On the comparative morphology of the juvenile avian skull: An assessment of squamosal shape across avian higher‐level taxa

The comparative morphology of juvenile avian skulls is poorly known. Here, we survey the shape of the squamosal (os squamosum) across juvenile skulls of avian higher‐level clades. In all palaeognathous birds, the rostral end of the squamosal does not surpass the parietal and does not reach the frontal. This morphology is likely to be plesiomorphic for neornithine birds. A short squamosal also occurs in some Neognathae, but in most neognathous birds the squamosal contacts the frontal, and in some taxa the bone is strongly elongated and distinctly surpasses the parietal. Some clades show a notable variation in squamosal morphology. This is, for example, true for Strigiformes, where the taxon Athene differs from the other examined owls in squamosal size, and for the Passeriformes, where Old World Suboscines are characterized by a distinctive squamosal morphology. A unique derived squamosal morphology is for the first time reported for the Apodidae and Hemiprocnidae, in which the bone forms a elongated rostral process that runs along most of the orbital rim. In non‐avian theropods, the squamosal articulates with the postorbital and delimits the upper temporal opening. Extant birds lack a postorbital, but a topological correlation between the squamosal and the postorbital process is maintained in most taxa of the Neognathae. The phylogenetic significance of squamosal morphology is diminished by the fact that closely related taxa often show very disparate shapes of the bone, and squamosal morphology appears to be determined by multiple functional constraints including skull geometry, brain morphology and, possibly, nostril type.

tightly co-ossified, so that their sizes and shapes can only be evaluated through the study of juvenile specimens. However, even though avian osteology has been intensely studied during the past two centuries, the comparative morphology of juvenile bird skulls still is insufficiently known.
Most of these analyses focus on a particular taxon and few comparative data exist on the variation across neornithine higher level clades. Although earlier authors emphasized the need for more detailed studies across a broader taxonomic range (e.g., Marugán-Lobón & Buscalioni, 2009;Zusi, 1993), such have not yet been performed, with comparative descriptions of particular skull bones being particularly scarce (Mayr, 2020;Zusi & Livezey, 2006).
The dorsal and lateral portions of the cranial vault of birds are mainly formed by four paired bones, the frontal (os frontale), parietal (os parietale), laterosphenoid (os laterosphenoidale; "orbitosphenoid" of Jollie, 1957, "pleurosphenoid" of Müller, 1963and Zusi, 1993, but see Clark, Welman, Gauthier, & Parrish, 1993), and the squamosal (os squamosum). The squamosal is the only bone of the cranial vault which derives from the caudal (posterior) portion of the mandibular-stream cranial neural crest, whereas the frontal, parietal, and laterosphenoid originate from the mesoderm and-in the case of the rostral portion of the frontal-the rostral (anterior) portion of the mandibular-stream cranial neural crest (Maddin, Piekarski, Sefton, & Hanken, 2016). The squamosal is situated in the laterocaudal portion of the neurocranium (Figure 1a). Caudally, it articulates with the exoccipital and supraoccipital, medially it is bounded by the parietal, and laterally it abuts the laterosphenoid; the relations to the frontal are variable across different neornithine taxa. The ventral portion of the squamosal bears the zygomatic process and a more caudally situated processus suprameaticus as well as the articulation facet for the capitulum squamosum of the quadrate.
Even though it was noted previously that the squamosal differs in its relation to the surrounding bones (Zusi, 1993), this variation remained unstudied. In the The squamosal is shaded gray. Abbreviations: exo, exoccipital; frt, frontal; lts, laterosphenoid; squ, squamosal; par, parietal; ppo, processus postorbitalis; qdr, quadrate. Scale bars equal 5 mm present study, we provide the first account of squamosal morphology across a broad number of avian higher-level taxa. We comment on the stem species pattern for neornithine and neognathous birds and discuss the phylogenetic and functional implications of the observed variation in squamosal shape.  (SMF 1986, SMF 19029, SMF 20206), Dryocopus martius (SMF 5550).

| MATERIAL AND METHODS
Juvenile skulls of Monias benschi (Mesitornithiformes; SMF 3744) and Rhynochetos jubatus (Eurypygiformes, Rhynochetidae; SMF 3771) were available as alcohol-preserved specimens. Data on the juvenile skull of other higher-level data were obtained from the literature.
The exact individual age of all specimens is unknown, but most are from hatchlings of a comparable, relatively late developmental stage, in which the skull bones are well ossified but are still separated by sutures. Skulls of differently aged juveniles were available for the Spheniscidae (Spheniscus humboldti), Phoenicopteridae, Laridae (Larus argentatus), and Gruidae (Balearica, Bugeranus).
Developmental data suggest that the avian parietal is actually homologous to the postparietal of other tetrapods and that the frontal is a frontoparietal (Maddin et al., 2016). However, to avoid unnecessary confusion in relation to earlier studies, the traditional terminology is maintained in the present article.

| RESULTS
As indicated by its Latin name, the squamosal usually has a squamiform (scale-like) shape. Even though the morphology of the bone appears to be consistent within neornithine "families", it shows considerable variation between superordinate higher-level taxa in shape, size, and orientation.
Crown group birds (Neornithes) are divided into the sister taxa Palaeognathae and Neognathae, with the latter being further split into Galloanseres and Neoaves (all other neognathous birds). In juveniles of the palaeognathous Apterygidae ( Figure 2a) and Casuariidae (Figure 2b), the squamosal has a subrectangular outline and bears a long processus zygomaticus in its rostroventral portion. A similar morphology occurs in other juvenile palaeognathous birds (Figure 3a), in which the zygomatic process is likewise the only well-developed projection of the squamosal. On the medial surface of the bone there is a fossa for a diverticulum of the middle ear air sac (Jollie, 1957).
In juveniles of neognathous birds, by contrast, squamosal shape is much more variable and the bone forms distinct processes. Often there is a long rostrodorsally directed process, which articulates with the frontal (Figure 2), but several taxa also exhibit a shorter caudomedial process, which separates the ventrolateral portion of the parietal from the exoccipital and occurs in, for example, the Charadriiformes and Passeriformes (Figure 2o-t). The development of the processus zygomaticus is highly variable in neognathous birds and this is also true for the processus suprameaticus; in some taxa, such as the Charadriiformes, Gruidae, Rallidae, and Strigiformes, processus zygomaticus and processus suprameaticus are of subequal length and form a Ushaped notch (Figure 2o,p). The fossa for the diverticulum of the middle ear air sac, on the medial surface of the bone, is well delimited in some neognathous taxa (e.g., the procellariiform Diomedeidae; Figure 2j), but less developed or absent in others, such as the Passeriformes ( Figure 2t).
The squamosal of neornithine birds not only differs in shape and size, but also in its relation to the surrounding bones. The different morphologies may, at least provisionally, be assigned to three categories, which are summarized in Table 1.
In the Palaeognathae, including the extinct Dinornithidae (Worthy & Scofield, 2012, figure 4) and Aepyornithidae (Balanoff & Rowe, 2007), the squamosal does not contact the frontal, and the laterosphenoid is therefore not separated from the parietal. This lack of contact between squamosal and frontal also occurs in some neognathous birds, and among the studied taxa it was found in the Caprimulgidae ( In all other studied Neognathae, by contrast, the squamosal separates the laterosphenoid from the parietal and contacts the frontal, even though the geometry of the involved bones shows much variation. In various only distantly related taxa, the squamosal does not extend beyond the boundary between parietal and frontal. In some taxa, the bone has a broad rostral end, so that the suture between squamosal and frontal (sutura frontosquamosalis) forms a line with that between frontal and parietal (sutura frontoparietalis). This morphology occurs in the Balaenicipitidae (Figure 4g Pycraft, 1903a;Posso & Donatelli, 2005), Falconidae (Figure 3l), Meropidae (Brusaferro & Simonetta, 1998), Alcedinidae, and Piciformes.
Squamosal morphology of juvenile Galloanseres was surveyed by Zusi and Livezey (2000). In these birds, the bone has a subrectangular shape, broadly contacts the frontal, and extends well beyond the suture between parietal and frontal (Figure 3b In several only distantly related neoavian taxa, the squamosal not only surpasses the parietal, but also extends rostrodorsally well beyond the postorbital process. Such a very long squamosal occurs in the procellariiform Hydrobatinae (Oceanodroma; Figure 4c), most Charadriiformes apart from the Burhinidae (Figure 5c), as well as in the Rallidae (Figure 5m), Columbidae (Figures 1g and 3h), Pteroclidae (Figure 3i), and Passeriformes (Figure 1a-c). Squamosal shape shows, however, much variation in these birds. The bone is particularly long in the charadriiform Laridae, Sternidae, and Alcidae, in which it projects beyond the cranial vault and forms a small projection in the adult skull, which caudolaterally delimits the fossae glandulae nasales (Figure 5d-j). This distinctive morphology also occurs in the procellariiform Hydrobatinae (Oceanodroma; Figure 4c). The rostrodorsal process of the squamosal is broad in oscine Passeriformes (Figure 1a,b), whereas it is narrow and of subfalcate shape in the suboscine Pittidae (Figure 1c). A narrow squamosal was also reported for Eurylaimus ochromalus (Eurylaimidae) by Pycraft (1905, pl. 2), but is not present in the only juvenile specimen of a New World suboscine available to us (cf. Lessonia rufa; Tyrannidae) and therefore appears to be a characteristic of Old World Suboscines. In both, Charadriiformes and Passeriformes, the long squamosal broadly overlaps with the frontal.
A unique and previously unreported squamosal morphology is finally found in the Hemiprocnidae and Apodidae, in which the bones is extremely elongated and forms a rostral process that runs along most of the orbital rim ( Figure 3e). In the adult skull of both Apodidae and Hemiprocnidae the long squamosal is reflected by a ridge-like rim along the orbit, which is medially bordered by a narrow sulcus for musculus cucullaris capitis (Zusi, 2013, figure 4). This rim and sulcus are also present in the Trochilidae, which suggests that hummingbirds, of which no juvenile skulls were available for study, have an equally elongated squamosal. By contrast, the squamosal is short in the Caprimulgidae (Figure 3d). The morphology in other representatives of the Strisores (the clade including Apodiformes and the paraphyletic "caprimulgiform" birds; Mayr, 2010) is unknown.

| Stem species pattern
In non-avian theropods, the squamosal articulates with the postorbital and delimits the ventral margin of the upper temporal opening; unlike in neornithine birds the bone is therefore not fully incorporated into the cranial vault (e.g., Elzanowski & Wellnhofer, 1996;Tsuihiji et al., 2014;Wang & Hu, 2017). The squamosal of Archaeopteryx still resembles that of non-avian theropods and exhibits a postorbital and a quadratojugal process (Elzanowski & Wellnhofer, 1996); the latter was considered homologous to the zygomatic process of neornithine birds (Elzanowski, 2001). Squamosal shape is poorly known in other Mesozoic birds and the bone is often left away in skull reconstructions (e.g., Wang & Hu, 2017 Chiappe, 2011;Sanz et al., 1997). Even in the Late Cretaceous ornithurine Ichthyornis, in which the postorbital is reduced, the squamosal seems to have not yet been fully integrated into the neurocranium (Field et al., 2018). The reconstruction of Field et al. (2018) furthermore indicates that the squamosal of Ichthyornis did not contact the frontal, so that the laterosphenoid abutted the parietal. Outgroup comparisons with Ichthyornis suggest that the short squamosal of palaeognathous birds is plesiomorphic for Neornithes as a whole. It is, however, not quite clear F I G U R E 5 Legend on next page. how the shape of the squamosal of Ichthyornis was determined by Field et al. (2018), because the cranial bones of this Late Cretaceous bird are co-ossified and sutures delimiting the squamosal cannot be discerned. If the squamosal morphology of Ichthyornis was actually deduced from comparisons with extant palaeognathous birds, any conclusions on the plesiomorphic morphology of neornithine birds based on Ichthyornis as an outgroup taxon would become circular.
For Neognathae, it is more parsimonious to assume that a squamosal contacting the frontal represents the plesiomorphic condition. A contact between squamosal F I G U R E 5 Juvenile neurocrania (left lateral view) of the Charadriiformes (a-j), Phoenicopteriformes (k), Podicipediformes (l), and Gruiformes (m-p). Most skulls are digitally brightened and the squamosal is highlighted in a darker tone; sutures delimiting the parietal are enhanced by fine dotted lines. (a) Himantopus himantopus (Recurvirostridae; SMF 14796); in this specimen the squamosal is already partly co-ossified with the laterosphenoid and parietal so that its exact shape could not be determined. (h) Dendrocopos major (Picidae; SMF 19029); right side, mirrored; hyoid digitally removed. Abbreviations: frt, frontal; lts, laterosphenoid; squ, squamosal; par, parietal; ppo, processus postorbitalis; qdr, quadrate. Scale bars equal 10 mm and frontal is absent in various taxa of the Aequornithes, the clade including most aquatic and semiaquatic birds, as well as in the Phaethontidae, Rhynochetidae, Caprimulgidae, Musophagidae, and Strigiformes (Table 1). In Aequornithes, there exists variation in the size of the squamosal, which is rostrodorsally shorter than the parietal in the Spheniscidae, Phalacrocoracidae, Sulidae, Threskiornithidae, and Ardeidae, as long as the parietal in the Ciconiidae and Balaenicipitidae (Böhm, 1930), and longer than the parietal in the Pelecanidae. Based on this distribution and current phylogenies (Hackett et al., 2008;Jarvis et al., 2014;Prum et al., 2015; Figure 7), it is most parsimonious to assume that a short squamosal is plesiomorphic for Aequornithes. The taxa of the Aequornithes form a clade together with Phaethontidae and Eurypygiformes (Rhynochetidae + Eurypygidae) in current molecular phylogenies (Figure 7; Jarvis et al., 2014;Prum et al., 2015) and a short squamosal is likely to have already been present in the stem species of this clade. The short squamosal of the Caprimulgidae may also be plesiomorphic for Strisores, the clade including "caprimulgiform" and apodiform birds, because the Caprimulgidae result as the sister taxon of other taxa of the Strisores in current molecular phylogenies (Prum et al., 2015).
The early divergences within Neoaves are controversially resolved in different molecular analyses (Ericson et al., 2006;Hackett et al., 2008;Jarvis et al., 2014;Prum et al., 2015), but in all phylogenies most taxa with a short squamosal are among the early diverging clades, and only the Strigiformes are nested within Telluraves, the clade including most arboreal landbirds (Figure 7). However, even though a short squamosal, which does not contact the frontal, is mainly found in phylogenetically more "basal" neognathous taxa, it is more parsimonious in all current phylogenies to assume that it represents a derived condition for Neognathae and evolved convergently in the clades showing this feature. In the phylogeny of Prum et al. (2015), for example, a short squamosal may have developed only four times independently in neognathous birds (in the Strisores, Musophagiformes, the clade formed by Eurypygiformes, Phaethontiformes, Aequornithes, and in Strigiformes; Figure 7). By contrast, at least nine independent origins are required for a squamosal reaching the frontal (in Galloanseres, Apodiformes, Otidiformes + Cuculiformes, Musophagiformes + Mesitornithiformes + Pterocliformes, Gruiformes, Phoenicopteriformes + Podicipediformes + Charadriiformes, Ciconiidae, Pelecaniformes, and Opisthocomiformes + Telluraves). From a parsimony point of view, it is therefore more likely that a squamosal, which F I G U R E 7 Phylogenetic interrelationships of neornithine birds as recovered in the molecular analysis of Prum et al. (2015). Taxa shown in blue have a short squamosal, which does not reach the frontal, whereas the bone contacts or exceeds the frontal in taxa highlighted in red. Squamosal morphology of taxa in light gray is unknown reaches the frontal, represents the plesiomorphic condition for neognathous birds.
Galloanseres, which are the sister taxon of all other neognathous birds, exhibit a long, rectangular-shaped squamosal, which surpasses the parietal, and this morphology is also found in multiple neoavian clades. However, these very long squamosals show a great diversity of shapes, so that a squamosal, which distinctly surpasses the parietal, is likely to have evolved independently in Galloanseres and various neoavian clades.
Although the postorbital bone is completely reduced in adult neornithine birds, in some juveniles a small ossification on the tip of the postorbital process of the laterosphenoid was considered to be a remnant of the bone (Bittner, 1912); this ossicle was, however, identified as a "secondary membral ossification" by Jollie (1957, p. 411). Irrespective of whether an embryological anlage of a postorbital is present in some Neornithes, a topological correlation between the squamosal and the postorbital process of the laterosphenoid seems to be maintained in most Neognathae, in which the rostral tip of the squamosal reaches to the postorbital process. The fact that the squamosal projects beyond the cranial vault and is therefore not fully integrated into the neurocranium in the charadriiform Lari (Laridae, Alcidae, and allies; Figure 5d-j) and the procellariiform Hydrobatinae (Oceanodroma; Figure 4c) may indicate an atavism, which reflects the evolutionary history of the bone. If the postorbital process is topologically equivalent to the postorbital bone, squamosal development may therefore still be influenced by ontogenetic factors correlated with the formation of the postorbital.

| Phylogenetic and functional considerations
The phylogenetic significance of squamosal shape is diminished by the fact that closely related taxa often show very disparate morphologies, which is the case for, for example, Sphenisciformes ( Figure 4a) and Procellariiformes (Figure 4b,c), Balaenicipitidae ( Figure 4g) and Pelecanidae (Figure 4h), or the strigiform taxon Athene (Figure 6c) and other Strigiformes (Figure 6a,b).
Moreover, there appears to be a great amount of homoplasy in the distribution of squamosal types among neornithine birds. However, even though it is difficult to identify derived squamosal morphologies characterizing major neoavian clades, a number of taxa exhibit distinctive squamosal shapes of potential phylogenetic and functional interest.
A unique apomorphy of the Apodidae, Hemiprocnidae, and-presumably (see above)-Trochilidae is the extreme elongation of the rostrodorsal process of the squamosal, which runs along the dorsal margin of the orbit (Figure 3e). In the adult skull of these apodiform birds, the rostrodorsal process of the bone borders a distinct sulcus for musculus cucullaris capitis (Zusi, 2013, figure 4), and we consider it possible that the extreme elongation of the squamosal is related to a particular development of this muscle.
Columbidae and Pteroclidae form a clade together with the Mesitornithidae in molecular phylogenies (Hackett et al., 2008;Jarvis et al., 2014;Prum et al., 2015). However, whereas the former two taxa have a greatly elongated squamosal, which exceeds the postorbital process in rostrodorsal direction, an elongated squamosal is absent in the Mesitornithidae (although the exact shape of the bone could not be determined in the alcohol-preserved specimen available to us). From a mere morphological point of view, this morphology may represent a synapomorphy of Pteroclidae and Columbidae. However, if the molecular phylogenies correctly reflect the interrelationships of the above taxa, the squamosal was either secondarily shortened in mesites or the morphology in Columbidae and Pteroclidae evolved convergently.
Certainly, functional and ontogenetic constraints on squamosal development are multifactorial and differ across Neognathae. Some of the observed variation may be due to different skull geometries of the involved taxa. It has been shown that the development and shape of the frontals and parietals is correlated with brain morphology (Fabbri et al., 2017;Marugán-Lobón & Buscalioni, 2009). Possibly, therefore, differences in brain shape and associated changes in the sizes of the frontral and parietal may affect the shape of the squamosal.
Different skull geometries are also likely to account for the observed variation of squamosal shape in strigiform birds (Figure 6a-c), which was already recognized by Pycraft (1903b) for the two species studied by him (Athene cunicularia and Strix aluco). As shown by the latter author and the present study, the taxon Athene has a much longer squamosal than other Strigiformes (Figure 6a-c). Current phylogenies support a sister group relationship between a clade including Athene, Surnia, Glaucidium, and Aegolius and all other Strigidae (e.g., Wood et al., 2016), and whether a long squamosal is a derived characteristic of this latter clade or represents an autapomorphy of Athene needs to be examined once juvenile skulls of Surnia, Glaucidium, and Aegolius become available. Whereas in Tyto and most Strigidae the portion of the skull roof that forms the caudodorsal margin of the orbital rim is flattened, it exhibits the usual neornithine morphology (sharp ridge) in Athene. Even though the short squamosal of strigiform birds is therefore associated with a derived cranial morphology, it is more parsimonious to assume that a short squamosal is plesiomorphic for Strigiformes (i.e., Strigidae + Tytonidae).
In the Charadriiformes, by contrast, all examined taxa except the Burhinidae have a greatly elongated squamosal (Figure 5a-j). Because the Burhinidae are phylogenetically nested within other Charadriiformes (e.g., Mayr, 2011), an elongated squamosal is certainly plesiomorphic for the clade. Burhinidae differ from other charadriiform taxa in that the squamosal does not form a long rostrodorsal process and in that the nostrils are holorhinal, whereas they are schizorhinal in other charadriiform birds (Mayr, 2011). Charadriiformes therefore not only provide an example for a secondary shortening of the squamosal (in the Burhinidae), but also indicate that a correlation may exist between squamosal shape and nostril type.
The squamosal is the only bone of the cranial vault that derives from the cranial neural crest, which also gives raise to the facial part of the skull, the quadrate, and the palatal area (Maddin et al., 2016). Therefore mutually interdependent ontogenetic constraints on squamosal morphology and the feeding apparatus may exist and this so much the more, since the squamosal provides an articulation facet for the quadrate. However, no obvious correlation exists between squamosal shape and beak length, and similar squamosal shapes occur in birds with very different beak shapes and feeding ecologies, such as the charadriiform Laridae and the Columbiformes.
We studied squamosal shape in a large number of birds, but the morphology of this bone remains unknown for various critical taxa, including many Strisores (Steatornithidae, Nyctibiidae, Podargidae, and Aegothelidae), some Aequornithes (Gaviidae, Fregatidae), the eurypygiform Eurypygidae, the accipitriform Cathartidae and Sagittariidae, as well as many taxa of Telluraves (Leptosomidae, Coraciidae, the piciform Galbulae). Examination of the squamosal morphology in these birds may further improve our understanding of the phylogenetic and functional significance of the observed variation shown by this bone, and future studies of the comparative morphology of other bones of the juvenile avian skull are likely to augment these data.

ACKNOWLEDGMENT
We thank S. Tränkner for taking the photographs. Comments from two anonymous reviewers improved the manuscript. Open access funding enabled and organized by Projekt DEAL.