen Tympani Canal Tegmen Tympani The tegmen tympani is a distinct, continuous shelf on the lateral aspect of the pars canalicularis of the petrosal and is continuous with the mastoid region in euungulates (Luo and Gingerich 1999; O’Leary 2010). In the taxa studied here the tegmen tympani is typically narrow relative to the width of the promontorium, an important exception being ?Helohyus, which has the widest tegmen tympani (Fig. 8a). In Diacodexis, the tegmen tympani is relatively small consisting only of a thin strut of bone. Anterior Process of the Tegmen Tympani All extant and fossil artiodactyls described in the literature so far exhibit an anterior process of the tegmen tympani (when this area is preserved). The anterior process of the tegmen tympani of dichobunoid taxa is relatively modest in size and blunt in shape. O’Leary (2010) reported the presence of this feature as a synapomorphy of a clade of Artiodactyla+ mesonychians based on the phylogenetic analysis of O’Leary and Gatesy (2008). Those authors did not record a score for this feature for either species of Diacodexis that they studied because the feature could not be assessed from the available data. We find that an anterior process of the tegmen tympani is, however, absent in North American Diacodexis, indicating that the character does show some variability, at least among extinct taxa that have been assigned to Artiodactyla. Should future phylogenetic studies corroborate the highly nested position of Diacodexis found in Spaulding et al. (2009), the absence of this feature would optimize as a reversal. This process is also absent in “condylarthran” ungulates such as Protungulatum sp. (O’Leary 2010:fig. 3), Meniscotherium (Cifelli 1982), and Hyopsodus (Gazin 1965), but is present in Orthaspidotherium (Ladevèze et al. 2010). Tegmen Tympani Fossa In Diacodexis, the tegmen tympani is excavated by a wide fossa, located dorsal to the hiatus Fallopii and separated from the subarcuate fossa by a sharp crista petrosa (Fig. 4a). This fossa is also present in Dichobune (Fig. 4c) and Homacodon (Figs. 1b and 4d), but is smaller and shallower than in Diacodexis. Acotherulum (Fig. 4b), Gobiohyus (Fig. 4e), and ?Helohyus (Fig. 4f) do not present this fossa. The tegmen tympani fossa is located lateral and anterior to the crista petrosa and is directed towards the cerebral cavity. This area likely corresponds in part to the fossa for trigeminal ganglion In all species described here, the tegmen tympani is perforated by a canal, the tegmen tympani canal (new term, this study). This passageway pierces the tegmen tympani from the ventrolateral surface of the petrosal adjacent to the secondary facial foramen, and opens onto its anterolateral surface (Fig. 4). The canal connects anteriorly to the orbitotemporal canal (Fig. S14). We have separately observed, but do not figure, that the tegmen tympani canal exists in Bunomeryx (AMNH 2070, Homacodontidae), and in the European Eocene artiodactyls Mouillacitherium (UM2 ACQ 6625, Hyperdichobuninae), and Cainotherium (UM2 PDS 3352). Dechaseaux (1974) and Coombs and Coombs (1982) also did not report the presence of this canal in the dichobunoid taxa they studied, and, with the exception of the five genera examined here, no other artiodactyls for which the petrosal is known has been described as having such a canal (e.g., Geisler and Luo 1998; Luo and Gingerich 1999; O’Leary 2010). The most similar structure is what Geisler and Luo (1998: fig. 2) identified in some extinct artiodactyls as the stapedial foramen, but that structure lies between the petrosal and the squamosal. Those authors also identified a stapedial foramen in the archaeocetes Pakicetus, Basilosaurus, and Dorudon. A tegmen tympani canal for the ramus superior of the stapedial artery has also been identified in the ‘condylarths’ pleuraspidotheriids (Ladevèze et al. 2010) and in notungulates (Billet and Muizon 2013). The orbitotemporal canal has been reported to transmit the anterior division of the ramus superior of the stapedial artery into the orbit (Wible and Gaudin 2004:162). It is possible that the tegmen tympani canal, connected to the orbitotemporal canal, gave passage to the ramus superior of the stapedial artery in life. This soft tissue – osteology correlation would also be consistent with the presence of a stapedial sulcus observed in these taxa. A tegmen tympani canal is absent in some of the most ancient taxa that may be members of or on the stem to crown clades within Artiodactyla: Suoidea (Perchoerus, Orliac 2012), Ruminantiamorpha (Leptomeryx AMNH 53596), Hippotamidamorpha (Heptacodon AMNH 12462). A canal for the ramus superior of the stapedial artery passing through the petrosal or petrosal/squamosal suture on dorsolateral edge of epitympanic recess is considered to be plesiomorphic for Artiodactyla and Perissodactyla (Ladevèze et al. 2010: character 11) and is, for example, present in the “condylarths” Pleuraspidotherium and Orthaspidotherium (Ladevèze et al. 2010). Loss of the tegmen tympani canal would therefore be derived among Artiodactyla and might have occurred several times in artiodactyl evolution. J Mammal Evol Hiatus Fallopii Although out of view in the illustrations except from the ventromedial view in Fig. 5, the hiatus Fallopii is typically a very small opening at the anterior edge of the tegmen tympani in all the species we examined (e.g., Fig. 1b). Only one of the taxa studied, Acotherulum, has an enlarged (relative to the total size of the petrosal) hiatus such as is found in hippopotamids and cetaceans (O’Leary 2010). Epitympanic Recess The epitympanic recess in the taxa described here is similar to that described in many terrestrial artiodactyls (O’Leary 2010), in that it is ventrally offset relative to the facial sulcus. The recess is also an indistinct and shallow basin. None of the early artiodactyls studied here exhibits a distinct fossa for head of malleus as seen in cetaceans. The fossa incudis can not be identified with certainty on any of the 3D reconstructions, but a very subtle depression could correspond to this fossa in Homacodon and Gobiohyus. Mastoid Region, Including Exposure The mastoid region is relatively large in all dichobunoids represented in this study. The proportion of the mastoid region compared to the total size of the petrosal is about the same in Diacodexis, Dichobune, and Acotherulum. Homacodon and ?Helohyus present a relatively greater development of the dorsal part of the mastoid region such that it accounts for more than half of the total size of the bone. In all dichobunoids, the mastoid region is broadly exposed on the ventral part of the skull at the base of the paracondylar process of the exoccipital. This region is also visible in lateral view, anterior to the paracondylar process, as described by Russell et al. (1983) in D. pakistanensis. More surprisingly, however, is that the mastoid region is also widely exposed on the temporal part of the skull and the dorsal portion of the bone is intercalated between the parietal and the occipital bones in Homacodon (Fig. 10b) and ?Helohyus (Fig. 10c). In Diacodexis (Fig. 10a), the mastoid region is also exposed on the temporal side; however, given the postmortem deformation of this specimen, it is uncertain whether this exposure is artifactual or real. The morphology of this area is also not clear in Dichobune, but if this region were exposed at all, it is likely the exposure would have been very small. We did not observe this exposure in the cebochoerids Acotherulum (MNHN Qu 16366) and Cebochoerus (MNHN Qu 17151), and, to our knowledge, this exposure has not been reported in other euungulates. Basicapsular Groove The basicapsular groove is present in all dichobunoids described here, extending from the medial edge of the bone to the base of the lateral process of the epitympanic wing. The groove is wholly situated on the dorsomedial surface of the bone as is found in the majority of terrestrial artiodactyls described so far (O’Leary 2010). This is in contrast to a more ventral position reported in Moschus, extant hippopotamids, Babyroussa, and Bothriogenys (O’Leary 2010). Various positions of this groove (hypothesized to house the inferior petrosal sinus, Presley 1979) relative to the basioccipital bone have been described within Artiodactyla (for review see Theodor 2010:fig. 4, “petrobasiliar canal”). For the dichobunoid species described here, with the exception of Gobiohyus, the location is similar to that described for Cainotherium (Theodor 2010:fig. 3c). The hypothesized primitive condition for artiodactyls - a dorsally positioned groove (O’Leary 2010) - is confirmed here, and a ventral location of the basicapsular groove appears to be a derived condition that may appear several times. Crista Petrosa A sharp and long crista petrosa occurs in Diacodexis, where it delineates the tegmen tympani fossa medially (Fig. 1c). The crista petrosa is also present in Homacodon (Figs. 1b and 4d) and Dichobune (Fig. 4c), but shorter and not as sharp as in Diacodexis. A shorter crista petrosa occurs in Gobiohyus (Fig. 4e), one that extends anteriorly to the anteromedial tuberosity of the petrosal. The crista petrosa is absent in Acotherulum (Fig. 4b) and ?Helohyus (Fig. 4f). In the studied sample, the presence of the crista petrosa co-occurs with the presence of a tegmen tympani fossa. O’Leary (2010) did not identify the crista petrosa as a particularly significant anatomical feature in the sample of Artiodactyla she described, and similarly Mead and Fordyce (2009) noted that this feature was typically absent in cetaceans. In extant artiodactyls the crista petrosa and the fossa on the tegmen tympani are variously present and have been hypothesized to coincide with a weak development of the tentorial crest (Mead and Fordyce 2009; e.g., Ruminantia, Antilocapra, illustrated by O’Leary 2010: fig. 2a). A sharp crista petrosa has been separately observed in some “condylarthrans” that may be closely related to Artiodactyla: Protungulatum (as visible on fig. 5 of O’Leary 2010: not labeled); Orthaspidotherium (Ladevèze et al. 2010: fig. 5). Subarcuate Fossa As noted above in other features, Acotherulum and Dichobune present a morphology very close to that of Diacodexis, in having a relatively deep and sharply defined subarcuate fossa. By contrast, Homacodon has a relatively small subarcuate fossa. The subarcuate fossa is completely open and shallow in ?Helohyus. With the exception of Diacodexis, the petromastoid canal (originally described by Gannon et al. 1988) is observed in all dichobunoid taxa of this study, even J Mammal Evol in species that have a very shallow subarcuate fossa (?Helohyus, Gobiohyus; Fig. 9). The presence of a petromastoid canal has been proposed to have appeared convergently several times within Artiodactyla (O’Leary 2010). Its occurrence in all dichobunoid taxa of this study except Diacodexis opens the possibility that more taxonomically comprehensive phylogenetic studies may show that this feature is primitive for Artiodactyla. Conclusions Dichobunoids are a paraphyletic assemblage of fossil taxa (Geisler et al. 2007; Spaulding et al. 2009; Gatesy et al. 2012) distributed in Eocene rocks over North America, Europe, and Asia that are either members of crown Artiodactyla or are on the stem to this clade. Our study shows that the petrosal morphology of these Eocene artiodactyls differs in several important ways from what has been observed in modern species. For the characters described here in five dichobunoid species, Diacodexis has more in common with Homacodon than with any other taxon, and ?Helohyus represents the most morphologically divergent taxon in these features. The dichobunoid species of our study share features that, to our knowledge, have not been observed in extant representatives of Artiodactyla: a lateral process of the epitympanic wing bordering the posterolateral border of the piriform fenestra, and a tegmen tympani canal. These characters are, however, observed in fossil representatives of other artiodactyl families and may be shown to be plesiomophic for Artiodactyla. Some of the species of this study (Diacodexis, Homacodon, and ?Helohyus) exhibit a dorsolateral exposure of the mastoid region of the petrosal on the temporal part of the cranium. This rare morphology has, to our knowledge, not been reported in another ungulate group. Given the current lack of resolution at the base of the artiodactyl tree, the distribution of this character must be determined in future comprehensive phylogenetic studies. Documenting morphotypes of the earliest artiodactyl species such as described here is crucial to improving the quality of hypotheses of homology between structures of morphologically distant taxa, and helps explain the morphology of extant taxa. Some or all of these species have been hypothesized to be basal to crown Artiodactyla, and, as such, integrating their anatomy into taxonomically broad phylogenetic studies will be essential for characterizing the basal morphotype of Artiodactyla. Acknowledgement We thank R. O’Leary (AMNH), Meng Jin (AMNH), C. Argot (MNHN), and B. Marandat (UM2) for access to the collections, J. Thostenson, R. Rudolph, and M. Hill for acquisition of raw CT data at the AMNH, and A-L Charruault and R. Lebrun for acquisition of raw CT scan data at the UM2. We are grateful to J Wible, J Geisler, and one anonymous reviewer for their enriching comments on earlier versions of the manuscript. We also thank the American Museum of Natural History for use of the high resolution CT-scanner (NSF MR1–R2 0959384 to N. Landman, D. Ebel, and D. Frost), and the Montpellier Rio Imaging platform (Montpellier, France) for use of their Skyscan machine. This is ISE-M publication 201X-XX. This research was supported by the ANR funding project Palasiafrica, headed by L. Marivaux. Appendix 1 Table 2 Museum accession numbers and scanning parameters of the specimens included in this study Family Taxon Specimen number CT scan institution Specimen side Voxel size (μm) Number of slices File size Dichobunidae Homacodontidae Helohyidae ?Helohyidae Diacodexeidae Cebochoeridae Dichobune leporina Homacodon vagans ?Helohyus plicodon Gobiohyus orientalis Diacodexis ilicis Acotherulum saturninum MNHN Qu 16586 USNM 482369 USNM 13079 USNM 26277 AMNH VP 16141 MNHN Qu 16366 UM2 AMNH AMNH AMNH AMNH UM2 left right right right left left 75.2 48.16 61.66 62.57 34.39 25.28 1104 994 981 988 1751 984 621×540 765×563 711×702 764×702 849×678 984×984 Appendix 2 Table 3 Alphabetical list of anatomical terms used and illustration of these terms in the different figures Term Sources other than Giannini et al. 2006 Anterior process of tegmen Luo and Gingerich (1999) tympani Anteromedial tuberosity of New term (= “knob” of O’Leary 2010) the petrosal Basicapsular groove Presley (1979) Ventrolateral Dorsomedial Anterior Ventromedial Fig.1 Fig.10 Fig. S14 Figs. 2, 6, 8 Figs. 3, 7, 9 Fig.4 Fig. 5 x x x x J Mammal Evol Table 3 (continued) Term Sources other than Giannini et al. 2006 Ventrolateral Dorsomedial Anterior Ventromedial Fig.1 Fig.10 Fig. S14 Figs. 2, 6, 8 Figs. 3, 7, 9 Fig.4 Fig. 5 Basioccipital Caudal tympanic process of petrosal Cochlear aqueduct Crista interfenestralis Crista parotica Crista petrosa Crista transversa Epitympanic recess Epitympanic wing External acoustic meatus Facial nerve Facial sulcus Fenestra cochleae Fenestra vestibuli Foramen acusticum inferius Foramen acusticum superius Fossa for the head of the malleus Fossa for stapedius muscle Fossa for the tensor tympani muscle Greater petrosal nerve Hiatus Fallopii Inferior (= ventral) petrosal sinus Internal acoustic meatus Internal carotid artery Lateral process of the epitympanic wing Mastoid plate Mastoid region Orbitotemporal canal Petromastoid canal Petrosal Pars canalicularis Pars cochlearis Posteromedial flange Prefacial commissure Prefacial commissure fossa Giannini et al. (2006) Giannini et al. (2006) x Processus muscularis Promontorium Ramus superior of the stapedial artery Secondary facial foramen Semicircular canals Stapedial artery Stapedial foramen Stylomastoid notch Stylomastoid foramen Sisson (1911) Wible et al. 1995 Giannini et al. (2006) Wible et al. (1995) Giannini et al. (2006) Giannini et al. (2006) MacPhee (1981) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) x x x x x x x x Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) x Giannini et al. (2006) NAV Giannini et al. (2006) Rose and Emry (1993) Giannini et al. (2006) NAV, foramen stylomastoideum x x x x x Giannini et al. (2006) Giannini et al. (2006) NAV Giannini et al. (2006) Giannini et al. (2006) x x x x x O’Leary (2010) McDowell (1958); MacPhee (1981) Wible and Gaudin (2004) Gannon et al. (1988) Giannini et al. (2006) Giannini et al. (2006) Giannini et al. (2006) O’Leary (2010) Giannini et al. (2006) O’Leary (2010) x x x Luo and Gingerich (1999) Giannini et al. (2006) Giannini et al. (2006) New term x x x x x x x x x x x x x x x x x x x x x x x x x x x J Mammal Evol Table 3 (continued) Term Sources other than Giannini et al. 2006 Subarcuate fossa Sulcus for stapedial artery Tegmen tympani Tegmen tympani canal Tegmen tympani fossa Tensor tympani muscle Transpromontorial sulcus Tympanic bulla Giannini et al. (2006) Novacek (1986) Giannini et al. (2006) New term New term Giannini et al. (2006) Wible (1986) NAV, bulla tympanica Tympanic nerve NAV, nervus tympanicus; Jacobsen’s nerve (Wilkie 1936) NAV, plexus tympanicus Wible (2003) Luo and Gingerich (1999) Luo and Gingerich (1999) Giannini et al. (2006) Tympanic plexus Tympanohyal Vascular groove Ventrolateral tuberosity Vestibular aqueduct Ventrolateral Dorsomedial Anterior Ventromedial Fig.1 Fig.10 Fig. S14 Figs. 2, 6, 8 Figs. 3, 7, 9 Fig.4 Fig. 5 x x x x x x x x x x x x x x x x x Terminology generally follows the Nomina Anatomica Veterinaria as applied by Giannini et al. (2006) and Wible (2003), with additions from O’Leary (2010) and Luo and Gingerich (1999) who listed standardized cranial anatomical terms in English for many Latin Nomina Anatomica Veterinaria (NAV) terms and synonymized many terms. Online figures can be found on MorphoBank, Project 946 References Asher RJ, McKenna MC, Emry RJ, Tabrum AR, Kron DG (2002) Morphology and relationships of Apternodus and other extinct zalambdodont placental mammals. Bull Am Mus Nat Hist 273:1117 Asher RJ, Bennett N, Lehmann T (2009) The new framework for understanding placental mammal evolution. BioEssays 31: 853-864 BioChroM’97 (1997) Synthèses et tableaux de corrélations. In: Aguilar J-P, Legendre S, Michaux J (eds) Actes du Congrès BiochroM’97. Mémoires et Travaux de l’Institut Pratique des Hautes-Études, Institut de Montpellier 21, Montpellier, pp. 769-805 Bryant HN, Russell AP (1992) The role of phylogenetic analysis in the inference of unpreserved attributes of extinct taxa. Phil Trans R Soc Lond B Biol Sci 337:405-418 Cifelli RL (1982) The petrosal structure of Hyopsodus with respect to that of some other ungulates, and its phylogenetic implications. J Paleontol 56:795-805 Coombs MC, Coombs WP (1982) Anatomy of the ear region of four Eocene artiodactyls, Gobiohyus, ?Helohyus, Diacodexis and Homacodon. J Vertebr Paleontol 2:219-236 De Queiroz K (2007) Toward an integrated system of clade names. Syst Biol 56:956-974 Dechaseaux C (1969) Moulages endocrâniens d’artiodactyles primitifs, essai sur l’histoire du néopallium. Ann Paléontol (Vertébr) 55:195-248 Dechaseaux C (1974) Artiodactyles primitifs des phosphorites du Quercy. Ann Paléontol (Vertébr) 60:59-100 Gannon PJ, Eden AR, Laitman JT (1988) The subarcuate fossa and cerebellum of extant primates: comparative study of a skull-brain interface. Am J Phys Anthropol 77:143-164 Gatesy J, Geisler JH, Chang J, Buell C, Berta A, Meredith RW (2012) A phylogenetic blueprint for a modern whale. Mol Phylogen Evol 66(2):479-506 Gazin LC (1965) A study of the early Tertiary condylarthran mammal Meniscotherium. Smithsonian Misc Collect 149(4-2):1-98 Geisler JH, Luo ZX (1998) Relationships of Cetacea to terrestrial ungulates and the evolution of cranial vasculature in Cete. In: Thewissen JGM (ed) The Emergence of Whales, Evolutionary Patterns in the Origin of Cetacea. Plenum Press, New York, pp 163- 212 Geisler JH, Theodor JM (2009) Hippopotamus and whale phylogeny. Nature 458(7236):E1-4 Geisler JH, Theodor JM, Uhen MD, Foss SE (2007) Phylogenetic relationships of Cetaceans to Terrestrial Artiodactyls. In: Prothero DR, Foss S (eds) The Evolution of Artiodactyla. John Hopkins University Press, Baltimore, pp 319-31 Giannini NP, Wible JR, Simmons NB (2006) On the cranial osteology of Chiroptera. I. Pteropus (Megachiroptera, Pteropodidae). Bull Am Mus Nat Hist 295:1-134 Gingerich PD (1989) New earliest Wasatchian mammalian fauna from Eocene of northwestern Wyoming, composition and diversity in a rarely sampled highfloodplain assemblage. Univ Mich Pap Paleontol 28:1-97 Hunt RM Jr (1974) The auditory bulla in Carnivora: an anatomical basis for reappraisal of carnivore evolution. J Morphol 143:21-76 Hunt RM Jr (2001) Basicranial anatomy of the living linsangs Prionodon and Poiana (Mammalia, Carnivora, Viverridae), with comments on the early evolution of aeluroid carnivorans. Am Mus Novitates 3330:1-24 Ladevèze S (2007) Petrosal bones of metatherian mammals from the late Paleocene of Itaboraí (Brazil), and a cladistic analysis of petrosal features in metatherians. Zool J Linn Soc 150:85-115 Ladevèze S, Missiaen P, Smith T (2010) First skull of Orthaspidotherium edwardsi (Mammalia, “Condylarthra”) from the late Paleocene of Berru (France) and phylogenetic affinities of the enigmatic European family Pleuraspidotheriidae. J Vertebr Paleontol 30: 1559-1578 Luo Z, Gingerich PD (1999) Terrestrial Mesonychia to aquatic Cetacea: transformation of the basicranium and evolution of hearing in whales. Univ Mich Pap Paleontol 31:1-98 J Mammal Evol MacIntyre GT (1972) The trisulcate petrosal pattern of mammals. In: Dobzhansky T, Hecht M, Steere WC (eds) Evolutionary Biology, Vol. 6. Appleton- Century-Crofts, New York, pp 275–303 MacPhee RDE (1981) Auditory regions of primates and eutherian insectivores: morphology, ontogeny, and character analysis. Contrib Primatol 18:1-282 McDowell SB Jr (1958) The Greater Antillean insectivores. Bull Am Mus Nat Hist 115:117-214 McKenna MC, Bell SK (1997) Classification of Mammals Above the Species Level. Columbia University Press, New York Mead JG, Fordyce RE (2009) The therian skull: a lexicon with emphasis on the odontocetes. Smithsonian Contrib Zool 627:1-248 O’Leary MA (2010). An anatomical and phylogenetic study of the osteology of the petrosal of extant and extinct artiodactylans (Mammalia) and relatives. Bull Am Mus Nat Hist 335:1-206 O’Leary MA, Bloch JI, Flynn JJ, Gaudin TJ, Giallombardo A, Giannini NP et al. (2013) The placental mammal ancestor and the Post-K-Pg radiation of placental. Science 339:662-667 O’Leary MA, Gatesy J (2008) Impact of increased character sampling on the phylogeny of Cetartiodactyla (Mammalia): combined analysis including fossils. Cladistics 24:397-442 Orliac MJ (2012) Osteology of the petrosal bone of Suoidea (Artiodactyla, Mammalia). J Syst Palaeontol 11(8):925-945 Orliac MJ, Ducrocq S (2011) Eocene raoellids (Mammalia, Cetartiodactyla) outside the Indian Subcontinent, palaeogeographical implications. Geol Mag 149:80-92 Presley R (1979) The primitive course of the internal carotid artery in mammals. Acta Anat 103:238-244 Rose KD (2006) The Beginning of the Age of Mammals. The John Hopkins University Press, Baltimore Russell DE, Thewissen JGM, Sigogneau-Russel D (1983) A new dichobunoid artiodactyl (Mammalia) from the Eocene of NorthWest Pakistan – II Cranial osteology. Proc K Ned Akad B Phys 3: 285-300 Sisson S (1911) A text-book of veterinary anatomy. WB Saunders, Philadelphia Spaulding M, O’Leary MA, Gatesy J (2009) Relationships of Cetacea (Artiodactyla) among mammals: increased taxon sampling alters interpretations of key fossils and character evolution. PloS ONE 4(9):1-14 Theodor JM (2010) Micro-computed tomographic scanning of the ear region of Cainotherium: character analysis and implications. J Vertebr Paleontol 30(1):236-243 Theodor JM, Erfurt J, Métais G (2007) The earliest artiodactyls. In: Prothero DR, Foss S (eds) The Evolution of Artiodactyla. John Hopkins University Press, Baltimore, pp 32-58 Thewissen JGM, Cooper LN, Clementz MT, Bajpai S, Tiwari BN (2007) Whales originated from aquatic artiodactyls in the Eocene epoch of India. Nature 6343(450):1190-1195 Thewissen JGM, Cooper LN, George JC, Bajpai S (2009) From land to water: the origin of whales, dolphins, and porpoises. Evo Edu Outreach 2:272-288 Wible JR (1986) Transformations in the extracranial course of the internal carotid artery in mammalian phylogeny. J Vertebr Paleontol 6:313325 Wible JR (2003) On the cranial osteology of the short-tailed opposum Monodelphis brevicaudata (Didelphidae, Marsupialia). Ann Carnegie Mus 72:137-202 Wible JR, Gaudin TJ (2004) On the cranial osteology of the yellow armadillo Euphractus sexcinctus (Dasypodidae, Xenarthra, Placentalia). Ann Carnegie Mus 73:117-196 Wible JR, Rougier GW, Novacek MJ, McKenna MC, Dashzeveg D (1995) A mammalian petrosal from the Early Cretaceous of Mongolia: implications for the evolution of the ear region and mammliamorph interrelationships. Am Mus Novitates 3149:1-19 Wible JR, Rougier GW, Novacek MJ, McKenna (2001) Earliest eutherian ear region: a petrosal referred to Prokennalestes from the Early Cretaceous of Mongolia. Am Mus Novitates 3322:1-44 Witmer LM (1995) The Extant Phylogenetic Bracket and the importance of reconstructing soft tissues in fossils. In: Thomason JJ (ed) Functional Morphology in Vertebrate Paleontology. Cambridge University Press, New York, pp 19-33