- The Paleontological Society
A neurological method for assessment of nasal cavity homologies in extant archosaurs is extended to lambeosaurine hadrosaurids (Dinosauria: Ornithischia) to test functional hypotheses associated with their hypertrophied nasal passages and highly derived cranial crests. The olfactory system and associated cranial nerve pathways that have consistent relationships to soft tissue divisions of the nasal cavity are reconstructed in lambeosaurines on the basis of new paleoneurological data and a comparative phylogenetic approach. The new model of the lambeosaurine olfactory system and nasal cavity shows that a significant portion of nasal cavity proper was located outside the crest cavities and that the primary olfactory region was located rostromedial to the orbits.
All available data indicate that the evolutionary hypertrophy of the nasal cavity occurred predominantly within the non-olfactory nasal vestibule, and that crest development was not causally associated with olfaction. The high level of interspecific and ontogenetic variation in crest shape and nasal vestibule development in lambeosaurine dinosaurs is most consistent with proposed behavioral functions, notably acoustic resonance for intraspecific communication. Despite significant modification to the nasal cavity within Archosauria and its extreme hypertrophy and supraorbital development in Lambeosaurinae, the neural olfactory system and the olfactory region of the nasal cavity proper retain their plesiomorphic positions and associations, suggesting that this system is highly conserved in vertebrate evolution.
The function of the lambeosaurine cranial crest has received considerable attention (Wiman 1931; Ostrom 1961; Sternberg 1964; Heaton 1972; Dodson 1975; Wheeler 1978; Weishampel 1981a,b, 1997; Whybrow 1981; Horner 1995; Sullivan and Williamson 1999; and see references in Ostrom 1962 and Hopson 1975). Although there is now consensus that the morphologically diverse crests likely served as sociosexual visual display structures (Hopson 1975; Dodson 1975; Sampson 1997, 1999; Sullivan and Williamson 1999; Sampson and Forster 2001), the function of the highly derived internal nasal passages (Fig. 1) is still debated. Many proposed functional hypotheses, including several related to underwater feeding, have been critically evaluated and rejected (Ostrom 1962; Hopson 1975; Weishampel 1981b; Sullivan and Williamson 1999). Other hypotheses, including acoustic resonance (Weishampel 1997; Diegert and Williamson 1998), thermoregulation (Wheeler 1978; Horner 1995; Sullivan and Williamson 1999), and olfaction (Ostrom 1962; Horner 1995) are potentially viable explanations for crest development that have yet to be adequately tested.
The structure of the olfactory system and its relationship to the crest cavities has been the most vexing piece of missing data in the debate on lambeosaurine crest function and evolution. The braincase has a close relationship to the crest cavities, but the endocranial anatomy of the forebrain region has remained obscure because it is often poorly preserved or unprepared in most specimens (Fig. 1). Ostrom (1961, 1962) based his reconstruction of the lambeosaurine neural olfactory system on that of lepidosaurs and crocodilians. The olfactory bulb was placed within the common median chamber of the crest, and an elongate olfactory peduncle was reconstructed to extend ventrally out of the crest and curve caudally to connect with the cerebral hemispheres (Ostrom 1962). Subsequent research noted the absence of an empirical osteological basis for Ostrom's reconstruction, and concluded that the location of the olfactory bulb and nerves in both lambeosaurine and hadrosaurine hadrosaurids was unknown (Hopson 1979).
In archosaurs and all tetrapods, cranial nerves have consistent relationships to soft tissue structures of the nasal cavity and, in part, define its histological/functional divisions into the nasal vestibule, nasal cavity proper, and nasopharyngeal duct (Parsons 1970; Witmer 1995b). Olfactory epithelium is associated with the nasal cavity proper, and afferent olfactory nerve branches extend from this region to synapse in the olfactory bulb. The bulb and the nerves can leave clear osteological correlates (e.g., cribriform plate, bulb impressions). Identification of osteological correlates of the olfactory system in archosaurs can clarify lambeosaurine nasal cavity homologies and therefore is critical to any discussion of crest function.
In this paper, new braincase material and the first available forebrain endocranial cast allow forebrain soft tissues and sensory nerve pathways to be reconstructed in Lambeosaurinae by use of a comparative phylogenetic approach. A neurological method for assessment of nasal cavity homologies in archosaurs is outlined and extended to lambeosaurine hadrosaurids. Nasal cavity homologies are reevaluated, and hypotheses of cranial crest function are tested for the first time with neurological data.
MATERIALS AND METHODS
The forebrain of lambeosaurines is described and reconstructed on the basis of isolated and articulated neurocranial material and an endocranial cast (Appendix). Examination of braincases referable to Corythosaurus, Lambeosaurus, Hypacrosaurus, and Parasaurolophus indicates that the morphology of this region is fundamentally similar within the clade. A forebrain endocast of an indeterminate corythosaurian lambeosaurine (ROM 1940, identified as Corythosaurus by Ostrom 1961) was obtained by using the casting technique of Chatterjee and Zheng (2002).
Olfactory neural anatomy and neural-nasal cavity relationships are surveyed in birds and crocodilians, and the olfactory system and associated nerve pathways in Lambeosaurinae are reconstructed using the new paleoneurological data and the extant phylogenetic bracket method, or EPB (Witmer 1995a). Rahman (2002) independently took a similar approach to olfactory reconstruction in Tyrannosauridae. Cranial nerve data and additional anatomical evidence interpreted in the context of the EPB are used to to infer the relative positions of soft tissue divisions of the nasal cavity in lambeosaurines.
Institutional Abbreviations: CM, Carnegie Museum, Pittsburgh; CMN, Canadian Museum of Nature, Ottawa; ROM, Royal Ontario Museum, Toronto; TMP, Royal Tyrrell Museum of Palaeontology, Drumheller; USNM, National Museum of Natural history, Smithsonian Institution, Washington.
Presphenoid Bone Morphology
The presphenoid is the most rostral paired endochondral ossification of the hadrosaurid braincase and contributes to the canal and fenestra that conduct the neural olfactory system (Fig. 2). The use of “presphenoid” here follows its historical and widespread usage in the hadrosaurid literature (Lambe 1920; Parks 1923; Ostrom 1961; Horner 1992) and does not necessarily imply homology with similarly named interorbital ossifications in other groups (e.g., mosasaurs, phytosaurs, pachycephalosaurs, mammals), although the presphenoid of hadrosaurine and lambeosaurine hadrosaurids is presumably homologous. This bone has also been referred to as the ethmoid in hadrosaurids (Gilmore 1937; Horner and Currie 1994) and the sphenethmoid in other dinosaurs (Currie 1997).
In lateral view of the skull the presphenoid is visible through the orbit. It articulates with the orbitosphenoid caudally, the frontal dorsolaterally, and its complement along the ventral midline of the braincase (Fig. 2). The external (ventrolateral) surface is concave and forms the dorsomedial wall of the orbital cavity. A sagittal flange projects ventrally to form a partial bony interorbital septum (e.g., ROM 777, ROM 702). Posteriorly, the interdigitate presphenoid-orbitosphenoid joint extends transversely rostral to the optic nerve foramen. In rostral view the presphenoid bifurcates dorsally into dorsolateral and dorsomedial processes (Fig. 3). The surface between the dorsal processes is rugose for articulation with the frontal at the confluence of its annular and lateral cerebral ridges (Fig. 4). The frontal encloses the olfactory canal of the braincase dorsally.
Endocranially, the presphenoid forms the lateral and ventral wall of fenestra and canal that conducted the neural olfactory system (Fig. 3). The endocranial surface consists of a rostrally tapering rostral flange and a shallow caudal depression. The delicate rostral flange has a series of prominent longitudinal septa that project into the endocranial cavity (Fig. 3). Septa define sulci that are hemispherical in transverse section. There are approximately six to eight sulci; the precise number is variable among specimens (eight in TMP 92.36.219, at least six in TMP 75.11.54). Sulci also vary in their relative prominence. The rostroventral surface of the frontal (the “olfactory depression” [Horner 1992]) forms the dorsal region of the olfactory foramen but does not have complementary sulci or foramina (Fig. 4). The depression caudal to the sulcate flange is shallow and rounded (Fig. 3B). The surface of the depression has a network of minute (<0.5 mm in diameter) transversely oriented grooves (vg, Fig. 3B).
A hemispherical notch in the rostral edge of TMP 92.36.219 represents the caudal margin of a foramen (Fig. 3A–C; interpreted here as the orbitonasal foramen, see below). The foramen does not correspond to any of the sulci of the rostral flange. Gilmore (1937: Fig. 34) illustrated a similar foramen in disarticulated presphenoids of a lambeosaurine from the Two Medicine Formation of Montana (USNM 11893). The caudal margin of the notch indicates that the foramen extends dorsomedially, opposite the direction of the sulci.
Forebrain Endocast Morphology
The forebrain endocast of an indeterminate corythosaurian lambeosaurine (ROM 1940) is the first available for Lambeosaurinae (Fig. 5). The braincase is from a large, presumably adult, individual. In extant reptiles the endocranium reflects forebrain surface morphology and thus the endocast is likely a relatively accurate representation of the shape of the telencephalon (Hopson 1979). Additionally, vascular valleculae on the cerebral fossa indicate a close relationship between the forebrain and endocranium in lambeosaurines (Fig. 4) (Evans 2005), although poor preparation of ROM 1940 obscures this surface detail. Synostosis of the presphenoid-orbitosphenoid complex does not allow the precise demarcation of the neurocranial braincase bones. The dorsal bifurcation of the presphenoid and the sulcate rostral flange are evident in rostral view. The flange is relatively complete, but septa are broken at their bases.
The endocast is hourglass-shaped in ventral and dorsal views (Fig. 5). Rostral and caudal expansions represent the olfactory and cerebral regions of the forebrain, respectively. The large cerebral hemispheres are transversely wide and dorsoventrally compressed. Two large, cylindrical protrusions near the ventral midline represent the optic nerves (CN II). A deep circular fissure separates the dorsal olfactory region from the cerebral hemispheres, but this separation is absent ventrally (Fig. 5C). The olfactory region is represented by an undivided median bulbous segment rostroventral to the hemispheres. The brain cavity expands abruptly at the rostral flange of the presphenoid, and a distinct cylindrical ridge on the left side (interpreted here to represent an olfactory nerve bundle; see below) corresponds to the only preserved presphenoid sulcus in the specimen (Fig. 5A). A series of tubular soft tissue structures would have radiated rostroventrally and rostrolaterally from the median olfactory region if the presphenoid sulci were complete and undamaged.
THE OLFACTORY SYSTEM OF EXTANT ARCHOSAURS
In tetrapods, many soft tissue structures occur in the area rostral to the braincase and olfactory region, including several sensory nerve pathways and vessels (de Beer 1937). Aves and crocodilians form the EPB of lambeosaurine hadrosaurids and all other non-avian dinosaurs (Weishampel 1997). The anatomy the neural olfactory region is reviewed here for birds and crocodilians, and osteological correlates are identified in order to reconstruct this region in lambeosaurines.
Vomeronasal Nerve (CN 0)
Most tetrapods possess an accessory olfactory tract, the vomeronasal system (CN 0), innervating the vomeronasal organ (Pearson and Pearson 1976; Kuhlenbeck 1977; Senter 2002). The vomeronasal organ is not present in postembryonic crocodilians and birds and appears to be absent in all archosaurs (Senter 2002). The vomeronasal system was probably absent in lambeosaurines and is therefore not considered further here.
Olfactory Nerve (CN I) and Tract
The anatomical components of the olfactory system are the largest and most prominent structures that occupy the rostralmost region of the braincase in archosaurs. The olfactory system consists of three basic components, the olfactory nerve, the olfactory bulb, and the olfactory peduncle, which are defined as follows (Kuhlenbeck 1977):
Olfactory nerve: Axons (or axon bundles) of primary neuroepithelial nerve cells that synapse with secondary olfactory neurons in the olfactory bulb. The olfactory sensory cell bodies and dendrites form part of the olfactory epithelium of the nasal cavity.
Olfactory bulb: A swelling of the olfactory tract containing secondary olfactory cell bodies. Primary sensory neurons of the olfactory nerve synapse in the olfactory bulb.
Olfactory peduncle (= olfactory stalk/tract): United secondary axons that carry impulses from the olfactory bulb to the cerebrum.
The olfactory nerves are short in crocodilians and extend caudodorsally from the nasal cavity proper through the cavum orbitonasale (an extracranial cavity between the fenestra olfactoria advehens of the nasal capsule and the fenestra olfactoria evehens of the braincase, = fenestra cribrosa [Shiino 1914]) to the olfactory bulbs (Meek 1911; Starck 1979; Klembara 1991). Many distinct olfactory nerve branches occur on the lateral and medial walls of the nasal cavity, and the afferent olfactory nerves join the lateral and ventral regions of the bulb (Shiino 1914). The bulbs occupy a dorsal position adjacent to the ventral surface of the frontals, rostromedial to the orbits, and a long olfactory peduncle traverses a narrow interorbital space (Starck 1979). Impressions of the olfactory bulb and peduncle are present on the rostroventral surface of the frontal (Iordansky 1973; Senter 2002). Ventrolaterally, the olfactory region is enclosed by the rostral sphenolateral plate cartilage (= orbitosphenoid [Parker 1882]), a paired derivative of the planum supraseptale (= preoptic root of the orbital cartilage [de Beer 1937; Iordansky 1973; Starck 1979; Klembara 1991]). True osteological correlates of the olfactory nerve are absent because the floor of braincase in this region does not ossify (Hopson 1979).
In Aves, the enlargement of the eye has resulted in reduction of loss of the preoptic cartilages (= rostral planum supraseptale) and a rostrocaudally elongate and modified extracranial space representative of the crocodilian cavum orbitonasale (de Beer 1937). Aves thus have a different general mode of olfactory organization than crocodilians, in which the olfactory bulbs have been displaced caudally and the olfactory nerve is elongated and traverses the orbit (Bang and Cobb 1968; Kuhlenbeck 1977; Bubien-Waluszewska 1981). Unlike in crocodilians, the bulbs are located caudomedial to the orbits (Bang and Cobb 1968). The olfactory tract is very short relative to that in crocodilians, and the olfactory bulb is not separated from the cerebrum by an elongate peduncle as it is in crocodilians. The bulb is located directly rostral to the hemisphere, as in turtles and mammals (Bang 1971). Birds are similar to crocodilians in that the bulbs are enclosed within the rostral sphenoid region (developed within the caudal planum supraseptale) and the frontals, except these and the ethmoid cartilages often ossify in birds (Baumel and Witmer 1993; Zusi 1993; Witmer 1995b). In many birds (e.g., ratites), a fossa on the ventral surface of the frontal is a correlate of the bulb (Parker 1893; Chamberlain 1943; Baumel and Witmer 1993).
In most birds, distinct olfactory nerve bundles unite to form a single elongate nerve that extends caudally through the medial orbitonasal foramen, located between the frontal and ectethmoid (an ossification in the lamina orbitonasalis) (de Beer 1937; Bubien-Waluszewska 1981, Baumel and Witmer 1993). The olfactory nerve traverses the interorbital space along the interorbital septum to reach the bulb through the olfactory foramen, the fenestra olfactoria evehens (Bubien-Waluszewska 1981; Wenzel 1987; Baumel and Witmer 1993). The olfactory groove is a prominent sulcus in the mesethmoid bone (= presphenoid [de Beer 1937]; = ethmoid [Bubien-Waluszewska 1981]) formed by ossification around the olfactory nerve (Bubien-Waluszewska 1981; Baumel and Witmer 1993).
This situation does not pertain to apterygids (kiwis), in which the eyes are small and the olfactory chonchae are well developed caudally (de Beer 1937; Bang 1971). In these birds, the elongate nasal capsule reaches the rostral region of the bulbs (Parker 1893; de Beer 1937; Bubien-Waluszewska 1981). Multiple olfactory nerve bundles pass directly to the olfactory bulb, as in crocodilians (Bubien-Waluszewska 1981; Wenzel 1987). The olfactory nerves of apterygids are particularly robust and perforate well-developed mesethmoid and ectethmoid bones to form a cribriform plate (Bubien-Waluszewska 1981; Zusi 1993).
Despite the differences between crocodilians and birds (and among birds), living archosaurs share the following general topographical relationships of the olfactory system: (1) the olfactory bulbs are contained within a cavity formed by cartilage or bone developed within the planum supraseptale and roofed by the frontal; (2) the olfactory bulb is connected to the cerebral hemisphere by a single olfactory tract connection; and (3) the olfactory nerve ramifies when it reaches the nasal cavity proper. The olfactory groove (and cribriform plate of apterygids) in the mesethmoid bone is an osteological correlate of the olfactory nerve in birds. Crocodilians do not have a mesethmoid and osteological correlates of the olfactory nerve are absent, although crocodilians do have a cartilaginous cribriform region derived from probably non-homologous cartilage. An osteological correlate of the olfactory bulb is a depression on the rostroventral region of the frontal in both birds and crocodilians. In crocodilians, impressions of the olfactory peduncle are also present on the ventral surface of the frontal.
Ophthalmic Nerve (CN V1)
In amniotes, the ophthalmic division of the trigeminal nerve passes rostrally through the dorsal portion of the orbit, between the eyeball laterally and the braincase or interorbital septum medially (Witmer 1995b). The ophthalmic nerve divides into two major rami, the ramus medialis nasi and the ramus lateralis nasi, upon reaching the nasal capsule (Witmer 1995b). The ramus medialis nasi enters the nasal cavity proper through the foramen olfactoria advehens with the olfactory nerve in sauropsids (de Beer 1937; Bubien-Waluszewska 1981; Witmer 1995b). The ramus lateralis nasi has an extracapsular course (Witmer 1995b).
The ophthalmic nerve of crocodilians extends along the rostroventral surface of the sphenolateral plate cartilage and enters the cavum orbitonasale in the region of the bulb, through the orbitonasal foramen (or fissure orbitonasalis in Crocodilus) (de Beer 1937; Romer 1956; Klembara 1991). Here the ophthalmic nerve divides into its two main rami. The ramus medialis nasi enters the nasal capsule through the fenestra olfactoria advehens along with the olfactory nerve, extends along the nasal septum, and exits the nasal capsule rostrally via the apical foramen in the rostral solum nasi (Shiino 1914; Klembara 1991; Witmer 1995b).
In birds, the ophthalmic nerve enters the orbit via a foramen in the laterosphenoid bone, a derivative of the posterior orbital cartilage (de Beer 1937; Bubien-Waluszewska 1981; Baumel and Witmer 1993). It extends along the interorbital septum medial to the eyeball, ventral to the olfactory nerve (Witmer 1995b). At the rostral level of the orbit, the ophthalmic nerve divides. The ramus medialis nasi passes through the medial orbitonasal foramen between the ectethmoid and frontal bones (Bubien-Waluszewska 1981; Baumel and Witmer 1993; Witmer 1995b). The ectethmoid ossifies within the lamina orbitonasalis (Witmer 1995b), and thus the medial orbitonasal foramen is homologous to the cartilaginous fenestra olfactoria advehens of crocodilians (de Beer 1937). The ramus medialis nasi and the olfactory nerve enter the nasal cavity proper through the medial orbitonasal foramen (Bubien-Waluszewska 1981; Witmer 1995b). After exiting the nasal capsule, the ramus medialis nasi extends into the rostral medial facial region where it innervates the ramphotheca in birds and the integument in crocodilians and other reptiles (Starck 1979; Witmer 1995b).
Osteological correlates of the ophthalmic nerve are few in crocodilians, in part because the ethmoid region and the nasal capsule do not ossify. The ophthalmic nerve enters the cavum orbitonsale through the orbitonasal foramen rostroventral to the olfactory bulb. Osteological correlates of the ophthalmic nerve in birds include the ophthalmic foramen, the ophthalmic groove in the mesethmoid bone, the medial orbitonasal foramen, and the lateral orbitonasal foramen. Foramina in the premaxilla are osteological correlates of the premaxillary divisions of the ophthalmic nerve in birds and crocodilians (Witmer 1995b).
The extent that intracranial vessels are reflected in the osteology of the bones (i.e., as vascular imprints or grooves) surrounding the forebrain in crocodilians and birds is difficult to determine. In an endocast of Caiman, vessels are not reflected to a significant degree, and the longitudinal sinus of the venous system is faint in the forebrain region (Hopson 1979). In birds, meningeal vessels can leave deep groves on the cerebral fossa, and a dorsal olfactory sinus drains the olfactory bulbs (West et al. 1981; Evans 2005).
THE LAMBEOSAURINE OLFACTORY SYSTEM
The lambeosaurine presphenoid is a paired endochondral ossification rostral to the optic nerve foramen and dorsal to the interorbital septum that conducts the olfactory system. This bone is therefore likely to have arisen within the embryonic planum supraseptale and thus is not homologous with the avian mesethmoid, an unpaired ossification that arises from the interorbital and nasal septa (de Beer 1937). The presphenoids enclose the rostralmost cavity of the endocranium, which is roofed dorsally by the frontals. The olfactory bulb and peduncle (when present) occupy the homologous region in extant archosaurs, below the frontals and rostral to the cerebral hemispheres. The limited space between the cerebral hemispheres and the rostral margin of the frontals suggests that the olfactory bulbs were in close proximity to the hemispheres. Distinct olfactory peduncles are not apparent from braincase morphology, and the frontals lack a rostrocaudally oriented groove that would indicate an elongate olfactory peduncle as in crocodilians. The endocast indicates that both bulbs occupied an undivided median cavity formed by the presphenoids and the frontals. The bulb was partially separated from the cerebral hemisphere dorsally by an annular ridge of the frontals (Fig. 4). The endocranial depression of the presphenoid is interpreted to have formed by ossification around the olfactory bulb. The bulbs were located in a dorsal interorbital position, as in pachycephalosaurids and other ornithischians (Hopson 1979; Giffin 1989).
Organization of the olfactory system into multiple discrete branches, the olfactory nerve, occurs only rostral to the bulb in archosaurs. The rostral septate region of the presphenoid indicates ramification of the olfactory system in a position that is consistent within archosaurs (i.e., distal to the bulb), and the presphenoid sulci are inferred here to represent ossification around distinct branches of the olfactory nerve. The olfactory nerve in extant archosaurs has morphologically similar correlates in the region rostral to the bulb (regardless of the homology of the bones involved in conducting the nerve). A longitudinal sulcus or a cribriform plate is an osteological correlate of the olfactory nerve on the mesethmoid bone of birds. The sulcate region of the lambeosaurine presphenoid is morphologically analogous and possibly homologous to the cartilaginous cribriform region of crocodilians, which lack a mesethmoid. Multiple olfactory nerve bundles joined the rostral bulb at the level of the sulcate rostral presphenoid flanges. The sulci extend ventrolaterally, indicating that a series of nerve bundles diverged rostroventrally and rostrolaterally, as shown in the endocast (Fig. 6B).
The lambeosaurine fenestra olfactoria evehens occupied a position homologous to that of extant archosaurs, between the frontal and rostral sphenoid cartilage/bone ventrally. The length of the sulcate rostral flanges suggests that they enclosed a significant portion of the cavum orbitonasale. The notch in the rostral margin of the presphenoid flange is consistent with the location of the orbitonasal foramen in crocodilians, and is interpreted here to represent the orbitonasal foramen for the passage of the ophthalmic nerve (V1) from the orbit into the cavum orbitonasale and then to the nasal cavity proper.
The precise pathways of the ophthalmic nerve divisions through and around the nasal cavity proper in lambeosaurines are less clear. Foramina within the premaxillary bone are an osteological correlate of the ophthalmic nerve in archosaurs (Witmer 1995b). Parks (1923: Fig. 1) illustrates a foramen on the caudal surface of the premaxilla in the Corythosaurus, a condition that seems to pertain to other lambeosaurines (Hypacrosaurus [Horner et al. 2001]). Horner et al. (2001) noted that the canal extends from the caudal wall of the premaxilla in the area of the s-loop, through the premaxilla alongside the nasal passage, to the tip of the rostrum. In this light, it is likely that the foramen and canal conducted the ophthalmic nerve and associated vessels into the rostral facial region and the ramphotheca.
The large number and large cross-sectional diameter of the presphenoid sulci preclude their interpretation as impressions of intracranial blood vessels that passed through the olfactory foramen. However, the endocranial surface of the presphenoid depression in TMP 92.36.219 preserves minute canals that may represent vessels that supplied the olfactory bulb (Fig. 3B). Paired networks of similarly sized impressions in the lateral region of the olfactory depression in TMP 97.12.95 may represent vessels associated with the dorsal bulb region (Fig. 4A). The position of the olfactory bulbs relative to the cranial roof, as seen in the endocast, suggests that the olfactory depression may have housed a dorsal olfactory sinus (Fig. 5C).
HOMOLOGIES OF THE NASAL CAVITY
The crest cavities have generally consistent morphological subdivisions in Lambeosaurinae, although there is considerable intertaxic variation in their relative sizes (Ostrom 1962; Hopson 1975; Weishampel 1981a). The bony external naris is located at the premaxillary rostrum. A long, tubular passage extends caudodorsally within each premaxilla from the bony naris to the lateral diverticulum (Corythosaurus, Lambeosaurus, Hypacrosaurus) or directly to the supraorbital common median crest chamber (Parasaurolophus) (Hopson 1975; Weishampel 1981a). The premaxillary passages are folded to form an “s-loop” in some taxa (Corythosaurus, Lambeosaurus) (Weishampel 1981a). The lateral diverticulum is continuous with the premaxillary passage rostrally and the common median chamber caudally in all taxa except Parasaurolophus, where it extends caudodorsally from the common chamber (Sullivan and Williamson 1999). The common median chamber is undivided and communicates with the antorbital space via a large, rostrocaudally elongate foramen in the ventral region of the common chamber (Ostrom 1962). The olfactory fenestra of the braincase is located immediately caudoventral to this foramen (Ostrom 1962; Horner et al. 2001). Despite intertaxic variation in nasal cavity morphology, the sulcate rostral flange of the presphenoid is present in both “corythosaurs” (TMP 75.11.54, ROM 1940) and Parasaurolophus (PMU.R1250), and the rostral braincase is fundamentally similar within Lambeosaurinae.
The nasal cavity of tetrapods can be regarded as a modified tube connecting the external naris to the choana. The nasal cavity consists of three morphological and histological/functional divisions: the nasal vestibule, the nasal cavity proper, and the nasopharyngeal duct (Parsons 1970; Witmer 1995b). The nasal vestibule occupies the narial region of the skull and is either a tube or chamber that connects the external naris to the nasal cavity proper (Witmer 1995b, 2001). In crocodilians, the vestibule is short and vertically oriented (Parsons 1970; Witmer 1995b), whereas in birds it is apomorphically expanded and contains the rostral concha (Bang 1971; Bang and Wenzel 1985; Witmer 1995b). In both archosaur groups, the vestibular epithelium is stratified squamous, and not olfactory (Parsons 1970; Bang 1971).
The archosaurian nasal cavity proper is a large, complex chamber that occupies the cartilaginous nasal capsule between the vestibule and the choana (Witmer 1995b). The nasal cavity proper can be divided into two histological regions: a rostral respiratory region and a caudal olfactory region (Parsons 1970; Bang 1971; Bang and Wenzel 1985; Witmer 2001). The nasal cavity proper contains the middle and caudal chonchae of birds and the concha and postconcha in crocodilians (Witmer 1995b). The caudalmost concha in both birds and crocodilians (the caudal and postconcha, respectively) is antorbital in position and is a primary location of the olfactory epithelium (Parsons 1970; Bang 1971; Bang and Wenzel 1985; Witmer 1995b).
The nasopharyngeal duct of crocodilians is a long tube that connects the nasal cavity proper to the pharynx (Parsons 1970). This duct is lined with respiratory epithelium and/ or oral epithelium (Parsons 1970; Witmer 1995b). In birds, the nasopharyngeal duct is very short (essentially absent) and the choana communicates directly with the oral cavity (Witmer 1995b).
Previous attempts to identify these soft tissue divisions in Lambeosaurinae are based upon internal morphological similarities between the crest cavities and the nasal cavities of extant reptiles (Fig. 6A). In his comprehensive work on lambeosaurine nasal cavity homologies, Weishampel (1981a, 1997) largely concurred with the soft tissue divisions proposed by Ostrom (1962) and Hopson (1975). The rostral tubular region was considered homologous to the reptilian nasal vestibule, and the lateral diverticula and common median chamber were considered to contain the region homologous with the nasal cavity proper (Weishampel 1997). A crocodilian-like nasopharyngeal duct was reconstructed to extend from the common median crest chamber to the pharynx (Ostrom 1962; Weishampel 1997). Unlike Ostrom (1962), Weishampel did not postulate on associated neurology.
The new paleoneurological data provide an independent test of these hypotheses of nasal cavity homologies. The soft tissue divisions of the nasal cavity are defined in part by their associated epithelium, and particular regions carry with them expectations in terms of the presence or absence of olfactory innervation. Only the nasal cavity proper is lined with olfactory epithelium; the vestibule and nasopharyngeal duct are not.
There is no evidence that the olfactory nerve entered the rostral premaxillary tubes and s-loop (when present). The premaxilla lacks foramina in its caudal wall that could correspond to the multiple branches of the olfactory nerve (e.g., ROM 776, ROM 777; personal observation 2003). Absence of olfactory innervation in this region is consistent with the hypothesis that the nasal passages within the premaxillae were homologues of the nasal vestibule (Ostrom 1962; Sternberg 1964; Hopson 1975; Weishampel 1997). In extant amniotes the vestibule tends to be restricted to the narial region and bony nostril of the skull (Witmer 2000, 2001). The rostral passages of lambeosaurines are confluent with the bony external naris and are conducted solely by the premaxillae (Sternberg 1964).
Homologies associated with the common median chamber are more problematic because osteological correlates are lacking. The point at which the nasal salt gland duct joins the nasal cavity marks the boundary between the nasal vestibule and the nasal cavity proper in archosaurs (Witmer 1995b). Although several hypotheses have been proposed (Osmolska 1979; Whybrow 1981), the location of the nasal gland and the course of its duct are uncertain in hadrosaurids (Witmer 1997). There is no osteological evidence that the olfactory nerve extended into the common median crest cavity. The paleoneurological test therefore rejects homology of this region of the nasal cavity with the nasal cavity proper. In extant archosaurs significant portions of the nasal cavity proper do not have olfactory epithelium (Parsons 1970; Witmer 1995b). The common median chamber and lateral diverticulum are enclosed by the nasal and the caudal processes of the premaxilla (Hopson 1975; Weishampel 1981a). Topographical and developmental associations between the nasal bone and the nasal capsule in extant vertebrates suggest that at least a portion of the nasal cavity proper was within the common median chamber (Shiino 1914; Shaner 1926; de Beer 1937; Sternberg 1964; Witmer 1995b). Here, the common median chamber is interpreted to have housed the predominantly non-olfactory region of the nasal cavity proper. This interpretation is also consistent with homology hypotheses based on internal crest morphology (Weishampel 1981a).
The nasal cavity proper houses the olfactory epithelium (Parsons 1970; Witmer 1995b). Regardless of the bulb position, ramification of the olfactory nerve occurs in association with the nasal cavity. The pattern and orientation of the olfactory nerve branches can therefore be used to infer the relative position of the primary olfactory region of the nasal cavity proper in lambeosaurines. Endocranial morphology indicates that numerous prominent olfactory nerve branches extended to an olfactory region located medial to the lacrimal at the rostral level of the orbits (Fig. 6B). Therefore, the nasal cavity proper extended ventrally much farther than previously hypothesized.
The position of the choana and the orientation of the nasolacrimal canal provide additional evidence that much of the nasal cavity proper was located outside of the crest cavities. In archosaurs, the primary choana marks the caudoventral extent of the cartilaginous nasal capsule and nasal cavity proper, as well as the boundary between the nasal cavity proper and the nasopharyngeal duct (Witmer 1995b). The primary choana of amniotes is surrounded by the maxilla, palatine, and vomer (Witmer 1995b, 1997). The primary choana of lambeosaurines is located deep to the dorsal process of the maxilla, at the level of the palate (Heaton 1972; Witmer 1997). The nasolacrimal duct opens into the nasal cavity proper rostral to the primary choana in both crocodilians and birds (Parsons 1970; Bang and Wenzel 1985; Witmer 1995b). In lambeosaurines, the lacrimal canal that transmitted the nasolacrimal duct extends rostroventromedially to open on the rostromedial surface of the lacrimal rostrodorsal to the primary choana (Fig. 6B). This relationship is consistent within the EPB.
The lambeosaurine nasal cavity proper therefore extended from the common median chamber to the choana at the level of the palate. The olfactory region of the nasal cavity occupied an antorbital position medial to the lacrimal, a location that is consistent within amniotes (Witmer 1995b). A large fossa across the caudal wall of the paired premaxillae indicates enlargement of the nasal cavity in this region (e.g., CMN 2869, ROM 776, ROM 777). As a consequence, lambeosaurines had a short nasopharyngeal duct relative to previous reconstructions. The nasopharyngeal duct may have extended caudally beneath the arched palatine by way of a soft palate (Heaton 1972).
Hadrosaurine hadrosaurids and the non-hadrosaurid ornithopods are successive outgroup taxa to Lambeosaurinae (Norman 2002; Godefroit et al. 2003). The premaxillae and nasals do not form a hollow crest in these groups, and they exhibit the plesiomorphic condition in the configuration of the olfactory system and nasal cavity with respect to lambeosaurines. The nasal cavity of hadrosaurines and other ornithopods was a relatively direct passage, and configured similarly to that of other reptiles (Heaton 1972; Norman 1980). The vestibule was enlarged to varying degrees in Hadrosaurinae, and diverticula may have been present in many taxa (Hopson 1975; Witmer and Sampson 1999; Witmer 2001; Ruben et al. 2003). The homology of the putative hadrosaurine diverticula with the lambeosaurine lateral diverticula has been suggested (Hopson 1975). The nasal vestibule is lengthened considerably in lambeosaurines with respect to other ornithopods.
The nasal cavity proper was located rostral to the orbits in hadrosaurines and other ornithopods. The structure of the palate and the position of the choana are similar in lambeosaurines, hadrosaurines, and non-hadrosaurid iguanodontians (Heaton 1972; Norman 1980). The boundary between the nasal cavity proper and nasopharyngeal duct is thus also similar to that seen in lambeosaurines. A large portion of the nasal cavity proper, including the olfactory region, retained its plesiomorphic position in Lambeosaurinae. The nasopharyngeal duct was not significantly modified in lambeosaurine evolution.
IMPLICATIONS FOR CREST FUNCTION
Knowledge of lambeosaurine forebrain neuroanatomy and clarification of crest cavity homologies allow for more informed interpretations of the evolution and possible functions of this historically enigmatic structure.
The shape of the crest is highly variable in Lambeosaurinae (Weishampel 1981a). The crest is fan-shaped in Corythosaurus, Hypacrosaurus, and Olorotitan, hatchet-shaped in Lambeosaurus, and tubular in Parasaurolophus (Ostrom 1961; Godefroit et al. 2003). Morphometric analyses indicate that crest growth is positively allometric, with accelerated development of the crest late in ontogeny (Dodson 1975; Evans 2003). The pattern of ontogenetic and interspecific variation in crest shape suggests that the crest served in broadside visual display (Hopson 1975). There is now consensus that the crest had an important function as a sociosexual display structure (Hopson 1975; Dodson 1975; Weishampel 1981b, 1997; Horner et al. 2004; Sampson 1997, 1999; Sullivan and Williamson 1999; Sampson and Forster 2001), and it has been suggested that it was the most significant factor in crest evolution (R. M. Sullivan in Sullivan and Williamson 1999).
Visual display, however, may only partially account for the variation and complexity of the highly derived internal nasal passages. Intertaxic differences in the size and shape of the nasal cavity proper are difficult to assess, but the vestibule is clearly the most variable region of the nasal cavity. The vestibular tubes are relatively short in Lambeosaurus and greatly lengthened in Parasaurolophus walkeri and P. tubicen (Weishampel 1981a,b). The vestibule is intermediate in length in P. cyrtocristatus. Hopson (1975) and Dodson (1975) showed that the development of visual display and nasal cavity components of the crest are partially independent in most taxa. The lateral diverticula are late developmental additions to the nasal cavity in Corythosaurus and Lambeosaurus (Weishampel 1997). Presumably, the diverticula would have had a negligible effect on visual display, but their development would have affected greatly the internal geometry of the crest (Weishampel 1981b). Thus, the hypertrophy and supracranial development of the nasal cavity, unique among vertebrates, does seem to necessitate special consideration in addition to visual display (T. E. Williamson in Sullivan and Williamson 1999).
Ostrom (1962) proposed that lambeosaurine crest development was associated with an increase in surface area for olfactory epithelium, and correspondingly evolved to heighten olfactory capabilities. The olfaction hypothesis has fallen out of favor (Sullivan and Williamson 1999), although no paleoneurological rationale for its rejection has been provided. Sternberg (1964) argued that the expansion of the nasal cavity took place within the non-olfactory region of the nasal tract (Sternberg 1964). Hopson (1975) also disputed the olfaction hypothesis and noted that selection for increased olfactory capability in extant vertebrates does not lead to the diversity of cranial structures such as is present in lambeosaurines.
The olfaction hypothesis (Fig. 6A) (Ostrom 1962) predicts that (1) the olfactory nerves proliferated within the crest passages, and (2) the common median crest cavity represents the enlarged olfactory region of the nasal capsule. There is no osteological evidence that olfactory nerves proliferated within the tubular premaxillary passages, and all available data suggest that the region of the nasal cavity contained within the premaxilla was homologous to the non-olfactory nasal vestibule. The close spatial relationship between the olfactory fenestra and the ventral crest foramen does not preclude the possibility that some olfactory nerve branches extended into the common median chamber of the crest (Horner et al. 2001). However, the position of the olfactory bulb and the orientation of the olfactory nerve sulci indicate that the region of the nasal cavity proper housing the majority of olfactory epithelium was outside the crest cavities. The primary olfactory region was medial to lacrimal at the rostral level of the orbit, a location that is consistent and plesiomorphic within amniotes (Fig. 6B) (Witmer 1995b). The gross organization of the olfactory system and the location of the olfactory region of the nasal cavity suggest that the greatly hypertrophied nasal passages are best interpreted as an evolutionary expansion of the predominantly non-olfactory regions of the nasal cavity, namely the nasal vestibule (Sternberg 1964; Weishampel 1997). Consequently, olfaction can be rejected as a significant factor in the evolution of the lambeosaurine cranial crest.
The size of the olfactory bulbs has been associated with the size of the olfactory organ in reptiles and with the relative importance of olfactory-related behaviors in birds (Bang 1960; Bang and Cobb 1968; Bang 1971; Starck 1979; Bang and Wenzel 1985). The framework for comparison of relative olfactory bulb size across dinosaurs is not yet established. Ellipsoids with axes measured from the lambeosaurine endocast provide a terse estimate of olfactory system and cerebral volumes. The cerebral hemispheres have a combined estimated volume of 134 ml. The olfactory system (6 ml) is approximately 4.5% of the total cerebral hemisphere volume.
Wheeler's (1978) hypothesis that the crest functioned in brain thermoregulation is not rejected by the new neurological data, and comparison with birds does suggest that physiological functions may have been important in lambeosaurine nasal cavity evolution. The vestibule is the most highly variable region of the nasal cavity in birds (Bang 1971; Witmer 1995b) as well as in lambeosaurines. Enlargement of the vestibule and respiratory rostral concha in birds is associated with countercurrent heat exchange and water retention, and has been linked to high metabolic rates in these animals (Bang and Wenzel 1985; Ruben et al. 1996, 2003; Geist 2000; Witmer and Sampson 1999; Witmer 2001). Cross-sectional area of the nasal cavity proper correlates with metabolic rate in extant amniotes (Ruben et al. 1996). The relatively small cross-sectional area of the vestibular tubes and common median chamber suggest that lambeosaurines had ectothermic metabolic rates (Ruben et al. 1996, 2003). Recalculation may be required as data presented here indicate that the nasal cavity proper was significantly larger than previously thought.
The vestibular region of the nasal cavity of lambeosaurine hadrosaurids is elongated compared to that of other dinosaurs, even those with relatively large nasal vestibules (Witmer and Sampson 1999; Witmer 2001). Horner (1995) also reported possible internal turbinate-like structures within the vestibule of Hypacrosaurus. Curiously, these turbinate-like structures have not been mentioned since (e.g., Ruben et al. 1996, 2003), and evidence of their presence in most taxa is currently absent. However, if “turbinates” are present, the distribution of the olfactory nerve suggests they were likely respiratory structures. Enlargement and variation in the vestibule does not reject a potentially important physiological component to crest function, such as countercurrent exchange, water economy, and brain thermoregulation (Witmer and Sampson 1999). Development of the crest may, at least initially, have been related to physiology.
The most widely accepted idea is that the hypertrophied nasal passages served as acoustic resonation chambers (Wiman 1931; Weishampel 1981b, 1997; T. E. Williamson in Sullivan and Williamson 1999; Horner et al. 2004). Low-frequency calls produced by resonation are hypothesized to have been important in intraspecific communication (Weishampel 1997). Elongation of the vocal tract to modify acoustic signals appears to be quite common in vertebrates and has occurred independently in a number of bird lineages (Fitch and Hauser 2003). Physical and computer models of crest passages demonstrate that resonation was possible (Weishampel 1981b, 1997; Diegert and Williamson 1998; Sullivan and Williamson 1999). Acoustic resonance remains the preferred functional hypothesis for the highly modified lambeosaurine nasal cavity. A behavioral explanation is most consistent with the late-maturing ontogeny and interspecific variation in the shape of the nasal passages (Dodson 1975; Hopson 1975; Weishampel 1981a,b, 1997).
Sullivan noted that resonation is an inherent property of the crest cavities (Sullivan and Williamson 1999). Whether sound production drove crest evolution or was an inevitable epiphenomenon of a structure that evolved primarily for different biological reasons, such as visual display, remains unanswered. Clearly, acoustic resonance and other functions are not mutually exclusive. Crest evolution was doubtlessly complex and almost certainly involved several functions (Brett-Surman 1999). The predominant function(s) of the crest should not be interpreted as static. Acoustic, visual display, and physiological functions can all be envisioned to have contributed to the evolution of the lambeosaurine crest and nasal cavity. The relative “significance” of these and other functions likely shifted in different lineages at different points in their evolution. The pattern of ontogenetic and intertaxic variation in nasal cavity development indicates an important behavioral function associated with vestibule hypertrophy, most likely intraspecific communication via acoustic resonance (Weishampel 1997). In at least the derived lambeosaurines, acoustic communication and visual display may have served equally important functional roles (Dodson 1975; Hopson 1975; Brett-Surman 1999; Horner et al. 2004). However, the current data set is restricted to forms with well-developed crests and comparison with noncrested outgroups. New fossils are required to shed light on the early evolution of this structure.
Ostrom (1961, 1962) hypothesized that, like their nasal cavities, the lambeosaurine's neural olfactory system was also highly modified (Fig. 6A). He associated the bulbs with the nasal bone, placing them high within the crest cavities, and correspondingly reconstructed long olfactory peduncles that extended out of the crest (Ostrom 1962). However, the olfactory system typically follows a more direct course in archosaurs, and the bulbs and peduncles are ventral to the frontals. New neurological data indicate that in lambeosaurines the olfactory bulbs were within the ossified braincase and beneath the frontals. Although relatively compact, the structure of the olfactory system is not strikingly different from that of other ornithischian dinosaurs.
Neurological data serve as an independent test of proposed homologies of the soft tissue that lined the crest cavities. The nasal vestibule extended from the external naris to the common median crest chamber. The nasal cavity proper was not restricted to the common chamber as previously thought (Weishampel 1997). Rostroventral radiation of the olfactory nerve and the position of the choana indicate that the nasal cavity proper extended from the common chamber to the level the palate, and that the primary olfactory region was rostromedial to the orbit and medial to the lacrimal (Fig. 6B). The nasopharyngeal duct was correspondingly short.
Clarification of nasal cavity homologies within Lambeosaurinae allows critical evalution of crest function hypotheses. Enhancement of olfactory capabilities can now be definitively rejected as a primary cause for crest development because neurological data confirm that narial hypertrophy occurred predominantly in the non-olfactory nasal vestibule. A behavioral explanation is most consistent with the late-maturing ontogeny of the crest cavities and interspecific variation in the nasal passages. Acoustic resonance remains the preferred functional hypothesis for the complexity of the nasal cavity (Weishampel 1997). Visual display also appears to have been an important function of the crest, and crest evolution likely involved multiple biological functions (Hopson 1975; Brett-Surman 1997).
The olfactory region of the nasal cavity proper consistently retains its plesiomorphic position in archosaurs, despite significant independent structural modification to the nasal cavity in crocodilians and birds and its extreme supraorbital development in Lambeosaurinae. The homologous olfactory conchae of birds (caudal concha) and crocodilians (concha; the postconcha is neomorphic [Witmer 1995b]) are structurally associated with the lacrimal bone (Witmer 1995b), as appears to be the case in lambeosaurines as judged from the inferred pathways of olfactory nerve bundles. Regardless of whether behavioral or other factors were primary in the extreme evolutionary modification of the nasal cavity and surrounding bones in lambeosaurine hadrosaurids, the olfactory system was not radically modified in lambeosaurine evolution, suggesting that this system is evolutionarily conservative among vertebrates.
All hadrosaurid braincase material in the collections of the Royal Tyrrell Museum, Royal Ontario Museum, and Canadian Museum of Nature was studied. Particularly informative specimens include the following: Corythosaurus, CM 11375, CMN 8676, ROM 776, ROM 777, ROM 1933, ROM 1947; Edmontosaurus, CMN 2289; Gryposaurus, TMP 80.22.01; Hypacrosaurus, CMN 8675, ROM 702; Lambeosaurus, CMN 2869; Lambeosaurinae indet., ROM 1940 (braincase and endocast), TMP 67.09.11, TMP 75.11.54, TMP 81.16.370, TMP 92.36.219, TMP 92.36.891, TMP 97.12.95, TMP 99.55.110; Parasaurolophus, PMU.R1250, CMN 8502.
I thank J. Gardner, R. R. Reisz, and M. J. Ryan for their comments and direction. L. Witmer, D. Dilkes, J. Müller, H. Larsson, T. Carr, J. R. Wagner, and M. Kundrat provided much appreciated discussions and comments. I am grateful to K. Seymour and H.-D. Sues (ROM), K. Shepherd (CMN), and D. Berman and A. Henrici (CM) for access to specimens in their care. K. Aulenback and D. Scott provided technical support with the figures, and J. Müller translated several German articles. Special thanks to A. Prieto-Marquez for providing photographs of PMU.R1250. Thanks also to H. Maddin, G. Conacher, M. Guenther, J. Fröbisch, V. Lam, T. Courtenay, D. Tanke, K. Folinsbee, L. Tsuji, M. Yamakawa, M. Bowie, M. Chapman, J. Snow, and E. Snively.
This paper resulted from undergraduate work done at the University of British Columbia (UBC) in conjunction with the Royal Tyrrell Museum. I am indebted to B. Milsom and C. L. Gass (UBC) and J. Gardner, D. Brinkman, P. J. Currie, A. Neuman, and B. Naylor (TMP). A grant from the National Science and Engineering Research Council of Canada provided funding for this project. I thank P. Dodson, R. M. Sullivan, J. A. Wilson, and an anonymous reviewer for comments that improved the manuscript.
- Accepted 15 June 2005.