The Endocranial Cavity of Oviraptorosaur Dinosaurs and the Increasingly Complex, Deep History of the Avian Brain

Original Paper Published online: August 10, 2018 The Endocranial Cavity of Oviraptorosaur Dinosaurs and the Increasingly Complex, Deep History of the Avian Brain Amy M. Balanoff a, b Mark A. Norell b Aneila V.C. Hogan a Gabriel S. Bever a, b a Center for Functional Anatomy and Evolution, Johns Hopkins University School of Medicine, Baltimore, MD, USA; b Division of Paleontology, American Museum of Natural History, New York, NY, USA Keywords Aves Cerebellum Endocast Forebrain Theropoda Abstract Unraveling the origins of the character complexes diagnosing major crown clades is one of the greatest challenges in evolutionary biology. These origination events tend to optimize along extraordinarily long stem lineages where the comparative biology of extant lineages is relatively weak in its heuristic power. Here we add to a growing paleontological literature on the evolutionary origins of the modern avian brain by describing the endocranial casts of two oviraptorosaur dinosaurs, Citipati osmolskae and Khaan mckennai. These fossil data confirm the antiquity of several avian features, including the expanded cerebrum. They also extend our appreciation of both the inherent variability in the brainskull relationship along the avian stem and the dynamic nature of these crown characters in the earliest history of their expression. 2018 S. Karger AG, Basel Introduction All phenotypes, from transcripts to complex adult morphologies, are composite structures whose individuated components reflect a succession of unique evolutionary origins. These origins may be clustered on a single phylogenetic stem, or they may span a series of nested crown clades. Under the latter scenario, the comparative biology of extant lineages is likely to reveal the timing, tempo, and perhaps mode of their stepwise acquisition. For example, the anatomy of extant cephalochordates establishes a complex model of forebrain evolution in which the diencephalon takes its origin along the chordate stem lineage, whereas the telencephalon emerges much later as a product of stem vertebrate evolution [Lacalli, 1996; Holland, 2015]. The alternative model, where key transformations are clustered on a single stem lineage, poses a much greater challenge and reveals the fundamental weakness of the comparative method as it is applied exclusively to the extant biota. Such comparisons are heuristically powerful in establishing the phylogenetic distribution of a targeted feature and thus the identity of the stem lineage housing its origin. But because that origin will always lie outside the phylogenetic bracket defined by the most inclusive crown clade in which the feature is expressed, the biology of extant lineages is poorly equipped to reveal other details critical to understanding the origin and early evolution of the targeted feature. These details certainly include the relative and absolute timing of transformations along the stem. They also include the initial functional role of the feature (if one existed), and the variational dynamics of the developmental system responsible for its expression, both of which may E-Mail karger@karger.com www.karger.com/bbe 2018 S. Karger AG, Basel Amy M. Balanoff or Gabriel S. Bever Johns Hopkins University 1830 E Monument Street Baltimore, MD 21205 (USA) E-Mail abalano2 @ jhmi.edu or gbever1 @ jhmi.edu

differ significantly from conditions found in the nearest crown clade and yet are critical for understanding the context in which a novel feature emerged [Gould and Lewontin, 1979; Gould and Vrba, 1982; Lauder, 1990, 1995; Witmer 1995; Bever, 2009]. Because increasingly long phylogenetic stems are likely to mask an increasing number and proportion of these details, the structural complexity underlying the origin of major clades must rank among the most challenging of evolutionary problems. The modern avian body plan emerged across a stem lineage whose total length easily exceeds 100 million years [Prum et al., 2015]. Its ongoing elucidation highlights the two primary approaches for circumventing (or at least reducing) the empirical shortcomings of the comparative biology of extant lineages. The first involves the experimental manipulation of model organisms to effectively engineer meaningful outgroups to the crown condition. This strategy remains in its infancy for bird origins [Bhullar et al., 2012] but is likely to expand its influence as technological advances such as CRISPR (clustered regularly interspaced short palindromic repeats) [Jinek et al., 2012] establish a firmer foundation. The other, well-traveled path takes advantage of the rich fossil record that populates the avian stem lineage and has proven effective at establishing the complex and deep origins of such iconic avian features as their bipedal stance, feathers, flight apparatus, and toothlessness (among many others, see Brusatte et al. [2014] for a recent review). The stem origins of the highly encephalized avian brain, as revealed by the endocranial morphology of nonavian theropod dinosaurs, are attracting the attention of an increasing number of comparative neuroanatomists. One theropod lineage that proves of particular interest in this pursuit is a morphologically unusual group from the early to late Cretaceous of North America and Asia called the oviraptorosaurs. These animals ranged from the catsized Caudipteryx zoui [Qiang et al., 1998] to the estimated 1,400 kg Gigantoraptor erlianensis [Xu et al., 2007]. They famously exhibited a number of avian features considered homologous with the crown group, including feathers and derived nesting behaviors [Norell et al., 1995]. They also express a number of apomorphic features in the skull, whose convergence on the modern avian condition was convincingly supported by the discovery of the basal-most oviraptorosaur Incisivosaurus gauthieri [Xu et al. 2002; Balanoff et al., 2009]. Extensive pneumatization, relative enlargement of the parietal, a double-headed otic process of the quadrate, and a host of other equally detailed skeletal features characterize crown birds and deeply nested oviraptorosaurs. These features tend to be absent in Incisivosaurus, which conserves a more plesiomorphic theropod morphology. In this study, we build on what is a limited understanding of the oviraptorosaur endocranial space by providing the first detailed endocast description for Citipati osmolskae [Clark et al., 2001, 2002] and Khaan mckennai [Clark et al., 2001; Balanoff and Norell, 2012]. We compare the morphological and volumetric details of these endocasts with those derived from Incisivosaurus and Conchoraptor to give a broader view of oviraptorosaur brain evolution and how it relates to the origin of the crown condition. Materials and Methods The endocast descriptions for both Citipati osmolskae and Khaan mckennai are based on specimens collected during the 1993 Joint Expedition of the American Museum of Natural History (AMNH) and Mongolian Paleontological Institute to the late Cretaceous Djadokhta Formation of the Gobi Desert. The Citipati specimen (IGM 100/978; Fig. 1a) is the holotype, whereas that of Khaan (IGM 100/973; Fig. 1b) is a referred specimen collected in close association with the holotype. Detailed descriptions of the collection, preparation, and cranial morphology of these specimens are provided in studies by Clark et al. [2002] and Balanoff and Norell [2012], respectively. High-resolution X-ray CT data for both Citipati and Khaan were collected at the High-Resolution X-Ray Computed Tomography Facility at The University of Texas at Austin (UTCT), with comparative datasets generated at both the Ohio University and AMNH Microscopy and Imaging Facility. Citipati was scanned at an interslice spacing of 0.25 mm, with a field of reconstruction of 201 mm and an image size of 1,024 1,024 pixels. These parameters result in a voxel that is 0.25 mm along the z-axis and 0.19 mm in the x- and y-axes. Khaan was scanned at an interslice spacing of 0.1637 mm, with a field of reconstruction of 78 mm and image size of 1,024 1,024 pixels. These parameters result in a voxel that is 0.1673 mm along the z-axis and 0.076 mm along the x- and y-axes. Scanning parameters for the comparative dataset can be found in Balanoff et al. [2013]. Endocasts were constructed and measured using the original TIFF or DICOM images in the volume-rendering program VG- StudioMax and following the recommended protocol of Balanoff et al. [2016]. For ease of description, we refer to the digitally constructed features of the endocranial space by the names of the softtissue neural structures they are inferred to reflect (e.g., cerebrum rather than cast of cerebrum). It is important to bear in mind, however, that even for those regions where endocast shape conforms closely with the expected morphology of the brain, the tissue interface reflected in the endocast surface is never actually brain to bone. It is phylogenetically justified to infer that these tissues were minimally separated by the meninges and in some areas by more volumetrically expansive structures such as dural sinuses and cerebral vessels [Balanoff and Bever, 2017]. We attempted to delimit the primary functional divisions of the brain within the endocast using a series of osteological landmarks [described in detail by Balanoff et al., 2013]. The landmarks were 126 Balanoff/Norell/Hogan/Bever

chosen based on the brain-skull relationship among extant birds, and they appear to conserve a high degree of accuracy for at least the most crown-ward, nonavian theropods, which include Oviraptorosauria. One can expect their fidelity to the actual brain regions to erode as the BEC index (brain-endocast index) [Balanoff et al., 2016] is reduced down the stem lineage and towards the ancestral crown archosaur. The rate and complexity of that declination remains unclear but represents an area of active research [Morhardt et al., 2012]. Results C. osmolskae (IGM 100/978) The highly derived and quite simply bizarre nature of the Citipati skull [Clark et al., 2002] is, to at least some degree, echoed in the morphology of its endocast (Fig. 1a, 2). Considering that Citipati is recovered in such a deeply nested position among coelurosaurian theropods and has such a short, compact skull, its endocast is remarkably tubular. The marked cephalic and pontine flexures expected in a maniraptoran dinosaur that correspond to the folding of neural tissue at the forebrain-midbrain and metencephelon-myelencephalon boundaries, respectively [Hopson, 1979], are not strongly expressed. The resultant elongation of the endocast renders a number of structural landmarks, including the exit points of various cranial nerves, separated by a larger-than-expected anteroposterior distance (Fig. 2b). The forebrain is represented by impressions of the olfactory bulbs, cerebrum, and parts of the diencephalon (Fig. 2). The olfactory bulbs and tracts are more clearly defined here than perhaps in any other published oviraptorosaur [Kundrát, 2007; Balanoff et al., 2014]. The bulbs are slightly upturned at their distal end and are relatively large, representing approximately 2.6% of the total endocranial volume. The tracts are short, which leaves the bulbs pressed tightly against the anteroventral surface of the cerebral hemispheres (Fig. 2b, c); this arrangement characterizes all other examined oviraptorosaurs [Kundrát, 2007; Balanoff et al., 2009, 2014]. The cerebral hemispheres are laterally expanded, extending well beyond the margins of the optic tectum (Fig. 2a). This makes the cerebrum the widest structure of the endocast, and it seems likely based on the vascular impressions evident on the deep surface of the braincase that their lateral surfaces filled this dimension of the cranial cavity [Osmólska, 2004; Evans, 2005]. The cerebrum is easily the volumetrically largest region of the endocast, comprising approximately 43% of total endocast volume. Despite their overall large size, there is little dorsoventral expansion of the cerebral hemispheres. Their a b 1 cm 1 cm Fig. 1. Right lateral views of the skulls of Citipati osmolskae (IGM 100/978) (a) and Khaan mckennai (IGM 100/973) (b). general shape is much more disk-like than in other coelurosaurs [Norell et al., 2009]. In dorsal view, a subtle, transverse constriction in the caudal half of the cerebrum is identified as the impression of the frontoparietal suture (Fig. 2a). A deep interhemispheric sulcus characterizes the rostral half of the cerebrum but dissipates caudally, perhaps obscured by an overlying dorsal sagittal sinus (Fig. 2a). The area adjacent to the posterior margin of the cerebrum lacks any evidence of a dorsal evagination of the epithalmic roof of the diencephalon. This was described for a specimen of Conchoraptor (ZPAL MgD-I/95) and identified as an epiphysis [Kundrát, 2007]. The pituitary fossa is visible in the two-dimensional CT slice images, although the walls of the basisphenoid that delimit its spatial relationships are not well ossified or at least not well preserved. The former hypothesis is most likely, as this region of the skull tends to be poorly ossified in oviraptorosaurs [Balanoff and Norell, 2012]. Its exact size and shape are difficult to discern, but the available data suggest a fossa morphology similar to that of Conchoraptor [Balanoff et al., 2014]. The optic nerves (CN II) exit through a single, midline opening defined by the paired laterosphenoids (Fig. 2c) [Lamanna et al., 2014]. Oviraptorosaur Endocasts and the Origins of the Bird Brain 127

a b Fig. 2. Endocranial cast of the oviraptorid Citipati osmolskae (IGM 100/978) in dorsal (a), lateral (b), and ventral views (c). c The optic lobes are the only identifiable structures of the midbrain. They are spherical and reside in a ventrolateral position on the endocast (posteroventral to the cerebral hemispheres). The lobes are large, representing approximately 16% of total endocranial volume, and extend laterally beyond the cerebellum. They do not, however, contact each other along the dorsal midline. The hindbrain is represented by the cerebellum, medulla, and associated cranial nerves and vessels (Fig. 2). The prominent cerebellum is expanded both anteriorly and laterally. Its dorsal surface remains low and rounded with several small indentations that possibly reflect cerebellar folia (Fig. 2b). A large, conically shaped flocculus extends caudolaterally from each side of the cerebellum to take up its expected position between the semicircular canals of the inner ear. The pontine end of the brain stem is expanded. As noted previously, the hindbrain lacks a distinctive pontine flexure, which leaves the trigeminal nerve separated from the hypoglossal and vagus nerves by a derived (large) distance. The nerve itself (CN V) is large and undivided on the endocast; it originates just anteroventral to the flocculus. The midpoint of the foramen is slightly constricted. The cast of the vagus foramen, which is inferred as containing cranial nerves IX XI, is located completely posterior to the flocculus (Fig. 2). 128 Balanoff/Norell/Hogan/Bever

a b Fig. 3. Endocranial cast of the oviraptorid Khaan mckennai (IGM 100/973) in dorsal (a), lateral (b), and ventral views (c). c Khaan mckennai (IGM 100/973) The skull of IGM 100/973 exhibits some degree of dorsoventral compression, but this appears to have had only minor effects on endocast morphology (Fig. 1b, 3). In general, the Khaan endocast is very similar to that of Citipati with an elongate, tubular morphology reflecting a shared relaxation of the cerebral and pontine flexures (Fig. 3). The short olfactory tracts are not directly visible but are evidenced by the contact between the olfactory bulbs and anteroventral surface of the cerebral hemispheres. The bulbs themselves are highly reduced (Fig. 3), much more so than in Citipati and certainly more so than in other nonavian coelurosaurs (Fig. 4). Their volume represents only approximately 0.23% of the total endocast. The cerebrum is dorsoventrally shallow and, in dorsal view, is more ovoid and less disk-like than in Citipati. Its breadth at the anteroposterior midpoint still represents the widest point of the endocast (Fig. 3a). The posterior dissipation of the midline sulcus may again reflect the rostral distribution of an overlying sinus. The pituitary fossa is even less visible in the two-dimensional CT scans of Khaan than in those of Citipati, Oviraptorosaur Endocasts and the Origins of the Bird Brain 129

suggesting once again that the basisphenoid either fails to fully ossify or occurs as an extremely late-stage postnatal event. The optic nerves, along with the midbrain optic tecta, are highly similar to those described for Citipati (Fig. 3). The relaxed flexure of the hindbrain in Khaan produces an endocast that approaches a straight line in lateral view (Fig. 3b). The low cerebellum falls short of the dorsal margin of the cerebrum when the medulla is horizontally positioned (Fig. 3b). The cerebellum is wide but less so than the medulla, although both structures may be constrained laterally by a combination of basioccipital pneumaticity and dural sinuses [Kundrát and Janáček, 2007]. The pontine region of the brain stem directly ventral to the cerebellum is expanded in Khaan relative to Citipati and closely approaches the condition seen in crown group birds. The only discernible cranial nerves in this region are CN V and the combined paths of CN IX XI (Fig. 3). The trigeminal cast is large and lies just posterior to the optic lobe as it does in other theropod dinosaurs. The cast of the vagus foramen lies well posterior to the base of the large, conical flocculus. Discussion The additions of Citipati and Khaan to the previously described endocasts of Conchoraptor [two specimens; Kundrát, 2007; Balanoff et al., 2014] and Incisivosaurus [Balanoff et al., 2009] establish a comparative framework that brackets nearly the entire oviraptorosaur radiation. So even with only four sampled taxa, these new data establish the first phylogenetically meaningful hypotheses of endocranial evolution for this interesting lineage and set a firm stage for future studies. Endocast Shape The flattening of the cephalic and pontine flexures in Khaan and Citipati are derived features within Oviraptorosauria that differ markedly from the in-group condition of Conchoraptor and Incisivosaurus, and from the endocasts of closely related deinonychosaurs (Fig. 4) [Norell et al., 2009]. The resultant morphology at least superficially resembles the plesiomorphically elongate endocasts of more basal theropod divergences, such as tyrannosaurids [Witmer and Ridgely, 2009; Bever et al., 2011a, 2013] or even noncoelurosaurian taxa [Sampson and Witmer, 2007]. Interestingly, if the shared morphology of Khaan and Citipati is to enjoy the status of secondary homology, then the S-shaped, highly flexed endocast of Conchoraptor must lose its secondary homology with Incisivosaurus and other maniraptorans. Both scenarios require the same number of steps under parsimony and the preferred evolutionary model reflects a choice between an accelerated or delayed transformation that can only be adjudicated with increased taxonomic sampling. It is worth noting that although these differences in endocast shape fail to correspond with obvious disparities in the external morphology of the involved crania (e.g., the presence or absence of a cranial crest), they do appear to correlate with the degree to which the surrounding dermal roofing bones are pneumatized (Fig. 4). Pneumatization and elongation roughly parallel each other across a spectrum running from Citipati with the most extreme pneumatization and the most elongate endocast to Khaan to Conchoraptor to Incisivosaurus, whose highly flexed endocast resides within a braincase nearly devoid of internal pneumatization [Clark et al, 2002; Kundrát and Janačék, 2007; Balanoff et al., 2009; Balanoff and Norell, 2012]. We thus may be witnessing the secondary effects of brain evolution in concert with transformations in cranial pneumatization. If so, oviraptorosaurs should be carefully considered when assessing the complex relationship between the morphologies of the brain and skull, and the structural, functional, and developmental constraints that influence that relationship. Reduction in the Olfactory System The components of the oviraptorosaur olfactory system exhibit marked reduction relative to those of more basal coelurosaurs and nonavian maniraptorans (Fig. 4). The extent of their morphological disparity is perhaps most clear when compared with the olfactory system of deinonychosaurs, whose relatively large olfactory bulbs reside at the distal end of extended olfactory tracts (Fig. 4e; see Zanabazar junior) [Norell et al., 2009: Fig. 28]. That Incisivosaurus also expresses these reductions establishes the attendant transformations as occurring at or near the base of the radiation, where they would diagnose the entirety of the group. Incisivosaurus and Khaan both have a volumetric ratio between the olfactory bulbs and total endocast volume under 1% (0.54 and 0.32%, respectively), which suggests that an apomorphically low value was likely conserved over much of the radiation (although not without variation). The olfactory bulbs of Conchoraptor are so small that they cannot be confidently segmented from the main body of the endocast [Kundrát, 2007; Balanoff et al., 2014], whereas those of Citipati appear to be secondarily enlarged. Their approximately 2.6% volumetric ratio is still smaller than expected for a nonavian 130 Balanoff/Norell/Hogan/Bever

a b c d e Fig. 4. Hypothesized relationships of Oviraptorosauria based on Pu et al. [2017]. Boxes denote the taxa whose endocasts form the basis of this study. Lateral views of endocasts of Conchoraptor gracilis (IGM 100/3006) (a); Khaan mckennai (IGM 100/973) (b); Citipati osmolskae (IGM 100/978) (c); Incisivosaurus gauthieri (IVPP V 13326) (d); and Zanabazar junior (IGM 100/1) (e) with skulls rendered semitransparent. maniraptoran, but it is the only value among oviraptorosaurs that falls outside the distribution of crown group birds [Balanoff et al., 2013]. This raises what is perhaps the most compelling point regarding the reduced olfactory system of oviraptorosaurs, which is that it parallels the olfactory reduction within modern birds. The extent of the observed overlap between the derived values of these two groups is striking, although clearly homoplastic, considering the well-established topology of the theropod tree and the large olfactory structures in the more crown-ward deinonychosaurs [Norell et al., 2009; Witmer and Ridgely, 2009; Bever et al., 2011a; Balanoff et al., 2014]. That this reduction in olfactory bulb size was due to the emergence of an omnivo- Oviraptorosaur Endocasts and the Origins of the Bird Brain 131

rous diet among oviraptorosaurs remains a viable hypothesis [Zelenitsky et al., 2009]. The olfactory features of the early avialan Archaeopteryx lithographica are transitional in their development smaller than those of deinonychosaurs but larger than their homologues in both crown birds and oviraptorosaurs [Balanoff et al., 2013]. The condition within Archaeopteryx, however, was interpreted as a reflection of a larger pattern of increased olfactory acuity among early Neornithes, which was subsequently reversed within living birds [Zelenitsky et al., 2011]. Cerebra The cerebra of Khaan, Citipati, and Conchoraptor are circular in dorsal view, with a diameter that establishes the widest overall point of the endocast. They lack the distinct dorsoventral expansion of other maniraptorans, which serves to enhance the overall disk-like morphology (Fig. 2 4). The cerebrum of Incisivosaurus, in contrast, has a distinctive dorsoventral expansion that aligns it morphologically with paravians. The cerebra of paravians are more pyriform in shape than those of Incisivosaurus. This morphology may well reflect some mediolateral expansion of the paravian cerebra but is also likely exaggerated by some mediolateral compression of the one and only Incisivosaurus skull [Balanoff et al., 2009: Fig. 16]. Crown group birds have cerebra that are deeper than those of oviraptorosaurs or early paravians (including Archaeopteryx) [Alonso et al., 2004], which firmly establishes the circular, disk-like morphology of deeply nested oviraptorosaurs as a derived feature within the group (Fig. 4). The structural influences responsible for this autapomorphic morphology among theropods are unclear; perhaps it is related to their unusually large, circular orbits, as the orbit strongly influences brain shape in living birds [Kawabe et al., 2013]. Optic Lobes The optic lobes of all oviraptorosaurs have the ventrolateral position expected of a maniraptoran theropod. This condition takes its place within a trend of displacement among coelurosaurs, whose early history is informed by tyrannosaurs [Witmer and Ridgely, 2009; Bever et al., 2011a; 2013]. Those of Citipati, Khaan, and Conchoraptor are circular in lateral view, whereas those of Incisivosaurus have a more rectangular outline, which may reflect some degree of postmortem mediolateral compression [Balanoff et al., 2009: Fig. 15]. The optic lobes of oviraptorosaurs, deinonychosaurs, and early avialans all fail to extend laterally beyond the cerebral hemispheres as they do in crown group birds [Alonso et al., 2004; Norell et al., 2009; Balanoff et al., 2013; Walsh et al., 2016]. Cerebellum The cerebellum in the observed oviraptorosaurs is volumetrically large (relative to the total endocast) when compared to other coelurosaurs, including crown group birds. That the oviraptorid cerebellum [Balanoff et al., 2013: Fig. 3b] is distinctly enlarged over that of Incisivosaurus indicates that cerebellar expansion continued inside the radiation. This increased volume may well be concentrated in the distinctly large flocculus. The more conical shape of the oviraptorid flocculus relative to the compressed structure of Incisivosaurus indicates that floccular diameter increased with length (though not isometrically; Fig. 2 4). The expanded floccular size of pterosaurs is possibly linked to the evolution of powered flight and the associated increase in afferent proprioceptive and cutaneous information from the flight apparatus [Witmer et al., 2003]. Within birds, the only significant correlations thus far established for the volume of the floccular fossa are with feeding strategy and nocturnality [Walsh et al., 2013; Ferreira-Cardosa et al., 2017]. Based on their postcranial skeletal morphology, it is highly unlikely that oviraptorosaurs were capable of even rudimentary flight [Osmólska et al., 2004; Balanoff and Norell, 2012]. The possibility remains, however, that their enlarged flocculus reflects an elaboration of somatosensory input from cutaneous structures such as feathers. The low profile of the cerebellum in Khaan and Citipati is found neither in Incisivosaurus nor Conchoraptor. The latter taxa resemble other maniraptorans whose cerebellar outline is characterized by a nearly 90 angle. This morphology is likely influenced by the pronounced pontine flexure (Fig. 4). The cerebellum endocast of Conchoraptor described by Kundrát [2007: Fig. 2] bears a triangular apex on its dorsal surface that may reflect an expanded occipital sinus as inferred for more basal coelurosaurs and some deinonychosaurs [Norell et al., 2009; Witmer and Ridgely, 2009; Bever et al., 2011a, 2013; Lautenschlager et al., 2012; Brusatte et al., 2016]. Whatever the structure represents, it must be subject to some form of variation as it is lacking in the other described specimen of that taxon [Balanoff et al., 2014] and other oviraptorosaurs (Fig. 4). Cerebellar ridges, indicating the possible presence of avian-like folia, are present in Incisivosaurus [Balanoff et al., 2014], one specimen of Conchoraptor [Kundrát, 2007], and now Citipati (Fig. 2b), although, to be sure, they are 132 Balanoff/Norell/Hogan/Bever

relatively faint in the latter form. The apparent absence of these structures among the earlier divergences within Theropoda and within the known deinonychosaurs may reflect the retention of an expanded occipital sinus that precludes their expression on the endocast [Franzosa and Rowe, 2005; Sampson and Witmer, 2007; Witmer and Ridgely, 2009; Bever et al., 2011a; 2013]. If this is correct, then the apomorphy shared by oviraptorosaurs and birds likely has more to do with the thinning or loss of that sinus than with a derived foliation, a conclusion supported by the observation that nearly all vertebrate groups possess some degree of cerebellar folding [Butler and Hodos, 2005]. Modern crocodilians lack these extensive foliations, making the plesiomorphic condition for crown group Archosauria and most successive clades along the avian stem lineage ambiguous. For now, we consider the expressed folding of the cerebellar endocast an ambiguous maniraptoran synapomorphy. Brain Stem The expanded pontine region of the brain stem in Khaan and Conchoraptor resembles the condition expressed in crown group birds (and crown mammals) [Butler and Hodos, 2005]. The enlarged pons likely reflects an increased number of axonal tracts originating in the pallial cerebrum and passing through the pons en route to the cerebellum. The lack of a distinct expansion in Incisivosaurus and Citipati suggests that the added tracts are derived deep within Oviraptoridae and thus nonhomologous with the bird condition. Posterior to the flocculi, the brain stems of Khaan and Citipati exhibit a derived expansion of the medulla endocasts (Fig. 2 4) that is absent in both Incisivosaurus and Conchoraptor. Whether convergent in Khaan and Citipati or lost in Conchoraptor, this apomorphic condition may reflect a transformation of the surrounding venous sinuses [Sinus foraminis magni; Baumel and Witmer, 1993] rather than one of the medulla itself. The pinched nature of the trigeminal foramen in Incisivosaurus, Citipati, and Conchoraptor (indiscernible in Khaan) suggests that CN V formed its constituent divisions prior to exiting the braincase. A similar condition is described for the caenagnathid Chirostenotes pergracilis, which also bears a small fossa on the deep surface of its laterosphenoid that likely housed the trigeminal ganglion [Sues, 1997]. An intracranial division of the trigeminal was considered a convergent feature shared by tyrannosaurs and crown group birds [Witmer and Ridgely 2009; Bever et al. 2011a, 2013], an interpretation reinforced by the apparently extracranial division of this nerve in the basal tyrannosaur Timurlengia [Brussatte et al., 2016]. Our data suggest an even greater degree of homoplasy for this feature. Conclusions The fossil record has proven itself an effective resource for breaking up the long phylogenetic branch connecting the ancestral crown group archosaur to the ancestral crown group bird and thereby revealing the origins of modern avian anatomy, physiology, and behavior. In terms of the avian brain, fossils indicate the trends responsible for the highly encephalized, highly flexed brain of modern birds began deep in their stem lineage, which is not terribly surprising considering that fossil discoveries will almost always push the origins of modern features progressively deeper in time and tree space. The new oviraptorosaur data presented here confirm the antiquity of a variety of derived avian neuroanatomical features, most notably an expanded cerebrum set within a highly pneumatized braincase. The morphology of these fossils also indicates that the deeply folded cerebellum and intracranial origin of the trigeminal divisions of modern birds may also have originated deeper in the stem than previously appreciated. Perhaps more importantly, the expanded sampling within oviraptorosaurs demonstrates that the early history of several features of the avian brain was marked by at least some level of homoplasy. This pattern includes apparent convergences with the crown bird condition (cerebellar/flocullar expansion, enlarged pons, and a reduced olfactory apparatus) but also at least one reversion to the plesiomorphic theropod state (elongate overall shape with only a subtle degree of cerebral and pontine flexure). This reversal piqued our interest because it may well be driven by a derived expansion of pneumaticity within the overlying dermal roof rather than by selective forces aimed directly at the brain. If correctly interpreted, this would call into question a generalized assumption that the functional significance of the brain and its regions gives it structural primacy over the surrounding head tissues. The complex signaling relationship between the brain and cranial mesenchyme is only beginning to be meaningfully outlined [Marcucio et al., 2011; Fabbri et al., 2017], and we are hopeful that the fossil record will serve as an increasingly useful resource in assessing the phenotypic potential of these integrated systems. An appropriate question is whether homoplasies, such as those discovered among Oviraptorosauria, should be Oviraptorosaur Endocasts and the Origins of the Bird Brain 133

expected. Classic models predict that the probability of convergence should scale positively with phylogenetic distance. In other words, given enough time, distinct lineages are likely to converge on the same structural solution to a shared problem. This model includes the concepts of long-branch attraction [see Bergsten, 2005] and morphological exhaustion [Wagner, 2000]. A second possibility, one that has received much less attention among morphologists, is that homoplasy levels will be relatively high during the early history of an apomorphic feature. This is sometimes referred to as phylogenetic inertia [Hoyal Cuthill, 2015], although one should recognize that this term has a long history during which it conveyed a variety of concepts [Blomberg and Garland, 2002]. It has also been discussed as a zone of variability, a model in which the concentration of homoplasy in the early history of a character is explicitly predicted as the phylogenetic conservation of the variability [sensu Wagner and Altenberg, 1996] expected in the development system involved in a derived transformation [Bever, 2009; Bever et al., 2011b]. The full implications of such a pattern have yet to be fully worked out but may well impact the fundamental roles of paleontological data within larger biological pursuits for example, establishing a meaningful distribution of posterior probabilities for the age of a crown clade based on the phylogenetic assessment of the known fossil record. The larger point that we want to emphasize here is simply that the fossil record can do more than establish the timing and relative order in which a set of diagnostic crown features take their origin along a stem lineage. It can also reveal refined patterns of phenotypic expression that inform on the dynamic nature of evolving developmental and functional systems, including those responsible for patterning the brain. Acknowledgments We thank Ashley Morhardt for convening the Karger Workshop and for graciously inviting us to participate. We also thank her for her editorial patience as we completed the manuscript. We would like to thank Steve Brusatte and an anonymous reviewer for providing comments that helped improve the manuscript. We gratefully acknowledge UTCT and the Ohio University for facilitating the scanning of specimens and Mick Ellison for taking photographs of the skulls. This project was supported by grants from the Division of Environmental Biology, National Science Foundation, to A.M.B., M.A.N., and G.S.B. (No. 1801224) and to A.M.B. and M.A.N. (Doctoral Dissertation Improvement Grant: DDIG DEB 0909970). The Division of Paleontology at the AMNH and the Mongolian Academy of Sciences are also thanked for their support of this project. Disclosure Statement The authors declare that they have no competing financial interests. References Alonso PD, Milner AC, Ketcham RA, Cookson MJ, Rowe TB (2004): The avian nature of the brain and inner ear of Archaeopteryx. Nature 430: 666 669. Balanoff AM, Bever GS (2017): The role of endocasts in the study of brain evolution; in Kaas J (ed): Evolution of Nervous Systems, ed 2. Oxford, Elsevier, pp 223 241. Balanoff AM, Bever GS, Colbert MW, Clarke JA, Field DJ, Gignac PM, Ksepka DT, Ridgely RC, Smith NA, Torres CR, Walsh S, Witmer LM (2016): Best practices for digitally constructing endocranial casts: examples from birds and their dinosaurian relatives. J Anat 229: 173 190. Balanoff AM, Bever GS, Norell MA (2014): Reconsidering the avian nature of the oviraptorosaur brain (Dinosauria: Theropoda). PLoS One 9:e113559. Balanoff AM, Bever GS, Rowe TB, Norell MA (2013): Evolutionary origins of the avian brain. Nature 501: 93 96. Balanoff AM, Norell MA (2012): Osteology of Khaan mckennai (Oviraptorosauria: Theropoda). Bull Am Mus Nat Hist 372: 1 77. Balanoff AM, Xu X, Kobayashi Y, Matsufune Y, Norell MA (2009): Cranial osteology of the theropod dinosaur Incisivosaurus gauthieri (Theropoda: Oviraptorosauria). Am Mus Nov 3651: 1 35. Baumel JJ, Witmer L (1993): Osteologia; in Baumel JJ, King AS, Breazile JE, Evans HE, Vanden Berge JC (eds): Handbook of Avian Anatomy: Nomina Anatomica Avium, ed 2. Cambridge, Nuttall Ornithological Club, No 23, pp 45 132. Bergsten J (2005): A review of long-branch attraction. Cladistics 21: 163 193. Bever GS (2009): The postnatal skull of the extant North American turtle Pseudemys texana (Cryptodira: Emydidae), with comments on the study of discrete intraspecific variation. J Morphol 270: 97 128. Bever GS, Brusatte SL, Balanoff AM, Norell MA (2011a): Variation, variability, and the origin of the avian endocranium: insights from the anatomy of Alioramus altai (Theropoda: Tyrannosauroidea). PLoS One 6:e23393. Bever GS, Brusatte SL, Carr TD, Xu X, Balanoff AM, Norell MA (2013): The braincase anatomy of the Late Cretaceous dinosaur Alioramus (Theropoda: Tyrannosauroidea). Bull Am Mus Nat Hist 376: 1 72. Bever GS, Gauthier JA, Wagner GP (2011b): Finding the frame shift: digit loss, developmental variability, and the origin of the avian hand. Evol Dev 13: 269 279. Bhullar B-AS, Marugán-Lobón J, Racimo F, Bever GS, Rowe TB, Norell MA, Abzhanov A (2012): Birds have paedomorphic dinosaur skulls. Nature 487: 223 226. Blomberg SP, Garland T (2002): Tempo and mode in evolution: phylogenetic inertia, adaptation and comparative methods. J Evol Biol 15: 899 910. 134 Balanoff/Norell/Hogan/Bever

Brusatte SL, Averianov A, Sues H-D, Muir A, Butler IB (2016): New tyrannosaur from the mid- Cretaceous of Uzbekistan clarifies evolution of giant body sizes and advanced senses in tyrant dinosaurs. Proc Natl Acad Sci USA 113: 3447 3452. Brusatte SL, Lloyd GT, Wang SC, Norell MA (2014): Gradual assembly of avian body plan culminated in rapid rates of evolution across the dinosaur-bird transition. Curr Biol 24: 2386 2392. Butler AB, Hodos W (2005): Comparative Vertebrate Neuroanatomy: Evolution and Adaptation, ed 2. Hoboken, Wiley. Clark JM, Norell MA, Barsbold R (2001): Two new oviraptorids (Theropoda: Oviraptorosauria), Upper Cretaceous Djadokhta Formation, Ukhaa Tolgod, Mongolia. J Vert Paleontol 21: 209 213. Clark JM, Norell MA, Rowe T (2002): Cranial anatomy of Citipati osmolskae (Theropoda, Oviraptorosauria), and a reinterpretation of the holotype of Oviraptor philoceratops. Am Mus Nov 3364: 1 24. Evans DC (2005): New evidence on brain-endocranial cavity relationships in ornithischian dinosaurs. Acta Palaeontol Pol 50: 617 622. Fabbri M, Koch NXSM, Pritchard AC, Hanson M, Hoffman E, Bever GS, Balanoff AM, Morris ZS, Field DJ, Camacho J, Rowe TB, Norell MA, Smith RM, Abzhanov A, Bhullar B-AS (2017): The skull roof tracks the brain during the evolution and development of reptiles including birds. Nat Ecol Evol 1: 1543 1550. Ferreira-Cardoso S, Araújo R, Martins NE, Martins GG, Walsh S, Martins RMS, Kardjilov N, Manke I, Hilger A, Castanhinha R (2017): Floccular fossa size is not a reliable proxy of ecology and behaviour in vertebrates. Sci Rep 7: 1 11. Franzosa J, Rowe T (2005): Cranial endocast of the Cretaceous theropod dinosaur Acrocanthosaurus atokensis. J Vert Paleontol 25: 859 864. Gould SJ, Lewontin RC (1979): The spandrels of San Marco and the panglossian paradigm: a critique of the adaptationist programme. Proc R Soc Lond B Biol Sci 205: 581 598. Gould SJ, Vrba ES (1982): Exaptation a missing term in the science of form. Paleobiology 8: 4 15. Holland LZ (2015): The origin and evolution of chordate nervous systems. Philos Trans R Soc Lond B Biol Sci 370: 20150048. Hopson JA (1979): Paleoneurology; in Glans C, Northcutt RG, Ulinski P (eds): Biology of the Reptilia. New York, Academic Press, pp 39 146. Hoyal Cuthill JF (2015): The morphological state space revisited: what do phylogenetic patterns in homoplasy tell us about the number of possible character states? Interface Focus 5: 20150049. Jinek M, Chylinksi K, Fonfara I, Hauer M, Doudna JA, Charpentier E (2012): A programmable dual-rna-guided DNA endonuclease in adaptive bacterial immunity. Science 337: 816 821. Kawabe S, Shimokawa T, Miki H, Matsuda S, Endo H (2013): Variation in avian brain shape: relationship with size and orbital shape. J Anat 223: 495 508. Kundrát M (2007): Avian-like attributes of a virtual brain model of the oviraptorid theropod Conchoraptor gracilis. Naturwissenschaften 94: 499 504. Kundrát M, Janáček J (2007): Cranial pneumatization and auditory perceptions of the oviraptorid dinosaur Conchoraptor gracilis (Theropoda, Maniraptora) from the Late Cretaceous of Mongolia. Naturwissenschaften 94: 769 778. Lacalli TC (1996): Frontal eye circuitry, rostral sensory pathways and brain organization in amphioxus larvae: evidence from 3D reconstructions. Philos Trans R Soc Lond B 351: 243 263. Lamanna MC, Sues H-D, Schachner ER, Lyson TR (2014): A new large-bodied oviraptorosaurian theropod dinosaur from the latest Cretaceous of western North America. PLoS One 9:e92022. Lauder GV (1990): Functional morphology and systematics: studying functional patterns in an historical context. Annu Rev Ecol Syst 2: 317 340. Lauder GV (1995): On the inference of function from structure; in Thomason JJ (ed): Functional Morphology in Vertebrate Paleontology. Cambridge, Cambridge University Press, pp 1 18. Lautenschlager S, Rayfield EJ, Altangerel P, Zanno LE, Witmer LM (2012): The endocranial anatomy of Therizinosauria and its implications for sensory and cognitive function. PLoS One 7:e52289. Marcucio RS, Young NM, Hu D, Hallgrimsson B (2011): Mechanisms that underlie co-variation of the brain and face. Genesis 49: 177 189. Morhardt AC, Ridgley RC, Witmer LM (2012): From endocast to brain: assessing brain size and structure in extinct archosaurs using gross anatomical brain region approximation (GABRA). J Vert Paleontol 32(suppl): 145. Norell MA, Clark JM, Chiappe LM, Dashzeveg D (1995): A nesting dinosaur. Nature 378: 774 776. Norell MA, Makovicky PJ, Bever GS, Balanoff AM (2009): A review of the Mongolian Cretaceous dinosaur Saurornithoides (Troodontidae: Theropoda). Am Mus Nov 3654: 1 63. Osmólska H (2004): Brief report: evidence on relation of brain to endocranial cavity in oviraptorid dinosaurs. Acta Palaeontol Pol 49: 321 324. Osmólska H, Currie PJ, Barsbold R (2004): Oviraptorosauria; in Weishampel DB, Dodson P, Osmólska H (eds): The Dinosauria, ed 2. Berkeley, University of California Press, pp 165 183. Pu H, Currie PJ, Carpenter K, Xu L, Koppelhus EB, Jia S, Xiao L, Chuang H, Li T, Kundrát M, Shen C (2017): Perinate and eggs of a giant caenagnathid dinosaur from the Late Cretaceous of central China. Nat Commun 8: 14952. Prum RO, Berv JS, Dornburg A, Field DJ, Townsend JP, Lemmon EM, Lemmon AR (2015): A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526: 569. Qiang J, Currie PJ, Norell MA, Shu-An J (1998): Two feathered dinosaurs from northeastern China. Nature 393: 753 761. Sampson SD, Witmer LM (2007): Craniofacial anatomy of Majungasaurus crenatissimus (Theropoda: Abelisauridae) from the Late Cretaceous of Madagascar. J Vert Paleontol 27: 32 104. Sues H-D (1997): On Chirostenotes, a Late Cretaceous oviraptorosaur (Dinosauria: Theropoda) from western North America. J Vert Paleontol 17: 698 716. Wagner GP, Altenberg L (1996): Perspective: complex adaptations and the evolution of evolvability. Evolution 50: 967 976. Wagner PJ (2000): Exhaustion of morphologic character states among fossil taxa. Evolution 54: 365 386. Walsh SA, Iwaniuk AN, Knoll MA, Bourdon E, Barrett PM, Milner AC, Nuds RL, Abel RL, Sterpaio PD (2013): Avian cerebellar floccular fossa size is not a proxy for flying ability in birds. PLoS ONE 8:e67176. Walsh SA, Milner AC, Bourdon E (2016): A reappraisal of Cerebavis cenomanica (Aves, Ornithurae), from Melovatka, Russia. J Anat 229: 215 227. Witmer LM (1995): The extant phylogenetic bracket and the importance of reconstructing soft tissues in fossils; in Thomason JJ (ed): Functional Morphology in Vertebrate Paleontology. Cambridge, Cambridge University Press, pp 19 33. Witmer LM, Chatterjee S, Franzosa J, Rowe T (2003): Neuroanatomy of flying reptiles and implications for flight, posture and behaviour. Nature 425: 950 953. Witmer LM, Ridgely RC (2009): New insights into the brain, braincase, and ear region of tyrannosaurs (Dinosauria, Theropoda), with implications for sensory organization and behavior. Anat Rec 292: 1266 1296. Xu X, Cheng YN, Wang XL, Chang CH (2002): An unusual oviraptorosaurian dinosaur from China. Nature 419: 291. Xu X, Tan Q-W, Wang J, Zhao X, Tan L (2007): A gigantic bird-like dinosaur from the Late Cretaceous of China. Nature 447: 844 847. Zelenitsky DK, Therrien F, Kobayashi Y (2009): Olfactory acuity in theropods: palaeobiological and evolutionary implications. Proc Biol Sci 276: 667 673. Zelenitsky DK, Therrien F, Ridgely RC, McGee AR, Witmer LM (2011): Evolution of olfaction in non-avian theropod dinosaurs and birds. Proc Biol Sci 278: 3625 3634. Oviraptorosaur Endocasts and the Origins of the Bird Brain 135