Homology of Facial Structures in Extant Archosaurs (Birds and Crocodilians), with Special Reference to Paranasal Pneumaticity and Nasal Conchae LAWRENCE M

JOURNAL OF MORPHOLOGY 225269-327 (1995)

Homology of Facial Structures in Extant Archosaurs (Birds and Crocodilians), With Special Reference to Paranasal Pneumaticity and Nasal Conchae LAWRENCE M. WITMER Department of Cell Biology and Anatomy, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205; Department ofdnatomy, New York College of Osteopathic Medicine, New York Institute of Technology, Old Westbury, New York 11568 ABSTRACT Homology of virtually all major components of facial anatomy is assessed in Archosauria in order to address the function of the antorbital cavity, an enigmatic structure that is diagnostic for the group. Proposed functions center on its being a housing for a gland, a muscle, or a paranasal air sinus. Homology is approached in the context of the Extant Phylogenetic Bracket method of reconstructing unpreserved aspects of extinct organisms. Facial anatomy and its ontogeny was studied in extant archosaurs (birds and crocodilians)to determine the osteological correlates of each soft-tissue component; resemblances between birds and crocodilians comprised the similarity test of homology. The congruence test of homology involved surveying phylogenetically relevant fossil archosaurs for these bony signatures. The facial anatomy of extant birds and crocodilians is examined in detail to provide background and to discover those apomorphic aspects that contribute to the divergent specialization of these two groups and thus obscure homologies. Birds apomorphically show enlarged eyeballs, expanded nasal vestibules, and reduced maxillae, whereas crocodilian faces are dorsoventrally flattened (due to nasal rotation) and elongated. Most facial attributes of archosaurs are demonstrably homologous and in fact characterize much more inclusive groups. Special emphasis has been placed on the nasal conchae and paranasal air sinuses. Within Amniota, the following conchal structures are homologous, and all others are neomorphs: avian caudal concha, crocodilian concha + preconcha, Sphenodon caudal concha, squamate concha, and probably the mammalian crista semicircularis. The avian antorbital paranasal air sinus is homologous with the crocodilian caviconchal sinus; the maxillary sinus of placental mammals is not homologous with the archosaurian paranasal sinus. With regard to the function of the antorbital cavity, archosaurs possess homologous nasal glands, dorsal pterygoideus muscles, and paranasal air sinuses, but the osteological correlates of only the paranasal sinus involve the antorbital fenestrae and fossae. Thus, the antorbital cavity is best interpreted as principally a pneumatic structure. o 1995 Wiley-Liss, Inc.

The age-old image of a tiny plover calmly vertebrates living today, during the Mesozoic gleaning the leeches from within the gaping Era extinct archosaurs (i.e., nonavian dino- mouth of a crocodile hardly suggests the no- saurs, pterosaurs, and a variety of early tion of any kinship between these two very forms) radiated into virtually all habitats and different vertebrates. However, among mod- by all measures were the dominant terres- ern vertebrates, birds and crocodilians are trial vertebrates. As a result of this radiation, indeed sister taxa, representing the only sur- viving clades of Archosauria. Although ex- Address reprint requests to Lawrence M.Witmer, Department tant archosaurs, with their 10,000 species, of Biological Sciences and College of Osteopathic Medicine, Ohio remain the most diverse group of terrestrial University,Athens, OH 45701. email: [email protected] o 1995 WILEY-LISS,INC 270 L.M. WITMER archosaurs present an extraordinary diver- called the antorbital fenestra and cavity, re- sity of skull morphology. The pattern of ar- spectively (Witmer, '94). The antorbital cav- chosaur phylogeny among extinct as well as ity (defined below) is ubiquitous in at least living members is beginning to be unraveled the basal members of all clades of archosaurs (Gauthier, '86; Sereno, '86, '91; Gauthier et and is a synapomorphy of a slightly more al., '88a; Benton and Clark, '88; Cracraft, inclusive group (Archosauriformes; Fig. 1; '88; Norell, '89; Novas, '92; Clark et al., '93; Gauthier et al., '88a; Benton and Clark, '88). Parrish, '93), and we are now in a good In some archosaurs (e.g., some theropod dino- position to understand the evolution of archosaurs) the antorbital cavity is very promi- saur craniofacial adaptation. nent, occupying somewhat more than half An important component of the diversity the total skull length, whereas in others (e.g., in skull morphology in archosaurs pertains some ornithischian dinosaurs) it is all but to the facial skeleton, in particular to an lost (see Witmer, in press). Ironically, the opening and space in the side of the snout function (and hence soft-tissue relations) of

Abbreviations a o cav cavitas antorbitalis nldu ductus nasolacrimalis a o fos fossa antorbitalis n 1 du ost ostium nasale, ductus nasolacrimalis a o sin sinus antorbitalis n 1gl fos fossa glandulae nasolacrimalis a o sin ost ostium, sinus antorbitalis npdu ductus nasopharyngeus acc cav cavitas accessorius nar apertura nasi ossea (= naris) ad co aditus conchae nas 0s nasale antorb sin sinus antorbitalis nas gl glandula nasalis ap cavico rec apertura recessus caviconchalis nas gl du ductus glandulae nasalis apnldu apertura ductus nasolacrimalis nas gl gr groove for glandula nasalis atr tur atrioturbinal nas tur nasoturbinal caud co concha nasalis caudalis neurovas neurovasculature caudolat rec recessus caudolateralis neurovas sp neurovascular space cav co cavum conchae olf bulb rec recessus bulbus olfactorius cavico rec recessus caviconchalis ophth N n. ophthalmicus cavico sin sinus caviconcbalis orb orbita cavico sin ost ostium, sinus caviconchalis Pal 0s palatinum cec rec recessus caeci pal bulge bulge of 0s palatinum ch choana pal pr rnax processus palatinus, 0s maxillare CNP cavum nasi proprium pal sin sinus palatinus CN Vi n. ophthalmicus Pare cartilago paranasalis co concha nasalis Pmx 0s premaxillare cr sem crista semicircularis pmx div diverticulum premaxillare eth tur ethmoturbinal pn for foramen pneumaticurn ex a o fen fenestra antorbitalis externa PO vest rec recessus postvestibularis ex add m. adductor mandibulae externus PO vest sin sinus postvestibularis ex co rec recessus extraconchalis postco postconcha fen nar fenestra narina postco cav cavitas postconchalis font a o fonticulus antorbitalis preco preconcha for epiph foramen epiphaniale preco rec recessus preconchalis fr 0s frontale Prf 0s prefrontale in a o fen fenestra antorbitalis interna prf rec recessus prefrontalis jo organum vomeronasale ( = Jacobson's prim ch choana prima (= primary choana) organ) PtC cartilago parietotectalis jug 0s jugale pter 0s pterygoideum jug bar arcus jugalis pter dor m. pterygoideus, pars dorsalis lac 0s lacrimale Ram lat nas ramus lateralis nasi. n. oDhthalmicus lac cav cavitas lacrimalis Ram med nas ramus medialis nasi, n. opbthalmicus la div diverticulum lacrimale rec du np recessus ducti nasopharyngei lam orb lamina orbitonasalis roof nas cap roof of cartilaginous nasal capsule lat Gr laterale Grenzfalte root co root of concha LTR lamina transversalis rostralis ros co concha nasalis rostralis mand mandibula scl sclera mand N n. mandibularis sec ch choana secundaria (= secondary choana) max 0s maxillare SeP septum male max N n. maxillaris suborb fen fenestra suborbitalis rnax sin ost ostium sinus maxillaris sep sulc sulcus septalis max tur maxilloturbinal tect nas tectum nasi mes 0s mesethmoidale vest vestibulum nasi mid co concha nasalis media vom vomer musc fos fossa muscularis, 0spalatinum FACIAL HOMOLOGIES IN ARCHOSAURS 271

a o cav lac Comparison of birds and crocodilians within the context of both amniote and archosaur phylogeny also provides the opportunity to study a striking example of divergent specialization superimposed upon a shared, inherited ground plan. Crocodilians have a “pri- mordial” look about them and are commonly portrayed as living fossils little changed since their origin over 200 million years ago (Romer, ’66).Indeed, they retain manyprimi- tive features such as the presence of teeth “\ and most of the original complement of skull max pter Pal bones. However, modern crocodilians exhibit numerous apomorphies in all regions of the Fig. 1. Eupurkeria capensis, a basal archosauriform skull (Langston, ’73) and particularly in the from the Triassic of South Africa, in left lateral view, face, probably in association with skull flat- showing the antorbital cavity within the lateral aspect of the snout. Modified after Ewer (’65) and Witmer (’87). tening and formation of a long nasopharyngeal duct. On the other hand, birds, despite being highly modified for flight, show some aspects of skull morphology that are actually the cavity is enigmatic and has been a matter primitive in comparison to crocodilians. For of some dispute (see Walker, ’61; Ewer, ’65; example, although birds apomorphically have Osmolska, ’85; Witmer, ’87; see Witmer, in press), the three major hypotheses being that lost several skull bones in association with the cavity housed 1) a gland, 2) a portion of the evolution of cranial kinesis, they primi- the jaw musculature, or 3) a paranasal air tively retain an external antorbital fenestra. sinus. The glandular hypothesis never has In fact, this mosaic pattern is a good illustra- had many adherents, whereas the muscular tion of why characters, not taxa, should be hypothesis has been by far the most popular regarded as “primitive” or “derived” (Brooks (summarized in Witmer, in press). More re- and McLennan, ’911. Specific apomorphic as- cently, a novel anatomical system, paranasal pects of facial development in extant birds pneumaticity, has been implicated, and pre- and crocodilians are examined in a later sec- liminary studies (Osmblska, ’85;Witmer, ’87) tion so that those aspects contributing to suggested that the pneumatic hypothesis is their “divergent specialization” may be iden- the best corroborated of the three. tified and accounted for when attempting to To address these hypotheses, this paper discover facial homologies. probes the facial anatomy of extant archo- Elucidation of facial homologies and char- saurs and other amniotes for homologous acterization of the disparity among extant soft-tissue attributes that leave reliable indi- archosaurs require an appropriate phyloge- cations of their presence on the bones (i.e., netic perspective. Historically, solving these “osteological correlates”) that can be as- problems has been greatly hampered by typo- sessed in fossil material. The detailed mor- logical thinking. Typology has been probably phology and distribution of these osteological the most influential factor in, for example, correlates in fossil archosaurs and their sig- the debate on the homologies of the nasal nificance for the function of the antorbital conchae and paranasal air sinuses. Numer- cavity is discussed in detail elsewhere (see ous workers (e.g., Gegenbaur, 1873; Meek, Witmer, in press) and summarized here in ’06, ’11; Matthes, ’34;Bertau, ’35; Schuller, the last section. Special attention will be paid ’39; May, ’61; among many others) searched here to the patterns and homologies of para- for features that could be homologized with nasal air sinuses as these are the most poorly named structures of mammals or, in some documented in the literature (Bellairs and cases, humans, apparently working implic- Kamal, ’811. The cartilaginous nasal conchae itly under the typological belief that all organ- are important morphological landmarks in isms will have the characteristics of the arche- the nasal region. In the course of this study, type. With phylogenetic thinking, however, it became apparent that the homologies of one expects to dscover novel attributes char- the conchae are not particularly clear, and, as acterizing just a subset of a taxon. a result, their homologies among amniotes A related problem plaguing this kind of also are examined. analysis is paraphyly. Many workers (e.g., 272 L.M. WITMER Romer, ’66; Colbert, ’80; Carroll, ’88)have treated Archosauria as paraphyletic, exclud- ing birds. Thus, extant crocodilians and many fossil archosaurs often have been compared to other “reptiles” rather than to their extant sister taxon, birds (e.g., Dollo, 1884; Meek, ’ 11;Anderson, ’36;Wettstein, ’37-’54; Parsons, ’59, ’70; many others). Clearly, paraphyletic classification can result in exclusion from the analysis of very relevant taxa. For example, in the debate on the function of the antorbital cavity, a paraphyletic Archosauria excludes the only extant taxon actually retaining an antorbital fenestra. As a result, all Uiapsida comparisons and inferences made here will be within the context of an independently corroborated phylogenetic hypothesis (see Materials section). FTrtrapoda This paper is organized in the following Fig. 2. Phylogenetic relationships of extant Amniota manner. First, the phylogenetic framework (topology after Gauthier et al., ’88a). and the archosaur taxa receiving detailed study are introduced in the Materials section. The next section presents the method- ological and theoretical foundation of this the course of the following analysis, several research, emphasizing its relationship to the extinct archosaur species will be mentioned, Extant Phylogenetic Bracket approach for and these also are indicated in Figure 4. reconstructing soft tissues in extinct taxa All of the extant taxa listed below were (Witmer, ’95). The following section outlines obtained as fresh or preserved whole ani- the various methods used to study the mor- mals, eggs, or heads. Among the avian sample, phology and ontogeny of the anatomical sys- five species received the greatest attention: tems. Next, the facial anatomy of extant ar- commercially raised, domestic breeds of 1) chosaurs is presented system by system; the White Leghorn chicken (Gallus gallus), 2) purpose here is to provide in one place (i.e., as Peking duck (Anas platyrhynchos), and 3) a reference) the comparative anatomical in- Greylag goose (Anser anser),4) commercially formation necessary for tackling the thornier raised ostrich (Struthio camelus),and 5)wild- issues of homology. Acknowledging the collected Laysan albatross embryos (Di- marked and obvious differences between birds omedea immutabilis). Several additional spe- and crocodilians, the following section identi- cies of birds were studied for comparison, fies those facial apomorphies that strongly each species sample comprising fewer than contribute to this disparity and obscure the three specimens: rhea (Rheaamericana), emu homologies. The next section gets down to (Dromaius novaehollandiae), ring-billed gull the business of assessing the homology of (Larus delawarensis), common crow (Coruus facial anatomical components-again, sys- brachyrhynchos), and mourning dove (Ze- tem by system-among archosaurs. The last naida macroura). Other extant taxa (not section summarizes these findings and exam- listed) were studied from dried skulls. ines their impact on the debate concerning Among crocodilians, three species received soft-tissue relations of the antorbital cavity. the greatest attention: 1) American alligator (Alligator mississippiensis), collected from the Rockefeller Wildlife Refuge, southwest- MATERIALS ern Louisiana, 2) saltwater crocodile (Croco- Figure 2 depicts the phylogenetic relation- dylus porosus), and 3) New Guinea freshwa- ships of extant Tetrapoda (see Gauthier et ter crocodile (C. novaeguineae); both species al., %a), and Figure 3 shows the relation- of crocodiles were collected from a crocodile ships of the extant taxa examined for this farm in Papua New Guinea. Additionally, study. The relationships of the major clades juvenile specimens of common caiman of archosaurs (and also nonarchosaurian ar- (Caiman cmcodilus) and singlejuvenile spec- chosauriforms) are provided in Figure 4. In imens of false gharial (Tomistoma schlegelii) FACIAL HOMOLOGIES IN ARCHOSAURS 273 9

Fig. 3. Phylogenetic relationships of the extant taxa studied (topology after Gauthier et al., '88a; GatTney and Meylan, '88; Benton and Clark, '88; Cracraft, '88). and gharial (Gavialisgangeticus) were sagit- staged according to Koecke ('58) and Starck tally sectioned and dissected for comparison. ('89).Although Gallus andhasconstituted Species selected for detailed ontogenetic the most extensive series, shorter series of study using clearing and staining were cho- Anser anser and Diomedea immutabilis also sen primarily based on the availability of were prepared. One Anser embryo each from carefully aged embryos (Table 1). Among days 20 to 27 of incubation (except day 26: birds, eggs of Gallus gallus and Anas platy- two embryos; 28-30 days total incubation rhynchos were obtained commercially and time) and two heads of approximately 3-day- incubated at 37°C in a forced-draft, humidi- old goslings were cleared and stained. Ten fied incubator. For Gallus, two to six eggs Diomedea embryos ranging in age from day (averaging about four) were removed from 17 to day 32 (approximately 63 days total the incubator every day (and at about the incubation time) were selected for clearing same time of day) from day 8 of incubation and staining. Anser and Diomedea embryos up to hatching (21 days). Gallus embryos were weighed prior to processing; no staging were weighed and staged according to the scheme is available for either species, and scheme of Hamburger and Hamilton ('51). none was devised here. Additionally, a late Additionally, posthatching chicks aged 1 day, Rhea americana embryo of unknown age 7 days, and 28 days (three each) were ob- (about 260 g) and a very young chick of tained for clearing and staining. For Anas, Struthio camelus were cleared and stained. two to seven eggs (averaging about four) Among crocodilians, Alligator mississippi- were removed from the incubator every day ensis eggs were retrieved from nests and from day 9 of incubation through day 17 of incubated at the Rockefeller Wildlife Refuge incubation; thereafter three to four eggs were (see Joanen and McNease, '77, '79). Two to removed every 2 days up to hatching (about three eggs were removed from the incubator 28 days). Anas embryos were weighed and every day from days 13 to 51 of incubation 274 L.M. WITMER

2 Crocodvlomoroha

YArchosauriformes

Fig. 4. Phylogenetic relationships of major clades of fossil archosauriforms,including those genera mentioned in the text (topology after Gauthier, '86; Cracraft, '86; Benton and Clark, '88; Sereno, '91; Sereno and Arcucci, '90; Sereno and Wild, '92; Wu and Chattejee, '93).

and one or two eggs for most days thereafter Ferguson's ('85) scheme. Additionally, until hatching (about 65 days) and fixed by hatchlings up to day 7 were collected (averag- injecting with 37% formaldehyde (100% for- ing five per day) and fixed in formalin. Of this malin) until the eggshell cracked and then sample, 45 were processed as cleared-and- submerging in the same solution (see Fergu- stained specimens. Additionally, a Caiman son, '81). Most embryos were weighed and crocodilus hatchling (187 mm SVL) also was measured, and all were staged according to processed in this fashion.

TABLE 1. Numbers of individuals of the major study taxa examined and their means ofm-euaration Serially Cleared and Latex Taxon Dissected Skeletonized sectioned stained injected

Alligator mississippiensis 17 29 312 45 33 24 Caiman crocodilus 3 4 0 1 0 Crocodylus porosus 7 17 1' 0 0 C. novaeguineae 2 6 1' 0 0 Struthio camelus 3 4 1' 1 14 Diomedea immutabilis 0 0 15 10 0 Gallus gallus 6 10 1' 83 83 64 Anas platyrhynchos 8 10 1" 43 63 44 Anser anser 7 10 21 11 1s 34

'Serial gross sectioning. 2Serial histologicalsections of numerous individuala ofAllzgator mississippiensis were studied in the laboratory of Dr. M.W.J. Ferguson, University of Manchester. "Clearedand stained after latex injection. 4Preparedas corrosion cast after latex injection. 5Serial histologicalsectioning. FACIAL HOMOLOGIES IN ARCHOSAURS 2 75 Most information on extant nonarchosau- these are the only forms for which details of rian amniotes was obtained from the litera- the soft tissues and their relationships to the ture. Literature reports were confirmed by skeleton can be observed directly. However, sagittal section and gross dissection of the not all extant taxa are equally relevant. In following species: 1) Squamata: Agama stel- particular, reference to (minimally) the first lio, Gekko gecko, and Acanthodactylus boski- two extant outgroups of the fossil taxon of anus; 2) Testudines: Sternotherus minor, interest is required (Fig. 5A) because at least Chrysemys picta, and Geochelone carbonaria; two outgroups are required to justify charac- and 3)Mammalia: Oryctolagus cuniculus and ter assessments at the outgroup node (i.e., Homo sapiens. the common ancestor of the fossil and its Fossil material of extinct archosaurs was extant sister group; see Maddison et al., ’84). studied in museum or university collections, Reorganizing Figure 5A so that the extant scored for its morphological features, photo- taxa flank the fossil taxon provides a heuris- graphed, and, in a few cases, X-rayed. tic, graphical representation (Fig. 5B) of an important implication of this approach- ANALYTICAL METHODS that is, the extant taxa “bracket” the fossil taxon and therefore constrain any soft-tissue Phylogenetics, extant taxa, and the inferences. In fact, the extant may be reconstruction of soft tissues in fossils taxa termed the “extant phylogenetic bracket” of As alluded to at the outset, the research the extinct taxon. presented here is part of a larger project that The anatomy of the extant taxa is studied explores the evolution of facial anatomy in with attention to the soft tissues and their archosaurs-both extinct and extant. Fur- relationships to the skeleton. In particular, thermore, it seeks to discover the function of causal associations of the soft tissues and the antorbital cavity, their most important bones are sought. There is mounting evi- facial apomorphy. Determining the function dence that the form (and in many cases even of the antorbital cavity clearly centers on the existence) of bony features is largely or en- soft-tissue relations of the bony cavity: viz., tirely under the morphogenetic control of whether it housed a gland, a muscle, a para- nonskeletal tissues (reviewed in Witmer, ’95; nasal air sinus, or some other structure. As a see also Herring, ’93). That is, many soft- result, this research requires an objective means of inferring soft anatomical components in the fossil remains of extinct organisms. A detailed methodology (the Extant Phylogenetic Bracket approach) is presented elsewhere (Witmer, ’92, ’95; see also Bryant and Russell, ’92) and is briefly outlined here. As will be seen, this approach is in many ways simply a specific application of the well-

understood (though still hotly debated) prin- rotation around ciples of homology determination. Thus, this ourgroup node study emphasizes the homologies among ex- 6 tant archosaurs, whereas a companion re- Extant Phylogenetic Bracket port (Witmer, in press) focuses more on fossil < rn archosaurs. I previously have discussed (Wit- mer, ’95) the relevance of soft-tissue consid- - erations for accurate interpretations of bony morphology, but these issues will not be addressed here in detail. The methodology for reconstructing soft tissues in fossils is outlined below 1) to provide the framework for the importance of extant taxa in studies of taxa known only as fossils, and 2) to highlight the kinds of data sought here from extant taxa (see Witmer, Fig. 5. A: Phylogenetic relationships of a fossil taxon and its first two extant sister groups. B Rotation around ’95, for elaboration). In order to infer soft the outgroup node in A brings the extant taxa to the tissues in a fossil taxon, information about periphery, forming the Extant Phylogenetic Bracket. soft anatomy must come from extant taxa as Modified after Witmer (’95). 276 L.M.WITMER tissue components produce (i.e., cause) par- amples and examines some of the difficulties ticular osteological attributes. In practice, that may be encountered. the goal is to discover unambiguous bony Reconstruction of ancestral (or "extinct") signatures left on the bones by known soft attributes presents no particular theoretical anatomical components, which thus are re- problems because it is based simply on the garded as the osteological correlates of that relation of biological homology. Thus, this component. Of course, the causal nature of method is basically a specific adaptation of these associations are assumptions or hypoth- the well-known principles for the elucidation eses amenable to testing. The point (ideally) of homologous characters. Homologies are is to identify those soft-tissue attributes that hypotheses subject to a series of tests (Patter- are both necessary and sufficient to explain son, '82, '88; Stevens, '84; Rieppel, '88, '94; particular bony features. Bock, '89): 1) the similarity test, whereby However, these attributes have an evolu- putative homologs must share particular to- tionary history. Once the causal associations pographical relationships to, or 1:1 correspon- are in hand one may return to the cladogram dences with, other structures; 2) the congru- and pose the following hypothesis: Any simi- ence test, whereby the homology must larities in the soft tissues and their osteologi- characterize a monophyletic group; and 3) cal correlates between the extant members of the conjunction test, whereby putative ho- the bracket were inherited from their com- mologs must not co-occur simultaneously in mon ancestor (located at the "bracket node"; the same organism. Fig. 5B) which itself had the same causal Clearly, the above methodology for recon- association (Fig. 6, dashed arrow). A predic- structing soft tissues in fossils is directly tion of this hypothesis is that other descen- comparable to these tests of homology (Wit- dants of the bracket ancestor also inherited mer, '95). The similarities between the living the soft-tissue attribute. The hypothesis is members of the extant phylogenetic bracket thus tested by surveying these other descen- in both soft anatomical attributes and osteo- dants-i.e., the fossil taxa-for the osteological correlates (Fig. 6, solid arrow), with parsi- logical correlates are appraised on the basis mony deciding the fate of the hypothesis. The of sharing specific 1:1 correspondences, i.e., technical paper (Witmer, '95) provides ex- the similarity test. The hypothesis that these similarities were present in the bracket ancestor is thus a hypothesis of homology. The congruence test involves surveying other de- Extant Phylogenetic Bracket scendants of the bracket ancestor (the fossil taxa) for the osteological correlates: If the fossil taxa exhibit the specified osteological correlates, then the bony feature and its correlated soft tissue characterize the monophyletic group comprising the bracket, and the hypothesis survives the congruence test. The conjunction test is not a formal part of the methodology, but co-occurrence of putative homologs presumably is discovered in association with the similarity test. Thus, determination of homologous osteological correlates, combined with causal association of the correlates with particular soft tissues, allows the inference of the soft tissue in taxa preserving only hard parts. Bock ('89) has emphasized the importance of the similarity test, and Fig. 6. Basic scheme of hypothesis formulation and thus this study provides detailed analysis of testing in the Extant Phylogenetic Bracket approach. Similarities between the components of the EPB are anatomical and ontogenetic similarities; con- hypothesized as being present in the bracket ancestor gruence with the phylogeny of extinct and (broken arrows). This hypothesis is tested for its congru- extant archosaurs is noted here but is more ence with the phylogenetic pattern by surveying the fully elaborated elsewhere (Witmer, in press). fossil taxa (solid arrows). The EPB approach provides phylogenetic justification for the inference of soft tissues The ontogeny of structures plays an impor- or other unpreserved features in fossil organisms. Modi- tant part in the discussion to follow. Al- fied after Witmer ('95). though some workers (e.g., Roth, '84) have FACIAL HOMOLOGIES IN ARCHOSAURS 277 argued that development is one of the clear- tion; embryonic material of Gallus gallus, est guides to homology, the use of ontoge- Anas platyrhynchos, Anser anser, Struthio netic patterns has its limitations and should camelus, and Alligator mississippiensis also not be given primacy over other evidence (see was dissected (see Table 1). Some specimens Alberch, ’85; Smith and Hall, ’90; Mabee, for all species were frozen solid and then ’93). Instead, ontogenetic information sim- sagittally sectioned with a band saw prior to ply provides additional, highly detailed data dissection; some crocodilian specimens were on the 1:l correspondences that go into the sectioned horizontally. A small, rotary, power similarity test of homology (Remane, ’52; saw was indispensable for dissecting the MacPhee, ’81; Stevens, ’84). In some cases, snouts of the crocodilians. Almost all of the the topographical relationships become so dissections were extensively photographed transformed during the later portions of on- throughout the procedure. Specimens were togeny that 1:1 correspondences are ob- studied with special attention to the soft scured; studying the pattern of ontogenetic tissues of the rostra1 portion of the head, and transformation often helps reveal these corre- in particular, their topographic relationships spondences. For example, many of the air to each other and to the skeleton; details of sinuses and diverticula of extant archosaurs the osteological correlates of the soft tissues merge and become broadly confluent in were noted. adults, whereas earlier in ontogeny they de- Some dissected specimens subsequently velop in relative isolation and are more easily were fixed in 10% neutral-buffered formalin characterized (Witmer, ’90, ’92;see below). for 2-5 days and then stored in 70% ethanol. Furthermore, comparison of entire ontog- All other dissected specimens (except those enies (or life cycles) is appropriate because destined for another technique) were skel- ontogeny is continuous and cannot be di- etonized (see Witmer, ’92, for details) so that vided nonarbitrarily into “stages” (de Quei- the osteological correlates of the observed roz, ’85; Alberch, ’85; Rieppel, ’88). Even soft tissues could be better assessed. “adult” is somewhat inappropriate as a devel- Serial gross and histological sectioning opmental stage in forms with indeterminate growth such as crocodilians. For example, Serially sectioned specimens give detailed the paranasal and nasopharyngeal sinuses of relational information on all tissues, and in crocodilians (see below) continue to expand particular are a critical complement to the and pneumatize more of the skull long after cleared-and-stained specimens. For example, many of the classic hallmarks of “adulthood” although cleared-and-stained specimens pro- are reached. Therefore, whole ontogenies will vide better data on the three-dimensional be considered here when possible, and, al- relationships of particular osteological corre- though this goal is seldom realized, the per- lates, serially sectioned specimens allow the spective is fruitful. soft-tissue components to be related more directly to their correlated bony (or cartilagi- ANATOMICAL METHODS nous) features. Adult and posthatching juve- Four major techniques were used to study nile specimens of Alligator mississippiensis, the soft tissues and their relationships to the Crocodylus porosus, C. nouaeguineae, and skeletal tissues (Table 1): 1) gross dissection Anser anser were serially sectioned grossly and skeletonization, 2) serial gross and histo- by freezing the head solid and transversely logical sectioning, 3) clearing and staining of sectioning the head on a band saw, cutting embryonic and young material, and 4) latex sections approximately 3-12 mm thick de- injection of various cavities (especially the pending on the length of the head. Heads of paranasal air sinuses). In some cases, more Gallus gallus, Struthio camelus, and small than one technique was performed on the juvenile Alligator were sectioned transversely same specimen. In all cases, the goal was to without freezing using a scalpel (Table 1). If understand particular aspects of the ontog- not already fixed, the sectioned specimens eny of facial anatomy. These four techniques were then fixed in 10% neutral-buffered for- are discussed in turn. malin and stored in 70% ethanol. One head each of Alligator and C. porosus was sec- Gross dissection and skeletonization tioned as outlined above, and then these Adults and posthatching juveniles of vari- transverse sections were sectioned sagittally; ous ages of all of the major study taxa (except one side of the head was fixed and the other Diomedea immutabilis)were studied through side was skeletonized (see above) so that the the well-known techniques of gross dissec- osteological correlates of the soft tissues and 2 78 L.M. WITMER the soft tissues themselves could be com- tained from other techniques (e.g., dssec- pared easily. tion, sectioning, latex injection, etc.), cleared- One embryo each of Diomedea immutabi- and-stained specimens provided a rather lis (25 days of incubation, total weight 3.35 g) detailed picture of the form and ontogeny of and Anas platyrhynchos (15 days of incuba- the facial bones and related soft anatomy. tion, total weight 6.54 g) were subjected to The following procedures are extracted routine serial histological sectioning (i.e., em- from Witmer ('92) which should be consulted bedded in paraffin, serially sectioned at 10-13 for more details. km, and stained with hematoxylin and eosin) and studied with light microscopy (Table 1). 1. Initial preparation In addition, numerous serially sectioned As a rule, specimens were skinned, enucle- specimens of Alligator mississippiensis in ated, eviscerated, and debrained before fixa- the collection of Dr. M.W.J. Ferguson were tion, although the Alligator mississippiensis studied at the University of Manchester, En- material was fixed first with no apparent ill gland, and many were photographed (see Fer- effects. guson, '81 for his techniques). 2. Bleaching Clearing and staining of embryonic Heavily pigmented and particularly large or young material specimens (e.g., Alligator and Anser anser Clearing and staining is a well-known tech- embryos, many hatchling birds) benefited nique for visualizing bone andlor cartilage in from bleaching and mild maceration in a whole-animal preparations by selectively solution of about one part 3%HzOz to 9 parts staining bone with alizarin red S and carti- 0.5% KOH for a period of rarely more than lage with alcian blue, and rendering the other 1.5-2.5 hours. tissues transparent by clearing with pancre- atic enzymes and a graded series of potas- 3. Fixation sium hydroxide (KOH) and glycerol solu- Five different fixatives were used: a) 37% tions. Special protocols for crocodilians were formaldehyde (100% formalin), b) 10% neu- required, and hence a new variant of the tral-buffered formalin (NBF), c) 95% etha- widely used procedures of Wassersug ('76) nol, d) 1%acetic acid in 95% ethanol, and e) a and Dingerkus and Uhler ('77) was devel- new fixative termed EFA (Witmer, '92), an oped (Witmer, '92). A sketch of this method acronym for ethanol-formalin-acid (90 parts is provided below. Within an ontogenetic se- 95% ethanol, 7 parts NBF, 3 parts acetic acid). ries, the earliest appearance of stain is a readily identifiable marker for the onset of 4. Fat removal chondrogenesis or ossification (Alberch and Since fat usually fails to clear completely, Alberch, '81; Hanken and Hall, '84) and was diffuse fatty tissue was removed by treat- used to establish the timing of these events. ment with acetone for 2-3 days. Cleared-and-stained specimens offer the advantages of 1) preserving three-dimensional 5. Cartilage staining relationships such that topographical rela- Specimens weighing less than 15-20 g were tionships can be assessed from all angles, and stained in a fresh solution of 11 mg alcian 2) being rapid enough that large numbers of blue, 80 ml 95% ethanol, and 20 ml acetic specimens can be prepared. Although the acid. Specimens weighing more than 15-20 g technique renders tissues other than carti- (and all Alligator specimens) were stained in lage and bone transparent, thus seemingly a fresh solution of 11 mg alcian blue, 77.5 ml obscuring the desired non-bony-tissue data, 95% ethanol, and 22.5 ml acetic acid. Stain- the specimens retain much of the evidence ing times were 1.5-2.0 times in hours the age for many of the soft tissues. For example, the of the specimen in days, up to 48 hours total. ostia of the paranasal air sinuses are apparent in the cartilaginous nasal capsule, and 6. Dehydration foramina in the bones or nasal capsule for Most specimens were dehydrated by 24-48 passage of nerves, vessels, and pneumatic hours in 95% ethanol, changing the solution diverticula also can be observed. Further- two to four times during this period. more, muscles, ligaments, and some other connective tissues often are visible with care- 7. Enzymatic clearing ful lighting as "ghosts." Thus, when further Specimens were taken to distilled water compared with soft-tissue information ob- through a graded ethanol series, and then FACIAL HOMOLOGIES IN ARCHOSAURS 279 treated with an enzyme solution (30 ml of easy; and 6) it is inexpensive and readily saturated aqueous sodium tetraborate available. Its major drawback is that it tends (Na2B407 . lOHzO), 70 ml of distilled water, to shrink somewhat, although this can be and 1 g of 4x pancreatin) and maintained at ameliorated (see below). 37°C in a water bath until much of the skel- The basic constituents of the injection me- eton was visible, changing the enzyme solu- dium were 1) about 65 g of latex, 2) 10 ml of tion every 4-5 days. distilled water (to decrease the viscosity), 3) 8. Bone staining 3.5 g of a finely particulate filler composed of amorphous fumed silica (to reduce shrink- The specimens were stained for bone tis- age; brand: Aerosil200), and 4) 150 drops of sue with a solution of alizarin red S in KOH colored latex-based drafting-pen ink (to add (15 drops of 0.1%aqueous alizarin red S in some contrasting color to the off-white latex). 100 ml of 0.5% KOH), spending 1-2 days in The relative proportions can be altered. The the solution. injection apparatus consisted simply of a stan- 9. Clearing dard 3-ml or 5-ml disposable syringe mounted with a modified 18- to 25-gauge needle. The Final clearing was achieved by taking the needle was modified by bending the distal 10 specimens to glycerol through a graded 0.5%- mm into about a 45" angle, being careful not KOH/glycerol series (i.e., 3:1, 1:1, 1:3, pure to crimp the lumen; then, under a micro- glycerol). scope, the point of the needle was ground 10. Storage down with a whetstone or rotary grinding Completed specimens were stored in clean, wheel until the aperture was completely ter- tightly lidded containers in fresh, pure glyc- minal and no burrs remained. This modified erol to which some thymol crystals were blunt needle thus could be directed into vari- added. ous subcavities within the sinuses without undue concern for puncturing the epithelial Latex injection of sinus cavities walls of the sinus. In crocodilians, the paranasal air sinuses Fresh material was more suitable than pre- are largely enclosed in bone, such that the served (fixed) material because the fresh tis- extent of the sinuses can be judged readily in sues were more supple and natural. Fresh or dried skull material. In birds, however, much thawed heads of several bird species were of the main sinus and its diverticula are sit- injected (Table 1). If various skull bones were uated either subcutaneously or among other suspected to be aerated and in communica- soft tissues, such that it is often difficult to tion with the chamber to be directly injected, visualize the form and extent of the parana- small pressure-release holes were drilled sal air sinus in birds. To address this prob- through the outer table of the bone. In a few lem, a technique was devised to inject a mass cases, the nasolacrimal duct and nasal cavity into the sinuses (and some other cavities) also were injected with contrasting colors. A that could withstand subsequent dissection, small hole was incised in the lateral aspect of clearing and staining, skeletonization, and the antorbital sinus in the region of the exter- total corrosion. The resulting casts allowed nal antorbital fenestra. The latex medium the detailed tabulation of the topographic was injected into the aperture and directed in relationships of the sinuses to surrounding various directions known or suspected to be structures. Polyester resin was attempted as locations of diverticula. When injection was an injection medium, and results were good; completed, the mandible was propped maxi- however, it was rather difficult to work with, mally abducted with modeling clay and the and it was feared that the heat of the curing incision was closed by swabbing it with 10% injection mass might adversely affect those acetic acid which immediately sealed the inci- specimens to be subsequently cleared and sion on contact. When both sides were in- stained. Therefore, latex was selected as an jected, the head was submerged in 10%acetic injection medium. Latex offers the following acid and refrigerated for a couple of hours to advantages: 1) It is water soluble, thus allow- allow the latex to solidify; if the volume of ing the viscosity to be altered easily; 2) the injected latex was high (e.g., in adult Anser injection medium is reasonably stable and anser), the specimen was removed from the can be stored for some time after mixing; 3) it acid and stored overnight in a refrigerator. sets immediately under acidic conditions; 4) The specimen usually was dissected at this it is relatively safe to the user; 5) cleanup is point. If a partial corrosion cast (i.e., preserv- 280 L.M. WITMER ing the skeleton) was desired, then the speci- seven bones (Fig. 7): premaxilla, nasal, max- men was skeletonized using routine proce- illa, lacrimal, jugal, palatine, and vomer. The dures (Witmer, '92). In a few cases, complete frontal and mesethmoid (an ossification of corrosion casts were obtained by immersing the interorbital septum) also encroach on the the entire head in 88%formic acid, yielding a region. The naris is caudally situated (due to sinus cast free of any other tissues in 2-4 the large premaxilla) and is surrounded by days. The casts faithfully reproduced the form premaxilla and nasal and floored partially by of the sinuses, preserving the tunnels and the maxillary bone. The osseous portion of grooves through which nerves, vessels, and the nasal cavity is roofed by the premaxilla, ligaments passed and even the striations of nasal, and dorsal lamina of the mesethmoid. the adjacent muscles. Its lateral wall is largely open in dried skulls, Injecting embryonic material for subse- but has partial walls formed rostrally by recip- quent clearing and staining employed the rocal (subnarial) processes of the maxilla and same basic technique but required a great nasal and the palatal process of the maxilla deal more care, most of the procedure taking (maxillopalatine) and caudally by the lacri- place under the dissecting microscope. Fresh mal. The nasal cavity is partially floored by material again gave the best results. It was the palatine, vomer, premaxilla, and palatal imperative that injection preceded enucle- process of the maxilla. The choana is bounded ation in that removal of the eyeball could by the vomer medially, palatine caudally and have damaged the antorbital sinus or its sub- laterally, and maxilla rostrally and some- orbital diverticulum. After injection was com- times medially; it opens caudally from the plete, the embryo was submerged in 10% nasal cavity into the oral cavity via a very acetic acid for 30-60 minutes, followed by short nasopharyngeal duct. enucleation, evisceration, debraining, fixing, The antorbital cavity is defined simply as etc. (see above). the space rostral to the orbit, external to the FACIAL ANATOMY IN BIRDS cartilaginous nasal capsule, and internal to AND CROCODILIANS the outer surface of the snout. Although The following description of adult facial rarely recognized as such, the antorbital cav- anatomy in extant archosaurs is intended to ity is an important cephalic compartment provide background for discussion of facial comparable to other compartments (e.g., the homologies but not to comprehensively de- orbit, adductor chamber, etc.) in having bony pict the complexity of this anatomy or its boundaries and a variety of contents. In birds, taxonomic diversity. More detailed anatomi- it is an irregular space bounded principally cal reviews are relatively plentiful for birds, by the maxilla rostrally, lacrimal and ecteth- although strongly biased toward domestic moid (when ossified) caudally, nasal dorsally, species (see Stresemann, '27-'34; Getty, '75; jugal arch ventrolaterally, and usually pala- Nickel et al., '77; Baumel et al., '79, '93; King tine ventromedially. Among the contents of and McLelland, '84; Zusi, '931, but few such the antorbital cavity are the antorbital para- works are available for crocodilians (seeWett- nasal air sinus, the nasolacrimal duct, some- stein, '37-'54). The descriptions here are times part of the jaw musculature, and the based primarily on original dissections of the nasal gland andlor its ducts (see below). Sev- major study taxa (Table 1) and focus on eral of the bones surrounding the antorbital anatomy rostral to the orbit (with some excep- cavity often are pneumatized by the antor- tions). Avian (and, as much as possible, croco- bital sinus: the palatal process and body of dilian) osteological terminology follows the maxilla, lacrimal, and less commonly the Baumel and Witmer ('93) and Witmer ('94), palatine (Fig. 7). The lateral aperture of the and other terminology follows Baumel et al. antorbital cavity, the external antorbital fe- ('93).It is assumed (but not rigorously tested) nestra, is closed by skin and is bounded by that the descriptions apply to the common the reciprocal processes of the maxilla and ancestor of each most-inclusive, monophy- nasal, lacrimal, and jugal arch. The medial letic, extant clade (i.e., Neornithes and Croco- aperture of the antorbital cavity, the internal dylia). antorbital fenestra, opens into the nasal cavity and is bounded by the palatal process of Birds the maxilla and sometimes the rostral edge of Bones the lacrimal (although its osseous borders The facial skeleton (including the rostral are usually difficult to define). The antorbital portion of the palate) of most birds includes cavity opens caudally into the orbit via the FACIAL HOMOLOGIES IN ARCHOSAURS 281

a o cav

Fig. 7. Anser anser. Rostral portion of the skull in left lateral view (top) and ventral view (bottom).Modified after Kom6rek (’79) and specimens. postnasal fenestra (Witmer, ’94, in press), orbit, where it opens via dual puncta, to the which is bounded by the lacrimal laterally, nasal cavity. It passes just lateral to the lacri- mesethmoid medially, palatine ventromedi- mal, which is usually faintly grooved for the ally, and ectethmoid dorsally (when ossified). duct. The duct itself passes rostrodorsally The postnasal fenestra should not be con- through the external antorbital fenestra, fused with the orbitonasal fissure and fora- crossing dorsally over the antorbital air sinus men which are openings generally dorsome- and aditus conchae and then turning ven- dial to the ectethmoid (or cartilago lamina trally medial to the sinus, perforating the orbitonasalis) €or passage of nerves and the capsule to open into the choanal region of the nasal gland duct(s) (Crompton, ’53; Baumel nasal cavity just ventral to the middle nasal and Witmer, ’93). concha and caudal to the crista nasalis (the “Schwelle” of the older literature). Bremer Nasolacrimal duct (’40) regarded the avian nasolacrimal duct as The form and course of the nasolacrimal an air sac, comparable to the antorbital sinus duct is very consistent in birds (Fig. 8). It is a (his “subocular sac”), but the only basis for thin-walled, spacious tube running from the this assertion is the large volume of the duct. 282 L.M. WITMER ad co nas gl

acc

".-' \ mand

Fig. 8. Anser anser. Dissection of the antorbital cav- have been removed from the coronoid process of the ity and adjacent regions in left lateral view. The antor- mandible. The jugal bar has been cut and the middle bital sinus (i.e., the epithelial air sac) is opened to view section removed. Arrows indicate the openings into the the internal structures. The external adductor muscles nasolacrimal duct. Scale bar = 1 cm.

As in other amniotes, it develops as a solid medial to the nasolacrimal duct, and caudo- epithelial cord that later cavitates, rather dorsal to the antorbital air sinus; it may than as an epithelial evagination of the nasal project a short distance into the antorbital cavity, the latter being a characteristic of sinus (e.g., Fregata magnificens; Bang, '71) pneumatic diverticula. or even into the cavum conchae (Gallusgal- lus and Rhea americana; see also Sandoval, Nasal gland '63; Muller, '61). If it has a more caudal Glandula nasalis occupies a number of po- position, the ducts pass into the antorbital sitions in birds (see Technau, '361, ranging cavity along with ramus lateralis nasi of the from being located completely preorbitally ophthalmic nerve via the lateral orbitonasal (as in some gruiforms and pelecaniforms) to foramen. In all cases, the ducts travel along extending far caudally in a supraorbital posi- with ramus lateralis nasi in the caudal portion (many marine birds; see also Fig. 8). The tion of the antorbital cavity, then pass rostro- usually paired ducts always open into the ventrally medial to (and sometimes grooving) caudal portion of the nasal vestibule (Bang, the reciprocal subnarial processes of the na- '71). When the gland is located preorbitally, sal and maxilla and lateral to the capsular the body of the gland is situated dorsally wall; the lateral duct enters the vestibule within the antorbital cavity, just ventral to directly, whereas the medial duct passes the mesethmoid and nasal, medial to the transversely through the crista nasalis to lacrimal, lateral to the nasal capsule, dorso- open medially into the vestibule adjacent to FACIAL. HOMOLOGIES IN ARCHOSAURS 283

"om'(cut) ch

Fig. 9. Anus platyrhynchos. Rostral half of the right side of a sagittally sectioned head in medial view, showing the parts of the nasal cavity, the nasal conchae, and the antorbital sinus ostium. Hatching denotes cut bone surfaces. Scale bar = 1 cm. the nasal septum (Marples, '32; Bang, '71; of the maxillary nerve travels through the Vorster, '89). ventrolateral portion of the orbit ventral to both the eyeball and the suborbital diverticu- Nerves lum of the antorbital sinus and dorsal to m. The nerves innervating the face are very pterygoideus, pars dorsalis (Fig. 8). The naso- consistent (Figs. 8-10] and are composed pri- palatine branch of the maxillary nerve car- marily of sensory branches of the trigeminal ries fibers from the nasal capsule, palate, and nerve (autonomics will not be discussed; see maxillary bone. Webb, '57; Bubien-Waluszewska, '81). The ophthalmic division of the trigeminal nerve Muscles (CN V,) passes through the orbit between As is true of all sauropsids, birds lack any the interorbital septum and eyeball and di- muscles attaching superficially to the facial vides into two major branches upon reaching skeleton (their "facial musculature" being the nasal capsule. The larger, medial branch, restricted only to those innervated by the ramus medialis nasi, is mostly intracapsular, seventh cranial nerve, viz. m. depressor man- entering the nasal capsule through the me- dibulae and m. columellae). However, one of dial orbitonasal foramen; it tends to run along the jaw adductors, m. pterygoideus, pars dor- the septum, usually dividing rostrally into salis, has been involved in the debate on the dorsal and ventral premaxillary nerves. The function of the antorbital cavity. Although smaller, lateral branch, ramus lateralis nasi, most avian anatomists (e.g., Hofer, '50;Sims, is entirely extracapsular, entering the antor- '55; Fisher and Goodman, '55; Goodman and bital cavity through the lateral orbitonasal Fisher, '62; Merz, '63; George and Berger, foramen along with the nasal gland ducts '66;Owre, '67; Richards and Bock, '73; Bhat- and rarely (e.g.,Passer domesticus and Melop- tacharyya, '82, '89) have found m. pterygoi- sittacus undulatus; Lang, '55) passing deus divided into pars dorsalis and pars ven- through a small foramen epiphaniale within tralis, in some cases the lateral portions of the cartilaginous capsule; it always passes pars dorsalis and pars ventralis are fused, just lateral or dorsal to the aditus conchae as suggesting mediolateral rather than dorso- it hugs the capsular wall. The maxillary divi- ventral division of the pterygoideus mass sion of the trigeminal nerve (CN V,) is rela- (Zusi, '62; Zusi and Storer, '69; Burton, '74; tively smaller than in most vertebrates, al- Elzanowski, '87). In a few taxa the dorsal most certainly owing to reduction of the pterygoideusis reduced or even absent (Webb, maxilla and loss of the teeth. Its supraorbital '57; Bhattacharyya, '82, '89; Elzanowski, '87). branch will be ignored here. The main branch In most birds, however, the dorsal pterygoi- 284 L.M. WITMER

mand a o sin

Ram tat nas

tooth cavico sin

Fig. 10. A Anser anser (in caudal view). B: Crocody- antorbital (caviconchal)sinus ostium into antorbital (cavi- lus porosus (in rostra1 view). Transverse sections of adult conchal) sinus, and into aditus and cavum conchae. Hatch- heads at the level of the aditus and cavum conchae. In A ing denotes cut bone surfaces. (and B), arrow passes from nasal cavity proper, through FACIAL HOMOLOGIES IN ARCHOSAURS 285 deus originates from the dorsolateral surface ’61). Some of these elements develop from of the palatine and pterygoid and extends somewhat independent anlage, which will be caudoventrally as a broad sheet to insert on discussed further later. The tectum nasi and the mandible (Figs. 8, 1OA). The muscle of- paries nasi together form a ventrally open ten reaches into the caudoventral portion of box, with the box closed caudally by the more the antorbital cavity where it contacts the or less transversely situated lamina orbitona- antorbital sinus. Within the orbit, the muscle salis; sometimes there is a rostral cupola is always just ventral to the maxillary nerve closing the box rostrally. The paries nasi and the suborbital diverticulum of the antor- occludes much of the internal antorbital fe- bital sinus (Figs. 8,lOA). nestra and forms part of the medial wall of the antorbital cavity. The nasal conchae Nasal cavity project medially from the paries nasi or tec- The nasal cavity and cartilaginous nasal tum (Fig. 9). The rostral concha is located in capsule of birds is very complex and variable the vestibule, whereas the middle and caudal (see Bang, ’71) but usually has the same conchae are located in the respiratory and basic elements. The nasal cavity may be di- olfactory portions of the nasal cavity proper, vided into the three major compartments respectively. The rostral and middle conchae described by Parsons (’59, ’70) for “reptiles” range from simple lamellar projections to (Fig. 9): 1) the vestibule rostrally, 2) nasal highly branched structures, with the most cavity proper (= cavum nasi proprium) cau- common morphology being simple scrolls (see dally, and 3) the nasopharyngeal duct caudo- Bang, ’71). The caudal concha is usually a ventral to the nasal cavity proper. The bound- hollow, bubble-shaped hillock in medial view, ary between the first two is usually taken to and is pneumatized by a diverticulum of the be the region into which the nasal gland antorbital sinus. The cavity within the cau- ducts open (Muller, ’61; Bang and Wenzel, dal concha is called the cavum conchae, and ’85) and which corresponds roughly to the the lateral entrance to the cavum is the adi- position of the crista nasalis. The vestibule is tus conchae (Figs. 8, 10). The caudal concha expanded in birds, probably in association tends to be caudodorsal to the middle concha, with development of a large rostral concha such that the latter extends caudally ventral (see below), and is enclosed within the narial to the former to also reach the lamina orbito- region of the skull, principally the premax- nasalis. The caudal concha has been lost, illa, nasal, maxilla, and in some birds (e.g., apparently independently, in a few clades of ratites) the vomer. The nasal cavity proper is small birds (e.g., swifts, some small passeri- sometimes subdivided into a rostral respira- forms; Engelbrecht, ’58;Bang, ’71). The na- tory or main cavity and a caudal olfactory sal cartilages only occasionally ossify in birds cavity (Matthes, ’34; Bang, ’71; Bang and (see Baumel and Witmer, ’93),with the nasal Wenzel, ’85). The nasal cavity proper extends septum and lamina orbitonasalis (ecteth- to the orbit, and the capsule is supported by moid) ossifying more commonly than the the maxilla, nasal, palatine, vomer, lacrimal, other elements. and mesethmoid. As mentioned earlier, the nasopharyngeal duct is usually very short in Air sacs birds, although in some birds (e.g., Diomedea Birds have a single major air-filled, epithe- immutabilis) caudoventral expansion of the lial diverticulum of the nasal cavity, the ant- crista ventralis of the palatine, forming a orbital sinus (Fig. ll),which itself has sev- choanal fossa, has the effect of elongating the eral subsidiary diverticula (Witmer, ’90, and duct somewhat (Baumel and Witmer, ’93). references therein). The antorbital air sac The nasal capsules in most clades of birds exits the nasal cavity via a small ostium in have the seven major cartilaginous elements the caudal portion of the nasal capsule just noted by Macke (’69), most of which are aptly ventral or caudoventral to the caudal concha named: septum nasi (nasal septum, dividing and just rostral to the lamina orbitonasalis the left and right nasal cavities), tectum nasi (Fig. 9). The ostium is directly opposite and (nasal roof), paries nasi (nasal side wall), usually close to the caudal part of the choana, lamina orbitonasalis ( =planum antorbitale, and in the vicinity of the entrance of the forming the caudal wall of the nasal cavity), maxillary nerve to the antorbital cavity (Fig. and three conchae or “turbinals”-caudal, 10A). The antorbital sinus fills most of the middle, and rostral concha-projecting into antorbital cavity ventral to the nasolacrimal the nasal cavity. Most birds have a very poorly duct and is directly in contact with the skin developed solum nasi (nasal floor; Muller, covering the external antorbital fenestra. The 286 L.M. WITMER itself is dorsally situated at the rostral end of the snout, and is enclosed mostly by the premaxilla and, to a variable extent, the nasal. Both maxilla and premaxilla have well- developed palatal processes forming a secondary palate (Fig. 12). The nasal cavity proper is very long, extending from the rostralmost tip of the skull to the orbit. As in birds, the ostium of the nasal gland ducts (see below) is regarded as the boundary between vestibule and nasal cavity proper. The rostral half to one-quarter of the nasal cavity proper (or Fig. 11. Aquzla chrysaetos. Schematic drawing of the more, depending on the extent of elongation antorbital sinus and its lacrimal, premaxillary, and suborbital diverticula in left lateral view. Modified after Wit- of the snout) is a tube surrounded by the mer (’87). premaxilla and maxilla ventrally and laterally and the nasal dorsally. More caudally, the nasal cavity opens caudolaterally into the antorbital sinus has several subsidiary diver- orbit via the postnasal fenestra; in this re- ticula that typically pneumatize the bones gion, the nasal cavity otherwise is surrounded surrounding the antorbital cavity: the pala- by the vomer and palatine ventrally, maxilla tal process of the maxilla, the body of the laterally, lacrimal, prefrontal, and frontal dor- maxilla (often leading rostrally into large sally, and the prefrontal pillar caudally. pneumatic cells within the premaxilla), the Crocodilians are characterized by a very lacrimal, and less commonly the palatine and long nasopharyngeal duct enclosed by ven- mesethmoid (Fig. 7; see Witmer, ’90). As tral laminae of the palatine and pterygoid, mentioned above, the sinus also sends a diver- which divert the opening of the airway far ticulum into the cavum of the caudal concha, caudally. The terminology for the nasal and entering via the aditus conchae. A final diver- pharyngeal openings of the duct has had a ticulum, only very rarely pneumatizing bone, long, confusing, and often contradictory his- is the suborbital diverticulum which exits the tory, using terms such as “primitive,” “pri- antorbital cavity caudally via the postnasal mary,” “secondary,” and “tertiary” choanae fenestra and expands into the often most (e.g., Born, 1879; Voeltzkow, 1899; Fuchs, voluminous of the cephalic air sacs, situated ’08; Plate, ’24; Wettstein, ’37-’54; Muller, rostral and especially ventral to the eyeball ’67). The rostral end of the nasopharyngeal (Fig. 11; see Bignon, 1889). The suborbital duct opens in the middle to caudal quarter of diverticulum commonly interleaves between the nasal cavity and is regarded here as the the dorsal pterygoideus and external adduc- primary choana, whereas the caudal opening tor muscles, sometimes even reaching the is the secondary choana (Fig. 12B,C). The region of the trigeminal foramen. primary choana is bounded by the vomer medially and caudally, palatine caudally and Crocodilians laterally, and maxilla laterally and rostrally; Bones and nasal cavity the secondary choana is completely within The facial skeleton of extant crocodilians is the pterygoids. composed of nine bony elements: premaxilla, The term “antorbital cavity” is applied nasal, frontal, prefrontal, lacrimal, maxilla, here to crocodilians based on the strictly mor- jugal, palatine, and vomer (Fig. 12). The phological definition given above for birds, snouts of crocodilians are remarkable for be- viz. the space between the orbit, nasal cap- ing both long and dorsoventrally flattened. sule, and surface of the snout. The antorbital Since the nasal cavity is largely enclosed in cavity is usually relatively small in extant bone in crocodilians, the general organiza- crocodilians,although it is sometimes moder- tion of the cavity will be introduced here ately large in Alligator. The antorbital cavity rather than with the description of the carti- is largely within the maxillary bone, with the laginous nasal structures, as was done for rostral portion of the lacrimal forming its birds. Parson’s (’70)tripartite division of the roof caudally and the palatine part of its nasal cavity works well for crocodilians. floor. There is no external antorbital fenes- The vestibule is small and restricted to the tra. Most of the cavity is occupied by the narial region, forming a short vertical tube caviconchal sinus, an epithelial paranasal air leading ventrally from the nostril. The naris sac (see below). The bony caviconchal recess FACIAL HOMOLOGIES IN ARCHOSAURS 287

\ sep SUIC’ vom

suborb fei sec\ ch

Do vest rec ao cavico rec I

,- ace cav I- n p du

Fig, 12. A Alligator mississippiensis skull in dorsal E: Crocodylus porosus skull, sagittally sectioned and in view. B: Same in ventral view. C: Same, horizontally see- medial view (rostral to left). Hatching denotes cut bone tioned snout with roof removed in dorsal view. D: same, surfaces. Scale bars in D and E = 2 em and 1cm, respec- sagittally sectioned and in medial view (rostral to left). tively. A and B after Wettstein (’37-’54) and specimens. opens caudomedially into the antorbital cav- 12C-E). This caviconchal aperture is always ity via a large aperture in the maxilla that is directly opposite the primary choana and cau- bordered dorsally by the rostral tip of the dal to the rostral opening of the nasolacrimal lacrimal and ventrally by the palatine (Fig. duct. The bony caviconchal recess often has 288 L.M. WITMER been referred to as the “maxillary sinus” (Wegner, ’58). Another bony cavity associ- (Gegenbaur, 1873; Meek, ’06; Nemours, ’30; ated with the nasal cavity proper is the “cau- Wegner, ’58; Ferguson, ’81; among others), dolateral recess” (Bertau, ’35) within the but this term has been applied also to a palatine bone just lateral to the nasopharyn- separate, rostral cavity in Alligator mississip- geal duct and ventral to the postconcha (Fig. piensis (Bertau, ’35; Parsons, ’70); it seems 12C). The caudolateral recess varies from best to avoid confusion by abandoning the being a discrete foramen that opens into a term “maxillary sinus” altogether for croco- chamber (e.g., Melanosuchus niger IWegner, dilians. The space within the antorbital cav- ’581, some Alligator mississippiensis) to one ity caudomedial to the caviconchal aperture or more deeply excavated fossae (e.g., some is small but important because several struc- Alligator and Caiman crocodilus). This re- tures (e.g., the nasolacrimal duct, sometimes cess was absent in my material of Crocody- the nasal gland) pass through it (see below). lus, but Wegner (’58) figured an apparent There are usually various pneumatic cells caudolateral recess in a large individual of C. (“accessory cavities”) associated with the niloticus. Finally, the prefrontal sinus is ap- caviconchal recess of the antorbital cavity, parently restricted to Alligator, where it the most consistent one being an extensive forms a large cavity within the prefrontal medial recess within the palatal process of communicating with the nasal cavity proper the maxilla (Fig. 12D). via a medial foramen within the prefrontal In addition to the caviconchal recess, there pillar just adjacent to the nasal septum and usually are other openings leading from the caudal to the postconcha (Fig. 12C,D).Weg- nasal cavity into pneumatic cavities within ner (’58)described for several species a “pre- the facial bones. In fact, crocodilians as a maxillary sinus;” however, this is almost cer- whole exhibit a greater diversity of such cavi- tainly the cavity for the enlarged premaxillary ties than perhaps any other group except vascular space (associated with the narial perhaps mammals (see Paulli, 1900; Dieu- cavernous tissue; Bellairs and Shute, ’53) laf6, ’05). Wegner (’58) provided a detailed and not a pneumatic feature at all. Similarly, description of most of these; his terminology, Iordansky (’73) labeled as “accessory air cav- however, is generally too cumbersome, as he ity” the non-pneumatic neurovascular fora- applied names to many structures that have men at the premaxillomaxillary suture. extreme intraspecific variability. There are There are also six consistent major re- five major recesses associated with the nasal cesses associated with the nasopharyngeal cavity proper, although no species has all of duct: 1) the vomerine bulla, 2) the palatine them: 1) the caviconchal recess (discussed sinus, 3) the pterygopalatine sinus, 4) the above), 2) the postvestibular recess, 3) maxil- palatine bulla, 5) the pterygoid bulla, and 6) lary cecal recesses, 4) the caudolateral recess the pterygoid sinus. Vomerine recesses or of the palatine, and 5) the prefrontal recess. bullae occur apparently only in Melanosu- As “maxillary sinus” is unacceptable for the chus niger (Howes, 1891; Wegner, ’58) and reasons noted above, the term “postvestibu- Alligator mississippiensis; they are of funda- lar recess and sinus” has been proposed for mentally different construction in the two the bony cavity and epithelial diverticulum, taxa (and of questionable homology), the respectively (Witmer, ’94, in press). The former being rostral to the primary choana postvestibular recess is a very common fea- and the latter being caudal to it. The palatine ture in Alligator, being a relatively small sinus is an often large cavity within the pala- cavity completely within the maxilla lateral tine bone opening medially into the nasopha- to the nasal cavity proper and rostral to the ryngeal duct just rostral or ventral to the caviconchal recess, and opening into the na- prefrontal pillar (Fig. 12D). Although in a sal cavity via a medial foramen opposite the few large Alligator skulls a communication second or third maxillary tooth (Fig. 12C,D). between palatine sinus and caudolateral re- In some large alligators, the postvestibular cess could be demonstrated, they were usu- and caviconchal recesses broadly communi- ally separated by a thin lamina; unfortu- cate, but usually they are separated by a nately, Wegner (’58) rarely differentiated transverse bony septum opposite the largest between the two cavities within the palatine (fourth) maxillary tooth. Crocodylus spp. of- and applied the same name to them. Within ten has numerous, similar lateral evagina- the orbit of many crocodilians, there is a tions of the nasal cavity into the maxilla, but marked inflation of the nasopharyngeal duct usually they are very short and end blindly- (both pterygoid and palatine) called the ptery- hence their name, maxillary cecal recesses gopalatine bulla (Mertens, ’43;Wegner, ’58). FACIAL HOMOLOGIES IN ARCHOSAURS 289 In A. sinensis only the palatine is involved, orbit to the nasal cavity. After exiting the forming a bulbous palatine bulla, whereas in lacrimal rostrally, the duct passes dorsomedi- Gavialis gangeticus only the pterygoid is in- ally through the antorbital cavity where it volved, forming an often enormous pterygoid lies against the caviconchal sinus ventrolater- bulla (Mertens, ’43;Wegner, ’58; Martin and ally and nasal capsule ventromedially; in the Bellairs, ’77). Finally, best-developed in Alli- more rostral portion, the nasal gland is inter- gator spp. (Wegner, ’58; Norell, ’891, the posed between the capsule and duct (Fig. 13). pterygoid sinus is an often multichambered The duct opens medially into the nasal cavity cavity dorsal to the nasopharyngeal duct, ven- just ventral to the preconcha, usually extend- tral to the basisphenoid, and communicating ing rostrally a short distance beyond its nasal with the airway in the vicinity of the second- ostium as the “Saccus nasolacrimalis” of ear- ary choana. lier authors (e.g., Rathke, 1866; Shiino, ’14; Bertau, ’35). Unlike other vertebrates, the Nasolacrimal duct epithelium of the nasolacrimal duct of extant The nasolacrimal duct in crocodilians crocodilians is greatly hypertrophied and passes through the lacrimal bone from the formed into tubular crypts such that the

-roof nas cap

-Ram lat nas

--as gl

pter dor postco

Fig. 13. Alligator mississippiensis. Horizontally sec- the postconcha, and the course of some branches of the tioned head in dorsal view, showing the paranasal air trigeminal nerve (CN V). Hatching denotes cut bone sinuses, nasolacrimal duct, nasal gland, parts of the nasal surfaces. Sale bar = 1 cm. capsule, the attachment of the dorsal pterygoideus onto 290 L.M. WITMER whole structure may be termed the nasolacri- maxillary nerve (CN V,) is larger, passing ma1 gland (Saint-Girons, '76); the duct runs dorsally over the dorsal pterygoideus muscle through the middle of the gland and the on its way through the orbit (Figs. 10A, 13, crypts open into the duct (Fig. 17B). The 14), and giving off here n. alveolaris dorsalis nasolacrimal gland contacts the maxilla ros- caudalis (Poglayen-Neuwall, '531, which trally, and in Crocodylus spp. (but not Alliga- passes through a foramen within the maxilla tor mississippiensis, Caiman crocodilus or (erroneously labeled as a pneumatic foramen the examined juvenile Tomistoma schlegelii by Iordansky, '73, Fig. 14).The rostral dorsal and Gavialis gangeticus) the gland (in par- alveolar branch of the maxillary nerve contin- ticular, the saccus) is lodged within a deep ues rostrally into the antorbital cavity, enter- medial cavity within the maxilla. ing a neurovascular foramen just lateral or dorsolateral to the aperture of the cavicon- Nasal gland chal recess (arrows in Fig. 12D,E). In Alliga- The nasal gland of crocodilians extends for tor and Caiman, the nerve and accompany- much of the length of the snout. In adults, ing vessels (the internal maxillary vessels; Hochstetter, '06) pass through the cavicon- the gland is a generally large, vascularized chal recess of the antorbital cavity lateral and structure running from the region dorsal to adjacent to the caviconchal air sinus, enter- the postconcha (in at least Alligator missis- ing a bony canal that courses dorsolateral to sippiensis) rostrally up to the caudal margin the postvestibular recess on its way to the of the narial region (i.e., the nasal vestibule) where its ducts empty (Fig. 13). Although premaxillary vascular space; in Crocodylus Rose (1893), Reese ('24), Plate ('24), and spp.,. the neurovasculature runs entirely Bertau ('35) reported that the gland has lim- within a bony canal. ited caudal extent (e.g., to about the level of Muscles the preconcha; see also Parsons, '701, they As in birds, the dorsal portion of the ptery- studied only embryonic or young animals, goideus muscle is the only muscle encroach- and in adults the caudal portion of the gland ing on the facial region (the narial muscles usually just reaches the antorbital cavity. may be ignored here; see Bellairs and Shute, The nasal gland is situated medial to the '53). Busbey ('89) noted that in Alligator nasolacrimal gland (and duct) and dorsal to mississippiensis the dorsal pterygoideus is the nasal capsule (Figs. 13, 17B). Ramus the longest and third largest (by volume) of lateralis nasi of the ophthalmic nerve passes all the jaw muscles. Only the rostral attach- through the substance of the gland (Bellairs ments of the muscle are described (see Schu- and Shute, '53). The gland generally runs macher, '73, for the caudal attachments). along the nasomaxillary suture and may The dorsal pterygoideus is a large mass pass- faintly groove the nasal (see Fig. 12E and ing dorsally over the pterygoid and ectoptery- Witmer, in press). goid bones and suborbital fenestra, ventral to the eyeball, and through the postnasal fenes- Nerves tra to terminate in the caudolateral portion The basic organization of the nerves closely of the antorbital cavity (Figs. 13,14). Within resembles that of birds (Figs. 10B, 13, 14, the orbit, it attaches to or is in contact with 17B). The ophthalmic nerve (CN V,) passes the internal surfaces of the pterygoid, through the orbit between the eyeball and ectopterygoid, jugal, maxilla, prefrontal pil- septum and then medial to the prefrontal to lar, the laminae of the pterygoid and palatine enter the facial region where it splits into bones roofing the nasopharyngeal duct, and medial and lateral branches. Ramus medialis the interorbital septum. Rostra1 to the post- nasi enters the cartilaginous nasal capsule nasal fenestra, the snout becomes flattened and travels between the nasal septum and and the muscle fills the caudolateral part of adjacent mucosa. Ramus lateralis nasi re- the antorbital cavity. Again, throughout its mains extracapsular, in Alligator mississippi- course, the maxillary nerve is just dorsal to ensis and Caiman crocodilus passing through the muscle. Within the antorbital cavity, the a foramen epiphaniale (Fig. 15A)just dorsal muscle attaches broadly to the caudolateral to the aditus conchae (see also Bertau, '35; surface of the bubble-shaped postconcha and Bellairs and Shute, '53; Klembara, '91); ros- to the palatine bone lateral to the dorsal tral to the epiphanial foramen it travels with ridge or crest that supports the postconcha. the nasal gland between the nasal capsule Dorsally, the muscle attaches to the lacrimal and nasal bone (Figs. 10A, 13, 17B). The bone lateral to an oblique ridge (the postcon- FACIAL HOMOLOGIES IN ARCHOSAURS 291

max N Fig. 14. AZligutor mississippiensis. Snout with portions resected to show the position of the dorsal pterygoideus muscle and the course of branches of the trigeminal nerve (CN V). Modified after Schumacher ('73) and specimens. Hatching denotes cut bone surfaces. Scale bar = 10 em.

cha is medial to the ridge) and also to the maxilla and jugal. Rostrally, the muscle tapers to a point where it attaches to the maxilla just lateral to the ostium of the caviconchal air sinus. In crocodylids and Gavialis gangeti- A cus, the muscle stops at the maxillary neurovascular foramen, whereas in Alligator and Caiman crocodilus the muscle enters the ventral part of the foramen and passes rostrally a short distance. In the latter case, the muscle is directly adjacent to the epithelial air sac of the caviconchal sinus.

Nasal capsule orb Crocodilians have eight major components in their cartilaginous nasal capsule, bearing, with the exception of the conchae, the same egg/-7-i-i- tooth ad co general names and relationships as those described previously for birds: septum nasi, tectum nasi, paries nasi, solum nasi, lamina orbitonasalis, postconcha, concha, and preconcha. These elements form an extensive capsule (Fig. 15) with major apertures 1) rostrolaterally, the fenestra narina, 2) caudoventrally, the fenestra basalis (through which pass the nasopharyngeal duct, caviconchal sinus, and the palatine ramus of the facial nerve), and 3) caudodorsally, the fenestra cribrosa (through which pass the olfactory nerves [CN I], ramus medialis nasi, and vas- culature: Meek, '11; Shiino, '14;Klembara, '91). The palatal process and medial lamina Pmx max of the ascending ramus of the maxilla are adjacent to the solum nasi and much of the Fig. 15. Alligator mississippiensis. A Rostral part of paries nasi, respectively. The paries nasi par- the chondrocranium of a 34-day embryo in left lateral view. B. Same with bones in place, showing the fonticu- tially occludes the aperture of the cavicon- lus antorbitalis. The arrow in A passes through the chal recess. The caudal portion of the paries foramen epiphaniale. Drawn with camera lucida from a nasi (i.e., the external surface of the postcon- cleared-and-stainedspecimen. Scale bar = 2 mm. 292 L.M. WITMER cha) fills most of the postnasal fenestra and is (and the young Tomistoma schlegelii speci- part of the origin of the dorsal pterygoideus men), the cavum conchae is much smaller muscle (see above and Fig. 13). The tectum and not pneumatic in adults (Figs. 16, 17). nasi is adjacent to the nasal, frontal, prefron- The nasolacrimal gland (and duct) passes tal, and lacrimal. The nasal septum is lodged just lateral to the concha (Fig. 17B). The ventrally in a median sulcus between the postconcha is a large, egg-shaped structure premaxillae, maxillae, vomers, and ptery- located well caudal to the primary choana goids (Fig. 12C).The fenestra basalis is largely and supported by the palatine ventrally, pre- coextensive with the primary choana. frontal pillar caudally and dorsally, and lacri- The nasal conchae project medially from mal dorsally; as mentioned, it bulges caudola- the paries nasi and tedum (Fig. 16). The terally through the postnasal fenestra (Figs. preconcha is a low swelling with its greatest 13,15-17). The postconchais hollow, commu- dimension caudally and diminishing ros- nicating rostrally with the nasal cavity proper trally. The nasal gland, saccus nasolacrima- via a cartilaginous tube ventrolaterally within lis, ramus lateralis nasi, and accompanying the postconcha (the post-turbinal sinus of vessels partially fill the preconcha, and the Meek "111 and later authors). There are four, nasolacrimal duct opens just ventral to its consistent, named, intracapsular spaces (i.e., caudal portion. The concha is a large, some- preconchal recess, extraconchal recess, post- what scrolled structure projecting freely into conchal sinus, and post-turbinal sinus; Fig. the cavity and attaching ventrolaterally to 161, but these do not merit further discussion the paries nasi via rostromedial and caudola- here (see Parsons, '70, and references teral limbs in the vicinity of the primary therein). choana; there is usually a fossa in the concha ventrally between the limbs. In Gavialis gan- Air sacs geticus and Crocodylus spp., the concha is The epithelial diverticula of the nasal cav- hollow, with a large cavum conchae filled ity were introduced earlier in the osteological with a diverticulum from the caviconchal air description because there usually is close cor- sinus (hence the name of the latter), entering respondence between the air sac and the bony the cavum via the aditus conchae; in Alliga- cavity. Their relationships to capsular struc- tor mississippiensis and Caiman crocodilus tures are outlined below. The ostium of the

preco

n I du ost

Fig. 16. Crocodylus nouaeguzneue. Right side of sagit- the cavum conchae; the arrow passes through the ostium tally sectioned snout in medial view, showing nasal con- of the caviconchal sinus into the cavum. The vomer is chae, intracapsular recesses, cecal recesses, and nasopha- removed. Hatching denotes cut bone surfaces. Scale bar = ryngeal duct. A window is cut into the concha, revealing 1 cm. FAClAL HOMOLOGIES IN ARCHOSAURS 293

nas

nas Ram for

-r co cavico sin

Ram med nas

/ / postco cav vom max postco CNP

Fig. 17. Alligator mississippiensis. Transverse sec- dorsal position. Note, in particular, the rotation of the tions at the level of the concha of A) a 36-day embryo and ascending ramus of the maxilla, the lacrimal bone, naso- B) an adult. Arrow shows direction of nasal rotation. lacrimal duct, caviconchal sinus, and concha. Hatching Structures that are lateral early in ontogeny rotate into a denotes cut bone surfaces. 294 L.M. WITMER caviconchal sinus is located immediately ven- briefly examine some of the morphogenetic tral or slightly caudoventral to the root of the factors potentially impacting on the apomor- concha just lateral to the primary choana phic aspects of their respective facial mor- (Figs. 12C-E, 16, 17). In fact, the fibrocarti- phologies and contributing to their divergent lage band that anchors the concha to the specialization. The point is to introduce these bony skull attaches within the primary factors and bear them in mind when later choana in at least Alligator mississippiensis trying to tease apart disparity and unity. and forms the rostral margin of the ostium. Reference to extant amniote outgroups (e.g., External to the ostium, the sinus enters the for birds: Crocodylia, Lepidosauria, Testu- antorbital cavity where it passes ventral to dines, Mammalia) confirms that the follow- the nasolacrimal gland (and duct) before tra- ing features are apomorphic. versing the aperture of the caviconchal re- Birds cess (Fig. 13); the ostium is always much smaller than the aperture. As mentioned, the Three factors that play a significant role in caviconchal sinus fills the bony recess in Ga- avian facial conformation are 1)the large size vialis gangeticus and Crocodylus spp., of the eye, 2) the prominence of the nasal whereas in Alligator and Caiman crocodilus vestibule, and 3) the reduction of the maxilla. the maxillary neurovasculature and the tip of The extent to which the last two are corre- the dorsal pterygoideus muscle are also (mi- lated is unknown. The morphogenetic impor- nor) contents of the recess. The postvestibu- tance of the eyeball is well documented for a lar sinus (perhaps Alligator only) evaginates number of vertebrate groups, and it is clearly the lateral wall of the nasal cavity proper often a potent functional matrix (Taylor, ’39; immediately caudal to the vestibule and ex- Malan, ’46; Coulombre and Crelin, ’58; Bel- pands within the maxillary bone (Figs. 12C, lairs and Kamal, ’81; Hanken, ’83). Since D, 13). The maxillary cecal recesses of Croco- birds have inordinately large eyes, even as dylus have the appearance of “aborted” pneu- adults (Walls, ’42;Pumphrey, ’61; Gans, ’88), matic recesses, being very variable in num- and the eyes appear very early in ontogeny ber and position. They often form a row that (Romanoff, ’60),it is not surprising that these starts rostral to the nasal ostium of the naso- structures strongly constrain the develop- lacrimal duct and extends rostrally for most ment of adjacent areas, such as the caudal of the length of the maxilla (Figs. 12E, 16). components of the face. For example, the The caudolateral sinus evaginates the floor of bulging eyeball encroaches on the antorbital the nasal cavity proper ventral to the postcon- cavity, displacing the vertical, orbital process cha, producing an extremely variable pattern of the lacrimal bone rostrally such that more of pneumatic recesses and/or foramina within medial portions of the nasal capsule project the palatine. The prefrontal sinus is a con- caudally beyond it (in other diapsids, the stant feature of adult Alligator, evaginating capsule and lacrimal have a similar caudal the nasal cavity proper at its caudomedial extent). The effects of the eye are not always corner, precisely at the angle where the pre- consistent, however. For example, the lamina frontal bone contacts the nasal septum; the orbitonasalis, which is directly adjacent to cavity expands within the bone, eventually the orbital contents, is normally (probably pneumatizing both the pillar and the dorsal plesiomorphically) transversely oriented, but lamina (Fig. lZC,D). due to compression from the eyeball may be A couple of errors in Parsons (’70) influen- either more sagittally oriented (e.g., Ardeu tial treatment should be noted: 1) He illus- cinerea, Struthio cumelus) or more horizon- trates (his Fig. 30) the caviconchal sinus os- tal (e.g., Rhea americana, Nyctisyrigmuspec- tium as rostral to the concha when it should toralis);in Rhea, the lamina is initially verti- be ventral or caudal to it, and 2) he labels (his cal and transverse, and later in ontogeny Fig. 30) the cavity within the palatal process becomes more horizontal and somewhat com- of the maxilla as communicating with the pressed laterally (Frank, ’54; Miiller, ’61). postvestibular sinus (his “maxillary sinus”) Similarly, in Struthio the eyeballs are truly when it is instead a diverticulum of the cavi- enormous and encroach on the nasal region conchal sinus. to such an extent that the caudal part of the capsule is basically interorbital, having the MAJOR INFLUENCES ON FACIAL MORPHOLOGY effect of rotating the caudal concha so that its IN EXTANT ARCHOSAURS cavum conchae is directed rostrally (rather Having just completed a survey of the fa- than more 1ateraIly) and diverting the ostium cial anatomy of birds and crocodilians, I will of the antorbital sinus ventrally (Frank, ’54; FACIAL HOMOLOGIES IN ARCHOSAURS 295 Lang, ‘55, ’56). The caudal concha is absent growth, but rather by a complex mechanism in some small bird species, and these species referred to here as nasal rotation: The nasal typically have the lowest olfactory ratios (a capsule and surrounding structures rotate rough measure of olfactory capability; Bang, internally such that lateral structures be- ’71)) perhaps leading one to the suspicion come dorsal (Fig. 17). This phenomenon was that those birds with a decreased sense of first noted in a brief statement by Meek (’ll), smell have lost the caudal concha. However, but its significance has not been appreciated. it seems possible that the relatively very large Nasal rotation appears to be unique to eye size in these small birds in effect “crowds crocodilians. It effects virtually all systems, out” the caudal concha, such that absence of so transforming facial anatomy during ontog- the latter is not really an adaptation for de- eny that, although 1:l correspondences are creased olfaction but rather only a conse- preserved for the most part, relational terms quence of large eye size. A similar “telescop- such as “lateral” and “dorsal” cannot be ing” of the caudal portion of the nasal capsule applied to the entire ontogeny of many struc- by the eye has been implicated in loss of the tures. Perhaps the most obvious example is concha in chamaeleonid squamates (Bellairs the nares which are initially situated later- and Kamal, ’81). ally and later “wander” (Wettstein, ’37-’54) Relative to most other sauropsids, the ves- into their dorsal positions, eventually becom- tibular portion of the nasal cavity is large and ing a single opening within the dried skulls of complicated in birds (Fig. 9), although the many crocodilians. Other conspicuous ex- vestibules of many squamates also are com- amples are the lacrimal bone and nasolacri- plex (see Parsons, ’70, and references mal duct, which initially are lateral and even- therein). In birds, the rostral concha is a tually rotate dorsally onto the skull roof (Fig. prominent structure and only rarely is ab- 17). Additional, more subtle, examples will sent (e.g., Sulidae; Bang, ’71). Since the naris be encountered in the next section. is somewhat caudally placed (Fig. 7), the Nasal rotation seems to occur somewhat vestibule by necessity encroaches on the na- at sal cavity proper. Thus, in many birds, the different rates in different parts of the snout. enlarged vestibule has the effect of, again, Rotation of the naris indligator mississippi- telescoping the nasal cavity and compressing ensis, for instance, is virtually complete by more caudal structures (Engelbrecht, ’58). about day 42 of incubation, a time at which This relationship is paralleled in some lepido- the ascending ramus of the maxilla is only saurs, where elaboration of vestibular struc- beginning to turn dorsomedially and the limbs tures is correlated with shortening of the of the concha are only starting to assume nasal capsule (Malan, ’46). their adult positions; the lacrimal and naso- Finally, birds generally have relatively lacrimal duct do not rotate into their final small maxillary bones, having lost the dorsal dorsal positions until between hatching and ramus of the nasal process of the maxilla the first several weeks posthatching. Thus, (i.e., the ascending ramus of other archo- to a first approximation, it seems that nasal saurs) at the phylogenetic level above Ar- rotation begins rostrally, and more caudal chaeopteryx lithographica, as well as teeth at regions lag behind. Interestingly, the bony the level of Neornithes (Cracraft, ’86; Wit- nasopharyngeal duct undergoes relative mer, ’90).In Gallus gallus, for instance, the growth in an opposite direction (i.e., the lat- body of the maxilla is not much more than eral edges of the palatine and pterygoid grow the intersection of its various processes. Al- ventromedially), suggesting that growth pat- though Gallus is extreme, in many birds, the terns are very complex in crocodilians. It is maxilla offers little support to structures in probably worth noting that nasal rotation the facial region, and its neurovasculature is apparently does not characterize all extinct greatly reduced relative to other amniotes. crocodylomorphs, because some forms (e.g., Pelagosaurus typus among many others) re- Crocodilians tain primitive positions for most facial fea- The most obvious aspect of crocodilian fa- tures (e.g., laterally situated lacrimals, etc.), cial morphology is the great length and, espe- although otherwise resembling extant croco- cially, flattening of the snout. Ontogeneti- dilians in many ways. Thus, nasal rotation cally, the effects of elongation are easily may appear at or near the phylogenetic level comprehended. Flattening, however, does not of Neosuchia (Witmer, in press). occur morphogenetically simply by dorsoven- Another uniquely derived feature observed tral compression or differential lateral in extant crocodilians is that the nasal cavity 296 L.M. WITMER proper extends far caudally beyond the ros- determine the most inclusive phylogenetic tral aperture of the nasopharyngeal duct (pri- level at which the features first appear; clearly mary choana), rather than terminating di- many of the features characterize vertebrate rectly dorsal to this aperture as in other taxa at high levels and are symplesiomor- sauropsids. This caudal extension of the nasal phies of archosaurs. cavity proper houses the postconcha (Fig. 16). Bones HOMOLOGIES Most of the facial bones of extant archo- As the facial anatomy of extant birds and saurs appeared much earlier in vertebrate crocodilians and their respective facial apo- phylogeny and have very consistent relation- morphies have been examined in some detail, ships with each other and surrounding struc- it remains to determine which anatomical tures (see de Beer, '37; Romer and Parsons, attributes are homologous. Of special inter- '86). Thus, the homology of the premaxilla, est here are those attributes that figure maxilla, jugal, nasal, palatine, and vomer in prominently in the debate on the soft-tissue birds and crocodilians has never been in dis- relations of the antorbital cavity-i.e., the pute. nasal glands, the dorsal pterygoideus muscles, Whereas crocodilians have both prefrontal and the paranasal air sinuses. Other aspects and lacrimal bones, birds have only a single of facial anatomy will be used as topographic "preorbital" bone that has been regarded as landmarks in assessing the homologies of either a prefrontal or a lacrimal (see Tables 2 these structures. Obviously, these landmarks and 3 in Muller, '63). Determination of the also must be homologous and, as elaborated homology of this element in birds has in- below, most of these are uncontroversial and volved three major issues (see Gaupp, '10; receive only brief treatment, the exception Gregory, '20; Muller, '63; and references being the nasal conchae. The circularity seem- therein) which will be discussed in turn: 1) ingly imposed by using putative homologs to the fact that numerous groups of extant am- assess other putative homologs is addressed niotes have lost either the prefrontal or the by assuming homology rather than homo- lacrimal, and it has not always been clear plasy at the outset (according to the Hennig's which bone has been lost; 2) the topographic "661 auxiliary principle) and testing their relationships of the bones to the nasolacri- congruence with phylogeny under the expec- ma1 duct; and 3) the phylogenetic history of tation that homologies will tend to covary the bones. with other homologies (Brooks and McLen- Among extant amniotes, only some squa- nan, '91). Since homologies are hypotheses, mates and crocodilians have both lacrimals the stepwise process of reciprocal illumina- and prefrontals. The single element in extant tion should reveal those putative homologies mammals has always been called a lacrimal, that in fact are homoplasies (Hennig, '66; and in extant turtles, Sphenodon punctatus, Wiley, '81). Fortunately, in the present case, and many squamates it has been called a the homologies of most of the topographical prefrontal (see Camp, '23; Wettstein, '31- landmarks can be assessed with little or no '37; Bellairs and Boyd, '50; Gaffney, '79; Estes reference to the other features under consid- et al., '88).Based on the distribution in ex- eration. For example, the homology of the tant amniotes and disregarding the tradi- nasolacrimal duct can be assessed without tional names, one would be forced by adher- using evidence from, say, the nasal gland. ence to a parsimony criterion to regard the The basic plan is as outlined above with ancestral amniote condition as having only respect to Patterson's ('82) three tests of one bone, with two bones (i.e-, the addition of homology: 1) similarity or 1:l correspon- a bone) being a homoplasy of crocodilians dence, 2) conjunction, and 3) congruence. and some squamates. This is clearly not the Congruence is tested first using the phylog- traditional interpretation, and more informa- eny of extant amniotes (see Fig. 2; Gauthier tion and other criteria are required (see be- et al., '88a); mammals will not be explicitly low). [It should be noted here that Erdmann treated if lepidosaurs and turtles provide a ('40) identified two ossification centers in his decisive assessment. Congruence with archo- material of Gallus gallus that he identified saur phylogeny (i.e., including fossils) is also as lacrimal and prefrontal. However, Jollie tested here, drawing mostly on the more ('57), May ('611, and Muller ('63) were not extensive treatment provided elsewhere (see convinced by these observations, and more Witmer, in press). No attempt is made to than one ossification center was not observed FACIAL HOMOLOGIES IN ARCHOSAURS 297 in my material of Gallus or any of the other both lacrimals and prefrontals, the inclusion avian species studied here. Extant birds in- of fossil taxa provides a more complete data deed have a single preorbital bone.] set for assessing the congruence of these The topographic relationships of the pre- hypotheses of homology within the context of frontal and/or lacrimal to the nasolacrimal amniote phylogeny. For example, consider- duct have figured prominently in previous ation of synapsid phylogeny reveals that arguments (Gaupp, 'lo; Gregory, '20;de Beer, mammals lack the prefrontal and retain the '37;Jollie, '57; Webb, '57; Muller, '63; Macke, lacrimal, loss of the prefrontal actually hav- '69). In those squamates with both bones, ing occurred at a more inclusive level than the nasolacrimal duct passes between the Mammalia (i.e., Rowe, '88; Gauthier et al., lacrimal and prefrontal (Bellairs and Boyd, %a; Hopson, '91). Similarly, the basal turtle ,501,and in crocodilians it passes through the Proganochelys quenstedti retains both ele- lacrimal bone (and hence lateral to the pre- ments and indicates that higher turtles have frontal). Thus, it is reasoned that because, as lost the lacrimal (Gaf€ney, '90). Finally, the described above, in most extant neornithine basal rhynchocephalian Gephyrosaurus birds the duct passes lateral to the preorbital bridensis retains a tiny lacrimal along with element (Figs. 7, 8), the element must be the its large prefrontal, confirmingthat the single prefrontal (Jollie, '57). In ratites, however, element of sphenodontidans is indeed a pre- most or all of the duct is enclosed within the frontal (Gauthier et al., '88b; Evans, '88; bone (Pycraft, 1900), which is more character- Fraser and Benton, '89). istic of mammalian and crocodilian lacri- Turning to the clade that includes birds, mals. Muller ('63) offered a compromise based Dinosauria also plesiomorphically retains on his study of the ontogeny of the skull of both lacrimal and prefrontal, and it appears the ratite Rhea. He suggested that the me- that birds are the only dinosaurs (indeed the dial portion of the bone could be homologous only archosaurs) to have definitely lost one of to the prefrontal and that the ventrolateral these bones. In general in archosauriforms, portion enclosing the duct could be homolo- the lacrimal is much larger than the prefron- gous to the lacrimal (despite there being only tal and forms a strong vertical strut between one ossification center); in the majority of the orbit and antorbital cavity, contacting other birds, there thus would be no lacrimal ventrally the maxilla and/or jugal in the re- component. Curiously, although Muller ('63) gion of their suture; the prefrontal is a smaller and Macke ('69) regarded the bone as mostly element (except in some crocodylomorphs), or completely homologous to the prefrontal, and braces the lacrimal medially (Fig. 1). In they referred to it as lacrimal. Theropoda, the lacrimal apomorphically is The question then becomes, how trustwor- enlarged further and is exposed dorsally on thy is this topographic relationship? In other the skull roof (Gauthier, '86),such that the words, is it an invariant attribute? Already in prefrontal becomes a relatively small ele- birds, we have seen that the nasolacrimal ment that is wedged in the rostrodorsal por- duct may pass through or lateral to the bone. tion of the orbit and that sends an often In Sphenodon punctatus, the duct passes slender ventral process medial to the lacri- more ventral to the prefrontal bone than mal, Furthermore, in maniraptoran thero- lateral to it (Wettstein, '31-'37; Gabe and pods (the clade of theropods including birds) Saint-Girons, '76; Bellairs and Kamal, '81). the prefrontal is further reduced and per- Squamates are variable, in some cases (e.g., haps even absent in some taxa (Gauthier, Varanus) piercing the lacrimal bone and in '86). Wellnhofer ('74) identified a small pre- other cases (e.g.,the amphisbaenian Monopel- frontal in the basal bird Archaeopteryx litho- tis;Vipera and possibly all other snakes) pierc- graphica, but most later workers have re- ing the prefrontal (Bellairs and Boyd, '50). garded the putative lacrimoprefrontal suture The point here is that the course of the as equivocal (Gauthier, '86;Witmer, '90; Mar- nasolacrimal duct in extant amniotes does tin, '91). Thus, based on the phylogenetic not provide unequivocal evidence for the ho- history of the lacrimal and prefrontal in the mology of the avian preorbital bone. clade leading to birds, there is little doubt The ingredient missing from the previous that the avian preorbital bone is homologous two paragraphs is adequate information to the lacrimal of extant crocodilians and about the phylogenetic history of the struc- other amniotes (indeed probably all Choa- tures involved. Because many extinct (often nata) and that the prefrontal bone has been more basal) members of amniote clades have lost. 298 L.M. WITMER Nasolacrimal duct bone or vice versa. In other words, no re- The ontogeny, morphology, course, and to- course can be made to a "new" duct in birds pographic relationships of the nasolacrimal to explain its unusual topographic relation- duct are very similar not only in extant birds ships to the bone. As mentioned above, al- and crocodilians but also in all other Tetra- though ratites have the duct partly enclosed poda. In all amniotes in which its develop- within the lacrimal, this is almost never the ment has been described, the duct arises as a case in other neornithine birds. Further- solid epithelial cord lateral to the cartilagi- more, in the next outgroup for which we have nous nasal capsule, running in the furrow data, the Cretaceous hesperornithid birds, between the lateral nasal and maxillary pro- the duct passes lateral to the lacrimal in a cesses of the early embryo; it later develops groove as in most non-ratite neornithines its lumen through canalization and gains (Witmer, '90). In Archaeopteryx Zitho- communication rostrally with the nasal cav- graphica, the lacrimal appears to be pierced ity (Born, 1879; Peter, '06; Matthes, '34; by the nasolacrimal duct (see Witmer, in Bellairs and Boyd, '50; Romanoff, '60; Fergu- press). In fact, this situation is found in virtu- son, '85). ally all dinosaurs (see Witmer, in press). Thus, The position of the nasal ostium is some- relative to Archaeopteryx and nonavian dino- what variable. In Lissamphibia and Mamma- saurs, the condition in hesperornithids and lia, the duct enters the nasal cavity in the non-ratite neornithines is derived. Whether region of the fenestra narina, rostral to a ratites represent a reversal or hesperorni- cartilaginous band called the zona annularis thids represent a convergent acquisition of or lamina transversalis rostralis (de Beer, the non-ratite neornithine condition is not '37; Zeller, '891, although many mammals entirely clear. Given that in ratites the lacri- (including humans) have a second, late- mal usually does not completely surround appearing ostium and subsequently close the the duct but is instead open laterally (often rostral ostium (Matthes, '34; Starck, '67). completed by connective tissue; Webb, '57), it Extant turtles lack nasolacrimal ducts (Mi- is possible that the ratite condition involves a halkovics, 1898). In living diapsids, the os- secondary "re-evolution" of the portion of tium of the duct enters the nasal cavity cau- the lacrimal lateral to the duct. dal to the lamina transversalis rostralis in The basal bird Archaeopteryx lithographica the region of the (primary) choana (Fuchs, and nonavian dinosaurs resemble extant '08; de Beer, '37;Bellairs and Boyd, '50). The crocodilians in having the nasolacrimal duct intimate association of the nasolacrimal duct pierce the lacrimal bone, suggesting that this with the vomeronasal organ is a derived fea- softlhard-tissue association is the ancestral ture of squamates (Bellairs and Boyd, '50; archosaurian condition. However, reference Gauthier et al., '88b). In birds and crocodil- to Crurotarsi, the archosaur clade that in- ians, the duct opens directly into the rostral cludes crocodilians (Sereno and Arcucci, '90; portion of the (primary) choana via an os- Sereno, '91; Parrish, ,931, indicates that this tium just ventral to the middle concha and condition arose within the crocodylomorph preconcha, respectively (Figs. 9,161. Crocodil- clade. In the following successively more basal ians are unique in their elaboration of the crurotarsans, the duct passes between the nasolacrimal epithelium into a glandular lacrimal and prefrontal bones (Witmer, in structure (Saint-Girons, '89) and in the dor- press): the basal crocodylomorph Dibothrosu- sal position of most of the duct in adults chus elaphros (Wu and Chatterjee, '931, the (associated with nasal rotation; see Fig. 17B). stagonolepidids Desmatosuchus haplocerus In summary, on the basis of ontogenetic and and Stagonolepis robertsoni (Walker, '61), topographical similarities and the phyloge- probably Ornithosuchus longidens, and the netic distribution of these similarities within parasuchian Phytosaurus sp. The course of extant amniotes, there is no doubt on the the nasolacrimal canal is not known in basal homology of the nasolacrimal duct. Ornithodira (i.e., pterosaurs and basal dino- Under the empirical "checking and re- sauromorphs). Thus, despite passing the simi- checking" (Hennig, '66) associated with the larity test (i.e., the lacrimal transmits the principle of reciprocal illumination, the ho- duct), the condition in extant crocodilians mologies of the lacrimal bone and nasolacri- and dinosaurs is judged as nonhomologous ma1 duct of birds with those of other amni- on the basis of the congruence test. otes confirm that in birds either the course of To summarize, it is difficult to postulate the duct has changed relative to the lacrimal homologous causal associations between the FACIAL HOMOLOGIES IN ARCHOSAURS 299 nasolacrimal duct and its osteological corre- nasal gland, but in extant birds and crocodil- lates in extant birds and crocodilians, be- ians (and squamates as well) there is no cause both taxa exhibit such apomorphic as- medial nasal gland, so the term requires no sociations that little is left in common. Again, modifiers; some turtles and Sphenodonpunc- in probably all birds higher than Archaeop- tutus have medial nasal glands (Saint-Gi- teryx lithographica, the duct passes lateral to rons, ’89).As described above, extant crocodil- the lacrimal, and at the most, grooves the ians and most birds have apomorphic lateral aspect of the bone (except in ratites). positions of the gland. In crocodilians, it ro- In crocodilians, the duct passes through the tates ontogenetically into a dorsal position, lacrimal, the nasolacrimal canal is greatly whereas in many birds the body of the gland enlarged to accommodate glandular tissue, is located within the orbit. and both duct and bone assume a dorsal The osteological correlates of the nasal position due to nasal rotation. Interestingly, gland are very subtle in extant sauropsids although the duct and its course from orbit (with the exception of the supraorbital fossa to choana are homologous in extant archo- in marine birds). In crocodilians there is usu- saurs (see above), there are no homologous ally a faint, often striated depression on the osteological correlates. Fortunately, regard- internal surface of the nasal bone lateral to less of whether the duct pierces the lacrimal the sulcus for the nasal septum and near and (as in dinosaurs and crocodilians) or passes often within the nasomaxillary suture; this between the lacrimal and prefrontal (most groove, which mostly lodges the gland but crurotarsans), this homologous course is indi- also ramus lateralis nasi and accompanying cated in fossil archosaurs and suggests that vessels, is broader caudally and narrows as it the ancestral archosaurian condition is to approaches the vestibular region. In birds, have the duct passing through the dorsome- the nasal gland ducts run on the medial sur- dial portion of the antorbital cavity as in face of the bones in the vicinity of the naso- extant archosaurs (see Witmer, in press). maxillary suture, and in some cases (e.g., Nasal gland Anser anser, Ardea herodias) groove the bones. In those squamates in which the nasal The homology of the nasal glands of tetra- gland is lodged within the cavum conchae pods was discussed by Parsons (’59) and to a (Gabe and Saint-Girons, ’76), there are no lesser extent by Gaupp (18881, Peter (’061, osteological correlates and perhaps only a and Plate (’24), all of whom concluded that faint groove otherwise. Some turtles (e.g., these named glands were probably homolo- Chrysemys picta) show a groove associated gous in all Tetrapoda based on their innerva- with the gland, although here it is along the tion and association with the nasal vestibule. prefrontomaxillary suture (extant turtles This homology may well be true, but there is other than chelids lack nasal bones; Gaffney, a fundamental dichotomy in that the gland is ’79). Thus, based on the distribution of these intracapsular in Lissamphibia and Mamma- bony features in extant sauropsids, a hypoth- lia and extracapsular in Sauropsida. Al- esis may be formulated that the ancestral though this dichotomy may reflect no more archosaurian condition is to have the nasal than the acquisition of an apomorphic posi- gland situated caudodorsolaterallywithin the tion in sauropsids (as argued by Gauthier et snout and possibly grooving the adjacent al., ’88a), I choose to make explicit reference bones in the vicinity of the nasomaxillary only to Sauropsida where the homology of suture. The few fossil archosaurs that can be the gland is more certain. sampled for these features corroborate this In all sauropsids that have been surveyed hypothesis (see Witmer, in press) and hence for this structure, the nasal gland 1) arises support the homology of both the nasal gland early in ontogeny as an evagination of the and its osteological correlates in extant birds nasal epithelium near the caudal margin of and crocodilians. the nasal vestibule, 2) grows caudally, assuming a position lateral to the cartilaginous Nerves nasal capsule, and 3) is innervated by autonomics traveling with the ramus lateralis The homology among vertebrates of the nasi of the ophthalmic nerve (Born, 1883; divisions of the trigeminal nerve and their Gaupp, 1888; Rose, 1893; Marples, ’32; major branches is virtually certain (de Beer, Hoppe, ’34; Bertau, ’35; Technau, ’36; Par- ’37; Romer and Parsons, ’86), and requires sons, ’59). It should be noted that this gland no further justification here. In fact, their is often called the “external” or “lateral” highly conserved topographic relationships 300 L.M. WITMER have long been used as landmarks in studies for turtles, Adams ('19) and Lakjer ('26) dis- such as this one. cuss only a single pterygoideus muscle that The general course of the nerves described occasionally reaches rostrally onto the dorsal above for birds and crocodilians pertains to surface of the palatal bones; Schumacher other amniotes as well (Gaupp, 1888; Wet- ('73) regarded this portion as a separate pars tstein, '31-'37; Lakjer, '26; Oelrich, '56; dorsalis. The muscle in Sphenodon puncta- Getty, '75). The ophthalmic nerve travels tus is generally similar, extending onto the through the dorsomedial portion of the orbit dorsal surface of the palate under the eyeball before dividing rostrally into medial and lat- for a short distance, but most workers have eral nasal rami near the juncture of the nasal not regarded the muscle as having separate and orbital cavities. The maxillary nerve trav- parts (Adams, '19; Lakjer, '26; Wettstein, els ventrally or ventrolaterally through the '31-'37; Lubosch, '33). The pterygoideus floor of the orbit dorsal to the pterygoideus muscle in Squamata is usually described as musculature (or its presumed partial mam- undivided or only occasionally divided into malian homolog, m. tensor veli palatini; Ad- superficial and deep portions. Thus, only in ams, '19; Lakjer, '26; Barghusen, '86) before extant birds and crocodilians is there a well- entering the maxillary bone. The only consis- demarcated division into a pars dorsalis and tent osteological correlates of the ophthalmic pars ventralis (Lakjer, '26; Lubosch, '33). In nerve in extant archosaurs are foramina fact, Lubosch ('33) argued that the pars dor- within the premaxilla transmitting nerves salis in birds and crocodilians is actually a carrying sensory information from the integu- portion of the m. pseudotemporalis (another ment; it could not be determined for other portion of m. adductor mandibulae internus) amniotes if this territory was covered consis- that has become separated from the main tently by the ophthalmic nerve or the maxil- mass and that this separation is absent in lary nerve. other sauropsids. Whether or not Lubosch's The maxillary nerve itself has almost con- ('33) idea is correct, it is clear that the pres- stant osteological correlates among extant ence of a more or less separate pars dorsalis amniotes in the form of foramina and canals within the maxilla (and often jugal, as well) and pars ventralis is a synapomorphy of ex- innervating dental and integumentary struc- tant archosaurs. tures. These foramina provide important As just demonstrated, the dorsal pterygoi- clues to the course of the homologous nerve deus muscle of extant birds and crocodilians in all amniotes, both fossil and extant. passes both the similarity test and the test of congruence among extant sauropsids. How- Muscles ever, unlike the situation with structures As described earlier, in both extant birds such as the nasolacrimal duct and nasal gland and crocodilians one of the adductor muscles that characterize extant vertebrate taxa at reaches into the facial region. In both cases it much higher levels, the dorsal pterygoideus has been referred to as m. pterygoideus, pars is restricted to Archosauria. As a result, rigor- dorsalis (although it has had a variety of ous testing for congruence with archosaur synonyms: m. adductor mandibulae internus phylogeny becomes more important. The os- pterygoideus anterior, m. pterygoideus inter- teological correlates of the dorsal pterygoi- nus, m. pterygoideus anterior, pterygoideus deus are relatively few because of the diver- D, etc.). The homology of the various jaw gent specialization of extant birds and muscles among sauropsids and other verte- crocodilians, but an excavation on the dorso- brates has been the subject of numerous lateral surface of the palatine near or within monographic treatments, and the evidence the caudoventral portion of the antorbital does not need to be reviewed again here (see cavity remains as a reliable indicator of the Adams, '19; Lakjer, '26; Lubosch, '33; Edge- presence of the muscle. As elaborated else- worth, '35; Kesteven, '45). These workers all where (Witmer, in press), in at least the basal agreed that the dorsal pterygoideus of extant members of all major clades of archosaurs birds and crocodilians is homologous, citing there is a well-developed palatine excavation evidence primarily from the ontogeny, topo- extending well within the ventral portion of graphic relationships, andlor pattern of in- the antorbital cavity. Given the causal asso- nervation of the muscles. ciation of the bony excavation with the dorsal There is disagreement, however, as to the pterygoideus in extant archosaurs, the simi- extent to which the pterygoideus mass is lar structure in the fossil taxa is almost cer- subdivided in other sauropsids. For example, tainly a muscular fossa. Thus, the hypothesis FACIAL HOMOLOGIES IN ARCHOSAURS 301 of homology is congruent not only with ex- birds (Plate, '24; Matthes, '34; Bellairs and tant amniote phylogeny but also archosaur Boyd, '50; Parsons, '59). No one has sug- phylogeny, and the homology of m. pterygoi- gested that the nasopharyngeal ducts of am- deus, pars dorsalis in extant archosaurs is niotes are homologous, and, given both the well founded. phylogenetic distribution and topographic re- Parts of the nasal cavity and the choana lationships of the structure, homology is indeed unlikely. Instead, its repeated develop- In accordance with Parsons' ('59, '70) revi- ment probably reflects a tendency among sion of the earlier (often confusing) terminol- many clades to divert the airway caudally, ogy of the parts of the nasal cavity, three usually associated with the evolution of a major components have been recognized here: bony "secondary" palate (Fuchs, '08; Riep- vestibule, nasal cavity proper, and nasopha- pel, '93). ryngeal duct. The boundaries between these The complicated nature of the issues sur- regions are sometimes more definitional and rounding the terminology of the choana of arbitrary than discrete, but the terms are amniotes was only hinted at above in the used so commonly that they are discussed description of crocodilian facial anatomy. It briefly here. The vestibule in virtually all gets even worse when more basal vertebrates extant sauropsids is rostral to the apertures are considered (see papers on the origin of of the nasal gland ducts (Parsons, '59; Muller, tetrapods in Schultze and Trueb, '91). For '61). In turtles, Sphenodon punctatus, many the matter at hand, the question centers on squamates, and crocodilians the vestibule is a the homology (and name) of the opening lo- small cavity within the nasal capsule, but in cated at the juncture of the nasal cavity proper some squamates (Pratt, '48; Bellairs and Ka- and nasopharyngeal duct, particularly in mal, '81) and birds it is apomorphically en- crocodilians. In forms such as Sphenodon, larged. In most cases, it is generally restricted to the narial region of the skull, but most squamates, and the vast majority of in many avian species a ventral recess ex- birds, in which the nasopharyngeal duct is tends caudally back to the crista nasalis poorly developed or absent, the issue is simple (where the nasal gland ducts enter), undercut- and the opening is the choana. Again, it more ting somewhat the nasal cavity proper (see or less coincides with the borders of the carti- also Bang, '71). laginous fenestra basalis (this fenestra is not In tetrapods, the nasal cavity proper is really identifiable in most birds because they fairly consistent, occupying the cartilaginous usually lack a solum nasi). nasal capsule from the vestibule to the pri- Osteologically, in those turtles lacking an mary choana and having respiratory as well extensive nasopharyngeal duct, in most ex- as olfactory epithelium (Plate, '24; Matthes, tant lepidosaurs, and in birds, the choana is '34; de Beer, '37; Parsons, '71). The capsule surrounded by the vomer medially, palatine extends all the way back to the rostral mar- caudally and laterally, and maxilla laterally gin of the orbit, in extant amniotes usually and/or rostrally; the rostral border of the attaching to the lacrimal and/or prefrontal. choana is variable, being either the maxilla or The homology of both the vestibule and nasal premaxilla depending on the presence and cavity proper in extant amniotes has never extent of palatal processes of the maxilla been contested. (Romer, '56; Gafhey, '79; Witmer and Mar- As its name implies, the nasopharyngeal tin, '87). In crocodilians, the aperture termed duct runs from the nasal cavity proper to the earlier the primary choana has precisely the pharynx. To my knowledge, it is not lined by same topographic relationships: vomer medi- the cartilage of the nasal capsule in any verte- ally, palatine caudolaterally, and maxilla ros- brates but rather is lined by respiratory or trolaterally (Fig. 12C). Thus, given this simi- perhaps partly oral or pharyngeal epithe- larity and its congruence with the phylogeny lium. As a result, its rostral end (the choana of extant sauropsids, the homology of the or primary choana; see below) is roughly crocodilian primary choana with that of other coextensive with the fenestra basalis of the sauropsids is well founded. Moreover, virtu- chondrocranium. It is variably developed in ally all fossil archosaurs (as well as extinct, extant amniotes. Most mammals have a well- more basal members of other amniote clades) developed nasopharyngeal duct, as do many exhibit this same morphology. In fact, only turtles, some squamates, all crocodilians, and within the phylogeny of Crocodylomorpha a few birds; it is essentially absent in Sphen- does the formation of a bony nasopharyngeal odon punctatus, most squamates, and most duct (and hence a secondary choana) obscure 302 L.M. WITMER the primary choana from ventral view. Thus, part in most of the succeeding important the hypothesis of homology of the (primary) reviews (e.g., Dieulafk, '05; Peter, '06; Mat- choanae of extant archosaurs is also congru- thes, '34; Parsons, '59, '70; Starck, '67; ent with archosaur phylogeny. Gauthier et al., '88a; Zeller, '89). Gegen- baur's (1873) evidence comes almost exclu- Nasal conchae sively from a very strict, typological defini- Historical background tion of his conception of a "Nasenmuschel": The outgrowths of the lateral wall of the a simple, more or less independent lamella nasal cavity (conchae, turbinals, etc.) vary projecting inward from the capsular wall and tremendously in morphology among extant supported by a cartilaginous or bony skel- vertebrates, and as a result, there is an exten- eton. Thus, he homologized the lamellar sive literature addressing their homologies structures seen in mammals (maxilloturbi- (see Table 2). Although it is far beyond the nal), squamates (concha), crocodilians (con- scope of this paper to adequately review the cha), and birds (middle concha). Following various arguments, the following discussion his strict definition, Gegenbaur eliminated focuses on a few of the more influential treat- the avian caudal concha and crocodilian post- ments, in particular those studies represent- concha from consideration because they are ing alternative approaches to the problem. not lamellar projections but rather invagina- The next subsection presents the results of tions and evaginations of the nasal cavity, my research, drawing evidence from the mode respectively; thus, he regarded both struc- of development of the structures, their topo- tures as independent neomorphs. Most later graphic relationships, and the congruence of workers (e.g., those cited in the first sentence this information with phylogeny. of this paragraph) who otherwise followed Before proceeding, however, it is necessary to introduce the named structures found in Gegenbaur argued that the avian caudal con- the higher taxa of amniotes. As described cha and crocodilian postconcha are in fact above, birds have rostral, middle, and caudal homologous. I will refer to this scheme as the conchae, and crocodilians have a preconcha, GegenbaurIMatthesscheme to recognize Ge- concha, and postconcha (Figs. 9, 16). Squa- genbaur's initial formulation and Matthes' mates have a single concha (Fig. 18A; al- ('34) subsequent influential review. though it is absent occasionally). Sphenodon Gegenbaur's (1873) selection of Alligator punctatus has rostral and caudal conchae mississippiensis as his crocodilian sample (Fig. 18B). Turtles have no named conchae, rather than, say, a species of Crocodylus had but the Muschelwulst and laterale Grenz- a profound impact on his results. Alligator falte have been compared to the squamate has a poorly developed cavum conchae, and concha (Fig. 18C; Gegenbaur, 1873; Parsons, hence its concha is indeed somewhat lamel- '59). Mammals have a very complex and vari- lar, thus conforming to his definition of a able nasal cavity with a large variety of named Nasenmuschel (Fig. 17B). Crocodylus, on the turbinals (see Starck, '67);of these, the max- other hand, has a well-developed cavum such illoturbinal, ethmoturbinal, nasoturbinal, that its concha more closely resembles an atrioturbinal, and the crista semicircularis avian caudal concha and thus would not ad- have figured most prominently in the debate here to his strict definition (Fig. 10B). Thus, on concha1 homologies (Fig. 18D). For mor- it is tempting to wonder what both his re- phological details of these structures in nonar- sults and the subsequent history of the de- chosaurian amniotes, see de Beer ('37), Par- bate would have been had he simply chosen sons ('59, '701, Starck ('671, and Moore ('81) Crocodylus rather than Alligator. and references therein. Although de Beer Although Solger (1876) found merit in Ge- ('37, p. 180) identified an "exceedingly rudi- genbaur's (1873) strict definition, Born (1879) mentary concha nasalis" in the urodele Am- and virtually all later workers found it too bystoma sp., conchae are usually regarded as restrictive and called any kind of projection being restricted to amniotes and absent from into the cavity a Muschel or concha. In fact, Lissamphibia (Dieulafb, '05; Peter, '06; Plate, rather than simply applying a definition, some '24; Matthes, '34; Parsons, '59; Jurgens, '71; workers sought alternative criteria on which Gauthier et al., '88a). to homologize the conchae. For example, Without a doubt, the most influential pa- Hoppe ('34) and Bertau ('35) used an ap- per on the homologies of the nasal conchae is proach advocated by Beecker ('03) whereby that by Gegenbaur (18731, whose proposed the various intracapsular spaces and recesses homologies have been followed for the most (Hohlraume)of the nasal cavity were used as FAClAL HOMOLOGIES IN ARCHOSAURS 303 TABLE 2. Summary ofproposed homologies for the nasal conchae ofdmniota' Avian rostral Reference concha Avian middle concha Avian caudal concha Other Gegenbaur Neomorph Crocodilian concha; squa- Neomorph Neomorph: crocodilian (1873) mate concha; mamma- postconcha lian maxilloturbinal Mihalkovics - Mammalian maxillotur- Squamate concha; mam- (1898) binal malian ethmoturbinal Dieulaf6 ('05) Neomorph Crocodilian concha; squa- Crocodilian postconcha; mate concha; mamma- mammalian ethmotur- lian maxilloturbinal bind Peter ('06) Neomorph Squamate concha; mam- Crocodilian postconcha; malian maxilloturbinal mammalian nasoturbinal Nemours Mammalian maxillotur- Crocodilian concha; squa- ('30) binal mate concha Hoppe ('34) Sphenodon rostral Sphenodon caudal concha; concha; squamate Lippe squamate concha am Choanengang Matthes ('34) Neomorph Crocodilian concha; squa- Crocodilian postconcha; mate concha; Sphen- mammalian nasotur- odon caudal concha; binal mammalian maxilloturbinal de Beer and Crocodilian preconcha Crocodilian concha; squa- Crocodilian postcon- Barrington (in part) mate concha; mamma- cha = mammalian ('34) lian crista semicircu- ethmoturbinal? laris Bertau ('35) Neomorph Crocodilian concha + pre- Crocodilian postconcha; concha; Sphenodon ros- Sphenodon caudal tral concha; squamate concha; squamate Lippe am Choanengang; concha; mammalian mammalian maxillotur- nasoturhinal binal de Beer ('37) Crocodilian Crocodilian preconcha (in Crocodilian concha; squa- Neomorph: crocodilian preconcha; part); mammalian max- mate concha; mamma- postconcha mammalian illoturbinal lian crista semicircu- atrioturbinal laris; urodele concha Lang ('55) Neomorph Crocodilian concha; squa- Mammalian ethmotur- mate concha; mamma- binal or maxilloturbinal lian maxilloturbinal Parsons Neomorph Crocodilian concha + pre- Crocodilian postconcha ('59, '70) concha; Sphenodon caudal concha; squamate concha; mammalian maxiiloturbinal Miiller ('611 Neomorph mammalian maxillotur- Crocodilian concha; squa- binal; crocodilian mate concha concha; squamate concha Starck ('67) Crocodilian concha; squa- Crocodilian postconcha; Neomorphs: mammalian mate concha; mamma- mammalian nasotur- frontoturbinal & eth- lian maxilloturbinal binal moturhinal Gardiner - Crocodilian concha; mam- - - ('82) malian maxilloturbinal Gauthier et Neomorp h Crocodilian concha + pre- Crocodilian postconcha al. ('88a) concha; Sphenodon caudal concha; squamate concha; ? turtle laterale Grenzfalte; mammalian maxillotur- bind This study Neomorph Neomorph Crocodilian concha + pre- Neomorphs: crocodilian concha; Sphenodon postconcha; Sphen- caudal concha; squa- odon rostral concha; ? mate concha; ? mamma- turtle laterale Gren- lian crista semicircu- zfalte; mammalian laris ethmoturbinal, nasoturbinal, maxillotur- hind

'Birds are selected arbitrarily here as the reference to which other amniotes are compared. 304 L.M. WITMER

caud co

’- CNP ch CNP ch C lat Gr D

np d u

CNP rec’du np max sin ost

Fig. 18. A) Lacerta viridis, B) Sphenodonpunctatus, C) Testudo graeca, and D) Oryctolagus cuniculus (late fetus). Right sides of sagittally sectioned snouts in medial view, showing the parts of the nasal cavity and the eonchal and turbid structures.A-C modified after Parsons (’70), B also after Hoppe (’34), and D after Starck (’671. Hatching denotes cut surfaces.

primary evidence to homologize the conchae call this group A), as are the avian middle rather than the epithelial or cartilaginous concha, crocodilian concha, and mammalian structures themselves. Hoppe (’34) homolo- maxilloturbinal (group B). They differ primar- gized the structures of Sphenodon punctatus ily in the status of the structures in lepido- to those of squamates and birds, and Bertau saurs. In the HoppeIBertau scheme, the cau- (’35) accommodated crocodilians and mam- dal concha of Sphenodon punctatus and the mals into this system. According to their squamate concha are in group A, whereas scheme, the following two homologies per- they are in group B according to the Gegen- tain (Table 2):1) the caudal concha of Sphen- baur1Matthes scheme. Likewise, the rostral odon with the squamate concha, the crocodil- concha of Sphenodon and the squamate Lippe ian postconcha, the avian caudal concha, and am Choanengang are in group B in the the mammalian nasoturbinal; and 2) the ros- Hoppe/Bertau scheme and group A in the tral concha of Sphenodon with the squamate Gegenbaur/Matthes scheme (the Lippe am Lippe am Choanengang, the crocodilian con- Choanengang was not discussed by either cha and preconcha together, the avian middle Gegenbaur or Matthes). There is no easy way concha, and the mammalian maxilloturbinal. to reconcile these differences, particularly The Hoppe/Bertau scheme bears some re- since squamates were always the standard of semblances to the Gegenbaur/Matthes comparison (again reflecting a typological tra- scheme. In both schemes, the avian caudal dition). The Hoppe/Bertau scheme has had concha, crocodilian postconcha, and mamma- no lasting impact on notions of conchal ho- lian nasoturbinal are homologous (for the mologies among amniotes, and their “Hohl- purpose of the immediate discussion, I will raume approach” received no further atten- FACIAL HOMOLOGIES IN ARCHOSAURS 305 tion. The Gegenbaur/Matthes scheme has some way in each group, or do some features held sway to the present. characterize particular groups? Which simi- The Gegenbaur/Matthes scheme, how- larities truly inlcate homology? Is the ab- ever, has not gone uncontested. De Beer (’37; sence of clear conchae in turtles primitive or see also especially de Beer and Barrington, derived, and how does it impact on hypoth- ’34) offered an alternative approach to the eses of homology in other amniotes? These problem, this time making use of the form and other questions can be addressed only and ontogeny of the components of the carti- when the detailed morphological information laginous nasal capsule and their relation- arising from similarity testing is examined ships to surrounding structures. It should be for its congruence with phylogeny. The ques- noted, however, that de Beer (’37) had his tions may still not be answered, but the is- own typological definitions, in some respects sues will be in better focus and will highlight the opposite of Gegenbaur’s (1873): He re- areas of future research. garded a “true concha” as a structure enclosing a cavity, whereas “turbinals” are lamel- A reanalysis of conchal homologies lar structures. De Beer offered the following homologies (Table 2): 1) avian caudal concha, The following discussion begins with an crocodilian concha, squamate concha, and analysis of the ontogeny of the cartilaginous mammalian crista semicircularis; 2) avian anlage of the nasal capsule, their timing of middle concha, ventral edge of the precon- development, and their topographic relation- chal recess of crocodilians, and mammalian ships. Birds will be used to introduce both maxilloturbinal; 3) avian rostral concha, the structures and the issues involved, and crocodilian preconcha, and mammalian atrio- then successively more basal groups of ex- turbinal; and 4) crocodilian postconcha and tant amniotes will be analyzed in these terms. possibly the mammalian ethmoturbinal. De Similarities in these features will be summa- Beer (’37) reported conchae as absent in rized, and finally, the congruence of the simi- turtles and Sphenodon punctatus. The only larities with amniote phylogeny will be tested. homology shared by all three schemes is the Birds. The various regions of the adult avian middle concha with the mammalian nasal capsule in extant archosaurs (described maxilloturbinal. The Hoppe/Bertau scheme above) have their developmental origins in and de Beer’s scheme additionally share the various somewhat independent anlage, ei- homology of the avian caudal concha and ther as mesenchymatous condensations or as squamate concha. There are no other similari- chondrification centers. Although there is ties between the Gegenbaur/Matthes scheme considerable variation within birds as to the and de Beer’s scheme. precise details of the associations of these In summary, despite approaching the prob- anlage, it is clear that the major structures of lem from three different perspectives, the the nasal capsule (e.g., the conchae) are homologies proposed by the above workers formed by various combinations of 1) the do not converge on a single answer for all, or parietotectal cartilage, 2) paranasal carti- even most, amniotes. Of the three approaches, lage, and 3) lamina orbitonasalis (= planum de Beer’s is probably the most enlightening antorbitale) (Fig. 19). Generally in birds, the because it takes into account a great deal of nasal septum derives from the trabecula com- morphological and ontogenetic detail-i.e., it munis, the tectum nasi from the septum and is an effective similarity test. Likewise, the parietotectal cartilage, and the paries nasi Hohlraume approach of Hoppe and Bertau from the parietotectal and paranasal carti- involves detailed topographic correspon- lages; the rostral and middle conchae arise dences, but unfortunately in this case, the from the parietotectal cartilage, and the cau- things being compared are hopelessly com- dal concha from the paranasal and parietotec- plex and highly variable, even within a major tal cartilages (see also de Beer and Barring- taxon. Gegenbaur’s approach appears to be ton, ’34; Muller, ’61; Macke, ’69).Given the the least justified in that it entails a typologi- remarkable conservation of the basic form cal, class-based definition of a structure as a and organization of the chondrocranium over basis for homology. It therefore is ironic that vertebrates (de Beer, ’37; Thorogood, ’881, Gegenbaur’s homologies, supplemented by there is the prospect that these components later workers, have been the most popular. of the nasal capsule will provide reliable What all of these studies lack is input from guides to the homology of the conchae. the phylogenetic relationships of amniotes. It is not quite this simple, however, be- Must all of these structures be present in cause there has been considerable debate as 306 L.M. WITMER par c ton (’53),Engelbrecht (’58),and Vorster (’89) reported that virtually none of the elements orb have true independence. Resolution of this A conflicting evidence is difficult. Because the nasal capsule forms relatively rapidly in birds (Vorster, ’89), it is conceivable that denser ontogenetic series would reveal independent anlage in all taxa. Perhaps more importantly, the “independence” issue is probably peripheral to the question of concha1 homologies. It cav co may not be particularly important whether anlage appear in isolation or as outgrowths of other structures. Instead, independence may orb have to do more with the patterning provided B by other tissues (e.g., the developing nasal epithelium, etc.) such that shifts in epithelial folding or even in the timing of neural crest migration may alter epitheliomesenchymal interactions enough to affect the “independence” of an anlage. What clearly is important is whether the various components are present or absent and what their form is. For example, although Frank (’54) could identify a separate paranasal cartilage in Struthio camelus whereas Muller (’61) could not in Rhea americana, the form of the caudal concha is very similar in these birds. Similarly, the absence of both the paranasal cartilage and caudal concha in Euplectes orix (Engel- brecht, ’58) and Melopsittacus undulatus (de Kock, ’87) is clearly a significant association. Thus, regardless of their status as “independent,” the contributions of the anlage to the nasal conchae are important data. As indicated above, in most birds the cau- Fig. 19. Anus plutyrhynehos. A) Dorsal and B) left dal concha forms at or very near the junction lateral view of rostra1 part of chondrocranium of 8.5-day embryo, showing the origin of the parietotectal cartilage, of the paranasal and parietotectal cartilages, paranasal cartilage, and lamina orbitonasalis. C: Rostra1 later reaching caudally to the lamina orbito- part of chondrmaniumof &day embryo of same species nasalis. The concha is formed mostly by the in left lateral view, showing further development of these paranasal cartilage, which projects medially structures, including the nasal conchae. A and B modified after de Beer and Barrington (’34); C is camera lucida into the nasal cavity proper as a cup-shaped, drawing of cleared-and-stained specimen. Scale bars = hemispherical swelling. The caudal concha is 1 mm. concave laterally and encloses an extracapsular cavity, the cavum conchae (Figs. 8, lOA, 19). The aditus conchae leads into the cavum to the independence of these anlage and the and is usually near the paranasal-parietotec- significance of independence (de Kock, ’87). tal juncture, although in Struthio camelus it In fact, it has become almost mandatory in is apparently completely within the parana- the primary literature to report whether or sal cartilage (Frank, ’54);whether the aditus not the anlage appear as independent enti- faces laterally or more rostrally is probably a ties. For birds, de Beer and Barrington (’34), function of the size of the eye (see above). Slab9 (’521,Macke (’691, and Goldschmid (’72) The avian middle concha is a derivative of the described the lamina orbitonasalis and para- dorsolateral edge of the parietotectal carti- nasal cartilage as appearing in isolation, and lage, projecting ventromedially and exhibit- Frank (’541, May (’61), and Toerien (’71) ing a variable amount of scrolling. Although reported that at least the paranasal cartilage Slaby (’51) and Muller (’61) regarded the is independent. On the other hand, Cromp- middle concha as taking its origin from the FACIAL HOMOLOGIES IN ARCHOSAURS 307 juncture of the parietotectal and paranasal transformation that takes place during ontog- cartilages, Toerien ('71) clearly showed that eny, the same basic components of the nasal this association of the middle concha with capsule observed in birds are also present in the paranasal cartilage is a secondary phe- crocodilians. The tectum nasi is formed by nomenon, appearing late in ontogeny. In fact, the nasal septum and parietotectal cartilage, in many birds the middle concha eventually and the paries nasi by the parietotectal and extends caudally, ventral to the caudal con- paranasal cartilages. Unlike most birds, there cha, to reach the region of the lamina orbito- is a solum nasi rostral to the fenestra basalis, nasalis (Figs. 9, 19; Frank, '54; Fourie, '55; formed by the septum, lamina transversalis May, '61). The rostral concha is formed in rostralis, and paraseptal cartilage (Klem- probably all birds by the parietotectal carti- bara, '9 1) . lage (Muller, '61). The concha initially is formed at the junc- The pattern of the timing of development ture of the paranasal and parietotectal carti- of the nasal cartilages and conchae is avail- lages, and eventually the lamina transversa- able for a number of avian species. This pat- lis rostralis also contributes to the concha tern is reasonably consistent, although be- (i.e., its rostromedial limb; see above). At cause nasal development is so rapid and age- first, it projects caudodorsomedially into the sampling is rarely complete, often more than nasal cavity proper but later rotates into its one structure seems to appear simultaneously adult position. The concha encloses an extra- (of course, simultaneous appearance may well capsular cavity (cavum conchae) laterally (at occur in some cases). The order of develop- least in embryos or young animals), and the ment is derived from original study of Anas aditus conchae is elongate (Fig. 15).The pre- platyrhynchos and Gallus gallus and informa- concha is clearly a rostral portion of the tion gleaned from the following studies: de concha that has become somewhat separated Beer and Barrington ('34), Brock ('371, by the formation of the preconchal sinus Crompton ('531, Frank ('541, Engelbrecht (probably in association with snout elonga- ('58), Muller ('61),Macke ('691,Toerien ('711, tion) and is not a separate derivative of one of and Goldschmid ('72). The nasal septum is the main cartilages (Shiino, '14). As a result, always the first to form. The next structure Bertau ('35) was correct in arguing that the to form is variable in the birds studied but concha and preconcha must be considered usually is one or more of the following: lamina together as a unit when discussing concha1 orbitonasalis, paranasal cartilage, parietotec- homologies, and hereafter "crocodilian con- tal cartilage (most workers included the cau- cha" will include the preconcha. The post- dal concha with the paranasal cartilage). The concha forms mostly from the paranasal middle concha always formed after these, cartilage, with the lamina orbitonasalis con- and the rostral concha was always the last to tributing to it caudally. Development of the appear. postconcha is complex. Originally it is a carti- The topographic relationships of these laginous shell (concave laterally) that thick- structures are very consistent in birds. The ens and then, according to Shiino ('141, be- caudal concha from the beginning is located gins to form an internal cavity via resorption; at the caudal portion of the nasal cavity later, an intracapsular epithelial diverticu- proper, just dorsal to the caudal portion of lum (the post-turbinal sinus) penetrates the the choana. Ramus lateralis nasi takes a simi- postconcha, inflating it into an air-filled, car- lar course and in a very few birds (see de tilaginous bubble that is concave both medi- Beer, '37) passes through a foramen epipha- ally and laterally. niale just dorsal to the aditus conchae. The The order of appearance of the various caudal concha is always attached or next to capsular structures is derived from my stud- the lacrimal bone. The middle concha ex- ies of Alligator mississippiensis. As in birds, tends for much of the length of the choana. the nasal septum is the first structure to The nasolacrimal duct opens into the choana appear. The paranasal cartilage, parietotec- just ventral to the middle concha. tal cartilage, and lamina transversalis rostra- Crocodilians. Fewer data are available for lis arise together very close in time, the first extant crocodilians, and the following discus- two forming the vast majority of the concha sion draws heavily on original study of onto- and the third also contributinglater in ontog- genetic series of Alligator mississippiensis eny. The lamina orbitonasalis appears next, and also the work of Meek ('ll),Shiino ('141, forming a separate chondrification. The be- Bertau ('35),and Klembara ('91). Despite the ginnings of the postconcha develop next, fol- 308 L.M. WITMER lowed by segmentation of the concha into sippiensis, see above). The concha projects preconcha and the definitive concha. medially into the nasal cavity and in general The topographical relationships (Fig. 16) is directly dorsal to the choana or choanal become extensively transformed during on- tube (Fig. 18A, see Malan, ’46; Bellairs and togeny owing to elongation of the snout and Boyd, ’50). Laterally, there is almost always a nasal rotation (see above). The postconcha is well-developed aditus and extracapsular ca- located completely behind the primary choana vum conchae. A foramen epiphaniale trans- and attaches to the prefrontal, lacrimal, and mitting the ramus lateralis nasi is normally palatine. The concha forms directly opposite present just dorsal to the aditus conchae. The the primary choana, and although it often lacrimal is reduced or absent in many squa- extends caudomedially far beyond the choana mates, and any particular association of this in adults (especially in Alligator mississippi- bone with the concha could not be discov- ensis), the limbs of the “root” of the concha ered. As already indicated, the nasolacrimal (see above) retain their embryonic associa- duct opens apomorphically in squamates in tion with the primary choana. In Alligator association with the vomeronasal organ. but not Crocodylus spp. (Bertau, ’35)’ramus Sphenodon. Sphenodon punctatus ap- lateralis nasi passes through a foramen pears to have the same basic components of epiphaniale located dorsolaterally, just dor- the nasal capsule as in other extant diapsids sal to the aditus conchae (Fig. 15). The naso- (Bellairs and Kamal, ’81). Although Fuchs lacrimal duct opens into the nasal cavity (’08) and Pratt (’48) identified only a single proper just ventral to the preconcha. The concha, Hoppe (’34) reported the presence of ascending ramus of the maxilla develops ros- two conchae. The precise ontogeny of these tral to the aditus and cavum conchae in asso- conchae is unclear in the literature, particu- ciation with that portion of the lamina trans- larly with regard to the cartilaginous compo- versalis rostralis contributing to the concha nents discussed above. Bellairs and Kamal (Fig. 15).Early in ontogeny the lacrimal bone (’81) labeled the cartilage forming both con- is attached laterally to the concha just caudal chae as the paranasal cartilage and suggested to the aditus conchae (Fig. 15;see also Shiino, that the two structures together may repre- ’14; Klembara, ’911, but later, with rostral sent a single concha. The caudal concha is expansion of the postconcha and nasal rota- the larger of the two and is located more tion, the lacrimal attains a more dorsomedial caudally, opposite the caudal portion of the position such that only the rostral tip reaches choana, whereas the rostral concha is a more the conchal region. horizontal ridge reaching almost to the vesti- Squamates. The situation in squamates bule (Fig. 18B; Hoppe, ’34). Neither concha is relatively simple because they have only a has a laterally open cavum conchae. Accord- single concha or none at all. Parsons (’59) ing to Malan (’461, there is a foramen epipha- effectively refuted the homology proposed by niale in the region of the caudal concha. Hoppe (’34) of the “Lippe am Choanengang” Turtles. The ontogeny of the nasal cap- of squamates with the rostral concha of Sphe- sules is poorly known in turtles. De Beer nodonpunctatus, suggesting, in effect, that it (’37)identified all the familiar components of fails the similarity test of homology. Squa- the capsule (e.g., paranasal cartilage, parieto- mates in general have the same components tectal cartilage, etc.), but these identifica- and basic organization of the nasal capsule as tions were based on the study of a “fully in archosaurs (e.g., parietotectal cartilage, formed” chondrocranium in which the vari- paranasal cartilage, lamina orbitonasalis, ous components are completely continuous. lamina transversalis rostralis, etc.), although Shaner (’26)identified two isolated nasal car- occasionally the paranasal cartilage is absent tilages that appear after the septum, a “para- (e.g., Agamidae, possibly all snakes; Malan, nasal cartilage” (which is clearly a parasep- ’46; Bellairs and Kamal, ’81). These compo- tal, not paranasal, cartilage) and a “lateral nents all arise within a short span of time, nasal cartilage” (which could be either pari- except for the nasal septum which precedes etotectal or paranasal or both). Turtles lack them (Kamal and Abdeen, ’72). According to any named concha. The laterale Grenzfalte Kamal and Abdeen (’72; see also Zada, ’81; (Fig. 18C) has occasionally been suggested to Bellairs and Kamal, ’Sl),the concha is formed be a rudimentary concha (e.g., Gegenbaur, primarily by the paranasal and parietotectal 1873),and Bellairs and Kamal(’81) refer to it cartilages, with the lamina transversalis ros- as a “conchal ridge.” In general, it runs for tralis also contributing (as in Alligator missis- most of the length of the nasal cavity proper FACIAL HOMOLOGIES IN ARCHOSAURS 309 (Parsons, '70). Possibly only in Testudo '89); the foramen is situated just dorsal to graeca does the laterale Grenzfalte project the crista semicircularis (de Beer, '37). Al- far into the nasal cavity and have any resem- though the nasolacrimal duct opens into the blance to a concha; it is a poorly developed nasal cavity ventral to the maxilloturbinal in structure in most turtles (Parsons, '70). De Homo and some others, this is usually re- Beer ('37) and Bellairs and Kamal ('81) regarded as a secondary opening, a derived port a foramen epiphaniale in turtles. character relative to the more rostral open- Mammals. The nasal capsules of mam- ing within the fenestra narina (see Starck, mals are extraordinarily complex, with a large '67, and nasolacrimal duct discussion above). number of named turbinals (Fig. 18D; Paulli, The lacrimal contacts the crista semicircu- 1900; Peter, '06; Plate, '24; Matthes, '34; de laris (which ossifies as the uncinate process Beer, '37; Negus, '58; Starck, '67; Moore, '81; of the ethmoid) in Oryctolagus and is very Novacek, '93). As in other amniotes, the na- close in Homo. sal capsule of mammals is reported to de- The Similarity test. The mode of develop- velop from the same basic components (e.g., ment of the various nasal conchae and turbi- paranasal cartilage, parietotectal cartilage, nals among amniotes exhibits some clear pat- lamina orbitonasalis, lamina transversalis terns. For example, the avian caudal concha, rostralis, etc.; see de Beer, '37; Starck, '67; crocodilian concha, squamate concha, and and Moore, '81). The ontogeny of the turbi- mammalian crista semicircularis form at or nals and crista semicircularis is as follows very near the juncture of the paranasal and (Fig. 18D; see de Beer, '37; Starck, '67): 1) parietotectal cartilages, and the lamina trans- The crista semicircularis forms at the junc- versalis rostralis contributes to the concha of tion of the paranasal and parietotectal carti- squamates and crocodilians (the situation in lages; 2) the ethmoturbinal (i.e,, "first" eth- turtles and Sphenodon punctatus is unclear). moturbinal) forms at the junction of the The crocodilian postconcha and mammalian lamina orbitonasalis and paranasal cartilage; ethmoturbinal both form from contributions 3) the maxilloturbinal is formed by an inroll- of the paranasal cartilage and lamina orbito- ing of the ventral edge of the rostral portion nasalis. Although no one has ever suggested of the paries nasi (presumably parietotectal their homology, the avian middle concha and or paraseptal cartilage); 4) the nasoturbinal mammalian nasoturbinal are both out- forms as a ventral process of the tectum nasi growths of the dorsolateral portion of the (presumably also parietotectal in origin); and tectum nasi. The mammalian maxilloturbi- 5) the atrioturbinal, when present, is either nal is usually regarded as having the greatest the rostral part of the maxilloturbinal (Starck, similarity to the avian middle concha (Table '67) or the inrolled margin of the fenestra 2)' but the mammalian structure forms in a narina (de Beer, '37). None of them enclose a much different manner (as a continuation of lateral cavity early in ontogeny, and thus do the ventral margin of the paries nasi). As not adhere to de Beer's ('37) definition of a most other workers dating back to Gegen- concha. The nasal septum is the first struc- baur (1873) have mentioned, the avian ros- ture to form, followed by the paranasal carti- tral concha exhibits no morphological simi- lage, parietotectal cartilage, and lamina orbi- larities with structures of other amniotes, tonasalis at about the same time; the crista the only possibility being with the mamma- and ethmoturbinal form next, and the other lian atrioturbinal (but here again they form turbinals form much later (de Beer, '37). from different components). The topographical relationships of these With regard to the timing of development, structures are somewhat variable and the in perhaps all extant amniotes, the paranasal discussion is based on study of Oryctolagus cartilage, parietotectal cartilage, and lamina cuniculus (Fig. 18D) and Homo sapiens, and orbitonasalis appear very early and at about information extracted from the literature. the same time. The structures they combine The crista semicircularis is more or less di- to construct thus also appear very early. For rectly opposite the rostral end of the nasopha- example, the avian caudal concha, crocodil- ryngeal duct (primary choana), and the eth- ian concha, squamate concha, caudal concha moturbinal is slightly caudal to it, the others of Sphenodon punctatus, and mammalian being more rostrally situated. Ramus latera- crista semicircularis all form very early. The lis nasi passes through a foramen epipha- mammalian ethmoturbinal also forms early. niale in virtually all major clades of extant Although it is occasionally regarded as the mammals, including monotremes (Zeller, first concha to appear in birds (e.g., Parsons, 310 L.M. WITMER '59; Gauthier et al., '88a), the avian middle as an apomorphy. Most birds also have an concha always forms after the caudal concha. apomorphic position of the nasal gland rela- The avian rostral concha, crocodilian postcon- tive to other amniotes. Since ramus lateralis cha, rostral concha of Sphenodon, and mam- nasi carries the autonomic nerves to the na- malian maxilloturbinal (Zeller, '89) are also sal gland, it is possible that there is some later-appearing structures. causal association between migration of the The following structures enclose a lateral, nasal gland and loss of the foramen epipha- extracapsular cavity, usually called the ca- niale. vum conchae: the avian caudal concha, the Gegenbaur (1873; see also Dieulafk, '05) crocodilian concha, and the squamate con- used the position of the nasolacrimal dud as cha. There is no extracapsular cavum in the a topographic criterion in his homologies, caudal concha of Sphenodon punctatus, the because the duct opens just ventral to the laterale Grenzfalte of turtles, or the crista avian middle concha, crocodilian concha (ac- semicircularis of mammals. The crocodilian tually the preconchal portion), and the mam- postconcha also encloses a very large cavity, malian maxilloturbinal. Gegenbaur (1873) did but this postconchal cavity arises from an not indicate which mammals he sampled, but intracapsular sinus (the post-turbinal sinus) it is likely that he was drawing on human and remains intracapsular. anatomy, because in humans (and some other The following structures are situated di- mammals, see Starck, '67) the nasolacrimal rectly opposite the caudal portion of the (pri- duct indeed opens below the maxilloturbinal. mary) choana: the avian caudal concha, avian However, as mentioned above, this position middle concha, crocodilian concha, squamate is probably an apomorphy within Mammalia, concha (in most cases), the caudal concha of and a more rostral ostium having no particu- Sphenodon punctatus, the laterale Gren- lar association with the maxilloturbinal is zfalte of turtles, and the mammalian crista likely the primitive mammalian condition. In semicircularis. The crocodilian postconcha fact, the nasal ostium of the nasolacrimal and mammalian ethmoturbinal are caudal to duct does not have a particularly consistent it. All other structures are rostral to the position in amniotes, other than that ob- (primary) choana, although in Sphenodon served in birds and crocodilians. and most birds the choana is long enough The lacrimal bone attaches to or is very that the rostral concha of the former and close to the avian caudal concha, crocodilian middle concha of the latter also are opposite concha (at least prior to nasal rotation), and the choana. mammalian crista semicircularis. The lacri- Ramus lateralis nasi passes through a fora- mal is absent in turtles and Sphenodonpunc- men epiphaniale just dorsal to the following tatus and is generally very reduced or absent structures: avian caudal concha (although in extant squamates. the vast majority of birds lack the foramen In summary, based on the above 1:l corre- epiphaniale), crocodilian concha (at least in spondences in ontogenetic development and Alligator mississippiensis), squamate con- topographic relationships of morphological cha, possibly the caudal concha of Sphen- structures, the following hypothesis of homol- odon punctatus, and the mammalian crista ogy passes the similarity test: avian caudal semicircularis. Turtles also have a foramen concha, crocodilian concha (includingthe pre- epiphaniale but its relationship to the nasal concha), squamate concha, caudal concha of structures are uncertain. In birds, crocodil- Sphenodonpunctatus, and mammalian crista ians, and squamates the foramen is just dor- semicircularis. In all these cases, the struc- sal to the aditus conchae. De Beer and Bar- ture 1) forms at or very near the juncture of rington ('34;see also de Beer, '37) used this the paranasal and parietotectal cartilages (un- character as a landmark for determining ho- known for Sphenodon),2) is the earliest con- mologies, and numerous ornithologists (e.g., chal structure to form, 3) is situated (at least Lang, '55; Muller, '61; Goldschmid, '72; de early in ontogeny) directly opposite the cau- Kock, '87; Weber, '90) objected because the dal portion of the (primary) choana (also true foramen is so rarely present in birds. Weber for the laterale Grenzfalte of turtles), 4) is ('90) is correct in noting that, based on in- just ventral to the foramen epiphaniale which group comparison, presence of the foramen is transmits ramus lateralis nasi (the foramen probably not primitive for neornithine birds. is rare in birds but the nerve takes the same Given its distribution in nonavian amniotes, course), and 5) contacts the lacrimal bone its general absence in birds must be regarded (the bone is absent in turtles, Sphenodon, FACIAL HOMOLOGIES IN ARCHOSAURS 311 and reduced or absent in most squamates). Furthermore, in birds, crocodilians,and squamates, the structure encloses an extracapsular sinus, the cavum conchae. For the sake of discussion, this variably named structure will be termed the “primary concha,” borrowing the usage from Peter (’06).The crocodilian postconcha and mammalian ethmoturbinal pass the similarity test of homology on the basis of being 1) formed by both paranasal cartilage and lamina orbitonasalis and 2) located caudal to the primary choana. No other structures have enough 1:1 correspondences to’survivethe similarity test. The Congruence Test. The congruence ti primary concha lost ethmoturbinalipostconcha present test requires that the putative homolog char- 0ethrnoturbinal/postconrha losr acterize a monophyletic group. At first glance, cavum conchae present this is clearly the case with regard to the primary concha-the primary concha charac- cavum conchae lost terizes and is indeed a synapomorphy of Am- niota. However, turtles present a problem because they lack a structure that passed the similarity test. On the basis of strict parsimony and assuming monophyly of Saurop- sida, two equally parsimonious solutions ob- tain: 1) Absence of a primary concha in turtles is a reversal (i.e., a derived absence) such that presence of a primary concha is indeed the ancestral amniote condition (Fig. 20A), or 2) the absence in turtles is truly the primitive condition such that a very similar structure arose independently in Mammalia and Diap- Fig. 20. Cladograms summarizing the congruence of sida (Fig. 20B). hypotheses of homology of nasal conchae and associated Strictly in cladistic terms, these two alter- structures with the phylogeny of extant Amniota. A natives cannot be resolved. The second alter- Cladogram favoring losses over parallelisms. B: Clado- native might seem more likely in that the gram favoring parallelisms over losses. mammalian crista semicircularis simply “looks different” from the structure in diapsids: The crista is inconspicuous and is liter- al. (’88a, p. 1271, who also discussed the ally “overshadowed” by the much more problems presented by turtles in establishing prominent turbinals surrounding it (Fig. conchal homologies, suggested that the later- 18D).However, the numerous detailed simi- ale Grenzfalte might be a primary concha, larities between the crista and the structures “albeit of a sort peculiar to turtles.” The in diapsids are compelling when taken to- analysis presented here offers no support ei- gether, and on this basis it seems likely that ther way as to the status of the laterale the original hypothesis that these structures Grenzfalte. are homologous and characterize Amniota is The primary conchae of extant diapsids correct. The nasal cavity of extant turtles is other than Sphenodon punctatus exhibit an apomorphidy greatly reduced in size rela- extracapsular cavity, the cavum conchae. Its tive to other amniotes and even relative to absence in Sphenodon presents precisely the the Triassic turtle Proganochelys quenstedti same problem as turtles did above. Presence (Gaffney, ’90). Given this apomorphy and the of a cavum may characterize Diapsida, such numerous other craniofacial apomorphies that its absence in Sphenodon is scored as a characterizing extant turtles (Gaffney and derived loss (Fig. 20A), or, equally parsimoni- Meylan, ’881, it is not unreasonable to regard ous in cladistic terms, Sphenodon may retain the absence of the primary concha in turtles the truly primitive condition (also mani- as a derived absence (Fig. 20A). Gauthier et fested by mammals) such that a cavum con- 3 12 L.M. WITMER chae is derived independently in squamates presence of various bony ridges within the and archosaurs (Fig. 20B). Deciding between nasal cavities of several nonmammalian ther- these two alternatives on morphological apsid synapsids have suggested to many work- grounds is not as easy as in the previous ers that turbinals predate the origin of Mam- situation, and largely depends on whether malia (reviewed by Miao, ’88; see also one prefers reversals or parallelisms (i.e., ac- Hillenius, ’92, ’941,but these are absent from celerated or delayed transformation, respec- basal synapsids (Romer and Price, ’40). A tively, in the jargon of character optimiza- concerted-often less than conservative- tion; see Swofford and Maddison, ’87). On effort to find such correlates in Sauropsida the strength of the similarities of the pri- revealed no consistent bony features that mary conchae in squamates and archosaurs, could be used to infer conchae in dried skull my opinion is that a cavum conchae was or fossil material. The only indication of the present in the ancestral diapsid and has been cartilaginous nasal capsule is the presence in lost in Sphenodon (Fig. 20A). Sphenodon many fossil archosaurs of a ridge for the remains poorly known with regard to many paries nasi on the palatine bone in the vicin- of the important features, and new studies ity of the choana (see Witmer, ’94, in press). could have an impact on this issue. Thus, the congruence of the above hypoth- Finally, it is impossible with the data at eses of homology with the phylogeny of both hand to deduce whether the ancestral amni- extinct as well as extant amniotes cannot be ote enclosed a cavum within its primary contested adequately. cha. Without outgroups possessing homolo- In summary, the Gegenbaur/Matthes gous primary conchae, presence and absence scheme, although the most popular and re- of a cavum conchae cannot be polarized. It is peated hypothesis for conchal homologies equally parsimonious to regard either mam- among amniotes, is no longer tenable-it mals or diapsids as having the primitive con- fails both the similarity and congruence tests. dition. The only component of it that survives is the Turning to the homologies of the other homology of the squamate and crocodilian structures, the crocodilian postconcha and conchae. Ironically, both Matthes (’34, p. 926) and Muller (’61, p. 253) regarded the homol- mammalian ethmoturbinal passed the simi- ogy of the avian middle concha and mamma- larity test of homology (although barely so, lian maxilloturbinal as the best-supported and only on two criteria). This hypothesis aspect of the hypothesis and “undisputed” clearly fails the congruence test. If the post- (they even both used “unbestritten”), yet concha and ethmoturbinal were homologous both structures prove to be neomorphs. On (a possibility raised by de Beer and Barring- the other hand, the hypothesis of homology ton, ’34; Table 2), then this homology would advocated here for the primary conchae have to characterize Amniota, necessitating among amniotes is precisely that suggested independent loss in turtles, lepidosaurs, and by de Beer (’37;see also de Beer and Barring- birds (Fig. 20A). Thus, on the basis of parsi- ton, ’34).It differs from de Beer’s in that the mony, these structures are judged as indepen- similarity tests involve many more criteria dent acquisitions and nonhomologous (Fig. and de Beer never tested congruence. The 20B). Likewise, most of the other conchal absence of a congruence test led him to sug- structures that over the years have been re- gest the homology of structures that are here garded as homologous in various amniotes regarded as neomorphs. Although some un- are judged on similar grounds as neomorphs certainties remain as a result of conflicting in this analysis: the avian rostral concha, information from turtles and Sphenodon avian middle concha, crocodilian postconcha, punctatus, the one point about which there is rostral concha of Sphenodonpunctatus, later- absolutely no doubt is that the primary con- ale Grenzfalte of turtles, and all the mamma- chae of extant archosaurs (i.e., the avian lian turbinals. Only a single structure, the caudal concha and crocodilian concha) are primary concha, can be homologized between homologous. any of the major clades. The osteological correlates of the nasal cap- Antorbital cavity sule and conchal structures are remarkably The antorbital cavity is a little different few. Although mammals and a few avian from the other structures whose homology is species ossify some or all of their conchae, being tested here because it is a compart- this is not the case in extant nonavian saurop- ment rather than a bone or soft-tissue ele- sids and indeed the majority of birds. The ment. But since the significance (even exis- FACIAL HOMOLOGIES IN ARCHOSAURS 313 tence) of this cavity has not been appreciated, ous to regard the cavities as having been it is discussed here. A simple definition of acquired independently in Mammalia and antorbital cavity was offered earlier and ap- Archosauria (i.e., no homology); homology plied to both birds and crocodilians. It is the would require the less parsimonious assump- space rostra1 to the orbit, external to the tion of independent loss in Testudines and nasal capsule, and internal to the surface of Lepidosauria. In either case, the antorbital the snout. But definitions, when applied typo- cavities of birds and crocodilians are judged logically, can lead to erroneous interpreta- as homologous. tions as was just demonstrated in the case of If this hypothesis is also congruent with Gegenbaur's (1873) conception of a Nasen- archosaur phylogeny, then the osteological muschel. The question then becomes, what is correlates of the antorbital cavity will be the history of the antorbital cavities of extant present in at least the basal members of all archosaurs? That is, are they homologous? the major clades of archosaurs, both fossil In both birds and crocodilians, the antorbital and extant. These osteological correlates were cavity is a large space readily discernable in mentioned above: 1) a cavity surrounded by a dried skulls. It is surrounded by the maxilla, particular group of bones, 2) the course of the lacrimal, and palatine, with the jugal and nasolacrimal canal, 3) a muscular fossa on prefrontal also taking part in birds and croco- the palatine, and 4) evidence for the position dilians, respectively. The cavities in both taxa of the cartilaginous paries nasi (e.g., crests have a variety of contents: the nasolacrimal associated with the choanal portion of the duct, the nasal gland and/or its ducts (al- palatine bone). The homology of most of these though the gland sometimes fails to reach correlates was demonstrated in previous sec- the cavity in some crocodilians), the maxil- tions. lary neurovasculature, ramus lateralis nasi, When fossil archosaurs are sampled for these features (see Witmer, in press), it be- often a portion of the dorsal pterygoideus comes apparent not only that the presence of musculature, and a paranasal air sac. a large antorbital cavity is indeed an ances- The antorbital cavity is virtually absent in tral feature of Archosauria (affirming its ho- nonarchosaurian sauropsids, because the car- mology in extant forms), but also that extant tilaginous nasal capsule in general com- birds more closely approximate the primitive pletely fills the preorbital portion of the snout archosaurian condition (Witmer, '87, '90, '92, (Plate, '24; Matthes, '34; Parsons, '59). Of in press). In birds, the antorbital cavity is course many of the same structures (e.g., open laterally in the skull via the external ramus lateralis nasi, the nasolacrimal duct, antorbital fenestra, and the maxilla and lacri- etc.) must pass rostrally from the orbit be- mal usually exhibit fossae associated with tween the nasal capsule and roofing bones, the cavity (i.e., antorbital fossae; Fig. 7). This but in lepidosaurs and turtles they pass basic morphology is widely distributed in fos- through a very small potential space. It might sil archosaurs, including basal crocodylo- seem justifiable to regard the antorbital cav- morphs (e.g., Sphenosuchus acutus, Walker, ity as present but very small in nonarchosau- '90; Protosuchus richardsoni, Crompton and rian sauropsids. However, in comparison, the Smith, '80; see Witmer, in press). In fact, cavity in extant archosaurs is so well devel- antorbital fenestrae and fossae are tradition- oped and so well demarcated osteologically ally regarded as diagnostic features of Archo- that it seems overly pedantic to consider lepi- sauria (e.g., Romer, '66; Carroll, '88; among dosaurs and turtles as having an antorbital many others). Thus, extant crocodilians are cavity. Most mammals, on the other hand, apomorphic in secondarily closing the exter- have an extracapsular space, often accommo- nal antorbital fenestra, while retaining the dating many of the same structures as in antorbital cavity (Witmer, in press). archosaurs; this space is best developed in prenatal material (Mihalkovics, 18981, with Air sacs many of the contents of the space later becom- Historical background ing enclosed within bone. The homology of the paranasal air sacs of Although the resemblance is not great, one extant amniotes has received little formal could consider the mammalian extracapsular treatment in the literature. Gegenbaur (1873) space and the antorbital cavities of birds and briefly considered the possibility that the crocodilians as passing the similarity test of avian antorbital sinus is homologous to the homology. According to the phylogeny of ex- crocodilian postconchal sinus, but rejected it tant amniotes, however, it is more parsimoni- because, in effect, the former is extracapsular 314 L.M. WITMER whereas the latter is intracapsular. Solger topographic relationships (Figs. 8,9, 1OA): 1) (1876), however, retained the possibility of just ventral or slightly caudoventral to the their homology, suggesting that perhaps in caudal concha and rostral to the lamina orbi- the ancestors of birds (or in early avian ontog- tonasalis (later in ontogeny, with caudal ex- eny) the sinuses were intracapsular, and be- pansion of the middle concha ventral to the came extracapsular during the course of phy- caudal concha, the ostium may come to lie logeny (or ontogeny) owing to regression of between the two conchae), 2)just ventral to the cartilage. Matthes ('34) and Bertau ('35) the olfactory epithelium (Bang, '71), 3) di- sided with Gegenbaur (1873) in denying the rectly opposite (i.e., dorsal to) the caudalmost homology. Bleicher and Legait ('32) sug- portion of the choana, 4) just medial to the gested the homology of the avian antorbital maxillary nerve and accompanying vessels and mammalian maxillary sinuses, but this (the diverticulum usually eventually sur- idea was refuted by Stadtmuller ('36) and rounds the nerve), and 5) well caudal to the received little further attention. Otherwise, opening of the nasolacrimal duct. The sinus the only other issue receiving much analysis expands nearly to fill the antorbital cavity; was the homology of the crocodilian parana- among all the contents of the cavity, the sal sinuses to those of mammals. For ex- antorbital sinus is the most voluminous. ample, Meek ('06, '11) homologized the croco- Within the antorbital cavity, the diverticu- dilian caviconchal sinus to the mammalian lum in most birds (Bang, '71) penetrates the maxillary sinus. Bertau ('35) and Parsons aditus and cavum conchae and fills (pneuma- ('70) dismissed this hypothesis with little tizes) the caudal concha. The lacrimal bone discussion. In its place, Bertau ('35) sug- forms just caudal to the sinus in the region of gested that the crocodilian caudolateral re- the aditus, and later assumes a relatively cess is homologous to the mammalian maxil- more rostral position owing to pressure from lary sinus, although this hypothesis was based the orbital contents and subsequent caudal entirely on the supposed homology of the growth of the nasal capsule (see above). The crocodilian postconcha with the mammalian sinus is situated just lateral to the nasal nasoturbinal (both of which were regarded as capsule, just deep to the skin, and is lodged neomorphs in a previous section). rostrally in the maxillary bone which is usually pneumatized by the sinus. The bones Extant amniotes comprising the external antorbital fenestra As described above, birds have a single (see above) form initially as splints surround- epithelial extracapsular diverticulum of the ing the antorbital sinus and associated with nasal cavity, whereas crocodilians have five particular portions of the nasal capsule. As a different types of diverticula. The questions result, the fenestra is viewed correctly as an remain, which, if any, of the crocodilian diver- embryonic fontanelle that never closes rather ticula are homologous to the avian antorbital than as a "hole" that opens up ontogeneti- sinus, and do any other amniotes have such a cally. Finally, the nasolacrimal duct passes sinus? Three main criteria will be used to dorsomedially around the sinus, the two com- assess the similarity of these air sacs: 1) the ing into direct contact for much of the course timing of appearance of the diverticulum, 2) ofthe duct. the position and topographic relationships of Crocodilians. In crocodilians,the only one the sinus ostium, and 3) the topographic of the diverticula that consistently appears relationships of the air sac to surrounding during embryonic development is the cavicon- structures. Birds and then crocodilians will chal sinus. In Alligator mississippiensis, the be analyzed first, followed by a discussion of sinus arises just after the first third of the the status of paranasal air sinuses in other embryonic period, at about the same time the amniotes. Any hypotheses of homology sur- concha1 structures are beginning to form. viving the similarity test then will be tested The caviconchal sinus ostium has the follow- for congruence with amniote and archosaur ing topographic relationships (Figs. 10B, phylogeny (including extinct taxa). 12C-E, 17): 1) just caudoventral to the root Birds. The avian antorbital sinus ap- of the concha, 2) immediately ventral to the pears very early in ontogeny, during the first olfactory epithelium (Bertau, '35), 3) directly third of embryonic development in the birds opposite the caudalmost portion of the pri- studied here (see also Lurje, '06; Schuller, mary choana, and 4) caudal to the opening of '39; Bremer, '40). The ostium of the antor- the nasolacrimal duct. Early in ontogeny, the bital sinus has the following almost invariant caviconchal sinus projects laterally beneath FACIAL HOMOLOGIES IN AKCHOSAURS 315 (Alligator;Figs. 15, 17) or perforates (Croco- of the nasal cavity proper and thus far caudal dylus spp.) the paries nasi and is directed to all of the other structures mentioned here; toward the maxillary nerve and vessels; it it leads directly into the prefrontal bone (Fig. eventually reaches the neurovasculature 12C,D). where it expands, perhaps following the het- Summary of extant archosaurs. Compar- erogeneity provided by these structures. The ing crocodilians and birds, it is clear that the sinus expands within the antorbital cavity in avian antorbital sinus shares a number of 1:l the area of the fontanelle between the devel- correspondences with the crocodilian cavicon- oping maxilla, lacrimal, nasal, and prefron- chal sinus: 1) It forms very early in embry- tal, here named the fonticulus antorbitalis onic development; 2) the sinus ostium has (Fig. 15). The caviconchal sinus occupies the virtually identical relationships to the ho- cavum conchae, with the fonticulus antorbit- mologous (primary) concha, olfactory epithe- alis progressively closing around the aditus lium, primary choana, and the aperture of conchae. The ascending ramus of the maxilla the nasolacrimal duct; 3) the sinus has an subsequently closes the fonticulus, probably intimate relationship with the maxillary neu- in association with nasal rotation. Most of rovasculature; 4) the sinus is lodged within the caviconchal sinus eventually occupies a the antorbital cavity and is the largest of the bony cavity within the maxilla, the cavicon- contents of the cavity; 5) the sinus enters the chal recess, passing through a bony aperture cavum conchae; 6) the sinus is just rostral to (Fig. 12C-El that forms around the diverticu- the lacrimal bone where the latter borders lum as it expands. The nasolacrimal duct the aditus conchae (at least early in crocodil- passes dorsomedially over the caviconchal ian development); 7) the sinus occupies a sinus, the two contacting each other briefly large cavity within the caudal portion of the (at least in adults) before the latter enters maxillary bone; and 8)the nasolacrimal duct the caviconchal recess (Fig. 13). contacts and passes dorsomedially over the The other paranasal air sinuses of crocodil- sinus. Furthermore, the caviconchal sinus is ians generally appear after hatching. Excep- the only sinus that has been found to be tions may include the caudolateral sinus in present in all extant crocodilian species that Melanosuchus niger, which seems to appear have been sampled for these attributes. Thus, in late embryonic development (Bertau, ’35), the hypothesis of homology of the avian ant- although the same sinus always appeared orbital sinus and crocodilian caviconchal si- posthatching in the Alligator mississippien- nus passes the similarity test, whereas Sol- sis material studied here. Similarly, Meek ger’s (1876) and Bertau’s (’35) hypotheses (1893) described the formation of a maxillary fail the test. cecal recess in a Crocodylus porosus embryo Lepidosauria. Judging from the litera- that was probably about one-half to two- ture and my dissections, Lepidosauria in gen- thirds of the way through embryonic develop- eral (if not in toto) lack extracapsular parana- ment. The postvestibular and prefrontal si- sal air sinuses. Among the few workers who nuses in Alligator do not appear until some have identified them is Mihalkovics (1898) time after the 1st week and several months who labeled the extraconchal recess of Lac- posthatching, respectively. The caudolateral erta agilis as a “maxillary sinus.” The extra- recess forms ventral to the postconcha and conchal recess is an intracapsular space lined hence caudal to the concha and caviconchal with nasal epithelium (Parsons, ’70) that in recess ostium; it eventually pneumatizes the some squamates (e.g., lacertids) is exposed by palatine bone in at least some alligatorids a deficiency in the capsular wall called the (Fig. 12C; see above for variations). As indi- fenestra lateralis nasi. Although it may have cated earlier, the maxillary cecal recesses are the appearance of a paranasal sinus in some very variable in number, always evaginating transverse sections (e.g., Mihalkovics, 1898, the lateral wall of the nasal cavity proper Fig. 14), it is clearly part of the nasal cavity rostral to the concha; they usually are rostral proper. Even though Nemours (’30) could to the nasolacrimal duct aperture but one or not find any such sinuses in his squamate two may be caudal to it (Figs. 12E, 16). The material, he considered that they indeed were postvestibular sinus ostium is within the present on Mihalkovics’ (1898) authority. The maxillary bone caudal to the vestibule and closest thing to a true paranasal air sinus in well rostral to the opening of the nasolacri- Lepidosauria is probably the recessus extra- mal duct (Fig. 12C,D). The prefrontal sinus capsularis described for Varanus niloticus by ostium is located at the caudomedial corner Malan (’46) and figured for V. monitor by 316 L.M. WITMER Bellairs and Kamal ('81). This structure is dal or caudodorsal to the crista semicircularis similar to the one described by Mihalkovics (which is probably a primary concha); this (1898) in that it is clearly also part of the relationship is actually a little different from extraconchal recess, but here it almost archosaurs where the ostium is never at all reaches the cavum conchae and thus ap- dorsal to the primary concha, but it remains proaches the condition in birds and crocodil- a potential resemblance. The maxillary nerve ians (Malan, '46). However, its topographic passes close to the ostium in Oryctolagus relationships to surrounding structures are cuniculus and through the sinus in Homo otherwise very different, and given the phylo- sapiens. Finally, as its name implies, the genetic position of Varanus within Squamata sinus excavates the maxilla. Thus, given these (Estes et al., '88), it is clearly an apomorphy 1:1 correspondences (and the homology of of the genus. Thus, Lepidosauria can be re- the crista semicircularis to the archosaur garded as lacking a paranasal air sinus ho- primary concha), the mammalian maxillary mologous to the one observed in birds and sinus, avian antorbital sinus, and crocodilian crocodilians. caviconchal sinus could be viewed as passing the similarity test of homology. Turtles. Mihalkovics (1898) also reported a "maxillary sinus" for turtles and was again Congruence with extant amniote phylogeny echoed by Nemours ('30). This structure is Congruence of these features within the clearly the recessus ducti nasopharyngei of context of extant amniote phylogeny reveals Parsons ('59; see Fig. 18C). As its name im- a complex pattern. The following structures plies, Parsons ('59) regarded this small epithe- are potentially homologous: the avian antor- lial cavity as deriving from the nasopharyn- bital sinus, crocodilian caviconchal sinus, and geal duct and not the nasal cavity proper. the mammalian maxillary sinus. It should be The recessus generally has a thick columnar- noted that although it has been referred to to-cuboidal epithelium rather than the thin above as the "mammalian" maxillary sinus, squamous epithelium of the paranasal air it reportedly is absent from extant mono- sinuses of archosaurs. It does not pneuma- tremes and is apparently restricted to Phasco- tize bone or have any osteological correlates larctos cinereus among extant Marsupialia in the material studied here or, judging from where it has rather different topographic re- the photographs published by Gaffney ('791, lationships and may not be homologous probably any other turtles. It does resemble (Paulli, 1900; see also Moore, '81; Wible, '91; the archosaur sinus, however, in being lo- Novacek, '93). Taking monotremes and mar- cated close to the juncture of the nasal cavity supials into account, the avian antorbital si- proper and the caudal border of the primary nus and crocodilian caviconchal sinus are choana and in appearing before hatching. homologous, being congruent with other evi- Since extant turtles lack unambiguous pri- dence that they comprise a monophyletic Ar- mary conchae, nasolacrimal ducts, and antor- chosauria, but the maxillary sinus of placen- bital cavities (which were important land- tal mammals and the paranasal air sinus of marks in archosaurs), the status of the Archosauria are seen as independent acquisi- recessus ducti nasopharyngei as a potential tions (Fig. 21). homolog of the avian antorbital sinus and Congruence with the phylogeny of extinct crocodilian caviconchal recess is problematic and extant amniotes and can be regarded as having failed the test of similarity. Parsons ('70) denied its homol- According to the above analysis, parsimoni- ogy with the mammalian maxillary sinus. ous application of Hennig's ('66) auxiliary principle argues for homology of the main Mammals. Of all the mammalian parana- paranasal air sacs of birds and crocodilians. sal sinuses (Paulli, 1900), only the maxillary However, given the uncertain status of the sinus has the potential to be homologous to sinus in most nonarchosaurian amniotes, it the main paranasal sinus of extant archo- would be reassuring to know that the sinus saurs. The maxillary sinus appears earliest in indeed characterizes all archosaurs and did ontogeny, prenatally in at least dogs, sheep, not evolve independently in extant birds and and humans, although the ethmoidal sinus crocodilians. Furthermore, consideration of sometimes appears at about the same time other fossil amniotes may shed light on the (Dieulafe, '05; Peter, '06). The maxillary si- status of the maxillary sinus of placental nus ostium is more or less opposite the pri- mammals and the recessus ducti nasopharyn- mary choana (Fig. 18D). The ostium is cau- gei of turtles. FACIAL HOMOLOGIES IN ARCHOSAURS 317

Hypothesis oiHoinology: , plarmtal mdiiimal maxillary crocodilian caviconchd siiiu\ j aLian aiitortiital sinus

Fig. 21. Cladogram summarizing the congruence of a hypothesis of homology of paranasal air sinuses with amniote phylogeny. Parsimony dictates that the mammalian maxillary sinus is not homologous to the antorbital sinus of archosaurs.

The following shared osteological correlates of the avian antorbital sinus and crocodilian caviconchal sinus are potentially ob- servable in fossils (see Witmer, in press, for Fig. 22. Omithosuchus longidens. Skull in left lateral more extensive discussion): 1) An antorbital view, showing features of the antorbital cavity. Modified cavity is present in the area of the skull after Walker ('641, Sereno ('91),and specimens. lateral to the nasal cavity and formed principally by the maxilla, lacrimal, and palatine; 2) a fossa or recess is excavated within the lates of the paranasal air sinus, an air sac can maxilla; 3) the antorbital cavity is located be inferred with little speculation in fossil directly opposite the primary choana; and 4) archosaurs (Witmer, in press). The hypoth- the nasolacrimal canal indicates that the na- esis of homology of the avian antorbital sinus solacrimal duct passed dorsomedially around and crocodilian caviconchal sinus therefore the cavity. Fossil archosaur taxa were sur- survives the congruence test over all of archo- veyed for these attributes and are discussed saur phylogeny and is indeed strongly cor- in detail elsewhere (Witmer, in press). The roborated. results of the survey indicate that at least the To my knowledge, paranasal air sinuses basal members of all major clades of archo- have not been reported in any extinct lepidosaurs manifest these osteological correlates sauromorphs, which is consistent with their of paranasal pneumaticity. As just one rather absence in the extant members of the clade. typical example, the Triassic crurotarsan Or- Likewise, sinuses resembling those of archo- nithosuchus longidens (Fig. 22; Walker, '64; saurs have not been described in any fossil Sereno, '91; Witmer, in press) has a very turtles, including Proganochelys quenstedti large antorbital cavity formed by the maxilla, (Gaffney, '90). As in extant turtles, if the lacrimal, and jugal and partially floored by recessus ducti nasopharyngei was present in the palatine. There is an extensive, smooth- fossil turtles, it did not pneumatize bone. In walled antorbital fossa excavated into not fact, the only evidence for any sort of parana- only the maxilla but also the lacrimal and sal pneumaticity in turtles is restricted to jugal. The antorbital cavity and internal ant- Meiolaniaplatyceps, which has a small cavity orbital fenestra are situated directly opposite within the nasal and maxilla in the region of the caudalmost portion of the choana. The the naris; this nasomaxillary sinus, if indeed nasolacrimal canal passes dorsomedial to the pneumatic, would thus be a diverticulum of cavity, probably running between the prefron- the vestibule rather than the nasal cavity tal and lacrimal bones (its rostral portion proper (Gaffney, '83). could not be traced). Thus, given the phyloge- The phylogenetic level within Synapsida at netic distribution of the osteological corre- which the maxillary sinus appears is unclear. 318 L.M. WITMER Rowe ('88) listed the presence of ethmoidal, nally, among extant Mammalia, only placen- frontal, and sphenoidal sinuses as a synapo- tals seem to have a maxillary sinus. Clearly, morphy of therian mammals, implying their more work is necessary to determine the absence in more basal synapsids, but did not homologies of these structures within Synap- comment on the maxillary sinus. Wible ('91, sida. Taken together, however, these data p. 13) subsequently excluded Rowe's ('88) suggest that paranasal air sinuses were not "pneumatic sinuses" character from his present in Synapsida ancestrally, but instead analysis, because, based on Paulli's (1900) appeared later, perhaps even more than once. study, "paranasal air sinuses do not occur widely in taxa other than placentals." How- Summary ever, a number of workers have identified " In summary, with the additional evidence maxillary sinuses" in various nonmamma- from the fossil record, some of the former lian therapsids (Fourie, '74; Kemp, '69, '79, uncertainty can be resolved. There is good '80, '82; Kermack et al. '81; Sues, '851, sug- evidence that the maxillary sinus arose within gesting an earlier origin of this structure. In Synapsida and the antorbitallcaviconchal si- some cases (e.g., Morganucodon watsoni, Ker- nus arose within Sauropsida. Thus, the para- mack et al., 'Sl), the structure really is little nasal air sinus of archosaurs clearly is not more than a depression on the floor of the homologous with the maxillary sinus of pla- nasal cavity, whereas in others (e.g., Lu- centals: The hypothesis may pass the similar- angwa drydalli, Kemp, '80) the bony cavity ity test, but it fails the congruence test. More- is large and connected to the nasal cavity over, there really is no longer any reason to proper via a relatively small ostium. Further- entertain the notion of homology of the reces- more, in most cases the structure has a close sus ducti nasopharyngei of turtles with ei- relationship with the maxillary nerve (Kemp, ther the mammalian or archosaur sinus: It '82; Sues, '851, and in the gorgonsopsian passes only the most superficial of similarity Arctognathus sp. the pterygoideus muscula- tests and is not congruent with amniote phy- ture is even reconstructed as encroaching on logeny. the cavity (Kemp, '69). Paranasal pneumaticity in sauropsids, how- In contrast, in more basal synapsids (i.e., ever, probably did indeed originate some- "pelycosaurs"), there is no indication of any what earlier than the common ancestor of of the maxillary pneumaticity observed in Archosauria-but not much earlier. The ma- therapsids (Romer and Price, '40; personal jor osteological correlates of paranasal pneu- communication from A.S. Romer to T.S. Par- maticity are present in the immediate out- sons: see Parsons, '59, '70). In fact, medial groups of Archosauria: Proterochampsidae, views of the maxillae of most basal synapsids Euparkeria capensis (Fig. l), Erythrosuchi- reveal a relatively simple morphology (Romer dae, and Proterosuchidae (Witmer, in press). and Price, '40; Reisz, '72, '75). It is thus very These basal archosauriforms have not been unexpected to find a basal synapsid with an emphasized here because they fall outside of antorbital fenestra-the Permian varanop- the extant phylogenetic bracket of birds and sid Varanodon agilis (Olson, '65). As in archo- crocodilians. According to the methodology saurs the fenestra in Varanodon is bounded for inferring soft tissues in fossils (see Wit- by the maxilla and lacrimal, but unlike archo- mer, '95, for details), since basal archosauri- saurs, the premaxilla, septomaxilla, and na- forms fall outside of the bracket, soft-tissue sal also bound the fenestra. The soft-tissue inferences must be more tentative for them relations of this structure are unknown. because the causal associations between soft Given its perhaps unique occurrence in Syn- and hard tissues are not directly testable by apsida, it is almost certainly an autapomor- study of extant forms. Nevertheless, given phy of V. agilis. the compelling morphological evidence pro- Thus, synapsids present a very compli- vided by the osteological correlates ofparana- cated picture with regard to the evolution of sal pneumaticity, there is a sound basis to paranasal pneumaticity. Basal synapsids ap- speculate that a homologous paranasal air pear to lack any of the osteological correlates, sac characterizes a somewhat more inclusive except for one form that has an antorbital group than just Archosauria. fenestra. Some nonmammalian therapsids seem to have pneumatic maxillae that might SUMMARY AND CONCLUSIONS pass the similarity test with archosaurs and Perhaps the most significant general con- placentals, whereas others do not. And fi- clusion arising from the foregoing analysis is FACIAL HOMOLOGIES IN ARCHOSAURS 319 that, for many questions in evolutionary biol- may test this hypothesis by surveying these ogy, the dichotomy between paleontology and other (in this case, fossil) descendants-i.e., neontology is not only artificial but actually test for the congruence of the hypothesis may impede an investigator’s arrival at appro- with a specified phylogenetic pattern. The priate inferences. Paleontology is tradition- hypothesis passes the congruence test if the ally the domain of only hard parts, whereas osteological correlates are found in the fos- neontology theoretically can examine virtu- sils. If the association between the soft tissue ally any aspect of an organism. This di- and its osteological correlates is indeed caus- chotomy thus presents a problem for paleon- ally based in the extant forms, then there is a tologists because certain critical information sound basis for inferring the same, homolo- (e.g., soft-tissue attributes) is seemingly out gous soft-tissue element in the fossil form. of reach. For researchers focusing on living This basic approach has been employed groups of organisms, this problem seemingly above to probe the facial anatomy of extant disappears because all tissues theoretically archosaurs (birds and crocodilians) for ho- are available for study. Actually, the neontolo- mologous causal associations between soft gist neither may be able to escape this prob- and hard tissues in the hopes of shedding lem in that all organisms have an evolution- light on the morphology of fossil archosaurs. ary history, and many questions ostensibly In particular, the goal has been to elucidate posed about extant taxa in fact require infor- the soft-tissue relations of the antorbital fe- mation from phylogenetically relevant taxa nestra and cavity, structures that always have known only as fossils. been diagnostic for archosaurs yet enigmatic Thus, in order to correctly interpret both in function. The three previous suggestions extinct and extant organisms, we are chal- for the primary soft-tissue contents of the lenged to expand the domains of traditional antorbital fenestra and cavity have been a osteology or paleontology and to recover what- gland, a portion of the pterygoideus muscle, ever relevant soft anatomical information we and a paranasal air sac. Extant birds retain can from fossils. Gauthier et al. (’88a)effec- antorbital fenestrae and cavities that gener- tively demonstrated “the importance of fos- ally resemble those of most fossil archosaurs, sils,, in phylogeny reconstruction. The Ex- but extant crocodilians lack any such open- tant Phylogenetic Bracket approach advocated ings in the surfaces of their snouts. Thus, the here and elsewhere (Witmer, ’95) is in some question was, what are the soft-tissue con- respects the other side of the same coin in tents of the cavity and fenestra in birds, and that it emphasizes “the importance of extant are there any related features that can be taa” in paleobiology. Only organisms living homologized with crocodilians and hence all today provide the detailed evidence for causal archosaurs? interactions between soft and hard tissues, Having sampled all the major organ sys- and, when the appropriate extant taxa are tems comprising the facial portion of the sampled (i.e., minimally the first two extant head and having established their homolo- outgroups), they constrain all inferences gies over all amniotes or at the very least about the soft-tissue conformation of the ana- among sauropsids, we are now in a better tomical systems of extinct organisms. position to assess the function of the antor- Reconstructing soft tissues in fossils is con- bital cavity of archosaurs. First of all, al- ceptually similar to reconstructing ancestral though extant crocodilians lack a fenestra in or nodal characters and, given that organic the side of their snouts, they retain a homolo- evolution occurs, is founded on nothing more gous bony antorbital cavity internally that complex than the homology relation. Evi- compares well with that of birds, albeit in dence of the causal association between a crocodilians it is highly modified in associa- soft-tissue element and its osteological corre- tion with skull flattening and nasal rotation. lates is provided by extant taxa. When stud- Turning to the three major candidates for ied within a comparative framework, a hy- the soft tissue lodged in the cavity, birds and pothesis can be formulated that similarities crocodilians share a homologous nasal gland in these associations among taxa are due to with all sauropsids (perhaps all tetrapods), a inheritance of the association from a com- homologous m. pterygoideus, pars dorsalis mon ancestor, that is, that the associations (which may be an archosaur apomorphy), are homologous. The hypothesis predicts that and a homologous paranasal air sinus (again, the common ancestor will pass this associa- probably an innovation of archosaurs). Fur- tion on to all of its descendants. Thus, we thermore, all three structures at least reach 320 L.M. WITMER the antorbital cavity in most birds and croco- external antorbital fenestra, yet this fenestra dilians and therefore may be classed as “con- is relatively small in comparison to other tents” of the cavity. Thus, the three major archosaurs, particularly relative to nonavian hypotheses for the function of the archosau- theropods which tend to have enormous ex- rian antorbital cavity all remain valid on ternal fenestrae. This reduction in the size of these counts. the external fenestra (and indeed of the cav- What are the homologous osteological cor- ity as a whole) is probably another manifesta- relates of these soft-tissue elements? With tion of the apomorphic expansion of both the respect to the nasal gland, homologous bony vestibular portion of the nasal cavity and the evidence for the gland is scant but tends to eyeball and its adnexa and the apomorphic involve grooving of the maxilla and nasal in reduction in the size of the maxilla. the vicinity of their suture; thus, the osteo- Furthermore, the analysis reaffirms that logical correlates of the nasal gland do not crocodilians have lost the external antorbital involve the external antorbital fenestra in fenestra. Extant crocodilians plesiomorphi- birds, and it is a minor component of the cally retain the paranasal air sac and the cavity in both extant clades. The major ho- antorbital cavity, but the embryonic fonticu- mologous osteological correlate of the dorsal lus antorbitalis-the embryonic structure pterygoideus muscle is a muscular fossa on most comparable to the external antorbital the palatine. In birds the muscle is never fenestra of other archosaurs-becomes closed associated with the antorbital fenestrae and laterally (almost certainly due to nasal rota- is a minor component of the cavity. A muscu- tion), thus internalizing the antorbital cav- lar fossa is found in fossil archosaurs as well, ity. Extant birds retain a large internal antor- yet there is very good evidence (Witmer, in bital fenestra, which is largely occluded by press) that, as in birds, the muscle was re- the cartilaginous paries nasi. The internal stricted to the caudoventral portion of the antorbital fenestra is retained in extant croco- cavity and did not attach to the antorbital dilians as well, at least in part, and is at least fenestra or fossae. Finally, the homologous the rostromedial portion of the maxillary ap- osteological correlates of paranasal pneuma- erture to the bony caviconchal recess (Fig. ticity include the antorbital cavity itself, its 12C-E). In both birds and crocodilians, the morphology (e.g., excavations of the maxilla internal antorbital fenestra transmits only and/or lacrimal), and its relationship to the the epithelial air sac. primary choana and bony nasolacrimal ca- A final matter relates to the causal nature nal. In both birds and crocodilians, the para- of the associations between particular soft nasal air sinus is the largest component of and hard tissues and suggests avenues of the antorbital cavity, and in birds it extends further research. Although “causal associa- to the margins of the external antorbital tions” have been stressed throughout, causa- fenestra. The presence of these osteological tion remains an assumption in most cases correlates in fossil archosaurs suggests that and has not been specifically tested by, say, they also had a large paranasal air sac filling experimentation. Nevertheless, the causal na- their antorbital cavities. In summary, based ture of many of the associations seems rela- on careful study of all aspects of facial tively straightforward. For example, the dor- anatomy in extant and fossil archosaurs and sal pterygoideus muscles probably indeed the determination of the homologies of par- cause fossae on the palatine bone, and the ticular soft-tissue attributes and their osteo- antorbital and caviconchal air sacs indeed logical correlates, the problem of the func- cause pneumatic cavities via pneumatically tion of the antorbital fenestra and cavity of induced processes of resorption. The very archosaurs is solved-the structures func- existence of the antorbital cavity in extant tioned primarily to house a paranasal air sac. archosaurs well may be caused by the pres- It also was emphasized earlier that an ap- ence of the paranasal air sac. For example, in proach such as this benefits the interpreta- birds, as mentioned earlier, many of the fa- tion not only of fossil organisms but also cial elements form around the air sac such extant organisms, for the simple reason that that the external antorbital fenestra can be extant taxa have a phylogenetic history. Thus, viewed as an embryonic fontanelle that fails the findings reported in the last paragraph to close. It is virtually certain that the air sac impact on our interpretations of modern birds does not provide the inductive interaction and crocodilians. For example, as indicated triggering ossification because other saurop- above, birds are primitive in retaining an sids, which lack the air sac, retain the bones; FACIAL HOMOLOGIES IN ARCHOSAURS 32 1 the bones probably ossify as a result of induc- M.A. Norell, D.B. Norman, K. Padian, J.M. tive interactions involving the nasal capsule, Parrish, P.C. Sereno, and D.B. Weishampel. oral epithelium, etc. However, evagination of For stimulating discussions and/or corre- the air sac slightly precedes the onset of spondence, I thank all of the above people ossification of the facial bones, and subse- and additionally J.J. Baumel, A.J. Charig, quent expansion of the sac probably is a W.P. Coombs, M.W.J. Ferguson, E.S. major factor in determining the position and Gaffney, M.K. Hecht, M.L. Moss, H. Osm61- form of the bones. In crocodilians, the mor- ska, G.S. Paul, S.F. Tarsitano, A.D. Walker, phogenetic process of nasal rotation appar- and C.A. Walker. For their kind hospitality I ently “overrides” the competency of the air thank K., N., and A.C. Padian and P.C. sac to prevent apposition of the facial bones, Sereno. and the external antorbital fenestra subse- For supplying specimens of extant archo- quently closes. A final association that seems saurs, thanks go to T. Joanen, R. Elsey, and to be causally based is the relationship of the L. McNease, Rockefeller Wildlife Refuge, ostium of the air sac to the primary choana. Louisiana Department of Wildlife and Fisher- In both clades of extant archosaurs, the air ies, Grand Chenier (Alligator); M. Staton, sac and choana are tightly linked, starting Mainland Holdings Pty. Ltd., Lae, Papua New from the earliest appearance of the air sac Guinea (Crocodylusspp.); R.D. Klemm, Kan- and continuing throughout ontogeny, de- sas State University via J.J. Baumel, Creigh- spite subsequent ontogenetic transformation ton University (Diomedea);Truslow Farms of surrounding structures. The morphoge- Inc., Chestertown, Maryland (Gullus and netic reason for this association of air sac and Anus); D. Frey, Culver Duck Farms, Inc., choana remains obscure, but is probably real. Middlebury, Indiana (Anus);M. Gebaur, Glen In summary, by studying fossil archosaurs Burnie, Maryland (Anus);M. Sindler, Dover in light of the information provided by their Poultry, Baltimore, Maryland (Gullus); K. extant relatives, it is possible to recover from Schiltz, Schiltz Goose Farm, Inc., Sisseton, the fossils a great deal of important evidence for the existence and form of various soft South Dakota (Anser);K. Owen, Cleburne, tissues. Reciprocally, these insights impact Texas (Struthio);and T. Bryant, San Benito, on our interpretation of the modern taxa. Texas (Struthio). Determination of the homologous causal as- For providing access to specimens in their sociations between soft tissues and their os- care, I thank W.E. Miller (Brigham Young teological correlates indicates that the antor- University, Provo); L.D. Martin and H.-P. bital cavity and fenestrae of fossil archosaurs Schultze (University of Kansas, Lawrence); were associated primarily with an epithelial R.L. Zusi, S.L. Olson, G.R. Zug, N. Hotton diverticulum of the nasal cavity (i.e., an air 111, and M. Brett-Surman (United States Na- sac), and the nasal gland and dorsal pterygoi- tional Museum of Natural History, Washing- deus muscles were only minor components of ton, D.C.); P.R. Bjork (South Dakota School the cavity. of Mines, Rapid City); J.R. Horner and P. Leiggi (Museum of the Rockies, Bozeman); ACKNOWLEDGMENTS J.H. Ostrom and M.A. Turner (Yale Peabody This work derives from doctoral disserta- Museum, New Haven); E.S. Gaffney, M.A. tion research undertaken at the Johns Hop- Norell, and C. Holton (American Museum of kins University School of Medicine under the Natural History, New York City); J.D. Stew- direction of D.B. Weishampel, who deserves art (Los Angeles County Museum, Los Ange- special thanks. For reading all or portions of les); K. Padian, J.H. Hutchison, and M. this paper I thank D.B. Weishampel, A. Greenwald (University of California Mu- Walker, P. Dodson, N.C. Fraser, R.L. Zusi, seum of Paleontology, Berkeley); D.S. Ber- and three anonymous referees. Additionally, man and A.C. Henrici (Carnegie Museum of M.L. Oster-Granitewas helpful both intellec- Natural History, Pittsburgh); A.W. Cromp- tually and in terms of materials and equip- ton, F.A. Jenkins, and C.R. Schaff (Museum ment. of Comparative Zoology, Harvard); J.R. Bolt For generously sharing unpublished infor- (Field Museum of Natural History, Chicago); mation and/or access to loan specimens in S. Chatterjee (Texas Tech University, Lub- their care, I thank S.C. Bennett, H.N. Bry- bock); D.A. Russell (Canadian Museum of ant, S. Chatterjee, J.M. Clark, P.J. Currie, Nature, Ottawa); P.J. Currie (Royal Tyrrell J.A. Gauthier, D. Gower, A.W.A. Kellner, J. Museum of Palaeontology, Dmmheller);A.C. Klembara, J.H. Madsen, Jr., R.E. Molnar, Milner, A.J. Charig, C.A. Walker, and S. 322 L.M. WITMER Chapman (Natural History Museum, Lon- Bellairs, A.d’A., and C.C.D. Shute (1953) Observations on the narial musculature of Crocodilia and its innerva- don); and M.W.J. Ferguson and M. Carette tion from the sympathetic system. J. Anat. 87:367-378. (University of Manchester, Manchester). Benton, M.J., and J.M. Clark (1988) Archosaur phylog- Edward Heck executed Figure 22. eny and the relationships of the Crocodylia. In M.J. Funding was provided by National Science Benton (ed): The Phylogeny and Classification of the Tetrapods, Vol. 1: Amphibians, Reptiles, Birds. Oxford: Foundation Dissertation Improvement Grant Clarendon Press, pp. 295-338. BSR-9112070, two awards from the Alex- Bertau, M. (1935) Zur Entwicklungsgeschichte des ander Wetmore Fund of the American Orni- Geruchsorgans der Krokodile. Z. Anat. Entwicklungsge- thologists’ Union, two Grants-In-Aid of Re- sch. 104:168-202. a Bhattacharyya, B.N. (1982) Functional morphology of search from Sigma Xi, fellowship from the the jaw muscles of some Indian insect-eating birds. Lucille P. Markey Charitable Trust, the New Gegenbaurs Morphol. Jahrb. 128:208-254. York College of Osteopathic Medicine and Bhattacharyya, B.N. (1989) Functional morphology of the Department of Biological Sciences and the jaw muscles of two species of Imperial Pigeons, College of Osteopathic Medicine of Ohio Uni- Ducula aenea nicobarica and Ducula badiainsignis. Gegenbaurs Morphol. Jahrb. 135573-618. versity. Bignon, F. (1889) Contribution a l’etude de la pneuma- ticite chez les oiseaux. Les cellules aeriennes cervico- LITERATURE CITED cephalique des oiseaux et leurs rapports avec les 0s de Adams, L.A. (1919)Amemoir on the phylogeny of the jaw la t6te. Mem. SOC.Zool. France 21260-320. muscles in Recent and fossil vertebrates. Ann. N.Y. Bleicher, M., and E. Legait (1932) De l’homologie de la Acad. Sci. 2851-166. cavite accessoire sous-cutanee des oiseaux avec le sinus Alberch, P. (1985) Problems with the interpretation of maxillaire des mammiferes. Compt. Rend. Assoc. Anat. developmental sequences. Syst. Zool. 34:46-58. 1932:43-5 1. Alberch, P., and J. Alberch (1981) Heterochronic mecha- Bock, W.J. (1989) The homology concept: its philosophi-

nisms of morphological diversification and evolution- cal foundation and uractical methodolom.I” Zool. Beitr. ary change in the Neotropical salamander, Bolitiglossa 32r327-353. occidentalis (Amphibia: Plethodontidae). J. Morphol. Born, G. (1879)Die Nasenhohlen und der Thrinennasen- 167:249-264. gang der amnioten Wirbelthiere. I & 11. Gegenbaurs Anderson, H.T. (1936)The jaw musculature of the phyto- Morphol. Jahrb. 562-140,401-429. saur, Machaeroprosopus. J. Morphol. 59549-587. Born, G. (1883)Die Nasenhohlen und der Thrinennasen- Bang, B.G. (1971) Functional anatomy of the olfactory gang der amnioten Wirbelthiere. 111. Gegenbaurs Mor- system in 23 orders of birds. Acta Anat. (Basel) [Suppl.] phol. Jahrb. 8t188-232. 79:l-76. Bremer, J.L. (1940) The pneumatization of the head of Bang, B.G., and B.M. Wenzel (1985) Nasal cavity and the common fowl. J. Morphol. 67;143-157. olfactory system. In AS. King and J. McLelland (eds): Brock, G.T. (1937) The morphology of the ostrich chon- Form and Function in Birds, Vol. 3. New York Aca- drocranium. Proc. Zool. Soc. Lond. 107:225-243. demic Press, pp. 195-225. Brooks, D.R., and D.A. McLennan (1991) Phylogeny, Barghusen, H.R. (1986) On the evolutionary origin of the Ecology, and Behavior: A Research Program in Com- therian tensor veli palatini and tensor tympani muscles. parative Biology. Chicago: University of Chicago Press. In N. Hotton 111, P.D. MacLean, J.J. Roth, and E.C. Bryant, H.N., and A.P. Russell (1992) The role of phylo- Roth (eds): The Ecology and Biology of Mammal-like genetic analysis in the inference of unpreserved at- Reptiles. Washington: Smithsonian Institution Press, tributes of extinct taxa. Philos. Trans. R. SOC.Lond. pp. 253-262. [Biol.] 337:405-418. Baumel, J.J., and L.M. Witmer (1993) Osteologia. In J.J. Bubien-Waluszewska, A. (1981) The cranial nerves. In E. Breazile, H.E. Evans, and J.C. A.S. King and J. McLelland (eds): Form and Function Handbook of Avian Anatomy: in Birds, Vol. 2. New York: Academic Press, pp... 385- Nomina Anatomica Avium, 2nd Edition. Cambridge: 438. Nuttall Ornithological Society, pp. 45-132. Burton, P.J.K. (1974) Feeding and the Feeding Appara- Baumel, J.J., AS. King, A.M. Lucas, J.E. Breazile, and tus in Waders: A Study of Anatomy and Adaptations in H.E. Evans (1979) Nomina Anatomica Avium: An An- the Charadrii. London: British Museum (Natural His- notated Anatomical Dictionary of Birds, 1st Edition. tory). New York: Academic Press. Busbey, A.B. 111 (1989) Form and function of the feeding Baumel, J.J., AS. King, J.E. Breazile, H.E. Evans, and apparatus of Alligator mississzppiensis. J. Morphol. J.C. Vanden Berge (1993)Handbook ofAvian Anatomy: 202r99-127. Nomina Anatomica Avium, 2nd Edition. Cambridge: Camp, C.L. (1923) Classification of lizards. Bull. Am. Nuttall Ornithological Society. MI&. Nat. Hist. 48:289-481. Beecker, A. (1903) Vergleichende Stilistik der Nasenre- Carroll, R.L. (1988) Vertebrate Paleontology and Evolu- gion bei den Sauriern, Vogeln und Saugerthieren. Ge- tion. New York: W.H. Freeman and Company. genbaurs Morphol. Jahrb. 31 :565-619. Clark, J.M., J. Welman, J.A. Gauthier, and J.M. Parrish Bellairs, A.d’A., and J.D. Boyd (1950) The lachrymal (1993) The laterosphenoid bone of early archosauri- apparatus in lizards and snakes-11. The anterior part forms. J. Vert. Paleontol. 13:48-57. of the lachrymal duct and its relationship with the Colbert, E.H. (1980) Evolution of the Vertebrates. A palate and with the nasal and vomeronasal organs. History of the Backboned Animals Through Time. 3rd Proc. Zool. SOC.Lond. 120t269-310. Edition. New York: John Wiley and Sons. Bellairs, A.d’A., and A. Kamal (1981) The chondrocra- Coulombre, A.J., and E.S. Crelin (1958) The role of the nium and development of the skull in Recent reptiles. developing eye in the morphogenesis of the avian skull. In C. Gans and T.S. Parsons (eds): Biology of the Am. J. Phys. Anthropol. 16:25-37. Reptilia, Volume 11, Morphology F. New York: Aca- Cracraft. J. (1986) The origin and early diversification of demic Press, pp. 1-264. birds. Paleobiology 12:383-399. . FACIAL HOMOLOGIES IN ARCHOSAURS 323 Cracraft, J. (1988) The major clades of birds. In M.J. Frank, G.H. (1954) The development of the chondrocra- Benton (ed): The Phylogeny and Classification of the nium of the ostrich. Ann. Univ. Stellenb. 30A:179-248. Tetrapods, Vol. 1: Amphibians, Reptiles, Birds. Oxford: Fraser, N.C., and M.J. Benton (1989) The Triassic rep- Clarendon Press, pp. 339-361. tiles Bruchyrhimdon and Polysphenodon and the rela- Crompton, A.W. (1953) The development of the chondro- tionships of the sphenodontids. Zool. J. Linn. SOC. cranium of Spheniscus demersus with special refer- 96t413-445. ence to the columella auris of birds. Acta Zool. 34171- Fuchs, H. (1908) Untersuchungen iiber Ontogenie und 146. Phylogenie der Gaumenbildungen bei den Wirbeltiere. Crompton, A.W., and K.K. Smith (1980) A new genus Zweite Mitteilung: Uber das Munddach der Rhyncho- and species of crocodilian from the Kayenta Formation cephalen, Saurier, Schlangen, Krokodile und Sauger (Late Triassic?) of northern Arizona. In L.L. Jacobs und den Zusammenhang zwischen Mund- und Nasen- (ed): Aspects of Vertebrate History. Flagstaff: Museum hohle bei diesen Tieren. 2. Morphol. Anthropol. IIr153- of Northern Arizona Press, pp. 193-217. 248. de Beer, G.R. (1937) The Development of the Vertebrate Gabe, M., and H. Saint-Girons (1976) Contribution a la Skull. London: Oxford University Press. morphologie comparee des fosses nasales et de leurs de Beer, G.R., and E.J.W. Barrington (1934)The segmen- annexes chez les 16pidosauriens. Mem. Mus. Natl. Hist. tation and chondrification of the skull of the duck. Nat. Ser. A Zool. 98:1-87. Philos. Trans. R. SOC.Lond. [Biol.]223t411-467. GaRney, E.S. (1979) Comparative cranial morphology of de Kock, J.M. (1987) The development of the chondrocra- Recent and fossil turtles. Bull. Am. Mus. Nat. Hist. nium ofMelopsittacus undulutus. Adv. Anat. Embryol. 164t65-376. Cell Biol. 104:l-70. Gaffney, E.S. (1983) The cranial morphology of the ex- de Queiroz, K. (1985) The ontogenetic method for deter- tinct horned turtle, Meioluniaplatyceps, from the Pleis- mining character polarity and its relevance to phyloge- tocene of Lord Howe Island, Australia. Bull. Am. Mus. netic systematics. Syst. Zool. 34:280-299. Nat. Hist. 175:361-480. Dieulaf6, L. (1905) Les fosses nasales des vertkbr6s (mor- Gaffney, E.S. (1990) The comparative osteology of the phologie et ernbryologie). J. Anat. Physiol. 41t102-112, Triassic turtle Proganochelys. Bull. Am. Mus. Nat. 300-318,478-560,658-680. Hist. 194r1-263. Dingerkus, G., and L.D. Uhler (1977) Enzyme clearing of Gaffney, E.S., and P.A. Meylan (1988) A phylogeny of alcian blue stained whole small vertebrates for demon- turtles. In M.J. Benton (ed):The Phylogeny and Classi- stration of cartilage. Stain Technol. 52229-232. fication of the Tetrapods, Vol. 1: Amphibians, Reptiles, Dollo, L. (1884) Cinquisme note sur les dinosauriens de Birds. Oxford: Clarendon Press, pp. 157-219. Bernissart. Bull. Mus. R. Hist. Nat. Belg. 3:129-146. Edgeworth, F.H. (1935) The Cranial Muscles of Verte- Gans, C. (1988) Craniofacial growth, evolutionary ques- brates. London: Cambridge University Press. tions. Dev. SUDD~.103:3-15. Elzanowsh, A. (1987) Cranial and eyelid muscles and Gardiner, B. (19kf2) Tetrapod classification. Zool. J. Linn. ligaments of the tinamous (Aves: Tinamiformes). Zool. Soc. 74:207-232. Jahrb. Anat. 116:63-118. Gaupp, E. (1888) Anatomische Untersuchungen uber die Engelbrecht, D.V.Z. (1958)The development of the chon- Ne&enversorgung der Mund- und NaseGhohlendrii- drocranium ofPyromelana orix orin. Acta Zool. 39t116- sen der Wirbelthiere. Gegenbaurs Morphol. Jahrb. 14: 199. 436-489. Erdmann, K. (1940) Zur Entwicklungsgeschichte der Gaupp, E. (1910) Das Lacrimale des Menschen und der Knochen im Schadel des Huhnes bis zum..Zeitpunkt Sauger und seine morphologische Bedeutung. Anat. des Ausschlupfens aus dem Ei. Z. Morphol. Okol. Tiere Anz. 36t529-555. 36:315400. Gauthier, J. (1986) Saurischian monophyly and the ori- Estes, R., K. de Queiroz, and J. Gauthier (1988) Phyloge- gin of birds. In K. Padian (ed): The Oripin of Birds and netic relationships within Squamata. R. Estes and G. ;he Evolution of Flight. San Francisco: Ealifornia Acad- Pregill (eds): Phylogenetic Relationships of the Lizard emy of Sciences, pp. 1-55. Families: Essays Commemorating Charles L. Camp. Gauthier, J.A., A.G. Kluge, and T. Rowe (1988a) Amniote Stanford: Stanford University Press, pp. 119-281. phylogeny and the importance of fossils. Cladistics Evans, S.E. (1988) The early history and relationships of 4t105-209. the Diapsida. In M.J. Benton (ed): The Phylogeny and Gauthier, J., R. Estes, and K. de Queiroz (1988b) A Classification of the Tetrapods, Vol. I. Oxford Claren- phylogenetic analysis of Lepidosauromorpha. In R. Es- don Press, pp. 221-260. tes and G. Pregill (eds): Phylogenetic Relationships of Ewer, R.F. (1965) The anatomy of the thecodont reptile the Lizard Families: Essays Commemorating Charles Euparkeria capensis Broom. Philos. Trans. R. SOC. L. Camp. Stanford: Stanford University Press, pp. 15- Lond. [Biol.] 248:379-435. 98. Ferguson, M.W.J. (1981) The structure and development Gegenbaur, C. (1873) Ueber die Nasenmuscheln der Vo- of the palate in Alligator mississippiensis. Arch. Oral gel. Jena. Z. Med. Naturwiss. 7tl-21. Biol. 26:427-443. George, J.C., and A.J. Berger (1966)Avian Myology. New Ferguson, M.W.J. (1985)Reproductive biology and embry- York: Academic Press. ology of the crocodilians. C. Gans, F. Billet, and P.F.A. Getty, R. (1975) Sisson and Grossman’s The Anatomy of Maderson (eds): Biology of the Reptilia, Vol. 14A. New the Domestic Animals, 5th Edition. Philadelphia: W.B. York: Johns Wiley & Sons, pp. 32S491. Saunders Company. Fisher, H.I., and D.C. Goodman (1955) The myology of Goldschmid, A. (1972) Die Entwicklung des Craniums the Whooping Crane, Grus umericana. Illin. Biol. der Mausvogel (Coliidae, Colliiformes, Aves): I. Die Monogr. 24t1-127. Friihentwicklung des Chondrocraniums; 11. Die En- Fourie, S. (1955) A contribution to the cranial morphol- twicklung des Chondrocraniums bis zum Auftreten der ogy of Nyctisyrigmuspectoralis pectoralis with special dermalen Verknocherungen; 111. Ausgestaltung und reference to the palate and cranial kinesis. Ann. Univ. Umbau des Chondrocraniums bis zum Auftreten der Stellenb. 31t179-215. Ersatzverknocheruneen: IV. Die Entwicklune des Os- Fourie, S. (1974) The cranial morphology of Thrin- teocraniums. Gegenbvaurs Morphol. Jahrb. ?18:105- axodon liorhinus Seeley. Ann. S. Afr. Mus. 65t337-400. 138,274-304,369-413,553-569. 324 L.M. WITMER

Goodman, D.C., and H.I. Fisher (1962) Functional Kemp, T.S. (1969) On the functional morphology of the Anatomy of the Feeding Apparatus in Waterfowl (Aves: gorgonopsid skull. Philos. Trans. R. SOC.Lond. tBiol.1 Anatidae). Carbondale: Southern Illinois University 256:1-83. Press. Kemp, T.S. (1979) The primitive cynodont Procynosu- Gregory, W.K. (1920) Studies in comparative myology chus: functional anatomy of the skull and relation- and osteology: No. 1V.-A review of the evolution of ships. Philos. Trans. R. SOC. Lond. [Biol.] 28573-122. the lacrymal bone of vertebrates with special reference Kemp, T.S. (1980) Aspects of the structure and func- to that of mammals. Bull. Am. Mus. Nat. Hist. 4295- tional anatomy of the Middle Triassic cynodont Lu- 263. angwa. J. Zool. 191:193-239. Hamburger, V., and H.L. Hamilton (1951) A series of Kemp, T.S. (1982) The relationships of mammals. Zool. normal stages in the development of the chick embryo. J. Linn. SOC.77:353-384. J. Morphol. 88:49-92. Kermack, K.A., F. Mussett, and H.W. Rigney (1981) The Hanken, J. (1983) Miniaturization and its effects on skull ofMorganucodon. Zool. J. Linn. SOC.71:l-158. cranial morphology in plethodontid salamanders, ge- Kesteven, H.L. (1945) The evolution of the skull and the nus Thorius (Amphibia, Plethodontidae): 11. The fate cephalic muscles. A comparative study of their develop- of the brain and sense organs and their role in skull ment and adult morphology. Part 111. The Sauria and morphogenesis and evolution. J. Morphol. 177:255-268. Part IV. TheTheria. Mem. Aust. Mus. 7:237-316. Hanken, J., and B.K. Hall (1984) Variation and timing of King, A.S., and J. McLelland (1984) Birds: Their Struc- the cranial ossification sequence of the oriental fire- ture and Function. Philadelphia: Baillihre Tindall. bellied toad Bombina orientalis (Amphibia, Discoglossi- Klembara, J. (1991) The cranial anatomy of early ontoge- dae). J. Morphol. 182:245-255. netic stages ofAlligator mississippiensis (Daudin, 1802) Hennig, W. (1966) Phylogenetic Systematics. Urbana: and the significance of some of its cranial structures for University of Illinois Press. the evolution of tetrapods. Palaeontogr. Abt. A215:103- Herring, S.W. (1993) Epigenetic and functional influ- 171. ences on skull growth. In J. Hanken and B.K. Hall Koecke, H.-U. (1958) Normalstadien der Embryonalent- (eds): The Skull, Vol. 1. Development. Chicago: Univer- wicklung bei der Hausente (Anas boschas domestica). sity of Chicago Press, pp. 153-206. Embryologia 4r55-78. Hillenius, W.J. (1992) The evolution of nasal turbinates Komarek, V. (1979) Anatomia Avium Domesticarum, and mammalian endothermy. Paleobiology 18:17-29. Volume 1. Bratislava: Priroda. Hillenius, W.J. (1994) Turbinates in therapsids: evidence Lakjer, T. (1926) Studien iiber die Trigeminus-versorgte for the Late Permian origins of mammalian endo- Kaumuskulatur der Sauropsiden. Copenhagen: C.A. thermy. Evolution 48:207-229. Reitzel. Hochstetter, F. (1906) Beitrage zur Anatomie und En- Lang, C. (1955) Beitriige zur Entwicklungsgeschichte des twickelungsgeschichte des Blutgefassystemes der Kro- Kopfskelettes von Melopsittucus undulatus. Gegen- kodile. Reise Ostafr. Wissensch. Ergebn. 4:l-139. baurs Morphol. Jahrb. 94r335-390. Hofer, H. (1950) Zur Morphologie der Kiefermusculatur Lang, C. (1956) Das Cranium der Ratiten mit besonderer der Vogel. Zool. Jahrb. Abt. Anat. Ont. Tiere 70:427- Beriicksichtigung von Struthio cumelus. Z. Wissensch. 556. Zool. 159:165-224. Hoppe, G. (1934) Das Geruchsorgan von Hatteria punc- Langston, W., Jr. (1973) The crocodilian skull in histori- tata. Z. Anat. Entwicklungsgesch. 102:434-461. cal perspective. In c. Gans and T.S. Parsons (eds): Hopson, J.A. (1991) Systematics of the nonmammalian Biology of the Reptilia, Vol. 4, Morphology D. New Synapsida and implications for patterns of evolution in York: Academic Press, pp. 263-284. synapsids. In H.-P. Schultze and L. Trueb (eds): Ori- Lubosch, W. (1933) Untersuchungen uber die Visceral- gins of the Higher Groups of Tetrapods: Controversy muskulatur der Sauropsiden. Gegenbaurs Morphol. and Consensus. Ithaca: Cornell University Press, pp. Jahrb. 72:584666. 635-693. Lurje, M. (1906) Uber die Pneumatisation des Tauben- Howes, G.B. (1891) On the probable existence of a Jacob- schadels. Anat. Hefte I. Abt. 93:l-61. son’s organ among the Crocodilia; with observations Mabee, P.M. (1993) Phylogenetic interpretation of onto- upon the skeleton of that organ in the Mammalia, and genetic change: sorting out the actual from the artefac- upon the basi-mandibular elements in the Vertebrata. tual in an empirical case study of centrarchid fishes. Proc. Zool. SOC.Lond. 2892:148-159. Zool. J. Linn. SOC.107r175-291. Iordansky, N.N. (1973) The skull of the Crocodilia. In C. Macke, T. (1969) Die Entwicklung des Craniums von Gans and T.S. Parsons (eds): Biology of the Reptilia, Fulica atra L. Gegenbaurs Morphol. Jahrb. 113:229- Vol. 4. New York: Academic Press, pp. 201-262. 294. Joanen, T., and L. McNease (1977) Artificial incubation MacPhee, R.D.E. (1981) Auditory regions of primates of alligator eggs and post-hatching culture in con- and eutherian insectivores: morphology, ontogeny, and trolled environmental chambers. Proc. World Maricult. character analysis. Contrib. Primatol. 18:l-282. SOC.8:483490. Maddison, W.P., M.J. Donoghue, and D.R. Maddison Joanen, T., and L. McNease (1979) Culture of the Ameri- (1984) Outgroup analysis and parsimony. Syst. Zool. can alligator, Alligator mississippiensis. Int. Zoo Yb. 33t83-103. 19:61-66. Malan, M.E. (1946) Contributions to the comparative Jollie, M.T. (1957) The head skeleton of the chicken and anatomy of the nasal capsule and the organ of Jacob- remarks on the anatomy of this region in other birds. J. son. Ann. Univ. Stellenb. 24A:69-137. Morphol. 100:389436. Marples, B.J. (1932) The structure and development of Jurgens, J.D. (1971) The morphology of the nasal region the nasal glands of birds. Proc. Zool. SOC.1932:829- of Amphibia and its bearing on the phylogeny of the 844. group. Ann. Univ. Stellenb. 46:1-146. Martin, L.D. (1991) Mesozoic birds and the origin of Kamal, A.M., and A.M. Abdeen (1972) The development birds. In H.-P. Schultze and L. Trueb (eds):Origins of of the chondrocranium of the lacertid lizard, Acantho- the Higher Groups of Tetrapods: Controversy and Con- dactylus boskiana. J. Morphol. I37;289-334. sensus. Ithaca: Cornell University Press, pp. 485-540. FACIAL HOMOLOGIES IN ARCHOSAURS 325 Martin, B.G.H., and A.d’A. Bellairs (1977) The narial Fleischer (eds): Vertebrate Morphology. New York excrescence and pterygoid bulla of the gharial, Gavialis Gustav Fischer Verlag, pp. 159-162. gangeticus (Crocodilia). J. Zool. 182:541-558. Owrc, O.T. (1967) Adaptations for locomotion and feed- Matthes, E. (1934) Geruchsorgan. In L. Balk, E. Goppert, ing in the Anhinga and Double-crested Cormorant. E. Kallius, and W. Lubosch (eds): Handbuch der Verglei- Ornithol. Monogr. 6:l-138. chenden Anatomie der Wirbeltiere. Berlin: Urban & Parrish, J.M. (1993) Phylogeny of the Crocodylotarsi, Schwarzenberg, pp. 879-948. with reference to archosaurian and crurotarsan mono- May, W. (1961) Die Morphologie des Chondrocraniums phyly. J. Vert. Paleontol. 13:287-308. und Osteocraniums eines Waldkauzembryos (Strix Parsons, T.S. (1959)Studies on the comparative embryol- aluco L.). Z. Wissensch. Zool. 166:134-202. ogy of the reptilian nose. Bull. Mus. Comp. Zool. 120: Meek, A. (1893) On the occurrence of a Jacobson’s organ, 101-277. with notes on the development of the nasal cavity, the Parsons, T.S. (1970) The nose and Jacobson’s organ. In lachrymal duct, and the Harderian gland in Cmcodilus C. Gans and T.S. Parsons (eds): Biology of the Reptilia, porosus. J. Anat. Physiol. 27:151-160. Vol. 2. New York: Academic Press, pp. 99-191. Meek, A. (1906) The olfactory organ of the crocodile and Parsons, T.S. (1971) Anatomy of nasal structures from a the homologies of the ethmoid region. Proc. Univ. comparative viewpoint. In H. Autrum, R. Jung, W.R. Durham Philos. Sac. 3:l-2. Loewenstein, D.M. MacKay, and H.L. Teuber (eds): Meek, A. (1911)On the morphogenesis of the head of the Handbook of Sensory Physiology, Vol. IV: Chemical crocodile (Cdilusporosw).J. Anat. Physiol. 45:357- Senses, Part 1. New York Springer-Verlag, pp. 1-26. 377. Patterson, C. (1982)Morphological characters and homol- Mertens, R. (1943) Die rezenten Krokodile des Natur- ogy. In K.A. Joysey and A.E. Friday (eds): Problems of Museums Senckenberg. Senckenbergiana 26:252-312. Phylogenetic Reconstruction. New York: Academic Merz, R.L. (1963) Jaw musculature of the Mourning and Press, pp. 21-74. White-winged Doves. Univ. Kansas Publ. Mus. Nat. Patterson, C. (1988) Homology in classical and molecular Hist. 12r521-551. biology. Mol. Blol. Evol. 5:603425. Miao, D. (1988) Skull morphology of Lambdopalis bulla Paulli, S. (1900) Uber die PneumaticitSt des Schadels bei (Mammalia, Multituberculata) and its implications to den Saugerthieren. Eine morphologische Studie: I. mammalian evolution. Contrib. Geol. Univ. Wyom. Uber den Bau des Siebbeins. Uber die Morphologie des Spec. Pap. 4:1-104. Siebbeins und die der Pneumaticitat bei ben Mono- Mihalkovics, V. (1898) Nasenhohle und Jacobsonsches tremen und den Marsupialiern; 11. Uber die Morpholo- Organ. Eine morphologische Studie. Anat. Hefte 11:l- pie des Siebbeins und die der Pneumaticitat bei den 107. Ungulaten und Probosciden; 111. Uber dir Morphologie Moore, W.J. (1981) The Mammalian Skull. Cambridge: des Siebbeins und die der Pneumaticitat bei den Insec- Cambridge University Press. tivoren, Hyracoideen, Chiropteren, Carnivoren, Pinni- Miiller, H.J. (1961) Die Morphologie und Entwicklung pedien, Edentaten, Rodentieren, Prosimieren und Pri- des Craniums von Rhea americana Linne. I. Das knor- maten, nebst einer zusammenfassendenUbersicht iiber pelige Neurocranium. Z. Wissensch. Zool. 165:221-319. dir Morphologie des Siebbeins und die der Pneuma- Miiller, H.J. (1963) Die Morphologie und Entwicklung des Craniums von Rhea americana Linne. 11. Viszeral- ticit& des Schadels bei den Saugerthieren. Gegenbaurs skelett, Mittelohr und Osteocranium. Z. Wissensch. Morphol. Jahrb. 28:147-178,179-251,483-564. ZOO^. 168:35-118. Peter, K. (1906) Die Entwickelung des Geruchsorgans Miiller, F. (1967) Zur embryonalen Kopfentwicklung van und Jakobson’schen Organs in der Reihe der Wirbelt- Crocodylus cataphructus Cuv. Rev. Suisse Zool. 74: 189- iere. Bildung der ausseren Nase und des Gaumens. In 294. 0. Hertwig (ed): Handbuch der vergleichenden und Negus, V.E. (1958)The ComparativeAnatomy and Physi- experimentellen Entwickelungslehre der Wirbeltiere, ology of the Nose and Paranasal Sinuses. London: Val. 2, Pt. 2. Jena: Gustav Fischer, pp. 1-82. Livingstone. Plate, L. (1924) Allgemeine Zoologie und Abstammungs- Nemours, P.R. (1930) Studies on the accessory nasal lehre. Jena: Gustav Fischer. sinuses: the comparative morphology of the nasal cavi- Poglayen-Neuwall, V.I. (1953) Untersuchungen der Kief- ties of reptiles and birds. Ann. Otol. Rhinol. Laryngol. ermuskulatur und deren Innervation an Krokodilen. 39: 108G1 108. Anat. Anz. 99:257-276. Nickel, R., A. Schummer, and E. Seiferle (1977) Anatomy Pratt, C.W.M. (1948) The morphology of the ethmoidal of the Domestic Birds (Translation by W.G. Siller and region of Sphenodon and lizards. Proc. 2001. SOC.Lond. P.A.L. Wight). Berlin: Verlag Paul Parey. 118:171-201. Norell, M.A. (1989) The higher level relationships of the Pumphrey, R.J. (1961) Sensory organs: Vision. In A.J. extant Crocodylia. J. Herpetol. 23:325-335. Marshall (ed): The Biology and Comparative Physiol- Novacek, M.J. (1993)Patterns ofdiversity in the mamma- ogy of Birds, Vol. 2. New York: Academic Press, pp. lian skull. In J. Hanken and B.K. Hall (eds): The Skull, 55-68. Vol. 2. Patterns of Structural and Systematic Diver- Pycraft, W.P. (1900) On the morphology and phylogeny sity. Chicago: University of Chicago Press, pp. 438-545. of the Palaeognathae (Ratitae and Crypturi) and Neog Novas, F.E. (1992)Phylogenetic relationships of the basal nathae (Carinatae). Trans. Zool. SOC.Lond. 15:149- dinosaurs, the Herrerasauridae. Palaeontology 35:51- 290. 62. Rathke, H. (1866) Untersuchungen iiber die Entwick- Oelrich, T.M. (1956) The anatomy of the head of Cteno- elung und den Korperbau der Krokodile. Braunsch- saura pectinatu (Iguanidae). Mix. Publ. Mus. Zool. weig Friedrich Viewegund Sohn. Univ. Mich. 94:l-122. Reese, A.M. (1924) The cephalic glands ofAUigator mis- Olson, E.C. (1965) New Permian vertebrates from the sissippiensis, Florida alligator, and of Agkistrodon, Chickasha Formation in Oklahoma. Oklahoma Geol. copperhead and moccasin. Biol. Gen. 1:482-500. Surv. Circ. 7O:l-70. Reisz, R. (1972) Pelycosaurian reptiles from the Middle Osmolska, H. (1985) Antorbital fenestra of archosaurs Pennsylvanian of North America. Bull. Mus. Comp. and its suggested function. In H.-R. Duncker and G. Zool. 144:27-62. 326 L.M. WITMER

Reisz, R. (1975) Pennsylvanian pelycosaurs from Linton, Shaner, R.F. (1926) The development of the skull of the Ohio and N@any, Czechoslovakia.J. Paleontol. 49:522- turtle, with remarks on fossil reptile skulls. Anat. Rec. 527. 32343-367. Remane, A. (1952) Die Grundlagen des natiirlichen Sys- Shiino, K. (1914) Das Chondrocranium van Crocodylus tems, der vergleichenden Anatomie und der Phylogen- porosus mit Beriicksichtigung der Gehirnnerven und tik. Leipzig: Akademische Verlagsgesellschaft. Kopfgefasse. Anat. Hefte 50:257-381. Richards, L.P., and W.J. Bock (1973) Functional anatomy Sims, R.W. (1955) The morphology of the head of the and adaptive evolution of the feeding apparatus in the hawfinch (Coccothruustes coccothraustes), with special Hawaiian Honeycreeper, genushops (Drepanididae). reference to the myology of the jaw. Bull. Br. Mus. Ornithol. Monogr. 15:l-173. (Nat. Hist.) Zool. 2371-393. Rieppel, 0. (1988) Fundamentals of Comparative Biol- Slaby, 0. (1951) Le developpement du chondrmbe du ogy. Boston: Birkhauser Verlag. cormorant (Phalucrocoraz curb L.) au point de vue de Rieppel, 0. ( 1993) Patterns of diversity in the reptilian l’bvolution. Bull. Int. Acad. Tcheque Sci. LII/9:147. skull. In J. Hanken and B.K. Hall (edsf: The Skull, Val. Slaby, 0. (1952) Einige interessante Bemerkungen zur 2. Patterns of Structural and Systematic Diversity. Entwicklung des Schadels beim Fischreiher (Ardeu Chicago: University of Chicago Press, pp. 344-390. cinerea L.). Acta Sac. Zool. Bohemoslov. 16:163-178. Rieppel, 0. (1994) The role of paleontological data in Smith, M.M., and B.K. Hall (1990) Development and testing homology by congruence. Ada Paleontol. Po- evolutionary origins of vertebrate skeletogenic and lon. 38:295-302. odontogenic tissues. Biol. Rev. 65:277-373. Romanoff, A.L. (1960) The Avian Embryo: Structural Solger, B. (1876)Beitrage zur Kenntniss der Nasenwand- and Functional Development. New York: Macmillan ung, und hesonders der Nasenmuscheln der Reptilien. Company. Gegenbaurs Morphol. Jahrb. 1:467-494. Romer, A.S. (1956) Osteology of the Reptiles. Chicago: Stadtmiiller, F. (1936) Kranium und Visceralskelett der University of Chicago Press. Saugertiere. In L. Bolk, E. Goppert, E. Kallius, and W. Romer, A.S. (1966) Vertebrate Paleontology, 3rd Edition. Lubosch (eds): Handbuch der vergleichenden Anato- Chicago: University of Chicago Press. mie der Wirbeltiere. Berlin: Urban and Schwarzen- Romer, AS., and L.I. Price (1940) Review of the Pelyco- berg, pp. 839-1016. sauria. Geol. Sac. Am. Spec. Pap. 28:l-538. Starck, D. (1967) Le crbe des Mammiferes. In P.-P. Romer, AS., and T.S. Parsons (1986) The Vertebrate Grasse (ed): Traite de Zoologie, Tome XVI: Mam- Body, 6th Edition. Philadelphia: Saunders College. miferes Teguments et Squelette. Paris: L’Academie Rose, C. (1893) Ueber die Nasendriise und die Gaumen- Saint-German de Medecine, pp. 405-549. driisen von Crocodilus porosus. Anat. Anz. 8:745-751. Starck, J.M. (1989)Zeitmuster der Ontogenesen bei nest- Roth, V.L. (1984) On homology. Biol. J. Linn. Sac. 22t13- fliichtenden und nesthockenden Vogeln. Cour. Forsch- 29. ungsinst. Senckenberg 114:l-318. Rowe, T. (1988) Definition, diagnosis, and the origin of Stevens, P.F. (1984) Homology and phylogeny: morphol- Mammalia. J. Vert. Paleontol. 8:241-264. ogy and systematics. Syst. Bot. 9:395-409. Saint-Girons, H. (1976) Donnees histologiques sur les Stresemann, E. (1927-1934) Sauropsida. Aves. In W. fosses nasales et leurs annexes chez Crocodylus niloti- Kiikenthal and T. Krumbach (eds), Handbuch der Zo- cus Laurenti et Caiman crocodilus (Linnaeus) (Rep- ologie, Volume 7(2). Berlin: de Gruyter. tilia, Crocodylidae). Zoomorphologie 84r301-318. Sues, H.-D. (1985) The relationships of the Tritylodonti- Saint-Girons, H. (1989) Les glandes cephalique exocrines dae (Synapsida).Zool. J. Linn. SOC.85r205-217. des Reptiles. 11.-Considerations fondionelles et evolu- Swofford, D.L., and W.P. Maddison (1987) Reconstruct- tives. Ann. Sci. Nat. Zool. Paris Ser. 13 1O:l-17. ing ancestral states under Wagner parsimony. Math. Sandoval,J. (1963) Estudio sobre la anatomia comparada Biosci. 87r199-229. y functional del esqueleto cefalico de la gallina y su Taylor, W.O.G. (1939) Effect of enucleation of one eye in morfogenesis. An. Anat. Facul. Med. Zaragoza 12.983- childhood upon the subsequent development of the 359. face. Trans. Ophthalmol. Sac. U.K. 59t361-371. Schiiller, H. (1939) Die Entwicklung des Geruchsorgans Technau, G. (1936) Die Nasendriise der Vogel. Zugleich bei der Sturmmove und der Seeschwalbe. Z. Anat. ein Beitrag zur Morphologie der Nasenhohle. J. Orni- Entwicklungsgesch. 109:75-98. thol. 84:511-617. Schultze, H.-P., and L. Trueb (1991) Origins of the Thorogood, P.V. (1988) The developmental specification Higher Groups of Tetrapods: Controversy and Consen- of the vertebrate skull. Dev. Suppl. 103:141-153. sus. Ithaca: Cornell University Press. Toerien, M.J. (1971) The developmental morphology of Schumacher, G.-H. (1973) The head muscles and hyo- the chondrocranium of Podiceps cristatus. Ann. Univ. laryngeal skeleton of turtles and crocodilians. In C. Stellenb. 46Atl-128. Gans and T.S. Parsons (eds): Biology of the Reptilia, Voeltzkow, A. (1899) Beitrage zur Entwickelungsge- Vol. 4, Morphology D. New York: Academic Press, pp. schichte der Reptilien. Biologie und Entwicklung der 101-199. ausseren Korperform van Crocodilus rnadugascarien- Sereno, P.C. (1986) Phylogeny of the bird-hipped dino- sis Grand. Abh. Senckenberg. Naturforsch. Gesellsch. saurs (Order Ornithischia). Natl. Geogr. Res. 2234- 26:l-150. 256. Vorster, W. (1989) The development of the chondrocra- Sereno, P.C. (1991) Basal archosaurs: phylogenetic rela- nium of Gallus gallus. Adv. Anat. Embryol. Cell Biol. tionships and functional implications. Sac. Vert. Pale- 113:1-77. onto]. Mem. 2, J. Vert. Paleontol. Suppl. 11:153. Walker, A.D. (1961) Triassic reptiles from the Elgin area: Sereno, P.C., and A.B. Arcucci (1990) The monophyly of Stagonolepis, Dasygnathus and their allies. Philos. crurotarsal archosaurs and the origin of bird and croco- Trans. R. Soc. Land. [Biol.] 244:103-204. dile ankle joints. N. Jahrb. Geol. Palaontol. Abh. 180: Walker, A.D. (1964) Triassic reptiles from the Elgin area: 21-52. Ornithosuchus and the origin of carnosaurs. Philos. Sereno. P.C.. and R. Wild (1992) Procomvsopnathus: Trans. R. Sac. Land. 1Biol.I 248t53-134. theropod, “thecodont” or both? J. Veri. P>leontol. Walker, A.D. (1990) A revision of Sphenosuchus aeutus 121435458, Haughton, a crocodylomorph reptile from the Elliot FACIAL HOMOLOGLIES IN ARCHOSAURS 327

Formation (late Triassic or early Jurassic) of South Witmer, L.M. (1990) The craniofacial air sac system of Africa. Philos. Trans. R. Sac. Lond. [Biol.] 330t1-120. Mesozoic birds (Aves). Zool. J. Linn. SOC.l00t327-378. Walls, G.L. (1942) The Vertebrate Eye and its Adaptive Witmer, L.M. (1992) Ontogeny, Phylogeny, and Air Sacs: Radiation. Bloomfield Hills, NJ: Cranbrook Press. The Importance of Soft-Tissue Inferences in the Inter- Wassersug, R.J. (1976)A procedure for differential stain- pretation of Facial Evolution in Archosauria. Unpub- ing of cartilage and bone in whole formalin-fixed verte- lished Ph.D. Dissertation. Baltimore: Johns Hopkins brates. Stain Techno]. 51:131-134. University School of Medicine. Ann Arbor: University Webb, M. (1957) The ontogeny of the cranial bones, Microfilms #93-13453. cranial peripheral and cranial parasympathetic nerves, Witmer, L.M. (1994) A revised terminology of the facial together with a study of the visceral muscles of Stmthio. osteology of archosauriforms. J. Vert. Paleontol. Suppl. Acta Zool. 38t81-203. 14:53A. Weber, E. (1990) Zur Kraniogenese bei der Lachmowe Witmer, L.M. (1995) The Extant Phylogenetic Bracket (Larusridibundus L.), Zugleich ein Beitrag zur Fkkon- and the importance of reconstructing soft tissues in struktion des Grundplans der Vogel. Gegenbaurs Mor- fossils. In J.J. Thomason (ed):Functional Morphology phol. Jahrb. 136:335-387. in Vertebrate Paleontology. New York: Cambridge Uni- Wegner, R.N. (1958) Die Nebenhohlen der Nase bei den versity Press, pp. 19-33. Krokodilen. Wissensch. Z. Ernst Moritz Arndt-Univ. Witmer, L.M. (in press) The evolution of the antorbital Greifswald 7:l-39. cavity of archosaurs. Mem. SOC.Vert. Paleont. Wellnhofer, P. (1974) Das fiinfte Skelettexemplar von Witmer, L.M., and L.D. Martin (1987) The primitive Archaeopteryx. Palaeontogr. Abt. A 147:169-216. features of the avian palate, with special reference to Wettstein, 0. van (1931-1937) Rhynchocephalia. In W. Mesozoic birds. Doc. Lab. Geol. Lyon 99:21-40. Kiikenthal and T. Krumbach Handbuch der Zo- Wu, X.-C.,and S. Chattejee (1993)Dibothrosuchuseluph- (eds): ros, a crocodylomorph from the Early Jurassic of China ologie: Eine Naturgeschichte der Stamme der Tier- and the phylogeny of the Sphenosuchia.J. Vert. Paleon- rieches, Val. 7: Sauropsida: Allgemeines Reptilia, Aves. tol. 13:5&89. Berlin: de Gruyter, pp. 1-235. Zada, S. (1981) The fully formed chondrocranium of the Wettstein, 0. von (1937-1954) Crocodilia. In W. Kiiken- agamid lizard, Agama pallida. J. Morphol. 17Ot43-54. thal, T..Krurnbach, J.-G. Helmcke, and H. von Leng- Zeller, U. (1989) Die Entwicklung und Morphologie des erken (eds): Handbuch der Zoologie: eine Naturge- Schadels van Ornithorhynchus unatinus (Mammalia: schichte der Stamme der Tierrieches, Vol. 7: Prototheria: Monotremata). Abh. Senckenb. Naturfor- Sauropsida: Allgemeines Reptilia, Aves. Berlin: de Gruy- sch. Ges. 545t1-188. ter, pp. 236-424. Zusi, R.L. (1962) Structural adaptations of the head and Wible, J.R. (1991) Origin of Mammalia: the craniodental neck in the Black Skimmer, Rynchops nigm Linnaeus. evidence reexamined. J.Vert. Paleontol. 1l:l-28. Publ. Nuttall Ornithol. Club 3;l-101. Wiley, E.O. (1981) Phylogenetics: The Theory and Prac- Zusi, R.L. (1993) Patterns of diversity in the avian skull. tice of Phylogenetic Systematics. New York: John Wiley In J. Hanken and B.K. Hall (eds): The Skull, Val. 2. and Sons. Patterns of Structural and Systematic Diversity. Chi- Witmer, L.M. (1987)The nature ofthe antorbital fossa of cago: University of Chicago Press, pp. 391437. archosaurs: Shifting the null hypothesis. In P.J. Currie Zusi, R.L., and R.W. Storer (1969) Osteology and myol- and E.H. Koster (eds): Fourth Symposium on Mesozoic ogy of the head and neck of the Pied-billed Grebes Terrestrial Ecosystems, Short Papers. Drumheller: (Podilymbus).Misc. Publ. Mus. Zool. Univ. Mich. 139: Royal Tyrrell Museum of Palaeontology, pp. 230-235. 149.