<%BANNER%>

New Cretaceous and Cenozoic Fossil Turtles from Colombia and Panama; Systematic Paleontology, Phylogenetical and Paleobi...

Permanent Link: http://ufdc.ufl.edu/UFE0025016/00001

Material Information

Title: New Cretaceous and Cenozoic Fossil Turtles from Colombia and Panama; Systematic Paleontology, Phylogenetical and Paleobiogeographical Implications
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Cadena, Edwin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Fossil vertebrates from the tropical part of the New World are well known from at least twentyfive Mesozoic and Cenozoic localities. In spite of this considerable number of fossil localities, many groups of vertebrates still having huge gaps in their fossil record. That is the case for turtles, which lack of Mesozoic and Paleogene fossils, leaving many questions unresolved as for example how far in time the extant fauna of turtles have inhabited the tropics?, how they have evolved and responded to major geological and climatic events, and at the same time how these events have influenced dispersal and interaction events between faunas from different land sources. In the last five years, intensive fieldwork has been performed by geologists and paleontologists from the Florida Museum of Natural History, the University of Florida and the Smithsonian Tropical Reseach Institute, in three different localities in Colombia and Panama, for which I described the fossil turtles. In chapter 1, I described a new specimen of the Early Cretaceous (Valanginian) Notoemys zapatocaensis species, which represents an allotype for this species and also the best preserved specimen for basal pleurodires or side-necked turtles, thus exhibiting a transitional morphological stage between primitive and derived turtles. Chapter 2 is dedicated to describe the first Middle to Late Paleocene turtles from Cerrejo acuten Coal Mine, Colombia; which represents a new genus and new species of podocnemidid turtle, named Cerrejonemys wayuunaiki and which is the sister taxa of the extant genus Podocnemis. Finally, the Chapter 3 focus in the description of the fossil turtles from the Early to Middle Miocene sedimentary sequences from the Panama Canal Basin. These fossils represent the earliest evidence of interaction between North-Central American (trionychids) and South American (podocnemidids) turtles; as well as the early diversification for the geoemydid genus Rhinoclemmys, genus for which a new species is described and named R. panamaensis. Thus, a new species of the kinosternid genus Staurotypus is also described, indicating a wider past geographical distribution for this genus in Central America. The last component of the turtle fauna fom the Panama Canal, are the giantic testudinids (tortoises), probably related to the genus Geochelone (Chelonoidis), reaching sizes bigger than the biggest modern representatives. Indicating that the gigantism in testudinis was early developed under ecological and geographical conditions different to those that characterized the islands where they are resctricted today.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Edwin Cadena.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Macfadden, Bruce J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0025016:00001

Permanent Link: http://ufdc.ufl.edu/UFE0025016/00001

Material Information

Title: New Cretaceous and Cenozoic Fossil Turtles from Colombia and Panama; Systematic Paleontology, Phylogenetical and Paleobiogeographical Implications
Physical Description: 1 online resource (133 p.)
Language: english
Creator: Cadena, Edwin
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Geological Sciences -- Dissertations, Academic -- UF
Genre: Geology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Fossil vertebrates from the tropical part of the New World are well known from at least twentyfive Mesozoic and Cenozoic localities. In spite of this considerable number of fossil localities, many groups of vertebrates still having huge gaps in their fossil record. That is the case for turtles, which lack of Mesozoic and Paleogene fossils, leaving many questions unresolved as for example how far in time the extant fauna of turtles have inhabited the tropics?, how they have evolved and responded to major geological and climatic events, and at the same time how these events have influenced dispersal and interaction events between faunas from different land sources. In the last five years, intensive fieldwork has been performed by geologists and paleontologists from the Florida Museum of Natural History, the University of Florida and the Smithsonian Tropical Reseach Institute, in three different localities in Colombia and Panama, for which I described the fossil turtles. In chapter 1, I described a new specimen of the Early Cretaceous (Valanginian) Notoemys zapatocaensis species, which represents an allotype for this species and also the best preserved specimen for basal pleurodires or side-necked turtles, thus exhibiting a transitional morphological stage between primitive and derived turtles. Chapter 2 is dedicated to describe the first Middle to Late Paleocene turtles from Cerrejo acuten Coal Mine, Colombia; which represents a new genus and new species of podocnemidid turtle, named Cerrejonemys wayuunaiki and which is the sister taxa of the extant genus Podocnemis. Finally, the Chapter 3 focus in the description of the fossil turtles from the Early to Middle Miocene sedimentary sequences from the Panama Canal Basin. These fossils represent the earliest evidence of interaction between North-Central American (trionychids) and South American (podocnemidids) turtles; as well as the early diversification for the geoemydid genus Rhinoclemmys, genus for which a new species is described and named R. panamaensis. Thus, a new species of the kinosternid genus Staurotypus is also described, indicating a wider past geographical distribution for this genus in Central America. The last component of the turtle fauna fom the Panama Canal, are the giantic testudinids (tortoises), probably related to the genus Geochelone (Chelonoidis), reaching sizes bigger than the biggest modern representatives. Indicating that the gigantism in testudinis was early developed under ecological and geographical conditions different to those that characterized the islands where they are resctricted today.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Edwin Cadena.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Macfadden, Bruce J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0025016:00001


This item has the following downloads:


Full Text

PAGE 1

1 NEW CRETACEOUS AND CENOZOIC FOSSIL TURTLES FROM COLOMBIA AND PANAMA; SYSTEMATIC PALEONTOLOGY, PHYLOGENETICAL AND PALEOBIOGEOGRAPHICAL IMPLICATIONS By EDWIN ALBERTO CADENA RUEDA A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Edwin Alberto Cadena Rueda

PAGE 3

3 To life and its magical evolutionary proc ess and of course to my Mom, friends and my Argentino

PAGE 4

4 ACKNOWLEDGMENTS Funding for this project came from the Smithsonian Paleobiology Endowment Fund, The Florida Museum of Natural History, The National Science Foundation grant DEB 0733725, the Miss Lu cy Dickinson Fellowship, the Fondo para la Investigacin de Ciencia y Tecnologa Banco de la Republica de Colombia, the Unrestricted Endowments Smithsonian Institution Grants, and Carbones del Cerrejn LLC. Thanks go to J. Bloch, C. Jaramillo, B. MacFadden L. de Broin, W. Joyce, H. Martin and J. Bourque for all the support and values comments that they did. C. Montes and Cerrejn Geologists team helped with logistic support and fieldwork. Thanks for access to collections to J. Arenas at Ingeominas, Dr. L de Broin paleontological collections (Musum national dhistoire naturelle, Paris, France); Dr. O. Castao and Dr. J. Lynch (Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Bogot, Colombia); Dr. E. Gaffney and C. Mehling, (Fossil Amphi bians, Reptiles, and Birds Collections, Division of Paleontology, American Museum of Natural History, New York, USA). Dr K. De Queiroz, Division of Amphibians and Reptiles, Smithsonian National Museum of Natural History, Washington DC, USA). Special thanks go to F. Herrera, A. Hastings, A. Rincon, S. Moron, L, Meza, I. Gutierrez, G. Bayona, C. Sanchez, T. Gaona, R. Hulbert and other all paleontologists and geologists working at the Cerrjon Coal Mine, at the Colombian Petroleum Institute, Smithsonian Tropica l Research Institute, and the Florida Museum of Natural History. Thanks to R. Rueda and M. Gonzalez for their continue support and source of inspiration.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT ........................................................................................................................................ 11 CHAPTER 1 INTRODUCTION ....................................................................................................................... 13 2 TWO NEW SPECIMENS OF THE PLATYCHELYDID TURTLE NOTOEMYS ZAPATOCAENSIS FROM THE EARLY CRETACEOUS (VALANGINIAN) OF COLOMBIA. ............................................................................................................................... 16 Introduction ................................................................................................................................. 16 Systematic Paleontology ............................................................................................................. 17 Comparative Description ............................................................................................................ 19 Phylogenetic Analysis ................................................................................................................. 28 Discussion .................................................................................................................................... 29 Phylogenetic Results ............................................................................................................ 29 Sexual dimorphism .............................................................................................................. 31 3 NEW PODOCNEMIDID TURTLE (TESTUDINES: PLEURODIRA) FROM THE MIDDLE PALEOCENE OF SOUTH AMERICA. ................................................................. 40 Introduction ................................................................................................................................. 40 Systematic Paleontology ............................................................................................................. 41 Description and Comparisons ..................................................................................................... 43 Skull ...................................................................................................................................... 43 Lower Jaw ............................................................................................................................ 51 Cervical Vertebrae ............................................................................................................... 54 Carapace ............................................................................................................................... 55 Plastron ................................................................................................................................. 58 Coracoid ............................................................................................................................... 58 Pelvic Girdle ........................................................................................................................ 59 Phylogenetic analysis .................................................................................................................. 59 Results .................................................................................................................................. 60 Disc ussion .................................................................................................................................... 60 Relationship between Extant Podocnemidids .................................................................... 64 Paleobiogeographical Scenario ........................................................................................... 65

PAGE 6

6 4 EARLY TO MIDDLE MIOCENE TURTLES FROM PANAMA; SYSTEMATICS AND PALEOBIOGEOGRAPHICAL IMPLICATIONS ......................................................... 79 Introduction ................................................................................................................................. 79 Systematic Paleontology ............................................................................................................. 81 Discussion .................................................................................................................................... 96 Fauna and provinciality ....................................................................................................... 97 Summary and Conclusions .................................................................................................. 99 5 CONCLUSIONS ....................................................................................................................... 108 Notoemys zapatocaensis Early Cretaceous of Colombia ...................................................... 108 Cerrejonemys wayuunaiki Middle to Late Paleocene of Colombia ..................................... 109 Pleurodires and Cryptodires from the Early to Middle Miocene of Panama ........................ 110 APPENDIX A CHAPTER 2 DATA ................................................................................................................. 112 Carapace..................................................................................................................................... 112 Plastron ...................................................................................................................................... 113 B CHAPTER 2 CHARACTER MATRIX .................................................................................. 115 C CHAPTER 3 DATA ................................................................................................................. 116 Skull ........................................................................................................................................... 116 Lower Jaw .................................................................................................................................. 120 Cervical Vertebrae .................................................................................................................... 120 Coracoid ..................................................................................................................................... 121 Carapace..................................................................................................................................... 121 Plastron ...................................................................................................................................... 121 D CHAPTER 4 CHARACTER MATRIX .................................................................................. 122 REFERENCES ................................................................................................................................. 125 BIOGRAPHICAL SKETCH ........................................................................................................... 133

PAGE 7

7 LIST OF TABLES Table page 2 1 Measurements for the platychelyrids, included the allotype of Notoemys zapatocaensis .......................................................................................................................... 39 3 1 Summary of the known fossil record of South American podocnemidinuran turt les. ....... 75 3 2 Measurements for UF/IGM 33, holotype of Cerrejonemys wayuunaki ............................. 78

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Map of the most equatorial part of Central and South America, showing the three localities for which fossil turtles are described in this study ............................................... 15 2 1 A. Location of Zapatoca town, Department of Santander, Colombia 6 50 35N, 73 13 50W. ........................................................................................................................ 32 2 2 Notoemys zapatocaensis allotype MGJRG IPN 15 EAC 150620061 ................................ 33 2 3 Notoemys zapatocaensis MGJRG IPN 15 EAC 150620062 ............................................... 34 2 4 Entoplastron and epiplastra evolution in testudines ............................................................. 35 2 5 Strict consensus cladogram showing the phylogenetic relationships between pleurodiran turtles .................................................................................................................. 36 2 6 One of the three most parsimonious trees obtained from the cladistic analysis u sing branch and bound search ........................................................................................................ 37 2 7 Differences in the xiphiplastra and fontanellas within platychelyrids, differences potentially related with sexual dimorphism .......................................................................... 38 3 1 Stratigraphic column for the middle late Paleocene Cerrejn formation .......................... 68 3 2 UF/IGM 33, Cerrejonemys wayuunaiki holoytpe. .............................................................. 69 3 3 UF/IGM 33, Cerrejonemys wayuunaiki holoytpe. .............................................................. 70 3 4 UF/IGM 33, Cerrejonemys wayuunaiki holoytpe ............................................................... 71 3 5 Strict consensus cladogram showing the phylogenetic relationships between podocnemidinurans turtles ..................................................................................................... 72 3 6 Left condylus mandibularis of quadrate in ventral view. .................................................... 73 3 7 Map showing the distribution of modern (grey shading) and extinct podocnemidids. Hexagons for Late Cretaceous; stars for Paleogene; and black circles for Neogene records ..................................................................................................................................... 74 4 1 Map of the Panama Canal, starts show the Early to Middle Miocene fossil localities from Culebra and Cucaracha formations, from which fossil turtles were collected. ....... 102 4 2 Rhinoclemys panamaensis Holotype UF 237887 ............................................................... 103 4 3 Rhinoclemys cf. Areolata UF242075 most anterior portion of the nuchal ....................... 104

PAGE 9

9 4 4 Testudinids Cf. Geochelone. USNM V23180 .................................................................... 105 4 5 Staurotypus moschus UF242076 ......................................................................................... 106 4 6 Podocnemidids G en. et sp. Indet. UF242176 ..................................................................... 107

PAGE 10

10 LIST OF ABBREVIATIONS AMNH American Museum of Natural History, New York, USA. ICN Instituto de Ciencias Naturales, Universidad Nacional de Colombia, Bogot, Colombia. IPN EAC Museo Ge olgico Jos Royo y Gomz Instituto Colombiano de Geologa y Minera Ingeominas, Bogot, Colombia. MACN Museo Argentino de Ciencias Naturales, Buenos Aires, Argentina. MNHN Musum national dhistoire naturelle, Paris, France, laboratoires: AC, Anatomie Compare, Z, Zoologie des Reptiles et Amphibiens. MNHNCu Museo Nacional de Historia Natural, La Habana, Cuba. MOZP Museo Prof. Dr. Olsacher Zapala. Argentina. UF/IGM University of Florida, Florida Museum of Natural History Vertebrate Paleontology Collections,, Gainesville, USA/ Museo Geolgico, at the Instituto Nacional de Investigaciones en Geociencias, Minera y Quimica, Bogot,Colombia. UF (H) University of Florida, Florida Museum of Natural History Herpetology Collections. UNEFM CIAPP Universidad Nacional Experimental Francisco de Miranda, Coro, Venezuela. USNM Smithsonian Nacional Museum of Natural History, Paleobiology, Washington, USA

PAGE 11

11 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science NEW CRETACEOUS AND CENOZOIC FOSSIL TURTLES FROM COLOMBIA AND PANAMA; SYSTEMATICS PALEONTOLOGY, PHYLOGENETICAL AND PALEOBIOGEOGRAPHICAL IMPLICATIONS. By Edwin Alberto Cadena R ueda August 2009 Chair: Jonathan Bloch Major: Geology Fossil vertebrates from the tropical part of the New World are well known from at least twentyfive Mesozoic and Cenozoic localities. In spite of this considerable number of fossil localities, many g roups of vertebrates still having huge gaps in their fossil record. That is the case for turtles, which lack of Mesozoic and Paleogene fossils, leaving many questions unresolved as for example how far in time the extant fauna of turtles have inhabited the tropics?, how they have evolved and responded to major geological and climatic events, and at the same time how these events have influenced dispersal and interaction events between faunas from different land sources. In the last five years, intensive fiel dwork has been performed by geologists and paleontologists from the Florida Museum of Natural History, the University of Florida and the Smithsonian Tropical Reseach Institute, in three different localities in Colombia and Panama, for which I described the fossil turtles. In chapter 1, I described a new specimen of the Early Cretaceous (Valanginian) Notoemys zapatocaensis species, which represents an allotype for this species and also the best preserved specimen for basal pleurodires or side necked turtles, thus exhibiting a transitional morphological stage between primitive and derived turtles. Chapter 2 is dedicated to describe the first Middle to Late Paleocene turtles from

PAGE 12

12 Cerrejn Coal Mine, Colombia; which represents a new genus and new species of podo cnemidid turtle, named Cerrejonemys wayuunaiki and which is the sister taxa of the extant genus Podocnemis Finally, the Chapter 3 focus in the description of the fossil turtles from the Early to Middle Miocene sedimentary sequences from the Panama Canal B asin. These fossils represent the earliest evidence of interaction between North-Central American (trionychids) and South American (podocnemidids) turtles; as well as the early diversification for the geoemydid genus Rhinoclemmys genus for which a new spe cies is described and named R. panamaensis Thus, a new species of the kinosternid genus Staurotypus is also described, indicating a wider past geographical distribution for this genus in Central America. The last component of the turtle fauna fom the Pana ma Canal, are the giantic testudinids (tortoises), probably related to the genus Geochelone (Chelonoidis ), reaching sizes bigger than the biggest modern representatives. Indicating that the gigantism in testudinis was early developed under ecological and geographical conditions different to those that characterized the islands where they are resctricted today.

PAGE 13

13 CHAPTER 1 INTRODUCTION Turtles are one of the most remarkable and successful reptiles; this is greatly due to the development of their shell, whic h is constituted by the dorsal carapace, the ventral plastron and the lateral bridge connecting those two. The turtle shell is considered as one of the most remarkable evolutionary novelty among vertebrates (Gilbert et al. 2008), and its origin as far as i t is known from the fossil record occurred during the late Triassic, with the apparition of Odontochelys from China, Carnian in age (228 Ma) (Li et al. 2008), which shows that the plastron evolved first than the carapace in marine environments. Proganochel ys and Proterochersis from the late Triassic of Germany, Norian in age (215 Ma) (Gaffney, 1990) are turtles with carapace and plastron completely developed, and adapted for more terrestrial environments demonstrated for Proganochelys based on bone histolog y. The evolution of the turtles not only of their shell, but also of the skull and post -cranial elements particularly the neck and pelvis continued along the Jurassic, and was precisely during the middle Jurassic when has been hypothesized occurred one of the most important splitting event, giving origin to the two largest groups of turtles, the Pancryptodires (formed by stem cryptodires plus the crown cryptodires or hidden -necked turtles which includes tortoises, soft shell turtles, sea turtles and some f reshwater turtles) and the Panpleurodires (formed by stem pleurodires plus crown pleurodires or side -necked turtles which includes marine to freshwater turtles) (Joyce, 2007; Danilov and Parham, 2008). The fossil record of turtles in the tropical part of Central and South America, as well as the Caribbean islands is very rare, particularly for the Mesozoic and mostly of the Cenozoic (Paleogene -Early Miocene) leaving huge gaps in the evolutionary history of turtles particularly for the tropics, which are ar eas that generally retain a long historical biogeography considered as

PAGE 14

14 the key factor determining the highest diversity found in lower latitudes (Wiens et al. 2009). Therefore, any new fossil discovery from the tropics particularly for the periods above me ntioned, has important significance, not only for the potential contribution for the understanding of the phylogeny of turtles and the evolution of the turtle shell, but also to analyze the response of tropical turtles to climatic and extinction events, as well as the chance to test biogeographical hypotheses of dispersal and vicariance. Since 2004, intensive fieldwork campaigns lead by a multidisciplinary team of geologists and paleontologists from the Smithsonian Tropical Research Institute, The Florida Museum of Natural History University of Florida, Smithsonian Institution and the Colombian Geological Survey (Ingeominas) have resulted in amazing fossil discoveries from at least three different localities, two of them in Colombia, and the third one in Panama, in all these localities, turtles have been one of the most abundant and better preserved fossil vertebrates. The first locality is Zapatoca, at the Eastern cordillera of Colombia (Fig.1 1), fossil turtles from this locality are early Cretaceous, ba se of the late Valanginian in age (138 Ma) (Fernando Etayo personal communication), exhibiting an exquisite preservation within of limestone deposited in shallow marine environments. The second locality is Cerrejn Coal Mine, at the Guajira Peninsula, Northeastern of Colombia (Fig 1 1), fossil turtles from this locality are middle to late Paleocene in age (55 Ma) (Jaramillo et al. 2007), mostly of them are found at the top of claystones and coal seams corresponding to deltaic systems. The third and last loc ality involve in this study, is the Panama Canal basin (Fig 1 1); fossil turtles from this basin are early to middle Miocene in age (23 15 Ma) (Kirby et al. 2008); found in deposits corresponding to delta front and delta plain, with important volcanic influence.

PAGE 15

15 The goal of this study was the description of the fossil turtles from the three localities previously mentioned, establishment their systematic paleontology, involving them in phylogenetical analyses based on morphological characters, and finally di scussing their biogeographical implications Figure 1 1. Map of the most equatorial part of Central and South America, showing the three localities for which fossil turtles are described in this study. 1. Early Cretaceous (Valanginian), Zapatoca town, Eastern Cordilleran, Colombia. 2. Middle to Late Paleocene, Cerrejn Coal Mine, Guajira Peninsula, Colombia. 3. Early to Middle Miocene, Panama Canal Basin, Panama.

PAGE 16

16 CHAPTER 2 TWO NEW SPECIMENS OF THE PLATYCHELYDID TURTLE NOTOEMYS ZAPATOCAENSIS FROM THE EARLY CRETACEOUS (VALANGINIAN) OF COLOMBIA. Introduction Pleurodire turtles appeared during the Late Jurassic (160 Million years ago) in shallow marine environments; posteriorly, during the Late Cretaceous and Cenozoic they diversified and coloni zed fluvial systems, in which the modern representatives live today. According to Gaffney et al. (2006) pleurodire turtles are constituted by two nanorders: Platychelira or most basal pleurodires and Eupleurodira or crown pleurodira. Platychelirid turtles are grouped in two families: Platychelyidae ( Platychelys ) and Notoemydidae ( Notoemys and Caribemys) following Lapparent de Broin et al. (2007) and De la Fuente (2007); and in a single family Platychelyidae (Notoemys and Platychelys ), following Cadena and G affney (2005), posteriorly ratified by Gaffney et al. (2006); and once again supported here. Notoemys represents the most diverse genus of platychelyids with three fossil species spread from high latitudes to the tropics. N. laticentralis from the late Jur assic (Tithonian) of Argentina (Fernandez and De la Fuente, 1988; 1994; Lapparent de Broin et al., 2007; De la Fuente, 2007) known by shells and a posterior partial skull; N. oxfordiensis from the late Jurassic (Oxfordian) of Cuba (De la Fuente and Iturral de, 2001) known by a single shell; and N. zapatocaensis from the early Cretaceous (Valanginian) of Colombia (Cadena and Gaffney, 2005) known by a nearly complete shell. Any of the three species of Notoemys has preserved completely the anterior plastral lob e, which has key morphological features to understand the evolution of the turtle shell, such as the intergular and gular scales, as well as the epiplastron and entoplastron bones. Two new specimens of Notoemys zapatocaensis are described here. The former is an almost complete and articulated shell, and the second is a partial carapace with some isolated

PAGE 17

17 plastral elements; both specimens were collected by me during 2006 in Zapatoca, Colombia (same locality and stratum as for the holotype) 6 50 35 N, 73 13 50 W (Fig. 1 1). The nearly complete shell constitutes an allotype or specimen of opposite sex to the holotype (ICZN, 1999: Recommendation 72A). Additionally, the excellent preservation of the anterior plastral and carapaceal elements allows to am end the diagnosis for this species, as well as to include it in a phylogenetic analysis focused in pleurodires. Systematic Paleontology Order TESTUDINES Linnaeus, 1758 or Batsch, 1788 Infraorder PLEURODIRA Cope, 1864 Nanorder PLATYCHELIRA Gaffney et al., 2 006 Family PLATYCHELYIDAE Brm, 1965 Genus NOTOEMYS Cattoi and Freiburg, 1961 NOTOEMYS ZAPATOCAENSIS Cadena and Gaffney, 2005 Allotype : MGJRG IPN 15 EAC 150620061 (Fig. 2 1, A D) Articulated shell (carapace and plastron), missing the right posterolateral p ortion of the carapace. Type Locality: El Caucho Farm (6 50 35 N, 73 13 50 W) northeast of Zapatoca town, Department of Santander, Colombia. Horizon and age : A limestone layer belonging to the upper segment of the shallow marine Rosablanca Formatio n (Guzman, 1985). The occurrence of the ammonite Saynoceras verrucosum (F. Etayo, personal commun, 2008.) indicates that this part of the Rosablanca Formation corresponds to the base of the late Valanginian in age, approximately 138 Ma according to the biochronostratigraphic framework of Ogg et al. (2008). Revised and referred specimens : MGJRG IPN 15 EAC 140120031 (holotype), A nearly complete shell, with the anteromedial region of the carapace and the anteromedial part of

PAGE 18

18 plastron missing. MGJRG IPN 15 EAC 150620062, partial central portion of the carapace including neurals 2 through 8, the most medial portion of the costals 2 through 7, and an isolated most medial portion of the left eighth costal preserving the iliac scar (Fig 2 2, A B) Amended Diagnosis : Notoemys zapatocaensis is recognized as a pleurodire turtle by the following characteristics: (1) one pair of mesoplastra laterally restricted, lacking a medial contact between each other, (2) well developed anal notch in U or V open shape, and (3) sutur al articulation of the pelvis with the shell. N. zapatocaensis is a platychelyid because has: (1) a costovertebral tunnel very wide in its entire length, (2) articulation tubercle on anterior face of the first thoracic rib, (3) carapace with posterior side s tapering medially, (4) second neural smaller than the rest of the neural series, (5) thoracic vertebrae smooth and flat ventrally, hexagonal in shape with notch centrolateral, (6) carapace with dorsal protuberances, located on posterior region of pleural and vertebral scales. It is recognized as a member of the Notoemys genus and differs from Platychelys by: (1) wider and shorter cervical scale, (2) lacking supramarginal scales, (3) smooth shell with lower dorsal protuberances of carapace, lacking radial striation, (4) relatively flatter shell, (5) larger suprapygal one, (6) neural 3 in posterolateral contact with costal 4, (7) iliac scar oval in shape and restricted to costal 8, (8) very reduce medial space between the first and the second thoracic ribs. Autopomorphies of Notoemys zapatocaensis are: (1) anterior plastral lobe margin with two pretty reduced lateral tuberosities, almost straight in outline, (2) shallow notch on the posterolateral margin of epiplastra, (3) gular scales rectangular in shape, m uch wider than long, (4) intergular scale long slightly touching the pectorals medially, separating completely the humerals, (5) posterolateral edge of epiplastra convex, (6) central plastral fontanel projected posteriorly into the xiphiplastral region and filled by thinner bone in males, and unfilled in females, (7) very small marginal 3, (8) neural 1 slightly

PAGE 19

19 shorter than neural 2 and exclusively in contact with costal 1 laterally, avoiding also a contact between neural 2 and costal 1, (9) vertebral scales narrower than in N. laticentralis N. oxfordiensis and Platychelys. (10) peripheral 3 lacks posteromedial contact with costal 2. Comparative Description Carapace: MGJRG IPN 15 EAC 150620061 has a shell cordiform in shape, having an anterior edge straigh t and posterior lateral sides tapering medially as in the holotype and the other platychelyids. Posterior edges are dentate at the contacts between marginal scales, as in the holotype, although much less pronounced than in Platychelys and slightly more pro nounced than in N. laticentralis MACN 18403, feature unknow for N. oxfordiensis due to bad preservation of the edges. Low protuberances are located at the posterior medial region of each vertebral and pleural scales, as in the holotype and MGJRG IPN 15 EAC 150620062 partial carapace referred here. In N. laticentralis the protuberances are slightly lower and due to the highly eroded surface of the carapace they are unrecognized in N. oxfordiensis high and well developed protuberances are characteristic of P latychelys The carapace surface is smooth, with a light microvermiculation more than granulation as in the holotype, similar to the condition present in N. latincentralis In contrast, Platychelys has a carapace surface very sculptured with radial striati ons originated at the center of the protuberances. The nuchal bone is hexagonal in shape and wider than long as in all other platychelyids and all primitive turtles for which the nuchal is known, including Kayentachelys, Eileanchelys Heckerochelys Indoc helys and Chengyuchelys By constrast, all eupleurodires have a nuchal bone relatively equidimensional or longer than wide. The neural series is composed of eight. Neural 1 is slightly shorter than neural 2, being the only neural in lateral contact with c ostal 1, condition also shared by the holotype, although with a neural 1 slightly larger. In contrast, the other two species of Notoemys and Platychelys have a neural 1 longer than neural 2, and in

PAGE 20

20 contact with costals 1 and 2 laterally; avoiding an antero lateral contact of neural 2 with costal 1; this is also the primitive condition present in Kayentachelys Eileanchelys Heckerochelys Indochelys Chengyuchelys and the basal podocnemidinuran Brasilemys In all eupleurodires, neural 2 contacts costal 1 an terolaterally, except in the chelid Hydromedusa tectifera and some species of Phrynops which remain the primitive condition. The rest of chelids and the podocnemidid Bairdemys lack completely the neural series. The neural 3 of the allotype of N. zapatocae nsis is large almost octagonal in shape, and posterolaterally in contact with costal 4 as for N. laticentralis. In the holotype of N. zapatocaensis neural 3 lacks the right posteroral contact with costal 4; same asymmetrical pattern is present in the Platychelys specimen figured in Lapparent de Broin (2001:fig 1), however the holotype of Platychelys lacks of posterolateral contact with costal 4 in both sides, thus exhibiting a more rectangular shape than in the other platychelyids. The neural series is fin ally complete with neurals 4 through 8, which exhibit the same pattern in shape and sutural contacts as for: the holotype, the other specimen of N. zapatocaensis referred here MGJRG IPN 15 EAC 150620062 (Fig 2 2, A B), N. laticentralis and Platychelys the last exhibiting neurals slightly more rectangular in shape. The neural series is unrecognizable for N. oxfordiensis due to its poor preservation Suprapygal 1 is rectangular in shape, slightly longer than wide as for the holotype, pretty similar than in P latychelys, the primitive testudines Condorchelys Indochelys and Kayentachelys By contrast, the suprapygal 1 is trapezoidal in shape, wider posteriorly than anteriorly in N. laticentralis All eupleurodires lack suprapygal 1. Suprapygal 2 is only preserv ed anteriorly in the allotype of N. zapatocaensis exhibiting the same pentagonal shape as in the holotype, and the other platychelyids. The pygal of N. zapatocaensis although missing for the allotype, but preserved and previously described for the holotype has a particular medial notch on its posterior edge, notch that is also present

PAGE 21

21 although shallower in the holotype of N. laticentralis MACN 18403, specimen reexamined directly by the senior author of this paper. This new interpretation differs from previ ous studies (De la Fuente and Iturralde 2001), that considered the posterior pygal notch absent for N. laticentralis The eight sets of costals are complete in both sides of the carapace of the allotype of N. zapatocaensis with the right sets slightly broken laterally. The shape of costals is similar as in the holotype and the other platychelyids. MGJRG IPN 15-EAC 150620062 preserves the left costal 8 with the iliac scar slightly oval, rounded and restricted to this costal, as in N. laticentralis and seems to be also the condition in N. oxfordiensis for which this region is badly preserved. By contrast, Platychelys has an iliac scar elongated and developed onto costal 8, suprapygal, and the medial margin of the peripherals. Eleven peripheral bones are recognized on the left side of the allotype. Peripherals 1 through 3 are in medial contact with costal 1, and particularly the peripheral 3 lacks a posteromedial contact with costal 2, differing from the other platychelids and other testudines for which periph eral 3 contacts posteromedially costal 2. A particular case of a small peripheral 3 restricted between peripheral 2 and 4 was defined as a potential diagnostic characteristic for the holotype of N. zapatocaensis (Cadena and Gaffney 2005), which has not pre served the anterior series of peripherals on the left side. The allotype described here shows a pretty well developed peripheral 3 in both sides of the carapace, indicating that the condition in the holotype corresponds to a pathology of that specimen, as was initially considered by Cadena and Gaffney, (2005). Peripherals 5 through 7 are longer than wide, whereas peripheral 8 and 10 are slightly larger than peripheral 9 and 11, as for the holotype and N. laticentralis In Platychelys only peripheral 10 is s lightly larger than peripheral 9 and 11. The cervical scale in the allotype of Notoemys zapatocaensis is rectangular in shape, much

PAGE 22

22 wider than long, as in the holotype and N. laticentralis The cervical is slightly shorter in Platychelys, which is also th e primitive condition present in Proganochelys, Proterochersis, Kayentachelys, Indochelys, Eileanchelys Heckerochelys Chengyuchelys at least for its middle cervical. In Dortoka the cervical is almost equidimensional, and in chelids is longer than wide, e xcept in the extant species Hydromedusa tectifera which has a large cervical enclosed between marginals 1, pleurals 1 and vertebral 1. All pelomedusoides turtles lack cervical scale, excluding euraxemydids for which the scales are unknown. Five vertebral scales are clearly visible on the dorsal aspect of the carapace in the allotype of Notoemys zapatocaensis ; vertebrals 1 through 3 are almost rectangular in shape as in the holotype and Platychelys much narrower than in N. laticentralis, and primitive tes tudines such as Proganochelys, Proterochersis, Kayentachelys, Heckerochelys, Indochelys, and Eileanchelys Condition unknown for N. oxfordiensis Vertebral 4 is nearly hexagonal in shape as in Platychelys and much narrower than in N. laticentralis and the primitive testudines Proganochelys, Proterochersis, Kayentachelys, Heckerochelys, Indochelys, and Eileanchelys The sulcus between vertebral 3 and 4 is on neural 6 and costals 6, as in Platychelys and the other species of Notoemys as well as in most of t he primitive testudines for which five neurals are recognized, (Character 74, Joyce, 2007; erroneously defined for vertebrals 2 and 3). Vertebral 5 although preserved only anterolaterally in the allotype of N. zapatocaensis seems to be heptagonal, as in th e holotype and N. laticentralis being octagonal in Platychelys Similar pattern of reduction in the width of the vertebral scales described for N. zapatocaensis and Platychelys is also share by the eupleurodires and eucryptodires. Four pleural scales are visible on the left portion of the carapace, exhibiting the same shape as in the holotype, N. laticentralis and Platychelys Pleural 4 is rectangular more than

PAGE 23

23 pentagonal in Platychelys which also has straighter medial edges for all pleurals. Twelve marg inal scales are visible on the left side of the carapace. Marginal 1 lacks a contact with pleural 1, as in the holotype, in contrast to N. laticentralis and Platychelys for which marginal 1 contacts pleural 1 posteriorly. Marginal 2 has the same shape and size as for the holotype, slightly longer than in N. laticentralis and Platychelys Marginal 3 in the allotype and also the holotype of N. zapatocaensis is smaller in contrast to the same scale in the others platychelyids and the others testudines. Margina ls 4 through 8, 10, and 12 are longer than wide, rectangular in shape as for the holotype for which marginals 10 and 12 were erroneously identified as marginals 9 and 11 by Cadena and Gaffney (2005). N. laticentralis the primitive Kayentachelys Condorche lys and the marginals 10 through 12 of Proterochersis also share the same pattern described for N. zapatocaensis By contrast, in Platychelys the same numbers of marginal scales are slightly more pentagonal in shape, due to the serrations in the outline of the carapace. Marginals 9 and 11 are pentagonal in shape in the allotype of N. zapatocaensis the holotype (marginals 8 and 10 erroneously identified by Cadena Rueda and Gaffney, 2005), N. laticentralis, Kayentachelys, Condorchelys, Heckerochelys, and the eupleurodires for which the posterior series of marginals are more equidimensionals due to an increase in the size of the peripherals. Plastron: The anterior plastral lobe of the allotype of Notoemys zapatocaensis is shorter than the posterior, having an anterior edge straight with very reduce tuberosities in both lateral corners, and a slight concavity at the medial margin. In Platychelys the anterior edge exhibits a very short tuberosity at the midline of the plastron, whereas than in N. oxfordiensis th e anterior edge lacks tuberosities. Both Platychelys and N. oxfordiensis have a slightly more convex anterior plastral edge than in N. zapatocaensis In the case of N. laticentralis the arrangement f

PAGE 24

24 bones and scales at anterior plastral lobe margin remains unknown because of neither the holotype MACN 18403 nor MOZP 2487 specimen figured in De la Fuente (2007) preserve completely this aspect. The primitive condition present in Odontochelys Proganochelys and Proterochesis is an anterior platral lobe with la rge tuberosities, defining a very dentate anterior margin. Tuberosities persist although much more reduce in number and size in Kayentachelys and Chengyuchelys and completely disappear in Indochelys which has a very straight anterior edge. Dortoka and th e others eupleurodires have an anterior plastral lobe very convex, with some exceptions as for example the bothremydid Taphrosphys, which has a slightly straight anterior plastral edge. The entoplastron of Notoemys zapatocaensis is diamond in shape, touching subtilely the edge of the anterior plastral lobe, and separating completely both epiplastral. The most primitive condition seen ventrally in Odontochelys is an entoplastron with high participation in the edge of the anterior plastral lobe, and both epi plastral meeting at midline posteriorly to the entoplastron. Proganochelys Proterochersis, Paleochersis, Kayentachelys Indochelys and N. zapatocaensis show a more progressive advance of the entoplastron inward the anterior plastral lobe, separating compl etely both epiplastral, being more advanced in N. zapatocaensis for which the entoplastron lacks of participation in the edge of the anterior plastral lobe. Eileanchelys, Heckerochelys, Chengyuchelys, Platychelys N.oxfordiensis and N. laticentralis (dubi ous for this species due lacking complete preservation of this area) show an entoplastron more inward advanced in the anterior plastral lobe, with both epiplatral meeting at midline anteriorly to the entoplastron, in a short contact that becomes longer in Dortoka and the others eupleurodires. A graphical reconstruction of the evolution of entoplastron in testudines is shown in Figure 2 3.

PAGE 25

25 The epiplastron in the allotype of Notoemys zapatocaensis is trapezoidal in shape with the posterior edge convex as in Chengyuchelys and Heckerochelys. In N. oxfordiensis N. laticentralis Platychelys, Dortoka and the others eupleurodires, the posterior edge of the epiplastron is straight to slightly concave, becoming highly concave in the primitive testudines Proganochely s and Proterochersis. The hyoplastron and hypoplastron are similar in shape than for the others platychelyids, but with the particularity that the central fontanela, developed from the central portion of hyoplastral until the anteromedial part of the xiphi plastral has been almost completely filled by bone thinner than for the rest of the shell, preserving its outline by a sulcus, similar to the specimen of N. laticentralis MOZP 2487 figured in Lapparent de Broin (2007:fig 1D). The allotype of N. zapatocaens is preserves unfilled by bone a small space of the central fontanela at the medial sutural point between the hypoplastral and xiphiplastral. The presence of central fontanela in the holotype of N. zapatocaensis remains dubious due to the highly broken marg ins for the hyoplastral and xiphiplastral principally at the midline, however if the central fontanela really existed in this specimen, should has been restricted to the central portion of the plastron and never went posteriorly extended toward the xiphipl astral region, as can be recognized by lacking of sulcus or differences in bone thickness indicating secondary filled of the fontanela. N. oxfordiensis Platychelys and the primitive testudines Sichuanchelys and Indochelys share with the holotype of N. zap atocaensis the presence of a central fontanela restricted to the sutural meeting area between hyoplastral and hypoplasral bones, and a posterior fontanela restricted between the sutural meeting area between hypoplastral and xiphiplastral bones, this last u nknown for N. oxfordiensis and absent in Indochelys and Heckerochelys By contrast, N. laticentralis and the allotype of the N. zapatocaensis share a large central fontanela posteriorly projected toward the xiphiplastral region. Primitive testudines such a s Odontochelys,

PAGE 26

26 Proganochelys Proterochersis, Paleochersis and Kayentachelys, lack plastral fontanelles, as well as Dortoka and the others eupleurodires except by Araripemys which has central and posterior fontanelles. Occasionally the fontanelles are de veloped in pretty young ontogenetic stages in modern pleurodires, beign filled by bone in later stages; examples of this process are seen in Podocnemis lewyana MNHN 1994286, and Chelus fimbriata MNHN A 5176. Mesoplastral in the allotype of Notoemys zapat ocaensis are triangular in shape, wider than long, lacking a midline contact and smaller than in N. latincentralis N. oxfordiensis and Platychelys. The primitive condition seen in Odontochelys is two pairs of mesoplastral meeting at midline of the platron condition erroneously considered by Li et al., (2008) as diagnostic for this genus, ignoring than is also the condition for Proterochersis. Proganochelys Kayentachely s, Eileanchelys Heckerochelys and Chengyuchelys have only one mesoplastral pair, meet ing at midline of the plastron or reaching the lateral border of the central fontanella in Sichuanchelys and Indochelys Dortoka together with chelids and Araripemys lack of mesoplastral, whereas than in all other eupleurodires the condition is one pair of mesoplastral, almost equidimensional, laterally restricted and lacking a midline contact. The posterior plastral lobe of the allotype of Notoemys zapatocaensis has a marked concavity, in contrast to the flat surface present in the holotype. The lateral ed ges are slightly rounded with two shallow embayment; one at the lateral expression of the sutural contact between the hypoplastron and xiphiplastron; and the other at the lateral expression of the sulcus between the femoral and the anal scale, this is also the condition for the holotype. Platychelys and N. laticentralis have a less marked embayment on the lateral edges of the posterior plastral lobe and for N. oxfordiensis the condition remains unknown due to the most of the posterior plastral lobe is missi ng.

PAGE 27

27 Xiphiplastral of the allotype of Notoemys zapatocaensis have a deep U shaped anal notch with posterior tips, similar to the specimen MOZP 2487 of N. laticentralis By contrast, the holotype of N. zapatocaensis and the specimen of Platychelys figured in Lapparaent de Broin (2001:fig 1B) have shallow wide V shaped anal notch, lacking well shaped posterior tips. All primitive testudines including Odontochelys lack of xiphiplastral anal notch, exhibiting a straight posterior edge, except in Proterochersis w hich has supernumerary scales at the most posterior margin of the plastron creating a very narrow false anal notch. All eupleurodires have well developed anal notch variable in size, shape and deep within each family. The intergular scale is pentagonal, e longated in shape, longer than wide, reaching the posteromedial corner of the entoplastron in the allotype of Notoemys zapatocaensis similar to the bothremydid Ummulisani figured in Gaffney et al. (2006:fig 269). By contrast, N. laticentralis N. oxfordie nsis and Platychelys have an intergular scale less advanced onto the posteromedial margin of the entoplastron, condition much less advanced in Dortoka and most of the eupleurodires for which the intergular only covers the most anteromedial corner of the en toplastron, and for the case of the podocnemidid Erymnochelys which has an intergular very small and restricted between the gulars. The intergular scale remains unknown for Odontochelys, and for the other primitive testudines such as Progranochelys Proter ochersis, and Heckerochelys Chengyuchelys has a particular condition of two small intergulars. The gulars in the allotype of N. zapatocaensis are almost rectangular in shape, much wider than long, exclusive condition within the testudines, and intermediat e from a pretty short squared and more laterally positioned gulars of Proganochelys and Proterochersis; and the triangular more medially positioned gulars of N. oxfordiensis Platychelys Dortoka and most of eupleurodires. The humeral scales are completel y separated medially by the intergular as in the bothremydid

PAGE 28

28 Ummulisani ; thus they are smaller than for the other platychelyids, Proganochelys Proterochersis, Chengyuchelys Heckerochelys, and the eupleurodires included Dortoka The humeropectoral sulcus in N. zapatocaensis is concave, slightly in contact with the posterior corner of the entoplastron as in N. laticentralis being more posteriorly positioned in N. oxfordiensis Platychelys Odontochelys Proterochersis, Proganochelys Chengyuchelys, Heckero chelys and Dortoka figured in Lapparent de Broin and Murelaga (1999:fig 4). In eupleurodires and Dortoka figured in Lapparent de Broin et al. (2004:plate III 4) the humeropectoral sulcus is more anteriorly advanced on the posterior region of the entoplastr on. The pectoroabdominal, abdominofemoral and femoroanal sulcus in the allotype of N. zapatocaensis as well as in N. laticentralis are interrupted at midline of the plastron by the large central fontanella. Phylogenetic Analysis In order to perform a cla distic analysis, I involved the new two specimens of Notoemys zapatocaensis above described plus the holotype, together with other 17 taxa, in a matrix of 36 shell morphological characters; 16 ingroup taxa and 1 outgroup taxon ( Odontochelys ), see Appendix A (list of characters) and Appendix B (character matrix). The characters were taken and in some cases modified from previously published character matrices and detailed systematic studies including: Lapparent de Broin and De la Fuente (2001), De la Fuente and Iturralde (2001), Cadena and Gaffney (2005), De la Fuente (2003), Joyce (2007), and Li et al. (2008). A few of these characters are new to this study and were defined based on direct examination of extant or fossil specimens. I constructed the characte r -taxon matrix using Mesquite Version 2.5 (Maddison and Maddison, 2008). For the phylogenetic analysis I used the parsimony algorithm of PAUP 4.0b10 (Swofford, 2002). All characters were equally weighted and unordered. Multistate characters were treated as polymorphic. I performed a branch and bound search, and

PAGE 29

29 finally I obtained boostraping percentages for 100 replicates. Bremer decay indecis were obtained using TreeRot version 3 (Sorenson and Franzosa, 2007). Discussion Phylogenetic R esults The cladistic analysis resulted in 3 most parsimonious trees (length =67 steps, consistency index =0.83,retention index =0.88, homoplasy index =0.16). The strict consensus of the 3 most parsimonious trees is shown in Figure 2 4. The consensus shows important differences in contrast to previous phylogenetic hypotheses, particularly to those proposed by Gaffney et al. (2006), Joyce (2007), Li et al. (2008), Sterli (2008), and Meylan and Gaffney (2009). The first big difference is the earlier position for Proterochersis ind icating that is most primitive than Progranochelys in terms of the shell morphology, particularly in terms of the plastral characters. This is principally notorious by the presence of two pairs of mesoplastral bones, which is also the condition present in Odontochelys considered as the most basal turtle so far known (Li et al., 2008). Considering that Proterochersis and Odontochelys share the presence of two pairs of mesoplastral bones, exhibiting the same shape and relative size, this character must be excluded as exclusive synapomorphy for Odontochelys The presence of a completely shaped shell (carpace and plastron) is an unambiguous character supporting the node A (Fig 2 4), which supports the clade Testudines. The placement of Kayentachelys and Indochel ys in the cladogram resembles the position obtained in previous studies for both taxa (Sterli 2008), indicating that they are derived in contrast to Proterochersis and Progranochelys but remain more primitive than the rest of testudines. The split between platychelirids and eupleurodires (node B) is well supported in the consensus by the presence of a pelvis strongly sutured to the plastron and carapace, synapomorphic character that supports the clade Pleurodira, and is agree with the cladogram

PAGE 30

30 presented i n Gaffney et al. (2006:fig 292). In contrast to Joyce (2007:fig 18) Platychelys Notoemys oxfordiensis (Caribemys see introduction of this Chapter) and N. laticentralis are stem pleurodires. This hypothesis is erroneous, considering that the shell characte r (presence of a round visceral contact of the ilium with the carapace) that supports the node 11 ( N. laticentralis plus crown pleurodira) is also shared by N. oxfordiensis making this character synapomorphic for the node 10 ( N. oxfordiensis N. laticentr alis and crown pleurodira) and not for the node 11. The node C in Fig 2 4 represents the platychelirids, including Platychelys and the three species of Notoemys, which are in an unresolved polytomy. Evaluating, the three most parsimoniosous trees, I favo red the one that shows N. laticentralis as sister taxa with N. zapatocaensis (Fig 2 5), the argument to chose this hypothesis is based on that these two taxa shares more characters in common than the other two possible combinations of taxa. The node D in t he consensus represents the clade Eupleurodira and is supported by the lateral position of the sulcus between vertebrals 3 and 4 on costals 5, and the iliac scar positioned on costal 8 and pygal, sometimes reaching costal 7. Within the eupleurodire clade, the most relevant difference in contrast to previous cladograms (Gaffney et al., 2006) is the basal position for Ararypemys, being more primitive than Dortoka and excluded from the clade Pelomedusoides. Node E represents the split between the most importan t groups of modern and fossil eupleurodires, the Cheloides and Pelomedusoides. Surprisingly, Bonaportemys bajobarrealis and Lomalatachelys neuquina considered as the most basal chelids (Lapparent de Broin and De la Fuente, 2001) appear together in a separa ted clade inside of the clade Pelomedusoides (node F). This means that they share more synapomorphic characters with pelomedusoides than whit cheloides if only shell characters are considered, and the only way to

PAGE 31

31 support their affinity with chelids should be found through skull material, which by now is not available for this two species. Sexual dimorphism Sexual dimorphism in turtles is expressed in several ways such as: difference in size between adult males and females, and particularly in a concave pla stron in males of terrestrial species (Pritchard, 2008). The new specimen of Notoemys zapatocaensis MGJRG IPN 15 EAC 150620061 together with N. laticentralis MOZP 2487 shares a posterior plastral lobe concave, and an anal notch well developed in U shaped, indicating that they represent males for each one of these two species. Thus, the large central fontanella present in both specimens, is potentially a morphological character associated with sexual dimorphism, in this case representing males. By contrast, the holotype of N. zapatocaensis Cadena and Gaffney (2005) and the specimen of Platychelys figured in Lapparent de Broin (2001:fig 1B) share a posterior plastral lobe flat, with narrower and V shaped anal notch, as well as smaller central and posterior fon tanelles, indicating that they represent females for each one of these two species. This could be also the case for the holotype of N. oxfordiensis figured in De la Fuente and Iturralde (2001:fig 3). A graphic reconstruction of the xiphiplastron for platyc helyids, as well as their differences associated to sexual dimorphism is shown in Figure 2 6. The identification of morphological variations associated with sexual dimorphism in fossil turtles has important implications in phylogenetic analysis. For exampl e, for Lapparent de Broin (2007) N. laticentralis differs from the other platychelyids by the wider and longer central fontanela, condition that as was mentioned above potentially represents a male morphological condition for N. laticentralis and N. zapatocaensis and possibly for all males belonging to platychelyids, making this characteristic useless for phylogenetic or systematic porposes, at least at species or genera taxomomic level.

PAGE 32

32 Figure 2 1. A. Location of Zapatoca town, Depart ment of Santander, Colombia 6 50 35N, 73 13 50W. B. Saynoceras verrucosum, ammonite indicator of the base of the Late Valanginian, collected at the same layer from Notoemys zapatocaensis holotype and allotype came from.

PAGE 33

33 FIGURE Figure 2 2. Notoemys zapatocaensis allotype MGJRG IPN 15 EAC 150620061. A, B. Carapace in dorsal view. C, D. Plastron in ventral view. Abbreviations: abd, abdominal; ce, cervical; co costal; ent, entoplastron; epi, epiplastron; fem, femoral; fon, fontanelle; gul, gular; hum humeral; hyo hyoplastron; hyp hypoplastron; intg intergular; ma, marginal; mes, mesoplastron; ne neural; pec pectoral scale; pe peripheral; pl, pleural; su, suprapygal; ve, vertebral xip xiphiplastron.

PAGE 34

34 Figure 2 3. Notoemys zapatocaensis MGJRG IPN 15 EAC 150620062 partial central portion of the carapace including neurals 2 through 8, the most medial portion of the costals 2 through 7, and an isolated most medial portion of the left costal 8 preserving the iliac scar. See areas shadowed on light grey in the turtle sketch. A. Ventral view. B. Dorsal view.

PAGE 35

35 Figure 2 4. Entoplastron and epiplastra evolution in testudines. Sketches of the plastr on were redrawn from previous publications, indicated after the species name. Entoplastron shadowed in black, and epiplastra in gray. A E represent primitive testudines. F H represent platychelyrids, I L represent eupleurodires. M P represent eucryptodires A. Odontochelys semitestacea Li et al. (2008), B. Proterochersis robusta Joyce (2007), C. Proganochelys quenstedti Joyce (2007), D. Kayentachelys aprix Gaffney (1990), E. Indochelys spatulata Datta et al. (2000), F. Platychelys orberndorferi Lapparent de Broin (2000). G. Notoemys zapatocaensis this study, H. Notoemys laticentralis Lapparent de Broin et al. (2007), I. Dortoka vasconica Lapparent de Broin et al. (2004), J. Chelodina oblonga Joyce (2007), K. Bonapartemys bajobarrealis Lapparent de Broin and De la Fuente (2001). L. Podocnemis sextuberculata Joyce (2007), M. Apalone ferox Joyce (2007), N. Eretmochelys imbricata Joyce (2007), O. Mauremys leprosa Claude et al. (2003), P. Kinosternon leucostomun Joyce (2007).

PAGE 36

36 Figure 2 5. Strict consensus cladogram showing the phylogenetic relationships between pleurodiran turtles The nodes are as follow: A Testudines, B Pleurodira, C Platychelira, D Eupleurodira, E node of divergence of Cheloides, F Pelomedusoides. Extant taxa indicat ed with a start superscript. Bootstrapping percentages (upper numbers) was run using 100 branch and bound replicates. Bremer decay indices (lower numbers) were obtained using TreeRot version 3 ( Sorenson and Franzosa, 2007).

PAGE 37

37 Figure 2 6. One of the three most parsimonious trees obtained from the cladistic analysis using branch and bound search, showing Notoemys laticentralis and N. zapatocaensis as sister taxa. Platychelirids shown on light grey rectangle and the three s pecies of Notoemys shown in dark grey.

PAGE 38

38 Figure 2 7. Differences in the xiphiplastra and fontanellas within platychelyrids, differences potentially related with sexual dimorphism. Males (circle with arrow symbol) are characterized by posterior plastral lobe concave, long and narrow posterior xiphiplastral tips, an anal notch well developed in U shaped a large central fontanella. Females (circle with cross down symbol) are characterized by posterior plastral lobe flat, short and wide posterior xipiplastral tips, anal notch in V -shaped, and two interrupted fontanellas. A. Notoemys laticentralis figured in Lapparent de Broin et al. (2007), B. Platychelys orberndorferi figured in Lapparent de Broin (2000), C. N. zapatocaens is this study, D. N. zapatocaensis figured in Cadena and Gaffney (2005)

PAGE 39

39 Table 2 1. Measurements for the platychelyrids, included the allotype of Notoemys zapatocaensis Measures in centimeters. Abbreviations: CL carapace length. CW, carapace width. PL, plastron length. PW, Plastron width. CLe, total carapace length estimated. CWe, total carapace width estimated. PLe, total length plastron estimated. PWe, total width plastron estimated. Taxon CL CW PL PW CLe CWe PLe PWe Notoemys zapatocaensi s MGRG IPN 15 EAC 150620061 this study 20 18 8 15 21 18 18 16 Notoemys zapatocaensis MGRG IPN 15 EAC 140120031 Figured in Cadena and Gaffney, (2005) 22 20 19 14 24 20 19 16 Notoemys laticentralis MOZP 2487. Figured in Fernandez and De la Fuente, (1994) 27 25 24 22 27 25 25 22 Notoemys oxfordiensis MNHNCu P 3209. Figured in De la Fuente and Iturralde, (2001) 25 23 20 20 25 23 22 20 Platychelys oberndorferi Figured in Lapparent de Broin, (2001). 20 17 17 13 20 17 17 13

PAGE 40

40 CHAPTER 3 NEW PODOCNEMIDID TUR TLE (TESTUDINES: PLE URODIRA) FROM THE MIDDLE PALEOCENE OF SOUTH A MERICA. Introduction Pleurodires or side -necked turtles, while currently restricted to freshwater environments of the southern hemisphere, have inhabited freshwater, brack ish, and near coastal environments of most continents since the Early Cretaceous (Danilov and Parham, 2008). They are known from at least 130 species classified in five families (Gaffney et al., 2006): Araripemydidae (Aptian Albian of Brazil), Chelidae (Ea rly Cretaceous Recent of South America and Australia), Euraxemydidae (Albian of Brazil, and Cenomanian of Morocco), Bothremydidae (Albian to Eocene of North America, South America, Europe, Africa and India), and Podocnemididae (Late Cretaceous present of S outh America, Europe, Caribbean, and Africa). Extant podecnemidids include Podocnemis and Peltocephalus of tropical South America, and Erymnochelys from Madagascar (Fig 3 7). Fossil relatives of extant podocnemidids (together termed podocnemidinurans; Ga ffney et al., 2006) include: Brasilemys, Hamadachelys, and Portezueloemys (Table 3 1). An important gap in the record of podocnemidids exists between the Late Cretaceous and the Neogene, particularly for the tropical part of South America. Here we describe the first known Paleogene podocnemidid from the northern neotropics (Figure 3 1) that not only fills this substantial gap in the fossil record, but also provides new morphological data that allow for a direct test of competing phylogenetic and biogeographic hypotheses for extant Podocnemididae. The first hypothesis to explain the current distribution, based on molecular data, indicates that extant podocnemidids are relicts of a once widespread Cretaceous radiation with subsequent local extinctions during t he Cenozoic. These data indicate a close relationship between the geographically disparate Podocnemis from South America and Erymnochelys from Madagascar

PAGE 41

41 (Noonan, 2000). A second hypothesis, based on morphological characteristics and includes fossils, indi cates that the origin and early diversification of the podocnemidinurans occurred in the east -central part of South America during the end of the Early Cretaceous, and was influenced by the separation between Africa and South America in the beginning of the early Cretaceous, combined with dispersal southward into southern South America, Madagascar, and probably Antarctica at that time (Lapparent de Broin, 2000; De la Fuente, 2003). Following this event and the complete isolation of India -Madagascar from Sou th America during the late Cretaceous, there was a split between Erymnochelyinae and Podocnemidinae (sensu Lapparent de Broin, 2000). Two different phylogenetic alternatives have been proposed to explain the relationships between the three extant genera. T he first suggests that both the South American taxa Peltocephalus and Podocnemis are closely related and had an autochthonous origination and speciation in South America, subsequently expanding their distribution northwards during the Cenozoic (Lapparent d e Broin, 2000; Romano and Azevedo, 2006; Lapparent de Broin et al., 2007). The second considers Peltocephalus as more closely related to Erymnochelys implying that both species are relicts of a more widespread clade (Gaffney and Meylan, 1988; Frana and L anger, 2006). Systematic Paleontology TESTUDINES Treviranus, 1802 PLEURODIRA Cope, 1864 PELOMEDUSOIDES Cope, 1868 PODOCNEMIDIDAE Gray, 1825 PODOCNEMIDINAE Cope, 1868 Included Genera: Podocnemis and Cerrejonemys gen. nov.

PAGE 42

42 Amended Diagnosis: Differs from all other known podecnemidids in having: (1) a parietal jugal contact resulting from a relatively reduced postorbital (all others lack this contact, with both bones completely separated by the postorbital), (2) a dorsolongitudinal ridge on the coracoid (all o thers have a smooth dorsal surface and lack the ridge). Remarks: Lapparent de Broin (2000) included Bauruemys aff. Roxochelys vilavilensis Podocnemis Stupendemys and Peltocephalus in the Podocnemidinae based on a single purported synapomorphic characte r: a cervical centra with saddle -shaped posterior condyles. We suggest that this character is shared by all podocnemidids for which the cervical vertebra is known except Erymnochelys (Frana and Langer, 2006), and thus is diagnostic of a more inclusive group than Podocnemidinae. However, a revised Podocnemidinae can be diagnosed based on the two synapomorphies listed in the emended diagnosis above that is restricted to Podocnemis and Cerrejonemys gen. nov. CERREJONEMYS, gen. nov. Etymology: From Cerrej n, t he name of the type locality, and emys from Greek for freshwater turtle. Type species: Cerrejonemys wayuunaiki, sp. nov. Diagnosis: As for the type and only species. CERREJONEMYS WAYUUNAIKI gen. et sp. nov. (Fig. 3 2 A D, 3 3 A D, 3 4 A J, 3 4 M O) Etymo logy: Named for the language (Wayuunaiki) of the Wayuu people from the Guajira Peninsula, Colombia. Type Locality: The La Puente Pit of the Cerrej n Coal Mine (11 08 30 N, 72 33 20 W), Guajira Peninsula, Colombia (Fig. 3 1 A).

PAGE 43

43 Horizon and Age: The fossils were recovered from a layer of claystone underlying Coal Seam 90 in the middle part of the brackish-continental Cerrej n Formation (Bayona et al 2004) (Fig. 3 1 B). The well -preserved palynoflora of the Cerrejn Formation includes Foveotricolpites perforatus Bombacacidites annae and the palynological assemblage indicates a middle late Paleocene age (palynological zone Cu 02; Jaramillo et al., 2007). Other vertebrates include the large boid snake Titanoboa cerrejonensis (Head et al., 2009), dyrosaurid crocodyliforms (Hastings et al., in press), and other pleurodire turtles (Bloch et al., 2005; Cadena et al., 2008). Holotype: UF/IGM 33: skull, lower jaw, anterior part of the carapace, middle part of the plastron, right coracoid, pelvic girdles, an d the sixth and seventh cervical vertebrae. See Table 3 2 for measurements. Diagnosis: Cerrejonemys wayuunaiki differs from all other podecnemidids in having small ventral ridges on the medial margin of the dentary, an acute symphyseal angle between the de ntaries, and a carapace and plastron both reaching a thickness of 35 mm. It further differs from Podocnemis in the absence of an interorbital sulcus at the sutural contact between both prefrontal, a relatively longer prefrontal bone, and the absence of acc essory ridges on the triturating surface of the dentary. Description and Comparisons For the description of Cerrejonemys wayuunaiki we adopted the format used by Gaffney et al. (2006), describing first the state of preservation of each bone, its contacts, and finally comparisons focused principally on podocnemidids. Skull The skull of C. wayuunaiki is known only from a single large (16.7 cm in length), relatively complete specimen (Fig 3 2 A D). The anteriormost portions of both maxillae, the

PAGE 44

44 posterior edg es of both squamosals, and the posterior end of the crista supraoccipitalis are missing. Due to substantial crushing, the left orbital opening is visible in ventral view, and most of the right cavum tympani is visible in dorsal view. Both prefrontals are preserved but are slightly broken. The posterior contact with the frontal is similar to that seen in Brasilemys, Hamadachelys, and all other podocnemidinurans except Dacquemys and Bairdemys in which it is much wider. The anterior protrusion projects slight ly over, and partially covers, the apertura narium externa, ending in an acute tip, similar to the condition in Podocnemis and aff. Roxochelys vilavilensis Bauruemys also has a similar conditon, although in this taxon the tip is less acute. By contrast, t he protrusion of the prefrontals of Daquemys Stereogenys, Bairdemys, Shweboemys antiqua, and especially Peltocephalus and Erymnochelys completely covers the apertura narium externa in dorsal view, with a generally convex anterior edge. The anteromedial c ontact of the prefrontal in Cerrejonemys lacks the interorbital sulcus seen in Podocnemis (Lapparent de Broin, 2000). Laterally the prefrontal contacts the maxilla. The medial length of the prefrontal is as long as that of the frontal, similar to the condi tion in all other podocnemidids except Podocnemis which has a very short prefrontal. In dorsal view, the prefrontal of Cerrejonemys is slightly wider than that of Podocnemis across the orbits, similar to that of aff. Roxochelys vilavilensis and Bauruemys, but narrower than that of other podocnemidids, in which the orbits are more laterally positioned with less dorsal roofing (e.g., Dacquemys Erymnochelys, Bairdemys and Peltocephalus ). The frontals are completely preserved but slightly damaged. The fronta l contacts the prefrontal anteriorly, the other frontal medially, forms part of the orbital margin and contacts the postorbital laterally, and the parietal posteriorly. As such, the frontal is similar to that of all other podocnemidinurans for which the re gion is known.

PAGE 45

45 Both postorbital bones are preserved in dorsal view. Whereas the right postorbital is complete, the left is slightly damaged laterally. As in Podocnemis the postorbital is small and forms part of the orbital margin anteriorly, contacts the frontal medially, the jugal laterally, and the parietal posteriorly. Whereas both parietals are preserved, they are slightly crushed. As a result, they are shifted anteriorly from their original position, resulting in total exposure of the roof of the otic chamber on the right side of the skull. Presumably the original condition of the parietals was more posterior, expanding the secondary roofing of the fossa temporalis (see Laparent de Broin et al., 2007:115 116, for explanation of the evolution of this fossa) and partially covering the roof of the otic chamber in dorsal view, with posterior concave margins as in Bauremys, aff. Roxochelys vilavilensis Bairdemys sanchezi and Podocnemis By contrast, Erymnochelys Peltocephalus Shweboemys antiqua, Neochelys Bairdemys venezuelensis B. hasrsteini, B. winklerae and Dacquemys exhibit secondary roofing of the fossa temporalis and possess more posteriorly expanded posterolateral temporal emargination of the parietals, with straight to convex posterior edges, and a parietal -squamosal contact in the case of Dacquemys In Brasilemys the parietals are highly concave and less advanced posteriorly, so that the roof of the otic chamber is entirely visible in dorsal aspect. This condition is also seen to a slightly mo re advanced degree in Hammadachelys and Portezueloemys The parietal of Cerrejonemys contacts the frontal and the postorbital anteriorly, the other parietal medially, the jugal and quadratojugal (as in Podocnemis ) laterally, and the supraoccipital posterom edially. In Podocnemis erythrocephala the secondary roofing of the fossa temporalis can be more posteriorly advanced with a slight contact between the quadrate and the parietal.

PAGE 46

46 Due to crushing, the contour of the cranial roof and development of a globosit y (sensu Lapparent de Broin, 2000) is indeterminate for Cerrejonemys. The right jugal is preserved and completely exposed on the dorsal surface, whereas the left is poorly preserved on the ventral surface due to crushing. The jugal contacts the maxilla and the orbit anteriorly, the postorbital and the parietal (as in Podocnemis ) dorsomedially, and the quadratojugal posterolaterally. The jugal plays a key role in the secondary lateral roofing of the fossa temporalis with a decrease in the amount of cheek or lateral emargination (see Lapparent de Broin, 2007:115 116, for explanation of the evolution of this character). In podocnemidids this lateral emargination is dominated by the jugal, and in bothremydids by the quadratojugal. Unfortunately, in Cerrejonemys the secondary closure of the cheek emargination is difficult to determine because of damage, but it seems to be much less advanced than in Erymnochelys and Peltocephalus and similar to that seen in Podocnemis Both quadratojugals are fairly well preserved in dorsal aspect, although the left is poorly preserved in ventral aspect. The quadratojugal contacts the jugal anteriorly, the parietal medially, and the quadrate and the squamosal posterolaterally. The posteromedial edge of the quadratojugal forms part of the temporal emargination. In all ways, the quadratojugal is similar to that of Podocnemis The right squamosal is visible in dorsal aspect, whereas the left is covered by the quadrate in ventral aspect of the skull and only its posteromedial aspect is visible. The squamosal of Cerrejonemys contacts the quadratojugal anteriorly, the quadrate anterolaterally, and the opisthotic medially. In this way, it is similar to all other known podocnemidids, although there is an additional contact with the parietal in Dacquemys

PAGE 47

47 The right premaxilla is missing and most of the left is obscured by the right maxilla because of crushing. However, a poorly developed anteroventral hook is present, as in all other known podocnemidids, particularly in Erymnochelys and Peltoc ephalus in which the premaxilla hook is highly developed. Both maxillae are present although slightly crushed, and the right is better preserved than the left. The dorsal surface of the left maxilla is visible in ventral view and covers part of the right maxilla and a large portion of the right premaxilla. The maxilla contacts the prefrontal medially and the jugal posteriorly. The ventral contacts are with the palatine posteromedially and with the jugal posteriorly. Cerrejonemys lacks accessory ridges on t he ventral surface of the right maxilla. It is similar to that of all podocnemidids except Podocnemis, for which two or more accessory ridges reach the premaxilla, and Dacquemys for which the ridges do not reach the premaxilla. On the dorsal surface of the skull the foramen supramaxillare appears in the lower posterior aspect of the orbit, as is the common condition in modern Podocnemis Peltocephalus aff. Roxochelys vilavilensis, Neochelys arenarum and probably other fossil podocnemidids for which this region is covered with matrix or not preserved. This suggests that the presence of a foramen supramaxillare is not exclusive to P. expansa (Joyce, 2007). The vomer is absent in Cerrejonemys, a condition similar to that described for most podocnemidids, exc ept Bauruemys aff. Roxochelys vilavilensis, Podocnemis bassleri, and Podocnemis vogli In Podocnemis unifilis presence of the vomer is variable. Both palatines are preserved. The right one is fully exposed but slightly damaged, whereas the left one is hea vily damaged and only partly discernible in medial aspect. The palatine contacts the maxilla anterolaterally, the other palatine medially, the jugal laterally and the

PAGE 48

48 pterygoid posteriorly. Anteriorly, the palatine forms the posterior margin of the apertur a narium interna. The foramen palatinum posterius is very close to or intercepts the palatine pterygoid suture in Cerrejonemys. A similar condition is present in Brasilemys, Portezueloemys, Hamadachelys, and in most podocnemidids, except Dacquemys, Stereog enys, and Shweboemys antiqua in which this condition is absent. In Podocnemis and Bairdemys (except for B. sanchezi which lacks the foramen) the foramen palatinum posterius is generally restricted to the palatine, well separated from the palatine pterygoi d suture. In Podocnemis expansa the foramen can be very close to the palatine pterygoid suture or it is restricted to the palatine as in the other species of Podocnemis Both pterygoids are preserved although only their ventral surfaces are visible. The pt erygoid contacts the palatine anteriorly, the other pterygoid medially, and the basisphenoid posteromedially. The processus trochelaris pterygoidei projects almost directly laterally into the center of the fossa temporalis. This is similar to the condition of most other podocnemidinurans except Bauruemys aff. Roxochelys vilavilensis and Peltocephalus in which the processus projects more obliquely with respect to the midline of the skull, and not as far into the fossa in the case of Peltocephalus The pte rygoid flange (Frana and Langer, 2006) or posterolateral wing (Lapparent de Broin, 2000) of the pterygoid although crushed in Cerrejonemys is well developed posterolaterally and almost completely covers the cavum pterygoidei (sensu Gaffney et al., 2006; fossa podocnemidoidof Lapparent de Broin, 2000) and extends to the caudal margin of the quadrate ramus. A similar condition is present in Bauruemys aff. Roxochelys vilavilensis

PAGE 49

49 Bairdemys, and Podocnemis bassleri In extant podocnemidids the pterygoid flange exhibits a similar condition, but often projects ventrally. The basisphenoid is completely preserved in Cerrejonemys, but only the ventral surface is clearly visible. It contacts both pterygoids anterolaterally, both quadrates posterolaterally, and the basioccipital posteriorly. In these features it is similar to that of all other podocnemidinurans. The basioccipital is complete in Cerrejonemys. Only the ventral surface and portions of the posterodorsal surfaces are clearly visible. The basioccipita l contacts the basisphenoid anteriorly, the quadrate laterally and, although the posterodorsal surface is completely crushed, appears to contact the exoccipital and participates in the structure of the condylus occipitalis. This is similar to the condition in all other podocnemidinurans and many other pleurodires, except in pelomedusids and some bothremydids for which the basioccipital does not form part of the condylus occipital. Both exoccipitals are preserved in Cerrejonemys. Only the right exoccipital e xhibits discernible contacts on the dorsal, posterior, and ventral surfaces. On the dorsal surface of the skull, the exoccipital is in contact with the supraoccipital dorsally, opisthotic laterally, quadrate ventrolaterally, and the basioccipitalis ventrom edially. On the posterior surface there is evidence for the entrance of the foramen jugulare posterius, but damage makes it impossible to determine the size or its direction into the bone. The exoccipital also constitutes a major part of the condylus occip italis, as in all other podocnemidids. The crista supraoccipitalis of the supraoccipital is distorted and its posterior tip is damaged. The entire structure has been rotated 90 from its original position, such that the dorsal edge of the crista supraoccipitalis is now oriented laterally. The supraoccipital contacts the prootic anterolaterally, the opisthotic laterally, and the exoccipital posterolaterally. There is

PAGE 50

50 slight dorsomedial contact with the parietal. The crista supraoccipitalis is long, flat, and maintains a uniform width along its ventral base from anterior to the posterior aspect, similar to the condition in most extant and fossil podocnemidids. Bairdemys differs from Cerrejonemys and all other podocnemidids in having a short crista supraoccipit alis that is wider posteroventrally than anteroventrally, and that ends in a bulbous shape in dorsal view. Both opisthotics are preserved, but only the right one is completely exposed on the dorsal aspect. The opisthotic contacts the quadrate anterolateral ly, the squamosal posterolaterally, the exoccipital posteromedially, the supraoccipital anteromedially, and the prootic anteriorly. These contacts are similar to the condition found in all podocnemidinurans except in Brasilemys in which there is no contac t between the opisthotic and the prootic beacuse these bones are separated by the supraoccipital. The processus paroccipitalis in Cerrejonemys is medially narrow, elongate, and projects beyond the squamosal, ending in a tip that is broken on both sides. A similarly shaped processus paroccipitalis is seen in most all podocnemidinurans, except Brasilemys, Shweboemys, and Bairdemys, which have a small, flat processus paroccipitalis that does not project beyond the squamosal. The right prootic is exposed in dor sal aspect, although its anterior end is obscured by the quadratojugal. It contacts the opisthotic posteriorly, the quadrate laterally, and the supraoccipital parietal medially. The foramen stapediotemporale is clearly visible in the contact between the pr ootic and the quadrate, as in all other pleurodires. Both quadrates are preserved, but only the right exhibits discernible contacts. Dorsally the quadrate contacts the prootic anteromedially, the opisthotic posteromedially, the squamosal posterodorsally, a nd the quadratojugal anterodorsally. Ventrally the contacts are with the

PAGE 51

51 pterygoid anteromedially, the basisphenoid medially, the opisthotic posteromedially, and squamosal posteriorly. The medial contact with the prootic is not visible. The quadrate is clo sed ventrolaterally around the cavum tympani, and is directed ventrally as in all podocnemidinurans except Brasilemys (Lapparent de Broin et al., 2007), although it is even more downwardly elongate in Shweboemys Stereogenys and in Bairdemys The right qu adrate preserves the cavum tympani. Due to crushing, the shape and position of the incisura, the columella auris, and Eustachian tube are not discernible. In right posterior aspect of the cavum tympani there is a shallow cavity that, while crushed and dist orted to appear somewhat smaller than that of other podocnemidids, is likely the fossa precolumeralis. A small antrum postoticum, similar in size to that of other known podocnemidids, is present on the posterior part of the right quadrate. The right condyl us mandibularis is crushed and deformed. The left is completely covered by the quadrate. Lower Jaw Whereas the lower jaw of UF/IGM 33 is considerably crushed dorsoventrally, it is still fairly complete, with only the most lateral portion of the left ramus at the processus coronoideus of the dentary and the area mandibularis of the right ramus missing (Fig 3 3 A D). The dentary contacts are indeterminate in the right ramus due to the slightly eroded bone surface, but are apparent in the left ramus. The den tary contacts the coronoid posterodorsally, the angular posteroventrally, and the surangular posterolaterally. Both dentaries are fused at the mandibular symphysis, as in all other podocnemidinurans. This is also very probably the condition in Brasilemys, for which only the left ramus is preserved (Lapparent de Broin, 2000). Both Cerrejonemys and a recently described indeterminate podocnemidid from the Miocene of Venezuela (UNEFM CIAPP 1399; Gaffney et al., 2008)

PAGE 52

52 have a very acute (less than 40) internal a ngle between rami in ventral view. In contrast, all other podocnemidinurans have a less acute angle (over 40) with the exception of Bairdemys in which this angle is greater than 90. In Cerrejonemys the triturating surface on the dorsal surface of the de ntary is consistently wide from the symphysis to the coronoid region, as in all other podocnemidinurans except Erymnochelys aff. Roxochelys vilavilensis and Neochelys in which the symphysis is slightly narrower, and in Bairdemys which has a much wider triturating surface at the symphysis than at the coronoid region. In addition, the triturating surface of Cerrejonemys lacks accessory ridges, as in most of podocnemidinurans apart from Podocnemis The triturating surface is bound by lingual and labial rid ges. As in other podocnemidids, the lingual ridge in Cerrejonemys is higher than the labial posteriorly. In contrast, aff. Roxochelys vilavilensis and Peltocephalus have lingual and labial ridges that are equally high posteriorly. The lingual ridge of Cerr ejonemys is nearly straight rather than the sigmoidal condition common to bothremydids (Gaffney and Foster, 2003). The sulcus cartilaginis meckeli is strongly marked on the medial surface of both dentaries in Cerrejonemys, and it is considerably elongated anteriorly, as in other podocnemidids. A narrow elongated ridge on the ventral surface is preserved on both dentaries of Cerrejonemys The ridge projects anteriorly from the medial margin of the ramus toward the symphysis area, at which point it disappears completely. These ridges are exclusive to Cerrejonemys within the podocnemidinurans. The anteroventral contact with the dentary is visible on the right angular. Otherwise, both angulars are severely crushed and all other sutural contacts are unrecognizabl e.

PAGE 53

53 Only the anteromedial part of the right angular is preserved, and participates in the lateral wall of the fossa meckeli. Its anterodorsal contact with the coronoid and anterolateral with the dentary are the only recognizable contacts for this bone. The right coronoid, although slightly crushed, is completely preserved, and is similar in height to Podocnemis and other podocnemidids. It contacts the dentary anterolaterally, the surangular posterolaterally, and the prearticular ventromedially. A very small dorsomedial portion of the left coronoid is preserved, but without any recognizable contacts. Both prearticulars are preserved, although slightly crushed, and their contacts with the angular and the articular are indeterminate. The anterodorsal process t hat covers the fossa meckelii and connects the prearticular with the coronoid is broken on both sides, exposing the fossa meckelii and the foramen intermandibularis. The left articular is fairly complete, whereas only the anterior end of the right is pres erved. The contacts with the subangular and prearticular are indeterminate. The processus retroarticularis, although poorly preserved, seems to project posteroventrally as in Podocnemis and aff. Roxochelys vilavilensis This is in contrast to all other podocnemidids plus Brasilemys and Hamadachelys in which the process extends more posteriorly, with variation in length among the different taxa. For example, in Peltocephalus the process is slightly shorter than in Erymnochelys The dorsal surface of the art icular, which articulates with the condilus mandibularis at the posteroventral region of the skull is slightly wider at its midpoint than at its lateral and medial margins, with a convex posterior edge. This could indicate that the condilus mandibularis of the quadrate was kidney -shaped, although more complete material is necessary to assess this interpretation more confidently. A kidney-shaped condilus mandibularis is exclusive to

PAGE 54

54 Podocnemis within the podocnemidinurans, and its presence in Cerrejonemys mi ght indicate a close relationship with that taxon. Cervical Vertebrae Fairly complete sixth and seventh cervical vertebrae constitute what is known of the axial skeleton of Cerrejonemys (Fig 3 4 G I, J K). The ventral portion of the sixth cervical, includ ing the posterior condyle, and part of both transverse apophyses are preserved, albeit considerably crushed. A notable feature of this vertebra, is the saddle -shaped posterior condyle, which is higher than wide, dimensions that are characteristic of cervical vertebrae of Peltocephalus Bauruemys aff. Roxochelys vilavilensis Podocnemis, and Stupendemys souzai (Williams, 1950; Lapparent de Broin, 2000). Although, Lapparent de Broin (2000) suggested less pronounced saddle -shaped condyles for the second throu gh sixth vertebra for Peltocephalus this is also the condition for Podocnemis expansa figured by Hoffstetter and Gasc (1969:fig 12) and Cerrejonemys. This indicates that the saddleshaped condyle for the seventh cervical in podocnemidids is variably prese nt. The left lateral part of the seventh cervical is nearly complete, except for the corner of the anterior articular surface of the centrum and the lateralmost margin of the transverse apophyses. However, only the medial aspect of the neural arch and the condylar region are preserved on the right side. The centrum of Cerrejonemys is similar to Podocnemis in being elongate, procoelus, and in lacking a ventral keel. The ventral keel is present in almost all other podocnemidids for which cervical vertebrae ar e known. A ventral keel has also been described for the bothremydid Acleistochelys (Gaffney et al., 2007). Similar to the condition in Podocnemis expansa (Hoffstetter and Gasc, 1969:fig 12), Peltocephalus, and variable for Erymnochelys the posterior condyle of the seventh cervical is spherical and slightly taller than wide with the dorsal edge slightly concave in Cerrejonemys.

PAGE 55

55 The prezygapophyses of the seventh cervical of Cerrejonemys are long and project almost vertically toward the vertebral centrum, a s in Podocnemis Erymnochelys, and Stupendemys souzai but in contrast to the slightly shorter prezygapophyses in Peltocephalus The transverse apophyses are located at the midline of the centrum as in all podocnemidinurans and the postzygapophyses are low and project posterodorsally, as in Podocnemis This differs from that of Peltocephalus, Stupendemys souzai and Erymnochelys, which have more vertically oriented postzygapophyses. Additionally, both postzygapophyses of Cerrejonemys are fused at the top of the pedicel, indicating the likely presence of collarette shape postzygapophyses as is common for podocnemidids (Lapparent de Broin et al., 2007). On the lateral surface of the pedicel, a deep concavity marks the juncture point of the prezygapophyses with the eighth cervical. Carapace The anterior region of the carapace is preserved for Cerrejonemys (Fig 3 4 A B), including the nuchal; right and left peripheral 1 and 2; right peripheral 3; neurals 1 3; right and left costal 1 and 2; and right costal 3. Whe reas a small portion of the lateral margin of left costal 1 is crushed, the original curvature of all other elements is preserved. The carapace is slightly oval in shape and forms a low dome, as in most podocnemidids. In Cerrejonemys the dorsal surface of the carapace is smooth, and thus is similar to that in all other podocnemidids except Roxochelys harrisi which exhibits marked reticulation in the form of small polygons or dichotomous sulci (Lapparent de Broin, 1991) on the dorsal surface. Cerrejonemys h as the thickest shell of all known podocnemidids, approaching an average thickness of 35 mm along the midline of the carapace and plastron. The nuchal bone is pentagonal in shape and wider than long, with a straight anterior edge and a slightly curved post erior margin. This is similar to the condition seen in all

PAGE 56

56 podocnemidinurans except Cambaremys, which has a longer than wide nuchal bone (Frana and Langer, 2005). Neural 1 is subrectangular in shape, almost twice as long as wide, slightly convex on its la teral and anterior edges, and with lateral contact restricted to costal 1 on both sides. This lateral contact is found in all podocnemidids except Bauruemys and the podocnemidinuran Brasilemys for which neural 1 laterally contacts right and left costal 1 and 2, and neural 2 is small square shaped. In the case of Brasilemys the neural series is more irregular in shape, a condition seen in basal pleurodires such as Platychelys and Notoemys (Cadena and Gaffney, 2005). A particular case is seen in the podocnem idinuran Portezueloemys (De la Fuente, 2003), which has neural 1 with a restricted lateral contact with costal 1 on its right lateral margin, as in most of podocnemidids, whereas on its left margin, the neural 1 contacts the costals 1 and 2 as in Brasilemy s and Bauruemys Whether or not this dual condition for the lateral contacts of neural 1 is a pathologic effect particular to that specimen of Portezueloemys or if is actually evidence for an intermediate stage in the evolution of the condition seen in pod ocnemidids will only be known with discovery of additional fossils of Portezueloemys. Neural 3 of Cerrejonemys is hexagonal in shape and contacts the costal 2 anterolaterally and would have contacted neural 4 posteriorly (although it is missing in this spe cimen). In Cerrejonemys costal 1 has convex anterior and posterior margins that meet laterally. The length of costal 1 is slightly more than twice the length of costal 2,a dimension that is similar to that of some species of Podocnemis. Peripheral 1 is su brectangular in shape with the anterior margin wider than the posterior, and a curved medial contact with the nuchal. Peripheral 2 is trapezoidal in shape and peripheral 3 is rectangular. The carapace of Cerrejonemys lacks the cervical scale, as do all pel omedusoides, but this is not exclusive to this group (Lapparent de Broin, 2000). Vertebral scale 1 is wider anteriorly,

PAGE 57

57 almost pentagonal in shape, with convex anterior and lateral edges. It covers most of the anteromedial corner of costal 1, the posterior area of peripheral 1, and the medial to posterior area of the nuchal. Vertebral scale 2 is hexagonal in shape. It medially covers the posterior area of neural 1, neural 2, and most of neural 3. It laterally covers the posteromedial corner of the costal 1, the medial portion of costal 2 and the anteromedial corner of costal 3. In all these respects, vertebral scale 2 is similar to that of all known podocnemidinurans. The marginal scales are confined to the peripherals. Marginal 1 is rectangular, wider than long, and covers the anteromedial part of the nuchal and a small portion of the anteromedial part of peripheral 1. Marginal 2 is larger than marginal 1, almost completely covering peripheral 1 and the anteromedial part of peripheral 2. The lateral contact between right marginal 3 and 4 occurs on Peripheral 3. The sulcus between the pleural scale 1 and 2 is poorly marked on both right and left costal 2, although it is clearer on the left costal. On the ventral surface the axillary buttress scar is deeply mar ked and located at the midline of costal 1 as in most of podocnemidids. In Roxochelys harrisi, aff. Roxochelys vilavilensis, and Erymnochelys the axillary scar is located slightly closer to the contact between costals 1 and 2. Peltocephalus has an axillary buttress scar situated more laterally on costal 1 than in the other podocnemidids. A particular case is present in Bairdemys in which the neural bones are completely absent, so that the axillary buttress scar is situated more medially on costal 1. In Cer rejonemys the projection of the axillary scar onto the peripherals reaches the anterior margin of peripheral 3, as in Podocnemis lewyana, Podocnemis negrii and Erymnochelys In all other podocnemidinurans the axillary scar projection enters onto the center or at the posterior margin of peripheral 3 or on peripheral 4, as is the most common condition for Peltocephalus

PAGE 58

58 Plastron Plastral bones recovered include the left and right hypoplastra, mesoplatra, and hyoplastra with the l ast slightly broken anteriorly (Fig 3 4 C D). As is the case in the carapace, the plastral elements are nearly 35 mm thick. The mesoplatra are hexagonal in shape with the posteromedial edge slightly curved, which is typical of that in other podocnemidids. In Cerrejonemys and most podoc nemidids the pectoroabdominal sulcus does not cross the mesoplastron; occasionally a slight contact with the anterior edge of the mesoplastron is seen in some of species of Podocnemis, but it never crosses onto the mesoplastron. An exception to the podocne midid condition is found in Neochelys in which the sulcus crosses the anteromedial margin of mesoplastron, and Peltocephalus in which both conditions are variably expressed. Coracoid The only element of the pectoral girdle preserved in Cerrejonemys is the right coracoid (Fig 3 4 E F). A small portion of its medial margin along the middle part of the bone and the most posterolateral corner are missing. The coracoid of Cerrejonemys is a long bone with a proximal articulation and a lateral body. It is cylindrical proximally and extends longitudinally toward the distal end where it is flatter and slightly divergent. The dorsal surface exhibits a marked longitudinal ridge, previously reported as being exclusive of Podocnemis by Frana and Langer (2006). However we have seen that the some specimens of Podocnemis vogli lack this ridge. The ventral surface of the coracoid of Cerrejonemys, Podocnemis and occasionally in Erymnochelys is concave, relatively deep laterally and flat distally. In contrast, the ventral surface of the coracoid of Peltocephalus, Cambaremys Bauremys, and aff. Roxochelys vilavilensis is nearly flat, without a marked concavity.

PAGE 59

59 Pelvic Girdle The left side of the pelvis is fairly complete but the antero and posteromost portions of the right side of the pelvis are missing (Fig 3 4 M N). The left side preserves a complete ilium and a pubis that is slightly broken on its distal margin. The epipubis and the most proximal area of the ischium are recognizable in the acetabulum capsule. The sutural contact between the ilium and the pubis is visible on lateral and medial surfaces. On the right side, the ilium and a considerably damaged part of the acetabulum capsule with the most proximal portions of the pubis and ischium are the only elements preser ved. In the comparable aspects for which the morphology is preserved the pelvis of Cerrejonemys is similar to that of all podocnemidids and other pleurodires. Phylogenetic analysis To examine the phylogenetic implications of Cerrejonemys we included it in a cladistic analysis with other known podocnemidinurans that are adequately known from skull, shell, or postcranial elements. Cambaremys, Shweboemys gaffneiy, Shweboemys pilgrimi, Shweboemys pisidurensis, Podocnemis pritchardi, Podocnemis medemi, Podocnem is negrii, Neochelys capellini, Roxochelys harrisi, and Stupendemys were excluded from this analysis due to missing data. A fragmentary skull of Podocnemis cf. P. expansa, which lacks a detailed published description and has been lost since its original pu blication (Wood, 1997), was also excluded for lack of data. However, most of the excluded taxa are considered in the comparisons. We assembled a matrix of 24 ingroup taxa (podocnemidinurans) and 5 outgroup taxa (Chelidae, Pelomedusidae, Euraxemydidae, Arar ipemydidae, and Bothremydidae; rooted to Chelidae) that were scored for the 53 morphological characters listed in Appendix C and coded in Appendix D. Most of the characters were modified from previously published character matrices and detailed systematic studies including Meylan (1996), Lapparent de Broin (2000),

PAGE 60

60 Gaffney et al. (2002), Gaffney and Forster (2003), De la Fuente (2003), Frana and Langer (2006), Gaffney et al. (2006), Lapparent de Broin et al. (2007), and Gaffney et al. (2008). A few of these characters are new to this study and were defined based on direct examination of fossil and modern specimens. The character matrix was constructed using Mesquite Version 2.5 (Maddison and Maddison, 2008) and analyzed using the parsimony algorithm of PAUP 4.0b10 (Swofford, 2002). The matrix is available as a Nexus file in the Supplementary Data 3 2. All characters were equally weighted and unordered. Multistate characters were treated as polymorphic. We performed a branch and bound search in PAUP. Decay indices were computed in TreeRot version 3 (Sorenson and Franzosa, 2007) and boostrap percentages were computed in PAUP (100 branch and bound replicates). Results The cladistic analysis resulted in 1,296 most parsimonious trees (length = 117 steps, consistency index = 0.83, retention index = 0.90, homoplasy index = 0.19). The strict consensus of the 1,269 most parsimonious trees is shown in Figure 3 5. In this consensus Cerrejonemys falls out as the sister taxon to a monophyletic (but unresolved) clade includ ing all species of Podocnemis Discussion Our phylogenetic results suggest that the presence of the cavum pterygoidei is a synapomorphy for the podocnemidinuran clade and is consistent with previous works in the basal placement of Brasilemys Hamadachelys, and Portezueloemys (e.g., Lapparent de Broin, 2000; De la Fuente, 2003; Romano and Azevedo, 2006; Gaffney et al., 2006). Brasilemys was recently excluded from the podocnemidinuran clade (Frana and Langer, 2006) based on the

PAGE 61

61 following characters: (1) pres ence of a large antrum postoticum, (2) lack of a contribution of the palatine to the triturating surface, and (3) presence of a well -developed pterygoid flange. However, we note that: (1) a large antrum postoticum is also present in some bothremydids, such as Galianemys and this character is generally widely variable within Pelomedusoides (Gaffney et al., 2006), (2) lack of a contribution of the palatine to the triturating surface, which is seen also in some bothremydids, such as Labrostochelys galkini and Taphrosphys ippolitoi (Gaffney et al., 2006) and may also be the condition in Podocnemis erythrocephala; and (3) the pterygoid flange is fairly developed in Brasilemys much less than as is suggested by Frana and Langer (2006). Thus, we regard Brasilemys and Hamadachelys to be podocnemidinurans based on the presence of a shallow cavum pterydoidei that is hidden anteromedially by the underlapping basisphenoid medially and the pterygoid laterally. Brasilemys and Hamadachelys are excluded from Podocnemididae because they lack a deep cavum pterygoid that is partially to totally covered by the pterygoid flange. Our results agree with those from previous analyses that exclude Bauruemys from Podocnemidinae (Frana and Langer, 2006; Romano and Azevedo, 2006). In c ontrast, Lapparent de Broin (2000) considered Bauruemys to be a member of the Podocnemidinae based on the presence of a cervical vertebra with a saddle -shaped condyle, a condition also shared by aff. Roxochelys vilavilensis, Cerrejonemys, Stupendemys souza i, Podocnemis, and Peltocephalus making this character a potential synapomorphy for the Podocnemididae family, with the exception of Erymnochelys, which exhibits the reversed condition (Frana and Langer, 2006). We note that many podocnemidid taxa are sti ll unknown for this character and that only further fossil discoveries will help to test the validity of this character as a synapomorphy for the Podocnemididae family.

PAGE 62

62 Results from our analysis suggest that Bauruemys and aff. Roxochelys vilavilensis shoul d be excluded from the rest of podocnemidids based on the presence of: (1) a coracoid bone that is slightly curved longitudinally and much wider distally; and (2) a secondary roofing of the fossa temporalis that is medially advanced with concave margins, p artially covering the otic chamber in dorsal view, a condition slightly more advanced in Cerrejonemys and Podocnemis. In contrast to the hypothesis of Lapparent de Broin (2000) our results suggest that Erymnochelyinae and Podocnemidinae constitute two clea rly separate clades. We suggest that the Erymnochelyinae, is constituted by Neochelys, Erymnochelyini ( Erymnochelys and Peltocephalus ), and Shweboemydini ( Dacquemys, Bairdemys, Shweboemys, and Stereogenys ), and is supported by two synapomorphic characteris tics: (1) a very advanced secondary roofing of the fossa temporalis, with convex to straight, tappering margins that totally cover the otic chamber roof in dorsal aspect (character 6, Appendix 3 1); and (2) an anterior protrusion of the prefrontal onto the apertura narium externa, totally covering the apertura, with its convex edge visible in dorsal view of the skull (character 7, Appendix 3 1). While the strict consensus tree shows unresolved positions for Neochelys arenarum and N. lapparenti within Erymno chelyinae, we favor the idea that Neochelys is more closely related to Erymnochelyini, and particularly related to Erymnochelys as was also suggested by Lapparent de Broin (2000) based on the presence of a large intergular scale, covering the anterior mar gin of the entoplastron and separating the gulars (character 53, Appendix 3 1), a condition present in Neochelys arenarum. Shweboemydini is supported by the lack of foramen palatinum posterius, with a reversal in the Bairdemys venezulensis B. harsteini, a nd B. winklerae (character 29, Appendix 3 1), which exhibit this foramen. Within Shweboemydini, Dacquemys is the most basal, in part because lacks a secondary palate (character 30, Appendix 3 1), in contrast to the others which have a

PAGE 63

63 secondary palate, wit h all Bairdemys species additionally having a second palate with ventral convexities. Additionally, all species of Bairdemys have a uniquely long downward projection of the quadrate that strongly separates the condylus mandibularis from the cavum tympani r egion (character 18, Appendix 3 1). It has been suggested that the evolution of a secondary palate may have happened more than once in Shweboemydini, possibly as an adaptation to facilitate the crushing of mollusks (Wood, 1984). If that were the case, then the support for Shweboemydini affinities of Bairdemys would be weak. However, the recently described Bairdemys sanchezi (Gaffney et al., 2008) retains the plesiomorphic condition seen in Dacquemys Shweboemys and Sterogenys in lacking a foramen palatinu m posterius, and thus seems to represent a morphologic and phylogenetic intermediate between the more primitive Shweboemydini and the more derived species of Bairdemys in which the foramen has re -evolved. A clade composed of Erymnochelys and Peltocephalus (Erymnochelyini), is supported by one clear synapomorphy: a very advanced secondary roofing of the cheek emargination by the descending jugal -quadratojugal. This condition results in a contact between the quadrate and the jugal (character 20, Appendix 3 1) In lateral view, the edge of the secondary roofing is almost parallel to the maxillary edge in most specimens, but occasionally a small notch is present at the posterolateral margin of the jugal with slightly less advanced secondary roofing. A second synapomorphy for Erymnochelyini has been discussed in the literature (i.e., Frana and Langer, 2006; Lapparent de Broin, 2000): the anteriorlyunrestricted roofing of an enlarged carotid canal, although the condition is less emphasized in Peltocephalus than E rymnochelys We note that the state of this character is unknown for most fossil podocnemidids, and for that reason we have excluded it from our phylogenetic analysis.

PAGE 64

64 A clade composed of Podocnemis and Cerrejonemys (Podocnemidinae, sensu stricto) is suppo rted here by the following synapomorphies: (1) a parietal -jugal contact related to a reduction of the postorbital (character 11, Appendix 3 1); and (2) a dorsal longitudinal ridge on the coracoid (character 44, Appendix 3 1). Among Podocnemidids, one of th e unique characteristics of Podocnemis is the presence of a slightly wider than long, kidney-shaped condylus mandibularis, with a straight to concave anterior edge and convex posterior edge (Fig. 3 6). However, because the region has not been recovered in Cerrejonemys, it is not yet possible to determine whether it represents an additional synapomorphy for Podocnemidinae. Relationship between Extant Podocnemidids The exact phylogenetic relationships between the three modern genera of the Podocnemididae ( Pod ocnemis Erymnochelys and Peltocephalus ) have been highly controversial (Noonan, 2000; Lapparent de Broin, 2000; De la Fuente, 2003). Results from our analysis provide support for a close relationship between Erymnochelys and Peltocephalus to the exclusion of Podocnemis This relationship is supported by two characters not previously used in phylogenetic analyses for the group: (1) the shape of the condylus mandibularis, in which Peltocephalus and Erymnochelys retain the primitive condition in being much w ider than long, with the anterior and posterior edges straight to concave making it shorter at the midline; this contrasts with the derived condition present in Podocnemis as described in previous section (Fig. 3 6 I F); and (2) the edge of the anterior p rotusion of the prefrontal is convex and completely covers the apertura narium externa in both Peltocephalus and Erymnochelys The latter condition is also present in Shweboemydini and in Neochelys In contrast, the edge of the anterior protrusion of the p refrontal projects slightly over, and partially covers, the apertura narium externa, ending in an acute tip in Podocnemis and Cerrejonemys.

PAGE 65

65 The morphological evidence presented here suggests that Peltocephalus and Erymnochelys are more closely related to e ach other than either are to Podocnemis However, it is only with new fossil discoveries, including elements such as cervical vertebrae, the coracoid, and skulls, will further resolution of podocnemidid phylogeny be possible. This is particularly the case for members of the Erymnochelyinae, Neochelys and the newly described Cerrejonemys Paleobiogeographical Scenario During the middle late Paleocene, the Cerrejn Formation was deposited as part of the Maracaibo crustal block, which at that time was in its southwestern most position, 5 6 degrees further south than today (approximately 11) (Montes et al., 2005:fig. 16). As such, the paleolatitude of the Cerrejn flora and fauna is firmly within the tropics. The oldest known podocnemidid is from the Upper Cre taceous of Brazil (Frana and Langer, 2006). Furthermore, based on the Late Cretaceous occurrence of the oldest erymnochelyine from Madagascar (Gaffney and Forster, 2003), the split between Podocnemidinae and Erymnochelyinae (subfamilies of Podocnemididae) must have occurred before then (Romano and Azevedo, 2006). However, prior to this study the oldest podocnemidine was from the Miocene of La Venta (Wood, 1997). Occurrence of the podocnemidine Cerrejonemys, the sister taxon of a clade that includes modern Podocnemis during the middle late Paleocene, reduces the ghost lineage for this clade by approximately 47 million years and provides strong support for the proposed vicariance scenario for the origin of these clades associated with the separation of South America and India/Madagascar at the end of the Cretaceous (Romano and Azevedo, 2006). As part of this model, it has also been suggested that Podocnemidinae would have originated in the southern part of South America, based on the southern occurrence of th e oldest known podocnemidid (Romano and Azevedo, 2006). Assuming this is true, and based on the occurrence of Cerrejonemys in the tropics, it is clear that

PAGE 66

66 podocnemidids must have moved north prior to the middle late Paleocene. What is less clear is the ti ming of dispersal for closelyrelated fossil taxa including Shweboemydini and Neochelys which were widely distributed during the Cenozoic. Despite the paucity of relevant data to test hypotheses about the timing and routes in which podocnemidids arrived a nd colonized the northernmost tropical corner of South America, we consider two possible routes. The first could have been from the southeastern part of the continent, moving northward along the eastern coastal margin of South America finally reaching the northeastern corner of the continent, in a similar way that other pelomedusoides such as bothremydids and Hamadachelys dispersed from the southeastern part of South America, towards the northwestern part of Africa and Western of Europe (Romano and Azevedo, 2006). The second possible dispersal route could have been from the southcentral part of South America moving northward using foreland basins developed in the Altiplano plateu during the Paleogene (Horton et al., 2001). The latter hypothesis may be suppor ted by the occurrence of the podocnemidid aff. Roxochelys vilavilensis from the early Paleocene, Tiupampa Basin, Bolivia (Lapparent de Broin, 1991). However, this scenario is complicated by the lack of evidence for a complete fluvial or seaway connection b etween the northern and southcentral basins of South America during the late Cretaceous Paleocene, which would have been required for the dispersal of aquatic faunas from Tiupampa northward. Following the Paleocene, the most important documented events in the geological history of the tropical part of South America occurred during the Neogene. These events had a strong influence over the distribution, diversification, and exctinction of aquatic vertebrates (e.g., Albert et al., 2006). The first of these events, corresponding to the uplift of the Eastern Cordillera (~ 12 Ma), would have isolated podocnemidids and chelids such as Podocnemis pritchardi, Podocnemis medemi, and Chelus

PAGE 67

67 colombiana, inhabiting the Magdalena Basin, from the podocnemidids and chelids inhabiting the proto -Orinoco river ( Bairdemys and Chelus lewisi). The second event (see Albert et al., 2006) is the hydrological capture of the Amazon River by the eastern Amazon Basin from the western Amazon Basin with the formation of the east -flowing m odern Amazon River (~ 9 Ma). This event, which may have resulted in a larger area and more diverse available habitats, could have influenced the diversification of Podocnemis The third event, the rise of the western portion of the Merida Andes (~ 8 Ma), i solated the modern Maracaibo and Orinoco basins. The fourth event was the rise of the Isthmus of Panama (~ 3 Ma). The latter two events events could have caused the geographic restriction of some species and also local extinctions due to an increase in eco logical competition with other freshwater turtles such as cryptodires arriving from North and Central America.

PAGE 68

68 Figure 3 1. Stratigraphic column for the middle late Paleocene Cerrejn formatio n and the stratigraphic horizon from which all known fossils of Cerrejonemys wayuunaiki were recovered. Stratigraphic column modified from Bayona et al. (2004).

PAGE 69

69 Figure 3 2. UF/IGM 33, Cerrejonemys wayuunaiki holoytpe. Skull, in A B dorsal and C D ventral views. Abbreviations: bo bassiocipital; bs basisphenoid; cm condylus mandibularis; co condylus occipitalis; cpt cavum pterygoidei; cs crista supraoccipitalis; ct cavum tympani; ex exoccipital; fon foramen or bito -nasale; fpc fossa precolumelaris; fpp foramen palatinum posterius; fr frontal; fsm foramen supramaxillare; fst, foramen stapedio temporale; ips interparietal scale; ju jugal; mx maxilla; op opisthotic; pa parietal; pal palatine; pf prefront al; po postorbital; pp, processus paraoccipitalis; pr prootic; pt pterygoid; ptp processus trochelaris pterygoidei; q quadrate; qj quadratojugal; so supraoccipital; sq squamosal.

PAGE 70

70 Figure 3 3. UF/IGM 33, Cerrejonemy s wayuunaiki holoytpe. Lower jaw, in A B dorsal and C D ventral views. Abbreviations: am area articularis mandibularis; an angular; art articular; cor coronoid; den, dentary; fmk fossa meckelii; lar labial ridge; lir lingual ridge; pra prearticular; prt processus retroarticularis; scm sulcus cartilaginis meckelii; sur surangular; vri ventral ridge.

PAGE 71

71 Figure 3 4 UF/IGM 33, Cerrejonemys wayuunaiki holoytpe. A B, carapace in dorsal view. C D plastron in ventral view. Right coracoid in E, dorsal and F ventral views. Sixth cervical vertebra in G ventral, H left lateral, and I posterior views. Seventh cervical vertebra in J posterior and K left lateral views. L, seven cervical vertebra of Podocnemis expansa AMNH 62947, in left lateral view. Pelvis of UF/IGM 33 in M left lateral and N right lateral views. Abbreviations: abd abdominal scale; cos costal bone; fem femoral scale; hyo hyoplastron; hyp hypoplastron; mar marginal scale; mes mesoplast ron; ne neural bone; nu nuchal bone; per peripheral bone; pec pectoral scale; ple pleural scale; ver vertebral scale.

PAGE 72

72 Figure 3 5. Strict consensus cladogram showing the phylogenetic relationships between podocnemidinurans turtles ; Unambiguous synapomorphies supporting each node are as follows (change is from 0 to 1 for binary characters; state indicated for multistate character): A (Pelomedusoides), 1(1), 2(1), 3(1), 45(1); B (Podocnemididae), 27(2); C (Erymnochelyinae), 6 (1),7(2); D (Shweboemydini), 29(1); E ( Erymnochelyini), 20(4); F (Podocnemidinae), 11(1), 41(1), 44(1). Extinct taxa indicated with a dagger superscript. Bootstrapping percentages (upper numbers) was run using 100 branch and bound replicates. Bremer decay indices (lower numbers) were obtained using TreeRot version 3 ( Sorenson and Franzosa, 2007).

PAGE 73

73 Figure 3 6. Left condylus mandibularis of quadrate in ventral view for A Erymnochelys madagascarensis YM 15398. B, Peltoc ephalus dumerilianus NFWFL 336. C Podocnemis unifilis AMNH 58195, and D Podocnemis bassleri AMNH 1622. A complete skull of Podocnemis expansa AMNH 97124, on the left for reference. Abbreviations: cm condylus mandibularis; cpt cavum pterygoidei; q quadrate.

PAGE 74

74 Figure 3 7. Map showing the distribution of modern (grey shading) and extinct podocnemidids. Hexagons for Late Cretaceous; stars for Paleogene; and black circles for Neogene records. Template obtained and po steriorly modified from Weinelt (1998), Ocean Drilling Stratigraphy Network through its Plate Tectonic Reconstruction Servic e. Available at www.odsn.de/odsn/services/paleomap/pale omap.html#Form Accessed January 29, 2009

PAGE 75

75 Table 3 1. Summary of the known fossil record of South American podocnemidinuran turtles. Fossil taxa Locality Age Material documented Sources Brasilemys josai Cear state, Brazil Aptian Albia n limit almost complete skull, carapace, two hyoid bones, left lower jaw, axis and third cervical vertebra Lapparent de Broin (2000) Portezueloemys patagonica Neuqun province, Argentina Late Turonian Early Coniacian partially preserved skull, carapace and plastron De la Fuente (2003) Bauremys elegans South Central, Bauru Group, Brazil Turonian Maastrichtian several skulls, lower jaws, shells, partial coracoid and cervical vertebra Suarez (1969), Kischlat (1994), Frana and Langer (2006) Bauremy s brasiliensis partial plastron Staesche (1937), Kischlat (1994) Roxochelys harrisi fragmentary carapace and plastron Pacheco (1913), Price (1953), Broin (1991) Cambaremys langertoni Minas Gerais, Brazil Maastrichtian partial carapace and plastron, coracoids, scapula, pelvis girdles and limb bones. Frana and Langer (2005) aff. Roxochelys vilavilensis Tiupampa Basin, Bolivia early Paleocene several skulls, lower jaws, shells, coracoid and cervical vertebra Broin (1991) Cerrejonemys wayuunaiki Cerr ejon Coal Mine, Colombia middle late Paleocene skull, lower jaw, partial carapace and plastron, right oracoid, pelvis girdle, two cervical vertebra this study Podocnemis pritchardi La Venta Fauna, Colombia middle Miocene nearly complete shell Wood (1997) Podocnemis medemi nearly complete plastron and partial carapace Podocnemis cf. expansa partially preserved cranium Bairdemys hartsteini Puerto Rico middle Miocene one skull almost complete Gaffney and Wood (2002)

PAGE 76

76 Table 3 1 Continued. Bairdemys venezuelensis Urumaco Fauna, Venezuela late Miocene several skulls and shells Wood and Diaz de Gamero, (1971) Gaffney and Wood (2002), Sanchez -Villagra and Winkler, (2006), Gaffney et al. (2008) Bairdemys sanchezi Urumaco, Fauna, Venezuela Late Miocene skull, lower jaws, anterior plastral fragment Gaffney et al. (2008) Bairdemys winklerae Urumaco, Fauna, Venezuela Late Miocene several skull, lower jaw Gaffney et al. (2008) Stupendemys geographic us Urumaco Fauna, Venezuela. late Miocene shell, humerus, femur, scapula, two cervical vertebrate Wood (1976) Stupendemys souzai Rio Acre, Peru Brazil late Miocene early Pliocene one costal bone, nuchal, right humerus, xiphiplastron, pelvic girdle and four cervical vertebra Lapparent de Broin et al. (1993), Gaffney et al. (1998), Bocquentin and Melo, (2006). Podocnemis bassleri Contamana Group, Peru late Miocene early Pliocene complete skull Williams (1956) Podocnemis negrii Acre state, Brazil lat e Miocene early Pliocene partial carapace and plastron, fragmentary pelvis girdle Carvalho et al. (2002) Fossi l taxa Locality Age Material documented Sources

PAGE 77

77 Table 3 1 Continued. Podocnemis expansa, P. erythrocephala, P. lewyana, P. sextuberculata, P unifilis, P. vogli Principal fluvial and lake systems of Northern South America recent complete skeleton Wagler (1830), Bonin et al. (2006) Peltocephalus dumerilianus Orinoco and Amazon Basins, Northern South America recent complete skeleton Schweigge r (1812), Bonin et al. (2006) Fossil taxa Locality Age Material documented Sources

PAGE 78

78 Table 3 2. Measurements for UF/IGM 33, holotype of Cerrejonemys wayuunaki in centimeters. Estimated Length for carapace and plastron are based on comparison to closely related forms (e.g., Podocnemis spp.). Measure UF/ IMG 33 Skull Maximum length. Indicated as I in (Gaffney et al., 2006. fig. 315) 16.7 Maximum width. Indicated as B in (Gaffney et al., 2006. fig. 315) 10.5 Lower jaw Maximum length. Indicated as B in (Gaffney et al., 2006. fig. 316) 11 Maximum widt h measured from the most lateral margins of the articular 8.5 Sixth cervical Maximum length in lateral view 3.5 Maximum width in dorsal view 1.9 Maximum high in posterior view Seventh cervical Maximum length in lateral view 5.5 Maximum width in dorsal view 3.9 Maximum high in posterior view 2 Coracoid Maximum length in dorsal view 10.2 Maximum width in dorsal view 2.5 Carapace Length 40.2 Length estimated for complete carapace 100 Width 50.2 Width estimated for complete carap ace 54 Thickness average of carapace measured in neurals, costals and peripherals 3 Plastron Length 32 Length estimated for complete plastron 80 Width 45 Width estimated for complete plastron 50 Thickness average of plastron 2.6

PAGE 79

79 CHAPTER 4 EARLY TO MIDDLE MIOC ENE TURTLES FROM PAN AMA; SYSTEMATICS AND PALEOBIOGEOGRAPHICAL IMPLICATIONS Introduction North and Central American turtles are dominated today by cryptodires or hidden -necked turtles, whereas pleurodires or side necked turtles are diver se in South America and cryptodires less so. This modern distribution of turtles in the New World is the result of a complex and poorly documented biogeographical history resulting in endemism, diversification, interchange and mixing of lineages from diffe rent geographic sources, all these events potentially influenced by geologic history. The most recent geological event with profound implications in the dispersal and interchange of biota between North Central America and South America, also causing the is olation between Caribbean and Pacific marine faunas was the formation of the Panama Isthmus by 3 Ma (Webb, 1985; Coates and Obando, 1996; Coates et al., 2004; Herrera et al., 2008). For turtles, the emergence of the Panama Isthmus allowed the arrival from North to South America of some representatives of cryptodires, including members of the families Kinosternidae, Chelydridae, Emydidae, and Geoemydidae (Pritchard, 1984), for geoemydids at least three different invasive events into South America are recogni zed after of the emergence of the isthmus, events suggested based on molecular data (Le and Mccord, 2008). By contrast, the arrival of members of the families Trionychidae and Testudinidae (including giant tortoises) has been suggested took place from the Southeastern portion of North America (particularly Florida), through the Antilles, finally reaching the Northernmost portion of South America (Pritchard, 1984; Head et al., 2006). Previous to the emergence of the Panama Isthmus, by the Early Miocene (ar ound 20 Ma), North Central America were separated from South America by the Culebra strait and Atrato

PAGE 80

80 seaway, with two long and narrow islands between them (Kirby et al., 2008:fig 11A). During the beginning of the Middle Miocene (around 15 Ma) the Culebra Cut was closed, giving more East continuity to the Central America Peninsula, however the Atrato seaway remained separating the peninsula from South America (Kirby et al., 2008:fig 11B). This condition of opening and closing of narrow straits inside the ea sternmost portion of the Central America Peninsula persisted until the final closer and formation of Panama isthmus by the Early Pliocene (3Ma) (Kirby et al., 2008). The faunas that colonized and inhabited the Early to Middle Miocene Central America Penin sula and the temporary islands system developed during the opening of the narrow straits are poorly documented, as well as their continental affinity, particularly for lower vertebrates (reptiles, amphibians and fish). In the particular case of turtles, th e fossil record is very rare and highly fragmentary for the Central American region and can be briefly summarized in the following previously described or documented fossils: (1) the Oligocene -Miocene Geochelone costarricensis from Costa Rica (Segura, 1944; Coto and Acua, 1986), (2) the Late Miocene carapace pieces of Rhinoclemmys sp. and shells of Geochelone sp. from Honduras (Web and Perrigo, 1984); (3) Pliocene isolated costal bones of a Trionychid from Costa Rica (Laurito et al., 2005), (4) the Late Pl eistocene Rhinoclemmys nicoyana from Costa Rica (Acua and Mora, 1996) and (5) the undescribed Miocene fossil turtles from the outcrops bordering the Panama Canal (Withmore and Stewart, 1965; MacFadden, 2006). Recently, new fieldwork campaigns lead by geologists and paleontologists from the Florida Museum of Natural History University of Florida, the Smithsonian Tropical Research Institute and the Panama Canal Authority allowed to relocate the Withmore Stewart original sites, as well as the discovery of ne w fossil sites with most complete and better preserved fossils,

PAGE 81

81 including new mammals taxa, crocodiles, snakes, turtles, fishes and plant remains; all this resulting in a better understanding of the paleontology of the Panama Canal basin. A first paper in this new stage of the paleontology from the Panama Canal described formally the fossil mammals collected by Withmore and Stewart during the 1960s, as well as new fossil taxa discovered in most recent years; all this mammalian fauna shows affinity with Nort h America Miocene faunas (MacFadden, 2006). In this paper, we describe the fossil turtles reported by Withmore and Stewart (1965) and MacFadden (2006), together with new fossils collected during our last three years of fieldwork in five different sites at the Galliard -Culebra cuts and Centenario bridge, in the Panama Canal Basin (Fig 4 1). Three of the sites correspond to outcrops of the Culebra formation, which is Early Miocene (19.8319.81 Ma) in age Culebra formation and the other two are Early Miocene to Middle Miocene (19 14 Ma) Cucaracha formation (sensu Kirby et al., 2008). Besides of the systematic paleontology of these turtles, we discuss their paleobiogeographical and paleoecological implications. Systematic Paleontology TESTUDINES Linnaeus, 1758 CRYPTODIRA Cope, 1868 TESTUDINOIDEA Batsch, 1788 (fide Baur, 1893) GEOEMYDIDAE Theobald 1868 RHINOCLEMMYS Fitzinger, 1835 RHINOCLEMMYS PANAMAENSIS sp. nov. FIG 4 2 A E Etymology: From Panama (The country from where the fossil was found).

PAGE 82

82 Holotype: Univers ity of Florida, Florida Museum of Natural History, Vertebrate Paleontology collections UF 237887, fairly complete articulated shell, missing most of the posterior portion of the carapace, and both xiphiplastra. Horizon, Locality and Age: Recovered from a s andstone layer, belonging to the upper part of the Early to Middle Miocene, Cucaracha Formation, stratigraphic section number (8) of Kirby et al (2008: fig 6). Site located just under the West side of the Centenario Bridge, at the Panama Canal Basin (9 1 47 N, 79 38 12W). Diagnosis: Differs from the other species of Rhinoclemmys and other geoemydid genera by gular and humeral scales much narrower dorsally, previous to the transition to the visceral surface of the anterior plastral lobe, condition th at is progressively increased medially. Discussion: It is recognized as a member of Testudinoidea Superfamily by lacking inframarginal scales (Character 38(3), Claude and Tong, 2004) and the presence of well developed axillary and inguinal buttresses reach ing the costal bones (Claude and Tong, 2004). As a member of the Geoemydidae Family by the presence of inguinal and axillary musk duct foramina (Character 36, Claude and Tong, 2004), at least recognizable on the right side; and as a member of Rhinoclemmys genus following Hutchison (2006) and Carr (1991) by: (1) a transition from dorsal margin of gular scale to the visceral surface almost smooth; (2) moderated to smooth transition between scaled and unscaled parts of peripherals on ventral view; (3) inguinal buttress contacts only costal 5; (4) dorsal parts of gular scales progressively narrowing medially; (5) cervical scale very narrow; and (6) presence of a very small axillary scale; (7) nuchal bone with strong posteromedial concavity. Comparative Descripti on: The type and only specimen of Rhinoclemmys panamaensis sp. nov is represented by an articulated shell (30 cm length x 25 cm width, maximum values),

PAGE 83

83 which is oval in shape with anterior margins of carapace and plastron straight, and smooth. Dorsal surfa ce slightly micro -sculptured in some areas, lacking lateral keels as in other Rhinoclemmys species and Echmatemys mentioned as the common condition in these two New World genera (Claude and Tong, 2004). The shell is slightly affected by crushing and cracki ng, particularly at the central portion of the plastron. The nuchal bone is wider than long, with a shallow embayment on its anterior margin having a strong posteromedial concavity on the ventral surface, in all these characteristics resembles other Rhin oclemmys species, the UF237892 nuchal bone referred in this paper and Rhinoclemmys sp. UF 46671 referred in Webb and Perrigo (1984). Neural bones 1 through 4 are preserved, neural 1 and 3 are rectangular, and neural 2 and 4 almost squared in shape with ver y short antero -lateral sides, in this particular aspect differs from the other species of Rhinoclemmys Brigderemys pusilla Hutchison 2006 and Echmatemys septaria Hay 1906 for which neurals 2 through 4 are hexagonals, with short sides located postero-later ally, but resembles the primitive condition seen in Palaeoemys Costal 1 through 5 are preserved in both sides of the carapace, right costal 5 only preserved in its most posterolateral margin. Peripheral 1 through 6 are preserved on both sides of the carap ace, plus peripheral 7 on the right side, additionally on the right side peripheral 3 through 7 and on the left side peripheral 4 through 6 are widely exposed in ventral view of the shell due to crushing. On the ventral surface the transition or slope betw een scaled and unscaled parts of peripherals 1 and 2 is moderate almost smooth, as in the habitual condition in Rhinoclemmys (Hutchison, 2006). The cervical scale is trapezoidal, narrower than wide, in contrast to Brigderemys pusilla which has a wider cer vical; the condition in primitive geomydids such as Palaeoemys and Echmatemys is a small squared cervical. Vertebral scale 1 through 3 are slightly longer than

PAGE 84

84 wide, being the vertebral 2 the longest of these three, condition that seems to be influenced by crushing; relatively longer than wide vertebral scales is the primitive condition seen in Palaeoemys and Echmatemys ; as long as wide or wider than long for most of derived geoemydids. Pleural scale 1 and 2 are completely preserved, as well as the most ant erior part of pleural 3, in both sides of the carapace. Claude and Tong, (2006) suggested a contact between pleural 2 and marginal 4 as a derived condition for geoemydids, however after examined a considerable number of specimens of Rhinoclemmys (see Appen dix 4 1) we conclude that this condition is highly variable and in most of the cases marginal 4 only contacts pleural 1; condition shared by R. panamaensis and the primitive geoemydids Palaeoemys and Bridgeremys pusilla. The marginal scales 1 through 7 are recognizable on the left side of the carapace, whereas on the right side only marginal 1 through 3 are clearly delimited. The plastron of Rhinoclemmys panamaensis resembles in shape those of other species of Rhinoclemmys and other geoemydids, being almos t rectangular. The epiplastra meet medially in a long contact, both having an almost parallel anterolateral and posteromedial margins, and a slightly concave anteromedial margin also present in R. annulata UF2532 and R. pulcherrima figured in Hutchison (2006:fig 8B). The anterolateral margin starts with a rounded and moderate step, which is an expression of the anterolateral contact between the gular and the humeral scale; similar condition is common in other species of Rhinoclemmys as for example: R. annulata and R. pulcherrima by contrast the rest of geoemydids and R. pulcerrima incisa UMNH 11393 have a stronger and straight step, except in the primitive geoemydid Palaeoemys which lacks the anterolateral step and has a continuous edge. The entoplastron has a bell -shaped, slightly longer than wide, with straight anterior sides and convex posterior sides; as in most of geoemydids particularly for R. nasuta mentioned in Carr (1991). The medial contact between both hyoplastral

PAGE 85

85 is shorter than the contact bet ween both hyoplastral and as long as in Palaeoemys and Echmatemys septaria, shorter in Bridgeremys pusilla and the other geoemydids. The axillary and inguinal buttresses are well developed, slightly crushed, and the musk duct foramina is recognizable for b oth buttresses on the left side of the plastron. Two pairs of musk duct foramina, in the axillary and inguinal buttresses are considered synopomorphic for all geoemydids turtles (Hirayama, 1984; Le and Mccord, 2008). Rhinoclemmys panamaensis has a gular t riangular in shape with its posterior tip overriding a very small area on the anterior portion of the entoplastron, condition highly variable within geoemydids; from a gular restricted only to epiplastral, through a gular overriding considerably the anteri or portion of the entoplastron. On the dorsal surface of the epiplastral (Fig 4 2 E F), the transition from the most posterolateral margin of the gular to the visceral surface is marked by an almost smooth step, as well as the medial contact between gulars is pretty short and differs from the all others geoemydids for which this contact is longer. The humero-pectoral sulcus crosses the entoplastron on its posterior portion and laterally is restricted to the hyoplastral as in most of geomydids, except in R. diademata, which has a humero pectoral sulcus that not only crosses the entoplastron, but also the most posteriomedial portion of the epiplastral (Carr, 1991). The pectoro abdominal sulcus is anterior to the hyoplastron-hypoplastron suture as in all other geoemydids, being medially very close to the suture in Bridgeremys pusilla. A very small axillary scale is present at the most anterolateral corner of both hyoplastral enclosed by the pectoral scale, as in all other species of Rhinoclemmys in contrast to Bridgeremys, Echmatemys and Palaeoemys all three lacking the small axillary scale. RHINOCLEMMYS cf. AREOLATA Referred Material: UF242075, most anterior portion of the nuchal bone.

PAGE 86

86 Locality and Age: same locality and age as for Rhinoclemmys panamaensis D escription and Remarks: UF242075 (Fig 4 3 A B), is the most anterior portion of a small nuchal bone, which has a small medial notch on the anterior margin. On the dorsal surface the cervical scale is very narrow and long, and the sulcus between vertebral s cale 1 and the marginal set 1 ends medially in an acute tip. On the ventral surface the transition from the posterior margin of marginal set 1 to the visceral surface is marked by a strong step. In all characteristics, UF242075 resembles juvenile specimens of Rhinoclemmys areolata Gray 1825, as for example UF(H)54199 (Fig 4 3 C D). RHINOCLEMMYS sp. Referred Material: USNM PAL171020A, left xiphiplastron; USNM PAL171020B, right peripheral 5; USNM PAL171020C, right xiphiplastron; USNM PAL171021, right costal 1; UF237892, nuchal bone; UF223583, right peripheral 1; UF237896, right epiplastron; UF237881, neural 3 or 5?; UF237894, neural bone 5; UF237890, left hypoplastron; UF237891, left hyoplastron, UF237888, left costal 1; UF237889, left costal 1; UF237893, rig ht costal 2 or 4?, UF237895, left costal 3 or 5; UF242092, right peripheral 9. Locality and Age: Same locality and age as for Rhinoclemmys panamaensis for all UF specimens referred. USNM PAL171020 (A -C) and USNM PAL171021 collected by Withmore and Stewart at the Culebra Reach, Station 1998 + 00, 600 feet W of center line of Panama Canal, Cucaracha Formation as taken from labels, Early to Middle Miocene in age according to Kirby et al (2008). Description and Remarks: USNM PAL171020A (Fig 4 3 E F) is a left xiphiplastron with rounded posterior tip and without evidence of kinesis with the hypoplastron; the femoroanal sulcus is strongly marked on its ventral surface and the anal scale overrides almost 60% of

PAGE 87

87 the bone, with indication of a long medial contact w ith the right anal scale. On the dorsal surface, the transition from the femoral and anal scale to the visceral surface is marked by a shallow groove almost parallel to the lateral margin of the xiphiplastron. In all previous characteristics USNM PAL171020A resembles the xiphiplastron of Rhinoclemmys and all other geoemydids. USNM PAL171020C (Fig 4 3 G H) is a right xiphiplastron, which resembles USNM PAL171020A in all its characteristics, except by the posterior tip of the xiphiplastron, which has a shal low wide embayment. USNM PAL171021 (Fig 4 3 I J), UF237888, and UF237889 are costal 1 bones, the former is right and the other two left, and all represent different individuals of similar size and at least two times smaller than the holotype of R. panamae nsis On the dorsal surface all share a strongly marked sulcus between the first and the second vertebral scales at the most posteromedial portion, and between those vertebrals and the first pleural very close to the medial edge of the costal. On the ventr al surface, the axillary buttress scar is well marked on the most lateral portion of the costal, and the projected head for the attachment with the thoracic rib is also pretty well developed medially. In all characteristics previously described, all these costal 1 resemble Rhinoclemmys and all other geoemydids. UF237892 (Fig 4 3 K L) is a complete nuchal bone, with a very narrow cervical scale, which is also notorious by a major thickness of the bone, which is associated with the development of the medial keel on the dorsal surface of the carapace. The sulcus between the marginal 1 set and the first vertebral scale is visible at the most anterior portion of the bone, and a very small corner of the pleural 1 is in contact with the vertebral 1 on the most la teral part of the bone. On the ventral surface, the beginning of the visceral surface is represented by a deep posteromedial concavity. UF237892 shares with UF 46671 (Fig 4 3 M) from the Late Miocene

PAGE 88

88 of Honduras (Web and Perrigo, 1984) and all modern speci es of Rhinoclemmys all characteristics described above, differing from R. panamaensis by a cervical scale much narrower and a well developed medial keel. UF237896 is a right epiplastron, corresponding to an individual at least two times smaller than the h olotype of R. panamaensis On the ventral surface, the gular -humeral sulcus crossing the posteromedial margin, indicating that the most posterior portion of the gular overrides the entoplastron as in Rhinoclemmys panamaensis and other geoemydids. On the do rsal surface, the step indicating the transition from the posteromedial edge of the gular and humeral scales to the visceral surface is slightly stronger than in R. panamaensis but similar than in other Rhinoclemmys species. Also on this same surface, the medial contact between gulars is as wide as the lateral contact with the humerals as in other species of Rhinoclemmys except R. panamaensis (see the condition described above for this species). UF223583 is a right peripheral 1. The posteromedial margin of the marginal scale 2 only reach the central portion of the peripheral, this is the most common condition for Rhinoclemmys and for most of geoemydids. In contrast to emydids for which the most common condition is a posteromedial margin of marginal 2 reac hing the most lateral portion of the nuchal or the sutural contact between the nuchal and peripheral 1. Other specimens attributed to Rhinoclemmys sp. are UF237881 (Fig 4 3 N) and UF237894, corresponding to two neural bones, probably neural 3 or 5? the f ormer and neural 5 the second based on the presence of the sulcus between vertebral scales, which has a very small notch anteriorly projected and located at the midline of the bone. Both neurals are hexagonals in shape with short posterolateral sides as i n most of the geoemydids, excluding R. panamaensis Palaeoemys and the second neural of Bridgeremys pusilla (MPM 3425) figured in Hutchison

PAGE 89

89 (2006). Finally, UF237893 and UF237895, consists of a right costal 2 or 4? and a left costal 3 or 5?. UF237893 is a right costal 2 or 4?, base on the presence on the dorsal surface of the sulcus between pleural scales at the posterolateral portion of the costal; the sulcus between the pleurals and the vertebral at the medial portion; and a rectangular shape with anteri or and posterior edges almost straight, and anteromedial short side for the contact with neural 2 or 3, differing from costal 6 and 8 which although also have the same scales sulcus pattern, their anterior margin is convex and the posterior margin concave, being medially shorter, and for the case of costal 6 having short anteromedial and posteromedial sides, and costal 8 commonly lacking short anteromedial or posteromedial sides. Similar comparative analysis between costals allows to defined UF237895 as a l eft costal 3 or 5?, with the particularity that on its dorsal surface the scales sulcus pattern is pleural scale in medial contact with two vertebral scales, and these vertebrals contacting one each other transversally. In all characteristics described, UF 237893 and UF237895 resemble the costals of Rhinoclemmys TESTUDINIDAE Batsch 1788 cf. GEOCHELONE Referred Material: USNM V23180, right coracoid; USNM PAL171017, right ulna; USNM PAL171020, right xiphiplastron; USNM V23146, claw. Locality and Age: All USN M specimens were collected by Withmore and Stewart at the Culebra Reach, Station 1998 + 00, 600 feet W of center line of Panama Canal, Cucaracha Formation as taken from labels, Early to Middle Miocene in age according to Kirby et al (2008). Description and Remarks: USNM V23180 (Fig 4 4 A B) is a large, and blade expanded right coracoid (19 cm length x 14 cm wide, maximum values), missing a portion of the posterolateral margin and most of the dorsal and ventral margin of the proximal articular area.

PAGE 90

90 The medi al edge is thinner than the lateral, and the blade is relatively short, with slightly rounded and flared distal margin. At the proximal area, the most central part of the glenoid surface is preserved, being slightly concave for the articulation with the hu merus. The sutural contact with the scapula is indeterminate due to highly eroded proximal area. USNM PAL171017 (Fig 4 4 E F), is a complete right ulna (l3 cm length x 4 cm width, maximum values), proximally dominated by the olecranon, which is rounded an d well developed. The proximal articular surface is curved and the bicipital tubercle is poorly recognized. Distally, the ulna ends in a slightly convex surface for articulation with the humerus, and the most dorsolateral margin is missing. In all characteristics mentioned above, USNM PAL171017 and USNM V23180 resemble the ulna and coracoid of testudinids, particularly those of the giant tortoises, as for example Chelonoidis elephantopus USMN 59867 (Fig 4 4 C D, G H). Although, at this point any further sys tematic resolution is possible, we confer these fossils to Geochelone and suggest their probably close relation with the South American subgenus Chelonoidis Additionally, an isolated claw USNM V23146 (Fig 4 4 I) and a right xiphiplastron USNM PAL171020 ( Fig 4 4 J K) are also included as testudinids and conferred to Geochelone genus. The xiphiplastron is relatively small (5 cm length x 4 cm width, maximum values), and on its ventral surface the most characteristic feature is the very small anal scale, which is trapezoidal in shape, much shorter medially. On the dorsal surface the transition from the margins of the femoral and anal scales to the visceral surface is strongly marked by a deep concavity. In all dorsal and ventral features, USNM PAL171020 resembles the xiphiplastron of species of Geochelone, particularly those of the Chelonoidis subgenus. CRYPTODIRA Cope, 1868

PAGE 91

91 KINOSTERNIDAE Baur, 1893 STAUROTYPUS Wagler 1830 STAUROTYPUS MOSCHUS sp. nov. Etymology: from Latin moschus for musk or musky in ref erence to the well developed anterior musk duct groove present in this form. Holotype: UF242076, left peripheral 2. Diagnosis: Presence of a well -developed and relatively deeply incised anterior musk duct groove that runs closely adjacent to the visceral s cale margin for Marginals 1 & 2; marginal scales 1 & 2 relatively narrow to the height of the peripheral in dorsal aspect; costiform process with only slight contact to the anteromost portion of P2 at the P1 P2 suture. Locality and horizon: same locality a nd age as for Rhinoclemmys panamaensis. Description and Comparisons: UF242076 (Fig 4 5A B) is that from an adult individual, comparable in size to extant members of the genus Staurotypus examined here with a carapace length between 2427 cm aproximately. O n the dorsal surface, marginal scales 1 and 2 are moderately bulbous and narrow,with a strong notch at the sulcus between them along the lateral margin of the peripheral. The outer rim of the marginals is notably thick and squared, with a distinct lip alon g the dorsal edge. In modern Staurotypus the outer rim is usually much thinner and tapered. Whether or not these features can be used in the diagnosis of the new taxon awaits further discovery of specimens. The surface of the bone has a fine microvermiculation as present in kinosternids (Cadena et al., 2007: fig 2L), on the dorsal and ventral surfaces of the marginals and Costal 1. Arguably, the most important feature on the element is the presence of the deeply incised musk duct on the visceral surface. The presence of an anterior musk duct groove is a synapomorphy for Kinosternidae (Joyce, 2007; character 66) musk generally runs from P4 and

PAGE 92

92 terminates on P1 (Hutchison, 1991; pers. obs.). Staurotypus moschus is the most southern occurring staurotypine yet known. Its extant congeners are Staurotypus salvini from the western lowlands of Central America and Staurotypus triporcatus from the base of the Yucatan Peninsula. Interestingly, these 3 taxa each exhibit different degrees of development of the anterior musk duct groove, with S. moschus being the most deeply incised, S. salvini being only moderately to weakly so, and S. triporcatus very weakly developed. This character state has been coded as weakly incised for the entire genus in previous phylogeneti c analyses (Hutchison, 1991). Where a weakly incised musk duct groove could be interpreted as a primitive condition, we feel that temporally older S. moschus provides evidence that this character state is a reversal and therefore synapomorphy in the extant taxa S. salvini and S. triporcatus The distance of this groove from the visceral marginal sulcus shares a similar progressive relationship, with the groove being closest in S. moschus moderately close in S. salvini and farthest from the sulcus in S. tr iporcatus For these reasons, S. moschus is interpreted here as having its closest affinities with the salvini group. A small pit that would recieve the very end of the costiform process is just visible on the visceral face of UF242076, at the P1 P2 sutur e. At and just posterior to this pit, is a slight swelling of the element, also present in S. salvini and S. triporcatus However, this pit is more substantial in the latter 2 taxa, with the costiform process usually more intrusive on P2 (Hutchison, 1991, fig. 5; pers. obs). CRYPTODIRA Cope, 1868 TRIONYCHIDAE Baur, 1893 Gen. et sp. Indet.

PAGE 93

93 Referred Material: UF242088 (Fig 4 5 C), costal bone, UF242106 (Fig 4 5 D), costal bone, UF242108 (Fig 4 5 E), left epiplastron. Locality and Age: same locality and age as for Rhinoclemmys panamaensis for UF242088. UF242108 and UF242106 collected at Lirio Norte site (9 3' 17" N, 79 39' 33" W), West margin of the Panama Canal, upper member Culebra formation, Early Miocene in age according to Kirby et al (2008). Descripti on and Remarks: All them characterized by: the complete absence of scales on the dorsal surface, as well as the sculptured ornamented pattern consisting of ridges, pits and knobs very distinct of trionychids or soft shell turtles. At this point the materia l collected is not enough to suggest any further taxonomic assignation. PLEURODIRA Cope, 1864 PELOMEDUSOIDES Cope, 1868 PODOCNEMIDIDAE Gray, 1825 Gen. et sp. Indet. Referred Material: UF242276, left articulated hyoplastron, hypoplastron and mesoplastron; U F242170, left and right articulated xiphiplastral, plus right hypoplastron, neural 2 and costal 3 or 5?; UF242175, fairly complete left side of the anterior plastral lobe; UF242174, right xiphiplastron; UF242160; right costal 6; UF242150, left costal 2; UF 242165, right side of a pelvic girdle; UF242171, distal and proximal portions of a right humerus; UF242097, proximal portion of a left femur; UF242111, right peripheral 2; UF242158, left peripheral 8; UF242168, neural 3 or 5?. Locality and Age: UF242097, and UF242170 collected at the Culebra Norte site (9 3' 4" N, 79 38' 58" W), East margin of the Panama Canal, conglometaic sandstones, Culebra

PAGE 94

94 formation, Early Miocene according to Kirby et al., (2008). All other UF specimens were collected at the Lirio N orte site (9 3' 17" N, 79 39' 33" W), West margin of the Panama Canal, from conglomerate channels belonging to the Culebra formation. Description and Remarks: UF24217 (Fig 4 6 A B) consists of a left articulated hyoplastron, hypoplastron and mesoplastron which is almost circular in shape and laterally positioned between the hyoplastron and hypoplastron, as in all other panpleurodires (Character 85, Joyce, 2007). On the ventral surface the pectoroabdominal sulcus is anterior to the mesoplastron, which is the most common condition in podocnemidids. The abdomino-femoral sulcus ends laterally at the hypoplastral notch. UF242170 is the unique turtle specimen so far collected at Culebra formation for which carapaceal and plastral elements were found associated including neural 2 (Fig. 4 6, C) both xiphiplastra and the right hypoplastron (Fig 4 6, D E), costal 3 or 5? and at least ten undifferentiated fragments of costals. The ventral surface is highly altered making impossible to recognize any sulcus between s cales; being the U open shaped anal notch and the concave outline for the mesoplastron at the anterolateral margin of the hypoplastron the most notorious features from this view. On the dorsal surface, both xiphiplastra preserve the pubic and ischial scars indicating a strongly sutured pelvis to the plastron, which is the undisputable synapormophy for panpleurodires (Character 125; Joyce, 2007). The pubic scar is oval -shaped, oriented almost parallel to the medial margin of the xiphiplatron, and the ischia l scar is triangular in outline, ending in acute tip medially very close to the sutural contact between both xiphiplastra, as is common in podocnemidids. The neural 2, is hexagonal elongated in shape with short anterolateral sides and dorsal surface lackin g sulcus between vertebral scales; at the most

PAGE 95

95 anteroventral margin two lateral like -horn projections are present for articulation with the neural 1. UF242175 (Fig 4 6 F G) is a fairly complete left side of the anterior plastral lobe, highly fractured. The most notable features of this specimen are the entoplastron in diamond shape, the gular scale triangular in shape overriding only the epiplastron, the humero -pectoral sulcus crossing the posterior portion of the entoplastron, and thus being posterior t o the epiplastron -hyoplastron suture. Two large costal bones UF242160 and UF242150 were found associated, and we assume that both belong to the same individual. UF242160 (Fig 4 6 H) is a right costal 6 preserving a moderate curvature seen from transversa l view and short posteromedial side. On its dorsal surface the sulcus between pleurals is visible, as well as most medially the sulcus between the vertebral and the two pleurals, which is more medially projected at the posterior margin of the costal. UF342150 (Fig 4 6 I) consist of a left costal 2, with anterior and posterior margins almost parallel, short posterolateral side, and a well preserved sulcus between pleurals, and most medially the sulcus between the vertebral and the pleural scales. UF242165 (Fig 4 6 J) is the right side of a pelvic girdle, including the complete ilium, the pubis and the most distal portion of the ischium. The acetabulum capsule is almost oval in shape with the triple sutural contact between the ilium, pubis and ischium, as in most pleurodires, and particularly similar to the pelvic girdle of the Podocnemis as for example Podocnemis expansa AMNH 62947 (Fig 4 6 K) Two limb bones are UF242171 and UF242097. UF242171 (Fig 4 6 L) consists of a right humerus preserved by the most di stal and proximal portions, missing the central portion. Proximally, the lateral process is missing and the medial is large and posteroproximally

PAGE 96

96 projected, the hemisphere for articulation is rounded and equidimensional, lacking a shoulder, considered as a synapomorphy for pleurodires turtles (Gaffney, 1990); the distal portion has a capitellum rounded and slightly convex for the articulation with the ulna and radius, and the distint ectepicondylar foramen on the anterodistal margin, characteristic of the humerus of all testudines. UF242097 (Fig 4 6 L) is a left femur, preserved on its proximal aspect. In dorsal view, the acetabular articulation head is oval in shape slightly elongated and inclined clockwise in respect of the longitudinal axis of the bone. A lso in dorsal view the throcanters major and minor are almost at the same horizontal level and well laterally projected. In ventral view, the interthrocanteric fossa is very shallow with a long scar on its posterior edge. In all aspects this femur resemble s podocnemidids particularly the femur of the extant genus Podocnemis Other isolated carapaceal elements are: UF242111 (Fig 4 6 M) which is a right peripheral 2; UF242158 (Fig 4 6 N) corresponding to a large (10 cm length x 8 cm width, maximum values) per ipheral 8; and UF242168 which is a neural 3 or 5? based on the presence of the sulcus between vertebral scales on its dorsal surface. All these carapaceal elements resemble podocnemidids in their shape and scales sulcus pattern. Discussion The fossil turtl es described in the present study reveal an extraordinary story of past colonization and faunal meeting between North Central America and South America previous to the Panama Isthmus formation, with events that not only started earlier than previously hypo thesized for some families of turtles, but also representing fossil evidence for molecular hypotheses of divergence for others. Besides, by contrast to mammals from the same localities, which show main North American affinity (MacFadden, 2006), turtles support evidence of interchange or at least habitat sharing between North -Central and South America representatives during the Early Miocene. All these aspects are discussed in detail in the following paragraphs.

PAGE 97

97 Fauna and provinciality Two different fauna s of turtles are recognized, the former, from the upper segment of the Early Miocene Culebra formation and the second from the Early to Middle Miocene Cucaracha formation. Turtles from Culebra Formation are represented by cryptodires (trionychids) and pleu rodires (podocnemidids), inhabiting prodelta to delta environments as inferred for the upper segment of the Culebra formation by Kirby et al (2008); however considering that most of the fossils correspond generally to complete but disarticulated shell elem ents, particularly plastral elements, and very few non -shell elements, mostly of them found inside conglomeratic channels, indicating pre -burial transport by streams, and potentially including also more fluvial environments, in same way that modern represe ntatives of these two families do (Boni et al., 2006). Trionychids turtles in the New World are restricted today to temperate subtropical regions of North America, with a single genus Apalone (Bonin et al., 2006). The Early Miocene occurrence of trionychi ds from the Panama Canal basin represents the southernmost advance for this family in the New World, and together with the record from the Castillo Formation in North western Venezuela (Sanchez Villagra et al., 2004) they represent the earliest record in t ropical environments of Central America and South America. In contrast to trionychids, podocnemidids are one of the most representative South American living and fossil turtles, and their occurrence in Early Miocene rocks from the Panama Canal basin repres ent the most western occurrence for this family, as well as the potential earliest record for Podocnemis genus, based on not only in their morphological similarity mentioned in the description, but also by the large size of the shell which could has reache d 100 cm at midline, being slightly bigger than the largest size reported for modern Podocnemis This size estimation for the Panamanian podocnemidids is based on the size of some isolated elements as for example

PAGE 98

98 the UF242160 (Fig 4 6 H) and UF242150 (Fig 4 6 I) costals, and the peripheral UF242158 (Fig 4 6 N). Additionally, the presence of neural bones in the specimen UF242179 and a long anterior plastral lobe of the UF242175 specimen, exclude them to belong to Bairdemys which was a the Late Miocene Carib bean South American podocnemidid characterized by lacking neural bones and a shorter anterior plastral lobe (Scheyer et al., 2008; Gaffney et al., 2006:fig 275). The second fauna of turtles is represented exclusively by cryptodires, including trionychids as in Culebra formation, thus of kinosternids, geoemydids, and testudinids, all them inhabiting delta plain environments as inferred for the Cucaracha formation by Kirby et al (2008). In contrast to the turtles from Culebra formation, the turtles from Cuca racha formation are represented for most complete and articulated specimens, as for example the shell of Rhinoclemmys panamaensis indicating relatively little pre -burial tansport. Living geoemydid turtles in the New World are represented by a single genu s, Rhinoclemmys with at least nine species inhabiting freshwater and terrestrial tropical environments of Central and South America (Bonin et al., 2006). The arrival of Rhinoclemmys to South America is relatively recent and related to the Panama Isthmus fo rmation (Pritchard, 1984; Savage, 2002). However, base on molecular data, a first dispersal event from Central to South America, involving the species R. nasuta is suggested for the Early Miocene (Le and Mccord, 2008). The occurrence of Rhinoclemmys panamaensis Rhinoclemmys cf areolata and Rhinoclemmys sp from the Early to Middle Miocene in the Panama Canal basin, not only represent the earliest record of geoemydids in Central America, but also support the Le and Mccord hypothesis in terms of the presence of this genus very close to South America for that time. Thus these fossils give evidence of the very early diversification of this genus.

PAGE 99

99 Closed related to geoemydids turtles are the Testudinids (Tortoises) for which has been suggested that arrived to Ce ntral and South America by passive flotation from North America and the Antilles during the Miocene (Pritchard, 1984) favored by their adaptations for over water dispersal (Meylan and Sterrer, 2000). The fossil testudinids described here, together with the Oligocene -Miocene record of Geochelone costarricensis Segura 1944, constitute the earliest record of land tortoises in Central America. Thus, the ulna, coracoid and claw described here are slightly bigger in size in contrast to modern giant tortoises as f or example Chelonoidis (Geochelone) elephantopus USNM 59867 from Galapagos islands and Dipsochelys (Geochelone) gigantea USNM 222495 from Aldabra island. This indicates that gigantism in tropical testudinids was early developed in their evolutionary histor y, under ecological conditions of abundant mammalian fauna, potential predators such as crocodiles (reported but undescribed by Withmore and Stewart, 1965; MacFadden, 2006), as well as the other cryptodires turtles described in this study; all them probabl y competing for same or similar food sources and relatively small area within the westernmost portion of Central America Peninsula or the temporal large -island created between this peninsula and South America due to the formation of the Culebra strait, acc ording to the paleogeographic scenario proposed by Kirby et al. (2008). The last member of the Cucaracha formation turtle fauna corresponds to kinosternidids or mud and musk turtles. The occurrence of Early to Middle Miocene kinosternidids from the Panama Canal basin represent the earliest record of this family in Central America, as well as indicates a wider past geographical distribution for the Staurotypus geneus, and a very early advance of kinosternidids into tropical environments, very near to South America for that time. Summary and Conclusions Fossil turtles from the Early Miocene Culebra formation represent the first and the earliest evidence of faunal meeting between North America (trionychids, probably Apalone ) and South

PAGE 100

100 America (podocnemidids, probably Podocnemis ) turtles, previous to the Panama isthmus emergence, in contrast to mammals, which only show North American affinity (MacFadden, Pecari and other taxon from Culebra fm REFERENCE). For the Early to Middle Miocene Cucaracha formation, th e trionychids turtles persisted inhabiting the Central America Peninsula, together with the arrival of North -Central America representatives of geoemydids ( Rhinoclemmys ), kinosternidids, and testudinids. Also is notorious the absence of podocnemidids, indi cating probably their retreating to South America as a consequence of the arrival of North American cryptodires. The occurrence of giant tortoises in presence of large predators (crocodiles), competing herbivores mammals, as well as other groups of turtl es, in an scenario with low geographical isolation, is a strong evidence that the gigantism in testudinids can be gained previous to the biogeographical isolation, as was suggested for the Aldabra island tortoises (Arnold, 1979). Thus, although much older in time, the giant tortoises from Panama are an excellent potential source for the modern Galapagos islands, and they could have been arrived to these islands through of the warm Panama Current, which should have had origin during the Pliocene as a consequence of the Panama Isthmus formation. The use of the Panama Current for the arrival of the giant tortoises to Galapagos was initially speculated by Pritchard (1984). In spite of the high fragmentation of most of the fossil turtles described in the present study, they preserved enough diagnostic features as to be assigned to family, subfamily and some cases genus and new species level, as for example Rhinoclemmys panamaensis Moreover, representing the earliest record for geoemydids, kinosternids, testudini ds, and trionychids in Central America, as well as expanding the biogeographical range for podocnemidids, and the

PAGE 101

101 subfamily Staurotypinae. Future paleontological fieldwork should be done in order to collected new and better preserve specimens, which potent ially could be included in phylogenetic analysis.

PAGE 102

102 Figure 4 1. Map of the Panama Canal, starts show the Early to Middle Miocene fossil localities from Culebra and Cucaracha formations, from which fossil turtles wer e collected.

PAGE 103

103 Figure 4 2. Rhinoclemys panamaensis Holotype UF 237887. A B. Carapace in dorsal view. C D plastron and carapace in ventral view. E F. Epipastra in dorsal view, showing gular and humeral scales muc h narrower dorsally, previous to the transition to the visceral surface of the anterior plastral lobe, diagnostic character of the species. Abbreviations: abd, abdominal; ce, cervical; cos costal; ent, entoplastron; epi, epiplastron; fem, femoral; gul, gu lar; hum humeral; hyo hyoplastron; hyp, hypoplastron; intg intergular; ma, marginal; mes, mesoplastron; ne neural; pec pectoral scale; pe peripheral; pl, pleural; ve, vertebral, vis, visceral surface. vis

PAGE 104

104 Figure 4 3. Rhinoclemy s cf. Areolata UF242075 most anterior portion of the nuchal A. Dorsal view, B. Ventral view. Rhinoclemmys areolata UF(H)54199 complete nuchal. C. Dorsal view, D. Ventral view. Rhinoclemmys sp. USMN PAL171020A, left xiphiplastron. E. Ventral view, F. Dorsal view. USMN PAL171020C, right xiphiplastron. G. Dorsal view, H. Ventral view. USNM PAL 171021, right costal 1. I. Ventral view, J. Dorsal view. UF237892, nuchal bone. K. Dorsal view, L. Ventral view. UF46671, nuchal bone referred in Web and Perrigo (1984), from Late Miocene of Honduras. M. Ventral view. UF237881, neural 3? or 5?. N. Ventral view. Scale bar on the left applies for A -F and on the right for G N.

PAGE 105

105 Figure 4 4. Testudinids Cf. Geochelone. USNM V23180, right cor acoid. A.Ventral view, B. Dorsal view. USNM PAL 171017, right ulna. E. Dorsal view, F. Ventral view. USNM V23146, claw. I. Lateral view. USNM PAL 171020, right xiphiplastron. J. Ventral view, K. Dorsal view. Chelonoidis elephantopus USMN 59867, right corac oid. C. Ventral view, D. Dorsal view. USMN 59867, right ulna. G. Dorsal view, H. Ventral view.

PAGE 106

10 6 Figure 4 5. Staurotypus moschus UF242076, left peripheral 2. A. Dorsal view, B. Ventral view showing the marked musk duck g roove. Trionychids Gen, et sp. Indet. UF242088, costal bone. C. Dorsal view. UF242106, costal bone. D. Dorsal view. UF212108, left epiplastron, E. Ventral view.

PAGE 107

107 Figure 4 6. Podocnemidids Gen. et sp. Indet. U F242176, left articulated hyoplastron, hypoplastron and mesoplastron. A B. Ventral view. UF242170, neural 2. C. Dorsal view; left and right xiphiplastra, together with the right hypoplastron. D E. Dorsal view. UF242175,left side anterior plastral lobe. F G Ventral view. UF242160, right costal 6. H. Dorsal view. UF342150, left costal 2. I. Dorsal view. UF242165, right side of a pelvic girdle. J. Lateral view. Podocnemis expansa AMNH 62942, right side of the pelvic girdle. K. Lateral view. Abbreviations : abd abdominal; ent, entoplastron; epi, epiplastron; fem, femoral; gu, gular; hum humeral; hyo hyoplastron; hyp hypoplastron; intg intergular; isch sc ischium scar; mes, mesoplastron; pec pectoral scale; pub sc, pubic scar.

PAGE 108

108 CHAPTER 5 CONCLUSIONS In the previous three chapters, I described new fossil turtles from tropical South America (Colombia and Panama), pointed out their evolutionary and paleobiogeographical implications. I dedicate the following paragraphs to summarize the most relevant about these fossil turtles, as well as potential future research topics to be developed using these fossils. Notoemys zapatocaensis Early Cretaceous of Colombia Notoemys zapatocaensis represents the youngest representative of basal pleurodires, a group of turtles that inhabited coastal to estuarine environments from the tropics to temperate zones from the Late Jurassic trhoug the Early Cretaceous. The best preserved and most complete specimen of N. zapatocaensis is the allotype (MGJRG IPN 15 EAC 150620061) described in Chapter 1, thus this specimen is the only representative of the genus so far recovered for which the plastron is completely preserved, allowing to code and to involve important characters in the phylogenetic analysis performed and it which showed as a r esult that N. zapatocaensis and N. laticentralis are sister taxa. Additionally, the arrangement of bones as well as scales of the anterior plastral lobe for N. zapatocaensis allotype shows a transitional step between stem testudines and the crown pleurodir es and cryptodires turtles, showing how the entoplastron was progressively being incorporated inside of the plastron until become totally enclosed by both epiplastral anteriorly and both hyoplastral bones. Other really interesting feature expressed by the shell, particularly the plastron of the allotype of N. zapatocaensis is the evidence of sexual dimorphism in basal pleurodires, which also corresponds to the earliest evidence of this phenotypical feature in early steps of the evolution of turtles.

PAGE 109

109 Notoe mys zapatocaensis was not the only turtle inhabiting the shallow marine margins of the proto -Caribbean seaway during the Valanginian. Also, fossil remains of primitive marine protostegids turtles have been also found at the same locality and stratigraphica l horizon (Cadena, submitted), additionally other undescribed material could correspond to a new genus of stem pleurodires. Other vertebrates found at the same type locality of N. zapatocaensis are pliosaurs and ichthyosaurs remains, fishes, and very prim itive crocodyliforms currently under study. All these vertebrates indicate pretty well developed trophic levels, with big predators such as the large marine reptiles and potential preys such as fishes and small turtles including N. zapatocaensis. A really big challenge for future paleontologic fieldwork campaigns will be to try to find any complete skull of Notoemys zapatocaensis that could be involve in a phylogenetic analysis in order to test the results ans phylogenetic hypothesis suggested here, based only in the available shells. The excellent preservation of the shells of N. zapatocaensis make them a potential target for histological studies, as well as to evaluated the potential preservation of stable isotopes signal in turtle bone, it which can be u sed as paleoenviromental or paleoclimatic proxy or inclusive to establish possible migration of tropical vertebrates faunas. Cerrejonemys wayuunaiki Middle to Late Paleocene of Colombia Cerrejonemys wayuunaiki is the first described turtle from the midd le to late Paleocene Cerrejn Formation, Cerrejn Coal Mine, Colombia; same locality and stratigraphical horizon as for Titanoboa cerrejonensis, the largest snake so far know (Head et al., 2009) and Cerrejonisuchus improcerus, a very short -snouted dyrosaur id crocodyloform (Hastings et al., in press). The cladistics analysis supports to Cerrejonemys as the sister taxon of the genus Podocnemis (Podocnemidinae), which ranges from the Miocene to Recent, implying a 47-

PAGE 110

110 million -year range extension of the ghost li neage for Podocnemis into the Paleocene of South America. The occurrence of Cerrejonemys during the middle to late Paleocene of Colombia indicates an early Cenozoic presence of podocnemidids in tropical environments of South America, and an neotropical o rigin for the Podocnemidinae subfamily, now constituted by Cerrejonemys and Podocnemis Cerrejonemys together with other undescribed turtles from Cerrejn including one of the largest pleurodire so far known, are excellent candidates to be used in paleoc limatic reconstructions in the same way as Titanoboa was used to establish mean annual temperature (Head et al., 2009). Additionally, the bone of these turtles is a potential source of paleoclimatic information through of the isotopic signal generally pres erved in the hydroapatite, which is the main mineral component of the bone after the fossildiagenesis process. The importance of the turtle bone as a proxy for paleoprecipation and paleotemperature has been tested for Paleocene Eocene fossil turtles from t emperate zones, analyzing oxygen isotopes (Matson and Fox, 2008). Pleurodires and Cryptodires from the Early to Middle Miocene of Panama The Early Miocene deltaic sediments of the Culebra Formation allowed the preservation of the earliest evidence of envi ronmental interaction between turtles with North America affinity (trionychids) and turtles with South America affinity (podocnemidids) in a tropical scenario. Podocnemidid turtles from Culebra Formation, could be close related with the extant genus Podocnemis However, at this point the fossil material recovered is not enough to get a better systematic resolution, particularly due to lacking skulls or complete shells. Other important aspect of the podocnemidids from Culebra is their large size, becoming inclusive bigger than the largest specimens reported for extant genus Podocnemis Trionychids or soft shell turtles from

PAGE 111

111 Culebra Formation, represent one of the earliest evidence of these turtles in tropical environments very close to Central and South Ameri ca. During the Early to Middle Miocene important changesin occurred in terms of the turtle fauna inhabiting the easternmost portion of the Central American Peninsula, particularly the arrival of abundant number of representative of North and Central Americ a cryptodires, such as kinosternidids, geoemyidids, testudinids, and the remaining of trionychids. Geoemydids are represented by the genus Rhinoclemmys with one new species R. panamaensis and at least to other taxa. Testudinids (tortoises) are probably re lated to Geochelone genus, and their size indicates that they were larger than the largest tortoises from Galapagos islands. The new species of the kinosternidids genus Staurotypus shows that this genus had a wider geographical distribution in Central Ame rica than today. The fossil turtles described in the chapters 1, 2 and 3 not only bring important morphological information for the understanding of the evolutionary history of turtles, particulary of pleurodires in tropical South America, but also have re levant palebiogeographical significance and potentially a tool for access to paleoclimatic information, which is extremely scarse in Mesozoic and Cenozoic neotropical marine and terrestrial sequences.

PAGE 112

112 APPENDIX A CHAPTER 2 DATA Description of characters u sed in phylogenetic analysis. Characters were polarized with respect to Odontochelys The source from where the character was taken, as well as if was modified or it is a new character is indicated and the end of its description. Carapace 1 Carapace: absent (0); present (1). Li et al. (2008) 2 Cervical scale: one or three cervicals, being the central cervical wider than long (0); central cervical much longer than wide (1), absent (2). Character modified from Laparent de Broin and Murelaga (1999) and Joyce (2007). 3 Nuchal bone: wider than long (0); as long as wide or longer than wide (1). Character taken from De la Fuente (2003). 4 Articulation of nuchal with neural spine of eight cervical vertebra: along a blunt facet (0); articulation absent (1). Character modifi ed from Joyce (2007). 5 Neural one: longer than wide (0); slightly wider than long, being the smallest neural of the series (1), neural one absent (2). New character. 6 Arrangement between neural one, neural two, costal one and costal two: neural one contacts costal one and two laterally, and neural two only contacts costal two (0); neural one and costal one exclusively in contact between each one, and neural two only contacts costal two (1); neural one only contacts costal one laterally, and neural two contact s costal one anterolaterally (2); neural series absent (3). New character. 7 Neural two: almost as long as wide (0); much wider than long (1); neural two absent (2). New character 8 Carapace posteriorly notched: present (0); absent (1). Pygal notch character f rom Cadena and Gaffney, (2005). 9 Supramarginal scales: full series of twelve, in both sides of the carapace (0); incomplete series, restricted to the anterior margin, in both sides of the carapace (1); absent (2). Character modified from Cadena and Gaffney, (2005). 10. Posterior lobe of the carapace: same width at the anterior lobe or slightly wider (0); tapering medially (1). Character modified from Cadena and Gaffney, (2005). 11. Articulation tubercle on anterior face of the first thoracic rib: absent, smooth ant erior face (0); present (1). Character modified from Cadena and Gaffney, (2005).

PAGE 113

113 12. Costovertebral tunnel: wide anteriorly and posteriorly (0); very wide entire length (1); small entire length (2). Character taken from Cadena and Gaffney, (2005). 13. Thoracic v ertebrae: cylindrical, longer than wide, keeled ventrally (0); smooth and flat ventrally, hexagonal in shape with central lateral notch (1). Character modified from Cadena and Gaffney, (2005). 14. Medial space between the first and the second thoracic ribs: v ery wide (0); moderately reduced (0); very reduced (1). New character 15. Axillary process: contacts peripherals only (0); contacts costal 1 or the sutural contact between costal 1 and 2 (1). Character modified from Cadena and Gaffney, (2005). 16. Dorsal knobs at the posterior region of pleural and vertebral scales: high, with radial striation pattern (0); low, with smooth dorsal surface (1); absent (2). New character 17. Suprapygal one: parallel sided (0); tapers anteriorly (1); absent (2). Character modified from C adena and Gaffney, (2005). 18. Vertebral scale two and three: hexagonal in shape, much wider than long (0); rectangular in shape, slightly wider than long (1). hexagonal or rectangular longer than wide (2). Character modified from Cadena and Gaffney, (2005). 19. Posterior edge of the carapace: deeply serrated (0); slightly serrated (1); smooth (2). Character modified from Cadena and Gaffney, (2005). 20. Medial contact of the posterior costals: absent (0); present (1); present due to complete absence of neural series (2). Character modified from Joyce (2007). 21. Lateral position of vertebrals 3 4 sulcus in taxa with five vertebrals: sulcus positioned on costals 6 (0); sulcus positioned on costals 5 (1). Character modified from Joyce (2007). 22. Pelural scales onto periphera ls: present (0); absent (1). Character modified from Lapparent de Broin and De la Fuente (2001). Plastron 1 Posterior epiplastral process: present (0); absent (1). Character from Cadena and Gaffney (2005). 2 Posterior entoplastral process: present (0); abse nt (1). Character from Cadena and Gaffney (2005). 3 Entoplastron participation in the anterior margin of the plastron: wide participation (0); short participation (1); lacking participation (2). Character modified from Cadena and Gaffney (2005). 4 Intergula r scales: two (0); one (1). Character from Cadena and Gaffney (2005).

PAGE 114

114 5 Intergular scales: covering slightly the anterior portion of the entoplastron (0); covering most of the entoplastron posteriorly (1); uncovering the entoplastron (2). Character modified from Cadena and Gaffney (2005). 6 Anterior plastral lobe border: defined by tuberosities, dentate margin (0); very reduced tuberosities, straight to slightly dentate margin (1); lacking tuberosities, smooth, highly convex margin (2). New character. 7 Mesopla stra: two pairs of mesoplastral with midline contact(0); one mesoplastra1l pair, with midline contact, or reaching the central fontanella border (1); one mesoplastral pair, no midline contact, wider than long (2); one mesoplastral pair, no midline contact, as long as wide (3), absent (4). Character modified from Cadena and Gaffney (2005). 8 Central plastral fontanella: absent (0); present (1). Character modified from Cadena and Gaffney (2005). 9 Inframarginal scales: present (0); absent (1). Character from De la Fuente and Iturralde (2001). 10. Anal notch: absent, straight to slightly concave posterior edge of the xiphiplastra (0); present, well developed in U or V open shape (1). Character modified from Joyce (2007). 11. Sutural articulation of pelvis to shell: abs ent (0); present (1). Character from Joyce (2007). 12. Iliac scar: absent (0); extends from costals onto the peripherals (1); restricted to the eighth costal (2); positioned on eighth costal and pygal, sometimes reaching the seventh costal (3). Character mod ified from Joyce (2007). 13. Humeral scales: medially in contact (0); lacking medial contact being separated by the intergular (1). New character 14. Shape of ilium articular site: narrow and pointed posteriorly (0); oval (1). Character from De la Fuente and Itu rralde (2001).

PAGE 115

115 APPENDIX B CHAPTER 2 CHARACTER MATRIX Character matrix ( taxa and 36 shell characters) used for phylogenetic analysis 000000000111111111122222222223333333 123456789012345678901234567890123456 Odontochelys 0??????????????????????00000 00?0???? Proganochelys 10?0???000000000?00???00100010?0000? Proterochersis 10?????010000?00?00???0010000000?00? Indochelys 10??000120??????00?000111??111100??? Kayentachelys 100000012000??020020001011011010000? Platychelys 1001000111111110010000 11210121111100 N. laticentralis 110100112111121110200011????21111201 N. zapatocaensis 110111012111121101100011211121111211 N. oxfordiensis 11010001211???11??????1?211221111201 Araripemys barretoi 1?01000120?00??201201010212241111301 Dortoka 10110201200?0?1222211011210240111301 Bonapartemys bajobarrealis 110?000120????1222211111210230111301 Lomalatachelys neuquina 110?00012000??1222211111210230111301 Brasilemys josai 1211000120000?1222211??????????????? Cearachelys placidoi 121100012000??1222201011210230111?0? Bauremys elegans 1211000?20000?1222211011210230111?01 Podocnemis expansa 121100012000021222211011210230111301 Pelomedusa subrufa 121102012000021222211011210230111301 Phrynops dahli 111123212000021222221011210240111301

PAGE 116

116 APPENDIX C CH APTER 3 DATA Description of characters used in phylogenetic analysis. Characters were polarized with respect to Chelidae, Bothremydidae, Araripemydidae, Euraxemydidae and Pelomedusidae. Characters are from Gaffney et al. (2006) unless indicated otherwise. Skull (1) Nasals: present (0); absent (1). (2) Prefrontals meet on midline: absent (0); present (1). (3) Quadratojugal: absent (0); present (1). (4) Squamosal parietal contact: present (0); absent (1). (5) Quadratojugal parietal contact: absent (0); pr esent (1). (6) Temporal emargination, secondary roofing of the fossa temporalis in dorsal view, not advanced and highly concave allowing the complete exposure of the otic chamber roof (0); medially advanced with posteriorly expanded posterolateral tempo ral emargination of the parietals and quadratojugal with concave margins, covering partially or almost totally the otic chamber roof (1); very advanced with convex to straight tapering margins completely covering the roof of the otic chamber (2). Character modified from Lapparent de Broin (2000). (7) Prefrontal, anterior overhang onto apertura narium externa: shaped by the nasals (0); by the prefrontals, covering a small portion of the posterior part of the apertura, ending in acute medial tip (1); by the p refrontals, completely covering the apertura, ending in a straight to convex edge (2). Character modified from Gaffney et al. (2002). (8) Prefrontal, interorbital sulcus at the sutural area between both prefrontals: absent (0); present (1). Character from Lapparent de Broin (2000).

PAGE 117

117 (9) Prefrontal at the interorbital space: wide (0); narrow (1). Character modified from Gaffney et al. (2006). (10) Frontal, orientation of the orbits: orbits facing more laterally (0); orbits facing more dorsally (1). (11) Pari etal jugal contact: absent (0); present (1). Character from De la Fuente, (2003). (12) Supraoccipital, crista supraoccipitalis: very short to absent (0); long, ventrally wider with uniform width from the anterior to the posterior aspect, ending in an acute tip in dorsal view (1); short, wider posteroventrally than anteroventrally, ending in a bulbous shape in dorsal view (2). In Dacquemis the crista supraoccipitalis is long, but is hidden dorsally by the large exposure of the supraoccipital at the posterio r roof of the skull, coded as 1. Character modified from Gaffney et al. (2006). (13) Interparietal scale, anterior margin: anterior to the frontal parietal suture (0); posterior to the frontal parietal suture (1). New character. (14) Condylus occipitalis: formed by exoccipitalis and the basioccipital(0); formed only by exoccipitalis (1). (15) Quadrate basioccipital contact: absent (0); present (1). (16) Quadrate, cavum tympani, incisura columella auris: open without posterior bony restrictions (0); enclo sed or with a narrow fissure separating the Eustachian tube and stapes (1); enclosed together with both stapes and Eustachian tube in the same oval dilated opening (2). Character modified from Gaffney et al. (2006) and Lapparent de Broin et al. (2007). Fo r the basal bothremydidae Cearachelys, the incisura is open and dilated including the Eustachian tube, but without the complete posterior closure of the quadrate as in Podocnemididae (Lapparent de Broin et al., 2007) plus Hamadachelys.

PAGE 118

118 (17) Quadrate, cavu m tympani, fossa precolumellaris: deep (0); shallow (1); absent (2). Character modified from Gaffney et al. (2006). (18) Quadrate, ventral projection: very short, condylus mandibularis very close to the cavum tympani region (0); short, condylus mandibular is slightly separated from the cavum tympani region (1); long, condylus mandibularis considerably separated from the cavum tympani region (2). New character. (19) Quadrate, eustachian tube separated by bone from the fenestra postotica: absent (0); present (1). Character from Gaffney and Wood (2002). (20) Cheek emargination, secondary lateral roofing of the fossa temporalis : fossa temporalis laterally exposed without secondary roofing (0); secondary roofing slightly advanced (1); secondary roofing moderately advanced by the descending of only the quadratojugal (2); secondary roofing moderately advanced by the descending of both the jugal and the quadratojugal (3); fossa temporalis completely roofed by the the jugal, resulting in a contact between the quadrate and the jugal; occasionally with a small notch at the posterolateral margin of the jugal (4). Character modified from Gaffney et al. (2006). (21) One or two accessory ridges on the ventral surface of the premaxilla: absent (0); present (1). One accessor y ridge on the ventral surface of the premaxilla is only seen outside of the podocnemidids in the recently described bothremydid Acleistochelys (Gaffney et al., 2007). Within Podocnemididae, only Dacquemys exhibits similar accessory ridges, curved anterior ly to join each other, but they are restricted to the maxilla. (Gaffney et al., 2002). New character. (22) Vomer: present (0); absent (1). (23) Basioccipital: long (0); short (1).

PAGE 119

119 (24) Opisthotic, processus paroccipitalis: small and flat, does not proj ect beyond the squamosal (0); medially narrowed and elongated, projects beyond the squamosal ending in a prominent tip (1). Character modified from Gaffney et al. (2006) and Lapparent de Broin et al. (2007). (25) Basisphenoid quadrate contact: absent (0); present (1). Character modified from Gaffney et al. (2006). (26) Basioccipital opisthotic contact: present (0); absent (1). (27) Pterygoid, cavum pterygoidei = fossa podocnemidoid of Lapparent de Broin, (2000): absent (0); shallow and slightly hidden a nteromedially by the underlapping basisphenoid medially and the pterygoid laterally (1); deep and partially to totally covered by the pterygoid flange (posterolateral wings of the pterygoid) (2). Character modified from Gaffney et al. (2006). (28) Pterygoi d, pterygoid flange = pterygoid wings (Lapparent de Broin, 2000): absent to very short (0); moderately developed (1); well developed reaching the caudal margin of the quadrate ramus of the bone and projected ventrally (2). Character modified from Frana and Langer (2006), and Lapparent de Broin (2000). (29) Palatine, foramen palatinum posterius: present (0); absent (1). Character modified from Gaffney et al. (2006). (30) Palatine, second palate: absent (0); present (1). (31) Quadrate, condylus mandibulari s shape: much wider than long, with anterior and posterior edges straight to concave making it shorter at midline (0); slightly wider than long in a kidney bean shape, with anterior edge straight to concave and posterior edge convex (1). New character. (32) Exoccipital quadrate contact: absent (0); extensive (1); narrow (2). Character modified from Gaffney et al. (2006).

PAGE 120

120 (33) Prootic quadrate contact: present (0); absent (1). Lower Jaw (34) Dentary, fused symphysis : absent (0); present (1). (35) Dentar y, internal angle between rami: acute, between 40 and 90 degrees (0); obtuse, over 90 degrees (1); very acute, less than 40 degrees (2). Internal angle between rami less to 40 degrees is also seen in UNEFM CIAPP 1399, an indeterminate podocnemidid (Gaffney et al., 2008), although it was excluded from this phylogenetic analysis due to poor preservation. New character. (36) Articular, processus retroarticularis: very short (0); short, projected posteriorly (1); short, projected ventrally (2); long, projected posteriorly (3). Character modified from Gaffney and Forster (2003) and Gaffney et al. (2006). (37) Surangular, well -extended anteriorly: absent (0); present (1). (38) Coronoid, wide lateral exposure: absent (0); present (1). (39) Dentary, accesory ri dges: absent (0); present (1). Character from Gaffney and Forster (2003). (40) Dentary, narrow and elongated ridge, located in the medial margin on the ventral surface: absent (0); present (1). New character. Cervical Vertebrae (41) Ventral keel at the p osterior condyle: protuberant (0); reduced almost absent (1). Character from Lapparent de Broin (2000). The condition for bothremydids is based on the recently described Acleistochelys (Gaffney et al., 2007). (42) Posterior condyle of the sixth or previou s cervical vertebrae in a horse -saddle shaped, higher than wide: absent(0); present(1). Character modified from Lapparent de Broin (2000).

PAGE 121

121 Coracoid (43) Coracoid shape: slightly curved longitudinally and much wider distally (0); narrow, almost straight lo ngitudinally and slightly wider distally (1). Character modified from Gaffney et al. (2006). (44) Coracoid, dorsal longitudinal ridge: absent (0); present (1). Character from Frana and Langer (2006). Carapace (45) Cervical scale: present (0); absent (1) (46) Nuchal bone: wider than long (0); longer than wide (1). Character from De la Fuente (2003) (47) Neural series composed of: eight or more bones (0); seven to 1 bones (1); neurals completely absent (2). Character modified from De la Fuente (2003) (48) Neural 2 present without a contact with costal 1 (0); present with a contact with costal 1 (1); neural 2 absent (2). Character modified from Fran a and Langer (2006). (49) Lateral thickness of t he shell: Frana and Langer (2006). Plastron (50) Plastral bridge: short (0); elongated (1). Character from De la Fuente (2003). (51) Pectoral scale contact with the epiplastron: absent (0); present (1). Ch aracter from De la Fuente (2003). (52) Pectoral scale, contact with the entoplastron: absent (0); present (1). Character from De la Fuente (2003). (53) Intergular scale: large, covering the anterior margin of entoplastron, separating the gulars (0); small, restricted between the gulars, lacking contact with entoplastron (1). Character modified from Gaffney et al. (2006).

PAGE 122

122 APPENDIX D CHAPTER 4 CHARACTER MATRIX Appendix 3 2. Character matrix (29 taxa and 53 characters) used for phylogenetic analysis (Nexus f ile as Supplementary Data 2, www.vertpaleo.org/jvp/JVPcontents.html). Polymorphic conditions abbreviated as follows: a = (0&1), b = (0&2), c = (1&2). Chelidae 00000000aa 00000b0000 0000000000 000a000000 0000000000 0a0 Araripemydidae 1111001011 00?0000001 0?00010000 0000000100 00??100100 000 Pelomedusidae 1111001011 00010b0003 0100a00000 0001000000 0000111001 000 Euraxemydidae 111a101011 00?0001003 000a010000 0200000000 00??110001 000 Bothremydidae 111aaac0aa 000a112002 0010110000 0111a3a0a0 00001aaa0a aa0 Brasilemys josai 1111001000 00?0100001 0?11101000 001??11000 ????11010? ??? Hamadachelys escuillie ?111101011 0?00100003 0011101000 0011011000 ????1????? ??? Portezueloemys patagonica ?11110???? ????1????? 0?11101?0? ?????????? ????111a01 010 Bauruemys el egans 1111101011 0100120003 0011102200 0011011000 0100111101 110

PAGE 123

123 Aff. Roxochelys vilavilensis 1111101011 0100120003 0011102200 0011021000 0100111001 110 Shweboemys antiqua 1111112000 0?00120103 0011102111 001??????? ?????????? ??? Stereogenys cromeri 11??? ?2?00 ??001201?3 0011102111 001??????? ?????????? ??? Daquemys paleomorpha 1110112000 0110120003 0011102110 001??????? ?????????? ??? Cerrejonemys wayuunaiki 111110101? 110012?0?3 ??11102200 ?011221001 111111?01? ?1? Neochelys arenarum 1111112000 0100120003 0011102100 0011011000 ????1110?1 11a Neochelys lapparenti 1111112000 0000120??3 0?11102?00 001??????? ?????????? ??? Podocnemis bassleri 1111101111 1100120003 1011102200 101??????? ?????????? ??? Bairdemys venezuelensis 1111112000 0200120213 0010102201 0 0111????0 ????112201 110 Bairdemys hartsteini 1111112000 0200120213 0010102201 001??????? ?????????? ??? Bairdemys sanchezi 11111?2000 ???0120213 0010102111 0?1112100? ?????????? ??0

PAGE 124

124 Bairdemys winklerae 1111112000 ?2??120213 0010102201 001111100? ?????????? ??? Erymnochelys madagascarensis 1111112000 0100120004 0011102200 1011011000 0010111001 111 Peltocephalus dumerilianus 1111112000 0100120004 0011102200 1011011000 0110111001 010 Podocnemis expansa 1111101111 110012b003 1011102200 1011021010 1111111001 110 Podocnemis sextuberculata 1111101111 1100120003 1011102200 1011021010 1111111001 110 Podocnemis vogli 1111101111 1100120003 1a11102200 1011021010 1111111001 110 Podocnemis lewyana 1111101111 1100120003 1011102200 1011021010 1111111001 110 Podocnemis eryt hrocephala 1111101111 1100120003 1011102200 1011021010 1111111001 110 Podocnemis unifilis 1111101111 1100120003 1011102200 1011021010 1111111001 110

PAGE 125

125 REFERENCES Albert, J. S., N. R. Lovejoy, and W. G. R. Crampton. 2006. Miocene tectonism and the separa tion of cis and trans -Andean river basins: evidence from Neotropical fishes. Journal of South American Earth Sciences 21:14 27. Arnold, E. N. 1979. Indian Ocean giant tortoises: their systematics and island adaptations. Philosophical Transactions of the R oyal Society. 286:127 145. Bayona, G., C. Jaramillo, M. Rueda, A. Pardo, A. Christie, and G. Hernandez. 2004. Important paleotectonic and paleogeographic considerations of the late Paleocene in the Northermost Andes as constrained by Paleogene rocks in the Cerrejon Coal Mine, Guajira, Colombia. III Convencin Tcnica ACGGP. La inversin en el conocimiento geolgico, P4, CD ROM, Bogot. Bloch, J., E. Cadena, F. Herrera, S Wing, and C. Jaramillo. 2005. Paleocene vertebrates from the Cerrejn Formation, Guaji ra Peninsula, n ortheastern Colombia. Journal of Vertebrate Paleontology 25(3, Supplement):3738A. Bocquentin, J., and J. Melo. 2006. Stupendemys souzai sp. nov. (Pleurodira, Podocnemididae) from the Miocene Pliocene of the Solimes Formation, Brazil. Revis ta Brasileira de Paleontologa 9:187 192. Bonin, F., B. Devaux, and A. Dupr. 2006. Toutes les tortues du monde. Delachaux et Niestl, Paris, 415 pp. Cadena, E. A., and E.S. Gaffney. 2005. Notoemys zapatocaensis a new side -necked turtle (Pleurodira:Platy chelyidae) from the Early Cretaceous of Colombia. American Museum Novitates, 3470:1 19. Cadena, E. A., M. Paramo, and C. Jaramillo. 2007. The first Late Pleistocene record of Kinosternon (Cryptodira: Kinosternidae) turtles for northern South America, Puben za Locality, Colombia. South American Journal of Herpetology 2:201 205. Cadena, E., J.Bloch, and C. Jaramillo. 2008. Paleocene turtles from Colombia: Phylogenetics, Paleobiogeographic and Paleoclimatic implications. Abstracts III Congreso Latinoamericano de Paleontologa de Vertebrados. Neuquen, Patagonia Argentina. 56 57. Carr, J.L. 1991 Phylogenetic analysis of the neotropical turtle genus Rhinoclemmys Fitzinger (Testudines: Emydidae). DPhil. Thesis, Southern Illinois University pp. 635 Carvalho, P., J. Bocquentin, and F. Lapparent de Broin. 2002. A new species of Podocnemis (Pleurodira, Podocnemididae) from the Neogene of the Solimes Formation, Acre, Brazil. Geobios 35:677 686. Claude, J, and H. Tong. 2004. Early Eocene testudinoid turtles from Saint P apoul, France with comments on the Early Evolution of modern Testudinoidea. Oryctes 5:3 45.

PAGE 126

126 Claude, J., E. Paradis, H. Tong, and J. Auffray. 2003. A geometric morphometric assessment of the effects of environment and cladogenesis on the evolution of the t urtle shell. Biological Journal of the Linnean Society 79:485 501. Coates A.G, L. Collins, M. Aubry, and W. Berggren. 2004. The geology of the Darien, Panama, and the late Miocene Pliocene collision of the Panama arc with northwestern South America. Geolog ical Society of America Bulletin 116:1327 1344. Coates A.G, and J. Obando. 1996. The geologic evolution of the Central American isthmus. In: Jackson JBC, Budd AF, Coates AG, (eds.) Evolution and environment in tropical America. Chicago, University of Chica go Press. pp 21 56. Coto, A. and R. Acua. 1986. Filogenia de Geochelone costarricensis y la familia Testudinidae (Reptilia: Testudines) en el continente americano. Revista de Biologia Tropical. 34:199 207. Cope, E. 1864. On the limits and relations of the Raniformes. Proceedings of the Academy of Natural Sciences of Philadelphia 16:181 183. Cope, E. 1868. On the origin of genera. Proceedings of the Academy of Natural Sciences of Philadelphia 20:242 300. Danilov, I, and J. Parham. 2008. A reassesment of some poorly known turtles from the middle Jurassic of China, with comments on the antiquity of extant turtles. Journal of Vertebrate Paleontology 28:306 318. Datta, P.M., P. Manna, S. Ghosh, adn D. Das. 2000. The first Jurassic turtle from India. Paleontolog y 43:99 109. De la Fuente, M. 2003. Two new pleurodiran turtles from the Portezuelo Formation (Upper Cretaceous) of Northern Patagonia, Argentina. Journal of Paleontology 77:559 575. De la Fuente, M. 2007. Testudines. In Z. Gasparini, L. Salgado and R. Cor ia (eds.), Patagonian Mesozoic Reptiles. Indiana University Press, Bloomington & Indianapolis. p. 50 86 De la Fuente, M. S., and M. Iturralde -Vinent. 2001. A new pleurodiran turtle from the Jagua formation (Oxfordian) of western Cuba. Journal of Paleontolo gy 75(4):860 869. Fernandez, M. S., and M. S. De la Fuente. 1988. Nueva tortuga (Cryptodira: Thalassemydidae) de la formacin Vaca Muerta (Jurasico, Tithoniano) de la Provincia del Neuquen, Argentina. Ameghiniana 25 (2):129 138. Fernandez, M. S., and M. S. De la Fuente. 1994. Redescription and phylogenetic position of Notoemys : the oldest Gondwanian pleurodiran turtle. Neues Jahrbuch fur Geologie und Palaontology Abhandlungen 193(1):81 105.

PAGE 127

127 Frana, M, and M. Langer. 2005. A new freshwater turtle (Reptilia, Pleurodira, Podocnemidae) from the Upper Cretaceous (Maastrichtian) of Minas Gerais, Brazil. Geodiversitas 27:391 411. Frana, M, and M. Langer. 2006. Phylogenetic relationships of the Bauru Group turtles (late Cretaceous of South Central Brazil). Revista Brasileira de Paleontologa 9:365 373. Gaffney, E.S. 1990. The comparative osteology of the Triassic turtle Proganochelys Bulletin of the American Museum of Natural History 194:1 263 Gaffney, E., and P. Meylan. 1988. A phylogeny of turtles. in M.J. Benton (ed.), The phylogeny and classification of the tetrapods. Vol. 1. Amphibians, reptiles, birds. Systematics Association Special Volume 35A:157 219. Gaffney, E., K. Campbell, and R. Wood. 1998. Pelomedusoid side -necked turtles from Late Miocene Sediments in southwestern Amazonia. American Museum Novitates 3245:1 12. Gaffney, E., D. DeBlieux, E. Simons, M. Snchez Villagra, and P. Meylan. 2002. Redescription of the skull of Dacquemys Williams, 1954, a podocnemidid sidenecked turtle from the Late Eocene of Egypt. American Museum Novitates 3372:1 16. Gaffney, E., and C. Forster. 2003. Side -necked turtle lower jaws (Podocnemididae, Bothremydidae) from the Late Cretaceous Maevarano Formation of Madagascar. American Museum Novitates 3397:1 13. Gaffney, E., and R. W ood. 2002. Bairdemys, a new side necked turtle (Pelomedusoides:Podocnemididae) from the Miocene of the Caribbean. American Museum Novitates 3359:1 28. Gaffney, E., H. Tong, and P. Meylan. 2006. Evolution of the side -necked turtles: the families Bothremydidae, Euraxemydidae, and Araripemydidae. Bulletin of the American Museum of Natural History 300:1 698. Gaffney, E., E. Roberts, F. Sissoko, M. Bouar, L. Tapanila, and M. OLeary. 2007. Acleistochelys a new side -necked turtle (Pelomedusoides: Bothremydidae) from the Paleocene of Mali. American Museum Novitates 3549:1 24. Gaffney, E., T. Scheyer, K. Johnson, J. Bocquentin, and O. Aguilera. 2008. Two new species of the side necked turtle genus, Bairdemys (Pleurodira, Podocnemididae), from the Miocene of Venezu ela. Palontologische Zeitschrift 82(2):209 229. Gray, J. 1825. A synopsis of the genera of reptiles and amphibia, with a description of some new species. Annals of Philosophy 10:193 217. Guzman, G. 1985. Los Grifeidos infracretacicos Aetostreon couloni y Ceratrostreon boussingaulti de la Formacin Rosablanca, como indicadores de oscilaciones marinas. In Etayo, S. Proyecto Cretcico Publicacion especial del Ingeominas. Bogot: pp. XII 1 (1 16).

PAGE 128

128 Hastings, A., J. Bloch, E. Cadena, and C. Jaramillo. In press. A new small short -snouted dyrosaurid (Crocodylomorpha, Mesoeucrocodylia) from the Paleocene of Northeastern. Journal of Vertebrate Paleontology. Head, J., O. Aguilera, and M. Sanchez-Villagra. 2006. Past colonization of South America by Trionychid turtles : Fossil evidence from the Neogene of Margarita Island, Venezuela. Journal of Herpetology 40:380 383. Head, J., J. Bloch, A. Hastings, J. Bourque., E. Cadena, F. Herrera, D. Polly, and C. Jaramillo. 2009. Giant boid snake from the Palaeocene neotropics rev eals hotter past equatorial temperatures. Nature 457(5):715 718. Herrera, A., M. Dick., J. Sanner., and J. Jackson. 2008. Neogene Cupuladriidae of Tropical America. II: Taxonomy of recent Discoporella from opposite side of the Isthmus of Panama. Journal of Paleontology 82:279 298. Hirayama R. 1984. Cladistic analysis of batagurine turtles (Batagurinae: Emydidae: Testudinidae); a preliminary result. Studia Geologica Salmanticensia 1:141 157. Hoffstetter, R., and Gasc, J. 1969. Vertebrae and ribs of modern reptiles. Biology of the Reptilia 1:201 301. Horton, B. K., B. Hampton, G. Waanders. 2001. Paleogene synorogenic sedimentation in the Altiplano plateau and implications for initial mountain building in the central Andes. GSA Bulletin 113:1387 1400. H utchison J. H.. 1991. Early Kinosterninae (Reptilia: Testudines) and their phylogenetic significance. Journal of Vertebrate Paleontology11:145 167. Hutchison, J.H. 2006. Bridgeremys (Geoemydidae,Testudines), a new genus from the middle Eocene of NorthAmerica. Foss il Turtle Research 1:63 83. ICZN. 1999. International Code of Zoological Nomenclature. Available at http://www.iczn.org/. Accessed on January 2009. Jaramillo, C., G. Bayona, A. Pardo, M. Rueda, G. Harrington, and G. Mora. 2007. Palynology of the Upper Pa leocene Cerrejn Formation, Northern Colombia. Palynology 31:153 189. Joyce, W. 2007. Phylogenetic relationships of Mesozoic turtles. Bulletin of Peabody Museum of Natural History 48(1):1 100. Kirby, M., D. Jones., and B. MacFadden. 2008. Lower Miocene str atigraphy along the Panama Canal and its bearing on the Central American Peninsula. PlosOne Vol 3:1 14. Kischlat, E. 1994. Observaes sobre Podocnemis elegans Suarez (Chelonii, Pleurodira, Podocnemididae) do Neocretceo do Brasil. Acta Geolgica Leopolden sia 17:345 351.

PAGE 129

129 Lapparent de Broin, F. de., 1991 Fossil Turtles from Bolivia; in R. Suarez Soruco (ed.), Fsiles y Facies de Bolivia, Vertebrados, Volume 12(34): 509 527. Revista Tcnica de los Yacimientos Petroliferos Fiscales Bolivianos. Lapparent de Br oin, F. de. 2000. The oldest pre -Podocnemidid turtle (Chelonii, Pleurodira), from the Early Cretaceous, Cear state, Brasil, and its environment. Threeballs del Museu de Geologa de Barcelona 9:43 95. Lapparent de Broin, F. 2001. The European turtle fauna from the Triassic to the Present. Dumerilia 4(3):155 217. L apparent de Broin, F ., and M. De la Fuente. 2001. Oldest world Chelidae (Chelonii, Pleurodira), from the Cretaceous of Patagonia. Comptes Rendus de lAcadmie des Sciences de Paris 333:463 470. Lap parent de Broin, F. de., M. de. La Fuente, and M. Fernandez. 2007. Notoemys laticentralis (Chelonii, Pleurodira), Late Jurassic of Argentina: new examination of the anatomical structures and comparisons. Revue de Palobiologie 26(1):99 136. Lapparent de Br oin, F. de., J. Bocquentin, and F. Negri. 1993. Gigantic turtles (Pleurodira, Podocnemididae) from the Late MioceneEarly Pliocene of south western Amazon. Bulletin de LInstitut franais dEtudes andines 23:657 670. Lapparent de Broin, F., and X. Murelaga 1999. Turtles from the Upper Cretaceous of La o (Iberian Peninsula). Estudios del Museo de Ciencias Naturales de Alava 14 (Nmero Especial 1):135 211. L apparent de B roin, F., X. M urelaga., and V. C odrea 2004. Presence of Dortokidae (Chelonii, Pleurodir a) in the earliest Tertiary of the Jibou Formation, Romania, Paleobiogeographical implications. Acta Paleontologica Romaniae 4:203 215. Laurito, C., A. Valerio., L. Gomz., J. Mead., E. Prez., and L Prez. 2005. A trionychidae (Reptilia: Testudines, Crypt odira) from the Pliocene of Costa Rica, Southern Central America. Revista Geolgica de Amrica Central. 32:7 11. Le, M. and W. Mccord. 2008. Phylogenetic relationships and biogeographical history of the genus Rhinoclemmys Fitzinger, 1835 and the monophyly of the turtle family Geoemydidae (Testudines:Testudinoidea). Zoological Journal of the Linnean Society 153:751 767. Li, C., X. Wu, O. Rieppel, L. Wang, and L. Zhao. 2008. An ancestral turtle from the Late Triassic of Southwestern China. Nature 456:497 501. MacFadden, B. 2006. North American Miocene land mammals from Panama. Journal of Vertebrate Paleontology 26:720 734. Maddison. W, and D. Maddison. 2008. Mesquite: a modular system for evolutionary analysis. Version 2.5(build j77)

PAGE 130

130 Meylan, P. A. 1996. Skel etal morphology and relationships of the Early Cretaceous side necked turtle, Araripemys barretoi (Testudines: Pelomedusoides: Araripemydidae), from the Santana Formation of Brazil. Journal of Vertebrate Paleontology 16:20 33. Meylan, P. A, and W. Sterrer. 2000. Hesperotestudo (Testudines: Testudinidae) from the Pleistocene of Bermuda, with comments on the phylogenetic position of the genus. Zoological Journal of the Linnean Society 128:51 76. Meylan, P. A., E. Gaffney, and D. De Almeida Campos. 2009. Caninemys a New side -necked turtle (Pelomedusoides: Podocnemididae) from the Miocene of Brazil. American Museum Novitates 2639:1 26. Montes, C., R. Hatcher, and P. Restrepo. 2005. Tectonic reconstruction of the northern Andean blocks: Oblique convergence and r otations derived from the kinematics of the Piedras Girardot area, Colombia. Tectonophysics 399:221 250. Noonan, B. 2000. Does the phylogeny of pelomedusoid turtles reflect vicariance due to continental drift?. Journal of Biogeography 27:1245 1249. Ogg, J. G., G. Ogg, and F. Gradsteing. 2008. The concise Geologic Time Scale. Cambridge University Press. Pacheco, J. 1913. Notas sobre la geologia do Vale do Rio Grande, a partir da foz do Rio Pardo at a sua confluncia com o Rio Paranahyba; pp 33 38 in J. Doura dos (ed.), Explorao do Rio Grande e de seus afluentes. Comisso de Geografia e Geologia. Price, L. 1953. Os quelnios da Formao Bauru, Cretceo terrestre do Brasil Meridional. Rio de Janeiro, Departamento Nacional da Produo Mineral, Diviso de Geologia e Mineralogia, Boletim 147, 34 pp. Pritchard, P. 1984. Evolution and zoogeography of South American turtles. Studia Geological Salmanticensia vol especial 1:225 233. Pritchard, P. 2008. Evolution and structure of the turtle shell In: J. Wyneken, M. God frey and V. Bels (eds.). Biology of turtles. CRC press. p 45 83 Romano, P. S., and S. A. Azevedo. 2006. Are podocnemidid turtles relicts of a spread Cretaceous ancestor? South American Journal of Herpetology 1(3):175 184. Snchez -Villagra, M., R. Asher, A. Rincn, A. Carlini, P. Meylan, and R. Purdy. 2004. New faunal reports from the Cerro La Cruz locality (Lower Miocene), north -western Venezuela. In M. R. Sanchez -Villagra and J. A. Clack (eds.), Fossils of the Miocene Castillo Formation, Venezuela: Contrib utions on Neotropical Paleontology. Special Papers in Paleontology 71:105 112. Snchez -Villagra, M., and J. Winkler. 2006. Cranial variation in Bairdemys turtles (Podocnemididae: Miocene of the Caribean Region) and description of a new material from Urumac o, Venezuela. Journal of Systematic Paleontology 4:241 253.

PAGE 131

131 Savage J.M. 2002. The reptiles and amphibians of Costa Rica. Chicago, IL: The University of Chicago Press pp. 934. Scheyer, T.M, B. Brllman, and M. Sanchez -Villagra. 2008. The ontogeny of the she ll in side necked turtles, with emphasis in the homologies of costal and neural bones. Journal of Morphology 269:1008 1021. Schweigger, A.F. 1812. Monographie Cheloniorum. in R. Knigsberg (ed.), Arch. Naturwiss. Math. 1:271 368, 406 462. Segura, A. 1944. Estudio de la primera especie nueva de tortuga fsil de Costa Rica, con algunas generalidades sobre el orden testudines. La Escuela de Farmacia 77 78:13 14. Sorenson, M. and E. Franzosa, 2007. TreeRot Version 3. Available at http://people.bu.edu/msoren/Tre eRot.html. Accessed May 10, 2009. Staesche, K. 1937. Podocnemis brasiliensis n.sp. aus der oberen Kreide Brasiliens. Neues Jahrbuch Mineralogie, Geologie und Paleontologie B 77:291 309. S terli J. 2008. A new, nearly complete stem turtle from the Jurassic of South America with implications for turtle evolution. Biology letters 22:1 4. Suarez J. 1969. Um quelnio da Formao Bauru; pp. 168 176 in Congresso Brasileiro de Geologia 23 Anais, Salvador. Swofford, D. L. 2002. PAUP*. Phylogenetic Analysis Using P arsimony (*and other methods). Version 4.0b10. Sinauer Associates, Sunderland, Massachusetts. Treviranus, G. 1802. Biologie, oder Philosophie der lebenden Natur, fr Naturforscher und Aerzte. Johann Friedrich Rwer, Gttingen, I, 477 p. Wagler, J. 1830. Na trliches System der Amphibien, mit Vorangehender Classification der Sugethiere und Vgel. J.C. Cottaschen Mnchen, Stuttgart, Tbingen 354 pp. Webb, S.D. 1985. Late Cenozoic mammal dispersals between the Americas. In: Stehli FG, Webb SD, (eds.) The grea t American biotic interchange. New York: Plenum Press. pp 357 386. Webb, S. D., and S. C. Perrigo. 1984. Late Cenozoic vertebrates from Honduras and El Salvador. Journal of Vertebrate Paleontology 4:237 254. Weinelt, M. 1998. Ocean Drilling Stratigraphy Ne twork, Plate Tectonic Reconstruction Servic e. Available at www.odsn.de/odsn/services/paleomap/paleomap.html#Form Accessed January 29, 2009. Williams, E. 1950. Variation and sele ction in the cervical central articulations of living turtles. Bulletin of the American Museum of Natural History 94:505 562.

PAGE 132

132 Williams, E. 1956. Podocnemis bassleri a new species of pelomedusid turtle from the Late Tertiary of Peru. American Museum Novita tes 1782:1 10. Whitmore, F. C., Jr., and R. H. Stewart. 1965. Miocene mammals and Central American seaways. Science 148:180 185. Wood, R. 1976. Stupendemys geographicus the worlds largest turtle. Breviora 436:1 31. Wood, R. 1984. Evolution of pelomedusid turtles; pp.120 135 in F. Lapparent de Broin, F. Jimenez, and E. Fuentes (ed.), Stvdia Palaeocheloniologica I. Ediciones Universidad de Salamanca. Wood, R. 1997. Turtles; pp. 155 170 in R. F. Kay, R.H. Madden, R.L. Cifelli, and J.J Flynn (ed.), Vertebrate paleontology in the Neotropics: The Miocene Fauna of La Venta, Colombia. Smithsonian Institution Press. Wood, R., and M. Diaz de Gamero. 1971. Podocnemis venezuelensis a new fossil pelomedusid (Testudines, Pleurodira) from the Pliocene of Venezuela and a review of the history of Podocnemis in South America. Breviora 376:1 23.

PAGE 133

133 BIOGRAPHICAL SKETCH Since 2001, Edwin has been enrolled in different research projects at different institutions. His first research experience was the thesis to obtain the Bachelor s degree in Geology, which was focused on taxonomy and the biostratigraphical value of a Late Eocene pollen species at the Middle Madgalena Valley Basin, Colombia, project funded by the Colombian Petroleum Institute. Then, he worked for two years at the C olombian Petroleum Institute as a junior geologist developing projects on quantitative biostratigraphy, palinostratigraphy and nanofossils. In 2004, he did an important discovery of an excellently preserved turtle from the Early Cretaceous (Valanginian), f rom Colombia, which he had the opportunity to study together with Dr Eugene Gaffney at the American Museum of Natural History, New York, supported by the Lerner Gray Memorial Fund also from the same institution, resulting in a paper published in AMNH Novit ates journal in 2005. It was also during 2005 that he moved to Panama, to start his research on fossil turtles from the Neotropics, participating in the intership program at the Center for Tropical Paleoecology and Archeology from the Smithsonian Tropical Research Institute. During this intership experience he participated in several fieldwork campaigns in Panama, Venezuela and Colombia, particularly at the Cerrejn Coal Mine, place where he discovered the biggest snake so far known, recently published toge ther with his colleagues in Nature. Also from this Paleocene locality, he collected turtles, which constitute the research topic of his Master of Science.