INTRODUCTION

During South America's Cenozoic isolation, an endemic fauna developed. Many of its components share morphological similarities with Holarctic taxa, which have been considered as cases of convergence (Scott, 1937; Simpson, 1980; Croft, 2016). One of the most diverse and widely spread groups corresponds to a potentially polyphyletic assemblage, the so-called South American native ungulates (hereafter SANUs; Bond, 2016; Croft et al., 2020). These native ungulates include placental herbivorous mammals that reached their greatest generic diversity throughout the Paleogene. Later, the number of suprageneric clades progressively declined throughout the Neogene until they went extinct in the Late Pleistocene/Early Holocene (Bond, 1986; Bond et al., 1995). Among the South American native ungulates, the order Litopterna is one of the most diverse and widely distributed, with some of its representatives even being present in West Antarctica throughout most of the Eocene (Bond et al., 2006, 2009; Gelfo, 2016; Gelfo et al., 2017).

Within Litopterna, the family Proterotheriidae includes small to medium-sized cursorial herbivores with low-crowned teeth (brachydont or mesodont) and reduced lateral digits (II and IV) (Scott, 1910; Bond et al., 2001; Soria, 2001; Villafañe et al., 2006, 2012; Ubilla et al., 2011; Schmidt, 2015; Morosi and Ubilla, 2017). Most of them occupied forested to open habitats of South America, in various locations in Argentina, Brazil, Bolivia, Chile, Colombia, Uruguay, and Venezuela (Carlini et al., 2006; Scherer et al., 2009, and references therein). Its fossil record exhibits two peaks of maximum diversity, the Early Miocene and Late Miocene (Villafañe et al., 2006; Ubilla et al., 2011). Finally, during the Pleistocene, the diversity of proterotheriids decreased to a couple of species that have been documented only in Argentina, Brazil, and Uruguay (Neolicaphrium recens and Uruguayodon alius; Scherer et al., 2009; Ubilla et al., 2011; Morosi and Ubilla, 2017; Corona et al., 2019; Agnolin et al., 2020).

Numerous proterotheriid species have been described (e.g., Soria, 2001; Villafañe et al., 2006, 2012). Some of these species are based on remarkably complete material (Scott, 1910; Simpson, 1932), but others are known mostly from isolated and incomplete remains. Moreover, most available descriptions primarily rely on dental evidence and, in fact, a comprehensive study of the cranial morphology of proterotheriids has not been carried out since Soria (2001), which is based on research conducted prior to his death in 1990. Furthermore, beyond some exceptional specimens from the Early Miocene of Patagonian Argentina (e.g., Thoatherium, Tetramerorhinus), knowledge of cranial material of even Neogene representatives is very limited.

The skull is a complex structure that houses the brain, sensory organs, and muscles responsible for mandibular movement (Moore, 1981; Emerson and Bramble, 1993; Fostowicz-Frelik and Tseng, 2023). Due to its complexity, it has been the focus of numerous studies that have investigated aspects of morphology and function in Mammalia, particularly among SANUs (e.g., Simpson, 1936; Cassini et al., 2012; MacPhee, 2014; Forasiepi et al., 2016; García-López et al., 2018; MacPhee et al., 2021). While cranial research has long captivated the scientific community, the exploration of cranial anatomy in juvenile individuals is of great importance, as it provides a unique opportunity to unravel the growth, development, and adaptation patterns during life's most transformative stage (e.g., Goswami, 2007; Sánchez-Villagra, 2010). As young individuals undergo cranial transformations, influenced by genetic predispositions and environmental factors, they not only offer insights into the specifics of development but also encapsulate the evolutionary history of their species (Gilbert, 2000; Goswami et al., 2012).

Among several cranial structures, the petrosal bone is a valuable source of information for analyzing phylogenetic relationships, adaptations to ecological niches, and even the auditory capabilities of mammals (e.g., Macrini et al., 2010; MacPhee, 2014; Billet et al., 2015; Forasiepi et al., 2016; García-López et al., 2018). Nevertheless, due to its location deep within the cranium, studying its features, such as shape, size, and the arrangement of structures, represents a significant challenge using traditional methods. Nowadays, noninvasive techniques for exploring internal cranial structures are widely employed in scientific research, particularly in paleontology. However, their use has been somewhat limited in litopterns in general (e.g., Billet et al., 2015; Forasiepi et al., 2016).

In this article, we analyze the cranial morphology of a juvenile individual of the late Neogene proterotheriid Neobrachytherium intermedium (Litopterna), with a focus on its cranial anatomy, particularly the orbitotemporal region, the mesocranium, and certain basicranial structures. Additionally, we make observations on ontogenetic stages within the genus based on cranial and dental morphology. Our study enables a discussion of features that have received limited attention in previous studies of proterotheriids, yielding potentially significant information for understanding the evolution of South American native ungulates.

MATERIALS AND METHODS

The studied specimen (PVL 7792) is stored at the Colección Paleontología de Vertebrados Lillo of the Facultad de Ciencias Naturales e Instituto Miguel Lillo, Universidad Nacional de Tucumán, Argentina. The anatomical study and comparisons were carried out mostly through direct observation, mainly using specimens of proterotheriids and litopterns stored in the mentioned collection and at the Museo de La Plata, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia,” Buenos Aires, Argentina, and the Field Museum, Chicago, Illinois. We also used historical descriptions and illustrations in the literature. Additionally, only from the perspective of cranial morphology, we made specific comparison to other SANUs (e.g., astrapotheres, pyrotheres, notoungulates) and extant ungulates such as equids, tapirids (considering the context of Pan-Perissodactyla proposed in recent contributions; see Buckley, 2015; Welker et al., 2015), and cervids, given the general resemblance observed with some taxa (e.g., Mazama).

Given the goals of the present study, we focused on the details of the cranial and dental anatomy of PVL 7792 without considering the postcranial remains, which are the subject of another contribution (i.e., Schmidt et al., 2024). The descriptions of the cranial bones are organized according to a topographical criterion from front to back. Anatomical terminology mostly follows the Nomina Anatomica Veterinaria (NAV; International Committee on Veterinary Gross Anatomical Nomenclature, 2017) and recent contributions on cranial morphology (e.g., Wible, 2003, 2007, 2011; Wible and Gaudin, 2004). Additionally, some terms that have varied in descriptions focused on SANUs have been summarized in table 1, in order to facilitate comparisons with other contributions. Regarding the dentition, we have followed the terms used by Soria (2001) in his extensive survey on proterotheriids, and Schmidt (2015: fig. 2). For abbreviations used in figures 1–14, see table 2.

In order to explore further details of the dentition and auditory region through non-invasive techniques, PVL 7792 was scanned at the Hospital Avellaneda of San Miguel de Tucumán (Tucumán, Argentina) using a Philips MX 16-slice clinical CT scanner (120 kV, 220 mA, slice thickness of 0.18 mm). The scan resulted in a final stack of 365 slices of 952 × 952 pixels. The program 3D Slicer (Fedorov et al., 2012) was used for 3D reconstruction and segmentation of the cranium and associated structures. External images were taken with a Nikon D5500 digital camera and measurements using a digital caliper (0.01 mm).

SYSTEMATIC PALEONTOLOGY

DESCRIPTION OF THE CRANIUM

The cranium of PVL 7792 is relatively well preserved except for the rostralmost regions of the skull and the skull vault, which are broken (figs. 2–12). There is a slight lateral diagenetic deformation towards the right side. PVL 7792 lacks the premaxilla, the right nasal and lacrimal, part of both zygomatic arches, the caudal border of the frontals, the dorsal surface of the squamosal, the parietals, the supraoccipital, and the left occipital condyle. The fact that the bones of the skull vault are not preserved allows access to the brain endocast (fig. 2); however, its original morphology was affected by the extraction process; therefore, detailed analyses cannot be done with accuracy.

In general terms, the cranium is longer than wide, with the rostrum representing roughly 33% of the total preserved cranial length and the large orbit occupying the central third in lateral view. The snout is long and narrows anteriorly (fig. 2). The maximum width of the cranium is at the level of the supraorbital foramen (across the zygomatic arches). The postorbital constriction is relatively weak (probably mostly due to the juvenile condition of the specimen). In lateral view, the dorsal outline of the cranium is slightly convex, descending rostrally; nevertheless, as the skull vault is broken, the real lateral outline of the specimen cannot be accurately described (fig. 3). The occipital condyles project caudally. In ventral view, the palate is narrow and the lingual margins of the postcanine rows are somewhat parallel; however, the buccal margins slightly diverge caudally, reaching their maximum width at the level of M1 (fig. 4). From this point caudally, at least the buccal margins of the toothrows likely would have converged toward the midsagittal plane, given the position of the unerupted M2 (although the position of this tooth inside the maxilla probably does not accurately foreshadow its position when completely erupted).

FIG. 1.

Geographical and geological context. A–B, location of the study site (Tucumán, Argentina) in the context of South America; C, aerial photograph (facing north) of the fossil-bearing levels; D, detail of Encalilla fossiliferous site and the exposure of the Andalhuala Formation in the area; E, photograph of PVL 7792 in situ.

Nasal: The nasal (paired dermal bone) forms the roof of the nasal cavity at the rostral end. In PVL 7792, the left nasal is preserved almost completely (without its rostral end). Caudally, it contacts the frontal by means of the sutura frontonasalis (fig. 2). This suture is serrate and V-shaped (W-shaped if both nasal bones are considered). In lateral view, the nasal contacts the maxilla (sutura nasomaxillaris) on its ventrocaudal third (fig. 3). The nasomaxillary suture is slightly serrated and continues in line with the suture between the lacrimal and frontal bones (see below).

The nasal of PVL 7792 is rostrocaudally elongated, with its width increasing toward the caudal end and reaching its maximum at the level of the rostral edge of the lacrimal, anterior to the rostralmost edge of the orbital rim. The surface of the nasal is slightly vaulted, without foramina. There is an elongated depression (the frontal sulcus) running rostrolaterally over the nasal surface. The rostral end of this sulcus reaches the rostrodorsal area of the facial process of the maxilla (rostral to the infraorbital foramen); caudally, the sulcus continues on the frontal surface (see below; figs. 2, 3).

Frontal: The frontal (paired dermal bone) forms the cranial roof between the orbits and contributes to the medial walls of the orbitotemporal fossa. Both frontals contact each other along the sagittal plane through a straight suture. In dorsal view, the frontal contacts the nasal rostrally, penetrating the caudomedial edge of that bone by means of a triangular wedge, which corresponds to the V-shaped part of the frontonasal suture (fig. 2). In lateral view, the frontal of PVL 7792 contacts the lacrimal rostrally (sutura frontolacrimalis); the palatine (sutura frontopalatina) and orbitosphenoid and alisphenoid (sutura sphenofrontalis) ventrally; and the squamosal caudally (inferred contact). There appears to be no contact with the maxilla on the rostral surface, though a rather small contact is possible (breakage obscures the precise relationship). The suture between the frontal and the lacrimal bones is irregular, wide, and clearly observable adjacent to the edge of the orbit; however, the other sutures on the lateral wall of the skull are weakly distinguishable (fig. 3).

In dorsal view, the surface of the frontal is smooth and slightly depressed, mainly at the interfrontal contact, and reaches its maximum width at the level of the postorbital bar (fig. 2). The lateral marginal portion of this surface is sloped downward on the zone of the frontolacrimal contact and the postorbital bar. The temporal line or crest (crista frontalis externa), on the frontal laterocaudal edge, is sharp and sigmoid and seems to continue ventrally, contributing to the dorsal part of the caudal edge of the postorbital process and bar. At least the preserved part of the temporal line is transverse and defines the rostrodorsal border of the temporal part of the orbitotemporal fossa and, therefore, the rostrodorsal border of the area of origin of the m. temporalis (e.g., Osgood, 1921; Turnbull, 1970).

On the dorsal surface of the frontal there are two conspicuous foramina: one in the laterocaudal area, on the postorbital process rostral to the postorbital bar, and the other at the center of the frontal surface, between the postorbital process and the rostral edge of the orbital rim (fig. 2). We interpret the posterior one as the supraorbital foramen. It is broken at its lateral wall, but it seems to be subcircular. It faces dorsolaterally and is larger than the anterior foramen on the frontal (fig. 3). The anterior foramen, on the other hand, is dorsally located and elliptical, with its major axis longitudinally oriented, and preceded by a wide and shallow frontal sulcus that runs anteriorly up to the nasal (see above; fig. 2). We refer to this aperture as the frontal foramen, following the terminology used by Dozo and Vera (2010) for a similar structure observed in macraucheniid litopterns. On the intraorbital surface of the frontal (roof of the orbit), a deep and relatively wide groove connects the supraorbital and frontal foramina (fig. 5A–D). Also, there is a small aperture on the medial wall of the groove. Given its position close to the supraorbital foramen, this aperture likely corresponds to the foramen for the frontal diploic vein (fig. 5D), an emissary of the dorsal cerebral vein/dorsal sagittal sinus or a vein issuing from the frontal diploë (Wible and Rougier, 2000; Wible, 2003, 2011; Wible et al., 2004, 2009). The supraorbital foramen represents a path for vessels and nerves (i.e., supraorbital nerves; Clark, 1925, 1926) from the orbit to the superficial region of the forehead (e.g., supplying the skin above the orbit, as in horses and ruminants; Sisson and Grossman, 1975; Barone, 1999). As for the frontal foramen, this would be related to branches of the external ethmoidal artery and vein, as noted by Wible and Gaudin (2004) in Euphractus.

The frontal sulcus is curved and runs rostrally from the frontal foramen, roughly parallel to the sagittal plane. It then turns laterally and ventrally over the nasal on its anterior end and vanishes on the dorsorostral aspect of the facial process of the maxilla at the level of dP1 (figs. 2–3).

Lacrimal: The lacrimal (paired dermal bone) forms the rostral edge of the orbit and has two processes: facial and intraorbital (figs. 2 and 5). The rostral orbital rim, represented by a sharp orbital crest, separates both. Only the left lacrimal is well preserved in PVL 7792.

FIG. 2.

Neobrachytherium intermedium PVL 7792, cranium in dorsal view.

The facial process of the lacrimal is a relatively large D-shaped plate that is roughly vertical and faces rostrolaterally (with two vertices on the orbital edge; fig. 3). Dorsally, the facial process of the lacrimal contacts the frontal by means of a long and serrated suture. Rostrally, the lacrimal is mostly surrounded by the maxilla (sutura lacrimomaxillaris) and, over its ventral-most part, it contacts the jugal by means of a relatively short and irregular suture (sutura lacrimozygomatica). At this point, there is a small knob, just ventral to the anterior lacrimal foramen (probably for the attachment of the palpebral ligament). The surface of the facial process is slightly concave and presents a short lacrimal tubercle, located at the rostral edge of the orbit, dorsal to the anterior lacrimal foramen on the facial exposure (fig. 3).

There are two lacrimal foramina, one on the facial surface and one on the intraorbital process (figs. 3, 5A, C). The former (anterior lacrimal foramen) is circular, opens on the ventrocaudal facial surface of the lacrimal, and is preceded by a short but deep groove. Its caudal wall is a sharp edge that forms the orbital border (fig. 3). The latter foramen (posterior lacrimal foramen) is smaller and opens inside the orbit, caudal and dorsal to the first aperture and just caudal to the orbital edge (figs. 3, 5A, C). Between both foramina, there is a small lacrimal tubercle, rostrocaudally compressed, located on the orbital rim. At least one of these foramina would be part of the lacrimal canal, carrying the nasolacrimal duct from the orbit to the nasal cavity (Sisson, 1965; Archer, 1976; Hiatt and Gartner, 2001). Potential connections between both lacrimal foramina could not be assessed.

FIG. 3.

Neobrachytherium intermedium PVL 7792, cranium in left lateral view. In this and subsequent figures, a small portion of the left coronoid process adherent to the cranium is highlighted in red.

The intraorbital process of the lacrimal forms part of the rostral internal wall of the orbits (fig. 5). It contacts ventrally with the jugal and the maxilla by vertical and transverse sutures, respectively, then forms an acute laterocaudal corner. On the dorsomedial area, the intraorbital process contacts the frontal by an irregular and oblique suture. The intraorbital surface of the lacrimal is concave, and its ventral part contributes to the dorsal edge of two apertures, the maxillary foramen and the sphenopalatine foramen (see below). Additionally, there is a small and elliptical opening, with its major axis dorsoventrally oriented, located laterodorsally relative to the maxillary foramen. This foramen is completely formed by the lacrimal and probably corresponds to a nutritive aperture.

TABLE 1.

Anatomical terms used in this contribution and their alternatives.

continued

Jugal: The jugal (paired dermal bone) forms most of the zygomatic arch or zygoma and contributes to the ventral border of the orbit. In PVL 7792, the jugal is a rostrocaudally elongated, laterally convex, and dorsally curved element. In lateral view, the jugal contacts rostrally with the lacrimal (sutura zygomaticolacrimalis) at the level of the mesostyle of dP4 (fig. 3). Ventrally, the contact with the maxilla (sutura zygomaticomaxillaris) is parallel to the ventral orbital rim, where the jugal forms a sharp crest (fig. 4). There is a small intraorbital component that contacts rostrally with the lacrimal and medially with the maxilla by means of a straight suture that is also parallel to the ventral orbital rim.

The ventral aspect of the jugal shows a flat surface that would have represented the origin of the m. masseter (pars profunda; Osgood, 1921; Turnbull, 1970; figs. 4, 5D). This surface is slightly concave and inclined in such a way that its surface is laterally visible (figs. 3, 4, 5D). It is defined laterally by a crest running rostrocaudally on the ventrolateral surface of the jugal and ventromedially by a second crest formed between the maxilla and the jugal. The ventromedial side of the described flat surface is separated by a smooth crest from another roughly flat and smooth surface that would have served as the attachment for the m. zygomaticomandibularis (figs. 4, 5D).

In lateral view, the caudal portion of the jugal has two processes, dorsally and ventrally developed (fig. 3). The dorsal process is the frontal process of the jugal. It is larger than the ventral one and contacts the postorbital process of the frontal dorsally, forming the postorbital bar. This area is not well preserved and cannot be described in detail. The ventral process is triangular and dorsally curved and located roughly on the midlateral surface of the zygoma. Its ventral surface is rough, representing the attachment area for the m. zygomaticomandibularis (Turnbull, 1970; figs. 4, 5D) as mentioned above.

TABLE 2.

Abbreviations.

continued

FIG. 4.

Neobrachytherium intermedium PVL 7792, cranium in ventral view. In this and subsequent figures, a distal fragment of the stylohyoid bone is highlighted in yellow. Texture-filled gray areas represent rugose surfaces for muscle attachments.

Maxilla: This paired dermal bone has four processes: facial, zygomatic, orbital, and palatine, which are exposed in the lateral (the first two processes), intraorbital, and ventral views, respectively (figs. 3–5). Since the left side is the better preserved in PVL 7792, we mainly base our description on the left maxillary bone.

The facial process of the maxilla forms the lateral wall of the nasal cavity and contains the alveolar row for the dP1–4 and the M1–2 (see Description of Upper Dentition below; fig. 3). It dorsally contacts with the nasal, caudally with the lacrimal (sutura lacrimomaxillaris), and ventrocaudally with the jugal (sutura zygomaticomaxillaris). As mentioned above (see Frontal), there is a possible point of contact with the frontal bone at the dorsocaudal edge of the facial process that is obscured due to breakage; nevertheless, such contact would have been very small or nonexistent. The root of the premolars and unerupted permanent teeth forms a conspicuous bulge (juga alveolaria) on the lateral aspect of the maxilla, extending caudally from dP1 to the level of dP4 (fig. 3). The intermolar process (the wedgelike process located between teeth on the lateral aspect of the postcanine row) is more marked among dP3, dP4, and M1, while the interradicular process (a similar but usually more blunt process between the roots of teeth on the lateral aspect of the postcanine row) is similarly developed throughout the toothrow. There are fracture lines above the infraorbital foramen that preclude recognition of the area of origin of the m. levator nasi and m. levator labii maxillaris (e.g., Minkoff et al., 1979).

FIG. 5.

Neobrachytherium intermedium PVL 7792, detail of left orbit and surrounding structures, in lateral (A), dorsorostrolateral (B), caudolateral (C), and ventral (D) views. In D, 1 indicates the ethmoidal foramen, which forms the aperture of a canal that leads to the frontal foramen on the dorsal surface of the frontal bone. Texture-filled gray areas represent rugose surfaces for muscle attachments.

The infraorbital foramen (rostral opening of the infraorbital canal) is located at the level of the contact dP2-dP3 (fig. 3). It is oval, with the major axis dorsoventrally oriented, and faces rostrally. This foramen would have transmitted the infraorbital nerve (branch of the maxillary division of the trigeminal nerve, V2) and the accompanying infraorbital artery (branch of the maxillary artery) and vein (e.g., Sisson, 1965; Hiatt and Gartner, 2001). The infraorbital canal is broad and relatively long (fig. 6). The caudal aperture of the infraorbital canal, which opens into the orbital cavity at the level of the contact between dP4 and M1, is the maxillary foramen. It is circular and larger than the infraorbital foramen and located lateral to the sphenopalatine foramen (figs. 3, 5C). As mentioned above, the floor and lateral wall of the maxillary foramen are formed by the maxilla, while the dorsal wall is composed of the lacrimal bone.

The short zygomatic process of the maxilla is visible mainly in ventral view (figs. 4, 5D). It contributes a minor portion to the anterior root of the zygomatic arch. The rostral edge of this process is at the level of the mesostyle of M1, and the caudal edge seems to extend caudally beyond the M2. The suture with the jugal is dorsorostrally-ventrocaudally oriented. In ventral view, the rostral surface of the zygomatic process of the maxilla shows a series of small tuberosities, rostromedial to the flat area that may have served as the origin of the deep masseter (see Jugal). These would have corresponded to the attachment of the pars superficialis of the m. masseter (Turnbull, 1970; Sosa and García-López, 2018). It is considerably narrower and smaller than the area for the deep masseter (figs. 4, 5D).

The orbital process of the maxilla contributes to the formation of the orbital platform (floor of the orbital cavity). It contacts rostrally with the lacrimal, laterally with the jugal, and medially with the palatine (sutura palatomaxillaris). The maxillary tuberosity, housing the roots of M1–2, forms the markedly convex floor of the orbit; the caudal edge of this tuberosity is conspicuously rounded (fig. 3). In fact, the tips of the roots of M2 are slightly exposed in the right orbit, probably by breakage. Medially, throughout the contact with the palatine bone, there is a concave and wide groove, which incorporates rostrally the maxillary and sphenopalatine foramina.

In ventral view, the palatine process of the maxilla represents more than 60% of the total surface of the preserved hard palate (fig. 4). Although difficult to recognize, the process seems to contact the palatine caudally via an irregular palatomaxillary suture located at the level of the mesial edge of the dP4. Caudally, the suture runs parallel to the alveolar line, on the lateral wall of the palate, toward the base of the palatine crest (see below). The surface of the palatine process is mostly smooth but concave, resulting in a vaulted transverse profile, deeper toward the caudal end.

FIG. 6.

Neobrachytherium intermedium PVL 7792, parasagittal segment through left side of the cranium (A) and transverse segment at the level of the infraorbital foramen (B). Note that the infraorbital canal is relatively long whereas its caudal aperture (maxillary foramen) opens into the orbital cavity at the level of the contact between dP4 and M1.

Palatine: The palatine (paired dermal bone) forms the caudal hard palate, the choanae, and contributes to the orbital medial wall and floor and the lateral walls of the nasopharynx. In PVL 7792, the left palatine preserves two processes: a horizontal process that contributes to the hard palate and an ascending (or intraorbital) process exposed in the orbital cavity (lateral view) and in the nasopharyngeal passage (partially filled with sediment).

The horizontal process forms the caudal third of the palate and the floor of the choanae; it contacts the maxilla rostrally and laterally, the opposite palatine in the midsagittal plane of the palate (not preserved in PVL 7792), the alisphenoid caudally, and the pterygoid mediocaudally (fig. 4). The surface of the palatine on the palatal zone is roughly rectangular and excavated. Rostrally, the major palatine foramen is located near the contact with the maxilla at the level of the distal edge of the dP4 and is preceded by a short groove. It is subcircular and faces rostroventrally. The major palatine artery and nerve (branches of the maxillary artery and nerve, respectively) reach the palate through the major palatine foramen (e.g., Sisson, 1965; Hiatt and Gartner, 2001). In the caudal area of the horizontal process, the choanal border describes a parabola whose rostral edge reaches the level of the contact between M1 and M2. Also, the choanal border presents a robust tubercle at the rostromedial zone (postpalatine torus; Novacek, 1986; Wible, 2003, 2011; Muizon et al., 2015), associated with a short tubercule on the opposite (lateral) wall (fig. 4) and defining a relatively wide palatal groove that is oriented caudolaterally-rostromedially. These structures likely indicate the area of attachment of the muscles of the uvula and palatine velum (e.g., Hiatt and Gartner, 2001; but see MacPhee et al., 2021, for additional functional interpretations of a similar structure present in Trigonostylops, an Eocene SANU). The caudal end of the horizontal process, along with the alisphenoid, make up the ectopterygoid crest, which projects ventrally (fig. 3), reaching the level of the alveolar line (right apex partially preserved; see below). At the base of this crest, medially, the palatine contacts the pterygoid bone, which is reduced to a small plate (fig. 4). Both the mediocaudal wall of the horizontal process of the palatine and the pterygoid (see below) form the entopterygoid crest. This latter structure is not well preserved but seems to be less developed than the ectopterygoid crest.

The ascending process of the palatine is observable in lateral view, contributing to the medial wall of the orbit (fig. 3). It contacts the maxilla ventrally, the frontal and orbitosphenoid dorsally, and the alisphenoid caudally. The palatine forms the mediocaudal edge of the sphenopalatine foramen, which is completed by the maxilla (laterally), frontal (medially), and lacrimal (dorsomedially). This foramen is located at the level of M1, being a large, kidney-shaped aperture that faces laterocaudally. It transmits the sphenopalatine and major palatine arteries (branches of the maxillary artery), veins, and branches of the maxillary nerve (second division of trigeminal nerve, V2): the major palatine nerve to the hard palate, the nasopalatine nerve, and minor nasal branches to the nasal cavity (e.g., Evans and de Lahunta, 2013).

Pterygoid: The pterygoid is a laminar structure that forms part of the roof and lateral walls of the nasopharyngeal passage behind the choanae. It consists of a vertical lamina and a horizontal basal plate. In PVL 7792, the pterygoid is reduced to a small subtriangular plate located on the internal wall of the ectopterygoid crest, forming the entopterygoid crest (figs. 4, 7), as in other SANUs (e.g., Billet et al., 2008; Billet, 2010; García-López and Powell, 2011). It contacts the palatine rostrally (sutura pterygopalatina) and the alisphenoid caudally (sutura pterygosphenoidalis). There is no evidence of a hamular process, but since this fragile structure is easily broken, this apparent absence is probably a preservational artifact. Between the entopterygoid crest of the pterygoid and the ectopterygoid crest of the alisphenoid is a deep and narrow fossa that generally receives the m. pterygoideus medialis in mammals (Barone, 1997; Evans and de Lahunta, 2013; figs. 4, 7).

Sphenoid complex: In general terms, the sphenoid complex is composed of four elements in mammals (e.g., Clark and Smith, 1993; Wible, 2003, 2007, 2011): the presphenoid on the rostral midline; the basisphenoid on the caudal midline; the paired orbitosphenoid opposite the presphenoid; and the paired alisphenoid opposite the basisphenoid. In PVL 7792, the borders of these bones are not clearly defined due to complete fusion of these elements and/or preservation bias.

The orbitosphenoid is a small endochondral bone exposed on the lateral wall of the cranium, rostral and medial to the sphenorbital fissure (figs. 3, 5). The orbitosphenoid contacts the palatine (sutura sphenopalatina) obliquely (dorsorostral-ventrocaudal), and the frontal dorsally. It is roughly wedge-shaped and contributes to the walls of three major apertures: the anterior opening of the orbitotemporal canal, the ethmoidal foramen, and the sphenorbital fissure. The optic foramen is located roughly in the middle of the orbitosphenoid surface (fig. 5A, B).

The anterior opening of the orbitotemporal canal, which serves as passage for the ramus supraorbitalis (anterior division of the stapedial artery and vein; Wible and Gaudin, 2004), is located at the level of contact between the frontal and orbitosphenoid, near the alisphenoid and rostrodorsal to the optic foramen (fig. 5A, B). This location is similar to that observed by Billet (2010) in Pyrotherium, who pointed out that is typical for many mammals. It is kidney shaped and smaller than the optic foramen. Rostral to the aperture for the orbitotemporal canal, there is a small and circular aperture that seems to be preceded by a groove (not well preserved). Based on its position near the wall of the olfactory tract and between frontal and orbitosphenoid (Billet, 2010; MacPhee et al., 2021), this opening is here interpreted as an ethmoidal foramen. In mammals, the ethmoidal foramen usually transmits a branch of the internal carotid artery (presumably a branch of the ophthalmic artery) and the ethmoidal nerve (a branch of the ophthalmic division of the trigeminal nerve, V1) from the orbit to the nasal cavity (e.g., Archer, 1976; Hiatt and Gartner, 2001; Wible, 2003). Apparently, the ethmoidal foramen would be internally related to the frontal foramen described above (fig. 5A, B).

The optic foramen is placed between the anterior opening of the orbitotemporal canal and sphenorbital fissure, roughly in the middle of the orbitosphenoid surface (fig. 5A, B). It is large and elliptical, with its major axis obliquely oriented (dorsorostral-ventrocaudal). The foramen has acute edges, almost forming a sharp continuous crest, which probably corresponds to the ossified ala hypochiasmatica (De Beer, 1937) or tendinous ring that provides attachment for extraocular eye muscles. In particular, the ventral edge is preceded by a robust crest, which could represent the attachment for the m. medial rectus and m. levator palpebrae superiorus (Giannini et al., 2006; Wible, 2011). The optic canal presents a smooth surface running ventrocaudally into the cranium. The optic foramen transmits the optic nerve (cranial nerve II).

The sphenorbital fissure is located ventral and caudal to the optic foramen, between the alisphenoid and orbitosphenoid (fig. 5B). This kidney-shaped fissure faces rostrally and represents the largest aperture on the lateral wall of the cranium. The sphenorbital fissure transmits nerves and vessels from the cavum epiptericum (Gregory, 1910; McDowell, 1958; Archer, 1976). In extant therians, the nervous and vascular contents of this opening vary dramatically and may include some combination of the oculomotor, trochlear, ophthalmic, maxillary, and abducens nerves (cranial nerves III, IV, V1, V2, and VI, respectively); the ramus infraorbitalis, arteria anastomotica, and ophthalmic artery; and the ophthalmic veins (e.g., Kuhn and Zeller, 1987; Wible, 2003; Wible and Gaudin, 2004). The lateral wall of the sphenorbital fissure is formed by the alisphenoid. There is a conspicuous crest that defines an acute edge of the fissure lateroventrally.

The alisphenoid is a strengthened endochondral bone that forms part of the orbitotemporal fossa in lateral view, where it has a mainly vertical disposition; in ventral view, it forms the lateral zone of the mesocranium, where it has a horizontal disposition (figs. 3, 4). The two regions of the alisphenoid are separated by a well-developed crest (figs. 4, 7, 8).

In lateral view, mostly inside the orbitotemporal fossa, the alisphenoid contacts the pterygoid ventromedially, the palatine rostroventrally, the frontal rostrodorsally, and the squamosal dorsocaudally (sutura sphenosquamosa). In this area, the surface of the alisphenoid forms the ventrolateral border of the sphenorbital fissure, as a sharp crest. The part of the alisphenoid forming the lateral wall of the ectopterygoid crest is concave and shows a rugose surface on the apex of that structure (figs. 4, 7, 8). This could correspond to the area of origin of the m. pterygoideus lateralis (Hiiemae and Jenkins, 1969; Turnbull, 1970).

FIG. 7.

Neobrachytherium intermedium PVL 7792, detail of basicranium region in left ventral view.

In ventral view, on the mesocranium, the alisphenoid contacts the squamosal rostrolaterally and laterally, the ectotympanic caudally, and the basisphenoid medially (fused; figs. 4, 7). The sphenosquamosal suture runs rostrally from the medial end of the glenoid fossa. It then bends dorsolaterally in a slightly obtuse angle toward the orbitotemporal fossa (figs. 4, 7–9). The surface of the ventral face of the alisphenoid is smooth and slightly depressed. There is no evidence of the presence of the foramen rotundum on the surface of the alisphenoid; therefore, the maxillary nerve (maxillary division of the trigeminal nerve/V2) would have exited the cavum epiptericum through the sphenorbital fissure. The foramen ovale is positioned caudally, at the level of the synchondrosis sphenooccipitalis (figs. 4, 7–9). It is large and elliptical, with its major axis obliquely oriented (laterocaudally-rostromedially). The foramen ovale transmits the mandibular division of the trigeminal nerve (cranial nerve V3). The medial wall of this aperture is a sharp crest that runs from the base of the ectopterygoid crest of the alisphenoid rostrally to the alisphenoid-squamosal contact caudally. At this point, this crest corresponds to the tympanic process of the alisphenoid (Kramarz et al., 2017). This tympanic process abuts a longitudinal elevation of the adjacent surface of the squamosal that may represent the entoglenoid process (see below). Both structures combined form the preotic crest for attachment of the ectotympanic ring (McDowell, 1956; Novacek, 1986; Kramarz et al., 2017). The preotic crest represents the lateral edge of a large groove formed mainly on the alisphenoid surface but also over the lateral wall of the basisphenoid. The groove can be divided in two segments, rostral and caudal (figs. 4, 7–9). The rostral segment shows a small crest on its surface, running toward the medial side of the ectopterygoid crest and hence, into the nasopharyngeal passage. At this point, features of the rostral end of the groove are hidden by sedimentary matrix and some bony elements attached to the roof of the nasopharyngeal passage as a preservation artifact. The caudal segment of the groove is divided in two ways, one toward the tympanic cavity (lateral) and the other (medial) expressed as a notch leading to the piriform fenestra (figs. 7–9). The lateral groove corresponds to the bony roof of the exit of the Eustachian tube and runs laterocaudally-rostromedially. The medial part may correspond to the passage of the nerves likely occupying the pterygoid canal (whose caudal opening would presumably be present on the rostral part of groove).

Caudally, at the level of the synchondrosis sphenooccipitalis (between the caudomedial end of the alisphenoid, the rostral extreme of the petrosal, and a short portion of the rostrolateral edge of the basioccipital) there is a large gap that correspond to the piriform fenestra (figs. 4, 7, 9).

The basisphenoid occupies the midline of the basicranium (central stem) between the presphenoid rostrally and the basioccipital caudally (synchondrosis sphenooccipitalis). In PVL 7792, it is completely fused with the paired alisphenoids laterally (figs. 4, 7–9). The ventral surface of the basisphenoid is smooth, cylindrical, and convex without any specific observable structures (there is no evidence of foramina).

Squamosal: The squamosal (paired dermal bone) contributes to the laterocaudal surface of the cranium, the caudal portion of the zygomatic arch, bears the glenoid fossa where the dentary articulates, and contributes to the auditory region by forming the wall of the epitympanic sinus (if present), the roof of the external acoustic meatus, and the postglenoid, entoglenoid, and posttympanic processes. This bone has two parts, the zygomatic process and the squamous process or squama, on the lateral and caudal wall of the skull vault, which mostly contain the glenoid and ear regions (figs. 3, 4, 7–9).

The zygomatic process of the squamosal contributes to the caudal portion of the zygomatic arch (figs. 3, 7, 8). In PVL 7792, it is completely broken on the right side of the skull and only the rostralmost and caudalmost portions of the zygomatic process are preserved on the left side (lacking the middle segment). In this left area, the zygomatic process is distorted at the level of the postorbital bar. Externally, the rostral tip of the zygomatic process of the squamosal is laterally overlapped by the jugal and shows a roughly triangular outline (fig. 3). Only the root of the caudal portion of the zygomatic arch is preserved, forming the roof of the glenoid fossa. Dorsally, there is a sharp squamosal crest (Moyano and Giannini, 2017) that runs medially from the root of the caudal portion of the zygomatic arch and merges caudally with the nuchal crest (figs. 3, 8, 10).

The squamous process is partially preserved in PVL 7792. In addition, the coronoid process of the mandible (with a broken dorsal tip) remains attached by matrix in this region, hiding the rostral part of the squamous process (figs. 3, 4, 8).

In lateral view, the squamosal process contacts the alisphenoid (observed on the right side of the skull) and the frontal (inferred contact) anteriorly, and the parietal rostrodorsally (only the sutura squamosa can be recognized, since the parietal is not preserved). In ventral view, the squamous process contacts the alisphenoid rostrally and medially, the petrosal (inside the tympanic cavity) medially and caudally, and the exoccipital mediocaudally (figs. 3, 7–9).

In lateral view, the squamous process is well developed, reaching the dorsal surface of the cranium. This is a convex-concave surface (from ventral to dorsal), smooth in the caudal portion. The suprameatal foramen (= subsquamosal foramen; e.g., Sinclair, 1906; Archer, 1976; Petter and Hoffstetter, 1983; Marshall and Muizon, 1995) opens rostrally to the roof of the external acoustic meatus, at the same horizontal level as the base of the postglenoid process (figs. 7, 10). This foramen corresponds to a small aperture facing dorsally and caudally, which is connected to a ventral groove leading to the postglenoid foramen. The suprameatal foramen would transmit a temporal branch of the postglenoid artery and accompanying vein to the temporal fossa (Archer, 1976; Wible, 2003).

In ventral view, the glenoid fossa is oval, wider than long, concave, and apparently obliquely oriented (rostroventral-caudodorsally; figs. 4, 7, 8). It is delimited by the root of the posterior portion of the zygomatic process rostrally (not preserved), by the postglenoid process of the squamosal posteriorly, and by the surface of the Glaserian fissure and the entoglenoid process of the squamosal medially. The postglenoid process is robust, relatively long and wide, and quadrangular in cross section (figs. 3, 4, 7, 8). Its laterocaudal surface is rugose and has tuberosities, including two robust tubercles.

The area dorsal to the glenoid fossa has a shallow depression for the attachment of the zygomatic portion of the temporal musculature (Turnbull, 1970). This area is caudally limited by a conspicuous nuchal crest that runs above the level of the suprameatal foramen and the external acoustic meatus.

The postglenoid foramen is placed at the base of the medial wall of the postglenoid process, rostral to the external acoustic meatus and rostroventromedial to the suprameatal foramen (figs. 4, 7, 9). It is slightly smaller than the suprameatal foramen and elliptical, with its major axis rostromedial-caudolaterally oriented. This foramen would have transmitted the capsuloparietal emissary vein (postglenoid vein) and the vein of the suprameatal foramen (Archer, 1976). There is a narrow groove, medial to the postglenoid process, running completely over the squamosal but near the sphenosquamosal suture. This groove connects the lateral area of the tympanic cavity ventral to the epitympanic recess and tegmen tympani (see petrosal) and runs rostrally toward the lateral surface of the alisphenoid (figs. 4, 7–9). This should represent the roof of the Glaserian fissure that transmits the chorda tympani (a ramus of the facial nerve or cranial nerve VII). As mentioned above, medial to the surface for the Glaserian fissure, the squamosal bears a longitudinal crest, the entoglenoid process, which, along with the tympanic process of the alisphenoid, forms the preotic crest (see Sphenoid complex).

The external acoustic meatus is a deep excavation on the skull that connects to the middle ear cavity. In PVL 7792, it is relatively wide laterally but becomes slightly narrow medioventrally (figs. 3, 7, 10). The rostral wall of the external acoustic meatus is defined by a short crest, caudomedially oriented, that rises caudal to the suprameatal foramen and forms the caudal edge of that aperture. The lateral half of the crest also forms the caudal wall of the groove connecting the suprameatal foramen (ventrally open) to the postglenoid foramen. The medial half of the crest becomes independent of the groove, forming a structure separated from the caudal lip of the postglenoid foramen.

On the posterior wall of the external acoustic meatus, the posttympanic process of the squamosal is flat on its rostral wall and relatively low (compared to the postglenoid process) and adjacent to mastoid exposure of the petrosal (figs. 8, 10).

The squamosal surface on the right side of PVL 7792 is broken, exposing part of the internal structure of this bone (fig. 9). At least two large gaps are visible. Rostrally, there is a concave wall, forming the medial surface of a cavity enclosed within the postglenoid process. This probably represents the rostroventral end of the temporal sinus or the space for the capsuloparietal emissary vein, also referred to as the postglenoid vein (Martínez et al., 2020) or retro-articular emissary vein (MacPhee et al., 2021). A mostly smooth wall evidences the second internal gap, caudal and dorsal to the latter. This surface also shows some low crests, mostly associated with a large aperture that leads into the tympanic cavity. All these structures indicate the presence of a well-developed epitympanic sinus, with the aperture corresponding to the aditus or pneumatic foramen connecting the sinus to the tympanic cavity proper. A possible connection between these cavities is difficult to ascertain, given the poor preservation of some areas of the inner walls.

Ectotympanic: The ectotympanic of PVL 7792 is partially preserved on both sides of the skull, but this description is mostly based on the left side, which is more complete (figs. 3, 4, 7, 8). The arrangement is also different on each side, due to the unequal effects of diagenetic deformation on the basicranium. On the right side, the ectotympanic (although incomplete) is separated from the central stem at the level of the basioccipital, revealing a large basicapsular fenestra (figs. 7, 9). In contrast, the ectotympanic on the left side is appressed against the basioccipital due to preservation. This bone is small and gracile, crescentlike in lateral view, and somewhat kidney shaped in cross section (mostly in the caudal half), without lateral development (figs. 4, 7–9). Hence, this element forms neither bulla nor ventral wall of the external acoustic meatus.

FIG. 8.

Neobrachytherium intermedium PVL 7792, detail of caudal cranium in (A) left lateral and (B) ventrolateral views.

Being crescentlike, the ectotympanic is clearly divided into two “legs” or crurae. The rostral crus of the ectotympanic contacts the preotic crest (formed by the alisphenoid and squamosal; see above) at the level of the caudal edge of the glenoid fossa by means of a straight and oblique suture. The caudal crus meets the rostromedial surface of the tympanohyal and the rostral surface of the paracondylar process (exoccipital). In lateral view, the rostral crus bears a rostral protuberance that protrudes rostroventrally, forming an almost straight angle with rounded vertex: the styliform process of the ectotympanic (fig. 8). The ventral surface of the ectotympanic of PVL 7792 is smooth (fig. 7).

The internal surface of the ectotympanic ring shows two sulci separated by a longitudinal crest. The lateral sulcus occupies most of the inner surface of the ectotympanic, while the medial one is restricted to the caudal crus. Considering its extent, we interpret the lateral sulcus as the sulcus tympanicus, the site of attachment of the tympanum (by means of the fibrocartilaginous ring; MacPhee, 1981). Thus, the mentioned crest is the crista tympanica, developed medially to the latter.

FIG. 9.

Neobrachytherium intermedium PVL 7792, detail of caudal cranium in (A) right lateral and (B) ventrolateral (B) view.

Petrosal: The petrosal (paired endochondral bone) houses the organs of hearing and equilibrium and provides area of attachment for the muscles and ligaments of the middle ear ossicles. In PVL 7792, the petrosal contacts the alisphenoid anterolaterally and the basioccipital medially by means of cartilaginous tissues (not preserved but evidenced by the piriform fenestra), the squamosal laterally, and the exoccipital posteriorly (fig. 11A, B).

Both petrosals are preserved. Diagenetic deformation on the left side resulted in breakage separation of this element in two halves. The rostral half includes the pars cochlearis and a segment of the pars canalicularis and is displaced inside the cranial cavity. The caudal half includes most of the pars canalicularis, with the tympanohyal area and the mastoid exposure still attached to the external surface of the basicranium and occiput (figs. 9, 11A, B). On the right side, deformation was less extensive. However, the tympanic aspect of the petrosal is heavily weathered; hence, most of the following description is based on the reconstruction of the left petrosal (fig. 11C–E). Although a tridimensional model of both petrosals was generated, the low resolution of the tomographies hampers a completely accurate assessment of the subtler and smaller traits of this element.

FIG. 10.

Neobrachytherium intermedium PVL 7792, detail of (A) occiput in posterior view and (B) caudal cranium in right ventrolateral view.

Tympanic surface: The promontorium is the most conspicuous structure of the pars cochlearis; it is weakly inflated, located at the rostromedial part of the petrosal, and projected ventrally (fig. 11E). It presents the shape of a hemiellipsoid, whose major axis is oriented rostrocaudally. The caudal portion of the promontorium seems to be partially damaged, hampering an estimate of its real dimensions. Its rostral apex is small and almost flat, becoming gradually indistinguishable near the rostral edge of the petrosal. The epitympanic wing is not discernable as a discrete outgrowth on the rostral end of the petrosal outline, and the medial flange is not extended in a bladelike structure as in most notoungulates (Billet and Muizon, 2013; García-López et al., 2018). Instead, we recognize thickened rostral and ventral edges (fig. 11D–E). There is a marked bump projected ventrally on the promontorium surface, here interpreted as the rostral tympanic process (fig. 11D–E). In the ventromedial aspect of the left petrosal there is a fracture line that can be also observed on the cerebellar surface but is absent on the right petrosal. The promontorium outline is more evident on its caudal portion around the fenestrae and, as mentioned above, the protuberance corresponding to the promontorial surface gently disappears toward the rostral side of the petrosal.

The opening of the hiatus Fallopii (the exit of the grater petrosal nerve) faces the rostral edge of the petrosal and it is located on the tympanic surface (fig. 11D, E). Its aperture is elliptical and particularly large (slightly larger than the fenestra vestibuli). This large diameter may indicate that the caudolateral wall of the hiatus Fallopii (floor of the cavum supracochleare, which encloses the geniculate ganglion of the facial nerve; Voit, 1909) is broken, precluding assessment of its true morphology; however, the presence of a large aperture can be seen on both petrosals, indicating that this is probably the actual condition (though our observations and comparisons of the hiatus Fallopii must be taken cautiously and are subject to confirmation after the study of new specimens). Medioventral to the location of the opening of the hiatus Fallopii and just anterior to the fenestra vestibuli there is a large fossa on the tympanic surface of the promontorium that most probably represents the m. tensor tympani fossa (fig. 11D, E). The rostrolateral and caudomedial edges of the fossa are parallel (with the later particularly marked); thus, the fossa shows a rostrocaudal (and slightly oblique) elongation and a somewhat quadrangular outline.

There is a small opening caudal to the hiatus Fallopii and laterally adjacent to the fenestra vestibuli. We interpret this as the secondary facial foramen (through which the facial nerve exits the petrosal towards the tympanic cavity; Billet and Muizon, 2013). This foramen is subcircular, about one third the size of the fenestra vestibuli, and opens dorsolaterally (fig. 11E). The facial sulcus for the facial nerve (MacPhee, 1981) is not well differentiated, likely due to low resolution. It is a wide and barely concave longitudinal depression, although it shows a transverse convexity near its longitudinal midpoint. The crista parotica, which borders the facial sulcus laterally, is barely visible on the tridimensional model. It is evidenced only by a robust, low elevation that runs straight, reaching the caudolateral corner of the stapedial fossa. From this point, it becomes more robust and ventrally projected and extends caudomedially. The caudal end of the crista parotica evidently corresponds to a breakage surface, since the areas of the tympanohyal, stylomastoid notch, and mastoid petrosal exposure are still attached to the basicranial surface, well separated from the rest of the petrosal (figs. 9, 11E).

The facial sulcus is caudally continuous with the stapedial fossa (for the m. stapedius; Wible et al., 2001). This is an oval depression, more marked than the facial sulcus, which is strictly located caudal to the fenestra vestibuli and cochlear fossula, dorsal and caudolateral to the postpromontorial tympanic sinus (see below), and laterocaudally bordered by the caudal end of the crista parotica (fig. 11E).

The fenestra vestibuli and external aperture of the cochlear fossula are located on the caudolateral and caudal side of the promontorium, respectively. The fenestra vestibuli is oval with its major axis rostrocaudally oriented and has a stapedial ratio of 1.78 (following Segall, 1970). This opening is rather small compared with the size of the promontorium (fig. 11E). The aperture of the cochlear fossula seems to have a semicircular shape, but its real outline is uncertain, since the caudal portion of the promontorium (on the zone of the medial edge of the fossula) seems to be broken. It is distinctly larger than the fenestra vestibuli and both are separated by a wide and robust crista interfenestralis (Wible et al., 1995). The postpromontorial tympanic sinus (sensu Wible et al., 2009) borders the aperture of the cochlear fossula caudally and occupies a small area. This surface is oval with its major axis mostly dorsoventrally oriented. Caudolaterally, it is bordered by a robust crest that represents a caudal projection of the crista interfenestralis and ends at the caudal tympanic process (see Ekdale et al., 2004; Wible et al., 2004; Billet et al., 2015).

As mentioned, the tympanohyal is separated by breakage from the pars cochlearis and attached to the mastoid petrosal process on the basicranial surface (figs. 7–9). It is moderately robust, cylindrical with a broadened ventral edge, and slightly curved rostromedially. The tympanohyal makes up the lateral surface of the stylomastoid notch (for the exit of cranial nerve VII), formed also by the mastoid surface of the petrosal (fig. 11E).

At the rostrolateral corner of the petrosal, the tegmen tympani (petrosal tympanic area rostrolateral to the cavum supracochleare and facial sulcus and anterior to the epitympanic recess sensu Billet and Muizon, 2013) is visible as a robust and barely inflated tuberosity that reaches the area near the aperture of the hiatus Fallopii (fig. 11D, E). Apparently the tegmen tympani is not completely preserved, and its actual morphology is probably distorted by the low resolution of the tridimensional model. Nevertheless, there is no evidence of a tegmen tympani canal (for the passage of the ramus superior of the stapedial artery; Wible et al., 2001; Billet et al., 2015), which should be composed of part of the cavum supracochleare.

Caudolateral to the secondary facial foramen is a shallow and subcircular depression referred to as the medial portion of the epitympanic recess (an excavation in the roof of the middle ear that lies superior to the head of the malleus and the body of the incus; Sisson, 1911; Ekdale et al., 2004). This fossa is bordered medially by the crista parotica and rostrally by the tegmen tympani process. At the caudal end of the epitympanic recess, there is a circular fossa that we interpret as the fossa incudis (fig. 11E). It lies in line with the middle portion of the facial sulcus, caudolateral to the fenestra vestibuli.

Cerebellar surface: Dorsally, the crista petrosa is well defined and mostly evident on the rostral edge of the subarcuate fossa and the rostromedial edge of the cerebellar surface (fig. 11C, D). The prefacial commissure is a conspicuous bulge, ventrally limited by a subtle notch on the crista petrosa.

The cerebellar surface contains two conspicuous structures: the subarcuate fossa, an excavation in the dorsomedial surface of the pars canalicularis, and the internal acoustic meatus, situated on the dorsomedial surface of the pars cochlearis (fig. 11C). The subarcuate fossa (that accommodates part of the parafloccular lobe of the cerebellum; O'Leary, 2010) is apparently partially preserved, with only its ventral half visible on the tridimensional model. This preserved part is relatively shallow and wide, with its maximum concavity dorsolateral to the level of the internal acoustic meatus dorsal edge. There are no signs of the presence of the petromastoid canal. The caudolateral rim of the subarcuate fossa (overlying the posterior semicircular canal) is a wide crest oriented dorsoventrally, and the caudomedial rim, which overlies the crus commune (the junction of the anterior and posterior semicircular canals; Wible, 2003), is a conspicuously robust bulge. Ventrally, the subarcuate fossa is bounded by a strong, rostrodorsally-caudoventrally oriented crest, separating it from the internal acoustic meatus. This longitudinal crest almost reaches the rostral edge of the petrosal, allowing the dorsal boundary of the prefacial commissure to be distinguished (fig. 11C).

FIG. 11.

Reconstructed CT scan of Neobrachytherium intermedium PVL 7792, showing locations of the petrosal bones (A–B) and reconstruction of the left petrosal (C–E). A, dorsolateral view of the left side of the cranium (left) and translucent digital rendering (right) showing the position of both petrosal bones. B, ventral view of the cranium (left) and translucent digital rendering (right) showing the position of both petrosal bones. Note that the right petrosal is broken and has been displaced by diagenetic deformation (A–B); the position of the stylohyoid bone prior to partial removal is also evident (B). C, cerebellar; D, rostral; and E, lateroventral view of the digital rendering of the reconstructed left petrosal. Dashed line indicates a fracture area.

The internal acoustic meatus is oval with its major axis dorsoventrally oriented. There are two rounded foramina within the meatus, the foramen acusticum superius and the foramen acusticum inferius (Wible et al., 2004), which are separated at a moderate depth by a conspicuous crista transversa (fig. 11C). This crest is slightly oblique to the longitudinal crest that limits the subarcuate fossa ventrally, forming an acute angle. The foramen acusticum superius contains the canal for the passage of the facial nerve (cranial nerve VII) toward the cavum supracochleare in the middle ear region (O'Leary, 2010; Billet and Muizon, 2013; Billet et al., 2015). The foramen acusticum inferius contains a deep area rostrally that may be interpreted as the tractus spiralis foraminosus, which gives passage to fibers of the cochlear nerve (part of cranial nerve VIII; Meng and Fox, 1995; O'Leary, 2010). Caudal to the foramen acusticum inferius, a smaller opening faces rostrally. It could represent the foramen singulare (Wible, 2010; Billet and Muizon, 2013); however, this location would be unusual because the opening is adjacent to the foramen acusticum inferius rather than within it (fig. 11C).

The caudomedioventral edge of the cerebellar aspect is characterized by the presence of a large notch (jugular notch; MacIntyre, 1972; O'Leary, 2010), forming a conspicuous right angle (fig. 11C). Although the cochlear canaliculus (passage of the perilymphatic duct; Cifelli, 1982; Billet and Muizon, 2013; Billet et al., 2015) is often related to the jugular notch, we cannot observe any obvious aperture in this case, probably due to the low resolution of the model.

There is a large slit located dorsolaterally on the cerebellar face, specifically on the caudolateral rim of the subarcuate fossa (fig. 11C). It represents the aperture of the vestibular aqueduct (passage for the endolymphatic duct; Cifelli, 1982; Billet and Muizon, 2013; Billet et al., 2015) and opens to the caudal aspect of the petrosal.

Petrosal mastoid process: There is a relatively narrow mastoid exposure of the petrosal visible on the lateral edge of the occiput between the squamosal and exoccipital (fig. 9). Although discernible, this surface is highly damaged and no features can be clearly observed. The exposed surface is developed as a vertical osseous slit, dorsally continuous with the nuchal crest.

Occipital complex: This complex (endochondral bones) is composed of the basioccipital (single), the exoccipital (paired), and the supraoccipital (single). In PVL 7792, only the basioccipital and part of the exoccipital are preserved.

The basioccipital forms the floor of the basicranium (as the caudal part of the central stem) and generally contributes to the medialmost portion of the occipital condyles and the ventral border of the foramen magnum in mammals. In PVL 7792, it is a thick bone of rectangular ventral outline (fig. 4). The basioccipital contacts the basisphenoid rostrally, and laterally it is adjacent to the tympanic cavity and associated elements. Laterocaudally, the basioccipital is fused with the exoccipital (there is no trace of the synchondrosis intraoccipitalis basilateralis).

In PVL 7792, the basioccipital shows a liplike elevation on the sphenooccipital synchondrosis. Moreover, the ventral basioccipital surface presents a conspicuous crest in the midsagittal plane (median keel; figs. 4, 7). This crest disappears caudally, near the occipital condyles. In notoungulates, Billet et al. (2009) pointed out that the median keel separates two fossae that might be for the origin of the m. rectus capitis ventralis (see Barone, 1999, for the insertion area of this muscle in domesticated mammals). The basioccipital surface is very rough and the laterocaudal border shows a longitudinal crest (visible on the right side of the skull; fig. 8) extending from the level of the petrosal promontorium (rostrally) to the level of the jugular notch (caudally). This crest probably corresponds to the area of attachment of muscles of the neck (m. rectus capitis; Scott, 1910), similar to the median keel. The crest also marks the ventral edge of a longitudinal depression, the ventral condyloid fossa (figs. 8, 10). It is developed as a wide sulcus that is rostrocaudally oriented with a small hypoglossal foramen located on the caudal end. The aperture of the hypoglossal foramen (almost hidden in ventral view) is oriented rostrally and laterally (fig. 10); it would have transmitted the hypoglossal nerve (cranial nerve XII) and accompanying vein, which connects the condyloid and vertebral veins (Wible, 2003). The ventral condyloid fossa is caudally limited by a transverse crest separating it from the dorsal condyloid fossa (see below). This crest extends from the lateral surface of the occipital condyle to the medial surface of the paracondylar process (fig. 10).

There is a notch on the rostral end of the ventral condyloid fossa that could correspond to the caudal edge of a jugular foramen (figs. 4, 10). This foramen would have opened between this notch and the petrosal and may have transmitted the glossopharyngeal, vagus, and accessory nerves (cranial nerves IX, X, and XI, respectively), as well as a small branch of the sigmoid sinus (see Evans, 1993; Wible, 2003).

In ventral view, the odontoid notch (= incisura intercondyloidea following Evans, 1993; figs. 4, 7), between the occipital condyles (of which only the left is entirely preserved), is V-shaped and bears a small, median groovelike shelf for the dens of the axis (second cervical vertebra).

Finally, the lateral edge of the basioccipital is separated from the petrosal medial surface by a gap, presumably corresponding to the basicapsular fenestra (figs. 4, 8, 10). This is a large gap only visible on the right auditory region; on the left side, the petrosal is pressed against the basioccipital, presumably by diagenetic deformation. However, as deformation has affected both sides of the cranium, the real development of this aperture cannot be accurately ascertained.

The exoccipital contacts the basioccipital medially and the squamosal laterally in ventral view, but the suture with the exoccipital is not clearly defined (fig. 4, 10). In caudal view, the exoccipital articulates with the supraoccipital dorsally and squamosal dorsolaterally (not preserved in PVL 7792).

The most characteristic structures of the exoccipitals are the occipital condyles for articulation with the atlas. The preserved left occipital condyle is oval, with the articular surface obliquely arranged and protruding caudally and ventrally (figs. 3, 4, 7, 9).

Laterally, the exoccipital forms the paracondylar process. This structure is incomplete on both sides of PVL 7792; therefore, its dorsoventral extent cannot be assessed (figs. 8, 10). The base of this process is robust, narrowing apically. The lateral aspect of the paracondylar process is convex, and its medial surface is slightly concave.

The dorsal condyloid fossa is located on the medial surface of the base of the paracondylar process (fig. 10). In caudal view, it is elliptical with its major axis obliquely oriented (dorsomedial-ventrolateral).

The original shape and dimensions of the foramen magnum cannot be defined since its dorsal walls are completely broken (fig. 10).

DESCRIPTION OF UPPER DENTITION

The upper dentition of the studied specimen is represented by well-preserved postcanine teeth (the incisors are not preserved; fig. 12). All erupted premolars are deciduous, the permanent teeth below them having been revealed by the CT scans (fig. 13A–D). The dP1–4 are complete, and wear decreases from dP1 to dP4, suggesting the eruption sequence; however, this cannot be clearly established, since deciduous teeth are well worn (fig. 12; see comments on ontogeny in discussion section). Molarization increases distally in the deciduous toothrow. The dP1–2 are longer than wide (fig. 12, table 3); their occlusal surfaces have two small enamel fossettes, with the anterior larger than the posterior and more marked in dP1 than in dP2. The anterior fossette might be serially homologous to the anteroposterior groove of more molarized teeth. The dP1 also presents small and shallow accessory medial and lingual fossettes (fig. 12). Both dP1 and dP2 have a lingual groove (better marked in dP2) that divides the tooth into two lobes, the posterior being wider than the anterior. The buccal faces have three styles, with the metastyle and parastyle more developed than the mesostyle. These features are also more conspicuous in dP2 than in dP1 (fig. 12). Additionally, dP1 shows a small vertical fold on the buccal wall, located between the mesostyle and metastyle; this structure might represent the metacone column and is not distinct in the dP2. Molarization is conspicuous in dP3–4, which are also quadrangular. The three buccal styles are subequally developed (with a slightly smaller metastyle) and divided by markedly concave paracone and metacone buccal walls. The protocone is mesiodistally extended and larger than the hypocone; both are well developed and separated by a wide posterolingual groove (fig. 12). Mesially, the protocone contacts a small paraconule, more distinct in dP4. The anteroposterior groove is smaller than the central fossette, although both have a similar depth. The occlusal development of paracone and metacone, plus the stylar cusps, results in the typical W-shaped ectoloph of proterotheriids and other litopterns. At the base of the crown, between the paraconule and protocone, there is a conspicuous anterolingual cingulum (fig. 12). In the case of dP3, the anterolingual cingulum occupies the anterolingual corner of the tooth, reaching the level of the center of the protocone surface lingually. As for dP4, this cingulum is more restricted to the anterior wall; however, a lingual cingulum is also present on this tooth, which occupies the base of the lingual wall of the protocone, reaching the posterolingual groove posteriorly. The cingula are discontinuous in dP4 although very close and barely separated (fig. 12). Distally, the metaconule is lophoid and transverse, connecting the metacone to the crest running between protocone and hypocone.

Regarding the molars, M1 is almost completely erupted, M2 is covered by a thin layer of bone, and there is no sign of M3 (fig. 12). The M1 resembles dP4 but clearly shows less wear plus some morphological differences. The lingual wall of M1 is less sloped and has a larger surface than in dP4; also, there is no trace of the lingual cingulum at the base of the protocone on the molar. The anterolingual cingulum is stronger on M1, showing a cusplike lingual end (protostyle; fig. 12). On the buccal side, the parastyle and mesostyle are well marked and sharp, extending far more buccally than in dP4. Finally, the arrangement of the metaconule is also different, being mostly transverse on dP4 but showing an anterobuccal-posterolingual orientation on M1, thus pointing to the mesial wall of the hypocone instead of the crest between this cusp and the protocone.

FIG. 12.

Neobrachytherium intermedium PVL 7792, detail of the dentition in occlusal view. Note that the mesostyle is bifid in left dP3 but simple in right dP3 and that the M2 is only partially erupted (indicated by an asterisk). Dental terminology is illustrated on a drawing of left M1.

FIG. 13.

Deciduous and permanent teeth of Neobrachytherium intermedium in PVL 7792 (A–D) and FMNH-P 14483 (E). Teeth in eruption are indicated by an asterisk (*). A, dorsolateral view of the left side of the cranium. B–C, lateral to medial sequence of parasagittal sections at the level of the right toothrow. D, detail of the right toothrow at the level of the sagittal plane of the dP1. E, lateral (top) and occlusal (bottom) views of N. intermedium (FMNH-P 14483). Note that the unerupted P3 is almost complete and the unerupted P4 is barely visible in PVL 7792 (B–D), whereas the P3 is exposed and the P4 seems to be close to erupting in FMNH-P 14483, which may be a bit older (E).

The M2 on the left side was uncovered by breakage, revealing part of the occlusal morphology (fig. 12). The visible part of M2 is virtually indistinguishable from the morphology of M1. As this tooth shows no wear, it can be observed that the apex of the paracone is higher than that of the metacone, though this condition might be related to the fact that the crown is still buried within the maxilla and not in its erupted position.

DISCUSSION

Comparisons of Cranial Features, Anatomical Inferences, and Phylogenetic Insights Neobrachytherium

The assignation of the specimen PVL 7792 to Neobrachytherium intermedium is based on the following characters: dP1–2 longer than wide; molariform dP3–4 with incipient lophoid metaconule interrupting the anteroposterior groove and well-developed buccal folds of paracone and metacone; M1–2 with parastyle and mesostyle better developed than metastyle, well-developed protocone and hypocone, reduced paraconule, and lophoid metaconule. These traits are considered as diagnostic for this taxon (Soria, 2001; Schmidt et al., 2024).

Although not complete, the remarkable preservation of the cranium of specimen PVL 7792 allowed us to study its morphology in detail. This is particularly true for areas such as the orbitotemporal region, the mesocranium (referring to the ventral surface between the choanae and the rostral limit of the otic region), and some other basicranial structures, which are usually absent in most proterotheriid species or have been overlooked in several previous studies. The description and interpretation of these cranial features in a juvenile Neobrachytherium intermedium permits comparisons that will aid in placing this species and the family Proterotheriidae within the context of eutherian cranial evolution. In the discussion that follows, we have selected specific characters that are conspicuous enough to allow for direct comparisons with other taxa. Some of these traits seem to follow a generalized pattern, while others may represent particular features that should be considered for inclusion in future phylogenetic analyses.

A very clear feature of the rostrodorsal area of the cranium is the presence of a frontal foramen associated with a large frontal sulcus (note that we use the term supraorbital foramen to refer to a different structure; see Frontal). Although the sulcus appears to be incompletely developed in the juvenile (see below), this structure is conspicuous and occupies a large portion of the frontal and nasal bones. As for the distribution of these traits, a frontal foramen is usual for several taxa, but the frontal sulcus varies in its development or can even be absent. For instance, the sulcus and foramen have been observed in members of the extinct clade Pantodonta such as Alcidedorbignya inopinata (Muizon et al., 2015). Among SANUs, notoungulates such as Protypotherium australe and Colbertia lumbrerense seem to lack the sulcus; some members of that order show a faint depression anterior to the frontal foramen (e.g., Simpsonotus praecursor, Dolichostylodon saltensis, Hegetotherium mirabile), while others exhibit a well-developed sulcus, such as Griphotherion peiranoi (see García-López and Powell, 2011) and Archaeohyrax suniensis (Billet et al., 2009). Similarly, a foramen and associated sulcus have been observed in PVL 7697, a specimen referred to the putative astrapothere Eoastrapostylops riolorense. In this case, although the sulcus is barely visible, it is clearly present. Among Litopterna, these frontal traits can be found in Paleogene forms such as Notonychops powelli and Adiantoides leali (once regarded as a separate clade, Notopterna, but more likely members of Litopterna; see discussion in Saade et al., 2023). The former shows a well-developed sulcus comparable to that of PVL 7792, but comparisons are hampered because the specimen is incompletely preserved. In Adiantoides, the frontal foramen is clearly present, but the sulcus is faint. In the holotype of another purported Notopterna, Indalecia grandensis (PVL 4186), frontal foramina can be observed, but the presence of frontal sulci cannot be assessed because the skull surface is broken rostral to those apertures. Frontal foramina can be observed in some representatives of Macraucheniidae, although a frontal sulcus usually is not particularly evident (e.g., Cramauchenia normalis, Huayqueriana cristata, Promacrauchenia antiquua, and P. calchaquiorum; Dozo and Vera, 2010; Forasiepi et al., 2016). Finally, the condition of a well-developed frontal foramen and sulcus is common to other proterotheriids (e.g., Thoatherium minusculum MACN-A 2996; Tetramerorhinus cingulatum MACN-A 5971; Eoauchenia primitiva according to Soria, 2001). A similar arrangement of these structures is present in some extant mammals, particularly artiodactyls such as pigs and ruminants, with the frontal sulcus regarded as sulcus supraorbitalis (NAV; Barone, 1999). In summary, despite its widespread distribution among mammals, the presence of this foramen and sulcus (as well as postorbital bar) contributes to superficial resemblances between the skull of Neobrachytherium (and other proterotheriids) and extant forms such as small cervids.

TABLE 3.

Upper tooth dimensions (in mm; right/left) of Neobrachytherium intermedium and inferred ontogenetic stage based on cranial morphology and timing of tooth eruption. *, approximate; A, adult individual; J, juvenile individual; L, length; S, subadult individual; W, width.

Another character widely distributed among eutherians and present in Neobrachytherium is the robust tubercle marking the choanal border known as the postpalatine torus. Despite the widespread occurrence of this structure, it has been used to distinguish among species in certain groups including sparassodont metatherians (Engelman et al., 2020), leptictids (Novacek, 1986), and pantodonts (Muizon et al., 2015). In the present case, it should be noted that, while clearly developed in the juvenile specimen of Neobrachytherium intermedium, it appears to be absent in subadults of the same taxon (FMNH-P 14361), and it is absent in adults of Neobrachytherium morenoi (MACN-Pv 8428). This structure is also present in a juvenile specimen of Diadiaphorus majusculus (MACN-A 9174-80). However, in the adult specimen MACN-A 9200 of the same taxon, there are only two small tubercles instead of a well-developed torus, suggesting an ontogenetic trend similar to that observed in N. intermedium. In any case, the distribution of a postpalatine torus in proterotheriids should be examined in greater detail to determine whether its presence or absence may be useful for assessing the taxonomic integrity of the genus Neobrachytherium.

As mentioned before, evaluating the structures of the orbitotemporal fossa and mesocranium in SANUs is often challenging. In this study, we were able to observe several structures that have been discussed and garnered attention in recent contributions on SANU cranial anatomy. These mostly include different structures of the mesocranium and prominently marked grooves for the passage of nerves and blood vessels.

A large and elliptical foramen ovale, caudally located in the mesocranium, is a feature of Neobrachytherium and other proterotheriids (e.g., Tetramerorhinus MACN-A 8970; Epitherium MACN-Pv 8001). In contrast, we identified no separate foramen rotundum in the juvenile specimen of Neobrachytherium intermedium described here. Therefore, the maxillary branch of the trigeminal nerve (V2) must have exited the cranial cavity via the sphenorbital fissure. This arrangement has been documented in several other mammal groups, and Novacek (1986) proposed that it is the plesiomorphic condition for eutherians. He later indicated that having a distinct foramen rotundum is “most characteristic of edentates, tupaiids, and euprimates” (Novacek, 1993: 458). More recently, Muizon et al. (2015) argued that it likely represents the plesiomorphic condition for therians considering that the condition of a foramen rotundum well separated from the sphenorbital fissure is widespread among early eutherians and metatherians. These authors also suggested that most groups of SANUs lack a separated foramen rotundum (i.e., foramen rotundum and sphenorbital fissure are coalescent), a condition possibly shared by some litopterns that would represent a derived state (Muizon et al., 2015: 546). MacPhee et al. (2021) did not observe a separate foramen rotundum in Trigonostylops (in accordance with Simpson, 1967, and Muizon et al., 2015) nor the astrapotheriid Granastrapotherium and pointed out some ontogenetic variations in the sphenorbital fissure and foramen rotundum in different perissodactyls. Forasiepi et al. (2016) confirmed this condition for Huayqueriana and other macraucheniids, supporting observations previously made by Simpson (1933) and Fernández de Álvarez (1940). Our study corroborates this condition in Neobrachytherium and other proterotheriids (e.g., Epitherium laternarium MACN-Pv 8001; Tetramerorhinus cingulatum MACN-A 8666), as well as the putative early litoptern Indalecia grandensis (PVL 4186). Thus, the absence of a foramen rotundum may represent a shared and distinctive condition among SANUs, especially considering that other Paleogene South American mammal groups (e.g., pantodonts) show the opposite (plesiomorphic) condition. Further studies focusing on the relationships among these South American mammals are necessary to clarify these issues.

Neobrachytherium intermedium exhibits a large piriform fenestra. This aperture has received some attention concerning its expression in other SANUs, particularly notoungulates (MacPhee, 2014). It is variably developed among other mammals. In extant perissodactyls (e.g., Equus; Sisson and Grossman, 1975; Tapirus, Moyano and Giannini, 2017), the fissure is broad, defining a medially extended piriform fenestra that notches the lateral edge of the basisphenoid (and usually more medial than observed in PVL 7792). A similar arrangement is observed among Neogene litopterns, particularly in forms such as Scalabrinitherium (MACN-Pv 13082; Forasiepi et al., 2016), but the petrosal forms a tighter junction with the alisphenoid in some other macraucheniids (e.g., Promacrauchenia MACN-Pv 7986, MACN-Pv 5528; Huayqueriana IANIGLA-PV 29) than in proterotheriids (Epitherium MACN-Pv 8001; Diadiaphorus MACN-A 9200-08; Tetramerorhinus MACN-A 5971). Hence, a larger piriform fenestra seems to be more typical for proterotheriids, but this polarity remains to be evaluated.

Another notable feature of the mesocranium of PVL 7792 is the presence of a large and conspicuous groove, mainly on the surface of the alisphenoid, that marks the passage of the Eustachian tube and pterygoid canal, at least on its caudal portion (see Sphenoid complex). This groove has been observed in other specimens of Neobrachytherium intermedium (FMNH-P 14361), Tetramerorhinus cingulatum (MACN-A 5971; MACN-A 8666), and Epitherium laternarium (MACN-A 8001). Additionally, a similar structure was described by Forasiepi et al. (2016) in Macraucheniidae, who emphasized its large size. These authors suggest that the canal served as a passage not only for the nerve of the pterygoid canal but also possibly for an artery (vidian or artery of the pterygoid canal; Forasiepi et al., 2016: 33), given its large caliber. A similar osseous configuration is present in specimens of extant Mazama (PVL-A 81; PVL-A 82). In addition, we note that, in Mazama, the posterior part of the groove is divided between the Eustachian tube laterally and passage of nerves and possibly vessels medially.

There are notable similarities in the arrangement of the foramen ovale, Glaserian fissure, preotic crest, and groove of the alisphenoid in the specimen PVL 7792 of Neobrachytherium intermedium and the putative basal astrapothere Eoastrapostylops riolorense. Specifically, Kramarz et al. (2017) described a large and rostrally facing foramen ovale in this taxon. Additionally, they noted the position of the preotic crest (formed by the combination of the tympanic process of the alisphenoid and entoglenoid process of the squamosal) medial to the glenoid fossa and Glaserian fissure as well as the clear development and position of the alisphenoid groove (regarded in that contribution as groove for the Eustachian tube). Although these resemblances between E. riolorense and the juvenile N. intermedium might be superficial, they highlight the necessity of continued detailed comparative surveys of cranial morphology in different SANUs, even those considered a priori to be distantly related to the group in question.

The elements of the otic region (including the squamosal) exhibit several features that have been analyzed in other SANUs in different contexts. Some of these features do not have a clear functional correlation or a wide phylogenetic distribution and can be mentioned only as unique features of Neobrachytherium. For example, the tuberosities of the postglenoid process (see squamosal above) are present in other specimens of the genus (e.g., FMNH-P 14361) but absent in other proterotheriids, other litopterns, and notoungulates. Moreover, we have not observed this kind of tuberosities in a sample of extant mammals that included certain perissodactyls, artiodactyls, xenarthrans, and carnivorans.

Considering other osteological features, the specimen PVL 7792 shows the clear development of an epitympanic sinus and an aditus connecting this vacuity with the tympanic cavity. The development of the epitympanic sinus has been investigated in other SANUs, particularly Notoungulata. In this group, a greatly inflated sinus is manifest as a globose surface on the laterocaudal wall of the squamosal (termed epitympanic theca by MacPhee, 2014). Such an inflation is not present in PVL 7792, as can be seen on the right side of the cranium. Therefore, although the sinus is well developed, it protrudes externally less than in the average notoungulate. This arrangement confirms observations (e.g., Forasiepi et al., 2016; MacPhee et al., 2021) that an epitympanic sinus is present in proterotheriids and other litopterns even though it is not evident externally. Although the epitympanic sinus is not particularly large in proterotheriids, it is larger than in astrapotheriids (e.g., Astrapotherium), where a small cavity has been reported (MacPhee et al., 2021). The configuration also seems to differ from that of the pyrothere Pyrotherium, which shows a large epitympanic sinus and other similarities with early-diverging Notoungulata (Billet, 2010). In summary, an epitympanic sinus is a generalized character among SANUs, sometimes with subtle variations, and only notoungulates have conspicuous dorsolateral inflation covered by the epitympanic theca.

Perhaps the most striking superficial difference among the otic regions of proterotheriid and macraucheniid litopterns and other SANUs is related to the morphology of the ectotympanic. It is a ringlike element in the former clades but varies greatly in other groups. In forms such as Indalecia (García-López and Babot, 2014) and Trigonostylops (Simpson, 1933, 1967; MacPhee et al., 2021), this element is more expanded and forms a cover for the tympanic cavity but not an inflated bulla as in notoungulates. In the latter group, there is a massive tympanic bulla that is firmly attached to the surrounding elements (Patterson, 1934; Gabbert, 2004; MacPhee, 2014). The bulla is usually formed by the ectotympanic within this order (Gabbert, 2004), although an entotympanic may also be present in some taxa (see MacPhee, 2014). Pyrotheres exhibit a very different morphology from litopterns, with the ectotympanic also forming a bulla (or part of it; see Billet, 2010). The ectotympanic ring is small and gracile in Neobrachytherium, as it is in other litopterns in which this element is known, such as Epitherium laternarium (Soria, 2001), Theosodon (Scott, 1910: pl. XVII), and Huayqueriana cf. H. cristata (Forasiepi et al., 2016). Regarding more discrete features of the ectotympanic, the styliform process (rostral protuberance of the rostral crus) of PVL 7792 is similar to that of Huayqueriana cf. H. cristata (see Forasiepi et al., 2016: fig. 16). The ventral surface of the ectotympanic is smooth in PVL 7792, whereas two additional blunt protuberances are present in Huayqueriana cf. H. cristata. Other pan-perissodactyls have an expanded ectotympanic or one that forms a bulla and thus differ from proterotheriids and macraucheniids in a manner similar to other SANUs (MacPhee et al., 2021). In summary, a ringlike ectotympanic seems to be particular to at least the two main litoptern lineages (i.e., Proterotheriidae and Macraucheniidae) and differs from other SANUs as well as extant perissodactyls. However, more extensive studies are needed to determine the significance of this trait as a possible defining feature for some litopterns.

The petrosal is an important element of the otic region that has been studied in several contributions dealing with systematics and phylogenetics of SANUs (Billet and Muizon, 2013; Billet et al., 2015; García-López et al., 2018). Indeed, Billet et al. (2015) provided a detailed survey that considered the basal morphotype of the litoptern petrosal and stated some possible synapomorphies for the group. The petrosal of PVL 7792 is the first described in detail for Proterotheriidae, and some observations can be made on the general distribution of some character states within eutherians and SANUs, as well as the expression of features potentially shared among litopterns.

The epitympanic wing and medial flange of PVL 7792 are not well developed; they are barely distinguishable as discrete structures. The lack of a well-defined medial flange on the petrosal has also been mentioned for basal Litopterna (cf. Miguelsoria parayirunhor; Billet et al., 2015), Eoastrapostylops, and astrapotheres (but see Kramarz et al., 2017) among SANUs, as well as in several other eutherian groups, including extant perissodactyls (Billet et al., 2015; MacPhee et al., 2021). A shelf morphology for this flange is present only in some basal eutherians (e.g., Maelestes; Wible et al., 2009; Billet et al., 2015) and notoungulates, where there is at least a bladelike and usually extended medial flange. In sum, although this condition is absent in basal litopterns, late-diverging proterotheriids (i.e., Neobrachytherium), and several eutherians, it represents at least a clear difference with regard to Notoungulata among SANUs (see Billet and Muizon, 2013).

Another easily recognizable petrosal structure, the promontorium, shows some variation within Litopterna. In Neobrachytherium, the protuberance of the promontorium is particularly evident on its caudal portion (near the petrosal fenestrae) and gently diminishes rostrally. In other Litopterna, such as the petrosals referred to cf. Miguelsoria parayirunhor (see Billet et al., 2015) as well as Tetramerorhinus cingulatum (MACN-A 5971) and the specimen MNHN-F-SCZ 205 (referred to Proterotherium by Billet et al., 2015; but see Kramarz et al., 2017, for its reconsideration as Tetramerorhinus), the petrosal shows a more extended promontorial surface, although still less developed than in Notoungulata (Billet and Muizon, 2013) and apparently more similar to the condition in Eoastrapostylops and Astrapotherium (Kramarz et al., 2017; MacPhee et al., 2021). In this regard, they seem to represent features unique to Neobrachytherium. Among extant eutherians, this morphology is comparable to camelids such as Lama and Camelus (see O'Leary, 2010), as well as the cervid Mazama (PVL-A 81, PVL-A 82). Regarding Perissodactyla, the condition of Neobrachytherium is barely comparable to that of Equus caballus and Tapirus terrestris (PVL-A 74), where there is just a subtle bump evidencing the position of the promontorium (see O'Leary, 2010).

The hiatus Fallopii is also a very striking feature of Neobrachytherium, as it is expressed as a large aperture. Although a large hiatus is also observed in some eutherians, such as Suidae (O'Leary, 2010), it is poorly known among other SANUs. Remarkably, the hiatus Fallopii seems to be smaller in the basal litoptern cf. Miguelsoria parayirunhor, although Billet et al. (2015) did not include any comment on this condition. This aperture might be located on the tympanic side of the petrosal in Neobrachytherium intermedium (although there is some evidence of possible breakage obscuring this condition; see above), a derived feature if the plesiomorphic condition for Litopterna is an opening at the anterior edge of the petrosal, as observed in a wide array of genera, including proterotheriids such as Diadiaphorus and Tetramerorhinus (Billet et al., 2015; Kramarz et al., 2017). This location also differs from that observed in several notoungulates (e.g., Colbertia, Dolichostylodon, Griphotherion, Periphragnis), in which it is closer to the rostral edge of the petrosal and rostrolateral to the promontorium. A hiatus Fallopii located on the tympanic side has also been documented in Eoastrapostylops, Alcidedorbignya (Pantodonta), Pleuraspidotherium (“Condylarthra”), Tapirus, and Palaeotherium (Perissodactyla) (Kramarz et al., 2017; Mateus, 2018; MacPhee et al., 2021).

Some other petrosal characters show a generalized distribution among different ungulate groups. The configuration of the tympanohyal (moderately robust and cylindrical with a broadened ventral edge) and the petrosal mastoid process (narrow and vertically arranged strip between the squamosal and the exoccipital) is very similar to that observed in bovids (e.g., Capra) and cervids such as Mazama (PVL-A 04; PVL-A 81; PVL-A 82). Among SANUs, the arrangement in PVL 7792 is very different from the more gracile and strongly medially curved tympanohyal of Eoastrapostylops, in which the distal end contacts the caudal tympanic process of the petrosal. However, the morphology observed in the latter taxon might be an artifact of preservation (Kramarz et al., 2017). Regarding Trigonostylops, the most striking difference is the apparently partial ossification of the tympanohyal cartilaginous precursor (Reichert's cartilage; see MacPhee et al., 2021). In astrapotheriids such as Astraponotus and Astrapotherium, the tympanohyal seems to be shorter but much more robust than in Neobrachytherium and, at least in Astraponotus, it more closely contacts the paracondylar process (Kramarz et al., 2010, 2017). Pyrotherium exhibits a large tympanohyal similar to that of notoungulates (Billet, 2010) but very different from the smaller element of PVL 7792. The tympanohyal of Pyrotherium is also positioned in a hyoid recess, whereas no such recess is present in PVL 7792 considering the ringlike morphology of the ectotympanic, the element that usually forms much of the recess in SANUs in which it is present. As for notoungulates, the always well-developed bulla dictates that the zone of the tympanohyal is related to the presence of a well-marked hyoid recess. Since the tympanohyal itself is often absent in that recess, MacPhee et al. (2021) pointed out that this element might have not been completely ossified in most taxa within that group (as in Trigonostylops). Hence, the morphology of notoungulates is also very different from that of PVL 7792. Other litopterns also present some differences in the tympanohyal relative to Neobrachytherium; in macraucheniids such as Scalabrinitherium and Huayqueriana, the tympanohyal is a relatively robust element (Forasiepi et al., 2016), although the general arrangement and its relation with the ectotympanic is more comparable to PVL 7792 than in the groups previous discussed. Extant perissodactyls have a morphology more comparable to that of astrapotheres and notoungulates; MacPhee et al. (2021) mention that the tympanohyal is very robust in ceratomorph perissodactyls and partially ossified in Equus.

The external exposure of the petrosal mastoid process cannot be accurately evaluated in some SANUs, such as Pyrotherium (Billet, 2010) and Eoastrapostylops (Kramarz et al., 2017). The mastoid is usually not exposed in astrapotheres and notoungulates, being very reduced in the caudal cranium (Gabbert, 2004; Billet, 2011; García-López, 2011; Kramarz et al., 2011; García-López et al., 2018); when it is present, as in Paedotherium (see MacPhee, 2014), it forms an extended surface that is relatively wider than in PVL 7792. Other litopterns observed have an arrangement of the mastoid exposure similar to the specimen studied here, including subadult specimens of Neobrachytherium intermedium (FMNH-P 14361), and macraucheniids such as Scalabrinitherium (MACN-Pv 13082) and Huayqueriana cf. H. cristata (IANIGLA-PV 29).

On the cerebellar side of the petrosal, the presence of a longitudinal crest ventral to the subarcuate fossa and the shape and location of the aperture of the vestibular aqueduct are features widely distributed among litopterns and notoungulates. However, the aforementioned crest is much weaker in PVL 7792 than in other Neogene SANUs; the proterotheriid Diadiaphorus (MACN-A 9200-08) and the hegetotheriid notoungulate Paedotherium (see MacPhee, 2014) show a stronger development of this structure, evidenced by a conspicuous fold. The slitlike structure of the aperture for the vestibular aqueduct is apparently common to a wide range of eutherians, as can be observed in different notoungulates (MacPhee, 2014), other litopterns (Billet et al., 2015), xenarthrans (see Babot et al., 2012), and artiodactyls (e.g., Leptomeryx; see O'Leary, 2010).

As for the arrangement of the cochlear fossula and postpromontorial tympanic sinus and the separation of the latter from the stapedial fossa, the condition observed in PVL 7792 is similar to that of cf. Miguelsoria parayirunhor and other Litopterna (Billet et al., 2015). Nevertheless, these features have not been mentioned as characters typical for the order. Information about these petrosal features in other SANUs is limited. Billet and Muizon (2013) described a merged postpromontorial tympanic sinus in an isolated petrosal from the Itaboraí Basin (MNHN-F-BRD 23) and considered this arrangement to be a notoungulate synapomorphy. García-López et al. (2018) observed some separation between these structures in the petrosal of an indeterminate Toxodontia from the Eocene of northwestern Argentina (IBIGEO-P 12) but pointed out that such separation was not as strong as in known Paleogene litopterns (Billet et al., 2015). In other words, known notoungulates display a different arrangement of these structures from that observed in PVL 7792. This difference can be also found in extant perissodactyls (e.g., the postpromontorial tympanic sinus and stapedial fossa are merged and almost indistinct in Tapirus and Equus; MacPhee et al., 2021).

Among the several petrosal characters mentioned by Billet et al. (2015) as potential synapomorphies for Litopterna, the position and shape of the m. tensor tympani fossa and the notch in the jugular area of the petrosal are clearly observable in PVL 7792. Additionally, a low stapedial ratio can be mentioned among the features of the latter specimen (see below) and was considered among potential synapomorphies. The fossa for the m. tensor tympani is located medial to the cavum supracochleare in PVL 7792, a condition widely distributed among Litopterna (e.g., cf. Miguelsoria, Macrauchenia, Tetramerorhinus, Diadiaphorus) and regarded as a litoptern symplesiomorphy (Billet et al., 2015). However, it represents a clear difference from Notoungulata, in which part of this fossa is located more lateral relative to the cavum supracochleare. Billet and Muizon (2013) indicated this as a possible synapomorphy for Notoungulata. Considering the positions of the fossa for the m. tensor tympani described in previous contributions (e.g., Billet, 2010; Kramarz et al., 2017; MacPhee et al., 2021), the observation of this condition in Neogene Litopterna reinforces the perception of this feature as a trait particular to Notoungulata, at least among SANUs. In a wider context within Pan-Perissodactyla, the fossa for the m. tensor tympani is also located more laterally in Equus. As for Tapirus, the condition seems to be somewhat variable (Mateus, 2018; MacPhee et al., 2021).

A subquadrangular and longitudinally elongated fossa for m. tensor tympani has been suggested as a potential synapomorphy for Litopterna (e.g., Diadiaphorus, Tetramerorhinus, Macrauchenia), although this feature is also observed in some perissodactyls and artiodactyls, as pointed out by Billet et al. (2015). Moreover, the expression of this character seems to be variable in basal Litopterna; it is present in the Itaboraí Basin taxon Miguelsoria but not clearly expressed in the early Eocene species Indalecia grandensis of the Lower Lumbrera Formation in northwestern Argentina (García-López, personal obs.). In the latter case, at least part of the m. tensor tympani fossa seems to be medial to the cavum supracochleare, but it does not have a defined subquadrangular outline.

It is important to underscore the uncertain phylogenetic position of Indalecia grandensis and other Paleogene forms that were grouped in the clade Notopterna by Soria (1989). In this sense, the differently shaped fossa may be a trait that distinguishes Notopterna from Litopterna. The validity and definition of Notopterna are currently under revision (Saade et al., in prep.), and further analyses, especially of basicranial traits, will be undertaken. Although PVL 7792 displays the same condition observed in many other litopterns (including other proterotheriids and macraucheniids), the utility of this character state (subquadrangular and longitudinally elongated fossa for m. tensor tympani) for defining the order relies on its presence in Paleogene species that, regrettably, are still poorly known.

A similar context is evident for the notch observed in the jugular area of the petrosal, also known as the jugular notch (MacIntyre, 1972; O'Leary, 2010), which subtends a conspicuous right angle in PVL 7792. This morphology was mentioned by Billet et al. (2015) as a possible synapomorphy for Litopterna. While it is present and conspicuous in the petrosal described as cf. Miguelsoria parayirunhor by those authors (in addition to Macrauchenia, Tetramerorhinus, and Diadiaphorus), in Indalecia grandensis, the notch is less defined than in any litopterns for which the petrosal morphology is known. Once again, understanding the distribution and importance of this character relies on resolving the validity and phylogenetic position of notopterns.

A low stapedial ratio (i.e., below 1.80) may be another litoptern synapomorphy (Billet et al., 2015). The ratio calculated for PVL 7792 is close to this value (1.78) and consistent with that interpretation. Low stapedial ratios have been considered to be more typical of metatherians than eutherians, with the latter generally having values greater than 1.8 (Segall, 1970; Wible, 1990; Rougier and Wible, 2006). Litopterns would seem to be unusual among eutherians in this sense, though as pointed out by Billet et al. (2015), low stapedial ratios are more common than previously appreciated among eutherians. For instance, this ratio is 1.6 in “zhelestids” (Ekdale et al., 2004) and 1.71 in Prokennalestes (Wible et al., 2001). Babot et al. (2012) reported even lower values among extant xenarthrans, including 1.25 in Bradypus and 1.6 in Tamandua and Tolypeutes. Among other SANUs and Perissodactyla, values above 1.8 have been reported for Eoastrapostylops (putative astrapothere), Cochilius and Hegetotherium (Notoungulata), and Equus, while low values (<1.8) correspond to forms such as Astrapotherium (Astrapotheria), Notostylops (Notoungulata), and Tapirus (Billet et al., 2015; MacPhee et al., 2021). The broad distribution of this character state likely diminishes its potential phylogenetic value, though it is important to note that this and other phylogenetic hypotheses are limited by the poor current fossil record of several families (with most Paleogene groups known exclusively from teeth). Therefore, the distributions and relevance of several morphological traits observed in SANUs can easily rise or fall as new discoveries and surveys provide new information on these taxa, a fact acknowledged in a number of contributions (e.g., Deraco and García-López, 2016; García-López et al., 2018).

Brief Comments On ONtogenetic Variation In Neobrachytherium

We were able to identify certain ontogenetic variation affecting the development of specific cranial traits (not taking into account general cranial proportions and development of large structures like the sagittal crest, which typically vary across different ages following tooth eruption and muscular growth; e.g., Cassini et al., 2012, and references therein). Unfortunately, the known sample of other Neobrachytherium intermedium individuals of varying ages is quite limited, consisting of only four specimens that represent a juvenile (FMNH-P 14341; fig. 14A), subadults (FMNH-P 14483 and FMNH-P 14361; figs. 13E, 14B, respectively), and an adult (FMNH-P 14500; fig. 14C). Furthermore, the caudal half of the skull in FMNH-P 14500 has been extensively reconstructed in plaster, FMNH-P 14341 represents the rostral half of the skull, FMNH-P 14483 is a left maxillary fragment with P3 and M1–2, and FMNH-P 14361 represents the caudal half. Some observations can be made considering other species within the family and extrapolating observable variations in the case of N. intermedium. However, this approach presents significant challenges, as observed differences may reflect taxonomic distinctions rather than ontogenetic variation (see Schmidt et al., 2024, for a detailed taxonomic context).

One of the most variable features in Neobrachytherium is the frontal sulcus, which is developed over the frontal and nasal bones in PVL 7792. It is present but less pronounced than in the adult (FMNH-P 14500). However, this structure might also be influenced by intraspecific variation, since the sulcus is also more pronounced in FMNH-P 14341, which appears to be of similar age to PVL 7792 based on tooth eruption and degree of wear. Scott (1910) suggested that this sulcus deepened with age. Soria (2001) observed these ontogenetic differences among specimens of Thoatherium minusculum. It is also worth noting that this trait has been discussed in interspecific comparisons as well; for instance, the frontal sulcus is more ventrally curved in T. minusculum and ends in the maxilla, close to the contact with the nasal, while in Eoauchenia primitiva, the frontal sulci converge toward the internasal sulcus and then bifurcate rostrally, reaching the maxillary surface (Soria, 2001).

Another frontal trait that exhibits age-related changes in Neobrachytherium intermedium is the supraorbital foramen (see frontal section), which is positioned medially and more caudally in the adult (FMNH-P 14500). Interestingly, this location has also been observed in adults of other genera, such as Diadiaphorus (MACN-A 9200-08). Unfortunately, the condition in juvenile (FMNH-P 14341) and subadult (FMNH-P 14361) specimens of N. intermedium cannot be assessed, as this part is missing in the available materials.

The development of the lacrimal facial process also appears to be associated with ontogenetic changes. In both juvenile specimens of N. intermedium, PVL 7792 and FMNH-P 14341, the frontal and maxilla apparently have no or only very limited rostral contact (this cannot be fully corroborated, as the surface is broken). In the adult specimen FMNH-P 14500, the frontal and maxilla show a larger rostral contact, and the lacrimal is relatively smaller due to the growth of the surrounding rostral elements.

On the other hand, despite the poor preservation of the palate in other Neobrachytherium species, we note a significant difference between the juvenile PVL 7792 and adult specimens of Neobrachytherium (N. morenoi MACN 8428, Neobrachytherium sp. MMP-M 4657) and other proterotheriids (e.g., Diadiaphorus majusculus, MACN-A 9200-08 and MACN-A 9174-80; Epitherium laternarium, MACN-Pv 8001). In PVL 7792, the choanal edge reaches the level of the contact between M1 and M2; by contrast, it reaches the level of the M3 in adult forms. This palate reconfiguration, closely related to the development and eruption of posterior molars, reflects substantial changes and marked allometric shifts affecting skull proportions in most mammals (Abdala et al., 2001; Cassini et al., 2012; Moyano and Gianini, 2017; Liuti and Dixon, 2020).

FIG. 14.

Specimens of Neobrachytherium intermedium in different life stages. A, FMNH-P 14341, a juvenile individual in dorsal (left) and ventral (right) views. B, FMNH-P 14361, a subadult individual in dorsal (left) and ventral (right) views. C, FMNH-P 14500, an adult individual in dorsal (top) and ventral (bottom) views. Note that portions of these specimens have been reconstructed in plaster. Measurements of teeth are detailed in table 3.

TABLE 4.

Summary of the main cranial characteristics observed in Neobrachytherium intermedium (PVL 7792) and their significance in a broader context.

Continued

Finally, we can make some comments regarding the dentition. As wear progresses in these brachydont and lophodont teeth, the occlusal pattern undergoes substantial changes. The lophs become wider, especially in the case of the ectoloph and protoloph. In the latter case, this is particularly important for occlusal morphology, as wear on the protocone, which shows a low lingual slope in this species, results in the lingual edge of the protocone and the mesiolingual edge of the protoloph being close to the lingual limit of the tooth outline. Therefore, it appears that adult molars have almost vertical walls. In this sense, it is important to note that this is due to the effects of wear.

Another important observation is related to the first upper premolar. The first deciduous premolar (dP1) is rarely replaced in eutherians (Ziegler, 1971; Luckett, 1993; Uhen, 2000; McKay et al., 2022). In fact, its replacement has only been documented in a few extinct (e.g., hyracoids and cetaceans; Uhen, 2000; Gheerbrant et al., 2007) and extant groups (e.g., Tapirus, Ziegler, 1971; Procavia, Asher et al., 2017). In general terms, a single generation in the first premolar characterizes many carnivorans, perissodactyls, talpids, and macroscelidids; otherwise, the first premolar locus is absent, at least in extant mammals (McKey et al., 2022). Another topic of debate arises regarding the nature of the preserved first premolar; for instance, in macroscelidids, the first premolar is a retained deciduous tooth, but in Canis it is a successional tooth with no deciduous precursor (McKey et al., 2022 and references therein).

Regarding SANUs, Ziegler (1971: 243) noted that the first premolar replacement was observed in “members of both families of litopterns and at least two genera of the notoungulate suborder Toxodonta.” Soria (2001) described the morphology of deciduous and permanent first premolars for some proterotheriid species. In PVL 7792, the computed tomography and subsequent reconstruction allowed us to observe the presence of a permanent tooth beneath the erupted dP1 in the maxilla. Along with Soria's (2001) studies, our observations suggest a common pattern among the members of the family, which should be investigated as new materials are discovered and studied. Tomographic images also allowed observing the developing P2–4 (below deciduous counterparts; fig. 13A–D). Of these, the P4 is barely evident. Interestingly, the specimen FMNH-P 14483, N. intermedium, has P3, M1, and M2, and an unerupted P4, where the P3 seems to be recently erupted and resembles the unerupted M2 of PVL 7792 (fig. 13E). The M1 is more worn than the M1 of PVL 7792. Both specimens may give us a glimpse of the eruption sequence, as the last permanent premolar is the least developed. Hence, the eruption sequence would likely be P1, P2, P3, and P4, and these would also precede the eruption of the M3. In this context, PVL 7792 is a bit younger than FMNH-P 14483.

Some variation can be also noted for the dP3; however, in this case, it represents variation within the same individual. In this tooth of PVL 7792, the mesostyle is bifid on the left side, while it is simple on the right side. Another feature, the metaconule, also differs, being more lophoid and better developed on the left side. The first variation is morphological, while the second can be attributed to differential wear on both sides of the cranium. Nevertheless, these observations, especially differences like the distinct structure of the mesostyle on both sides, should be carefully documented and considered when analyzing isolated dental pieces for taxonomic and phylogenetic research.

CONCLUSION

The study of the juvenile specimen of Neobrachytherium intermedium PVL 7792 has provided valuable insights into the cranial morphology of this species and implications for the broader context of South American native ungulates. The specimen exhibits remarkable preservation allowing for detailed observations of dorsal and orbitotemporal regions, mesocranium, and basicranial structures. Table 4 briefly summarizes more relevant features. Some of these traits are distributed beyond Litopterna and are shared with other eutherians and even some metatherians. On the other hand, a number of traits previously mentioned as potential synapomorphies for Litopterna differ in Indalecia. Since this taxon may be an early diverging member of Litopterna, assessing its phylogenetic position is important for resolving the status and distribution of such traits.

Furthermore, although our observations are limited by the available sample, we have noted that ontogenetic variation can have a significant impact on the assessment of specific cranial traits. Certain features, such as the frontal sulcus and the supraorbital foramen on the frontal bones, as well as changes in palate configuration during the development and eruptions of posterior molars, exhibit age-related variations in Neobrachytherium. Additionally, dental occlusal patterns undergo substantial changes as wear progresses, particularly affecting loph morphology. Variations in dP3 are noted also in the same individual, with morphological differences in the mesostyle and the metaconule between the left and right sides. These findings underscore the importance of considering these sources of variations when analyzing fragmentary cranial remains and/or isolated teeth for taxonomic and phylogenetic purposes.

Another noteworthy observation of this study is the presence of the permanent counterpart beneath the erupted dP1 within the maxilla, representing a rare case of replacement of the first postcanine tooth among eutherians.

The study of cranial features and ontogenetic variations in Neobrachytherium intermedium contributes valuable information to future understanding of this species and its evolutionary context within Litopterna. It emphasizes the need for more comprehensive research on this subject, as studies on cranial evolution of SANUs are still a crucial issue, even after more than a century of continuous and valuable contributions.