New Remains of Scandiavis mikkelseni Inform Avian Phylogenetic Relationships and Brain Evolution

Although an increasing number of studies are combining skeletal and neural morphology data in a phylogenetic context, most studies do not include extinct taxa due to the rarity of preserved endocasts. The early Eocene avifauna of the Fur Formation of Denmark presents an excellent opportunity for further study of extinct osteological and endocranial morphology as fossils are often exceptionally preserved in three dimensions. Here, we use X-ray computed tomography to present additional material of the previously described taxon Scandiavis mikkelseni and reassess its phylogenetic placement using a previously published dataset. The new specimen provides novel insights into the osteological morphology and brain anatomy of Scandiavis. The virtual endocast exhibits a morphology comparable to that of modern avian species. Endocranial evaluation shows that it was remarkably similar to that of certain extant Charadriiformes, yet also possessed a novel combination of traits. This may mean that traits previously proposed to be the result of shifts in ecology later in the evolutionary history of Charadriiformes may instead show a more complex distribution in stem Charadriiformes and/or Gruiformes depending on the interrelationships of these important clades. Evaluation of skeletal and endocranial character state changes within a previously published phylogeny confirms both S. mikkelseni and a putative extinct charadriiform, Nahmavis grandei, as charadriiform. Results bolster the likelihood that both taxa are critical fossils for divergence dating and highlight a biogeographic pattern similar to that of Gruiformes.


Introduction
The early Eocene Fur Formation of north-western Jutland, Denmark, is well-known for its rich fossil bird fauna that includes three-dimensionally preserved, articulated specimens, as well as occasional soft tissues [1][2][3]. These fossils constitute one of the oldest diverse Cenozoic bird faunas [1,3], existing only 11 Ma after the Cretaceous-Paleogene mass extinction during a time of rapid neornithine radiation that saw the appearance of most major extant avian clades [4][5][6]. The diatomaceous sediments of the Fur Formation are also unusual in the Paleogene fossil record of birds in deriving from a marine offshore environment [3]. Articulated, three-dimensional bird skeletons in this formation are typically preserved within hard calcareous carbonate concretions, believed to have protected the remains from compaction [3,7].
The age and exceptional level of anatomical detail preserved within recovered fossils from the Fur Formation makes them critical to better understanding early neornithine evolution. Despite this, many of the taxa have not been studied in detail. Previous research has identified different taxa assigned to some of the major clades of modern birds, including Apodiformes [8,9], Pelecaniformes [10], Coliiformes [8], Galliformes [11], Table 2. The sampled taxa and publications for brain and osseous labyrinth characters used for the phylogenetic analysis.

Character Matrix
The data matrix is based on that of Musser and Clarke [18], an expansion on Musser and Cracraft [44]. We added 28 additional discrete morphological characters characterizing brain anatomy from Smith and Clarke [56], as the new specimen allowed for description and scoring of brain morphology. Character descriptions are provided in the Appendix A and the data matrix has been made publicly available on Morphobank [61] under Project 3614.

Phylogenetic Analysis
We first performed unconstrained parsimony analyses of the dataset in PAUP [62] Version 4.0a169, build 164 (86) using 10,000 random taxon addition replicates per run. Heuristic search algorithms were used. Tree bisection reconnection branch swapping was employed and minimum branch lengths valued at zero were collapsed. No character weighting was applied. Characters 245, 320, and 688 were ordered, following Bertelli et al. [14] and Musser et al. [15]. Bootstrap analyses were performed using 500 bootstrap replicates each with 10 random taxon addition replicates as in [63]. In addition to unconstrained analyses, constrained analyses following those of Musser and Clarke [18] were performed using  ). Bone is unfilled, the matrix is grey, and large voids in the bone are delimited in black. Anatomical abbreviations: fr, frontal bone; fm, foramen magnum; mnd, mandible; na, nares; nc, neural cavity; or, orbit.

Measurements
See Table 3. Table 3. Linear measurements in millimeters (mm) of NMHD 625345, taken from the 3D rendering, compared with measurements of Scandiavis mikkelseni and Nahmavis grandei. Total skull length is measured from the anterior end of the premaxilla to the cerebellar prominence.  Table 3. Table 3. Linear measurements in millimeters (mm) of NMHD 625345, taken from the 3D rendering, compared with measurements of Scandiavis mikkelseni and Nahmavis grandei. Total skull length is measured from the anterior end of the premaxilla to the cerebellar prominence.

Referral and Emended Diagnosis
Prior to phylogenetic analysis, we assigned NHMD 625345 to Scandiavis based on the following diagnostic characters from Bertelli et al. [17]: (1) skull with long narial openings, (2) pars symphysialis of mandible with a flat ventral surface, and (3) presence of a dorsally recurved processus retroarticularis of the mandible (character 238: state 1 and 239:1 of Musser and Clarke [18]. We additionally assigned NHMD 625345 to Scandiavis based on the following combination of characters from Musser and Clarke [18]: a rostrum that is slightly curved ventrally at the apex (2:1), nares that are both rostrally (13:1) and caudally (10:2) rounded, nares that are over half the length of that of the rostrum (11:3), nares that are holorhinal or rostral to the zona flexoria craniofacialis (15:1), presence of furrowing along the midline of the interorbital area (32:2), presence of a supraorbital crest (49:1) that is dorsally projected (50:1), presence of a craniocaudally extensive fonticulus interorbitalis (90:2) that is not confluent with the fonticulus orbitocranialis (93:1), an occipital that is subequal in rostrocaudal position to that of the nuchal crest (158:2), a nuchal crest that is ventral to the dorsal base of the postorbital process (159:1), and a symphysis of the mandible that is less than 1/5 the length of the mandible (216:1). Phylogenetic analysis of NHMD 625345 were additionally performed with NHMD 625345 as a separate taxon. Results placed NHMD 625345 as the sister-taxon of Scandiavis.
3.1.6. Differential Diagnosis from Nahmavis grandei Scandiavis mikkelseni differs from Nahmavis grandei in the following combination of character states: (1) nares that are more rostrocaudally and dorsoventrally extensive, (2) a more obtuse antorbital angle that is approximately 90 degrees, (3) a more rostrocaudally truncate cranium, (4) a rostroventrally oriented postorbital process [18], (5) a dorsoventrally wider rostral fenestra of the mandible, (6) a more prominent dorsal mandibular angle, (7) an articular of the mandible that is located markedly ventral to the ramus, (8) a more cranially projected crista cnemialis cranialis of the tibiotarsus [18], (9) a crista cnemialis cranialis with a more rounded distal apex than that of N. grandei, (10) presence of a notch along the distal rim of the medial condyle of the tibiotarsus, (11) a more shallow fossa parahypotarsalis lateralis in the tarsometatarsus, and (12) a shorter femur and tibiotarsus.

Cranium
Three-dimensional skeletal remains of NHMD 625345 are visible on the exposed surfaces of the slabs, including the endocranial cavity, the frontal and orbital region, as well as most of the bill and mandible ( Figure 1). CT data (see Appendix A) revealed skull morphology hidden by the sediment matrix ( Figure 2). The maximum length of the skull is 47.2 mm. The nares are extensive, taking up almost the entire length and height of the rostrum as noted in the diagnosis of Scandiavis based on the holotype specimen [17]. This is similar to the condition in Nahmavis grandei, Burhinus, Charadrius, and Jacana, although N. grandei and the extant taxa exhibit more of the terminal premaxilla that is not perforated by the nares. This is in contrast to Haematopus and many Gruiformes including Psophia, Heliornis, Sarothrura, and Himantornis that exhibit nares that are approximately half the length of the rostrum. As in Burhinus and N. grandei, the nares are rostral to the zona flexoria craniofacialis and would have likely been rounded at the caudal terminus, as in holorhinal taxa. All Gruiformes are holorhinal as well. Schizorhinal, caudally acuminate nares that extend caudal to the zona flexoria craniofacialis are present in Haematopus, Jacana and Charadrius. The caudal margins of the nares are not well preserved but were rounded. The rostral terminus of the premaxilla has been pushed dorsally due to diagenesis. However, it is clear that the rostral terminus of the premaxilla is swollen dorsally and recurved as in the holotype. This condition is also present in N. grandei, Jacana, and Charadrius. In Scandiavis, the olfactory bulbs have left an impression on the internal surface of frontals, but their ventral and lateral margins are not defined by bone, and the segmen tion was thus stopped at the lower limits of the ossified margins in accordance with B anoff et al. [24]. The olfactory bulbs are positioned at the rostralmost apex of the te cephalon without clear bifurcation and demarcated from the telencephalon by a dep sion ( Figure 4A). The dorsal surface displays a shallow inter-lobe sulcus. The relative of the olfactory bulbs in birds is known to correlate to olfactory capabilities [34,72, Overall, the olfactory bulbs of Scandiavis are small compared to some extant avian tax well as fossil birds such as Halcyornis. Nonetheless, the relative proportions are similar e.g., Pluvianus and Charadrius, and well within the size range of living birds. The lac clear outlines in this area means the extent of the olfactory nerve (CNI) could not be termined. (C) shows the right quadrate in lateral aspect. Anatomical abbreviations: ec, ectethmoid; fm, foramen magnum; fr, frontal bone, jug, jugal; mnd, mandible; na, nares; oc, occipital condyle; iof, interorbital fonticulus; porb, postorbital process; prx, premaxilla; qdr, quadrate.
The frontals are largely intact and show dorsally projected supraorbital crests. This condition is also present in N. grandei, Jacana, Burhinus, and Charadrius. In dorsal aspect the interorbital area is deeply furrowed but without foramina. In lateral aspect, a welldeveloped ectethmoid is visible and appears to have been ankylosed to the lacrimal as in Jacana, Charadrius, Haematopus and Burhinus; however, no lacrimal is preserved. This condition is present in many Charadriiformes and some Gruiformes, such as Psophia and Himantornis. Parts of the interorbital septum are visible within the orbital cavity; the sclerotic rings are lost. Straight, thin jugals can be observed on both sides of the skull. The postorbital process is truncate and dorsal to the dorsal apex of the nuchal crest, like in Scandiavis. The latter condition is present in Charadrius, Haematopus, and Psophia but not in Burhinus, Jacana, Heliornis, Sarothrura, or Himantornis. The zygomatic process is truncate, aciminate at the terminus, and rostrally directed. The occipital region is well preserved and visible in the CT data ( Figure 2). The fonticuli occipitalis are not present. This is consistent with the condition in Charadrius, Burhinus, Jacana, Haematopus, Psophia, Himantornis, Sarothrura, and Heliornis. The foramen magnum is round and the occipital condyle is reniform. The occipital condyle is subequal in rostrocaudal position to the paroccipital processes.
Both quadrates are well preserved. The capitula of the quadrate are moderately spaced and separated by a notch. The medial otic capitulum is deflected caudally, as in Charadriiformes. The capitula of the quadrate are moderately spaced and separated by a notch. The capitulum squamosum is rostrocaudally flattened and the capitulum oticum is mediolaterally elongate as in all examined Charadriiformes. The otic process is slender and recurved along the dorsal margin as in N. grandei. Prominent lateral and tympanic cristae are present, as in N. grandei, Haematopus, Jacana, Himantornis, Sarothrura, Heliornis, and Psophia. An elongate orbital process that is subequal in length to the otic process is present. It appears to have a blunted, sub-rectangular terminus. This is similar to the condition in Burhinus, Jacana, and Haematopus, although their orbital processes exhibit broader width. It is most similar to the condition in Heliornis. A caudal condyle is present as in Neoaves. The caudal condyle does not appear to be confluent with the lateral condyle, unlike the condition in Nahmavis. The fovea of the cotyla quadratojugalis appears to be shallow and laterally directed as in most Charadriiformes and Gruiformes. Similarly, as in most Charadriiformes and Gruiformes, a rounded processus lateralis present along the caudodorsal rim projects over the fovea quadratojugalis.
The mandible is complete. The symphysis appears to be truncate, less than 1 /5 of the length of the mandible. This is most like the condition in Jacana, Heliornis, Sarothrura, Himantornis, and Psophia. The dorsal angle of the mandible is prominent, as in Scandiavis, and is more prominent than those of Haematopus, Jacana, Burhinus, and Charadrius and the examined Gruiformes. The rostral mandibular fenestrae appear to have been elongate and perforate, like the condition in Haematopus, Jacana, Charadrius, Burhinus, and Psophia and unlike the slit-like condition in N. grandei and examined ralloids. Rounded, perforate caudal mandibular fenestrae are also present. The caudal articular area displays neoavian morphology. A hook-like caudolateral process is present on the articular [15,18]. It appears to form a pseudo-retroarticular process like that of Scandiavis. Haematopus similarly exhibits a pseudo-retroarticular process, although it is more ventrally directed.

Endocranial Anatomy
The reconstructed endocast includes the forebrain, midbrain, hindbrain, and osseous labyrinths as well as some cranial nerves and the arrangement of the carotid artery, but excludes full reconstruction of the olfactory bulbs and pathways of the optic nerve. As such, preservation of NHMD 625345 allowed reconstruction of almost the entire endocranial cast (Figures 3 and 4). Excluding cranial nerves, carotid artery and osseous labyrinths, the total endocranial volume is 1.09 mL. However, this should be considered a minimum value since the optic nerve and the olfactory bulbs were incompletely reconstructed. Unless otherwise stated, anatomical comparisons with extant charadriiform birds refer to brains and endocasts illustrated by Stingelin [52] and Smith and Clark [56]. sion ( Figure 4A). The dorsal surface displays a shallow inter-lobe sulcus. The relative size of the olfactory bulbs in birds is known to correlate to olfactory capabilities [34,72,73]. Overall, the olfactory bulbs of Scandiavis are small compared to some extant avian taxa as well as fossil birds such as Halcyornis. Nonetheless, the relative proportions are similar to, e.g., Pluvianus and Charadrius, and well within the size range of living birds. The lack of clear outlines in this area means the extent of the olfactory nerve (CNI) could not be determined. Figure 3. Visualization of the skull of Scandiavis mikkelseni (NHMD 625345) rendered transparent to show the position of the virtual brain endocast (grey) together with the osseous labyrinth (blue), carotid artery (red), and cranial nerves (yellow). Anatomical abbreviations: asc, anterior semicircular canal; c, cerebellum; ca, carotid artery; mnd, mandible; t, telencephalon.   The opening for the optic nerve (CNII) consists of a single central relatively wide foramen. As such, it conforms to the Type 2 optic nerve exit sensu Hall et al. [74]. This opening reflects the position of the optic chiasm rather than the size of the optic nerve, and is therefore considered unreliable for estimating the size of the nerve bundle. As a result, the optic nerves were not possible to determine. This type of nerve exit is present in different orders of birds [74]. However, a large optic tract is typical for Charadriiformes [56], and most families within this clade are characterized by a Type 2 optic foramen [74].
The endocast reveals no evidence of a pineal organ, indicating that it did not leave an impression on the skull roof (in similarity with extant birds). The pituitary fossa is readily distinguishable, and the pituitary gland is slightly cone-shaped ( Figure 4B). The bony tunnels that housed the carotid artery are preserved. The two arteries merge and form a longitudinal vessel before entering the pituitary gland fossa as a single artery. As  In Scandiavis, the telencephalon is expanded dorsolaterally as in modern birds, and largely stacked on top of the mesencephalon and rhombencephalon in lateral view. The relative dimensions and positions of brain regions, including the olfactory lobes, are highly congruent with certain charadriiform taxa such as Pluvialis and Charadrius. The brain axis is relatively vertically oriented, but the position of the horizontal semicircular canal of the osseous labyrinth indicates a more ventrally tilted in vivo head posture in Scandiavis than in Charadrius. In dorsal view, the two telencephalic hemispheres are anteriorly tapered and form a spade-like structure as in extant charadriiforms ( Figure 4D). In caudal view, the telencephalon shows a similar outline to one of the relatively complete natural endocast from the early Eocene of the German Baltic coast figured by Hoch [29]. However, the optic lobes appear relatively smaller in that specimen, and the cerebellum looks more rounded and has more distinct foliation.
Scandiavis also shows the presence of eminentia sagittalis or wulst; an expansion of the dorsal surface of the telencephalic hemispheres found in extant birds. Based on its position on the dorsal telencephalon, Stingelin [52] described two different morphotypes. Type A is rostrally positioned, and is found in several of the previously described Eocene taxa (Odontopteryx, Halcyornis, Numenius, and Prophaethon; [30,31]). In contrast, Type B is more caudally positioned, but also includes intermediate forms where the eminentia sagittalis is centrally located on the telencephalic hemisphere. In Scandiavis, it is relatively weakly developed dorsally ( Figure 4B), but broad and rostrocaudally extended without contacting the olfactory bulb nor the cerebellum ( Figure 4D). As such, Scandiavis is best described as possessing an intermediate Type B development. The telencephalon has a welldefined interhemispheric fissure, stretching from a rostral position to just dorsally to the contact telencephalon-cerebellum. In extant charadriiforms, the wulst shows considerable variation [56]. In dorsal projection, the wulst of Scandiavis resembles both that of aquatic foraging taxa, including alcids, and that of terrestrially foraging taxa, such as Charadrius and Stiltia. However, the rostrocaudal development of the wulst present in Scandiavis can only be found in the latter two. Its lateral and dorsal expansion is somewhat similar to Prophaethon, but its morphology is otherwise different from that of other Paleogene taxa for which wulst development is available [28,[30][31][32]36,41,59].
In Scandiavis, the olfactory bulbs have left an impression on the internal surface of the frontals, but their ventral and lateral margins are not defined by bone, and the segmentation was thus stopped at the lower limits of the ossified margins in accordance with Balanoff et al. [24]. The olfactory bulbs are positioned at the rostralmost apex of the telencephalon without clear bifurcation and demarcated from the telencephalon by a depression ( Figure 4A). The dorsal surface displays a shallow inter-lobe sulcus. The relative size of the olfactory bulbs in birds is known to correlate to olfactory capabilities [34,72,73]. Overall, the olfactory bulbs of Scandiavis are small compared to some extant avian taxa as well as fossil birds such as Halcyornis. Nonetheless, the relative proportions are similar to, e.g., Pluvianus and Charadrius, and well within the size range of living birds. The lack of clear outlines in this area means the extent of the olfactory nerve (CNI) could not be determined.
The opening for the optic nerve (CNII) consists of a single central relatively wide foramen. As such, it conforms to the Type 2 optic nerve exit sensu Hall et al. [74]. This opening reflects the position of the optic chiasm rather than the size of the optic nerve, and is therefore considered unreliable for estimating the size of the nerve bundle. As a result, the optic nerves were not possible to determine. This type of nerve exit is present in different orders of birds [74]. However, a large optic tract is typical for Charadriiformes [56], and most families within this clade are characterized by a Type 2 optic foramen [74].
The endocast reveals no evidence of a pineal organ, indicating that it did not leave an impression on the skull roof (in similarity with extant birds). The pituitary fossa is readily distinguishable, and the pituitary gland is slightly cone-shaped ( Figure 4B). The bony tunnels that housed the carotid artery are preserved. The two arteries merge and form a longitudinal vessel before entering the pituitary gland fossa as a single artery. As such, it conforms to the type I described by Baumel and Gerchman [75]. Ventrally, they bifurcate into two narrow tunnels that curve gently and extend caudolaterally ( Figure 4E).
The optic lobes of Scandiavis are of considerable size in relation to the telencephalon, and their relative size appears to exceed that of most Charadriiformes and other Paleogene taxa such as Odontopteryx, Prophaethon, and Halcyornis. In lateral view, the optic lobes are globular and slightly lunate. The apparent contact between the optic lobe and the cerebellum is wide, similarly to Charadrius and Stiltia, whereas the caudal portion of the optic lobe is tapered in most other charadriiform taxa. The optic lobes are visible in dorsal view in Scandiavis as well as basal taxa such as Charadrius, Pluvianus, Pluvialis, Vanellus, and Burhinus. This character is notably missing from other Charadriiformes. The semicircular vein is impressed on the caudo-ventral surface of the mesencephalon and extends onto the lateral surface of the cerebellum ( Figure 4C).
In lateral view, the medulla is globe-shaped and extends along the rostral half of the optic lobes. It bears a defined ventral sulcus and displays clear depressions on the lateral surfaces where the cochleae are positioned. The trigeminal nerve (CNV) is easily distinguished ( Figure 4B). Although the three branches cannot be distinguished with certainty, it appears that at least V 1 is clearly delimited. The abducens nerve (CNVI) is small and narrow, exiting from the rostralmost end of the medulla ( Figure 4B). The facial nerve (CNVII) and the vestibulocochlear nerve (CNVIII) are difficult to distinguish in this dataset, but one or both appear to be represented by a relatively small protrusion on the lateral medulla ( Figure 4A). The glossopharyngeal (CNIX), vagus (CNX), and accessory (CNXI) nerves do not appear to differentiate before leaving the skull. In Scandiavis, the cerebellum is oval-shaped in caudal view, but tapers caudally ( Figure 4C). It bears at least three distinct folia along with an impression of an occipital sinus, which is distinguishable primarily in the caudal region. The cerebellum is heavily restricted dorsally in lateral view and shows only very limited contact with the caudal margin of the telencephalon ( Figure 4A), similar to Vanellus, Pluvialis, and Pluvianus. The cerebellum is significantly more expanded in most other charadriiforms and in many other birds, including Fulica and Balearica, the only gruiform taxa for which published brain morphology were found [33,50,55]. The impression of the semicircular vein (present on the caudo-ventral surface of the optic lobe) extends to the cerebellum, where it can be traced ventrally along the lateral surface ( Figure 4A).
The flocculus is relatively large and projects from the lateral surface of the cerebellum through the arch of the anterior semicircular canal. The base exhibits some torsion, from where it is directed caudolaterally before tapering distally. Within Charadriiformes, this elongated morphology is similar to those of wing-propelled diving Alcinae and bears similarities to Larus philadelphia [33], and is significantly different from the reduced and truncated state in the terrestrially foraging Charadrius and Stiltia. In terms of relative mediolateral length, shape of the base, degree of tapering distally, and size, it closely resembles that of the more aquatic foraging gruiform Fulica americana. The reconstruction of the flocculus of Scandiavis shows that it may have fenestration distally, but the CT data was difficult to interpret in this region and the state of this character therefore remains uncertain.
In the endocast of the osseous labyrinth of Scandiavis, all semicircular canals as well as the cochlear duct are intact and preserved in detail on both sides of the skull. The structure is positioned largely ventral to the optic lobe ( Figure 3). All canals are relatively long and narrow with well-defined ampullae ( Figure 5). As in living birds, the anterior semicircular canal is more expanded than the others, and in addition distinctly medially angled. The cochlear duct is relatively long and somewhat arched. The distal tip is swollen as in Stiltia and Charadrius rather than tapered as in all other charadriiforms. Overall, the labyrinth endocast appears nearly identical to the two aforementioned taxa, but with a less pronounced swelling of the cochlear tip than in Charadrius. In comparison with previously described labyrinths of Eocene taxa (Odontopteryx, Prophaethon and Halcyornis), the semicircular canals are both relatively long and thin. The overall morphology of the inner ear is also significantly different from those observed in fossil stem penguins [32,36,59]. Diversity 2021, 13, x 11 of 56
Analysis applying a backbone constraint representing major subclade relationships of Kimball et al. [66] Table 1. Pellornis mikkelseni Bertelli et al. [14] was recovered as a messelornithid across all analyses, consistent with prior placements [14,15]. Messelornithidae (P. mikkelseni + Messelornis cristata Hesse [9876]) was placed as the sister taxon of Ralloidea under the Kimball et al. [66] and Reddy et al. [65] constraints. These results are consistent with those of Musser and Clarke [18] with the exception of the Prum et al. [64] constraints causing collapse of Messelornithidae within a polytomy containing extant Ralloidea and Songzia acutunguis.
Analysis applying a backbone constraint representing major subclade relationships of Kimball et al. [66] Table A1. Pellornis mikkelseni Bertelli et al. [14] was recovered as a messelornithid across all analyses, consistent with prior placements [14,15]. Messelornithidae (P. mikkelseni + Messelornis cristata Hesse [76]) was placed as the sister taxon of Ralloidea under the Kimball et al. [66] and Reddy et al. [65] constraints. These results are consistent with those of Musser and Clarke [18] with the exception of the Prum et al. [64] constraints causing collapse of Messelornithidae within a polytomy containing extant Ralloidea and Songzia acutunguis.
Analyses using a backbone constraint based on the major clade relationships of Prum et al. [64] [66] constrained analyses. Within the results constrained using Reddy et al. [65], N. grandei is placed as the sister taxon of a S. mikkelseni + Charadriiformes as a part of Pan-Charadriiformes. Analyses employing the Prum et al. [64] constraint resulted in an N. grandei + S. mikkelseni as sister taxon of Pluvianus aegyptius. This group is then collapsed into a polytomy of Charadriiformes. Bootstrap support for placement of extinct taxa was less than 50% across all analyses, with the exception of Messelornithidae earning a 94% or higher bootstrap score in each result.
Placement of Salmila robusta Mayr [46] and both Eocene humeri remain unchanged from the results of Musser and Clarke [18] with the exception of their placement in the tree resulting from analyses employing the Reddy et al. [65] constraint. In Musser and Clarke [18], Salmila is the sister-taxon of Eurypyga helias + Rhynochetos jubatus, whereas it is placed the sister-taxon of Opisthocomus hoazin + Leptosomus discolor in our results. Our results place IGM 100/1435 as the sister-taxon of Chionis alba, whereas it is collapsed into a polytomy of Charadriiformes in Musser and Clarke [18].   Bootstrap support values greater than 50% are denoted above branches. taxon of Pluvianus aegyptius. This group is then collapsed into a polytomy of Charadriiformes. Bootstrap support for placement of extinct taxa was less than 50% across all analyses, with the exception of Messelornithidae earning a 94% or higher bootstrap score in each result. Placement of Salmila robusta Mayr [46] and both Eocene humeri remain unchanged from the results of Musser and Clarke [18] with the exception of their placement in the tree resulting from analyses employing the Reddy et al. [65] constraint. In Musser and Clarke [18], Salmila is the sister-taxon of Eurypyga helias + Rhynochetos jubatus, whereas it is placed the sister-taxon of Opisthocomus hoazin + Leptosomus discolor in our results. Our results place IGM 100/1435 as the sister-taxon of Chionis alba, whereas it is collapsed into a polytomy of Charadriiformes in Musser and Clarke [18].
Nine unambiguous and five ambiguous optimized synapomorphies with CI < 1.0 support placement of S. mikkelseni with Charadriiformes using the Kimball et al. [66] constraint ( Figure 6). The rostral apex of the premaxilla is not dorsally inflated (character 4:state 1). This is similar to the condition in most included Charadriiformes; it is inflated in most included Gruiformes (4:2). The apex of the postorbital process is oriented rostroventrally as in most Charadriiformes (55:1), whereas it is angled ventrally in most Gruiformes (55:2). The supraorbital process of the lacrimal is tapered toward the caudal apex as in most included Charadriiformes (68:1). It is broad in most included Gruiformes (68:2). The ectethmoid is ankylosed to the lacrimal (86:1), a feature that is more common in Charadriiformes than in Gruiformes (typically unankylosed, 86:2). The fonticulus orbitocranialis is caudally extensive and continues along the coronal plane of the squamosal region (91:2). This feature is common in both Gruiformes and Charadriiformes. The lateral condyle of the quadrate terminates well ventral to the caudal condyle as in most Charadriiformes (206:2). Gruiformes typically exhibit a lateral condyle that is only slightly ventrally positioned with respect to the caudal condyle (206:1). The cervical vertebrae are not Nine unambiguous and five ambiguous optimized synapomorphies with CI < 1.0 support placement of S. mikkelseni with Charadriiformes using the Kimball et al. [66] constraint ( Figure 6). The rostral apex of the premaxilla is not dorsally inflated (character 4:state 1). This is similar to the condition in most included Charadriiformes; it is inflated in most included Gruiformes (4:2). The apex of the postorbital process is oriented rostroventrally as in most Charadriiformes (55:1), whereas it is angled ventrally in most Gruiformes (55:2). The supraorbital process of the lacrimal is tapered toward the caudal apex as in most included Charadriiformes (68:1). It is broad in most included Gruiformes (68:2). The ectethmoid is ankylosed to the lacrimal (86:1), a feature that is more common in Charadriiformes than in Gruiformes (typically unankylosed, 86:2). The fonticulus orbitocranialis is caudally extensive and continues along the coronal plane of the squamosal region (91:2). This feature is common in both Gruiformes and Charadriiformes. The lateral condyle of the quadrate terminates well ventral to the caudal condyle as in most Charadriiformes (206:2). Gruiformes typically exhibit a lateral condyle that is only slightly ventrally positioned with respect to the caudal condyle (206:1). The cervical vertebrae are not heterogeneously elongate, as in most Charadriiformes (253:2); most Gruiformes present heterogeneous vertebrae that are relatively elongated (253:1). The torus dorsalis of section II of the cervical vertebrae is concave as in most included Charadriiformes (261:2), whereas it is convex in most included Gruiformes (261:1). The caudalmost presacral vertebrae exhibit deep lateral excavations which are also present in all included Charadriiformes (279:2). These excavations are not present in included Gruiformes (279:1). The spina interna rostri of the sternum is present as in most included Charadriiformes (299:1); it is lost in all included Gruiformes (299:2). The iliac blades and synsacrum of the pelvis are unfused as in almost all included Charadriiformes (498:2), whereas those of included Gruiformes are fused (498:1). Pedal digit I: phalanx 1 is about half the length of III:1 as in most included Charadriiformes (691:2). Most included Gruiformes present a I:1 that is about half the length of III:1 (691:1). Three common crus of the semicircular canals are present as in Charadrius, Larus, Haematopus, and Stercorarius (719:1). Grus conversely only exhibits two common crus (719:0). The cochlear curvature of the endosseous labyrinth in lateral aspect is curved as in Charadrius, Larus, and Stercorarius (720:0). It is straight in Grus (720:1). As in Musser and Clarke [18], placement of Turnix nigricollis as the sister-taxon of N. grandei is not consistent with recent phylogenies, which recover Turnix as the sister-taxon of a group containing Glareola, Larus, and Dromas [64][65][66]. Incongruous placement of this taxon in our results is likely due to a lack of constraining taxon relationships within major subclades and a need for more taxon and character sampling. Turnix is also known to be problematic in morphological analyses; Livezey [77] similarly recovered Turnix as the sister-group to all other Charadriiformes, and Mayr [78] recovered Turnicidae as either within a polytomy in Charadriiformes or as the sister-taxon of a clade containing Jacanidae, Scolopacidae, Rostratulidae, Thinocoridae, and Pedionomidae. Synapomorphies for additional constraint analyses can be found in the Appendix A.

Discussion
Recovery of additional material and new phylogenetic analysis of Scandiavis mikkelseni more robustly establishes both S. mikkelseni and Nahmavis grandei as charadriiform, and most consistently places them as stem-charadriiform taxa. This is critical to reconstructing phylogenetic relationships of Paleogene fossils, as previous studies produced conflicting results as to whether either taxon was charadriiform or gruiform [17,18]. While it remains unclear whether S. mikkelseni or N. grandei is more basal, both taxa come from the early Eocene and may represent important divergence date calibration fossils for Pan-Charadriiformes.
The relative dimension of a brain region in vertebrates is proportional to the relative importance of that particular region in a species [79,80], allowing inferences to be made about the neural development and sensory capabilities of extinct taxa. The endocast of Scandiavis thus provides valuable new insights into avian neural architecture in the early Paleogene. It shares general features with modern birds, such as laterally expanded mesencephalic and telencephalic lobes, and a more vertically flexed brain axis as in many extant birds [81]. Comparisons with extant and Paleogene taxa reveal that this endocast shows numerous similarities with extant basal crown charadriiforms, and clearly possesses the apomorphies that have been proposed to characterize the endocranium of charadriiform birds. This includes more dorsally positioned olfactory bulbs, an anteriorly tapered telencephalon, relatively wider cerebella, and a larger optic tract [56].
The relative dimensions of the brain regions of Scandiavis and their positions relative to each other, including the olfactory lobes, are congruent with taxa such as Pluvialis, Vanellus and Charadrius. In particular, there is a striking resemblance between the endocasts of Scandiavis and Charadrius vociferus, which share all identical character states relating to both the brain and the osseous labyrinth with the exception of the floccular region (characters [711][712][713]. The flocculus in Scandiavis does not exhibit the reduced state seen in Charadrius. Instead, it is relatively large and elongated, resembling that of Fulica. The flocculus has been proposed to play a pivotal role in gaze stabilization and swift head movements, facilitating acrobatic maneuverability during flight [19,82]. However, floccular volumes have also been found to be unreliable for inferences on flight ability and ecology in extant taxa [83,84]. In contrast to Fulica, Scandiavis may have fenestration distally (character 712), but this was scored as a missing character due to difficulties interpreting the CT data in this region. This may indicate that a large floculus may be ancestral for Charadriiformes + Gruiformes if this sister-relationship is correct. Intriguingly, finfoots are some of the most basal gruiforms and have a similar ecology to that of Fulica. The endocast of Scandiavis additionally exhibits one endocranial character state that supports an 'aquatic charadriiform clade' including the Pan-Alcidae, Stercorariidae, Sternidae, Laridae, and Rynchopidae in Smith and Clarke [56]: a telencephalon with < 50% of its caudal margin in contact with the cerebellum. However, this character state is also present in several basal charadriifom taxa including Charadrius, Burhinus, Pluvianus and Vanellus.
The eminentia sagittalis is a dorsal telencephalic expansion unique to the brains of birds. It is commonly linked to visual field processing and tends to be well-developed in species that rely heavily on vision [85,86], but is also generally involved in somatosensory perception [85]. The eminentia sagittalis is found in all extant birds, but morphology and development greatly vary. This structure has been observed in several Paleogene birds (e.g., [30,31,59,72]) and recently also in a Cretaceous specimen [87]. Although Scandiavis possessed a relatively weakly developed eminentia sagittalis, it is within the range of development observed among extant birds. The dorsal expansion appears at the lower end of the scale, but still mirrors that of Charadrius vociferus and closely related taxa. The overall morphology is different from that of other Paleogene taxa and thus adds to a growing range of morphologies of the eminentia sagittalis within Aves by the early Eocene, suggesting that the diversification of this structure may have been initiated much earlier [31]. In addition, Scandiavis possessed large optic lobes. This region receives and processes much of the retinal input [88], and may thus suggest a need for optical input.
The morphology of the osseous labyrinth has been hypothesized to reflect flight style in modern birds, and is therefore commonly targeted in paleontological studies for generating ecological interpretations. Relatively long and thin semicircular canals often occur in taxa that engage in 'acrobatic flight' (e.g., Larus and Columba) whereas shorter and broader canals (e.g., Anas and Gallus) are normally associated with straightline flight [30,81,89,90]. Additionally, the ampullae, cross-sectional shape, and angles between the semicircular canals have been used to infer flying ability [30,81,89,90]. The relatively long and thin semicircular canals of Scandiavis as well as the close resemblance to terrestrially foraging charariiform birds may thus suggest locomotory abilities and maneuverability similar to these taxa. The ampullae are also relatively well-developed and the anterior semicircular canal is distinctly inclined medially, all conditions seen in birds regarded as good fliers [81,89]. However, potential relationships between labyrinth morphology and locomotory modes in birds were recently questioned. Benson et al. [53] suggested that avian labyrinth shape is determined more by spatial constraints in the braincase than flying style. This in turn was questioned for not adequately capturing the relevant patterns of locomotion [91], and at present relationships remain unclear [92].
Recovery of S. mikkelseni and N. grandei as stem-Charadriiformes presents a similar biogeographic pattern as that of Paleogene Ralloidea (Gruiformes: rails, finfoots and flufftails) [14,15,18,93], with Eocene representatives being located in Europe and North America. Messelornithidae are currently the only robustly placed stem gruiform taxa and comprise Pellornis mikkelseni of the earliest Eocene Fur Formation in Denmark, the Messelornis nearctica [94] in the Eocene Green River Formation of North America [94][95][96][97], Messelornis cristata [76] of the Eocene Messel Formation of Germany, Messelornis russelli [98] of the Paleocene of France, and Itardiornis hessae [98] of the late Eocene to early Oligocene Quercy fissure fillings of France. Such fossil evidence suggests two possible biogeographic hypotheses if Aves originated in West Gondwana: the North American Gateway hypothesis [99] or a Laurasian Gateway hypothesis. If Gruiformes and Charadriiformes originated in West Gondwana and were present in South America by the end of the middle Paleocene (~59.2 Ma), these clades could have dispersed through and diversified in North America coincident with the first appearances of Laurasian metatherian and placental mammals in South America [100,101] during the Paleocene to early middle Eocene (the North American Gateway hypothesis [99]). These avifaunas would then have spread to Europe by the end of the Paleocene and earliest Eocene~56 to 53 Ma, reaching Europe via a North Atlantic corridor with land connections along northeastern North America, Greenland and Europe [102,103]. Sea-floor spreading between northeastern Greenland and Europe would have created an ocean barrier around 53 to 52 Ma [102,104]. Conversely, Gruiformes and Charadriiformes could have dispersed through North America and crossed the Bering land bridge, which would have been emergent during much of the Tertiary [105,106]. Eocene-Oligocene fossils of stem-Gruoidea (Gruiformes: trumpeters, cranes and limpkins; [107]) in North America and Asia and a putative ralloid (Songzia [48,108]) in China may support this hypothesis for Gruiformes. Charadriiform fossils from Asia comprise the pan-charadriiform humeri from the earliest Eocene of Mongolia [69] and Jiliniornis huadianensis from the middle Eocene of China [109][110][111]. At the same time, presence of these fossils in Asia could still be consistent with the North American Gateway hypothesis with these avifaunas expanding first into Europe and then Asia [99], especially as both charadriiform fossils from Asia were recovered within crown Charadriiformes [18,109] and a putative specimen of Songzia has been recovered from the Paleocene of France [112]. If these avifaunas did not originate in Gondwana, fossil evidence may be indicative of stem clades originating in Eurasia and rapidly expanding into North America by the end of the early Eocene [18]. More fossil evidence from Gondwana and Asia is ultimately necessary to more robustly support one of these hypotheses or a non-Gondwanan based hypothesis.
In conclusion, Scandiavis mikkelseni is an important Paleogene fossil for understanding evolutionary relationships, identifying divergence dates and biogeography, and better understanding the evolution of charadriiform and possibly gruiform brain structure. The brain of Scandiavis is largely similar to those of basal extant charadriiforms, and it probably also possessed similar sensory capabilities. This endocast demonstrates that by the early Eocene, avian brain development was close to that of extant birds. Confirmation of Scandiavis and Nahmavis grandei as charadriiform presents a biogeographic pattern similar to that within the fossil record of Gruiformes [14,15,18,93]. This evidence is consistent with several biogeographic hypotheses, and further evidence supporting the sister-relationship of these two important clades is thus necessary to better understand their biogeographic histories. Acknowledgments: We thank Bent Lindow (NHMD) and Bo Schultz (FUM) for access to NHMD 625345, invaluable discussion and feedback, and for providing stratigraphic information regarding the specimen. We would also like to thank Randolph De La Garza for assisting with photographing the fossil, and Johan Lindgren and Christopher Torres for constructive comments and discussions. We additionally thank the editor and anonymous reviewers for thoughtful feedback.

Conflicts of Interest:
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Appendix A
Character Descriptions State (0) is reserved for absence. Citations are provided where significant overlap with a previously created character has occurred and/or where the character has been Diversity 2021, 13, 651 18 of 60 modified based on assessment of previously created characters. Citations of previously created characters are not meant to represent a comprehensive list of character overlap, and largely comprise the characters (and citations therein) of Musser and Cracraft [44], due to our building on this dataset. Characters are anglicized as much as possible.
8. Rostrum, ventral aspect, tomial crest, caudomedial extent and ventral enclosure of rostrum and consequent shape of ventromedial fenestra: almost completely fused along midline to border of antorbital angle so that rostrum essentially closed by tomial crest ventrally, slit-like fenestra along midline may be visible in ventral aspect (1); fused along about 1/2 of midline or less near rostral apex so that pair of cylindrical cavities at apex visible but separated toward antorbital fenestra, creating "leaf-shaped" fenestra that is wider near antorbital fenestra (2); ventral enclosure of rostrum lost completely (3); like state 2 but over 1/2 of rostral portion is enclosed, pair of concavities concealed by tomial crest (4); completely closed ventrally by inflated bone (5).
50. Orbital margin, supraorbital crest (if present): full, projecting dorsolaterally with a sharp, crista-like margin, creates a sharp and prominent supraorbital crest that extends from the postorbital process to the lacrimal (1); partial, like state only present along caudal half of orbit (2); crista-like margin prominent and present but projected ventrally/laterally, does not extend dorsal to frontal bone as in states 2 and 3 (3  (1); lost (2). Not comparable in absence of pterygoid process. Musser and Cracraft [44], character 57.
238. Area of articulation with quadrate, dorsally projecting hook-like accessory projection present caudolateral to cotyla caudalis: absent (0); present, dorsoventrally elongate and hook-like (1); present but truncate and rounded, sometimes present as a pedestal-like projection with a flattened dorsal surface (2). Noncomparable where articulation for caudal condyle of quadrate is absent in mandible. Note: This character is not to be confused with a true retroarticular process seen in e.g., galloanserines and Phoenicopteridae. The difference is that while the hook-like process seen in Neoaves (e.g., Gruiformes) is only present caudolateral to the caudal condyle (between area of articulation of caudal and lateral condyles of quadrate), the true retroarticular process is medial to or within the margin of the caudal cotyla or medial to the cotyla lateralis where the articulation for the caudal condyle is absent. The true retroarticular process must incorporate medial lamina of the mandible. Some Sphenisciformes appear to have a true retroarticular process, but that is actually the area of the caudal fossa of the mandible that has been elevated and flattened dorsally. Musser and Cracraft [44], character 129; Bertelli et al. [14], character 26. 239. Dorsally projecting hook-like accessory projection present caudolateral to cotyla caudalis (if present), extreme caudal elongation of all dorsoventral lamina: absent or only slight caudal projection present (0); present, creates mediolaterally flattened "pseudo processus retroarticularius" (1). See note above. Noncomparable where absent.

Group Name
Kimball et al. [