Chilcacetus cavirhinus, Lambert & Muizon & Bianucci, 2015

Lambert, Olivier, Muizon, Christian De & Bianucci, Giovanni, 2015, A new archaic homodont toothed cetacean (Mammalia, Cetacea, Odontoceti) from the early Miocene of Peru, Geodiversitas 37 (1), pp. 79-108 : 81-96

Chilcacetus cavirhinus n. sp.

( Figs 1-12 ; Tables 1, 2)

HOLOTYPE. — MNHN.F.PRU11, a nearly complete skull only lacking the anterior part of the rostrum and the ventral-most region of the basicranium, with the associated right mandible (anterior end missing), and the manubrium (or presternum). Ŋe right periotic and tympanic bulla, as well as the left periotic were originally present; unfortunately they were lost at an early stage of the preparation of the skull and mandible (see comment above).

REFERRED SPECIMEN. — MUSM 1401, a nearly complete skull, only lacking fragments of the supraorbital processes and the ventral-most part of the basicranium (including the earbones),and with most of the ventral surface heavily abraded, with associated fragments of both mandibles, eight partly preserved vertebrae, and rib fragments.Locality of Ullujaya (see below). Approximate geographic coordinates: 14°34’S- 75°38’W.

ETYMOLOGY. — From “cavi”, hole, cavity, and “rhinus”, nose in Latin. A reference to the excavation between nasals and mesethmoid along the nasal tracts observed in the new species.

TYPE LOCALITY. — Ŋe holotype was collected in 1988 close to Ullujaya, Pisco Basin, Peru, a few kilometers south of where the referred specimen MUSM 1401 was recovered. Approximate geographic coordinates: 14°37’S- 75°38’20’’W.

TYPE HORIZON AND AGE. — Chilcatay Formation, latest Oligocene to early Miocene ( Dunbar et al. 1990; DeVries 1998, 2001). Although the precise locality of the holotype is unknown, the referred specimen MUSM 1401 originates from within the Chilcatay Formation, several meters under the Chilcatay Formation - Pisco Formation disconformity, at Ullujaya, a few kilometers north to the type locality. Based on a published section a short distance from Ullujaya, the Chilcatay Formation layers of this locality (which preserves the fossilized remains of other marine vertebrates including mysticetes, kentriodontids, the squalodelphinids Huaridelphis raimondii Lambert, Bianucci & Urbina, 2014 and Notocetus vanbenedeni Moreno, 1892 , and a stem physeteroid) are tentatively dated to the early Miocene ( Alván De la Cruz 2008; Lambert et al. 2014; Bianucci et al. 2015).

D IAGNOSIS. — Chilcacetus cavirhinus n. gen., n. sp. is a longsnouted homodont odontocete (ratio between rostrum length and condylobasal length> 0.70) differing from all other odontocetes in the presence of a cavity between nasals and mesethmoid on the posterior wall of the bony nares. It differs from members of other long-snouted homodont extinct families ( Allodelphinidae , Eoplatanistidae , Eurhinodelphinidae , and Platanistidae ) in the absence of a deep lateral groove along the rostrum and in the absence of ankylosis for the symphysis of the mandibles.

It further differs from Eurhinodelphinidae in lacking an extended edentulous anterior premaxillary portion of the rostrum, in the nasals partly overhanging the bony nares, in the more anteriorly elongated zygomatic process of the squamosal (ratio between the height of the process and the length of the process ≥ 1.10), in the cranium distinctly longer than wide, in the lesser transverse widening of the supraoccipital shield (ratio between the maximum width of the supraoccipital at the lateral corners of the nuchal crest and the postorbital width <0.70), in the posterior margin of the postorbital process being vertical, in the top of the temporal fossa being nearly as high as the nuchal crest, in the palatines being separated anteromedially for a long distance at rostrum base, and in the longer and more laterally directed posterior process of the periotic. It differs from Eoplatanistidae in the premaxillary foramen roughly at the level of the antorbital notch, in the thinner and flatter antorbital process, in the acute anterior margin of the nasal partly overhanging the bony nares, in the transversely concave and less anteriorly projected anterodorsal portion of the supraoccipital shield, in the deep anterior bullar facet of the periotic, in the elongated posterior process of the periotic, and in bearing a conspicuous median furrow on the tympanic bulla. It differs from Argyrocetus patagonicus in the reduced widening of the premaxillae at the rostrum base, in lacking a wide dorsal opening of the mesorostral groove, in the premaxillary foramen roughly at the level of the antorbital notch, in the angle formed by the basioccipital crests in ventral view <50°, in the top of the temporal fossa being nearly as high as the nuchal crest, and in the absence of ankylosis for the symphysis of the mandibles. It differs from ‘ Argyrocetus ’ bakersfieldensis in lacking a deep lateral groove on the rostrum, in lacking a deep sulcus anterior to the main dorsal infraorbital foramen at rostrum base, and in the palatines being separated anteromedially for a long distance at rostrum base. It differs from ‘ Argyrocetus ’ joaquinensis in the dorsal opening of the mesorostral groove anterior to the rostrum base narrower than the premaxilla, in the presence of more than one dorsal infraorbital foramen at rostrum base, in lacking a deep sulcus anterior to the main dorsal infraorbital foramen at rostrum base, in the proportionally shorter and wider nasal, in the nasal partly overhanging the bony nares, in the wide exposure of the frontal on the vertex, in the posterior margin of the postorbital process being vertical, in the palatines separated anteromedially for a longer distance at rostrum base, in the significantly shorter hamular process of the pterygoid, and in the ventral margin of the postglenoid process of the squamosal being approximately at the same level as the ventral margin of the exoccipital in lateral view. It differs from Macrodelphinus in its smaller size, in the premaxillary portion of the rostrum making less than 10 per cent of its total

A

length, in the premaxillary foramen roughly at the level of the antorbital notch, in the nasal proportionally longer compared to the frontal on the vertex, in the palatines separated anteromedially for a long distance at rostrum base. It differs from Papahu Aguirre- Fernández & Fordyce, 2014 in the rostrum being proportionally dorsoventrally thicker in its proximal part, in the single premaxillary foramen roughly at the level of the antorbital notch, in the dorsal exposure of the premaxilla wider than the exposure of the maxilla at rostrum base, in the proportionally wider ascending process of the premaxilla, in the anterodorsal elevation of the dorsal surface of the nasal, in the posterolateral projection of the nasal, in the elongate postorbital process of the frontal, in the long and deeper anterior bulla facet of the periotic, in the posterior elevation of the dorsal margin of the mandible being progressive.

It differs from Platanistidae and Squalodelphinidae in that its cranium is as long as wide, in lacking a deep medial depression in the dorsal surface of the premaxillae at rostrum base, in the posterior dorsal infraorbital foramen being lateral to the lateral margin of the premaxillary sac fossa, in the nasal rising anterodorsally and partly overhanging the bony nares, in the strong reduction of the lateral lamina of the pterygoid, in the absence of an extension of the pterygoid sinus in the orbit roof; in the zygomatic process of the squamosal not being dorsoventrally inflated, in the proportionally longer posterior process of the periotic, in lacking an individualized anterior spine on the tympanic, and in the lower involucrum of the tympanic in medial view, cut by an indentation at mid-length. It further differs from Platanistidae in lacking a dorsal crest on the antorbital process. It differs from Allodelphinidae in lacking a deep medial depression in the dorsal surface of the premaxillae at rostrum base, in the cranium as long as wide, in the proportionally wider vertex, in the strong reduction of the lateral lamina of the pterygoid, in the anteriorly long zygomatic process of the squamosal, and in the top of the temporal fossa being nearly as high as the nuchal crest.

A

DESCRIPTION

Skull

Ŋe holotype MNHN.F.PRU11 ( Figs 1-10 ) and MUSM 1401 ( Figs 11 ; 12 ) have a medium size skull, with a postorbital width of 250 mm in the holotype and a width across the occipital condyles of 95 and 93 mm respectively (see Table 1). Except for the vertex and the supraoccipital shield (see below), measurements in MUSM 1401 are generally slightly greater than in the holotype. Ŋe sutures between the well-ossified cranial bones are closed, with no indication of movement for individual bones, and the maxillary alveoli are ossified, suggesting adult individuals. Ŋe complete rostrum of MUSM 1401 is long, comprising 72% of the condylobasal length. Furthermore, the width of the base of the rostrum is slightly greater than in the holotype ( Table 1). On the anterior part of the rostrum, transverse sections have a semi-circular outline, with the dorsal surface regularly convex ( Fig. 11G ). Based on the outline of the section of the premaxillae and maxillae at

A

the preserved apex of the holotype, it is likely that its rostrum was originally similarly long. Ŋe brain cavity is somewhat dorsoventrally crushed in both specimens, as indicated by horizontal cracks in the supraoccipital shield and the medial wall of the temporal fossae. Ŋe cranium is nearly as wide as long, with a weakly elevated vertex (ratio between vertical distance from dorsal margin of rostrum to top of vertex and width of the premaxillary sac fossae <0.70). Ŋe lateral edge of the frontal, posterior to the postorbital process is rectilinear and oblique. Dorsally, it does not fully cover the long and high, drop-shaped temporal fossa and consequently, most of the zygomatic process of the squamosal is exposed in dorsal view. Ŋis condition results in the facial region of Chilcacetus n. gen. being roughly trapezoidal (rather than semicircular or subrectangular), which is observed in several ancient odontocetes (e.g., Papahu , Prosqualodon Lydekker, 1893 , Squalodon Grateloup, 1840 , Squaloziphius Muizon, 1991 , and Waipatia Fordyce, 1994 ), but not in eurhinodelphinids.

Premaxilla. As mentioned above, the anterior transverse section of the rostrum is nearly semicircular. Ŋe transversely convex premaxillae contact each other dorsomedially in the holotype, whereas the mesorostral groove is not completely roofed dorsally in MUSM 1401, retaining a dorsomedial opening along its whole extent. Ŋe medial contact in the holotype was likely not sutural, as indicated by a minor overlap of the right premaxilla by the left. In MUSM 1401, a longitudinal sulcus marks the dorsal surface of the distal portion of each premaxilla, for about 150 mm. Ŋe premaxilla-maxilla suture is discernible on the lateral surface of the rostrum of MUSM 1401 until a level 130 mm from the apex ( Fig. 11B ). Considering the eroded ventral surface of the maxillae, the distal premaxillary portion of the rostrum likely did not exceed 100 mm. Furthermore alveoli are present in the alveolar groove to the end of the rostrum ( Fig. 11C ). Ŋis condition, also observed in the holotype of Macrodelphinus kelloggi Wilson, 1935 (at least 7 distinct alveoli in the premaxillary portion of the rostrum), is a major difference with eurhinodelphinids, characterized by a long premaxillary portion of the rostrum lacking alveoli ( Muizon 1991; Lambert 2004, 2005a,b). Ŋe number of alveoli in the premaxilla of MUSM 1401 cannot be determined precisely, due to the obscured end of the premaxilla-maxilla suture. Ŋe orientation of these anterior alveoli indicates that the anterior-most teeth were not procumbent in Chilcacetus cavirhinus n. gen., n. sp. In the holotype, the dorsolateral surface of the premaxilla is covered with shallow convoluted sulci for the distal 150 mm anteriorly; these sulci course anterodorsally from the maxillapremaxilla suture ( Fig. 1 ). As in Macrodelphinus , this suture is not margined by a lateral groove in either specimen, differing from allodelphinids, eoplatanistids, eurhinodelphinids, and platanistids. Only a few foramina and short sulci are preserved along the premaxilla-maxilla suture of MUSM 1401. In lateral view, this suture raises distinctly backwards until about the midpoint in the length of the rostrum. Consequently, the lateral exposure of the premaxilla is reduced at this level, in association with its narrower dorsal exposure. From this midpoint, the progressively diverging premaxillae flatten and widen towards the base of the rostrum.

A slight concavity of the lateral margin of each premaxilla is observed a few centimeters anterior to the premaxillary foramen in both specimens. Situated at the level of the antorbital notches, the premaxillary foramen is approximately centered on the premaxilla of the holotype, whereas it is closer to the maxilla in MUSM 1401. Ŋe anterior-most premaxillary foramen is more anterior in Eoplatanista Dal Piaz, 1916 , Papahu , Squalodon and Waipatia , whereas it is more posterior in Argyrocetus patagonicus and Macrodelphinus . Ŋe narrow anteromedial sulcus margins laterally the elongated prenarial triangle, whereas the wide posterolateral sulcus extends beyond the posterior margin of the bony nares, rising high along the lateral surface of the ascending process of the premaxilla. Ŋe shallow posteromedial sulcus is barely visible. Ŋe maximum opening of the mesorostral groove is in front of the level of the premaxillary foramina in the holotype, and slightly more anterior in MUSM 1401. In A. patagonicus , the dorsal opening of the groove is much wider, a feature possibly accentuated by dorsoventral crushing of the rostrum ( Cabrera 1926). Furthermore, in the latter the premaxillary sac fossae slope ventrolaterally.Ŋe anterior limit of the bony nares of Chilcacetus cavirhinus n. gen., n. sp. is V-shaped, but poorly defined. Ŋe bony nares roughly end at the level of the middle of the orbit, where the medial edges of the premaxillae are clearly approximated and present a medial angulation. Ŋe surface of the premaxillary sac fossa is distinctly concave longitudinally. Ŋe transverse concavity is somewhat more pronounced in the holotype, with the depression extending posterodorsally beyond two thirds of the length of the bony nares. Ŋe left premaxillary sac fossa is slightly wider than the right fossa in the holotype, whereas the opposite is measured in MUSM 1401. However, these weak asymmetries (ratio between width of right and left premaxillary sac fossae 0.94 and 1.04 respectively) are not considered significant. Ŋe ascending process is longitudinally short, with an abrupt posterodorsal elevation. It is transversely wide (more so than in Papahu and Waipatia ), individualized from the posterolateral plate of the premaxilla by the posterolateral sulcus. Nevertheless, the lateral margin of the ascending process does not overhang the maxilla. Ŋe ascending process is not marked by a longitudinal premaxillary cleft, differing from Papahu , Waipatia and several platanistids and squalodelphinids ( Fordyce 1994; Lambert et al. 2014). On the vertex, the premaxilla is wide lateral to the nasal in the holotype. Our interpretation of the sutures of the vertex ( Fig. 2 ), remaining conjectural due to the presence of numerous longitudinal grooves along the nasal-frontal suture, suggests a wide contact between premaxilla and frontal. A narrower and shorter posterior part of the premaxilla on the vertex is observed in MUSM 1401; in the latter the premaxilla-frontal contact is either reduced or absent.

Maxilla. At the preserved apex of the rostrum of the holotype, the laterally facing exposure of the maxilla has roughly the same height as the premaxilla. As previously mentioned, the suture between the two bones rises posteriorly, until a level where the ventrolateral margin of the maxilla starts its elevation towards the antorbital notch. On each maxilla of the holotype, one large dorsal infraorbital foramen, wider on the right side, is located anterior to the antorbital notch. Ŋis foramen, not followed anteriorly by a sulcus, is surrounded by at least 5-6 smaller foramina on the left side and at least 2-3 on the less well-preserved right side. Similar small dorsal infraorbital foramina are detected in MUSM 1401, six on the right maxilla and five on the left. Better preserved in the holotype, the antorbital notch is deep, U-shaped, with a long anterolateral projection of the maxilla on the antorbital process ( Fig. 1 ), as also seen in Macrodelphinus . Ŋe right projection was likely broken and artificially slightly shifted forwards. Posterior and posterolateral to the notch, the dorsal surface of the maxilla is only slightly elevated, without a distinct maxillary crest, and remaining lower than the premaxilla. A dorsal infraorbital foramen is located close to the maxilla-premaxilla suture at the level of the posterior margin of the bony nares, posterior to the level of the postorbital process of the frontal. Above the temporal fossa the lateral margin of the maxilla is directed posteromedially, nearly reaching the supraoccipital at a level posterior to the anterior-most level of the nuchal crest. Ŋe medial wall of the maxilla along the vertex is more erect on the left side, whereas it displays a more pronounced transverse convexity on the right side.

On the anterior margin of the bony nares, each maxilla sends a triangular ventromedial projection towards the mesethmoid, closing anterolaterally the bony nares at a level dorsoventrally lower than the medial margin of the premaxilla.

From the apex of the complete rostrum of MUSM 1401, a total tooth count of 41 is estimated on the right side and more than 38 on the left (as mentioned above, an uncertain number of the anterior alveoli originate in the premaxilla) ( Fig. 11C ). All alveoli correspond to single-rooted teeth; their transverse diameters range from 6 to 8 mm and alveoli are separated by transverse septa of 3-4 mm. In the holotype, 21-22 alveoli for single-rooted teeth are counted on 276 mm of the incomplete right maxillary alveolar groove, whereas 20 alveoli are counted along 239 mm of the left groove. Ŋe transverse diameter of the roughly circular alveoli ranges from 6 to 7.5 mm (except for the smaller posterior alveoli), with transverse septa 5-7 mm long. Ŋe last posterior alveolus is closer to the antorbital notch in the holotype than in MUSM 1401 (see Table 1). Small, circular to oval depressions posterolateral to each alveolus along the anterior half of the rostrum of the holotype (best seen on the left side) are interpreted as embrasure pits created by the apex of opposing teeth.

More completely preserved in the holotype, the vomerine trough is deep and wide, limited laterally by shallow sulci originating and passing forwards from two foramina located at about half the preserved length of the rostrum ( Fig. 5C ). At least two additional pairs of foramina are visible, the posteriormost being located at the anterior tip of the palatine-maxilla suture and corresponding to the major palatine foramina.

Between the palatine and the ventral infraorbital foramen, the ventral surface of the maxilla (and possibly the lateral-most part of the palatine) is crossed by an oblique crest, smoothly recurved anterolaterally in the rostrum base ( Figs 6 ; 7 ). A similar crest is seen in various odontocete groups, including eurhinodelphinids (see Lambert 2005a for Eurhinodelphis cocheteuxi du Bus, 1867 ). Ŋis crest is interpreted as the anterior limit of an area of origin for the internal or external pterygoid muscles (two of the adductors of the mandible; see Fraser & Purves 1960; Seagars 1982). Seagars (1982) found that the internal pterygoid muscle is the most powerful muscle for the closure of the jaws in several delphinids, including longsnouted forms with numerous teeth.

Ŋe ventral infraorbital foramen is elongated, oblique to the long axis of the skull. 25 mm posterior to its posterior end, close to the orbitosphenoid, a small foramen, followed anteriorly by a sulcus towards the infraorbital foramen, might correspond to the sphenopalatine foramen ( Fig. 6 ).

Mesethmoid. Between the triangular ventromedial projections of the maxillae on the anterior part of the bony nares of the holotype, the mesethmoid is visible as a low 30 mm-long element with an irregular dorsal surface ( Figs 2 ; 3). Ŋis region is separated from the keel of the nasal septum by an abrupt step; the empty space between the premaxillary sac fossae was likely originally occupied by cartilage. Completely hidden in dorsal view, due to the overhanging nasals, the posterior plate of the mesethmoid is excavated in both specimens by a pair of large crescent-like olfactory openings ( Figs 3; 11D ), as observed in many platanistoids ( Hoch 2000; Godfrey 2013; Godfrey et al. 2013; Lambert et al. 2014), at least some eurhinodelphinids (e.g. Schizodelphis morckhoviensis du Bus, 1872 ; Lambert 2004), a specimen referred to ‘ Argyrocetus ’ bakersfieldensis (YPM 13410; Wilson 1935), Macrodelphinus (OL pers. obs.), and Waipatia ( Fordyce 1994) . A relatively large gap separates the upper margin of the mesethmoid plate from the smooth ventral surface of the nasals, forming a horizontal and transversely wide cavity on the posterior wall of the bony nares. Such a cavity is absent in ‘ Argyrocetus ’ joaquinensis and Macrodelphinus , and to our knowledge has not been described in any other odontocete. Ŋe dorsal surface of the mesethmoid plate is irregular in MUSM 1401, excavated by small fossae. Ŋe texture of this surface, reminiscent of the dorsal surface of the bony nasal septum, might indicate that the cavity between mesethmoid and nasal was originally filled with cartilage in Chilcacetus cavirhinus n. gen., n. sp.

Nasal. Ŋe nasals occupy a large surface on the vertex: they are anteroposteriorly shorter medially, due to the anteromedial projection of the frontals ( Figs 2 ; 11E ). Ŋeir lateral part extends farther posteriorly, even if their exact extent is difficult to determine due to the complex organization of the nasal-frontal suture; this is different from the condition in Papahu . At least in the holotype, the right nasal is narrower than the left, with a ratio between their respective maximum widths of 0.90. Ŋe dorsal surface of each nasal rises gently anterodorsomedially, forming a convex dorsal outline in anterior view.Better preserved in MUSM 1401, the anterior margin of the joined nasals is regularly convex in dorsal view.Overhanging the bony nares, it forms an acute horizontal edge, with the ventral surface of the nasal progressively ascending forwards to the dorsal surface. A similar nasal morphology is seen in ‘ Argyrocetus ’ bakersfieldensis, ‘ A. ’ joaquinensis, and Macrodelphinus ; it probably corresponds to a remnant of the primitive condition of the posteroventrally directed bony nares in archaic neocetes and archaeocetes (see Albertocetus Uhen, 2008 and Simocetus Fordyce, 2002 ). Ŋe nasals of the holotype and only referred specimen of Argyrocetus patagonicus are now lost (GB pers. obs.), but in the original illustrations ( Lydekker 1893: pl. 5) they overhang the bony nares with an acute anterior margin, as clearly stated by the author. Ŋe nasals of Papahu and Waipatia are not as elevated anterodorsally, but the acute anterior margin similarly overhangs the bony nares ( Fordyce 1994; Aguirre-Fernández & Fordyce 2014). In some eurhinodelphinids, for example Ziphiodelpis abeli Dal Piaz in Bassani & Misuri, 1912, the nasal is higher anteriorly, but it does not overhang the bony nares.

Frontal. On the vertex the frontal is slightly lower than the nasal, even if the presumed anteromedial corners of the joined frontals form a prominent region in the holotype. Based on our interpretation of the sutures of the holotype ( Fig. 2 ), each frontal is wider on the vertex than the corresponding nasal. In MUSM 1401, displaying a significantly narrower vertex as demonstrated by the shorter distance between maxillae across the vertex, the nasal is slightly wider than the corresponding frontal. Ŋe anteroposterior extent of the frontal on the vertex is also shorter in MUSM 1401. Ŋe right frontal of the holotype is narrower than the left, with a ratio between the respective widths at mid-length of 0.84. In the latter, the anterior part of the interfrontal suture and the internasal suture are shifted slightly to the left side as compared to the posterior end of the interfrontal suture. Posteriorly, the frontal is lower than the marked nuchal crest, comprised of the anteromedial margin of the supraoccipital.

Ŋe antorbital process of the frontal of the holotype is robust, triangular in lateral view ( Figs 4 ; 5A ), and partly exposed lateral to the maxilla in dorsal view. In lateral view, the medium size orbit displays a strongly concave outline, in relation with the ventrally long postorbital process (vertical length from the contact with the maxilla = 37 mm, proportionally longer than in Papahu ). Ŋe upper part of the lateral surface of the process is excavated by a triangular depression. Ŋe posterior margin of the process is roughly vertical. Ŋe contact with the zygomatic process of the squamosal, seen on the right side, is exaggerated to some degree by dorsoventral crushing of the cerebral cavity.

On the roof of the orbit, bordered posteriorly by the high medial portion of the infratemporal crest (= postorbital ridge), the frontal groove is directed posteromedially and slightly ven - trally ( Figs 5C ; 6 ). In eurhinodelphinids, the groove is more transversely and ventrally directed. No trace could be found to indicate invasion of the orbit or temporal region of the frontal by the pterygoid sinus (see below for the alisphenoid and pterygoid limits of the sinus).

Supraoccipital. In dorsal view, the lateral extent of the supraoccipital shield is similar to the condition in ‘ Argyrocetus ’ bakersfieldensis and Eoplatanista , proportionally greater than in Squalodon and Waipatia , but less than in eurhinodelphinids. Ŋe outline of the acute nuchal crest is regularly convex anteriorly in dorsal view, limiting a transversely and longitudinally concave dorsal part of the supraoccipital shield. As in Macrodelphinus , the crest becomes thicker in the dorsolateral region where the surface of the supraoccipital shield is more uneven ( Figs 1 ; 5D ; 11A, F ), likely for the insertion of neck muscles (possibly m. semispinalis or m. rhomboideus sensu Schulte & Smith 1918). From there, in the holotype the prominent descending temporal crest is directed posteriorly and slightly posteromedially, with the lateralmost surface of the supraoccipital facing posteromedially. In MUSM 1401, the temporal crest projects farther posteriorly (best seen in dorsal view; Fig. 11A ). In posterior view the minimum distance between the temporal fossae across the supraoccipital shield is at mid-height of the temporal fossa; this distance is

A

greater in the holotype, which has a wider supraoccipital shield ( Figs 5D ; 11F ). Ŋis difference between the two specimens may be partly explained by the degrees to which the skulls were crushed dorsoventrally.

Vomer. Ŋe mesorostral groove is partly filled with indurated sediment in both specimens, hiding the dorsal exposure of the

vomer. In the deep vomerine trough, a keeled narrow stripe of the vomer is visible in ventral view. Ŋe vomer forms the medial and posterior walls of the choanae, reaching posteriorly to a level in line with the carotid foramen.

Palatine. Ŋe anterior tip of the palatine is 33 mm anterior to the antorbital notch in the holotype. From that level and for 40 mm backwards, left and right palatines are separated medially by a narrow exposure of the maxillae. Ŋe palatinemaxilla suture is more difficult to follow laterally; an interpretation based on the two sides is provided in Figure 7 , with a posterolaterally and laterally directed suture. Ŋe palatine is exposed in the anterior portion of the pterygoid sinus fossa, in the part of the medial wall not covered by the pterygoid ( Figs 6 ; 7 ), a condition also seen in Squaloziphius and several extant ziphiids ( Lambert et al. 2013). Ŋe palatine part of the palate is distinctly keeled medially, contrasting with the lower palate of Squalodon , Squaloziphius , and Waipatia , possibly related to the elongated snout, as seen in Eoplatanista , eurhinodelphinids, and platanistids.

Pterygoid. Ŋe anterior end of the pterygoid is just posterior to the level of the antorbital notch; consequently the pterygoid sinus fossa, fully covered by the pterygoid at its anterior end, does not extend farther forward than the notch. Ŋe anteromedial margin of the pterygoid is prominent compared to the adjacent palatine; it becomes transversely thicker posteromedially, forming a wide shelf, triangular in ventral view, ventrally overhanging the anterior part of the pterygoid sinus fossa ( Figs 5C ; 6 ; 7 ). Following a transverse constriction, this medial wall of the pterygoid widens considerably for a short distance, until the anterior margin of the ventral choanae. From there, the bone tapers to form a short and robust hamular process, less than 20 mm long from the anterior limit of the choanae, along the sagittal plane of the skull. Similarly joined thick hamular processes, not excavated by the pterygoid sinus fossa, are observed, even if longer, in ‘ Argyrocetus ’ joaquinensis, Simocetus , and Squaloziphius ( Kellogg 1932; Muizon 1991; Fordyce 2002). Ŋe medial lamina of the pterygoid forms the lateral wall of the choana. Posterior to the eustachian notch, it overlaps the basioccipital crest to the level of the carotid foramen. Ŋe lateral lamina of the pterygoid is preserved roughly to the same extent on both sides of the holotype, partly limiting laterally the transversely narrow pterygoid sinus fossa for only half the length of the fossa anterior to the choana. More posteriorly, the lateral lamina becomes a low and thin ridge along the subtemporal crest, joining the lateral margin of the alisphenoid ( Fig. 7 ). Ŋerefore, we consider that Chilcacetus cavirhinus n. gen., n. sp. did not possess a complete lateral lamina of the pterygoid contacting the falciform process of the squamosal, contrary to platanistids, squalodelphinids, and at least some eurhinodelphinids (e.g., Schizodelphis Gervais, 1861 and Xiphiacetus Lambert, 2005 ); a similar condition is proposed for Waipatia ( Fordyce 1994) . Along the medial wall of the thin ridge along the subtemporal crest described above, the pterygoid is excavated by a few small and shallow fossae, interpreted as extensions for the pterygoid sinus reminiscent of the condition in some ziphiids (e.g., Ninoziphius Muizon, 1983 ; Lambert et al. 2013). Dorsally, a thin plate of the pterygoid covers more than half the wide subcircular and concave area of the alisphenoid anterior to the foramen ovale; the alisphenoid is exposed mainly in the posterolateral portion of this depression for the pterygoid sinus.

Jugal/lacrimal. Ŋe left jugal-maxilla suture is visible posteromedially and medially to the antorbital notch ( Fig. 6 ). Ŋe base of the styliform part of the jugal is preserved in section, forming the posteroventral wall of the antorbital notch. On the antorbital process, only an incomplete thin plate of the lacrimojugal is observed, ventrally lining the anteroventral margin of the maxilla. Based on the section of this preserved fragment and from the nature of the anteroventral surface of the frontal on the antorbital process, we suspect the lacrimojugal was originally shorter longitudinally than transversely wide, with an outline in ventral view similar, for example, to Waipatia ( Fordyce 1994) . In lateral view, the lacrimojugal was probably limited to the anteroventral side of the antorbital process of the frontal, not or only barely dorsally exposed, differing on that point from Waipatia . Ŋe posterior portion of the right styliform part of the jugal is preserved between the mandible and the zygomatic process of the squamosal; no significant thickening is observed at that level ( Fig. 7 ).

Orbitosphenoid. Ŋe orbitosphenoid-frontal suture is difficult to follow, even if a thin plate along the anterior wall of the frontal groove likely corresponds to the former. At the end of the frontal groove, two main openings are observed. Ŋe slightly larger anterior opening, the origin and continuation of the frontal groove, is interpreted as the optic foramen, whereas the posterior opening is probably the orbital fissure (sphenoidal fissure, with superior and inferior fissures not distinguished), separated from the optic foramen by the metoptic process ( Fig. 6 ). Ŋe foramen rotundum is not observed, either coalescent with the orbital fissure or obscured due to the slight dorsoventral crushing of this region.

Alisphenoid. In ventral view, the alisphenoid is exposed in the posterolateral part of the large subcircular depression (= part of the pterygoid sinus fossa) anterior to the foramen ovale, a feature seen in both specimens ( Figs 7 ; 11C ). Ŋe lateral extent of this depression is similar to the condition in Papahu and Waipatia ( Fordyce 1994; Aguirre-Fernández & Fordyce 2014). Ŋe lateral margin of the alisphenoid along the subtemporal crest bears a thin ridge that is a continuation of the ridge on the pterygoid, limiting the pterygoid sinus fossa laterally. Ŋis ridge turns posteriorly and posteromedially and forms the anterior wall of the path for the mandibular nerve V3 and of the foramen ovale. Ŋe latter is distinctly anterior to the level of the carotid foramen.Ŋe alisphenoid-squamosal suture is difficult to follow in the holotype. In MUSM 1401, the foramen ovale is separated from the posterior lacerate foramen by a strip of bone at least 18 mm wide. Ŋe greatest diameter of the posterior lacerate foramen is 20 mm.

Squamosal. Relative to the long temporal fossa, the zygomatic process of the squamosal is elongate anteriorly, more so than in eurhinodelphinids, and only moderately elevated towards the postorbital process of the frontal ( Figs 4 ; 5A ; 11B ). Its dorsal margin is longitudinally convex and transversely thick. Ŋis margin dips smoothly towards the floor of the temporal fossa, lacking the posterior elevation seen in physeteroids.

A

Ŋe floor of the temporal fossa (= squamosal fossa) is longitudinally concave.

Ŋe mandibular fossa is wide, anteroventrally and slightly medially facing. Its lateral margin forms a prominent ventrally developed crest, increasing the height of the zygomatic process in lateral view, but not to the extent of platanistids and squalodelphinids. Medially, the mandibular fossa is only poorly separated by a subtle step from the tympanosquamosal recess ( Fig. 7 ). Ŋe latter does not extend to the medial face of the zygomatic process beyond the level of the anterior limit of the floor of the temporal fossa. Preserved in the holotype, the falciform process of the squamosal is a wide and long plate, directed anteroventromedially, extending ventrally nearly to the level of the basioccipital crest at the same longitudinal level. On the better-preserved left side, there is no indication of any major break or any sutural contact surface for the lateral lamina of the pterygoid on the anterior margin of the falciform process. Both the postglenoid process and spiny process were lost on the two sides of both specimens. Based on the interpretation of the morphology of the basicranium in Waipatia ( Fordyce 1994) we identify in the holotype an anteroposteriorly long (24 mm on the right side) periotic fossa in the medial portion of the squamosal. Ŋe fossa is divided in two by an elevated oblique ridge (?supratubercular ridge sensu Fordyce 1994), which corresponds to the posteromedial continuation of the falciform process. Ŋe posterior portion of the periotic fossa, deeper and more clearly defined with a longitudinal diameter of 17 mm, contains a foramen along the oblique ridge. Ŋis fossa seems homologous to the subcircular fossa described by Muizon (1987) in Notocetus Moreno, 1892 . Ŋe shallower anterior portion of the fossa, only distinct on the right side, is pierced by one or two foramina, one of them likely corresponding to the foramen spinosum noted by Fordyce (1994) in Waipatia . Anterior to the fossa, at mid-length between the presumed foramen spinosum and foramen ovale, is a tiny foramen in a position similar to foramen 1 in Waipatia .

Exoccipital. Ŋe occipital condyles are prominent, with a distinct condylar neck. Ŋe dorsal condyloid fossa is moderately deep, possibly somewhat exaggerated by the dorsoventral crushing of the cerebral cavity in both specimens ( Figs 5D ; 11F ). Ŋe paroccipital process is lost on both sides of the holotype, as well as the ventral part of the long lateral portion of the bone. In MUSM 1401, the exoccipital is projected laterally, with an acute dorsolateral angle towards the sternomastoideus fossa.

Basioccipital. Ŋe basioccipital crests of the holotype are heavily worn ventrally. Ŋe basioccipital basin is wide anteriorly; posteriorly the crests diverge moderately to form an angle of about 40° ( Fig. 5C ). On the posterolateral region of the basin, along the base of the crest, a shallow depression is limited anteromedially by a low but distinct crest turning anterolaterally, probably corresponding to the insertion of the cervical vertebrae muscle rectus capitis ventralis (see Evans & Lahunta 2013).

Ear bones

Based on preliminary drawings and observations (see comments above), several features of the now lost periotics and tympanic bulla of the holotype can be provided.

Periotic. Ŋe total length of the right periotic is 41 mm. On the moderately long and pointed anterior process, the anterior bullar facet is a long, well-excavated fossa ( Fig. 8A ) reminiscent of the condition in eurhinodelphinids, platanistids, squalodelphinids, and ziphiids, deeper though than in Papahu , Simocetus , allodelphinids, and squalodontids. In dorsomedial view, the anterior process is somewhat transversely swollen towards its base, with a ventromedial margin more convex than in most eurhinodelphinids. Ŋe pars cochlearis is large with a nearly rectilinear anterior margin and a slightly convex medial margin in ventral view. Ŋe lateral tuberosity is elongated anterolateral to the mallear fossa, more so than in eurhinodelphinids; a long lateral tuberosity is observed in many ziphiids, but it is directed more laterally than anterolaterally. Ŋe long posterior process appears roughly rectilinear, projecting posterolateroventrally. Ŋis process is distinctly longer than in all known eoplatanistids, eurhinodelphinids, platanistids, ziphiids, and even squalodelphinids, and it is directed more laterally, more similar to the condition in the xenorophiid Albertocetus (see Uhen 2008) and probably Simocetus (obscured in the latter by the posterior process of the tympanic; Fordyce 2002). On the dorsolateral margin of the bone, posterior to the epitympanic recess, is a low tubercle ( Fig. 9B ) that might correspond to the incipient articular rim of Waipatia ( Fordyce 1994) and eurhinodelphinids. Ŋe articular rim/ process is more developed in platanistids, squalodelphinids, and at least some squalodontids, than in eurhinodelphinids and Waipatia , with a deeper and wider groove between this element and the lateral margin of the posterior bullar facet (see Muizon 1987); it is generally lower or absent in ziphiids, and possibly absent in eoplatanistids. Ŋe internal acoustic meatus is roughly oval but only slightly pointed anterolaterally ( Fig.8C, D ). In both the left and right periotics, a small opening is noted anterolateral to the tip of the meatus; it is interpreted as homologous to the?foramen for greater petrosal observed in Eurhinodelphis cocheteuxi and Waipatia ( Fordyce 1994; Lambert 2005; hiatus Fallopii sensu Mead & Fordyce 2009). Along the posterolateral margin of the meatus is a small aperture for the endolymphatic duct; the aperture for the cochlear duct was possibly larger than the latter, close to the posteromedial margin of the hiatus.

Tympanic bulla. Ŋe length of the left tympanic, without the posterior process, is 39 mm. In ventral view, it is oliveshape ( Fig. 8E ), with a pointed anterior margin lacking a defined anterior spine (considering the dorsoventrally flattened anterior margin) and without the deep anterolateral notch seen in pomatodelphinines and squalodelphinids (see Muizon 1987). Inner and outer posterior prominences have the same posterior extent, differing from platanistids. Ŋe inner prominence is slightly narrower than the outer prominence, a difference not as marked as in several, but not all, eurhinodelphinids and ziphiids ( Lambert 2004, 2005a, b; Bianucci et al. 2010). Ŋe median furrow is deep and extends anteriorly farther than the mid-length of the bone, roughly similar to Macrodelphinus , eurhinodelphinids, squalodontids, and ziphiids. Ŋe median furrow is anteriorly longer in platanistids and squalodelphinids, whereas it is roughly absent in eoplatanistids ( Muizon 1988b). Ŋe dorsal margin of the involucrum is cut by a vertical indentation at mid-length ( Fig. 8F ), as in eoplatanistids and eurhinodelphinids. In ziphiids, the indentation is usually mostly seen in dorsal view, whereas it is visible in both dorsal and medial views in eoplatanistids and eurhinodelphinids. Interestingly, even if lower, a similar indentation is also observed in Albertocetus , whereas no clear indentation is present in Waipatia , platanistids, squalodelphinids, and squalodontids. In dorsomedial view, the sigmoid process is directed anteromedially.

Teeth

21 teeth of the holotype are preserved in situ in the maxillae; among them, only nine are complete to subcomplete. All the teeth, from the anterior to the posterior-most are simple, conical, and single-rooted ( Figs 4 ; 5A, B ; 9 ), contrasting with Squalodon and Waipatia . Ŋe teeth from the anterior third of the preserved length of the alveolar groove are the highest, with the height outside the alveolus reaching 13.4 mm. Ŋey are slightly less robust, with transverse and anteroposterior diameters respectively ranging from 6 to 6.4 and from 5.6 to 6.8 mm. Ŋe teeth from the second third are lower, with the height outside the alveolus reaching 11.8 mm in the last teeth. Ŋe transverse and anteroposterior diameters are

A

apical wear

generally larger, up to 7.2 and 7.1 mm respectively. In the posterior third, these diameters decrease to 5.5 and 5.7 mm respectively in the last preserved teeth (on the left alveolar row of the holotype, only the last posterior tooth is lost). Posterior teeth are also somewhat more ventromedially curved. Low mesial and distal keels are observed in all the preserved teeth; these subtle keels are neither crenulated nor do they bear accessory denticles. Ŋe enamel is nearly smooth ( Fig. 9 ), only marked by longitudinal grooves lighter than in Macrodelphinus . Exposed roots of anterior teeth are not inflated, but they do display a distinctly posteriorly projecting apex.

Tooth wear

Two anterior teeth display an intact apex whereas a nearly vertical wear surface is present on the apex of two more curved posterior teeth ( Fig. 9B ). Ŋis wear surface might correspond to an occlusion facet, instead of a tooth-to-food item contact. As in Macrodelphinus , the crown base is marked by a shallow occlusion facet on the distal surface of several teeth within the second third of the tooth row, corroborating the hypothesis for the presence of embrasure pits posterior to some maxillary teeth.

Mandible

An in situ 488 mm long portion of the right mandible of the holotype includes a part of the long symphysis, as well as most of the post-symphyseal region (355 mm). Ŋe latter is somewhat dorsoventrally crushed and lacks the angular process ( Figs 4 ; 7 ). Ŋe main mandibular fragment from MUSM 1401 ( Fig. 12A, B ) is a portion of the right dentary that includes at least 267 mm of the symphyseal region. Ŋe symphyseal surface, marked by longitudinal grooves and ridges without indication of major break, indicates that dentaries were not ankylosed in both the holotype and MUSM 1401, differing from the condition of allodelphinids, eoplatanistids, eurhinodelphinids, platanistids, squalodelphinids, and Argyrocetus patagonicus , but similar to Macrodelphinus . Ŋe height and width of the section of the dentary at the posterior end of the symphysis are 34 and 26 mm respectively in the holotype; taken together, the two dentaries were wider than high at that level moving forwards.Where preserved, the ventral margin of the mandible is roughly rectilinear. Ŋe dorsal margin raises gently towards the coronoid process, less abruptly than in Papahu . A slightly convex portion of this margin, posterior to the last alveolus, is not as developed as the precoronoid crest observed in ziphiids and some delphinoids ( Fordyce et al. 2002; Lambert et al. 2013). Ŋe maximum height of the coronoid process above the mandibular condyle of the holotype is approximately 35mm, proportionally lower than the posterodorsally elevated process in Albertocetus and, in a lesser extent, Waipatia . Ŋe condyle is 51 mm high, with a ventral margin distinct from the ventral-most margin of the ramus, differing on this point from physeteroids ( Bianucci & Landini 2006).

On the lateral surface of the mandible of the holotype, at least eight small mental foramina are counted, in an oblique line from the coronoid region to the anteroventral margin. As in Macrodelphinus , no lateral groove is detected in either of the two specimens, differing from eoplatanistids, eurhinodelphinids, and platanistids.

Ŋe preserved part of the symphyseal region of the holotype contains more than 10 alveoli, whereas 10 simple alveoli were counted on the post-symphyseal region. Ŋeir transverse diameters range from 6.1 to 6.9 mm, and interalveolar septa are 6 to 8.5 mm thick. In MUSM 1401, 22 alveoli are counted on 296 mm, with transverse diameters of 8-11 mm. Ŋe rounded anterior margin of the dorsoventrally high mandibular foramen of the holotype reaches a level 125 mm from the posterior surface of the mandibular condyle ( Fig. 7 ). With the mandible virtually repositioned in articulation with the skull, this anterior limit roughly corresponds to the anterior end of the pterygoid sinus fossa on the palate.

Manubrium (presternum)

This first segment of the sternum of the holotype, not frequently described in fossil odontocetes, is free from the second segment and roughly complete ( Fig. 10 ). The maximum length is 132 mm, the maximum width at the level of the articulation surface with the first rib is 88 mm, the maximum posterior width is 71 mm, the minimum width of the body is 53 mm, and the maximum dorsoventral thickness at mid-length is 18 mm. This is larger but thinner than in the pomatodelphinine Zarhachis Cope, 1868 (see Kellogg 1924). The general outline and proportions of the manubrium, are more similar to the ancient physeteroid Ferecetotherium Mchedlidze, 1970 (see Mchedlidze 1976) and the stem-ziphiid Ninoziphius (see Muizon 1984), proportionally narrower anteriorly than in the extant sperm whale Physeter Linnaeus, 1758 , and less stocky than in extant ziphiids, with a more abrupt widening towards the articulation surface with the first rib than in the latter. This articular surface is positioned more posteriorly on the manubrium as compared to the same articular surface in Dorudon Gibbes, 1985 (see Uhen 2004), Kogia Gray, 1846 (see Schulte & Smith 1918), Platanista Wagler, 1830 , Physeter , and Sulakocetus Mchedlidze, 1976 , a possible waipatiid ( Fordyce, 1994). Differing from Ninoziphius , most other ziphiids, and several other odontocetes (including kentriodontids, Inia , Pontoporia Gray, 1846 , monodontids and several delphinids, see for example Muizon 1988a: fig. 107), the anterior margin is not cut by a distinct notch and was likely originally smoothly convex transversely. The posterior margin was originally rectilinear. The posterolateral corners are thick, corresponding to the areas of articulation of the second pair of ribs. The ventral surface is concave between these thicker regions. A barely visible ridge marks the sagittal plane on part of the posterior half of the ventral surface, whereas a thin and short groove is observed more anteriorly, up to the level of the articular surface for the first rib. From that level onwards, a strong sagittal crest divides the ventral surface into two depressed areas that face anteroventrolaterally. A slight crest is observed in the same region in some mysticetes, for example Balaenoptera musculus and Caperea ; a wide and deeply concave anterolateral area is also described by Wilson (1935; not figured) in Macrodelphinus , but with a 23 mm long anterior projection lacking in the holotype of Chilcacetus cavirhinus n. gen., n. sp. In lateral view, the anterolateral margin of each depression curves anteroventrally, as seen in several delphinids. This anterior region is moderately asymmetric, with the left side more anterolaterally developed than the right side. The depressed areas probably correspond to the regions of origin of the sternohyoid muscle, inserting on the basihyal and thyrohyal in extant odontocetes; contractions of this muscle are thought to pull posteroventrally the hyoid apparatus, as well as the tongue and floor of the mouth ( Reidenberg & Laitman 1994). The moderate asymmetry observed in the holotype of Chilcacetus cavirhinus n. gen., n. sp. might be related to the laryngeal asymmetry generally observed in odontocetes, allowing them to swallow larger prey around the larynx ( Reidenberg & Laitman 1994; McLeod et al. 2007). On the dorsal surface, three tiny foramina are observed in the middle of the body. A few more tiny foramina are located posterior to a deep and narrow pit just before the anterior margin of the manubrium. This depression, not completely piercing the bone, might be a remnant either of a larger foramen, present in the same area in several delphinids (e.g., Globicephala melas Traill, 1809 and Lagenorhynchus acutus [Gray, 1828]), or of a deep notch as seen in several ziphiids.

Vertebrae

Eight vertebrae of MUSM 1401 are partly preserved; the transverse apophyses are usually broken at the base and even the centra are incomplete. Based on the few elements at hand, it is difficult to identify their position, even approximate, along the vertebral column. Ŋree of these vertebrae are illustrated in Figure 12 C-F and a few measurements are provided in Table 2. When the epiphyses are preserved, they are ankylosed to the corresponding centrum, suggesting that the animal was adult.

Ribs

Only a few fragments of ribs of MUSM 1401 are preserved. Two of them are illustrated in Figure 12 G-H.Two shark tooth marks, simple and roughly rectilinear, are observed on one of the rib fragment. Such tooth marks are commonly observed in the field on bony remains of small odontocetes from the Chilcatay Formation.

Figures

FIG. 1. — Skull and part of right mandible of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype),in dorsal view: A, photograph; B, interpretive drawing. Hatching for surfaces of break; dashed line for incomplete elements; shading for sediment. Scale bar: 100 mm.

FIG. 2. — Detail of the vertex of the skull of Chilcacetus cavirhinus n. gen.,n. sp., MNHN.F.PRU11 (holotype),in dorsal view:A, photograph;B, interpretive drawing. Dashed line for incomplete elements;shading for sediment.Scale bar: 20 mm.

FIG. 11. — Skull of Chilcacetus cavirhinus n. gen., n. sp., MUSM 1401, in dorsal (A), right lateral (B), ventral (C), and posterior (F) views; D, E, detail of the vertex in anterodorsal and slightly lateral (D) and dorsal (E) views; G, transverse section in the rostrum at one third of its length. Scale bars: A-C, F, 100 mm; D, E, 20 mm. G, enlarged 2x compared to A.

FIG. 12. — Mandible and postcranial remains of Chilcacetus cavirhinus n. gen., n. sp., MUSM 1401: A, B, fragment of right dentary in lateral (A) and dorsal (B) views; C-F, vertebrae (corresponding to vertebrae 1-3 in Table 2); C,?cervical in anterior or posterior views; D, cervical or thoracic in anterior or posterior views; E, F, thoracic in anterior or posterior views (E) and dorsal view (F); G, H, rib fragments. Note the shark tooth marks in H (arrows). Scale bars: A, B, 50 mm; C-F, G, H, 20 mm.

FIG. 5. — Skull and part of right mandible of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype): A, left lateral view; B, oblique anterolateral and slightly dorsal view; C, ventral view; D, posterior view. Scale bar: 100 mm.

FIG. 6. — Detail of the left orbit and palate of the skull of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype), in ventrolateral view: A, photograph; B, interpretive drawing. Hatching for surfaces of break; dashed line for incomplete elements; shading for sediment. Scale bar: 50 mm.

FIG. 7. — Detail of the palate and basicranium of the skull and part of right mandible of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype), in right ventrolateral view: A, photograph; B, interpretive drawing. Hatching for surfaces of break; shading for sediment. Scale bar: 50 mm.

FIG. 4. — Skull and part of right mandible of Chilcacetus cavirhinus n. gen., n. sp. MNHN.F.PRU11 (holotype), in right lateral view: A, photograph; B, interpretive drawing; C, detail of the anterior part of the rostrum with the anterior portion of the right mandible removed. Hatching for surfaces of break; dashed line for incomplete elements; shading for sediment. Scale bar: 100 mm.

FIG. 8. — Interpretive drawings of the ear bones of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype), based on preliminary drawings: A-C, right periotic in ventral (A), dorsolateral (B), and anterodorsal and slightly medial (C) views; D, left periotic in anterodorsal and slightly medial view; E, F, right tympanic bulla in ventral (E) and mediodorsal (F) views. Hatching for surfaces of break. Scale bars: 10 mm.

FIG. 9. — Right posterior maxillary teeth of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype): A, lateral view; B, ventromedial view. Scale bars: 10 mm.

FIG. 10. — Manubrium (presternum) of Chilcacetus cavirhinus n. gen., n. sp., MNHN.F.PRU11 (holotype), in dorsal (A), ventral (B), and right lateral (C) views.Scale bar: 20 mm.