Description and phylogenetic analysis of a new alligatoroid from the Eocene of Laredo, Texas

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Spring 2014 Description and phylogenetic analysis of a new alligatoroid from the Eocene of Laredo, Texas Rachel L. Guest University of Iowa Copyright 2014 Rachel Leah Guest This thesis is available at Iowa Research Online: Recommended Citation Guest, Rachel L.. "Description and phylogenetic analysis of a new alligatoroid from the Eocene of Laredo, Texas." MS (Master of Science) thesis, University of Iowa, Follow this and additional works at: Part of the Geology Commons

2 DESCRIPTION AND PHYLOGENETIC ANALYSIS OF A NEW ALLIGATOROID FROM THE EOCENE OF LAREDO, TEXAS by Rachel L. Guest 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 2014 Thesis Supervisor: Associate Professor Christopher A. Brochu

3 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 Rachel L. Guest has been approved by the Examining Committee for the thesis requirement for the Master of Science degree in Geoscience at the May 2014 graduation. Thesis Committee: Christopher A. Brochu, Thesis Supervisor Jonathan M. Adrain Justin S. Sipla

4 To my family for the never ending support, encouragement, love, and back scratches. ii

5 ACKNOWLEDGEMENTS There are so many people that I need to thank for help with this project. Of course, I have to thanks my committee members, Chris Brochu, Jonathan Adrain, and Justin Sipla without whom the world of phylogenetics would likely still be a mystery. Jonathan taught me so much about TNT, WINCLADA, and phylogenetic analyses in general. Chris has been such a wonderful advisor. He is so incredibly approachable and open to questions and wants to share his excitement about the science with his students- a mark of a truly great teacher. I hope that by taking a risk and letting me work on a portion of the tree of life near and dear to his heart, I have made him proud. For access to specimens, I thank Carl Mehling, Alana Gishlick, David Kizirian, Margaret Arnold, Alan Turner, J. Chris Sagebiel, Tim Rowe, Matt Brown, Kenneth Bader, Amy Henrici, Jackie Hoff, Bruce Erickson, and Alan, Resetar. I cannot leave out many of the people who inspired and nurtured my interest in paleontology early on. I owe much of my fascination with geology to Bruce Rueger and the entire geology department at Colby College. I owe my specific love of vertebrate paleontology entirely to the most wonderful mentor, Bill Parker. Why he took a chance on a small girl who insisted on wearing shorts on our long treks into the Petrified Forest day after day and then allowed me to come back for another season continues to amaze me. The time he, in addition to Jeffery Martz, Matt Smith, and Kenneth Bader, took to teach me the ins and outs of field work, preparation, element description, and the joys of driving on a washboard road sitting with a freshly made plaster jacket in your lap has hooked me on vertebrate paleontology for life. I also am indebted to those who have taken me on as a volunteer. Shelly Cox and Trevor Valle at the Page Museum were always an encouraging force in the delicate matter of excavating and cleaning bones from tar. Thanks is owed to Randy Irmis, iii

6 Sterling Nesbitt, Alan Turner, Nate Smith, Alex Downs, and numerous others for allowing me the joy of working the Hayden quarry on a few occasions. And lastly Mike Getty, Eric Lund, and the entire University of Utah crew for some of the most exciting digs I have participated in, including my first helicopter ride, and for providing some of the best around-the-fire entertainment you could imagine. Work on this project was greatly enhanced by discussions with the Brochu Lab discussion group and the paleoseminar at the university of Iowa, but specifically thanks is owed in large part to David Tarailo, Jessica Miller-Camp, and Matt Burkey. I also want to thank the amazing group of Brochu lab group graduates that have come before me and helped to build the program up to the amazing standard that it possess today. Watching you moving on in your careers from where I sit today to holding some of the best positions in paleontology in the country is inspiring. I owe so much of my success in life to my wonderful family, Mom, Dad, Sarah, and Mark. Every member is endlessly encouraging and always available for a pep talk. Mom, each and every over-the-phone study session and freak out session brought me one step closer to this goal. I want to say a huge thank you to Woodrow for allowing me to not only pick up and leave Utah to pursue my dreams but encouraging me to keep it going when things felt difficult. Surprise packages, the endless I m so proud of you s and helping me keep the end in sight has been so motivational. Financial support for this thesis was provided by the University of Iowa Littlefield Fund, the Graduate Student Senate Research Grant, and the Department of Earth and Environmental Sciences at the University of Iowa. iv

7 TABLE OF CONTENTS LIST OF TABLES...vi LIST OF FIGURES...vii LIST OF ABBREVIATIONS...x CHAPTER I. INTRODUCTION...1 Review of Alligatoroid Taxonomy..2 Review of North American Caimans..2 Present Work 3 Material and Methods 10 Geology and Stratigraphic Provenance 10 Methods...12 Outgroup selection...12 Operational Taxonomic Units (OTUs)...13 Character Selection and Analysis II. III. IV. SYSTEMATIC PALEONTOLOGY.18 DESCRIPTION.20 COMPARISON.25 V. PHYLOGENETIC ANALYSIS 27 Results 27 VI. DISCUSSION 32 APPENDIX A. SPECIMENS EXAMINED FOR STUDY..43 APPENDIX B. PHYLOGENETIC CHARACTERS AND DESCRIPTIONS.47 APPENDIX C. CHARACTER-TAXON MATRIX..66 APPENDIX D. MPTs 69 REFERENCES...85 v

8 Table 1. LIST OF TABLES List of operational taxonomic units (OTUs) name, age, occurrence, and source of data for matrix codings 13 Table A1. List of specimens personally examined with elements present and associated locality information vi

9 LIST OF FIGURES Figure 1. North American Land Mammal Ages (NALMA) scale. The Uintan, the age of the Laredo quarry, and the Wasatchian, the age of Tsoabichi greenriverensis and Orthogenysuchus olseni, the two other North American caimans. Modified from and The Geologic Timescale 2012, Volume Figure 2. Cladogram showing current hypothesis of relationships within Alligatoroidea. Modified from Hastings et al., Figure 3. TMM Skull in dorsal view with line interpretation of cranial sutures. Scale = 1 cm Figure 4. TMM Skull in occipital view with line interpretation of cranial sutures. Scale = 1cm Figure 5. TMM Skull in right lateral view with line interpretation of cranial sutures. Interpretations of sutures were based on a recent Caiman yacare skull. Scale = 1cm Figure 6. Geologic Cross Section of the Lakeshore and Spillway Section, Laredo, Texas modified from Westgate Boxed numbers represent TMM vertebrate fossil localities Figure 7. Strict consensus of 8 MPTs with length of 174, CI= 0.62, RI= GC Bootstrap support values are listed above the node on the left while Bremer support values are listed below the node. indicates an extinct species Figure 8. Adams consensus tree Figure 9. Strict consensus with biogeography overlain on relevant nodes suggesting possible dispersal events. AS= Asia, NA= North America, SA= South America Figure 10. Minimum dispersals implied by the topology when TMM groups with Purussaurus. NA= North America, SA = South America vii

10 Figure 11. Minimum dispersals implied by the topology when TMM groups with the Paleosuchus + Tsoabichi clade. NA= North America, SA = South America Figure 12. Possible dispersals implied by the topology when TMM groups with the Paleosuchus + Tsoabichi clade. NA= North America, SA = South America Figure 13. Possible dispersals implied by the topology when TMM groups with the Paleosuchus + Tsoabichi clade. NA= North America, SA = South America Figure 14. Possible dispersals implied by the topology when TMM groups with the Paleosuchus + Tsoabichi clade. NA= North America, SA = South America Figure 15. Possible dispersals implied by the topology when TMM is sister to Tsoabichi greenriverensis. NA= North America, SA = South America Figure D1. MPT 1 with ACCTRAN optimization.69 Figure D2. MPT 2 with ACCTRAN optimization.70 Figure D3. MPT 3 with ACCTRAN optimization.71 Figure D4. MPT 4 with ACCTRAN optimization.72 Figure D5. MPT 5 with ACCTRAN optimization.73 Figure D6. MPT 6 with ACCTRAN optimization 74 Figure D7. MPT 7 with ACCTRAN optimization 75 Figure D8. MPT 8 with ACCTRAN optimization.76 Figure D9. MPT 1 with DELTRAN optimization.77 Figure D10. MPT 2 with DELTRAN optimization...78 Figure D11. MPT 3 with DELTRAN optimization...79 Figure D12. MPT 4 with DELTRAN optimization...80 Figure D13. MPT 5 with DELTRAN optimization...81 Figure D14. MPT 6 with DELTRAN optimization...82 viii

11 Figure D15. MPT 7 with DELTRAN optimization...83 Figure D16. MPT 8 with DELTRAN optimization...84 ix

12 Institutional Abbreviations LIST OF ABBREVIATIONS AMNH = American Museum of Natural History, New York CM FMNH IGM LACM MCZN SDSM SMM TMM USNM = Carnegie Museum of Natural History, Pittsburgh = Field Museum of Natural History, Chicago = Geological Survey of Columbia = Natural History Museum of Los Angeles County, Los Angeles = Museum of Comparative Zoology, Harvard University, Cambridge = Museum of Geology, South Dakota School of Mines, Rapid City = Science Museum of Minnesota, St. Paul = Texas Memorial Museum, Austin = U.S. National Museum of Natural History, Washington, DC YPM-PU = Yale Peabody Museum of Natural History, New Haven Anatomical abbreviations used in figures bo bs eo = Basioccipital = Basisphenoid = Exoccipital FM = Foramen magnum f j = Frontal = Jugal lcf = Lateral carotid foramen ls = Laterosphenoid x

13 oc = Occipital condyle pa = Parietal po = Postorbital pot = Prootic pt q = Pterygoid = Quadrate qfa = Quadrate foramen aereum so = Supraoccipital sq = Squamosal STF = Supratemporal fenestra vf = Vagus foramen xi

14 1 CHAPTER I INTRODUCTION Living caimanines, Caiman crocodilus Linnaeus, 1758 and alligatorids more closely related to it than to Alligator mississippiensis (Daudin, 1802), consist of eight or nine currently recognized species and are known exclusively from Latin America with only one form known to extend as far north as Central America (Hrbek et al., 2007; Venegas-Anaya et al., 2008; Velasco and Ayarzagüena, 2010) ). The group first appears in South America during the Paleocene, and although poorly known in the Paleogene, their fossil record in the South American Neogene is good (e.g., Langston, 1965; Riff et al. 2010, Scheyer et al., 2013; Bona et al., 2013). In contrast, Alligator mississippiensis and its closest relatives (alligatorines) are all Laurasian (Norell, et al., 1994; Brochu, 1999, 2004, 2011; Martin, 2007). Caimans are biogeographically problematic. From living species alone, it would be easy to hypothesize a single dispersal event from North to South America early in the Cenozoic. But there are several possible caimanines from the Paleogene of North America, and phylogenetic analyses have not put them at the root of Caimaninae, implying multiple dispersal events (Brochu, 1999, 2010; Stocker et al., 2012; Hastings et al., 2013). A simple scenario of vicariance does not appear to explain the divergence and distribution of alligatorines and caimanines as the molecular divergence estimates conflict with paleogeographic data (Brochu, 2011) land masses were already separate by the time the alligatorine-divergence split occurred at or near the Cretaceous-Paleogene boundary. But like other alligatorids, living caimanines are salt intolerant (Taplin and Grigg, 1989). The principle of parsimony would lead us to assume that their fossil relatives were also salt intolerant, making it more difficult for this group to cross large ocean barriers, leading to many questions about modes of dispersal during the Paleogene. In this thesis, I describe a fossil from the middle Eocene of southern Texas. It appears to be a caimanine resembling another North American caimanine, early Eocene

15 2 Tsoabichi greenriverensis Brochu, It adds further evidence that the biogeographic history of early caimanines was complex. Review of Alligatoroid Taxonomy Alligatoroidea Gray, 1844 includes Alligator mississippiensis (Daudin, 1802) and all crocodylians more closely related to it than to Crocodylus niloticus Laurenti, 1768 or Gavialis gangeticus Gmelin, 1789 (Norell et al., 1994; Brochu, 1999). Alligatoridae is a node-based crown group consisting of the last common ancestor of Alligator, Caiman, Melanosuchus and Paleosuchus and all of its descendants (Brochu, 2003). Molecular and morphological data both support monophyly of Alligatoridae and a sister-group relationship between Alligatoroidea and Crocodyloidea (Dumeril, 1806; Densmore 1983; Norell, 1989; Brochu, 1999). Alligatoroidea has a rich fossil record due to the fact that members of this group, who are semiaquatic animals, tend to live in depositional environments. Alligatoroids have been found as far back as the Campanian (Erickson, 1972; Norell et al., 1994; Wu et al., 1996; Williamson, 1996; Lillegraven, 1976; Brochu 1997b). This group has been important in the research of the phylogenetics of crocodylians (Brochu, 1999). Review of North American Caimans Historically, several North American fossils were at one point thought to be closely related to caimans, such as Procaimanoidea kayi Mook, 1941, but most are now regarded to be alligatorines. Nevertheless, a few North American fossils appear to be caimanines. One of these is a specimen from the Wasatchian NALMA (North American Land Mammal Age) of Wyoming, Orthogenysuchus olseni Mook, Phylogenetic analyses revealed a close relationship between Orthogenysuchus olseni and Mourasuchus, a South American form (Brochu, 1999, 2011). Other fossils including Listrognathosuchus multidentatus (Mook, 1930) from the late Paleocene Torrejonian NALMA of New Mexico, and another specimen from the

16 3 middle Eocene Uintan NALMA of Utah may be closely related to Orthogenysuchus. Both Listrognathosuchus multidentatus and Orthogenysuchus olseni were animals with long but broad snouts and a large number of teeth, but they are known from nonoverlapping samples. Listrognathosuchus multidentatus is known from a nearly complete lower jaw and postcranium, but very little of the skull is preserved. Conversely, only the skull of O. olseni is known. An unusual alligatorid in the Uintan of Utah (Masters et al., 2010) has a jaw resembling that of Listrognathosuchus and skull resembling that of Orthogenysuchus, suggesting a very close relationship between L. multidentatus and O. olseni and, in turn, a caimanine affinity for L. multidentatus. The Utah material has not yet been described. Another of these enigmatic North American fossils is Tsoabichi greenriverensis from the Wasatchian Green River Formation of Utah. Although its relationships are not well understood, analyses have shown that Tsoabichi is neither basal to all South American caimanines nor closely related to Orthogenysuchus olseni (Brochu, 2010, 2011; Hastings et al., 2013; Scheyer et al., 2013). Present Work A fossil that outwardly resembles Tsoabichi greenriverensis is Texas Memorial Museum (TMM) from the Uintan of Texas. TMM was collected by James Westgate in the 1980 s at a site near Laredo, Texas. Although incomplete, what remains of the specimen is exceptionally preserved with little to no crushing or deformation due to the fossilization process. Other unidentifiable alligator material was collected by Westgate at the same time in addition to numerous chondrichthyan, osteichthyan, amphibian, reptilian and even one bird fossil. The specimen was figured in an overview of the fauna (Westgate, 1989) and presented in a talk at a professional meeting (Busbey, 1989), but has not been formally described. The purpose of this thesis is to formally describe the specimen and conduct the first phylogenetic analysis of the fossil to place it within Alligatoridae, and discuss its biogeographic implications. The exceptional three- dimensional preservation of this fossil

17 4 braincase allows for an opportunity to study phylogenetically informative characters that are rarely preserved in the fossil record.

18 Figure 1. North American Land Mammal Ages (NALMA) scale. The Uintan, the age of the Laredo quarry, and the Wasatchian, the age of Tsoabichi greenriverensis and Orthogenysuchus olseni, the two other North American caimans. Modified from and The Geologic Timescale 2012, Volume 1. 5

19 Figure 2. Cladogram showing current hypothesis of relationships within Alligatoroidea. Modified from Hastings et al.,

20 Figure 3. TMM Skull in dorsal view with line interpretation of cranial sutures. Scale = 1 cm. 7

21 Figure 4. TMM Skull in occipital view with line interpretation of cranial sutures. Scale = 1cm. 8

22 Figure 5. TMM Skull in right lateral view with line interpretation of cranial sutures. Interpretations of sutures were based on a recent Caiman yacare skull. Scale = 1cm. 9

23 10 Material and Methods Geology and Stratigraphic Provenance TMM is housed in the Vertebrate Paleontology Laboratory, Texas Memorial Museum collections, Austin, Texas, USA. It was first collected as part of the Lake Casa Blanca Local Fauna, which is preserved in the middle Eocene Laredo Formation at Lake Casa Blanca International State Park near Laredo, Texas by Westgate in The sediments from which it came are described as marginal marine sands and muds of the middle Eocene Clairbone Group (Westgate, 1989). The specimen was collected from the southeast side of Lake Casa Blanca in Webb County, Texas (Westgate, 1989). Based on the plant, vertebrate, and invertebrate remains collected from the unit, the middle Eocene in Laredo has been reconstructed as a mangrove swamp-like environment bordering a coastal rainforest. A tropical or sub-tropical climate was proposed in part due to the crocodylian and turtle fossils uncovered in the area (Westgate, 2012). The depositional environment can be compared to the calm, warm waters of present-day southern New Guinea and northern Australia, or those of southern Mexico and northern Central America (Westgate 1990). The specimen was cataloged as an unidentified medium-sized alligatorid, but was not described further. Busbey (1989) identified this specimen as the earliest known caiman in an abstract, but no formal description of the fossil was written at that time.

24 Figure 6. Geologic Cross Section of the Lakeshore and Spillway Section, Laredo, Texas modified from Westgate Boxed numbers represent TMM vertebrate fossil localities. 11

25 12 Methods All photographs were taken by a Canon PowerShot SX20IS and the photos were edited in Adobe Photoshop CS5. Line drawings and other figures were created in Adobe Illustrator CS5. The character matrix used in the phylogenetic analysis was modified from Brochu, Data were collected over the past year upon visits to museum collections by the author. When not possible to view the specimens in person, I utilized codings from Brochu s 2011 paper. The only significant changes in codings from the Brochu 2011 matrix relates to those of Procaimanoidea kayi. I had an opportunity to examine this specimen with a fellow graduate student and together, decided that several of the codings should be changed. In discussions with Brochu, the changes were agreed upon based on photographs taken by the students. As P. kayi was one of the first specimens ever coded by Brochu, he was not surprised to find that many of the codings needed modification from his original work (pers. comm.) The data for this analysis were entered into a matrix in Mesquite (Maddison and Maddison, 2011). A maximum parsimony analysis examining the phylogentic relationships of TMM was conducted. Heuristic searches based on 100 random addition sequence replicates were performed in TNT (Goloboff, P., S. Farris, and K. Nixon. 2000). All characters had equal weight and multistate characters were unordered. Strict consensus trees were generated in WINCLADA (Nixon, K. C ), and the Adams consensus was generated with PAUP*, version 4.0b10; Swofford, Outgroup Selection Brachychampsa montana Gilmore, 1911 and Albertochampsa langstoni Erickson, 1972 were chosen as the outgroup taxa which are standardly used as outgroups taxa in phylogenetic analyses of Alligatoridae.

26 13 Operational Taxonomic Units (OTUs) Table 1. List of operational taxonomic units (OTUs) name, age, occurrence, and source of data for matrix codings. NAME AGE OCCURENCE SOURCE OF CODINGS Brachychampsa Late Cretaceous North Dakota Personal montana Gilmore, (Maastrichtian) observation 1911 Albertochampsa Late Cretaceous Alberta Personal langstoni Erickson, (Campanian) observation 1972 Navajosuchus mooki Simpson, 1930 Lower Paleocene (Puercan) New Mexico Brochu 2011 Allognathosuchus wartheni Case, 1925 Lower Eocene (Wasatchian) Wyoming Brochu, 2011 Procaimanoidea Middle Eocene Wyoming Personal kayi Mook, 1941 (Bridgerian) observation Procaimanoidea utahensis Gilmore, 1946 Middle Eocene (Uintan) Utah Brochu, 2011

27 14 Table 1 continued Wannaganosuchus Upper Paleocene North Dakota Personal brachymanus (Tiffanian) observation Erickson, 1982 Alligator prenasalis Loomis, 1904 Late Eocene (Chadronian) South Dakota Brochu, 2011 Alligator mcgrewi Lower Miocene Nebraska Personal Schmidt, 1941 (Hemingfordian) observation Alligator sinensis Recent Eastern China Brochu, 2011 Fauvel, 1879 Alligator Recent Southeastern United Brochu, 2011 mississippiensis States (Daudin, 1802) Alligator mefferdi Pliocene Nebraska Personal Mook, 1946 (Clarendonian) observation Eocaiman cavernensis Early Eocene Argentina Personal observation Simpson, 1933

28 15 Table 1 continued Necrosuchus Early Paleocene Argentina Brochu, 2011 ionensis Simpson, 1937 Tsoabichi greenriverensis Brochu, 2010 Lower Eocene (Wasatchian) Utah Brochu, 2010 Paleosuchus Recent Northern and Personal palpebrosus Central South observation (Cuvier, 1807) America Paleosuchus trigonatas Recent South America Personal observation (Schneider, 1801) TMM Middle Eocene Texas Personal (Uintan) observation Purussaurus Late Miocene South America Brochu, 2011 mirandi Aguilera et al., 2006 Purussaurus Late Miocene Colombia Brochu, 2011 neivensis (Mook, 1941)

29 16 Table 1 continued Orthogenysuchus olseni Mook, 1924 Lower Eocene (Wasatchian) Wyoming Brochu, 2011 Mourasuchus sp. Miocene Columbia Brochu, 2011 Price, 1964 Caiman yacare Recent Central South Personal Daudin, 1802 America observation Caiman crocodilus Recent Central and South Personal (Linnaeus 1758) America observation/ Brochu, 2011 Melanosuchus Pliocene Venezuela Brochu, 2011 fisheri Medina, 1976 Melanosuchus niger Recent South America Brochu, 2011 (Spix, 1825) Caiman latirostris Recent Argentina Brochu, 2011 Daudin, (1801) Caiman lutescens Brochu, 2011 (Rovereto, 1912)

30 17 Character Selection and Analysis Character 146 from Brochu 2011 was deleted from the analysis after discussion with Brochu determined that it was uninformative as written and needed to be reevaluated. The matrix consists of 28 taxa and 180 discrete characters. This portion of the matrix relative to this specimen included characters 137 through 180. The remaining characters were coded as unknown for TMM while the characters remained in the analysis for the best possible resolution of the remaining taxa. The table of codings can be found in Appendix B.

31 18 CHAPTER II SYSTEMATIC PALEONTOLOGY Reptilia Laurenti, 1768 Crocodylomorpha Hay, 1930 Neosuchia Benton and Clark, 1988 Eusuchia Huxley, 1875 Crocodylia Gmelin, 1788 (sensu Benton and Clark, 1988) Alligatoroidea Gray, 1844 (sensu Norell et al., 1994) Caimaninae Brochu, 2003 (Following Norell, 1988) New gen., new sp. Referred Specimens: TMM , skull table with braincase. Occurence: This specimen was collected in the 1980s as part of the Casa Blanca Fauna from Webb County, Texas. The quarry is located at N, W. Gastropod biostratigraphy places the quarry site at a location approximately 85 meters below the top of the Laredo Formation (Westgate, 1989). The beds of the quarry are now considered to belong to the middle member of the Laredo Formation (Westgate, 2012). The Laredo Formation along with the El Pico Clay and the Yegua Formation make up a portion of the Claiborne Group. The Claiborne Group is part of the Rio Grande Embayment and the formations are dated to the latest middle Eocene (Westgate, 2012). Ash layers above the gastropod zone in a correlated section were dated with 40 Ar/ 39 AR to Ma and Ma by Yancey et al., This prompted Westgate to estimate the age of the quarry at approximately Ma. Diagnosis: Alligatoroid sharing abrupt supratemporal fenestral rim, large supraoccipital exposure on the skull table, and linear frontoparietal suture with caimanines. It shares a frontoparietal suture located entirely on the skull table and a postorbital-squamosal suture passing medially ventral to the skull table with other

32 19 alligatorids. Supraoccipital exposure on the skull table smaller than other caimans of the same size, and lateral margins of skull table are bowed outwardly.

33 20 CHAPTER III DESCRIPTION TMM is a complete braincase and skull table showing excellent threedimensional preservation with minimal apparent flattening, although depressions on the skull table surface may be the result of postmortem compression. Sutures are clearly visible on the skull table but are less easily discerned on the braincase. The frontal is partially preserved in TMM The dorsal surface of the frontal is concave between the orbits and the orbital rims are modestly upturned. The element contacts the parietal posteriorly along a mediolaterally linear suture and the postorbitals posterolaterally. A sagitally-oriented sulcus on the ventral surface represents the pathway of the olfactory tract. The anterior process is not preserved. Part of the broken anterior surface, projecting into the right orbit, might be the posteriormost extent of the sutural contact for the prefrontal, but this is unclear. The postorbital is broadly crescentic in dorsal view, forming the anterolateral corner of the skull table and the anterolateral margin of the supratemporal fenestra. The dorsal surfaces appear to be modestly reflected dorsolaterally, but whether this is due to preservational processes is unclear; whether natural or not, the reflection accentuates the concavity of the frontal between the orbits. Based on the dorsalmost base of the postorbital bar, which is preserved on both sides, the bars were slender, inset from the skull table margin, and hemicylindrical with a convex anterolateral surface. The parietal contacts the frontal along a mediolaterally linear suture that is excluded from the supratemporal fenestrae. It passes along the ventral surface of each postorbital for a short distance to the lateral tip of the laterosphenoid capitate process. The dorsal surface is generally planar, though it is concave posteriorly where it contacts the supraoccipital. Because the supratemporal fenestrae are small and have constricted rims, it is unclear whether the surface of the parietal within the fenestrae was perforate. The sutures between the squamosals and parietal are linear antero-posteriorly and extend

34 21 from the posterior margin of the supratemporal fenestra to the skull table, but the posteriormost extent of the parietal is not preserved. Whether the parietal was excluded from the skull table posterior margin is thus ambiguous. Nevertheless, a strong case can be made for a condition resembling that of extant Paleosuchus, in which the parietal extended between the squamosals and supraoccipital to the posterior limit of the skull. This is largely because although large, the dorsal exposure of the supraoccipital is modest compared with what is typically seen in alligatoroids in which the squamosals and supraoccipital are in contact dorsally (see below). The parietal-squamosal suture posterior to the supratemporal fenestra is linear and oriented parasagitally, and it would not have intersected the supraoccipital suture unless it veered sharply medially close to the edge of the skull table. The squamosals make up the posterolateral portion of the skull table, along with the posterolateral border of the supratemporal fenestra. As with the postorbital, the dorsal surface of the squamosal appears to be oriented dorsolaterally rather than dorsally. Sutural contact between the squamosal and parietal is elevated, which may explain both the dorsolateral orientation of the squamosal and the concave dorsal surface of the parietal, but again, it is unclear whether this was normal morphology for the animal or a preservational artifact. The squamosal forms the roof of the external otic recess, and in lateral view it extends anteriorly to terminate dorsal to the postorbital bar. The lateral margin is covered with shallow pits and bears a deep horizontal groove on the anterior half. The squamosal also forms the posterior margin of the triangular external otic aperture. The lateral margins curve posteriolaterally away from the skull table. Neither squamosal is complete, but preserved portions suggest that the squamosal would form the anterolateral surface of the paroccipital process. Very little of the quadratojugal is preserved. A small portion of the quadratequadratojugal suture can be seen on the right side, and a sutural surface on the quadrate extends to the dorsal angle of the infratemporal fenestra from the breakage point of the quadratojugal, but whether this surface was entirely for the quadratojugal or for a

35 22 descending lamina of the postorbital is unknown. In either case, the quadrate did not contribute to the posterior border of the infratemporal fenestra. The quadrate forms at least the ventral margin of the external otic aperture and the posterior border of the trigeminal foramen. There is no indication of a preotic foramen, but the area where it would be located is not well exposed. The quadrate bears a robust posterolateral ramus, which bears a modest boss on its ventral surface indicating the origin site of some of the adductor musculature. Sutural borders with the quadratojugal can be seen on both sides, even where the quadratojugal has been lost. The lateral quadrate hemicondyle is larger than the medial hemicondyle and the foramen aereum is inset laterally from the dorsomedial edge of the quadrate. The medial hemicondyle is located on a more ventral plane than the lateral hemicondyle. Details of the quadrateexoccipital contact, and of the cranioquadrate canal, are poorly preserved. The external otic apertures are visible on both sides, but sutures are not well preserved on the left side. Both are triangular in shape. On the right, the squamosalquadrate suture appears to intersect the aperture at its posteroventral corner, but the suture can be seen passing dorsally along the posterior margin for a short distance when examined anteroposteriorly. The supraoccipital contributes to the dorsal surface of the skull table with a triangular shape. While there is a clear contact between the parietal at the anteriormost portion of the supraoccipital, the anterolateral borders of the supraoccipital have not been preserved. Nevertheless, it is very unlikely that the supraoccipital contacted the squamosals. The posterior margin of the supraoccipital is concave in dorsal view. The supraoccipital is also broadly exposed on the occipital surface, contacting the exoccipitals ventrally. Only the floors of the posttemporal fenestrae are preserved. A prominent sagittal crest runs from the exoccipital contact to the skull table. The exoccipitals surround the foramen magnum dorsally and laterally, and they contact the supraoccipital dorsomedially. The exoccipitals contact the posterior portions of the squamosals dorsolaterally, although portions of the squamosal are missing from

36 23 this specimen near the sutural contact. The medial portion of the posterior wings of the exoccipitals curve upward slightly and the posteriormost portions of the exocciptials are located laterally. Each wing adjacent to the squamosal forms the paroccipital process, and its surface toward its lateralmost extent is markedly concave. A long slender process of the exoccipital extends ventrally lateral to the basioccipital, reaching the basioccipital tubera, though this is most apparent on the left side. The vagus foramen penetrates the exoccipital lateral to the foramen magnum. A small hole on the left exoccipital, adjacent to the border of the foramen magnum, may be a foramen for the twelfth cranial nerve. The lateral carotid foramen is a circular hole ventral to the vagus foramen. The foramen magnum, bound dorsolaterally by the exoccipitals and ventrally by the basioccipital, is wider mediolaterally than it is tall dorsoventrally, giving the opening an ovate shape in occipital view. The opening displays a slight bowing dorsally and a come to a very gradual point on the dorsal surface of the occipital condyle forming a shape reminiscent of a heart. The opening is surrounded ventrally by the basioccipital and dorsally and laterally by the exoccipitals. The basioccipital is the ventral-most portion of the skull preserved in this specimen. The entirety of the occipital condyle is formed by the basioccipital. A definitive vertically-oriented sagittal ridge runs from the base of the occipital condyle to the base of the basiocciptial tubera. The ventrolateral margins are also expanded. The tubera both flare outward at the base and are slightly constricted under the occipital condyle. A circular opening at the ventral tip, visible in ventral view, is the median Eustachian channel, and notches on the posterolateral margins of the basioccipital indicate the positions of the lateral Eustachian openings. Part of the basisphenoid is visible anteroventrally. The basisphenoid rostrum is broken off, but its base is preserved, with a pair of circular foramina opening into a sulcus dorsal to it. The openings are the anterior (exit) foramina for the carotid arteries, and the sulcus is part of the sella turcica.

37 24 The laterosphenoids are preserved, but have been compressed into the endocranial cavity. The capitate processes are oriented anterolaterally. Laterosphenoid bridges are apparent on both sides, and they did not include a contribution from the pterygoid. Although crushed, the medial edges of the laterosphenoids as preserved are nearly parallel and approach each other closely. This suggests a relatively mature animal at death. The laterosphenoids are incompletely ossified in hatchling crocodylians, and ossification progresses during ontogeny until the bones contact each other, or come close to doing so, along the sagittal plane between the pituitary fossa and olfactory tract. Sutures are generally not possible to make out on the braincase. The trigeminal foramina are preserved on both sides, and they were bound posteriorly by the quadrates and (presumably) anteriorly by the laterosphenoids, but whether the prootics would have been exposed externally is not known. A depression ventral to the left trigeminal foramen could be interpreted as the prootic, but no sutural contacts are visible in its vicinity.

38 25 CHAPTER IV COMPARISON TMM has a dorsally shifted quadrate foramen aereum on the quadrate ramus which is the derived condition found within alligatoroids, strongly suggesting that it is an alligatoroid (Brochu, 1997b) although independent derivations in other crocodyliformes have been recognized (Brochu, 1999, Buscalioni et al., 2001). Among alligatoroids, caimanines typically display a long slender process of the exoccipital descending alongside the basiocipital and reaching the basiocipital tubera in some instances (Brochu, 2011). TMM displays this process, placing it among caimanines. Brochu (2010) compared the basic geometry and proportions of TMM to Tsoabichi greenriverensis, but was unable to fully assess a possible relationship given the incomplete nature of TMM and the fact that, at the time, his comparisons were based on photographs of the specimen and not direct observation. Upon further study of the specimens, I am inclined to assert that there are some geometric differences between T. greenriverensis and TMM While damage to the posterior skull table of TMM causes some uncertainty, it appears that the exposure of the supraoccipital on the skull table is much smaller than in T. greenriverensis of similar size. The sides of the skull table when viewed dorsally are much more parallel in Tsoabichi greenriverensis than in TMM The posterior end of the preserved squamosal of TMM flares laterally more than the squamosals in T. greenriverensis, although some of this may be attributed to ontogenetic variation. Also, in the specimens of T. greenriverensis examined, orientation of the dorsal expression of the supratemporal fenestra differs from that of TMM The parietal of T. greenriverensis is pinched in posterior to the supratemporal fenestrae whereas the lines of the parietal are extremely linear and do not pinch in on the skull table of TMM The supratemporal fenestrae in T. greenriverensis appear to deviate only slightly from an anterior-posterior direction, exhibiting some angulation

39 26 laterally at the most anterior portion of the fenestrae. In contrast, the supratemporal fenestrae of TMM are oriented at approximately 20 degrees laterally from the midline, significantly more than in T. greenriverensis. Orientation of the supratemporal fenestral long axis changes during ontogeny in crocodylians (Kälin, 1933), but the observed differences are among individuals of similar size. TMM has a combination of character states not found in any living caiman. A triangular supraoccipital that does not contact the squamosals is seen in Paleosuchus, but unlike Paleosuchus, TMM bears constricted, but nevertheless open supratemporal fenestrae in what appears to be a mature individual. The supratemporal fenestrae are open in Paleosuchus at hatching, but close relatively early in posthatching development. Unlike Purussaurus Barbosa-Rodrigues, 1892, TMM does not have deeply concave posterior margins of the skull table (Brochu, 1999). The lack of foramina in the medial parietal wall of the supratemporal fenestrae excludes TMM from the jacareans, but several characters dealing with the nasals and external naris provide further resolution among the other basal caimans that is simply unknown for the TMM Based on these, I conclude that TMM is a species distinct from T. greenriverensis and other caimanines. But although it resembles T. greenriverensis, I cannot at present refer TMM to Tsoabichi. Most diagnostic features of Tsoabichi greenriverensis involve regions of the skull not preserved in TMM A prominent concavity on the skull table dorsal surface, in the region of the supraoccipital and posterior part of the parietal, is also seen in two derived caimanines from the Miocene of Argentina Caiman lutescens and Caiman gasparinae (Bona and Carabajal, 2013; Bona et al., 2013). The Argentinian fossils are jacareans with trapezoidal supraoccipitals that contact the squamosals, and the concavity is more tightly restricted to the supraoccipital dorsal surface. Moreover, the concavity seen in TMM may be a preservational artifact.

40 27 CHAPTER V PHYLOGENETIC ANALYSIS Results A phylogenetic analysis of Alligatoroidea using 180 characters and 28 OTUs recovered 8 most parsimonious trees (MPTs) with length of 174, consistency index (CI) of 0.62, retention index (RI) of 0.81 and rescaled consistency index (RC) of GC bootstraping was performed in TNT with 10,000 pseudoreplicates selected. GC scores in addition to Bremer support values are shown on Figure 7. In all analyses, TMM was recovered as a caimanine. TMM is labile due to its incomplete preservation. Supraoccipital exposure on the skull table as well as the long descending process of the exoccipital places it within Caimaninae, but neither the splenials nor the angular-surangular suture is preserved, and these might permit more precise phylogenetic placement of the fossil. The lability of Necrosuchus encountered by Brochu in 2011 does not appear to be as problematic as it was in the original study, but the strict consensus draws Tsoabichi, TMM , and crown caimans into a polytomy similar to that seen in the 2011 study. This is likely due in large part to the inability to code certain characters in both Tsoabichi and TMM As mentioned in Brochu 2010, the codings of the splenial symphysis for Tsoabichi do no change the topology, but increase the treelength of several of the MPTs by one step. For this reason, I have left the coding of the splenial symphysis of Tsoabichi as unknown in this analysis. ACCTRAN and DELTRAN optimizations of all eight MPTs were carried out in WinClada. Optimized MPTs can be found in appendix D.

41 Figure 7. Strict consensus of 8 MPTs with length of 174, CI= 0.62, RI= GC Bootstrap support values are listed above the node on the left while Bremer support values are listed below the node. indicates an extinct species. 28

42 29 The placement of the quadrate foramen aereum unambiguously places this fossil within Alligatoroidea (Brochu, 1997b) although independent derivations in other crocodyliformes have been recognized (Brochu, 1999, Buscalioni et al., 2001). Alligatoridae is defined by one unambiguous synapomorphy and two ambiguous synapomorphies in this analysis: it is supported unambiguously by the preservation of the frontoparietal suture entirely on the skull table and ambiguously by a short doprsal premaxillary process and a maxilla with the posterior process within the lacrimal. Of these, only the position of the frontoparietal suture is preserved within TMM None of the unambiguous character states for Caimaninae optimized with ACCTRAN were preserved in TMM due to the incompleteness of the fossil. Several of the ambiguous synapomorphies from the ACCTRAN optimization are preserved in TMM including dorsal edges of the orbits upturned, a linear frontoparietal suture, and a long, descending process of the exoccipital. While both ACCTRAN and DELTRAN optimizations place the character of a large supraoccipital that excludes the parietal from the edge of the skull table as an unambiguous synapomorphy for Caimaninae, TMM along with several other caimans show a different character state in which there was a transition to a state of a large supraoccipital that doesn t contact the squamosals. The pleisiomorphic state is retained in nettosuchids, and living caimans other than Paleosuchus. Brochu (2010) mentioned that although this character was coded as an unordered character with four separate states, it is possible that the relationships between the supraoccipital, the parietal, and the squamosal may be a gradational one and the state shown in TMM may be a reversal or may be an end-member in this gradational series. When Caimaninae is examined at the level above Eocaiman, ACCTRAN optimizations show an unambiguous synapomorphy of dermal bones of the skull roof overhanging the rim of the supratemporal fenestrae. This is a feature well-preserved in TMM , confirming its placement above the most basal member of Caimaniane.

43 30 There are two overarching topologies recovered in the eight MPTs -one in which TMM groups with the Paleosuchus + Tsoabichi clade, and one in which TMM groups with Purussaurus. Within these two topologies, there is some variation as to the placement of TMM In the context of the trees supporting a relationship with Paleosuchus +Tsoabichi, two MPTs place TMM as sister to the clade including Paleosuchus and Tsoabichi (See Figures D3, D7, D11, and D15 for characters optimized with ACCTRAN and DELTRAN on these two MPTs). The character uniting all four taxa is supraoccipital exposure on dorsal skull table large, which is a change from the pleisiomorphic state within Caimaninae of supraoccipital so large that it is in direct contact with the squamosals and excludes the parietal from the dorsal edge of the skull table. Two MPTs place TMM as sister to Paleosuchus (See Figures D1, D4, D9, and D12 for characters optimized with ACCTRAN and DELTRAN on these two MPTs). When this topology is achieved, Tsoabichi is separated out in both ACCTRAN and DELTRAN by its non-keeled single ventral osteoderms and the presence of a thin bony crest circumscribing the external naris. Other trees link TMM as sister to Tsoabichi greenriverensis (See Figures D2, D6, D10, and D14 for characters optimized with ACCTRAN and DELTRAN on these two MPTs). In this topology, the clade comprised of T. greenriverensis, TMM , Paleosuchus palpebrosus, and Paleosuchus trigonatus is united by a large exposure of the supraoccipital on the skull table, but the lack of teeth that are circular in cross-section, in addition to the open supratemporal fenestra at maturity excludes T. greenriverensis and TMM from being included within the Paleosuchus clade. The second class of topologies seen in the MPTs places TMM as sister to Purussaurus (See Figures D5, D8, D13 and D16 for characters optimized with ACCTRAN and DELTRAN on these two MPTs). I am highly skeptical of this topology due to the shape of the back of the skull margin in addition to the shape of the frontoparietal suture and the placement is likely due largely to the incompleteness of

44 31 TMM rather than actual phylogenetic similarity, although both bear a concavity between the squamosal-parietal sutures on the skull table, though this may be due to damage in TMM In the trees where TMM groups with Purussaurus, it is supported once again by a large exposure of the supraoccipital on the skull table, but it is excluded from the Purussaurus clade by the lack of the concavoconvex frontoparietal suture diagnosing Purussaurus.

45 32 CHAPTER VI DISCUSSION From the analysis, we can see that TMM is a basal caimanine, although the resolution of the exact placement within the group is unclear. An Adams tree (Figure 8) forces TMM into a polytomy with crown caimans and the Tsoabichi + Paleosuchus clade. Much of the resolution of the other taxa within Caimanine is due to characteristics of the anterior portion of the cranium and postcranial material, which limits their utility for understanding the phylogenetic placement of TMM Also of note, as seen previously in other phylogenetic studies, Caiman is paraphyletic due to Melanosuchus (Hastings, 2013; Brochu 2010); however, molecular data generally supports a monophyletic Caiman (e.g., Gatesy et al., 2003; Oaks, 2011; Erickson et al., 2014). Biogeography within Caimaninae paints a confusing picture. The group is derived from North American ancestors, but achieved widespread success in South America. The two oldest and most basal caimans, late Paleocene-early Eocene Eocaiman and late Paleocene Necrosuchus, are both from South America (Simpson, 1933, 1937; Bona 2007; Brochu, 2011; Pinheiro et al., 2013). Molecular divergence dates and fossil data put the caimanine-alligatorine split at or near the Paleogene boundary (Hass et al., 1992; Roos et al., 2007; Brochu, 2011, Oaks, 2011), but North and South America were not directly connected by land at this time. Unlike their crocodylid relatives, alligatorids lack functional glands on the tongue that excrete excess salt, rendering them less tolerant of salt water, though they have been known to live near salty water and could possibly tolerate it for short periods of time during migrations between islands (Grigg et al., 1998; Elsey, 2005; Mazzotti et al., 2009). The lack of salt glands in alligatorids suggests a mechanism of dispersal other than across a substantial marine barrier.

46 33 One explanation for the multiple dispersals based on island hopping on the part of caimanines in addition to many other saltwater intolerant animals including snakes, marsupials, and dinosaurs close to the end of the Cretaceous (Rage, 1986). Conversely, Bayona et al.,(2011) and Cardona et al.,(2010) have suggested the presence of a shortlived volcanic arc between North and South America which may have allowed the dispersal of caimanines back from South America to North America in the late Paleocene or Early Eocene (Hastings et al., 2013). Trees supporting a single dispersal are less than optimal in all published analyses, and in this one, such trees where the North American caimanines are basal to a monophyletic South American clade are seven steps longer than optimal. Because TMM assumes four sets of positions in the caimanine tree, it is difficult to assess the number of dispersals to or from South America. This analysis hints at a possible third dispersal as TMM is not sister to Orthogenysuchus olseni and may not be sister to Tsoabichi. Brochu (2011) suggested that there had to be a minimum of two back dispersals from South to North America of caimanines during the Paleogene to explain the presence the two unrelated caimans in Wyoming. Some trees in this analysis those in which Tsoabichi and TMM are sister taxa support the same scenario, but others hint at a possible third dispersal (Figures 10-15). In most trees, dispersal events for each North American lineage are among the most parsimonious arrangements. In trees where Tsoabichi and TMM are successive sister groups to Paleosuchus, a single dispersal to North America followed by a dispersal back to South America for Paleosuchus is equally parsimonious. A close relationship between Tsoabichi and TMM , which may yet be recovered when more complete material is known, would require only three dispersals the original dispersal bringing caimanines to South America and two dispersals back to North America (Figure 15). Either scenario is more complex than the simple single-dispersal or vicariant model supported when living species are analyzed in isolation.