IVPP V20378 is a nearly complete skeleton of Sinovenator changii, preserved in a sleeping posture resembling another troodontid (Fig. 1C), Mei long (Xu & Norell 2004). Although the skull is isolated from the skeleton, most of the postcranial elements are articulated and without deformation (Fig. 1A&B, 2). IVPP V20378 shows typical basal troodontid features such as relatively long forelimbs, elongation of caudal vertebrae, and the third phalanges of the second pedal digits forming enlarged claws.
Cranium
The skull is three-dimensionally preserved and only missing some of the posterior left elements including the postorbital, parietal, and squamosal, while the rest is well-preserved. The posterior half of the skull is tilted to the right side due to preservation deformation.
In lateral view, the skull measures 84mm in length from the anteriormost extent of the premaxilla to the posterior end of the occipital condyle, with the orbit having an orbit height of approximately 23mm. The rostrum is slightly dorsoventrally compressed with the anterior part of the mandible being largely inserted in between the upper jaws. In dorsal view, the sagittal sutures between braincase elements and nasals are clearly visible (Fig. 2A). The right frontal and parietal are well-preserved while the posterior part of left frontal and anterior part of left parietal are missing. The posterior half of the skull is deformed along the right side as the sagittal suture is tilted to the right side but the overall morphology of the right half of the braincase seems to be less impacted. Since the left side of the braincase is largely missing or distorted, the description and comparative analysis are predominantly based on the right side.
Both Xu et al. (2002) and Yin et al. (2018) indicated that the foramen magnum is dorsoventrally taller than wide, but the foramen magnum of IVPP V20378 is wider than its dorsoventral height. This difference could be due to intraspecific variation or the result of preservational deformation. A shallow drop-shape subotic recess is preserved on the right side of the braincase but the corresponding area on the left side is not preserved (Fig. 2B). An even shallower basisphenoid recess than in PMOL-AD00102 is observed in both the ventral view photograph and the CT scan (Fig. 2B&3B), and the otosphenoidal crest on both sides are well preserved. In general, the braincase morphology of IVPP V20378 resembles that of PMOL-AD00102 more in having traits that are not present in the holotype, including having a subotic recess, otosphenoidal crest and basisphenoid recess. Although there are differences between the holotype IVPP V 12615, PMOL-AD00102, and IVPP V 20378, there is no reason to question that they all pertain to the same species, and the overall morphology supports an intermediate status of Sinovenator between other troodontids and other early paravian taxa.
Endocranium
To evaluate the brain structure of IVPP V20378, we used the cranial endocast as a proxy for the actual brain surface anatomy, based on evidence that the brains of maniraptorans largely filled the bony endocranial cavity (Witmer & Ridgely 2009, Balanoff et al. 2013), but with consideration of possible bone misalignment and the existence of other soft tissues within the cranium. The division of neuroanatomical regions follow Balanoff et al. (2013). Both the olfactory bulbs and tracts are proportionately elongated anteriorly and show slight ventral curvature from lateral view. Maniraptorans had a decreasing trend of olfactory capabilities during their evolution, with these structures being highly reduced in crown birds, although some bird lineages showed a secondary increase (Zelenitsky et al. 2011; Balanoff et al. 2013). In the case of IVPP V20378, the olfactory bulbs are comparably enlarged relative to crown birds. What differentiates Sinovenator from more modern avians is the presence of a gracile, elongated olfactory tract that connects the olfactory bulbs to the main body of the endocast via the cerebrum. The olfactory apparatus (the olfactory bulbs and the olfactory tract) are more reminiscent of the same found in Archaeopteryx that that found in modern birds or other troodontids (Balanoff et al. 2013). The large size of the olfactory bulbs and anteriorly long olfactory tracts in Sinovenator seems to be a plesiomorphic trait for paravians and generally for archosaurs. However, the olfactory apparatus in Sinovenator is considerably smaller than in large theropods such as T. rex or Allosaurus, but resembles Deinonychus (Witmer & Ridgely 2009) and bears more curvature from lateral view (Fig. 3C). However, the boundary between the olfactory bulb and tract is not clearly defined like other paravians such as Zanabazar (Balanoff et al. 2013) and Latenivenatrixs (Witmer & Ridgely 2009; Pauline-Carabajal et al. 2023 edited by Dozo et al.).
The cerebrum is dorsoventrally flat in lateral view but wide along the coronal axis. Posteriorly, the cerebrum is wide and would have formed a triangular with the apex tapering anteriorly to meet the olfactory apparatus. Based on other paravian endocrania including that of Zanabazar (Balanoff et al. 2013) and Troodon (Witmer & Ridgely 2009), the flat configuration of the cerebrum is probably a result from preservational deformation in Sinovenator, which likely caused the midbrain shifting to the right, but the apparent flatness of many maniraptoran cerebra is probably an artifact of the ventral portion of the cerebrum not being walled by bone and hence not being preserved in fossils. Anteriorly, a sulcus can be found medially in dorsal view near the junction of the olfactory tract and the cerebrum. This sulcus separates two clearly raised left and right sections before being obstructed by the missing section of the braincase. The medial and left lateral parts of the cerebrum are not preserved in IVPP V20378, thus we cannot know whether the two cerebral hemispheres are clearly defined by the inter-hemispherical sulcus or not.
The optic lobe has increased in size significantly during the evolution of maniraptorans and occupis a large proportion of modern bird brain volume, implying greater emphasis on the sense of vision. The optic lobes are not clearly defined in most non-maniraptoran theropods, for example Majungasaurus (Sampson & Witmer 2007) and Struthiomimus (Witmer & Ridgely 2009). Maniraptorans have larger and more clearly defined optic lobes relative to the rest of the brain, and the optic lobes gradually move ventrolaterally to form a globose brain, and multiple lineages of modern birds have their optic lobes positioned completely under the cerebrum in lateral view (Balanoff et al. 2013). Consistent with this trend, the optic lobe of Sinovenator is well-defined from the rest of the brain in both lateral and ventral view (Fig. 3C). In right lateral view, the preserved optic lobe of IVPP V20378 is positioned more laterally than in more basal archosaurs (Hu et al. 2021) and ventral to the posterior section of the cerebrum. Its position of the optic lobes in IVPP V20378 is intermediate between non-avialan theropods such as oviraptorosaurian and extant birds and seems to be at least comparable to that of Archaeopteryx (Domínguez Alonso et al. 2004; Balanoff et al. 2013).
The posterior part of the braincase has a misalignment between the skull roof and the supraoccipital, artifactually ntroducing space between the parietal and supraoccipital, and the squamosal is also detached from the parietal (Fig. 3B). Moreover, given that venous sinuses that dorsally overlie the cerebellum and the hindbrain are widespread but have inconsistent morphology in extant archosaurs (Witmer et al. 2008; Watanabe et al. 2019), the hindbrain of Sinovenator was reconstructed with other paravian models as references and not closely reflecting the endocast morphology. The cerebellum in Sinovenator is less laterally expanded than both the optic lobe and cerebral area. However, unlike the condition in the troodontid Zanabazar (Balanoff et al. 2013; Torres et al. 2021) and the dromaeosaurid Velociraptor (King et al. 2020) in which the cerebellum had acquired volumetric expansion, the cerebellum in Sinovenator is flatter dorsoventrally and without a well-defined boundary between other endocranial compartments (Fig. 3D). When viewed laterally, Sinovenator has its cerebellum located posteriodorsally to the optic lobe and is separated distally from the cerebrum. The cerebellum in Sinovenator has a position resembling Archaeopteryx and extant birds more-so than the previously mentioned maniraptorans but has obviously less volume. The acquisition of avian brain architecture may be better described in a mosaic manner than a gradually linear process. The volumetric expansion and anterior shift of the cerebellum are at least decoupled in at least the lineage of troodontids. Based on the better preserved right side of the fossil, the flocculus extends posterior-ventrally with a slight curvature, having a total length of approximately 6mm. The distal end of the flocculus is within the lateral extent of the optic lobe in ventral view. Although the flocculus endocast is robust, the actual size may not be as large as the endocast indicates (Walsh et al. 2013; Ferreira-Cardoso et al. 2017), its general morphology resembles the dromaeosaurid Velociraptor (King et al. 2021), suggesting a comparable level of visual tracking movements of the eyes, head, and neck. The brainstem is directed ventrally to the optic lobe, indicating a strong dorsoventral flexure contrasting with the more linear profile in non-maniraptoran theropods. It extends posterior-laterally with a slight ventral orientation. Posteriorly, the foramen magnum creates an oval shape with the long axis extending laterally. Cranial nerves IX-XI are present along the ventral margin of the cerebellar area (Fig. 3D) though they are mostly unremarkable. A ventral midline fissure is clearly present in the brain stem in ventral view, which is common for modern birds but seems to be overlooked in non-avian dinosaurs.
Endosseous labyrinth
Only the right endosseous labyrinth is preserved in IVPP V20378 with the left missing the vestibular portion of the labyrinth due to poor preservation of the braincase. Broadly, the inner ear morphology resembles that in other paravians such as Velociraptor (King et al. 2020; Choiniere et al. 2021). The labyrinth is roughly triangular like that in Velociraptor (King et al. 2020) though it bulges more anteriorly due to the larger anteroposterior shape of the anterior semicircular canal. All of the semicircular canals are roughly orthogonal to each other. The posterior semicircular canal bends anteriorly prior to its dorsal connection with the crus communis. The lateral semicircular canal bows dorsally near its midpoint between the ampulla of the lateral semicircular canal and the posterior semicircular canal.
All of the semicircular canals maintain an even lumen thickness though none of the canals share the same thickness. This is to say that the anterior semicircular canal is the thickest canal with the posterior semicircular canal being smaller and lateral semicircular canal being smaller still. All of the canals widen as they meet their respective ampullae at the vestibule. The ampulla of the posterior semicircular canal and the ampulla of the lateral semicircular canal are dorsoventrally compressed. The ampulla of the anterior semicircular canal is a thicker than the other ampullae, though this may be due to the lack of resolution between the ampulla of the lateral semicircular canal, the ampulla of the anterior semicircular canal, and the vestibule.
The cochlear duct projects ventrally from the vestibule and deflects medially. When compared to the vestibular portion of the endosseous labyrinth, the cochlear duct is approximately the same length with no evidence of twisting or further curvature beyond its medial deflection. The cochlear duct is also thick as is the case in Velociraptor (King et al. 2020). Both the left and right cochlear ducts are similar in size, shape, and angle. Laterally, the fenestra vestibuli and fenestra cochleae are not visible but do not seem to have created a small profile that took do not seem to have taken up a large area – unlike those found in Velociraptor (King et al. 2020)
Geometric Morphometric Analysis
Prior to collection of landmark data from Sinovenator, the endocast was virtually mirrored to enable collection of artificially ‘left’ sided landmarks, then subsequently retrodeformed using median and several pairs of bilaterally symmetric landmarks. While the retrodefomed virtual endocast still exhibited some deformation, we conducted shape analysis on clearly identifiable landmarks that characterize the overall configuration of the neuroanatomical regions. Based on 13 anatomically defined landmarks (see Materials & Methods), the morphospace separates the non-avialan and avian dinosaurs along the first principal component (PC) axis (Fig. 4B). Notably, Sinovenator occupies an intermediate area of the morphospace, most closely resembling the endocranial shape of Archaeopteryx among the taxa sampled, whereas most of the sampled theropods are grouped with Alligator according to PC1 and PC2. The other juvenile troodontid specimen (IGM 100/1126; Balanoff et al. 2013) is also similar in shape to those of Sinovenator and Archaeopteryx, which all converge to the endocranial shape variation of extant birds.