Paleoneurology of stem palaeognaths clarifies the plesiomorphic condition of the crown bird central nervous system

Lithornithidae, an assemblage of volant Palaeogene fossil birds, provide our clearest insights into the early evolutionary history of Palaeognathae, the clade that today includes the flightless ratites and volant tinamous. The neotype specimen of Lithornis vulturinus, from the early Eocene (approximately 53 million years ago) of Europe, includes a partial neurocranium that has never been thoroughly investigated. Here, we describe these cranial remains including the nearly complete digital endocasts of the brain and bony labyrinth. The telencephalon of Lithornis is expanded and its optic lobes are ventrally shifted, as is typical for crown birds. The foramen magnum is positioned caudally, rather than flexed ventrally as in some crown birds, with the optic lobes, cerebellum, and foramen magnum shifted further ventrally. The overall brain shape is similar to that of tinamous, the only extant clade of flying palaeognaths, suggesting that several aspects of tinamou neuroanatomy may have been evolutionarily conserved since at least the early Cenozoic. The estimated ratio of the optic lobe's surface area relative to the total brain suggests a diurnal ecology. Lithornis may provide the clearest insights to date into the neuroanatomy of the ancestral crown bird, combining an ancestrally unflexed brain with a caudally oriented connection with the spinal cord, a moderately enlarged telencephalon, and ventrally shifted, enlarged optic lobes.

. Although direct evidence of total-group palaeognaths from the Mesozoic is still lacking, the presence of fossil neognaths in the latest Cretaceous (Clarke et al., 2005;Field, Benito et al., 2020) implies that at least one lineage of total-group palaeognaths, possibly including lithornithids, must have survived the Cretaceous-Palaeogene mass extinction event approximately 66 million years ago (Widrig & Field, 2022).
The larger and more complex brains of Neornithes relative to other reptiles may be related to their capacity for complex cognitive adaptability (Sayol et al., 2016;van Dongen, 1998).The canonical circuit of the avian forebrain is functionally equivalent to the mammalian neocortex, a region of the brain responsible for higher cognitive functions, indicating a remarkable degree of functional convergence between birds and mammals in terms of cognitive capacity (Güntürkün & Bugnyar, 2016;Jarvis et al., 2005;Stacho et al., 2020).Some birds, especially parrots (Psittaciformes) and corvids (Passeriformes: Corvidae), exhibit cognitive capabilities and encephalization quotients on par with those of nonhuman primates (Balakhonov & Rose, 2017;Emery, 2006;Emery & Clayton, 2004;Güntürkün & Bugnyar, 2016;Lambert et al., 2019;Pika et al., 2020).
Unlike their neornithine counterparts, no stem group birds are known to have survived the end-Cretaceous mass extinction (Longrich et al., 2011).Several interrelated hypothesis have been put forward to help explain the implied differential survival among early crown birds and stem birds, including differences in body size (Berv & Field, 2018;Field, Benito, et al., 2020;Field, Berv, et al., 2020), diet (Larson et al., 2016), habitat preferences (Field, Bercovici, et al., 2018;Hughes et al., 2021;Mayr, 2017), and a more derived neuroanatomy and sensory system in Neornithes that may have contributed to greater behavioral complexity and adaptability (Ksepka et al., 2020;Sayol et al., 2016;Torres et al., 2021).Therefore, insights into ancestral neornithine endocranial morphology from early total-clade palaeognaths and neognaths have exciting potential to help test this latter hypothesis; nonetheless, such insights remain elusive due to a lack of suitably complete and three-dimensional cranial remains from phylogenetically relevant taxa.
Here, we describe the external and internal morphology of the braincase from the neotype specimen of Lithornis vulturinus, a stem palaeognath from the early Eocene (Ypresian) London Clay Formation.As probable early diverging stem group representatives of Palaeognathae (Nesbitt & Clarke, 2016;Yonezawa et al., 2017), the sister group to all other extant birds, comparisons of lithornithid brains with those of other early crown birds should clarify the neuroanatomy of the ancestral crown bird and may therefore reveal insights into factors contributing to neornithine survival across the end-Cretaceous mass extinction event.This work constitutes the first detailed description of a lithornithid endocast, building upon a brief description of the external morphology of the L. vulturinus neurocranium (Wharton, 2002), and a previous investigation into the olfactory bulbs (Zelenitsky et al., 2011) and optic lobes (Early, Iwaniuk, et al., 2020) in Lithornis plebius.

| MATERIALS AND METHODS
The neotype of L. vulturinus, NHMUK A5204, was described by Houde (1988) and designated as the replacement for the holotype of the species which was destroyed in World War II.The cranial material investigated here, which is separate from the phosphatic nodule containing the majority of the neotype material but also designated as NHMUK A5204, was not described in that publication.Houde (1988) states that NHMUK A5204 (formerly BMNH A 5204) was first referred to Promusophaga magnifica by Harrison and Walker (1977), though we were unable to find any mention of the specimen in the latter publication.The earliest mention of the NHMUK A5204 neurocranium in the literature is by Wharton (2002), who attempted a description of the endocast based on external features of the braincase.Wharton (2002) states that the neotype specimen was collected from Warden Point on the Isle of Sheppey in Kent, England, though more specific locality information as well as the division of the London Clay Formation from which it derives, is unknown.Nonetheless, its geographic provenance indicates that it likely derives from one of London Clay divisions C through F (Collinson et al., 2016;King, 1981King, , 1984King, , 2016)), thus corresponding to an absolute age between 53.5 and 51.5 Ma (Vandenberghe et al., 2012).Despite the wealth of bird material known from this formation, to our knowledge this is only the fourth avian endocast to be described from the London Clay, with two described previously by Milner and Walsh (2009) and one by Walsh and Milner (2011b).2016).We determined total brain surface area and optic lobe surface area using the "compute geometric measures" function in the open-source software Meshlab (Cignoni et al., 2008), after applying screened Poisson surface reconstruction under MeshLab default settings to the original segmented endocast to remove artificial surface texture from segmentation.Reconstruction of the surface area of the missing olfactory bulbs was achieved by isometrically scaling that value from the 3D mesh of the congeneric L. plebius (USNM 336534) from Zelenitsky et al. (2011) to match the larger brain surface area of L. vulturinus (scaling factor = 1.0398).This surface area was then added to that of the remaining L. vulturinus endocast to estimate its total brain surface area.Encephalization Quotient (EQ) was calculated using the equation of Smith and Clarke (2012): EQ = ECV/0.137W 0.568 , where ECV is the endocranial volume in cm 3 , and W is the body mass in g.The endocranial volume for L. plebius was obtained from Ksepka et al. (2020), and scaled isometrically to match the larger body size of L. vulturinus.However, we note that the surface area and endocranial volume of L. plebius are also approximations, as damaged sections in the original fossil were interpolated to span gaps in the reconstructed endocast (Early, Ridgely, et al., 2020).
Body mass estimates were calculated for the L. vulturinus neotype (NHMUK A5204) and L. plebius (USNM 336534) based on the equations of Field et al. (2013).The body mass of L. vulturinus was estimated from the humeral circumference and that of L. plebius from the length of the humeral articular facet, as these are the least error prone skeletal correlates of body mass that could be obtained for each specimen (Field et al., 2013).

| Anatomical description
Anatomical terms follow Baumel and Witmer (1993), with English equivalents used where possible.

| Braincase (Figure 1)
As in other lithornithids, the overall anatomy of the braincase of NHMUK A5204 is similar to that of tinamous (Houde, 1988).The temporal region is somewhat weathered on both sides of the skull, but the zygomatic process of the left squamosal is nearly complete.This process is rostroventrally directed, exhibits a notched ventral margin, and is not mediolaterally compressed.The intact postorbital process on the left side is anterodorsally compressed and tapered, similar in morphology to that of tinamous (Bertelli et al., 2014).A portion of the posterodorsal mesethmoid is preserved, which separates the orbitocranial fonticuli and the optic nerve foramen.
The parasphenoidal lamina is mostly complete.As in all palaeognaths, the rostral openings of the eustachian tubes are widely separated (Baumel & Witmer, 1993;Houde, 1988).The parasphenoidal rostrum is broken at its cranial end.As expected for palaeognaths, the basipterygoid processes are robust, and are similar in shape and relative size to those of other lithornithids (Houde, 1988).The left basipterygoid process is undamaged and exhibits a flat articular surface for its articulation with the pterygoid.The ala parasphenoidalis is intact on the left and entirely missing on the right.It is triangular and well developed in ventral view, matching that of L. plebius (Houde, 1988;Zelenitsky et al., 2011).In dorsal view, damage to the roof of the skull clearly follows the unfused suture between the frontals and the parietals.Wharton (2002) interpreted these open sutures as evidence that this specimen was not skeletally mature; however, we consider this interpretation unlikely as visible sutures are also retained in adult tinamous, and are seen in all other known lithornithid crania (Houde, 1988;Nesbitt & Clarke, 2016).Most of the right frontal is missing, while the left is damaged.Damage to the medial portion of the frontals and laterosphenoids exposes the endocranial cavity, which is infilled with sediment.We did not observe the dendrite-shaped furrows for blood vessels on the dorsal surface of the cranium noted by Mayr (2009) for Lithornis specimen IRSNB Av82.
In caudal view, the foramen magnum is subcircular, with a flattened ventral edge.The foramen magnum opens caudally, forming an angle of 120°with the skull base.The external occipital vein foramina are located dorsal to the foramen magnum.A dorsal position of these foramina relative to the cerebellar prominence has been interpreted as a synapomorphy of Tinamidae (Bertelli, 2017), yet the condition in both NHMUK A5204 and L. plebius (USNM 336534) (Zelenitsky et al., 2011) suggests this trait is shared with Lithornithidae.Nonetheless, this cannot be fully confirmed as the cerebellar prominence itself has been lost to weathering in NHMUK A5204.The occipital condyle is about one-third the width of the foramen magnum, and the subcondylar fossa is deep.This contrasts with L. plebius, in which the occipital condyle is more than half the width of the foramen magnum (Zelenitsky et al., 2011).As expected for a palaeognath, no occipital fonticuli are present (Baumel & Witmer, 1993).

| Brain endocast (Figure 2)
Although incomplete due to missing portions of the frontals and laterosphenoids, the endocast of NHMUK A5204 clearly illustrates a brain morphology typical of crown birds, as evidenced by its ventrally shifted optic lobes and expanded telencephalon relative to stem group birds (Torres et al., 2021).Based on the external morphology of the braincase, Wharton (2002) speculated that the endocast would reflect the crown bird condition; with the ability to view the endocast itself via CT scanning we can now confirm this to be true.For comparative purposes, we examined digital endocasts belonging to exemplars from each major extant palaeognath subclade: Struthioniformes (Struthio), Rheiformes (Rhea), Casuariiformes (Dromaius), Apterygiformes (Apteryx), and the two major subclades that make up Tinamiformes: the open habitat-dwelling Nothurinae (Eudromia), and the forest-dwelling Tinaminae (Crypturellus).Of these, the endocast of NHMUK A5204 most closely resembles those of the tinamous Eudromia and Crypturellus in overall shape (Figure 3).It is also similar to its congener L. plebius (USNM 336534; Zelenitsky et al., 2011; Figure 3).The brain is unflexed; that is, the medulla opens caudally with the telencephalon and optic tectum cranial to the cerebellum and the medulla, as in most extant palaeognaths, galloanserans, and some neoavian lineages, such as Phaethoquornithes (Aequornithes + Phaethontimorphae; Stiller et al., 2024) and Gruiformes (Chiappe, Navalón et al., 2022).As preserved, the surface area of the endocast is 12.30 cm 2 .Assuming the same olfactory ratio as L.
plebius to correct for the missing olfactory bulbs in NHMUK A5204, the total surface area of the brain would have been approximately 13.14 cm 2 (Table 1).As preserved, the endocranial volume of NHMUK A5204 is 2.76 cm 3 .The endocranial volume of L. plebius is 2.51 cm 3 (Ksepka et al., 2020).The estimated body mass of the L. vulturinus neotype is 956 g, comparable to the mean mass of male Great Tinamous (Tinamus major), which average 960 g (Dunning, 2007).Assuming the same brain to body size ratio for the L. vulturinus neotype as L. plebius provides a total endocranial volume of 2.93 cm 3 for L. vulturinus.The encephalization quotient (EQ) of L. plebius (and therefore the estimated EQ of L. vulturinus) is 0.418, which falls well within the range of EQs of extant palaeognaths (Table 1).

Telencephalon (Figure 2)
Due to damage to the dorsal telencephalon, the size and shape of the eminentiae sagittales, commonly known as the Wulst, cannot be determined for NHMUK A5204.These paired features of the dorsal telencephalon represent the endpoint of the thalamofugal visual pathway in birds (Degrange et al., 2023;Güntürkün et al., 1993), and are low and weakly defined in L. plebius (USNM 336534) (Zelenitsky et al., 2011).The olfactory bulbs and nerves are also missing due to the damaged frontals.The olfactory bulbs are large in L. plebius (USNM 336534) (Zelenitsky et al., 2011), and we hypothesize that both the Wulst and the olfactory bulbs in L. vulturinus would have been similar to those of its congener.Despite the damage sustained by the fossil, the overall shape of the telencephalon is clear.It is broader than it is long, and is widest caudally as in most extant birds (Walsh & Milner, 2011a;Walsh et al., 2016) and all of the palaeognaths in our sample with the exception of Apteryx, where the telencephalon is especially elongate (Corfield et al., 2008;Early, Ridgely, et al., 2020;Torres & Clarke, 2018) (Figure 3).There are no casts of blood vessels or defined sinuses visible on the telencephalon's surface.Due to abrasion of the endocast's dorsal surface, we cannot comment on the state of the interhemispheric fissure, but in  L. plebius (USNM 336534) (Zelenitsky et al., 2011) the interhemispheric fissure is well defined though shallower than in some extant palaeognaths such as Dromaius and Rhea, most closely resembling those of the tinamous Crypturellus and Eudromia (Figure 3).

Mesencephalon
Most of the mesencephalon visible on the surface of the endocast is taken up by the large, intact optic lobes, which each have a surface area of 0.74 cm 2 .The optic lobes are positioned ventral to the telencephalic hemispheres, as it is typical of crown birds, and at least some putative stem birds such as Cerebavis cenomanica (Balanoff et al., 2013;Chiappe, Navalón et al., 2022;Field, Hanson, et al., 2018;Kurochkin et al., 2006;Torres et al., 2021;Walsh et al., 2016).They are similar in relative size and share their rounded teardrop shape in lateral view with L. plebius (USNM 336534) (Zelenitsky et al., 2011) (Figure 3), in which the surface area of each is 0.865 cm 2 (Early, Ridgely, et al., 2020).Among extant palaeognaths, they are again most similar in shape to Crypturellus and Eudromia (Figure 3).

Diencephalon (Figure 2)
The pituitary gland is large and appears identical in shape to that of L.
plebius (USNM 336534) (Zelenitsky et al., 2011).It is elongate, subrectangular, and caudally projected, similar to Crypturellus and Eudromia and unlike the more rounded glands of Rhea and Dromaius or the rostrally projecting glands of Struthio and Apteryx (Figure 3).As in other extant birds, there is no expression of the pineal gland on the endocast surface (Walsh et al., 2016).We were able to identify both the left and right optic nerves (Figure 2), but unable to determine the morphology of their exit type according to Hall et al. (2009) because the orbit and most of the interorbital septum are not preserved.A portion of the carotid artery is visible on the exterior of the endocast and can be partially traced within the endocast, with the intercarotid anastomosis corresponding to type I of Baumel and Gerchman (1968), in which the two carotids are merged for a considerable distance.
Cerebellum (Figure 2) The cerebellum is intact, as are the occipital veins lying just ventral to it.Wharton (2002) expected the cerebellum to be tilted along the anterodorsal-posteroventral axis based on the morphology of the neurocranium, and we confirm this to be the case.In both L. plebius and L. vulturinus, the sulcus separating the cerebellum and telencephalon is more pronounced than in Crypturellus and Eudromia, and most closely resembles the condition in Dromaius among extant palaeognaths (Figure 3).The cerebellar folia are difficult to discern, likely due to poor contrast between matrix and bone, a common problem encountered with London Clay fossils (Beckett et al., 2016).
The cerebellum is not partially covered by the telencephalon, unlike ventrocaudally directed but wider and less elongate (Figures 5 and 6).
However, the vestibule of NHMUK A5204 is substantially larger than it is in Hesperornis or Rhynchotus, and more similar in size and shape to those of Nothoprocta, Crypturellus, and Eudromia (Figures 5 and 6).
In lateral view, the anterior semicircular canal of Lithornis is more sample (Figures 5 and 6).As in most extant birds, this is the most expanded of the three semicircular canals (Walsh & Milner, 2011b).
The anterior, posterior, and lateral ampullae are all well developed and distinct.The posterior canal is small and subtriangular in both Lithornis and Hesperornis, though the position of this canal with respect to the rest of the inner ear differs in these taxa.In Hesperornis, the ascending portion of the posterior canal faces posteriorly, while in Lithornis the ascending canal faces posterolaterally and is more similar to the condition in Dromaius.The crus commune in Lithornis is long due to the caudal origin of the posterior semicircular canal being located further dorsally than in taxa such as the tinamous Rhynchotus, Nothoprocta, and Eudromia, and is more similar to the condition in Crypturellus (Figure 5).The anterior and posterior canals are elliptical in cross-section, while the lateral canal is subcircular.While the semicircular canals of most extant birds are circular in cross-section, elliptical canals are seen in some palaeognaths (e.g., Struthio) and neognaths (e.g., Galloanserae), but the phylogenetic distribution of these traits remains incompletely understood (Walsh & Milner, 2011b).The presence of a Wulst has also been proposed to be a plesiomorphic trait of Neornithes (Balanoff et al., 2013;Milner & Walsh, 2009;Torres et al., 2021;Walsh & Milner, 2011b), and although its presence cannot be confirmed in L. vulturinus, a Wulst does appear to be present in L. plebius (USNM 336534) (Zelenitsky et al., 2011).The phylogenetic distribution of these neuroanatomical features among extant birds, as well as their presence in early stem group representatives of Palaeognathae, support the hypothesis that they would have been present in the most recent common ancestor of crown group birds.The presence of the type three floccular fossa shape in lithornithids as well as extant palaeognaths (Walsh et al., 2013) suggests retention of the plesiomorphic floccular morphology for crown Palaeognathae since the early Cenozoic.
Although complete olfactory bulbs are missing in NHMUK A5204, the olfactory bulbs of L. plebius (USNM 336534) are greatly enlarged compared with most extant birds, falling within the range of variation of olfaction-reliant taxa such as tubenoses (Procellariiformes) or kiwis (Zelenitsky et al., 2011).Although previous ancestral state reconstructions suggested that large olfactory bulbs would have been present at the base of Neornithes (Zelenitsky et al., 2011), we treat that inference with caution and suggest that enlarged olfactory bulbs may have evolved independently among lithornithids, perhaps associated with an olfaction-dependent foraging mode.
The volume of the L. plebius endocast (USNM 336534) (Zelenitsky et al., 2011) is approximately 2.51 cm 3 (Ksepka et al., 2020), though this is a conservative estimate due to some compression damage to the right telencephalic hemisphere.We estimated a mean body mass for L. plebius of ~808 g from the length of the humeral articular facet of the coracoid using the equations of within the expected range for extant palaeognaths (Table 1), in which only Apteryx approaches an EQ of 1, representing a clear outlier with respect to the remainder of Palaeognathae (Corfield et al., 2008;Ksepka et al., 2020).The encephalization quotient of L.
vulturinus as preserved is 0.409, though it should be noted that this is a slight underestimate due to the missing olfactory bulbs.An EQ less than 1 indicates a smaller brain size than expected for a given body mass and greater than 1 indicates a larger brain size than expected (Jerison, 1973).By contrast, birds known for their intelligence such as the Common Raven, Corvus corax, and the Red-and-green Macaw, Ara chloropterus, exhibit EQs of 2.68 and 3.03, respectively (Carril et al., 2016;Ksepka et al., 2020).Based on these estimates, L. plebius and L. vulturinus fall along the brain-body size regression line for nonavian theropods and early crown birds, indicating that lithornithids are not more encephalized compared with extant palaeognaths and deeply diverging neognath lineages such as Galloanserae.This result corroborates hypotheses of a common allometric scaling relationship characterizing the origin of crown birds as well as stem birds and nonavian theropods, despite disparate brain morphologies (Ksepka et al., 2020).Unlike palaeognaths and galloanserans, several subclades within Neoaves appear to have attained higher relative brain-to-body size ratios through reduction of body size across the K-Pg transition (Berv & Field, 2018;Ksepka et al., 2020).The retention of a plesiomorphic scaling relationship between brain volume and body mass in totalgroup palaeognaths may be indicative of a reduced capacity for behavioral complexity in the aftermath of the end-Cretaceous mass extinction event, which may have limited their capacity for ecological niche-filling during an interval of ecological release on a global scale.Neoaves, by contrast, with comparatively high encephalization quotients (Ksepka et al., 2020), may have quickly colonized vacant niche space and rapidly diversified in the extinction's aftermath (Prum et al., 2015;Stiller et al., 2024).

| Evolution of brain flexion
In many extant birds, the brain is flexed along its rostral-caudal axis with a ventralized connection to the spinal column.This condition contrasts with the stem avian and nonavian theropod condition in which the rostral and caudal extremities of the brain are roughly aligned, and the brain connects caudally with the spinal column (Bever et al., 2011;Chiappe, Navalón et al., 2022;Hofer, 1952;Witmer & Ridgely, 2009;Witmer et al., 2008;Yu et al., 2024).Both Lithornis endocasts exhibit a caudally positioned occiput and show little flexion in the brains, which is typical of all extant palaeognaths (Chiappe, Navalón et al., 2022;Corfield et al., 2008).Among neognaths, the brain morphology of Lithornis most closely resembles the unflexed brains of Galliformes such as Megapodius, Meleagris, and Gallus, though these lack well-developed olfactory bulbs (Figure 3).
Many Anseriformes like Anas (Figure 7) also exhibit unflexed brains, though the optic lobes are shifted further proximally than in Lithornis, and the cerebellum appears reduced.Notably, the early diverging anseriform Anhima exhibits a slightly more flexed brain (the optic lobes located directly ventral to the telencephalon) with a more ventralized occiput than in Lithornis, revealing a degree of ventralization even within lineages generally exhibiting non-ventralized brains (Figure 3) (Chiappe, Navalón et al., 2022).Although the variation of brain flexion and ventralization in crown birds is a continuum, it seems clear that relatively nonventralized morphologies likely characterized the ancestral conditions of crown Palaeognathae and Galloanserae, while more extreme ventralization of the occiput only evolved within neoavian clades outside Phaethoquornithes (Figure 7).
For example, hummingbirds and many passerines both exhibit high degrees of ventralization.Within Phaethoquornithes, birds such as tropicbirds (Phaethon) and the stem-group tropicbird Prophaethon do not show a strong degree of ventralization.This is in keeping with the general observation of reduced ventralization in aquatic birds (Chiappe, Navalón et al., 2022).In light of this variability and lingering phylogenetic uncertainty regarding the interrelationships of the major subclades of Neoaves (Jarvis et al., 2014;Prum et al., 2015;Reddy et al., 2017;Stiller et al., 2024), it remains challenging to confidently infer whether the last common ancestor of Neoaves would have exhibited a high degree of brain flexion and occiput ventralization.
Little Mesozoic avialan endocranial material exists for comparison with Lithornis.The brains of dromaeosaurid theropods and early avialans such as Archaeopteryx are clearly unflexed, with the foramen magnum and thus the occiput positioned caudally (Balanoff et al., 2013;Chiappe, Navalón et al., 2022;Yu et al., 2024) (Figure 7).The endocasts of the crownward ornithurine Ichthyornis (Torres et al., 2021) and the putative ornithurine Cerebavis are also unflexed, with the foramen magnum opening caudally, although the precise phylogenetic affinities of the latter are uncertain due to the holotype specimen's incompleteness (Kurochkin et al., 2006;Walsh et al., 2016).Closer to the crown group, the early hesperornithine Enaliornis also exhibits a nonventralized occiput with an angle between the cranial base and foramen magnum of ~120° (Chiappe, Navalón et al., 2022;Elzanowski & Galton, 1991).Although the recent discovery of an enantiornithine with a more ventralized foramen magnum than most non-neoavian crown birds raised the question of whether a flexed brain and ventralized occiput evolved independently in enantiornithines and within crown birds (Chiappe, Navalón et al., 2022), or whether these traits would have been present in the last common ancestor of Ornithothoraces (Figure 7), we suggest that balance of available evidence from avialans support the hypothesis that the earliest crown birds retained an ancestrally unflexed brain with a caudally oriented connection with the spinal cord.
The determinants of marked ventrally directed flexure of the brain in many extant neoavian birds remain incompletely understood, but may involve spatial competition within the developing skull as a result of the interplay between eye size, body size, and the size of the brain and cranial base (Kawabe et al., 2015;Marugán-Lobón et al., 2022).Evidence from chickens shows that although the foramen magnum is oriented caudally posthatching, it is more ventralized during earlier stages of embryonic development, suggesting a degree of covariation between macroevolution and development (Kawabe et al., 2015).
Halcyornithidae has been proposed to represent a clade of stem group Psittaciformes (Ksepka et al., 2011;Mayr, 2002Mayr, , 2022)), the sister group to Psittacopasseres (Ksepka et al., 2019;Mayr, 2020Mayr, , 2021)), or may represent stem group members of Eufalconimorphae (Mayr & Kitchener, 2023).Walsh and Milner (2011b) found the brain shape of Halcyornis comparable to extant Laridae, though Mayr and Kitchener (2023) note those similarities may reflect retained neornithine plesiomorphies.The foramen magnum exhibits a comparable degree of flexion to Lithornis (Walsh & Milner, 2011b), while the Wulst is well developed and the olfactory bulbs relatively large (Walsh & Milner, 2011b).Given the inferred phylogenetic position for Halcyornis as a derived telluravian, its relatively plesiomorphic degree of foramen magnum ventralization may be indicative of multiple independent evolutionary acquisitions of pronounced brain flexion within Neoaves.
Among extant and recently extinct palaeognaths, the optic lobes are greatly reduced in elephant birds and nearly obsolete in kiwi, a feature likely correlated with a nocturnal or crepuscular lifestyle and a heavy reliance on smell and proprioception for locating food (Torres & Clarke, 2018).They are also reduced in some species of moa, though less dramatically so (Early, Ridgely, et al., 2020;Torres & Clarke, 2018).We observe no such reduction of the optic lobes in either L. vulturinus or in L. plebius, indicating that both species were likely reliant on visual cues despite having enlarged olfactory lobes.
The optic ratio of L. vulturinus, calculated as the surface area of one optic lobe versus the entire brain's surface area, is 0.056 (Table 2).
This is slightly less than that of tinamous, which range between 0.08 and 0.09 in our sample, but is similar to the Greater Rhea (Rhea americana), a diurnal bird that is heavily reliant on visual input (Torres & Clarke, 2018).
The cerebellar flocculi of NHMUK A5204 are intact.These paired structures, termed auricula cerebelli by Baumel and Witmer (1993), integrate visual and vestibular stimuli and are responsible for T A B L E 2 Optic lobe surface area, total brain surface area, and optic ratios of selected palaeognaths.gaze stabilization through the vestibulo-collic and vestibulo-ocular reflexes (Voogd & Wylie, 2004).Despite their completeness, they unfortunately provide little insight into the ecology of L. vulturinus.
Relative flocculus size is not related to flight style in extant birds, nor are there significant differences between flightless and volant taxa (Walsh et al., 2013), and no correlation has been found between the relative size of floccular fossae and behavior and ecology in extant birds and mammals (Ferreira-Cardoso et al., 2017).The intercarotid anastomosis, which is preserved intact in NHMUK A5204, is seemingly also ecologically and phylogenetically uninformative (Degrange et al., 2023).
Conflicting information exists as to whether the morphology of the semicircular canals can yield information pertinent to the ecology and flight style of fossil birds.Using a sample of 64 extant avian crania, Benson et al. (2017) found little correlation between flight style or wing kinematics and semicircular canal size and shape.
Instead, the shape of the canals is influenced mainly by spatial constraints imposed by the braincase, and larger birds tend towards exhibiting proportionally longer canals (Benson et al., 2017).Similar results were obtained by Bronzati et al. (2021) using a data set of both extant and extinct archosaurs, who found that only 10.7% of geometric shape variation was attributable to mode of locomotion.
The authors concluded that semicircular canal shape is affected mainly by the spatial constraints imposed by the brain and braincase.The portion of the cranium overlying the probable location of the Wulst is damaged in NHMUK A5204, and the presence of a Wulst cannot be unequivocally assessed.However, based on the condition in L. plebius, we expect that a weakly developed Wulst would also have been present in L. vulturinus.The Wulst has been linked to somatosensory integration and higher cognition (Reiner et al., 2005), and heightened activity in the Wulst during aerial locomotion suggests that its evolutionary origin may be associated with the refinement of avian flight (Balanoff et al., 2013;Gold et al., 2016).
The crownward non-neornithine avialan Ichthyornis appears to have had an incipient Wulst, indicating that the presence of a Wulst may be plesiomorphic for Ornithurae (Torres et al., 2021).In L. plebius (USNM 336534) (Zelenitsky et al., 2011), the Wulst is not as prominent as in many extant birds and occupies an intermediate position between the rostrally positioned Type A and caudally positioned Type B Wulsts as characterized by Stingelin (1958), which is also observed in extant penguins (Sphenisciformes) (Degrange et al., 2023;Tambussi et al., 2015).
As with the Wulst, the olfactory bulbs of L. vulturinus cannot be directly assessed due to damage in NHMUK A5204.However, given the high degree of similarity between the preserved portions of the endocasts of L. vulturinus and L. plebius, it seems reasonable to assume that the endocast of L. vulturinus would have originally exhibited comparably large olfactory bulbs, consistent with a welldeveloped sense of smell (Zelenitsky et al., 2011).The olfactory ratio of L. plebius, calculated as the length of the olfactory bulb relative to that of the cerebral hemisphere (Cobb, 1968), is 37.1 (Zelenitsky et al., 2011).This ratio is similar to that of extant olfactory foraging taxa such as kiwi and Procellariiformes, and suggests that smell could have played an important role in both food location and navigation in Lithornis (Zelenitsky et al., 2011).
Large olfactory bulbs in birds have been linked with both foraging over a wide area and with low light conditions such as densely forested habitats, where olfaction plays a greater role relative to visual cues (Martin et al., 2007;Nevitt et al., 2008;Torres & Clarke, 2018).The paleoenvironment of the London Clay Formation typifies the hothouse conditions of the early Eocene, with the shoreline covered in coastal mangroves and palms (Collinson, 1983), and inland vegetation generally resembling present-day Southeast Asian rainforests (Collinson, 2000).In the dense understory of the mangrove swamps and lush paratropical forests of early Eocene England, a keen sense of smell may therefore have been advantageous for lithornithids, which are thought to have been largely terrestrial probe-feeders (du Toit et al., 2020;Houde, 1988).This inferred foraging ecology is reminiscent of the extant kiwi, and, given that ground-feeding birds are more likely to evolve flightlessness in predator-depauperate contexts than arboreal taxa (Wright & Steadman, 2012), the numerous inferred transitions toward flightlessness among extant palaeognaths may be related to this ancestral ecology.

| CONCLUSIONS
Lithornithid neuroanatomy shares many features in common with that of crown group palaeognaths and galloanserans, as well as some early fossil neoavians.Features such as an expanded telencephalon and optic bulbs, and an unflexed brain with a caudally directed connection with the spinal cord, likely reflect the condition of the ancestral crown bird.We interpret L. vulturinus as a diurnal bird with acute vision that likely had a well-developed sense of smell.If lithornithid brain morphology approximates the ancestral condition for crown group palaeognaths, then the brains of tinamous appear to have changed little throughout palaeognath evolutionary history, with the exception of their reduced olfactory lobes.Questions regarding the morphology of the ancestral crown bird brain remain, yet continued investigations of adequately preserved early crown birds as well as crownward stem birds will bring us ever closer to understanding how and when the characteristic brains of crown group birds evolved.
NHMUK A5204 was scanned on a Nikon Metrology XT H 225 ST High Resolution CT Scanner at the Natural History Museum, London (175 kV, 257 uA, 85 frames, 8 frames per projection, voxel size = 12 μm).Scans were digitally segmented (i.e., digital reconstruction of three-dimensional volumes) and rendered in VGSTUDIO MAX v3.4 (Volume Graphics), following best practices for the generation of the digital endocasts of the brain and inner ear as outlined by Balanoff et al. (

F
I G U R E 4 Left endosseous inner ear labyrinth of Lithornis vulturinus (NHMUK A5204) in lateral, anterior, medial, posterior, dorsal, and ventral views.Anterior semicircular canal in blue, posterior canal in green, and lateral canal in yellow.Scale bar = 5 mm.the condition in some neoavians in which the brain flexes upon itself to partially or completely conceal the cerebellum below the telencephalon in dorsal view(Degrange et al., 2023).As with the other palaeognaths in our sample including L. plebius (USNM 336534)(Zelenitsky et al., 2011), there is no visible trace of the occipital sinus in our reconstructed endocast of NHMUK A5204.Both flocculi projecting from the cerebellum are preserved intact; these structures are much larger and more strongly projected laterally than in either tinamou taxon examined and are nearly identical to those of L. plebius (USNM 336534)(Zelenitsky et al., 2011) (Figure3).The fossae correspond to avian floccular fossa type three ofWalsh et al. (2013), exhibiting the following characteristics: the fossae enclose the arterial loop, the base of the fossae are not domed, the fossae are elongate and approximately circular in cross section, and the distal ends of the fossae are not tapered.The semicircular canals of the inner ear encircle the fossae.Medulla (Figure 2)The medulla is subcircular in ventral view, with no detectable median sulcus.Within our sample of palaeognaths, the medulla is reminiscent of Crypturellus and Eudromia and differs from the more elongate medullae of Struthio and Rhea.We were able to locate the trigeminal nerve (cranial nerve V) but could not distinguish its three main branches.The abducens nerve (CN VI) is narrow, and the glossopharyngeal, vagus, and accessory nerve bundle (CN IX-XI) are not differentiated over the length of the preserved cranial remains.These nerves, as well as the optic nerve (CN II), are the only cranial nerves that can be traced in NHMUK A5204.None can be traced far from their exits on the endocast surface.Inner ear (Figure 4) Almost the entirety of the bony labyrinth of the left inner ear is preserved intact save for some slight abrasion to the ascending posterior canal.The right cochlear duct was difficult to trace due to particularly low contrast between bone and matrix in this area.Descriptions of the cochlear duct are therefore based on the left inner ear.The overall shape of the semicircular canals and cochlear duct are remarkably similar to those of the Late Cretaceous ornithurine Hesperornis and the inferred ancestral state for Neornithes hypothesized by Hanson et al. (2021) (Figure 5).Like the condition in Hesperornis, the cochlear duct of NHMUK A5204 is elongate and ventrally directed, unlike the condition in the tinamou Rhynchotus in which the canal is directed ventrorostrally, and Nothoprocta, Crypturellus, and Eudromia in which the canal is ventrally or elliptical in shape than in Hesperornis or other palaeognaths such as Rhynchotus or Struthio and is most similar to Nothoprocta within our F I G U R E 5 Left endosseous inner ear labyrinths of (a) Lithornis vulturinus, (b) Hesperornis regalis, (c) Struthio camelus, (d) Dromaius novaehollandiae, (e) Rhynchotus rufescens, (f) Nothoprocta perdicaria UMZC unaccessioned, (g) Crypturellus tataupa UMZC unaccessioned, and (h) Eudromia elegans UMZC 156966 in left lateral view.Hesperornis, Rhea, Dromaius, and Rhynchotus redrawn from Hanson et al. (2021).Anterior semicircular canal in blue, posterior canal in green, and lateral canal in yellow.Not to scale.

|
Ancestral crown bird endocranial morphologyAs a probable stem-group member of the sister taxon to all other crown birds, L. vulturinus has potential to provide important insight into the neuroanatomy of the ancestral neornithine.Neuroanatomical traits observable in NHMUK A5204, such as a laterally expanded telencephalon and enlarged optic lobes, have been proposed to represent synapomorphies of Neornithes(Torres & Clarke, 2018;Torres et al., 2021).

F
I G U R E 6 Time-calibrated phylogenetic tree illustrating the evolution of the inner ear and semicircular canals across Avialae.Taxa with orange nodes and arrows exhibit substantial brain flexion and occiput ventralization, reflected by the caudal extension of the anterior semicircular canal.Yellow bar indicates time interval encompassed by the London Clay Formation.Dashed red line indicates the K-Pg boundary.MPM-334-1 redrawn from Chiappe, Navalón et al. (2022), Cerebavis redrawn from Walsh et al. (2016), Selasphorus and Passer redrawn from Benson et al. (2017), Rhea and Archaeopteryx redrawn from Hanson et al. (2021), Dasornis and Prophaethon redrawn fromMilner and Walsh (2009), Meleagris redrawn fromCerio and Witmer (2019).Crypturellus tataupa and Anas crecca drawn from unaccessioned UMZC specimens.Node ages and interrelationships of crown birds followPrum et al. (2015).

F
I G U R E 7 Time-calibrated phylogenetic tree illustrating the evolution of the avian brain and occiput ventralization across Avialae.Arrows indicate orientation of the foramen magnum.Taxa with orange nodes and arrows exhibit substantial brain flexion and occiput ventralization.Optic lobes are highlighted in teal.Yellow bar indicates time interval encompassed by the London Clay Formation.Dashed red line indicates the K-Pg boundary.Archaeopteryx redrawn from Balanoff et al. (2013), MPM-334-1 and Selasphorus redrawn from Chiappe, Navalón et al. (2022), Cerebavis redrawn from Walsh et al. (2016), Ichthyornis redrawn from Torres et al. (2021), Rhea and Crypturellus redrawn from Torres and Clarke (2018), Dasornis and Prophaethon redrawn from Milner and Walsh (2009), Dromornis redrawn from Handley and Worthy (2021).Node ages and interrelationships of crown birds follow Prum et al. (2015).
Contrary to these findings,Hanson et al. (2021) report three vestibular morphotypes corresponding to quadrupeds, bipeds and simple fliers, and maneuverable fliers.Though we did not undertake a geometric morphometric shape analysis of the semicircular canals, the lack of extension of the anterior semicircular canal of L. vulturinus places it inHanson et al. (2021)'s simple fliers group, suggesting that it would not have exhibited highly acrobatic flight.Nonetheless, arguments thatHanson et al. (2021) did not adequately control for the effect of shared ancestry within their taxonomic sample, as well as functional analyses demonstrating that different semicircular duct shapes can perform comparable functions(David et al., 2022), cast doubt on the strength of functional interpretations from inner ear morphology alone.