Cranial osteology and braincase morphometrics of Gavialis gangeticus: implications for crocodylian phylogenetics

University of Iowa Iowa Research Online Theses and Dissertations 2011 Cranial osteology and braincase morphometrics of Gavialis gangeticus: implications for crocodylian phylogenetics Maria Eugenia Leone Gold University of Iowa Copyright 2011 Maria Eugenia Leone Gold This thesis is available at Iowa Research Online: https://ir.uiowa.edu/etd/2504 Recommended Citation Gold, Maria Eugenia Leone. "Cranial osteology and braincase morphometrics of Gavialis gangeticus: implications for crocodylian phylogenetics." MS (Master of Science) thesis, University of Iowa, 2011. https://ir.uiowa.edu/etd/2504. Follow this and additional works at: https://ir.uiowa.edu/etd Part of the Geology Commons

CRANIAL OSTEOLOGY AND BRAINCASE MORPHOMETRICS OF GAVIALIS GANGETICUS: IMPLICATIONS FOR CROCODYLIAN PHYLOGENETICS by Maria Eugenia Leone Gold A thesis submitted in partial fulfillment of the requirements for the Master of Science degree in Geoscience in the Graduate College of The University of Iowa May 2011 Thesis Supervisor: Associate Professor Christopher A. Brochu

Copyright by MARIA EUGENIA LEONE GOLD 2011 All Rights Reserved

Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL MASTER S THESIS This is to certify that the Master s thesis of Maria Eugenia Leone Gold has been approved by the Examining Committee for the thesis requirement for the Master of Science degree in Geoscience at the May 2011 graduation. Thesis Committee: Christopher A. Brochu, Thesis Supervisor Jonathan Adrain Justin Sipla

To Josh, Rex, Maia, and Choya, without whose constant love and support I would have lost my mind long ago ii

ACKNOWLEDGMENTS I am thankful for the guidance of my advisor, Chris Brochu, and committee members Jonathan Adrain and Justin Sipla. The Department of Geosciences at the University of Iowa helped fund this project, as did the Littlefield Foundation and Tim Rowe. I thank Yale University and the Peabody Museum for lending me the specimen. I thank Jessie Maisano for CT scanning and processing the specimen and Matt Colbert for rendering the images. Nina Triche and Dave Dufeau were extremely helpful with answering my random questions about software and thesis writing. The graduate students of the Department of Geosciences at U. of Iowa were very supportive with edits and software help, especially J. Bowen, S. Casebolt, and J. Camp. I thank E. Guzman for always having my back. And lastly I d like to thank my family (Leones, Huldtgrens, and Golds) and for moral and financial support throughout my life. iii

TABLE OF CONTENTS LIST OF TABLES... vi LIST OF FIGURES... vii LIST OF ABBREVIATIONS... x CHAPTER 1: INTRODUCTION... 1 Problems In Crocodylia... 1 This Study... 4 CHAPTER 2: MATERIALS AND METHODS... 9 Specimens... 9 Scanning Parameters... 9 Osteological Descriptions... 10 Morphometric Analyses... 11 CHAPTER 3: CRANIAL OSTEOLOGY OF Gavialis gangeticus... 19 General Description... 19 General Shape and Characteristics... 19 Cranial Openings... 19 Regions... 20 Elements... 21 Premaxilla... 21 Maxilla... 21 Nasal... 22 Palatine... 23 Vomer... 23 Pterygoid... 23 Ectopterygoid... 24 Jugal... 24 Quadratojugal... 25 Quadrate... 25 Lacrimal... 26 Prefrontal... 26 Palpebral... 26 Frontal... 27 Postorbital... 27 Parietal... 28 Squamosal... 28 Braincase... 28 Laterosphenoid... 28 Basisphenoid... 29 Prootic... 29 Supraoccipital... 30 Exoccipital... 30 Basioccipital... 31 Mandible... 32 iv

Dentary... 32 Splenial... 32 Angular... 32 Surangular... 33 Coronoid... 33 Articular... 33 Pneumaticity and Neurovascular System... 34 Eustachian System... 34 Cartoid Artery... 35 Dentition... 35 Hyoid... 36 Ontogenetic Variation... 36 CHAPTER 4: RESULTS FROM THE MORPHOMETRIC ANALYSES... 50 CHAPTER 5: DISCUSSION AND CONCLUSIONS... 70 Discussion... 70 Conclusions... 72 APPENDIX A: MIDSAGITTAL SKULL RENDERINGS... 74 APPENDIX B: MIDSAGITTAL SLICES... 84 REFERENCES... 86 v

LIST OF TABLES Table 1: Specimens used in this study, with the date of the scan and age of the specimen indicated...14 Table 2: Landmark numbering scheme and descriptions....15 Table 3: Landmark explanations for the second morphometric analyses...15 Table 4: Eigenvalues and variance explained by each axis in the SAS PCA...53 Table 5: Variable loadings on principal component axes...54 Table 6: Results of the Kruskal-Walis one-way ANOVA...55 Table 7: Eigenvalues for relative warp analysis with Caiman....55 Table 8: Eigenvalues for relative warp analysis without Caiman....55 Table 9: Landmark weights of the relative warp analysis with Caiman and without Caiman...56 vi

LIST OF FIGURES Figure 1: Phylogenetic trees showing relationships among crocodylians using molecular data (right) and morphological data (left) (from Brochu 2003)...7 Figure 2: Phylogeny of specimens in this analysis (based on Brochu 2003)...8 Figure 3: Figure 4: Figure 5: Figure 6: Gavialis gangeticus, specimen YPM HERR 008438, in dorsal view...16 Gavialis gangeticus, specimen YPM HERR 008438, in ventral view...17 Landmark scheme for 2-dimensional morphometric analyses...18 Landmark scheme for second morphometric analyses...18 Figure 7: Dorsal view of skull....37 Figure 8: Left lateral view of skull....38 Figure 9: Palatal view of skull....39 Figure 10: Coronal section through the snout...40 Figure 11: Coronal section through supratemporal fenestrae...41 Figure 12: Lower temporal region exposed...42 Figure 13: Posterior view of skull...43 Figure 14: Coronal section through the orbits....44 Figure 15: Midsagittal cut of skull...45 Figure 16: Transverse section through braincase...46 Figure 17: Ventral view of skull....47 Figure 18: Comparison between hatchling (top) and adult (bottom) Gavialis in dorsal view...48 Figure 19: Comparison of hatchling (top) and adult (bottom) Gavialis in lateral view...49 Figure 20: Plot of principal component 1 versus 2...57 Figure 21: Plot of principal component 1 versus 3...58 Figure 22: Thin plate spline showing deformation along PC1....59 Figure 23: Thin plate spline showing deformation along PC2....60 Figure 24: Thin plate spline showing deformation along PC3....61 Figure 25: RW1 vs. RW2...62 vii

Figure 26: RW1 vs. RW3...62 Figure 27: RW2 vs. RW3...63 Figure 28: Deformation along RW1, negative at left, positive at right....64 Figure 29: Deformation along RW2, negative at right, positive at left....64 Figure 30: Deformation along RW3, negative at right, positive at left....65 Figure 31: RW1 vs. RW2 with Caiman excluded....65 Figure 32: RW1 vs. RW3 with Caiman excluded....66 Figure 33: RW2 vs. RW3 with Caiman excluded....67 Figure 34: Deformation along RW1 with Caiman excluded...68 Figure 35: Deformation along RW2 with Caiman excluded...68 Figure 36: Deformation along RW3 with Caiman excluded...69 Figure A1: Gavialis gangeticus (YPM 008438), A) Midsagittal cut; B) Sutural interpretation...74 Figure A2: Gavialis gangeticus (TMM-M-5490), A) Midsagittal cut; B) Sutural interpretation...75 Figure A3: Alligator mississipiensis (TMM-M-6723), A) Midsagittal view; B) Sutural interpretation....76 Figure A4 : Alligator mississipiensis (TMM- uncatalogued), A) Midsagittal view; B) Sutural interpretation....77 Figure A5: Caiman crocodilus (FMNH-73711), A) Midsagittal view; B) Sutural interpretation...78 Figure A6: Tomistoma schlegelii (TMM-M-6342), A) Midsagittal cut; B) Sutural interpretation...79 Figure A7: Osteolaemus tetraspis (FMNH-93986), A) Midsagittal view; B) Sutural interpretation...80 Figure A8: Mecistops cataphractus, (TMM-M-3529), A) Midsagittal view, B) Sutural interpretation....81 Figure A9: Crocodylus moreletii (TMM-M-4980), A) Midsagittal view; B) Sutural interpretation...82 Figure A10: Crocodylus johnstonii (TMM-M-6807), A) Midsagittal view; B) Sutural interpretation...83 viii

Figure B1: Midsagittal slices of G. gangeticus, A. mississippiensis, and C. crocodilus showing the Eustachian system in red....84 Figure B 2: Midsagittal slices of the indicated species, showing the Eustachian system in red....85 ix

Anatomical Abbreviations an: Angular ar: Articular blpt: Pterygoid Bulla bo: Basioccipital bs: Basisphenoid bsr: Basisphenoid Rostrum ca: carotid artery d: Dentary ec: Ectopterygoid emf: External Mandibular Fenestra eo: Exoccipital f: Frontal fm: Foramen Magnum crocodyloids: Hyoid ic: Internal Choanae itf: Infratemporal Fenestra j: Jugal la: Lacrimal ls: Laterosphenoid m: Maxilla mep: Medial Eustachian Passage met: Anterior Medial Eustachian Tube metp: Posterior Medial Eustachian Tube n: Nasal orb: Orbital p: Parietal pal: Palatine pm: Premaxilla po: Pstorbital pot: Prootic pp: Palpebral prf: Prefrontal prfp: Prefrontal Pillar ptf: Posttemporal Fenestra pt: Pterygoid q: Quadrate qj: Quadratojugal roe: External Otic Recess san: Surangular so: Supraoccipital spl: Splenial sq: Squamosal Institutional Abbreviations FMNH: Field Museum of Natural History NJSM: New Jersey State Museum TMM: Texas Memorial Museum YPM: Yale Peabody Museum LIST OF ABBREVIATIONS x

1 CHAPTER 1: INTRODUCTION Problems In Crocodylia Crocodylians and birds represent the two living groups of archosaurs. Much work has been done investigating the comparative structure and function of these animals, and their relationships with the theropod lineage in order to better understand the diversity of modern dinosaurs. Similar studies have been performed for the pseudosuchian lineage, which includes the living crocodylians and anything more closely related to them than birds. Once regarded as so-called living fossils, crocodylians and their close relatives are now known to have achieved great morphological and ecological diversity throughout their long history, including fully marine, bipedal, armored, hoofed, predatory, and herbivorous forms (Brochu 2001). As these various forms are discovered and described, details of their anatomy are incorporated into phylogenetic analyses, so as to develop a more complete idea of their evolutionary history. These analyses are not without problems, however. Within the crown group, molecular and morphological data result in similar phylogenies that differ only in a couple of key points (Figure 1). These points correspond to the placement of the modern gharials: the Indian gharial, Gavialis gangeticus, and the false gharial, Tomistoma schlegelii (Brochu 2003). Using morphologic characters, phylogenetic analyses place Tomistoma among crocodyloids and Gavialis is the basalmost modern crocodylian (Brochu 2003). If this hypothesis is correct, then the conditions seen in Gavialis (e.g. non-verticalized skull, discussed below) could either be plesiomorphic, paedomorphic (with other crocodylians showing the normal conditions), or peramorphic (thereby obscuring what the juvenile condition really is). Conversely, using molecular data, Gavialis and Tomistoma are reconstructed as sister taxa sharing a more recent common ancestor with crocodyloids than with alligatoroids (Brochu 2003). If this is the case, then some of the primitive features seen

2 in Gavialis would be secondary reversals. Gavialis is already known to possess ecological and morphological reversals from ancestral conditions, such as reduced salt water tolerance (Buffetaut 1978) and reduced nasal length (Brochu 2006b). Some argue that this sister group relationship with Tomistoma may be due to convergent skull shape rather than shared ancestry (Brochu 2003). Tarsitano (1989) argued that not enough of the genome of Gavialis and Tomistoma is known and that the genome that is known contains too many plesiomorphies to assess the phylogeny, but subsequent analyses based on nucleotide data sets and more sophisticated analytical techniques (Gatesy and Amato 1992; Gatesy et al. 2003; Willis et al. 2007) have not resolved the issue. In either case, examining a variety of crocodylians, while paying close attention to stem gavialoids and hatchling or young juvenile specimens, could provide important data to help illuminate the solution. All animals change physically as they mature from hatchling (or newborn, as the case may be) to adult. Derived eusuchians undergo an ontogenetic change where the bones of their braincase shift and grow relative to each other. In particular, the basisphenoid grows ventrally relative to the basioccipital, which moves the suture caudodorsally, resulting in a more vertically oriented braincase (Tarsitano 1985). This change is accompanied by a re-orientation of the quadrate and the jaw musculature, and internally shifts pneumatic passages within the braincase. Tarsitano (1985) argued that Tomistoma shares a verticalization pattern with crocodyloids and alligatoroids, and that Gavialis does not undergo this change and is therefore more plesiomorphic than the other eusuchians. Ontogenetic changes may also obscure features that have phylogenetic significance. To identify if these features are present in Gavialis, hatchling specimens must be compared to specimens at other stages of growth (namely juveniles and adults) and any changes need to be characterized and analyzed for phylogenetic inplications. Many archosaurs have pneumatized skulls (Colbert 1946). In crocodylians, three canals open between the basisphenoid and basioccipital along the posteroventral margin

3 of the braincase. In life these bear epithelial tubes originating from the pharynx that course into the ventrolateral braincase wall and eventually into the quadrate, where they empty into the tympanic space (Hecht and Tarsitano 1983, Tarsitano 1985). The openings are called Eustachian foramina, but they are unrelated to the Eustachian structures in mammals. The foramina in juvenile alligatorids and crocodyloids with unverticalized braincases are much shorter, (Hecht and Tarsitano 1983), implying that changes in the pneumatic system correlate to changes in the verticalization of the braincase during ontogeny. In contrast, Gavialis is thought to retain very short Eustachian canals into adulthood, thereby retaining the juvenile (i.e. plesiomorphic) condition. More recent studies of extinct relatives of Gavialis, such as the Campanian Maastrichtian gavialoid Eothoracosaurus, also suggest a verticalized braincase condition (Brochu 2004). This suggests that either Gavialis is secondarily reversed, or that other transformations in the gharial skull are masquerading as plesiomorphic characters. This could signify that there was a conflict between early interpretations of polarity and placement in a phylogeny in accordance with a misinterpreted character set. One possible way to expand morphological data sets is to identify soft tissue structures that interact with the skeleton in modern crocodylians, characterize the bony channels that surround them, and use that information to assess potential bony correlates for use in fossil species. This has been done in some cases, e.g., with the inclusion of information from muscle attachment scars on limb bones (Brochu 2006; Salisbury et al. 2006), but very little such information has been added to date. Another way is to look at the ontogenetic changes that take place in complex osteological systems on which these data sets rely, such as basicranial pneumaticity. If ontogenetic trajectories are different, we will have to rethink our homology assessments. A recent geometric morphometric analysis was performed on the skulls of four species of crocodylians (Gavialis, Tomistoma, Crocodylus acutus, and Mecistops

4 cataphractus) using 3-dimensional landmarks on the dorsal, ventral, caudal, and lateral sides of over 90 specimens (Piras et al. 2010). This analysis also used Euclidean distances to estimate allometric trajectories and to calculate predicted shapes for hatchlings. They found that Gavialis had a significantly different growth trajectory and used this to argue that Gavialis and Tomistoma should not be united as sister taxa (Piras et al. 2010). This study focused on the external landmarks of the skull, leaving the internal structures untested. The skull of any animal is subjected to the environmental and ecological pressures that come from natural selection. Differences in diets or environments can cause drastic changes in the shape of a skull (Sadlier 2009). The braincase, however, is functionally constrained due to its close relationship with the brain. Since the braincase is less apt to change phenotypically due to pressures from natural selection, evolutionary relationships may be preserved within the morphology of this region. Therefore, key to resolving the labiality of Gavialis lies in close examination of the braincase. This Study The purpose of this study is to describe the skull of a juvenile (near hatchling) Gavialis gangeticus using high-resolution Computed Tomography (CT) scanning to image the specimen with particular attention to the cranial pneumaticity. Because of the endangered status of G. gangeticus, access to specimens has been extremely limited, and the addition of this hatchling could prove pivotal in parsing out solutions to the phylogenetic issues of Crocodylia. The pneumatization of the posterior portion of the crocodylian skull has been known for almost two centuries, yet current technology has yet to be employed to facilitate accurate restoration of these passages. High resolution CT scanning has made it possible to visualize the internal structures of skulls without having to destroy the specimens (Conroy and Vannier 1984). A planar fan of x-rays passes through the specimen and is picked up on the other side by detectors as it rotates (Clark and Morrison

5 1994). Variation in the ability of the tissues and sediment to absorb x-rays creates a differentiated image of a two-dimensional slice of the specimen. Taking the twodimensional slices of specimens at regular intervals and splicing those images together allows researchers to recreate a three-dimensional digital replica of the specimen. The precision of these scans allows researchers to gather accurate distance and volumetric data from the images obtained (Hounsfield 1973; Clark and Morrison 1994). This technology has been used to better understand embryological development and to recreate soft tissues in extinct organisms, including pneumatic recesses and the brain and surrounding tissues (Witmer 2008). Specialized computer software takes the collection of two-dimensional slices and allows the user to analyze them simultaneously as a three-dimensional model (Hounsfield 1973; Clark and Morrison 1994). The employment of CT scanning allows paleontologists to access the delicate interior structures of skulls without destroying rare specimens. Once the digital braincases are exposed, they can be analyzed in many ways. Geometric morphometrics is an ideal method for assessing if and how the shape of the braincase changes through ontogeny in these animals. Morphometrics provides a way to quantify shape change using simple software packages and statistics. Initially, morphometric data included linear measurements (e.g., length, width, height) and ratios. (Zelditch et al. 2004; Adams et al. 2004). Because many of these data either overlap or originate from the same point on the specimen, the issue of non-independence made this technique questionable (Adams et al. 2004). Also, by using these types of data, the original shape of the organism was lost during the analyses, and with it, potentially valuable information (Zelditch et al. 2004). Discussion of these problems led morphometricians to develop a different technique geometric morphometrics whereby the coordinates of homologous landmarks are analyzed instead of the linear measurements that connect them (Zelditch et al. 2004). There are three types of landmarks: a Type 1 landmark is placed at the

6 intersection of three tissues, Type 2 is placed at the maximum or minimum of curvature, and Type 3 is placed at the furthest point from a reference. Mathematically and biologically Type 1 landmarks are the more stable and most likely to be homologous, whereas Type 3 are the least stable (Zelditch et al. 2004). Using landmarks allows for the use of two-dimensional and three-dimensional data, while still being able to apply the same statistical techniques (Zelditch et al. 2004). A principal components analysis (PCA) can be performed once the landmarks are chosen and digitized. This analysis reduces the dimensionality of the data by redrawing the first axis through the region with the most data, then drawing axes orthogonal to the first one. Each axis represents a complex variable and clustering patterns along these axes are easily visualized (Zelditch et al. 2004). This study analyzes homologous landmarks in ten crocodylians (Figure 2) of varying age with Procrustes superimposition, principal components analyses, and nonparametric statistical tests. Methods for applying these data to cladistics analyses are still in their infancy, but are being actively pursued (Goloboff and Catalano 2010; Catalano et al. 2010).

Figure 1: Phylogenetic trees showing relationships among crocodylians using molecular data (right) and morphological data (left) (from Brochu 2003). 7

Figure 2: Phylogeny of specimens in this analysis (based on Brochu 2003). Red are gavialoids, blue are alligatoroids, green is Tomistoma, and yellow are crocodyloids. 8

9 CHAPTER 2: MATERIALS AND METHODS Specimens The specimen scanned here is YPM HERR-008438 (Figure 3, Figure 4). This specimen is a captive-bred juvenile Gavialis gangeticus preserved in 10% formaldehyde and 70% alcohol. It is of near-hatchling age, which makes it a valuable addition to the ontogenetic CT data of Crocodylia. It was CT scanned at the high-resolution CT scanning facility at the University of Texas at Austin in August of 2010. This specimen had ingested large sedimentary particles upon death, which have been segmented out of many of the images. Usually, more than one specimen is needed to get a full osteologic description. In this case, the specimen (YPM HERR-008438) is in exceptional condition due to the preservation used. The only potential bias is that this specimen was captive-bred, and therefore may have morphological features that are not present in wild specimens (Sadlier 2009). Additional juvenile specimens of G. gangeticus were not scanned, so this bias remains untested. Other specimens used for the morphometric analyses include an adult Gavialis gangeticus (TMM-M-5490), Caiman crocodilus (FMNH-73711), Alligator mississippiensis (two specimens: TMM-M-6723, and an uncatalogued specimen, also from TMM), and Eosuchus minor (NJSM-15437), Tomistoma schlegelii (TMM-M- 6342), Osteolaemus tetraspis (FMNH-98936), Mecistops cataphractus (TMM-M-3529), Crocodylus johnstoni (TMM-M-6807), and Crocodylus moreletii (TMM-M-4980). Table 1 includes the specimens and their ages. The relationships of these taxa can be seen in Figure. These specimens were previously scanned at different facilities for other projects. Scanning Parameters Specimen YPM HERR-008438 was scanned on August 23, 2010 by Jessie Maisano. It was scanned in the coronal plane using X-ray energy of 200kB and 0.18mA.

10 The slice thickness was 0.05573mm, with a field of reconstruction of 26mm and a pixel dimension of 1024 by 1024. In total, 1680 slices were processed. These included the entire skull and the first three vertebrae as well as the postoccipitals and the first two pairs of osteoderms. Due to the excellent preservation of this specimen, all of the soft tissues of the head are also present in the scan. The CT slices can be made available by the author. Images of the hatchling G. gangeticus were produced by Matt Colbert of the University of Texas at Austin in VGStudioMax 2.0 and were labeled in Adobe Illustrator CS5 using the standard conventions of bone in red, other structures (teeth, fenestrae, etc.) in yellow, nasal passages in purple, and Eustachian passages in blue. All abbreviations follow Iordansky (1973). Image processing of the other specimens was done using Avizo 6.3. These were also labeled in Adobe Illustrator CS5, but bones and sutural interpretations were labeled in black, with less certain sutures indicated by dashed lines (Appendix A). The skull of this and other specimens were cropped along the midsagittal plane and 3-dimensional digital copies were rendered. These skulls were used for comparison and also for the morphometric analysis. ImageJ 1.43u (Rasband 2011) was used as a quick slice visualization tool. Osteological Descriptions Cranial descriptions follow the basic model of Iordansky (1973) and Walker (1990). Each bone is described in the order it appears from anterior to posterior. All surfaces are characterized, followed by foramina and any protrusions. Comparisons were made primarily with an adult G. gangeticus (TMM-M-5490), which was scanned in 1996, but also with published stem gavialoids (Brochu 2006a and b; Delfino 2005, 2010; Velez-Juarbe et al. 2007).

11 Morphometric Analyses In order to view the braincase, the skull had to be cropped along the midsagittal plane. This way, the internal sutures of the braincase are visible and can be analyzed with landmarks. A preliminary study analyzed the braincase utilizing the right lateral view. However, this view demonstrated that the quadrate and pterygoid obscure a majority of the sutures critical to this analysis. Therefore, a midsagittal view was chosen to maximize the number of landmarks and increase the power of these tests. The computational limitations were such that the entire skull could not be rendered at once, so skulls were sectioned away rostral to the frontal nasal suture. This permitted simultaneous analysis of the entire length of the braincase while decreasing computation usage. Midsagittal images of each braincase were created using the Isosurface function in Avizo, then cropping the surface down the midsagittal plane. Images were then taken using the screencapture function. A scale bar was included in each of the images using the 2- dimensional measure tool to make a length of 10 units. Images were then processed using Adobe Photoshop CS5 to standardize the background, and Adobe Illustrator CS5 to label bones and add landmarks. Using ImageJ 1.43u (Rasband 2011), I digitized 16 landmarks (12 Type 1, 4 Type 2; Table 2; Figure 7) on each of the ten specimens (Table 1). Eosuchus was not included in this morphometric analyses because its braincase was filled with sediment and Avizo would not allow me to keep portions of the braincase rendered while digitally removing the sediment. In addition, portions of its braincase were missing, and I was unwilling to remove landmarks to be able to include it. The landmark coordinate data were compiled into a tps file and run through CoordGen6 (Sheets 2010). Landmarks were superimposed using Procrustes superimposition, and the Procrustes coordinates were saved into a new file. Procrustes superimposition was chosen over Bookstein shape coordinates (BC) and sliding baseline registration (SBR) because of two factors: 1) I am looking at

12 midsagittal views of the braincase, so there is no convenient baseline (e.g., a plane of symmetry) to use for BC and SBR, and 2) it made more biological sense to allow all landmarks to vary, thereby capturing the movement of individual landmarks, instead of forcing two landmarks into not varying and potentially losing information. The Procrustes coordinates were then used in a principal components analysis (PCA) using PCAGen (Sheets 2010). Principal component plots and deformation grids were imaged using this program as a baseline for comparison with other statistical programs. In comparing these plots, I noticed that the IMP software mirrors the landmark scheme, and rotates it 30 clockwise in showing the Procrustes superimpositon. The thin plate splines are equally affected, but labeling each landmark on these diagrams should ameliorate the problem of interpretation. The statistical package SAS 9.2 was used to perform a PCA and a Kruskal-Wallis test using the raw Procrustes coordinates as the initial data instead of exporting the PC scores from IMP. To use a parametric test of variance, the data would need to be normally distributed; whereas a non-parametric test ranks the data, which allows for testing data that do not follow a normal distribution. The small number of specimens in this study precluded parametric tests; hence, the non-parametric Kruskal-Wallis test was used. The PC scores were used because insignificant PCs could be overlooked. If partial warp scores were used, all of the axes would need to be interpreted. This allowed me to compare the results from the two analyses to check for any errors while also receiving axis loadings from the analysis. In order to include Eosuchus in the comparison, a separate morphometrics analysis was performed using one Type 1 landmark and four Type 2 landmarks as well as sliding semilandmarks in between these (19 landmarks total; Table 3; Figure 9) on the midsagittal slices (Appendix B). TpsUtil 1.46 (Rohlf 2010) was used to create a tps file, then TpsDig 2.16 (Rohlf 2010) was used to digitize the landmarks. After that, TpsUtil was employed once more to make the sliding semilandmarks file. Finally, RelW 1.49

13 (Rohlf 2010) was used to perform the relative warp analysis and create the plots and thin plate splines. This analysis was repeated twice because the first iteration showed Caiman to be an outlier. The second iteration removed Caiman from the analysis. The results from both iterations are presented in Chapter 4.

14 Table 1: Specimens used in this study, with the date of the scan and age of the specimen indicated. Species Gavialis gangeticus Gavialis gangeticus Caiman crocodilus Alligator mississippiensis Alligator mississippiensis Tomistoma schlegelii Osteolaemus tetraspis Crocodylus johnstoni Mecistops cataphractus Crocodylus moreletii Eosuchus minor Note: Indicates extinct. Common Name Indian Gharial Indian Gharial Specimen Number YPM HERR- 008438 TMM-M- 5490 Date Scanned Age of Specimen Specimen number in Analyses PCA RW 8/2010 Hatchling 1 5 2/1996 Adult 2 4 Caiman FMNH-73711 2002 Adult 3 11 Alligator Uncatalogued 2005 Hatchling 4 9 Alligator TMM-M- 6723 TMM-M- 6342 1999 Hatchling 5 10 2005 Adult 6 8 FMNH-98936 2005 Adult 7 7 TMM-M- 6807 False Gharial Dwarf Crocodile Freshwater Crocodile Slendersnouted Crocodile Mexican Crocodile Crocodile Dawn TMM-M- 3529 TMM-M- 4980 1996 Adult 8 2 2005 Adult 9 1 2003 Adult 10 6 NJSM-15437 N/A Adult N/A 3

15 Table 2: Landmark numbering scheme and descriptions. Landmark Number Landmark Type Explanation 1 Type 1 Lateral suture of eo and bo 2 Type 1 Intersection of medial suture of bo and bs with cutting plane 3 Type 2 Intersection of bsr with bs (apex of the angle formed) 4 Type 2 Ventral maximum curvature of bsr 5 Type 2 Dorsal maximum curvature of bsr 6 Type 1 Anterior suture of bs with ls 7 Type 1 Ventral suture of pot with ls 8 Type 1 Ventral suture of pot with bs 9 Type 1 Suture of bo with cranial nerve canal 10 Type 1 Intersection of pot and ls suture with cranial nerve V opening 11 Type 1 Suture of ls with f 12 Type 1 Dorsal suture of pot with ls 13 Type 1 Suture of pot with so 14 Type 1 Suture of pot with eo 15 Type 1 Suture of so with eo 16 Type 2 Caudalmost midsagittal point on skull Note: Abbreviations as in List of Abbreviations. Table 3: Landmark explanations for the second morphometric analyses. Landmark Number Landmark Type Explanation 1 Type 1 Suture between pt and bs 2 Type 2 Maximum curvature of met 3 Type 1 Suture between bs and bo 4 Type 2 Maximum curvature of metp 5 Type 2 Maximum curvature of ventral bo Note: Abbreviations as in List of Abbreviations.

Figure 3: Gavialis gangeticus, specimen YPM HERR 008438, in dorsal view. 16

Figure 4: Gavialis gangeticus, specimen YPM HERR 008438, in ventral view. 17

18 Figure 5: Landmark scheme for 2-dimensional morphometric analyses. Figure 6: Landmark scheme for second morphometric analyses. Red dots are Type 1 and Type 2, blue dots are sliding semi-landmarks.

19 CHAPTER 3: CRANIAL OSTEOLOGY OF Gavialis gangeticus General Description General Shape and Characteristics Gavialis gangeticus is one of the primary examples for the longirostrine (long and slender snouted) condition within Crocodylia. The extensive premaxillae and maxillae make up most of the rostrum, and the maxillae only contact the rest of the skull near the orbits. The lower jaw is equally long and thin, with the two rami of the dentary and splenials joined until approximately two-thirds the way down the mandible. The braincase only occupies the caudal third of the skull and because of the young age of the specimen, some structures occupy relatively more space in the braincase than they do in adults (e.g. the semicircular canals). Also, the bone in general is spongier and thinner than in the adult. The skull table in the adult is much more planar, though it maintains a slight downward slope away from the midline. In the juvenile, it has a distinct ventral curve. Finally, and unlike Crocodylus, the frontal bends dorsally between the orbits, displacing the eyes upward relative to the nares. In other Recent crocodylians, the eyes and nares lie on the same plane. Cranial Openings The nares are ovoid in dorsal view (Figure 7) with the smaller end pointing posteriorly. Laterally, the nares are only slightly raised above the dorsal maximum of the rostrum. The foramen incisivum is triangular in ventral view. Both the nares and the foramen incisivum are completely enclosed by the premaxillae. The internal choanae are contained within the pterygoids and are situated at the midpoint of the conjoined pterygoid wings. The orbits are large: half the length of the last third of the skull. They are circular in dorsal view, but triangular in lateral view. They are rimmed by the lacrimals, prefrontals, jugals, postorbitals, and frontals. In the adult, each of these bones forms a

20 raised ridge along the orbit margin ( telescoped orbit ); whereas, in the juvenile it is much less pronounced. Also, the juvenile orbit is proportionally larger than that of the adult. Crocodylians follow the basic diapsid condition, possessing two pairs of temporal fenestrae: the supratemporal and infratemporal. The supratemporal fenestra is anteroposteriorly elongated, forming an oval opening that spans neary the length of the parietal. It narrows mediolaterally as it deepens ventrally. In the adult form, this fenestra is more D shaped, with the flat portion on the lateral side, and the curve forming the medial edge. The infratemporal fenestra is triangular and bordered by the squamosal, postorbital, jugal, and quadrate. It does not change shape during ontogeny. These two fenestrae expose the adductor chamber. The posttemporal fenestrae are present, but only as thin slits on the posteriormost dorsal surface of the skull. They open caudally. The foramen magnum is large and almost circular, forming an opening about one-third the total height of the skull. There are several openings in the back of the skull for the vagus and other cranial nerves, and Eustachian passages (discussed below). The mandibular fenestra (Figure 8) of the lower jaw comprises the dentary at its rostral margin, the surangular at its dorsal edge, and the angular ventrally. On the inside of the mandible, the foramen intermandibularis medius should sit at the suture between the coronoid and splenial, but cannot be exposed in this specimen due to the configuration of the jaw. Regions The skull is divided into three basic regions: the external skull elements; the braincase; and the mandible. Each of these regions contains multiple elements that are described in detail in the sections below. The nasal passages, pneumaticity, and dentition are characterized separately as they cross element boundaries. The hyoids are also described in a separate section, as they did not fit in with the rest of the regions.

21 Elements Premaxilla The two premaxillae (slices 19 413; Figure 7) as in all other crocodylians, exclusively form the front of the snout. The premaxillae form an extended diamond shape, both dorsally and ventrally, vaguely like a spear tip, with the tip facing caudally, forming an oblique scarf joint with the maxillae bracing each side. On the dorsal surface, the premaxillae enclose the naris, forming a single opening with no internarial bar. On the ventral surface (Figure 9), these encircle the foramen incisivum, and form the anteriormost portion of the hard palate. Where the two premaxillae meet on the ventral midline, they form a dorsoventrally straight contact. The premaxillary contact with the maxilla begins medially, and distally expands laterally until the premaxillae are only present as thin plates dorsally and ventrally. At this point, the contacts with the maxilla are jagged. The premaxillae pinch out ventrally prior to their disappearance on the dorsal surface. The dorsolateral surface of the premaxilla is pitted (Figure 8), with a distinct deeper pit on each half directly rostral to the naris. In other crocodylians, these correspond to pits on the ventral surface that accommodate dentary teeth (Iordansky 1973), though, in G. gangeticus, they only appear on the dorsal surface. The ventral surface is smooth, with the foramen incisivum opening rostral to the anterior edge of the nares. Each premaxilla contains five teeth. The medialmost tooth from each premaxilla projects from the rostromedial suture. Distally, they are separated from the next pair of teeth by the first dentary tooth. Following those pairs, each premaxillary tooth is interfingered with one from the lower jaw. Each premaxillary tooth is curved rostroventrally. This reflects the adult dentition of G. gangeticus (Iordansky 1973). Maxilla The paired maxillae (slices 211 1195; Figure 7, Figure 8) and contact each other at the midline (both dorsally and ventrally), except at points of contact with other bones,

22 such as the premaxillae anteriorly and nasals posteriorly, where the maxillae diverge from each other, forming a general x shape. Together with the premaxillae, the maxillae form the majority of the upper jaw. The width of the upper jaw stays constant posteriorly until approximately two-thirds of the distance of the maxillae, where it begins to widen. The dorsal surface is lightly pitted; whereas, the dorsolateral surface is pitted more deeply, with the pits concentrated above the alveoli. The ventral surface is smooth, as in the premaxillae, and continues to form the hard palate. The ventromedial contact of the maxillary palatal processes is dorsoventrally zig-zagged anteriorly and straightens out posteriorly. There are 24 teeth on each side, which interfinger with teeth from the dentary. Contained within the conjoined maxillae is the internal nasal passage (Figure 10). The maxillae contact the nasals in a scarf joint, and the jugals in an oblique butt joint that thins posteriorly. Ventrally (Figure 9), the maxillary palatal laminae contact the palatines in a dorsoventrally flat, but laterally angled contact. The maxilla does not separate the two tips of the prefrontal, as observed by Delfino (2010) in at least one specimen of G. gangeticus. Nasal The nasals (slices 593 942, Figure 7) are paired, and together form a cartoon heart-shaped element with the tip of the heart pointing toward the front of the snout and the curved portions projecting posterolaterally from the midline. In cross-section, the nasals are thin and plate-like rostrally and mediolaterally curved posteriorly. Laterally, the nasals contact the maxillae in a butt joint (Figure 8). The posterior curved edges contact the lacrimal, prefrontal, and frontal bones in butt joints. In comparison to stem gavialoids (Eosuchus lerichei and E. minor), the nasals in Gavialis are highly reduced, reaching only to the 12 th maxillary alveolus anteriorly, versus reaching the second maxillary tooth in E. lerichei (Delfino 2005) and actually contacting the premaxillae in E. minor (Brochu 2006).

23 Palatine The paired palatines (slices 908 1248; Figure 9) begin as a tapered wedge between the two maxillary palatal processes. The beginning of this taper seems to correspond to the point on the lower jaw where the two rami begin to diverge. They are thin plates of bone that curve slightly dorsolaterally to form the ventral portion of the nasopharyngeal duct. The two palatines bulge where they meet at the midline and their contact is dorsoventrally straight. Posteriorly, the midline contact decreases in height, thereby forming a single passage for the nasopharyngeal duct. The palatines flatten out towards their oblique, scarf joint contact with the pterygoid. Vomer The vomers exposed on the ventral palatal surface in some crocodylians, for example, Tomistoma schlegelii (Iordansky 1973), are not exposed on the palatal surface in G. gangeticus. A more thorough examination of the palate was not in the scope of this study, so a detailed description of the vomer is not included here. Pterygoid The pterygoids (slices 950-1367; Figure 9) are sutured to the palatines anteriorly, the ectopterygoids laterally, and the basisphenoid and basioccipital dorsally. Each pterygoid flange flares ventrolaterally and sutures in an acute angle with the ectopterygoids, which also make up part of the flange. The maximum ventral descent of the pterygoid flanges is just over 100 degrees ventral from the midline. As the ventral processes flare out, the dorsal surface extends dorsally to meet the basisphenoid and basioccipital. These sutures provide extra support for the braincase, keeping it stabilized with the rest of the skull. The internal choanae are circular and have no midline processes. They are fully enclosed by the pterygoids and sit about halfway back from the palatine-pterygoid suture. The pterygoids are relatively smooth on all surfaces.

24 Ectopterygoid The ectopterygoids (slices 1114 1322; Figure 9) begin as small ovals on either side of the cheek on the medial side of the jugal. They do not contact the maxilla as they do in C. niloticus. They expand posteriorly to form wedge-shaped bones in cross section, with the thicker portion of the wedge suturing vertically with the jugal, and dorsomedial to that, suturing in a scarf joint with the postorbital to form the postorbital bar. Posterior to this, the ectopterygoids continue medially, becoming more horizontal. Just before they reach a horizontal orientation, they contact the pterygoids medially (slice 1254). Posterior to this contact, the ectopterygoids begin to taper as they flare ventrally with the pterygoids, creating the pterygoid flange. Jugal The jugal (slices 932 1400; Figure 7) is elongate and though it is more plate-like (taller than wide) rostrally, posteriorly it becomes more of a rod (circular in cross section), especially near its contact with the quadratojugal. It sutures rostrally with the maxilla and lacrimal with a forward plate-like projection. It forms the ventrolateral edge to the orbit, bulging dorsally and forming a wall as it supports the eye. In the adult, the jugal has a more pronounced dorsal expansion that forces the eye into a more dorsally oriented opening. Behind the orbit, the jugal has a nearly vertical process that is triangular in cross section, which sutures with the postorbital to complete the orbital margin. The jugal forms a convex surface that fits into nook formed by a concave surface of the postorbital. The postorbital also has two rounded descending processes that brace each side of the nook it creates. Its posterior extension forms the ventral margin of the infratemporal fenestra and sutures with the quadratojugal medially within this opening. The jugal does not form part of the articulation with the lower jaw.

25 Quadratojugal The quadratojugal (slices 1280 1445; Figure 8, Figure 11) extends posteriorly from the infratemporal fenestra and, along with the jugal, forms the lateral part of the quadrate ramus. Anteriorly, it is thin and plate-like with two rami, one forming the postero-dorsal edge of the infratemporal fenestra and the other forming the ventral edge. There is a thin, spur-like projection of bone at the intersection of the rami, extending rostrally and bisecting the posterior tip of the infratemporal fenestra. Posterior to this fenestra, the quadratrojugal becomes thicker, with a D shaped cross-section, and is sandwiched between the jugal and quadrate. At its final point, it forms the posterolateralmost point of the skull. Quadrate The quadrate (slices 1250 1458; Figure 8, Figure 12) is a very complex element. At its anterior portion, it forms a dorsally concave feature that sutures with the parietal and squamosal, creating the internal surface of the supratemporal fenestra. As it curves posteriorly, it becomes more robust, and finally forms the articulation surface for the lower jaw. Dorsal to the articulation, the quadrate sutures once more with the squamosal and closes off the posterior edge to the temporal arcade. More medially, the quadrate sutures to the braincase overlapping the laterosphenoids, the lateral exposure of the prootics, and the exoccipitals. At its ventromedial edge, the quadrate sutures to the pterygoid to complete the internal support for the bones of the braincase. Caudally and medial to the lower jaw articulation, the quadrate expands medially to suture with the ventral edge of the exoccipitals and with the basioccipital, forming the ventrocaudal wall of the skull. Within the temporal arcade, the quadrate surrounds the opening to the otic region of the skull (Figure 12). The caudal wall of the quadrate contains the openings for cranial nerves X XII (Figure 13). There is also a small, circular foramen near the articulation

26 with the lower jaw that corresponds to the foramen aerum on the articular. Lastly, the quadrate forms the lateral edges of the foramen magnum. Lacrimal The paired lacrimals (slices 877 1044; Figure 7) each form a triangular wedge, with the tip contacting the prefrontal medially and the maxilla laterally. These bones also contribute to the orbit, with the thickest side forming the rostral edge of the fenestra. Like the prefrontals, the dorsal and ventral surfaces are smooth. There are three foramina located on the inside edge of the orbit, one of which is much larger than the other two, that are relatively equal sized. These are the lacrimal ducts, which form a tunnel from the orbits to the nasal passages (Iordansky 1973). The lacrimal is about twice the length of the prefrontal. Travelling laterally, each lacrimal contacts the nasals, maxilla, and jugal. Prefrontal The paired prefrontals (slices 913 1062; Figure 7) lie lateral to the frontal and medial to the lacrimals, creating the anteromedial margin of each orbit. The dorsal surface, where the bone contacts the orbit, becomes rugose, but not pitted as in the upper jaw. The prefrontals are generally wedge shaped, with the thickest portion forming the orbit. Ventrally, the prefrontals extend and create the prefrontal pillar and contact the corresponding palatine at its widest point. The two pillars do not meet at the midline, though they have small, medially extending processes. There are three foramina in the prefrontal: one on the dorsal surface and two on the inside edge of the orbit. These are most likely nutrient foramina. Rostrally, they contact the corresponding nasal. Laterally, the prefrontal contacts the lacrimal. Caudally, each contacts the parietal in a jagged scarf joint. Palpebral Both palpebrals (slices 992 1018; Figure 7, Figure 8) are preserved in this specimen. Each is thin and platelike, forming a semicircle of bone contained within the eyelids. In resting position, they sit dorsally to the prefrontals without contacting any

27 other bone as they are encased in soft tissue. The palpebrals are relatively simple structures, with no foramina or other such surficial ornamentation. Frontal The single frontal bone (slices 852 1264; Figure 7) forms a bridge between the upper jaw and the braincase, supporting both orbits on either side. It has a smooth dorsal surface, except in between the orbits, where the surface becomes pitted. There is also a slight ridge that forms at the midline between the orbits. As it slopes upwards towards the skull roof, it thickens to form the medial rim of each orbit. The ventral surface is smooth. Midway through the orbits, the frontal descends into two ventral processes (Figure ). It is within these processes that the rostralmost extension of the brain, the olfactory tract, sits (Iordansky 1973). It contacts both nasals rostrally, the prefrontals laterally, and the parietal caudally in a jagged butt joint. Along the midline of the frontal parietal suture, there is a distinct notch where the parietal invades the frontal. Caudal to the orbits, the frontal contacts the postorbitals in a zig-zagged (in dorsal view) sutural surface. The frontal does not participate in the supratemporal fenestrae. Postorbital The two postorbitals (slices 1162 1314; Figure 7, Figure 14) form supports for the orbits and the supratemporal and infratemporal fenestrae. The postorbital forms a medially-curved, T shape, with each of the top branches forming the anterior edge to the supratemporal fenestra. It contacts the frontal and parietal medially with a suture that is firm due to the robustness of the postorbital at this edge. It also has a lateral oblique suture with the squamosal, forming the anterolateral margin of the supratemporal fenestra. The descending process (Figure 14) forms the dorsal portion of the postorbital bar, and it contacts and sutures with the jugal at its base, forming the back margin of the orbit and the anterior margin of the infratemporal fenestra. The dorsolateral surface is slightly pitted.

28 In some specimens of Gavialis bengawanicus (Delfino 2010), the postorbital is pentagonal in dorsal view, though in others, it is shaped more like that found in the hatchling specimen of G. gangeticus. Parietal The parietal is a single bone (slices 1245 1439; Figure 7) forming part of the skull roof. In dorsal view, the basic shape is that of a rectangle with pinched-in sides, like a fat hourglass. It is almost as wide mediolaterally as the frontal, though the posterior edge of the parietal is less wide than the anterior edge. It has ten distinct, deep, bilaterally symmetrical pits placed on its dorsal surface. The ventral surface is smooth and forms part of the roof of the endocranial cavity. Laterally, the parietal forms the medial margin of the supratemporal fenestrae, expanding ventrally into the opening. This bone contacts the frontal and postorbitals rostrally, the supraoccipital caudally, and the squamosal caudolaterally. Squamosal Each squamosal (slices 1261 1471; Figure 7, Figure 11) has three processes that are orthogonal to each other, with the origin forming the posterolateral corner of the skull table. These corners are more rounded than they are in the adult. The longest process is plate-like and extends rostrally to contact the postorbital and form the lateral margin to the supratemporal fenestra. The dorsoventral process forms the back lateral edge of the skull table and contacts the quadrate at its vetral-most edge. The mediolateral process contacts the exoccipital and the supraocciptal medially, and forms the dorsocaudal edge of the skull table. Braincase Laterosphenoid Due to the youth of this specimen, the laterosphenoids (slices 1180 1338; Figure 15) are barely ossified. In the adult, they are triangular in lateral view, with a distinct posterior ridge that is medially displaced relative to the rest of the element. This ridge

29 runs dorsoventrally. In dorsal or ventral view, the laterosphenoids form two semicircular cups that protect the optic lobes of the brain (Figure 15). In the juvenile, the cup portion of the laterosphenoids has not yet fully ossified, though a thin wisp of bone is present. The posterior ridge is ossified completely in the juvenile. The laterosphenoids form the anterior margin of the opening for cranial nerve V. It should also contain the openings for cranial nerves II IV, but due to its incomplete ossification, these are not yet visible. Each laterosphenoid is firmly sutured to the frontal and parietal bones dorsally, basisphenoid rostrally, and quadrate laterally. Basisphenoid The single basisphenoid (slices 1210 1396; Figure 16) is elongate and oriented transversely. The midsagittal basisphenoid rostrum is developed. In the adult form, the basisphenoid rostrum adopts a somewhat anterodorsal orientation, pointing more toward the frontals than in the juvenile, where the rostrum points to the tip of the snout. At the intersection of the laterosphenoid and basisphenoid, there is a wide groove, the hypophyseal fossa, in which the pituitary gland sits. This groove is not as visible in the juvenile, and it may be that as the basiphenoid rostrum tilts upwards it hollows out the groove with its movement or that the groove becomes more recessed as the hypophysis grows into it. The basisphenoid forms a dorsally concave structure that protects the bottom anterior portion of the brain. Beneath the basisphenoid, the pterygoids come up and suture with both the basisphenoid and the basioccipital, forming a brace for the bottom of the braincase. The basisphenoid also sutures with the laterosphenoids, prootics, quadrates, and basioccipital. Prootic The prootic (slices 1319 1415; Figure 15, Figure 16) is one of three bones that surround the otic capsule. It has a complex shape that is triangular overall, but with a bend in the middle that gives it an s shape when viewed from the side. The prootic

30 plays host to many cranial nerve openings. Three of these, for cranial nerves VII (one opening) and VIII (two openings), are bunched near the center of the element. The foramen for cranial nerve VII opens laterally, whereas the other two open down and forward. Its anterior suture with the laterosphenoid also features an opening for cranial nerve V, with the laterosphenoid at its anterior border, and the prootic forming the rest of it. Posteriorly, the prootic sutures with the exoccipital. This suture is not a firm one because of a distinct groove that runs along the suture, forming a space for the cranial nerves. This opening may be filled with cartilage in life and may also slowly ossify during ontogeny, but never close completely. In lateral view of this opening, the prootic is only visible as a small tongue that spills ventrally from the trigeminal foramen. The rest of this element is obscured by the quadrate in lateral view. Dorsally, it sutures with the parietal and the supraoccipital. Supraoccipital The supraoccipital (slices 1360 1482; Figure 7, Figure 8) is triangular in lateral view and in caudal view. It forms the dorsal element to the otic capsule and in the junction between this element, the prootic, and the exoccipital, there is an opening instead of a distinct suture (Figure ). In dorsal view, the supraoccipital has a limited exposure, with a pronounced, rounded projection into the parietal. The dorsal suture of the supraoccipital is Y shaped, with the two outer apices pointing to the suture between the squamosals and the parietal, and the internal apex forming the projection into the parietal. In caudal view, the two posttemporal fenestra lie along the dorsal suture of this element with the parietal, though the squamosal contributes to the lateral margin of each fenestra. Exoccipital The exoccipital (slices 1419 1493) forms the caudal surface of the skull. It starts as a limited element within the braincase that articulates with the basioccipital, prootic, and supraoccipital (Figure 15). It expands caudally and laterally, and connects the

31 braincase with the dermal bones of the rest of the skull with sutures to the quadrate and squamosal. Midsagittally, it houses internal foramina for cranial nerves IX, X, and XII. On its external, caudal surface, the jugular vein and cranial nerves X, and XII exit the skull (Figure 15). Where the two exoccipitals meet at the posteriormost end of the skull, they are plate-like and form most of the occipital surface. At the ventralmost portion of their midline suture, they create the opening for the foramen magnum (along with the basioccipitals and quadrates), which is nearly circular. Inside the braincase (slices 1422-1455), along the dorsal and caudolateral edges, there is a pair of ossifications that are plate-like in sagittal view, but oblong in coronal view. These are probably portions of the exoccipital, which later co-ossify to form a single element on each side of the skull. This ossification is not the footplate of the stapes because the stapes medial termination lies rostral and ventral to the starting position of this ossification. Basioccipital The single basioccipital (slices 1408 1503; Figure 15) forms the final portion of the braincase. It suture with the basisphenoids anteriorly to structure the bottom of the braincase, and with the exoccipitals to create the foramen magnum. This element is spoon-shaped, with the concave-up surface contacting the brain, and the ventral convex surface forming the base of the skull. The caudal-most surface of the basioccipital forms a rounded projection, the occipital condyle, which articulates with the beginning of the vertebrae. Sagittally, the occipital condyle is rounded, but appears oval-shaped in coronal sections (Figure 15). The rest of the element is exposed ventrally to the occipital condyle in caudal view.

32 Mandible Dentary The dentary (slices 20 1255; Figure 7) is the only tooth-bearing element of the lower jaw. The two rami of the dentary are firmly sutured at the midline (slices 20 571) and diverge posteriorly when they encounter the wedge-shaped paired splenials (Figure 17). The dentary remains relatively uniform in width up to the point of divergence (slice 954). Medially, it contacts the coronoid at its caudal interior edge. At its posterior edge, it forms the margin for the mandibular fenestra, and sutures with the surangular dorsally and angular ventrally. Splenial The splenials (slices 571 1352) are exposed ventrally, but are most prominent medially, where they form vertical plates that suture with the dentary and with each other at the midline. In cross-section, each splenial is bracket shaped, with the short arms of the bracket supporting the dentary, and the dentary fitting into the U shaped canal in the center (Figure 14). The splenials become thin plates posteriorly along the medial surface of the dentaries until they taper off below the coronoids. The foramen intermandibularis oralis is present just prior to where the two tami of the mandible diverge. It is ovoid with the smaller end pointing anteriorly. The ventral surface is mostly smooth with a few distinct pits near the point of lateral expansion. Angular The angular (slices 887 1601) is the ventralmost mandibular element beneath the skull table. In lateral view, it is elongate and curved dorsally at each end (Figure 7). The whole element is curved medially. Rostrally, it sutures with the back, ventral edge of the splenial. At this suture lies the foramen intermandibularis caudalis. Its shape could not be accurately assessed due to software limitations, but its location is certain. It forms the ventral edge of the mandibular fenestra, and posterior to this opening, it sutures with the surangular for the rest of its length, finishing in a flat point at the caudal most edge of

33 the mandible. Medially, the angular is curved inwards and up, forming a groove on the inside of the lower jaw. This curvature dissipates posteriorly. In the adult, the curvature of the angular is more pronounced and the element is dorsocaudally expanded along with the articular. Surangular The surangular (slices 1276 1561) is robust and elongate. It contacts the dentary rostrally in a scarf joint, directly posterior to the last dentary alveolus. It curves dorsally along its length, with the apex above and slightly caudal to the mandibular fenestra. It forms the dorsal edge of this fenestra, with the angular forming the ventral margin (Figure 8). Posterior to the fenestra, the surangular sutures to the angular along the rest of its length. Also posterior to the fenestra, the surangular becomes more plate-like and sutures medially with the articular. At this point, the surangular becomes a pointed process between the bones that surround it. Coronoid The coronoids (slices 1229 1357) are extremely thin plates of bone that partially close off the mandibular groove. They are shaped like triangles, with the short edge of the triangle forming the anterior margin to the medial opening of the mandibular foramen. The point of the triangle projects forward into the splenial anteriorly. At the apex of the triangle lies the foramen intermandibularis medius. Like the foramen intermandibularis caudalis, its shape could not be accurately assessed due to software limitations, though its location is certain. Also, it is relatively smaller than the foramen intermandibularis caudalis. Ventrally, the coronoid sutures with the angular. Dorsally, the coronoid sutures with the surangular. Articular The articular (slices 1424 1596; Figure 8) is the posteriormost element of the lower jaw. It has two dorsal flat surfaces separated by a flat, pointed process. The rostralmost surface forms the articulation with the quadrate, and the pointed process helps

34 to stabilize this articulation. Posteromedial to the articulation lies a small pneumatic foramen, the foramen aerum, which continues its passage in the quadrate. The posterior flat edge forms the ventral surface of the back edge of the jaw. A ridge forms in the middle of the flat edge, which extends posteriorly, finally ending in a more square-shaped jaw tip. Ventrally, the articular is more wedge-shaped and triangular in cross-section. It sutures with the angular at its ventral margin. Laterally, it sutures with the surangular. Pneumaticity and Neurovascular System Eustachian System The Eustachian system is best viewed in transverse section. It begins as a ventral opening, the median Eustachian foramen that lies posterior to, and is much smaller than, the internal choanae (Figure 9). Dorsally, it opens into two chambers: an anterior chamber that proceeds into the basisphenoid, and a posterior chamber that moves into the basioccipital. Where these two passages meet and form a single chamber, the passage is horizontally extended, unlike in the adult specimen, where these passages form a slightly more verticalized opening (Figure 15). Both these tubes expand laterally to be about the width of the bone they occupy. At their widest point, the anterior and posterior Eustachian tubes each divide into two lateral channels that, with their lateral projections, connect each of the larger chambers to the rhomboid sinus. The Eustachian system does not progress dorsally past where these drain the sinus. Two lateral Eustachian foramina should also be present in the skull, but were difficult to locate. In comparison with the other specimens in this study, the two specimens of Gavialis have a much more horizontally extended medial Eustachian canal. The difference is especially visible in the specimens of Crocodylus, where the medial Eustachian passage is anteroposteriorly compressed and dorsally expanded. In hatchling Alligator, the medial Eustachian passage is still relatively horizontal in orientation, but still more horizontally compressed than in Gavialis. In Osteolaemus, the passage

35 illustrates a more intermediate condition of the two extremes shown in Gavialis and Crocodylus. Eosuchus possesses a medial Eustachian passage that is more similar to Osteolaemus than to either Gavialis or Crocodylus. Cartoid Artery Dorsal to the median Eustachian foramen, there is another system of passages (Figure 16). It begins as a pair of openings at the intersection of the basisphenoid rostrum and the laterosphenoids. This passage widens caudally into a large, almost square, chamber at the base of the braincase. Moving caudally, the chamber shrinks by having its ventral surface move dorsally, supposedly by expanding the basisphenoid dorsally. Posteriorlythis passage elongates and eventually splits into two circular tunnels that lead into the pneumatic space between the braincase and the outer skull. Dentition Gavialis hatches without erupted teeth (Whitaker, pers. comm.). The teeth erupt within the first three weeks post-hatching (Whitaker, pers. comm.). This specimen has almost a full complement of erupted teeth and therefore is around (or just over) threeweeks old. The premaxillae each contain three teeth. The first tooth on each premaxilla lies adjacent to each other at the anteromost midline suture of the snout. A tooth from the dentary occludes against the anterolateral tip of each premaxilla, separating the first and second premaxillary alveoli (Figure 7). Following this, each premaxillary tooth is interfingered with one from the dentary. The fifth premaxillary tooth is the largest of all the teeth in the series. There are 24 maxillary teeth that interfinger with dentary counterparts. They become thinner and more delicate posteriorly, though the last ten or so are more robust than those preceding them. There are 26 dentary teeth, some of which are not fully erupted. The first two teeth and the fourth are more robust than the teeth further back in the jaw, which are all relatively equal in size and shape. Each tooth projects laterally from its alveolus, before

36 curving upwards and back. Each tooth is curved and interfingers with the tooth from the opposing jaw, so that the jaw resembles a series of interlocking hooks. Hyoid Both hyoids (slice 1286 1450; Figure 7) are present in this specimen. They are elongate rods that bulge at each extreme and curve in medially, so that each forms a broad U. The hyoids would, in life, be parallel structures in the throat that support the pharynx. In this specimen, the throat is slightly compressed on one side, distorting the position of the hyoids. The hyoids are delicate, unattached structures that are usually not preserved in fossilized specimens due to taphonomic processes. Ontogenetic Variation The hatchling has a wider rostrum, larger orbits, and rostrocaudally expanded supratemporal fenestrae (Figure 18). The adult has a longer, mediolaterally thinner rostrum. The supratemporal fenestrae are more D shaped than ovular in the adult. The skull table is more angular, becoming almost rectangular in dorsal view. The two caudolateral points of the skull table become protrusions in the adult. The caudal third of the skull becomes more triangular shaped in the adult whereas the juvenile is more rounded. In lateral view (Figure 19), the rostrum becomes dorsoventrally compressed in the adult. The orbits are directed more dorsally. The infratemporal fenestra retains its basic shape from the hatchling, but becomes more angular. The caudal portion of the skull table becomes level with the rostral portion, thereby creating a flat dorsal surface. The curvature of the ventral margin of the angular becomes more pronounced in the adult. The teeth do not change in size or robustness, and therefore appear relatively thinner and smaller in the adult. The external mandibular foramen becomes more ovoid in shape, with the tip pointing anteriorly. Finally, the caudal tips of the articular expand dorsally.

Figure 7: Dorsal view of skull. Anterior at bottom, with bones in red and other structures in yellow. Abbreviations as in List of Abbreviations. Scale is 2 mm. 37

38 pm m la pp j f po sq qj q d orb an itf emf hy roe san ar Figure 8: Left lateral view of skull. Anterior at bottom, with bones in red and other structures in yellow. Abbreviations as in List of Abbreviations. Scale is 2 mm.

Figure 9: Palatal view of skull. Anterior at bottom. Bones in red, other structures in yellow, Eustachian system in blue. Abbreviations as in List of Abbreviations 39

Figure 10: Coronal section through the snout. Upper picture shows the location of the cut. Bones in red, other structures in yellow, nasal passages in purple. Scale is 2mm. Abbreviations as in List of Abbreviations. 40

Figure 11: Coronal section through supratemporal fenestrae. Upper right picture shows the location of the cut. Bones in red, other structures in yellow, Eustachian passages in blue. Scale is 2mm. Abbreviations as in List of Abbreviations. 41

Figure 12: Lower temporal region exposed. Upper picture shows location of cut. Bones in red, other structures in yellow. Scale is 2mm. Abbreviations as in List of Abbreviations. 42

Figure 13: Posterior view of skull. Bones in red and other structures in yellow. 43

Figure 14: Coronal section through the orbits. Upper right picture shows the location of the cut. Bones in red, other structures in yellow, nasal passages in purple. Scale is 2mm. Abbreviations as in List of Abbreviations. 44

45 XII eo so pot p ls f j pp n m pm VII V bo bs ec mep bsr metp met spl d Figure 15: Midsagittal cut of skull. Anterior at top, with bones in red, Eustachian system in blue, and other structures in yellow. Abbreviations as in List of Abbreviations. Scale is 2 mm.

Figure 16: Transverse section through braincase. Anterior at bottom. Upper picture shows location of the cut. Bones in red, other structures in yellow. Scale is 2mm. Abbreviations as in List of Abbreviations. 46

Figure 17: Ventral view of skull. Anterior at bottom. Bones in red, other structures in yellow. Scale is 2mm. Abbreviations as in List of Abbreviations. 47

Figure 18: Comparison between hatchling (top) and adult (bottom) Gavialis in dorsal view. 48

Figure 19: Comparison of hatchling (top) and adult (bottom) Gavialis in lateral view 49

50 CHAPTER 4: RESULTS FROM THE MORPHOMETRIC ANALYSES The principal components analysis in IMP (Sheets 2010) resulted in three significant axes that explained 47%, 26%, and 9% of the variation, respectively. The first PC separates the specimens based on size, with the smaller specimens on the positive section of PC1 and the larger ones on the negative side of the realized range; the two specimens of Alligator, the hatchling Gavialis, and Osteolaemus all fall out on the positive region of PC1 (Figure 20). Also along PC1, the adult Gavialis and the specimen of Tomistoma form a group along the negative range of both axes. Caiman clusters with the species of Crocodylus at the negative range of PC1 and positive range of PC2. It would be nice to think that PC2 was separating specimens based on snout shape (or variables in the braincase that are associated with snout shape), because the two Gavialis and Tomistoma cluster along PC2, but Mecistops cataphractus does not despite also being longirostrine. Therefore, snout shape does not play an obvious role in loading PC2. PC3 does not separate the specimens, instead keeping them clustered around the zero value for that axis (Figure 21). The plot of PC2 versus PC3 shows a distinct clustering of Gavialis and Tomistoma, and a grouping of the alligatoroid and crocodyloids, akin to the graph of 1 versus 2. Deformation along PC1 shows an elongation of the dorsal portion of the braincase relative to the ventral portion while also decreasing the distance between the anterior margin of the basisphenoid rostrum and the rest of the braincase (Figure 22). Deformation along PC2 results in stretching the midline suture of the basisphenoid and basioccipital away from the other landmarks (Figure 23). There is also minute shifting in the braincase causing the posterior half to move ventrally. Lastly, deformation along PC3 exhibits a shearing deformation where the ventral landmarks (numbers 1-10) moved caudally relative to the dorsal ones (landmarks 11-16) (Figure 24).

51 The SAS PCA results in six axes that described 95% of the variation (Table 5). In descending order, they describe 42%, 22%, 12%, 8%, 6%, and 3.5% of the variation, respectively. The plots resulting from this analysis are equivalent to those from the IMP analysis and are not figured here. The loadings are somewhat illuminating (Table 6). In the first PC axis, only the y-coordinate of the dorsal corner of the basisphenoid rostrum appears to affect the placement of the specimens. The only variable loading the second principal component axis is the landmark placed at the intersection of the basisphenoid rostrum and the pterygoid. The third principal component is affected by the x-coordinate of the ventral corner of the basisphenoid rostrum, and the y-coordinates of the contact between the laterosphenoid and the basisphenoid, and the contact between the prootic, laterosphenoid, and the opening for cranial nerve V. Along the fourth principal component axis, only the x-coordinate of the basioccipital-basisphenoid contact, the y- coordinate of the contact between the prootic, laterosphenoid, and the opening for cranial nerve V, and the x-coordinate of the lateral basisphenoid-basioccipital contact. The fifth principal component axis is affected by the y-coordinate of the basioccipital-exoccipital suture, the position of the laterosphenoid-basisphenoid suture, and the x-coordinate of the prootic-supraoccipital suture. Lastly, the sixth principal component axis is only affected by the x-coordinate of the dorsal prootic-laterosphenoid suture. The non-parametric one-way analysis of variance results in Chi-square values that all had p-values greater than 0.1, meaning that none of the axes are significant (Table 7). No regressions were performed in order to attempt to remove size differences from the sample due to the lack of a sufficient sample size. The second sets of analyses were performed in two iterations. The first (with Caiman) resulted in three axes describing 77.96%, 8.38%, and 6.79% of the variation, respectively (Table 8). The first relative warp separates the specimens based on size: the younger and smaller specimens clustered at the positive range of RW1, while the larger ones were towards the negative range (Figure 25). RW2 separates Caiman, Eosuchus,

52 and the adult Gavialis from the others. The third relative warp shows the gavialoids and Tomistoma at the positive range, and the alligatoroids and crocodyloids at the negative end (Figure 26). One pattern that is seen in all three of these plots, but is especially evident in the plot of RW2 versus RW3, is the grouping of the gavialoids and Tomistoma. In the plot of RW1 versus RW2, and RW1 versus RW3, Eosuchus falls out near Tomistoma. In the plot of RW2 versus RW3, the hatchling Gavialis and Tomistoma are almost directly on top of each other (Figure 27). In the same plot, Eosuchus and the adult Gavialis form a cluster together, away from the rest of the specimens. In this analysis, Caiman seemed to fall out well away from the other specimens, so I performed a second iteration with Caiman excluded. This second iteration resulted in three significant axes, each explaining 85.53%, 7.08%, and 2.68% of the variation, respectively (Table 9). The first relative warp shows almost the exact same results as the first iteration, except with the size of the specimen reversed on the axis: the small specimens were at the negative range and the larger ones on the positive range (Figure 31). Relative warp 2 separates the gavialoids and Tomistoma at the positive range from the alligatoroid and the crocodyloids at the negative range. The third relative warp clusters the specimens along the neutral portion of the axis (Figure 32). This analysis also returns the cluster of Eosuchus with Tomistoma in the plots of RW1 versus RW2, and RW1 versus RW3. The plot of RW2 versus RW3 showed the showed a distinct cluster of the gavialoids with Tomistoma (again with the hatchling grouping closest with Tomistoma) away from the crocodyloids and alligatoroids (Figure 33).

Table 4: Eigenvalues and variance explained by each axis in the SAS PCA. Axis Eigenvalue Proportion Cumulative % 1 13.5 0.42 42 2 7.24 0.22 65 3 3.85 0.12 77 4 2.56 0.08 85 5 2.09 0.065 91 6 1.11 0.035 95 53

54 Table 5: Variable loadings on principal component axes. Variable PC1 PC2 PC3 PC4 PC5 PC6 x1-0.212642-0.206062-0.009029-0.094518-0.096225-0.126201 y1 0.148348-0.227398-0.077173-0.13714 0.321359-0.05449 x2-0.135821-0.244562 0.018843 0.310428 0.00865 0.019541 y2-0.225259 0.138442-0.048003-0.148941 0.017552 0.123855 x3-0.009334 0.286453 0.221702-0.062331-0.203487-0.276052 y3 0.081372-0.343767 0.010898-0.101076 0.0047 0.136367 x4 0.147752-0.167785 0.308651-0.022131-0.234182 0.079567 y4-0.158825-0.188406-0.036904-0.239511-0.23695-0.147937 x5-0.186517 0.166934 0.045109 0.021239 0.276396-0.314365 y5-0.248108-0.022651-0.145478 0.032223-0.12262-0.17537 x6-0.12156 0.122248 0.238712-0.163015-0.322093 0.219379 y6-0.221169 0.053191 0.037627 0.135466 0.334295-0.037611 x7-0.205494 0.136035-0.210326-0.134302-0.052569-0.085219 y7-0.0705-0.025758 0.342868 0.33894 0.080406 0.351371 x8-0.195577 0.163458-0.123626 0.191614 0.015438 0.230959 y8-0.165989 0.182111 0.201888-0.119111 0.222714 0.148287 x9-0.142587-0.159342-0.167751 0.372044 0.062434-0.121635 y9-0.189833 0.200549 0.113897-0.149374 0.128846 0.077902 x10-0.052931 0.211922-0.266619-0.065747-0.085198 0.520675 y10 0.174857 0.126294 0.309209 0.096238-0.115687-0.135343 x11 0.234867-0.071039-0.119757 0.210179-0.007961 0.043073 y11 0.138647 0.198145 0.193232 0.263897 0.132997 0.099126 x12 0.105025 0.173609-0.321104-0.199148-0.033307 0.169923 y12 0.231095 0.152772-0.022245-0.101096-0.062891-0.160872 x13 0.219204 0.021705 0.027366-0.215604 0.324712 0.01147 y13 0.214695 0.136211-0.096475 0.198734-0.21294 0.014803 x14 0.229712 0.003944 0.031448-0.164634 0.284514 0.065573 y14 0.153866 0.236195-0.11035 0.19603-0.116852-0.193361 x15 0.246399 0.008756 0.068372-0.154068 0.144336-0.011597 y15 0.231052 0.159545-0.077037 0.057408-0.145109-0.010543 x16-0.068587-0.238285 0.300008-0.183534-0.112128 0.110391 y16 0.137224-0.23742-0.242984 0.035734-0.069796 0.16124 Note: Bold values are significant.

55 Table 6: Results of the Kruskal-Walis one-way ANOVA. Axis Chi 2 DF P 1 3.6764 2 0.159 2 4.4945 2 0.1557 3 3.0218 2 0.2207 4 1.0909 2 0.5796 5 1.3236 2 0.5159 Table 7: Eigenvalues for relative warp analysis with Caiman. Axis Eigenvalue SV Proportion Cumulative 1 1.22E+03 1.05749 77.96% 77.96% 2 4.30E+02 0.34665 8.38% 86.34% 3 3.82E+02 0.31207 6.79% 93.13% 4 1.83E+02 0.20379 2.90% 96.02% 5 1.39E+02 0.16241 1.84% 97.86% 6 9.25E+01 0.12558 1.10% 98.96% 7 7.29E+01 0.08082 0.46% 99.42% 8 5.17E+01 0.06879 0.33% 99.75% 9 2.98E+01 0.05556 0.22% 99.96% Table 8: Eigenvalues for relative warp analysis without Caiman. Axis Eigenvalue SV Proportion Cumulative 1 4.73E+02 1.05303 85.53% 85.53% 2 1.51E+02 0.30308 7.08% 92.61% 3 1.44E+02 0.18631 2.68% 95.29% 4 1.16E+02 0.17719 2.42% 97.71% 5 1.02E+02 0.11987 1.11% 98.82% 6 7.44E+01 0.08341 0.54% 99.36% 7 6.72E+01 0.06849 0.36% 99.72% 8 5.56E+01 0.04669 0.17% 99.89% 9 2.86E+01 0.03857 0.11% 100.00%

56 Table 9: Landmark weights of the relative warp analysis with Caiman and without Caiman. Landmark Number Landmark Weights With Caiman Without Caiman 1 0.00193 0.01708 2 0.02648 0.10524 3 0.00439 0.03821 4 0.00278 0.02849 5 0.00314 0.00783 6 0.28737 0.06145 7 0.43238 0.03554 8 0.11267 0.01285 9 0.00371 0.02574 10 0.00183 0.01803 11 0.02766 0.09176 12 0.00213 0.01547 13 0.00253 0.03111 14 0.00427 0.0485 15 0.00195 0.01309 16 0.00552 0.02324 17 0.01889 0.1124 18 0.04137 0.24247 19 0.01898 0.07152

Figure 20: Plot of principal component 1 versus 2. Black circles are Gavialis, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. 57

Figure 21: Plot of principal component 1 versus 3. Black circles are Gavialis, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. 58

Figure 22: Thin plate spline showing deformation along PC1. Landmarks as in Fig 5. 59

Figure 23: Thin plate spline showing deformation along PC2. Landmarks as in Fig 5. 60

Figure 24: Thin plate spline showing deformation along PC3. Landmarks as in Fig 5. 61

62 Figure 25: RW1 vs. RW2. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. Figure 26: RW1 vs. RW3. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1.

Figure 27: RW2 vs. RW3. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. 63

64 Figure 28: Deformation along RW1, negative at left, positive at right. Figure 29: Deformation along RW2, negative at right, positive at left.

65 Figure 30: Deformation along RW3, negative at right, positive at left. Figure 31: RW1 vs. RW2 with Caiman excluded. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1.

Figure 32: RW1 vs. RW3 with Caiman excluded. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. 66

Figure 33: RW2 vs. RW3 with Caiman excluded. Black circles are gavialoids, red star is Tomistoma, blue x s are alligatoroids, pink squares are crocodyloids. Specimen numbers as in Table 1. 67

68 Figure 34: Deformation along RW1 with Caiman excluded. Figure 35: Deformation along RW2 with Caiman excluded.

Figure 36: Deformation along RW3 with Caiman excluded. 69

70 CHAPTER 5: DISCUSSION AND CONCLUSIONS Discussion The principal components analyses of the two-dimensional landmark analysis resulted in three to six axes necessary for describing the variation in the ten-specimen sample. However, the Kruskal-Wallis test returned statistically insignificant values for each of the axes. One interpretation of these results is that these taxa truly do not have differences in the shape of their braincases, and therefore, statistically insignificant p- values are actually true to life. If this is the case, then searching for phylogenetic characters within the braincases may be futile as any changes therein are superficial and phylogenetically uninformative. High p-values could also be due to the small sample size used this study. With the addition of more specimens, small patterns seen in these analyses may become more pronounced and statistically significant. Just as easily, addition of more specimens may completely obliterate any pattern found herein. Therefore, this morphometric analysis is to be interpreted as an exploratory study, one that requires more data in order to have meaningful results. It is clear that size has the biggest influence on the first principal component, and even though size in crocodylians tends to correlate with age, this variable needs to be factored out to truly evaluate shape changes in these taxa. It is curious that Tomistoma consistently clustered with the specimens of Gavialis even within this small sample, even though other longirostrine crocodylians (e.g. M. cataphractus) did not. This is the first morphological evidence that Tomistoma and Gavialis may be more closely related than is currently revealed by morphological phylogenetics. The relative warp analysis of sliding semilandmarks on the Eustachian system also reflected this grouping. Two, independent landmark-based methods therefore returned similar results, both clustering Tomistoma with Gavialis and Eosuchus. Eosuchus and Tomistoma were

71 probably only clustered along RW1 due to a similarity in size, since that was the biggest factor weighting that axis, but size was less of a factor along RW2 and RW3, and the consistent clustering of Tomistoma with the hatchling Gavialis may have important implications for the phylogeny of Crocodylia. These analyses show some reconciliation between morphological and molecular phylogenetics, though more data are needed to corroborate this result. Even though most morphological phylogenies of Crocodylia place Gavialis at the base and Tomistoma within crocodyloids (e.g., Brochu 2007), these analyses may have missed key features of the braincase, such as the pneumatic system, that could have a critical bearing on phylogenetic reconstructions. To date, only molecular data or combined analyses have shown the close relationship of Tomistoma and Gavialis (Gatesy and Amato 1992; Gatesy et al. 2003, Willis et al. 2007). If this is the case, then Gavialis has morphological features that are secondarily reversed. The results presented here are among the first to lend morphological evidence to this hypothesis. Eosuchus, a stem gavialoid, consistently clustered with the crocodyloids in the morphometric analyses, even though Gavialis did not. This suggests that Gavialis is secondarily reversed in comparison with older gavialoids, which corroborates the molecular hypotheses of Crocodylian relationships. These results go against what was found in the Piras et al (2010) study, indicating that landmarks may not result in an accurate representation of morphospace when placed on morphologically plastic features. Even though superficial evidence from the ontogenetic changes in the pneumatic system agree with morphological phylogenetic hypotheses, a more thorough look at the shape of the pneumatic system in these specimens has shown that Gavialis and Tomistoma may share more in common than recently thought (contra Tarstitano 1985, 1989). My results indicate that Eosuchus has a more similar pneumatic system to crocodyloids than it does to Gavialis, which supports a secondarily reversed state for Gavialis.

72 It remains hard to assess these results without a more comprehensive study on braincase morphometrics. Future work should include a larger sampling of ontogenetic stages of modern crocodylians, while including as many stem species as possible. Better three-dimensional renderings would provide more sutures for landmark-based, 3- dimensional morphometric techniques. Conclusions Gavialis gangeticus is critically endangered and access to specimens is rare. High-resolution CT scanning allows for imaging of the internal structures of this and other rare specimens without destroying them. Digital renditions of the skull of YPM HERR-008438 are an invaluable source of information on Gavialis, especially due to the young age and extremely good preservation of this specimen. Ontogenetic differences show that the adult form is generally more angular than the hatchling, especially in the skull table. The rostrum becomes thinner and relatively longer in the adult. The angular has a more pronounced dorsal curvature in the adult. Finally, the posterior articular processes are dorsally expanded. Geometric morphometric analyses of the braincase of ten specimens of crocodylians show distinct groupings of Gavialis with Tomistoma and C. crocodilus with crocodyloids along PC1 and PC2. Size is the most important variable weighting PC1, as seen by the smaller specimens clustering at the positive range, and larger ones on the negative range. PC3 clustered the specimens along the zero region. Non-parametric statistical tests showed that none of the PC axes were significant (p-values > 0.1). This could be due to small sample sizes, but could also be a sign that the braincase changes little amongst crown-group crocodylians, and that we should look elsewhere for phylogenetically informative characters. In either case, larger sample sizes are necessary to better illuminate crocodylian relationships. Results from the second morphometric analysis again show a distinct grouping of Eosuchus and Tomistoma along the first and second relative warps, and a grouping of the

73 hatchling Gavialis with Tomistoma along the second and third relative warps. These results agree with the results from the principal components analysis even though they used a completely different set of landmarks and morphometric tests. This suggests that even though the first statistical tests returned insignificant p-values, there may be biologically significant characters in the structure of the braincase revealing deeper evolutionary relationships that were previously recovered by molecular phylogenetic analyses (Gatesy et al. 2003). Previous crocodylian morphometric analyses have failed to recover a close relationship between Tomistoma and Gavialis (Piras et al. 2010). These analyses have mostly focused on the external features of the skull, which are subject to ecological and environmental forces. The internal structure of the braincase is more constrained by the required maintenance of the brain and is therefore more able to retain evolutionary information. The results of this exploratory morphometric study shown that using geometric morphometric techniques on these structures may help reconcile morphological and molecular phylogenetics if larger sample sizes are used. Future work with more data could either corroborate the results discussed here, or further add to the problematic nature of the phylogenetic placement of Gavialis.

74 APPENDIX A: MIDSAGITTAL SKULL RENDERINGS Figure A1: Gavialis gangeticus (YPM 008438), A) Midsagittal cut; B) Sutural interpretation.

Figure A2: Gavialis gangeticus (TMM-M-5490), A) Midsagittal cut; B) Sutural interpretation. 75

Figure A3: Alligator mississipiensis (TMM-M-6723), A) Midsagittal view; B) Sutural interpretation. 76

Figure A4: Alligator mississipiensis (TMM- uncatalogued), A) Midsagittal view; B) Sutural interpretation. 77

Figure A5: Caiman crocodilus (FMNH-73711), A) Midsagittal view; B) Sutural interpretation. 78

Figure A6: Tomistoma schlegelii (TMM-M-6342), A) Midsagittal cut; B) Sutural interpretation. 79

Figure A7: Osteolaemus tetraspis (FMNH-93986), A) Midsagittal view; B) Sutural interpretation. 80

Figure A8: Mecistops cataphractus, (TMM-M-3529), A) Midsagittal view, B) Sutural interpretation. 81

Figure A9: Crocodylus moreletii (TMM-M-4980), A) Midsagittal view; B) Sutural interpretation. 82

Figure A10: Crocodylus johnstonii (TMM-M-6807), A) Midsagittal view; B) Sutural interpretation. 83

84 APPENDIX B: MIDSAGITTAL SLICES Figure B1: Midsagittal slices of G. gangeticus, A. mississippiensis, and C. crocodilus showing the Eustachian system in red.

Figure B 2: Midsagittal slices of the indicated species, showing the Eustachian system in red. 85