BULLETIN OF THE FLORIDA STATE MUSEUM BIOLOGICAL SCIENCES Volume 7 Number 2 FOSSIL TESTUDININE TURTLES OF FLORIDA GENERA GEOCHELONE AND FLORIDEMYS Walter Auffenberg <0*3 -V 13.3*f='5% I . * 185% * UNIVERSITY OF FLORIDA Gainesville 1963 Numbers of the BULLETIN OF THE FLORIDA STATE MUSEUM are. published at irregular intervals. Volumes contain about 800 pages and are not necessarily completed in any one calendar year. WILLIAM J. RIEMER, Managing Editor OLIVER L . AUSTIN, JR ., Editor Consultant for this issue Rainer Zangerl Communications concerning purchase or exchange of the publication and all manuscripts should be addressed to the Managing Editor of the Bulletin, Florida State Museum, Seagle Building, Gainesville, Florida. Published 25 April 1963 Price for this. issue 60¢ FOSSIL TESTUDININE TURTLES OF FLORIDA GENERA GEOCHELONE AND FLORIDEMYS WALTER AUFFENBERGI SYNOPSIS: Recent paleontological work in Florida has made available a large series of fossil tortoises from one relatively small area. Two distinct apecies of tortoises of the genus Geochetone are represented in the khown Pleistocene localities of the area, Geochelone crassiscutata C Leidy) and G. incisa ( Hay). The only valid Pliocene species of this genus is Geochelone havi (Sellards). The Florida Miocene contains only Geochelone tedwhitei C WilliamS ) These four species seem to represent two main evolutionary lines. Both lines are recognized as distinct subgenera, Hesperotestudo and Caudochet!/s; the · latter,is new. G. incisa is, the last representative of Hesperotestudo, a line pos- sibly beginning with G. ami)hithorax (Cope ) of the Oligocene of Colorado, and continuing through G. osborniana (Hay) and related forms of the late Tertiary., The new subgenus Caudochetus is represented in the Florida Pleistocene by G. crassiscutata: The subgenus. can be traced from G. ligonia (Cope) of the Oligocene of Colorado through G. tedwhitei of the late Tertiary of Florida, and related species of middle and western North America. Floridemys nanus (Hay) is known only from the Florida Pliocene. Its relation- ships remain obscure. 1Walter Auffenberg is Associate Director of the Biological Sciences Curriculum Study, University of Colorado. Most of his researches have dealt with the fossil history of amphibians and reptiles. Earlier contributions to the Bulletin concern fossil Terrapene and the musculature of salamanders. Manus-cript submitted 26 August 1961.-ED. Auffenberg, Walter, 1963. Fossil testudinine turtles of Florida. Genefa Geochelone and Floridemys Bull. Florida State Mul, vol. 7, no. 2, pp. 53-97. 54 BULLETIN FLORIDA STATE MUSEUM Vol. 7 TABLE OF CONTENTS Introduction 55 Abbreviations and kacknowledgments ,. 56 Material examined ................ . 56 Anatomical r6sum6 .. . 58 Systematics .. 69 Genus Geochetone 69 Subgenus Caudochetvs 69 Geochelone orassiscutata 70 Geochelone havi . 78 Geochdone tediohitei . 80 Subgenus Hesperotestudo 81 Geochelone incisa ... 82 Genus Floridemus ,....._- 83 Floridemus nanus ....... ,88 Discussion . 87 Literature cited 96 1963 AUFFENBERG: . FOSSIL TESTUDININES 55 INTRODUCTION Fossil turtles, especially Pleistocene forms, have been so little studied that much remains to be learned of them. No single comprehensive study of tortoises, fosdl or Recent, has ever been attempted. The best analysis of the generic relationships within the group is presented by Williams ( 1950, 1952) and Loveridge and Williams ( 1957). Many of the early publications tend to obscure relationships of fossil and Recent forms under a profusion of new names, often on such frag- mentary individuals that identification of new material is almost impossible. This is particularly true in Florida, where six species have been described from Pleistocene deposits alone. As outlined by Loveridge and Williams ( 1957) the genus Geoche- lone seems to represent a natural group. The Recent range includes the Galapagos Islands, South America, Africa, Madagascar, and cer- tain islands of the Indian Ocean. The fossil record is fairly extensive; tortoises are well represented in the Tertiary and Pleistocene of North America and in the Tertiary of Europd, Asia, and Africa. In South America fossil tortoises date from the Miocene, and in the West Indies from the Pleistocene. The genus includes 13 subgenera ( Wifliams, 1950, 1952; Loyeridge and Williams, 1957). Five occur in the New Wdrld ( Hadrianus, Hes- perotestudo, Chelonoides, Caudochelys, and Monachelys). The only extant New World subgenus ( Chelonoides) is known from the Mio- cene to the Recent of South America, and the Pleistocene and Recent of the West Indies. Monachelys is a distinctive subgenus known only from the Pleistocene of the island of Mona, West Indies. So far as known, all fossil species of the genus from North America belong to the subgenera Hadrianus, Caudochelys, or Hesperotestudo. The first of these is considered to be the most primitive of the entire genus, and may be ancestral to the other New World groups. At present it is known only from the Eocene of North America. Hesperotestudo, the subgenus to which Geochelone incisa belongs, represents a. highly specialized line that differentiated early and is now extinct. The number of valid species is uncertain, but the group can be traced back into. the Lower Oligocene of Colorado. The subgenus Caudochelys ( described herein as new) is repre- sented in the Pleistocene, Pliocene, and Miocene of Florida, as well as in deposits of the Great Plains. It also can be traced to the Lower Oligocene of Colorado. Unlike Hesperotestudo, Caudochelys is not known from Asia. The subgenus differs from Hesperotestudo chiefly in that the dermal ossicles of the supracaudal area are not fused. 56 BULLETIN FLORIDA STATE MUSEUM Vol. 7 The following forms have been described from Florida. Miocene testudinine remains from Florida are known at present only from the Thomas Farm, Gilchrist County. Only one species is ~ known from this locality, Geochelone tedwhitei ( Williams, 1958). Only three testudinine turtles are described from Pliocene deposits of Florida. These are Geoche/one hayi, G. louisekressmani, and Flor- idemys nanus. It is shown later that G. louisekressmani is probably a synonym of G. hayi. Six testudinine turtles are described from the Pleistocene of Florida. These are Geochelone incisa, G. crassiscutata, G. sellardsi, G. luciae, G. ocalana, and G. distans. G. incisa remains a valid species as described. All the others are considered synonyms of G. crassiscutata. Another nominal testudinine genus must be considered in discussing the Florida material. Eupachemys obtusus Leidy ( 1877) is based on a single peripheral of a large turtle believed to be a testudinine from the Ashley River beds of South Carolina. These beds contain fossils of various ages, but most of the terrestrial vertebrates are Pleistocene. As the frag- ment lacks definitive characteristics, it cannot be assigned with cer- tainty to any particular genus. Leidy believed it represented a large aquatic emydine turtle. Hay ( 1908) refers it to Testudo C = Geoche- lone) on the basis of size and suggests it may belong to the same species Leidy ( 1889) described as Testudo crassiscutata. This inter- pretation is also followed in the present study. ACKNOWLEDGMENTS AND ABBREVIATIONS The foll6wing abbreviations are used when referring to collections: ACM-Amherst College Museum; AMNH-American Museum of Nat- ural History; FCS-Florida Geological Survey; MCZ-Museum of Comparative Zoology; UF-University of Florida Collections; UMMP -University of Michigan Museum of Paleontology; USNM-United States National Museum. I wish to thank the persons responsible for the collections in these institutions for the loan of specimens in their care. Special acknowledgment is due for the criticisms and sugges- tions made by Claude Hibbard, Ernest E. Williams, William Milstead, Clayton Ray, William J. Riemer, and Thomas Oelrich, and to the National Science Foundation for financial assisfance ( G-17618). MATERIAL EXAMINED This study is based on the recent discovery of much better material of Geochelone than has heretofore been available. This new material 1968 AUFFENBERC: FOSSIL TESTUDININES 57 includes many complete but isolated elements; several plastra, cara- paces, and complete shells, many associated with limb elements and vertebrae. Some specimens are so well preserved that for the Brst time it is possible to describe the skull and the limb and circumeloacal armor. I have examined all the type material from Florida ( except that of G. louisekressmani) as well as 18 complete shells, 18 complete plastra, and numerous isolated elements. The University of Florida Collections contains a Bne series of 12 complete or almost complete shells ( some with skulls) from chrono- logically equivalent Pleistocene localities near Haile, Alachua County, Florida ( Haile VIII A, IX, and XI). The series constitutes the largest and best sample yet assembled of any Pleistocene testudinine from a single locality in North America. Also in the University of Florida Collections are complete shells from scattered localities in Florida, and, a series of plastra from a locality near Reddick, Marion County„ Florida ( Reddick I A). In addition to- this excellent material, many scattered elements from other localities were examined. SPECIMENS OF G. CRASSISCUTATA AND G. INCISA EXAMINED Type, C. crassiscutata USNM 983, along Peace Creek, near Area(lia, Hardee Co., Florida. Type, G, incisa USNM 8821, near Ocala, Marion Co., Florida. Arredondo II, G. incisa UF uncat., two epiplastra. Eichelberger Cave B. G. crassiscutata UF 1457, peripheral. Haile II B. C. incisa UF 3005, buckler and caudal vertebrae. Haile VIII A. G. incisa UF 2986, 3029, 3132, partial shells; UF 3073, 3077, sheH and limbs; UF 3141, complete skeleton; UF 3258, 3857, bucklers, UF uncat., two :xiphiplastra ; UF 3228, partial shell and limb elements . G . crassiscutata UF 3133, two scapulae; UF uncat., scapula; UF 3028, 3151, 3226, 3623, shells and limbs; UF 3139, 3623, shells; UF 3146, plastron and partial shell; UF uncat., three epiplastra; UF 8248, partial plastron; UF 8797, humerus; UF 3825, 3830, xiphiplastra; UF uncat., six xiphiplastra; UF uncat., thigh arm6r. Haile IX A. G. incisa UF 3234, 3235, two shells with bucklers and -parts of skeletons. Haile XI A. G. incisa UF 3462, complete shell, buckler, and ~all limb elements; UF 3844, buckler. Hillsborough River State Park. G. trassiscutata uncataloged, large shell in collee- tion of Hillsborough River State Park. Itchtucknee River. G. inciss UF 2713, xiphiplastion. G. crassiscutata UF 2328, xiphiplastron; U.F 1994. epiplastron. 5 mi. W,Jupiter Inlet. G. crassiscutata UF 8006, xiphiplastron. Kanapaha C. G. crassiscutata UF 9907, scapula. Mefford Cave II. G. crassiscutata UF 6523, miscellaneous'parts.of shell. Melbourne. C. crass·iscutata, large complate shell in Amherst College Museum. Prairie Creek. G. crassiscutata UF 1816, parts of plastron. 58 BULLETIN FLORIDA STATE MUSEUM Vol. 7 Reddick I. G. incisa UF 4748, peripherals; UF 2545, epi- and xiphi-plastron. G. crassiscutata UF 2477, 2532, partial carapace (?crassiscutata); UF 2516, 2517, 3082, 3083, plastra; UF 2397, 2420, 2480, plastra with carapace parts; UF 2460, carapace parts; UF 2851, scaptila; UF 3784, feinur, UF 2475, cir- cumcloacal armor; UF 8091, xiphiplastron; UF 2598, peripherals; UF 2521, suprapygal and entoplastron. Reddick II. G. crassiscutata UF 9908, small shell with crushed skull and limbs. Rock Springs. G. cmssiscutata USNM 11311, parts of carapace and plastron of large specimen. Zuber. G. incisa UF 5020, epiplastron. Because of the Sne series of specimens from the Haile VIII A locality, I decided that the best approach was to study these relatively com- plete specimens intensively. Fragmentary specimens, isolated Hnds of reasonably complete individuals, and isolated single elements could then be interpreted in the light of information gleaned from this series. ANATOMICAL RESUME OF THE HAILE SERIES Even casual observation of the shells available from this locality leads one to the conclusion that two forms are represented. Shells of both forms occur throughout the deposit, and there is no evidence of morphological intermediacy. The degrees of plastral concavity show that males and females of both forms are present. Although one form apparehtly attains a much greater size than the other, all small indi- viduals can be readily separated inte two groups. These data are interpreted as being evidence of differences at the species level. The type specimens of previously described Pleistoc6ne species of this genus from Florida, in almost every case, can be placed in one group or the other. The earliest names for identifiabie fragments belonging to'these groups are Geochelone incisa ( Hay ) and Geochelone crassiscutata ( Leidy ). A comprehensive description of the specimens referred to each of these species among the Haile VIII A series is given below. The reasons for allocation of the particular names usedfor these two forms, and the disposition of remaining named Pleistocene forms will follow. In this series Bve complete or almost complete specimens are referred to G. incisa. while six are referred to G. crassiscutata. SHELL Specimens of this series placed in Geochelone incisa are much smaller than the largest placed in G. crassiscutata. See table 1. The largest specimen of G. incisa in the Haile series ( UF 8141) measures 241 mm in carapacial length. The largest complete shell of G . crassiscutata from the same locality ( UF 8623 ) measures 1210 mm in carapacial TABLE 1. Shell measurements ( in millimeters) 1963 A U FFE N B E R G : FO S S IL TE S TU D IN IN E S 59 C. incisa G. crassiscutata UF UF UF UF uF UF UF UF UF UF UF UF UF UF 3462 3141 3077 '3073 3029 2986 3235 3151 2480 2397 3139 3226 2426 3028 Carapacial length __ 241 231 228 224 216 212 327 - -- 192 180 __ 168 Plastral length 264 205 214 204 206 192 196 304 259 257 180 169 164 153 Entoplastral length 52 41 42 40,5 41 39 33 72 59 55 46 38' 34 39 Entoplastral width 46 36 36 36 40 40 34 71 57 55 39 39 31 32 Cular length 46 34 36 37 30 30 34 54 39 43 31 27 24 28 Humeral length 37 36 48 32 35 34 33 54 51 45 31 30 29 29 Pectoral length 11 18 12 10 7.5 12 11 19 20 14 9 7 12 13 Abdominal length 94 66 59 74 65 70 68 115 79 89 68 63 59 52 Femoral length 42 38 32 32 28 32 28 41 43 31 23 21 19 24 Anal length 16 10 9 8 12 7 12 17 22 21 9 13 19 12 Xiphiplastral notch depti 28 22 23 21 23 12.5 20 24 12 13 14 6 12 8 Xiphiplastral external height 28 23 22 21 29 19 18 24 21 21 16 14 13 12 Xiphiplastral length 33 35 35 30 35 29 32 51 43 45 33 30 33 26 Epiplastral length 51 45 50 41 49 39 40 68 60 51 29 32 32 31 Epiplastral width 47 44 43 49 40 39 39 63 57 49 38 34 31 33 Epiplastral lip thickness 36 29 28 27 31 25 21 35 31 31 17 22 19 21 Epiplastral lip length 36 33 40 31 37 39 ___ 45 36 36 27 23 19 21 Nuchal scute length 20 17 15 14.5 11 10 11 20 16 11 10 10 Nuchal scute width 886739 4.5 13 99 7 9 Pygal length 30 29 31 22 27 22 44 ·22 21 19 Pygal dorsal width 40 39 46 31 40 28 48 26 26 20 Pygal ventral width 27 23 28 19 24 19 36 19 15 14 Second suprapygal length 24 19 21 19 31 20 27 15 14 . 24 Spcond suprapygal width 62 46 55 41 57 44 45 24 22 34 60 BULLETIN FLORIDA STATE MUSEUM Vol. 7 length, though fragments of much larger specimens have been found. Specimens referred to in G. incisa are more rugose than similar-sized or even larger specimens of G. crassiscutata. Sulcal growth patterns are much more evident in the former, the central areas of the costals are more raised, and small bosses are developed in several parts of the shell, particularly on or near the peripherals. The latter are con- siderably thickened, particularly the posterior members. The plastron is also quite different in the two species. In G. incisa the xiphiplastral notch is much deeper ( fig. 1),, with the free tips more pointed. The epiplastral beak is proportionately thicker. In 1 13 - 1 Riddick I »*/4-~50-':1*WAZM*ABAVOllmWL-ZA~OR. PL IO CE NE PL EI ST O C EN E 12 1 Holli ZI[A crassiscutata incisa ~4 turgida rexroadensis AGE to IS 2.0 2.5 3.03.5 4.0 4.5 5.0 5.5 6,0 6.5 70 7.5 808.5 Median length tiphiplostron Deplh of xiphiplastral notch FIGURE 1. Specimens of Geochelone from the Pleistocene of Florida can be separated on the basis of the proportional depth of the xiphiplastral notch. The Great Plain5 species, G. turgida and C. rexroadensis, seem most closely related to G. incisa and G. crassiscutatd respectively. , most specimens the tips of the beak are slightly divergent, and a keel may be present on each of the two elements, both dorsally and ven- trally; and running out to the tips. The lateral parts of the hyo-, hypo-, and xiphi-plastral elements are thickened much more in G. incisa than in G. crassiscutata (fig. 2 ). PECTORAL GIRDLE AND LIMBS The scapulae of G. crassiscutata and G. incisa are similar. In G. incisa the procorocoid process is proportionally shorter than in G . crassis- cutata. The scapular portion of the glenoid cavity is proportionally 1968 AUFFENBERG: FOSSIL TESTUDININES 61 wider in G. incisa than in G. crassiscutata ( fig. 3). The included angle of the blade of the corcoid in G. incisa is usually smaller than in G. crassiscutata ( measurement made with the apex of the angle located 3.4 32- W e 30- .al 25- 24- 22 - , 41 -: 20 - 7.'8- _ ~ 4,8 i16- / 2% . 6,4 12- crass,scutoto remadensis incisal turdida P. LEISTOCENE PLIOCENE Median length xiphiplostrdn. E„ernal height wiphiploslon FIGURE 2. G. incisa and G. crassiscutata are readily separated on the basis of the prclportidnal external height of the xiphi- plastron. on the shaft). Holman ( 1959) suggests that two humeral forms in the Florida Pleistocene indicate the presence of two species of Geoche- lone. However, information obtained from complete specimens of all sizes of the two species indicates that the differences are usually subtle, and much ontogenetic and individual variation exists. The tuberosity in G. incisa seems higher than in G. crassiscutata in the few individ- uals of equal size which can be compared. The head is more rounded and set at less 9f an angle in G. crassiscutata. In early members of the Incisa group ( G. osborni(ma, G. orthopygia ), the humeral head is like that in G. crassiscutata. Lack of material prevents comparison of the ulna. and radius. PELVIC GIRDLE The lateral extent of the ventroposterior process of the ischium is greater in G. incisa than in G. crassiscutata. The extension is more acute and more twisted in the former. The anteroexternal process of the pubis tends to have a shorter articular surface in G. incisa, though there is age variation. Small. specimens have· a proportionally longer 62 BULLETIN FLORIDA STATE MUSEUM Vol. 7 surface. The curvature between this process and the midline is great- est in G. incisa ( fig. 8). f : .l·· 5 ...\ -Plailt·5*.1 < - -- ---1 A- ~ MI-. ' --2--1 Wild,U#A'Fyi'Lit_-r JXS~W' :AmE.......INWW 1 ' I. b- --' -A 1 '. 6;,Icap dri W ./bj FIGURE 8. Top left) Six fused and expanded caudal vertebrae of,a specimen of G. incisa ( UF 3077). Top right) Proportional width of the scapular portion of the glenoid cavity in ( left) G. incisa (UF 8077) and ( right) G. crassiscutata (UF 3151 ) 1968 AUFFENBERG: FOSSIL TESTUDININES 68 ~he distal end of the ilium is generally thinner and less twisted in end view in G. incisa than in G. crassiscutata (fg. 8). Small specimens of the latter have ilia that are less twisted, but as far as is known, they are never as straight as in G. incisa. The femur of G. incisa is more curved from the side; and the epi- condyles are developed more than in specimens of G. crassiscutata of similar size. The intertrochanteric fossa is relatively much larger in G. incisa. The angle between the head and the external surface of the greater trochanter is greater in G. incisa than in G. cras#icutata. In the largest available specimen of G. incisa ( UF 8462), the fibu- lare is strongly co-ossified with the astragalo-calcaneum. A slightly smaller specimen of the same species ( UF 3141) has this co-ossifica- tion on the left limb, but on the right the bones are only weakly united. In a similar-sized specimen ( UF 3073 ) they are not fused. In a single spedimen of G. crassiscutata in which the foot bones are preserved ( UF 8151 ) these bones are not fused. Whether such fusion occurs in larger specimens of G. crassiscutata is unknown. Young specimens of G. incisa probably lack this fusion. In UF 3462 ( C. incisa) tarsal 3 is missing. Tarsals 4 and 5 seem to be fused . Such fusion is not present. in the best hindfoot of G. crassis- cutata examined ( UF 3151). Metatarsal 8 and phalange III are fused in UF 8462. VERTEBRAE In G. incisa the terminal portion of the tail is expanded to support the heavy, co-ossified supracaudal ossicles ( fig. 8). The last 5 verte- , brae of G. osborniana have lengthened lateral processes. In G. ortho- pygia at least 6 vertebrae show these elongated processes. The last - 8 vertebrae are so modified in the specimens of G. incisa examined. Thus, the available material of this phyletic line suggests an increase from the Miocene to Pleistocene in the number of vertebrae support- ing the supracaudal buckler. The inference by Hay (1908:426) that the lengthened transverse processes of G. osborniana differ basically from the "distinct, true caudal ribs" of Gopherus polyphemim is apparently in error. In UF 3462, an excellent specimen of G. incisa, most of the processes are Middle left) Shape of the distal end of the ilium in (left) G. incisa ( UF 8077) and (right) G. crassiscutata (UF 8151) Middle right) The supracaudal buckler of G. incisa (UF 3077), Lower left ) The forelimb armor of G. indisa (UF 3141). r Lower right) Pelvic girdle of (left) G. crassiscutata ( UF 3151) and ( right) G. incisa ( UF 3141) 64 BULLETIN FLORIDA STATE MUSEUM Vol. 7 firmly ankylosed to the vertebrae. However, several of the vertebrae clearly show a thickened seam at the juncture of the base of the proc- ess and the vertebral body. Furthermore, a few of the vertebrae anterior to the buckler lack the processes, whereas adjacent members possesses them. Where they are missing the articular surface of the caudal rib is clearly evident. In a small specimen of G. incisa (UF 8141) all the vertebrae anterior to the buckler lack the ribs but have the articular surface. In the Vertebrae .supporting the buckler the ribs are all Brmly fused to the centra. The ribs are not fused to the centra in specimens of G. crassiscutata of the same size. However, in large specimens of this species these ribs are all firmly fused to the vertebrae. Thus the character seems to be associated with size, at least in the more anterior vertebrae. Complete cervical series of both G. incisa and G. crassiscutata are available. Most individual members of the series show no differences between the two species, but the second verfebra does show a few differential characters. In G. incisa the prezygapophysial articular surfaces are almost horizontal when viewed from the front, In C. crassiscutata the surfaces are at an angle of about 45 degrees. In G. incisa the anterior articular surfaces of this element are sometimes biconca*e ( 2 out of 4 complete series associated with complete shells). In G. crassiscutata the surfaces are biconvex in all 4 specimens exam- ined. Many more vertebra] series associated with shells are needed Pmt=LI=,=zatesm .i -' & B'Mul#LL~ * i -/. p ../...99//32// V - C A ill"IFT.F. 41'r=„il Ta.#ffr.2 5~3/hy, B+Z.·4:il/47..19*jllimilill'8 -xe~----tns-·-· D FIGURE 4. The skull of G. incisa ( UF 3141) 1968 AUFFENBERG: FOSSIL TESTUDININES 65 to validate this suspected proportional difference in shape of the articular surfaces. SKULL A complete skull of G. incisa, associated with the rest of the skele- ton and shell ( UF 8141) from Haile VIII A, is rather 16ng and narrow (fig. 4, and table 2), but less so than that in G. osborniana. The sides are almost parallel from the squamosal processes t6 the posterior border of the orbits. Near the middle of the orbits the maxil- laries converge to the snout and are slightly concave from above. From the side the dorsal surface is slightly concave behind the orbits, and strongly convex above and in front of them. The cutting edge of the maxilla is concave and finely serrated. There is a,low, wide notch with slightly larger serrations on either side anteriorly. The squamosal processes are slightly compressed laterally, less so internally. TABLE 2. Measurements· of the skull of G. incisa ( UF 3141) Snout to end of supraoccipital process 51 mm Snout to occipital condyle 42 Width at base of quadrates 35 : Width at posterior end of maxillae 33 Width of interorbital space 14 Width of nasaI opaning 12 Anteroposterior diameter of orbit 14 Maxillary edge to hjghest part of frontal 18 Medjan length of prefrontals 9 Median length of frontals 11 Width of jugal arch 7 The upper outer border rises above the upper surface of the paroccipital. - The height of the skull from the cutting edge of the maxilla to the upper surface of the frontals is contained in the length from the snout to the condyle 2.3 times. The anteroposterior extent of the otic region is 12 mm, and is contained in the length of the skull to the condyles 8.5 times. The palatal region is highly vaulted, the median fossa having a width of 13 mm. The narrowest part of the pterygoids is 11 mm. The alveolar surface of the maxilla is provided with 2 longitudinaI ridges and 2 grooves. The outer, larger ridge is separated from the cutting edge by the wider groove. The ridge is sharp, slightly ser- rated, and runs forward and inward to the premaxillary, is then directed anteriorly to the cutting edge of the same element. The inner ridge is higher than that in G. osborniana. It is broader posteriorly 66 BULLETIN FLORIDA STATE MUSEUN[ Vol. 7 I 4-r #,1 '* M 4 . f » 5 J f I 4 r/ FIGURE 5. The circumeloacal armor of G. crassiscutata ( UF 2475). than anteriorly and runs along the inner edge of the maxillary. There is no median prein:~x i 1 lary ridge. The skull of G. incisa is more like that of G. osborniana than that of G. gilberti, but the jugal arch is wider still than in the former. The otic region, like that in G. osbomiana, is less elongated than iii G. gilberti. The skull of G. impensa differs from G. inc·isa in the structure of the maxillary ridges, iii having a narrower jugal arch, a narrower palatal fossa, wider prefrontals, and longer squamosal processes. The skull of G. orthopygia is like that of G. osborniana, and thus similar to that of G. incisa. Iii G. orthopygia the skiill is wider than in C. incisa; the prefrontals are wider, and the frontals proportionally shorter. The cutting surface of the mandible is longer in G. inc:isa than in G. orthopygia. A small specimen of G. crassiscutata in the University of Florida Collections from Reddick II ( UF 9908) includes a smashed skull. Little can be determined from this specimen except that the frontal is proportionally longer than in G. incisci. LIMB ARMOR UF 8151, a half-grown G. crassiscutata shows some of the armor of both the front and hind limbs. The bottom of the hind foot is covered with fairly large, flat, bony plates. Those at the posterior periphery of the foot are much larger, and more spurlike than those over the rest 1968 AUFFENBERG: FOSSIL TESTUDININES 67 X.- : FIGURE 6. A diagramatic illustration of the posterior limb and caudal armor shown. in black in ( top) G. inciss and ( bottom) G. crassiscutata. of the foot. The armor of the front iimbs below the humerus is unknown in G. crassiscutata. However, a number of large, lanceolate, Hattened scutes found with sevefal large specimens suggests, that a part of the forelimb was armored. The thigh armor of G. crassiscutata is now well known. Two groups of ossicles from this region, associated with shells, are available. Both specimens are large. In each, several ossicles are fused together'to form a plate on the posterior surface of the thigh. The plate is composed of approximately 12 large, highly peaked ossicles ( fgs. 5,6). Around this plate the ossicles are usually smaller, though a few large spurs occur between the plate-and the base of the tail. The heel armor of G. incisa is unknown. The armor of the forelimb shows in UF 8141 ( fig. 3) and seems to be composed of proportionally larger ossicles than in G. orthopygia. In fact, the ossicles on the outer edge of the limb are proportionally larger than in most extant testu- dinines except for a few Old World forms. The largest ossicle is con- tained in the length of the ulna 2.5 times. As far as known the ossicles on the posterior surface of the thigh of G. incisa. are not fused as they are in G. crassiscutata. Instead, the area seems to be covered with large, loose, conelike ossicles. As in large specimens of G. crassiscutata, spurs are present between this area and the base of the tail. 68 BULLETIN FLORIDA STATE MUSEUM Vol. 7 CAUDAL BUCKLER One of the most diagnostic features of G. incisa is a fused series of dermal scutes immediately above the tail (fig. 8). The entire series comprises a subcircular patch composed of from 80 to 40 scutes. All the elements are quite firmly ankylosed in larger specimens. The scutes are pentagonal, hexagonal, or octagonal. Although there is no definite alignment, the central scutes are usually largest. Each scute near the center usually has a small spur, frequently off center. The scutes close to the edge are more conelike. On the basis of several complete specimens in which this buckler was found in place, in life it was obviously located near the median posterior edge of the carapace. The pygal and the adjacent carapacial bones possess a transverse depression on their inner surfaces. It is presumed that the plate was hinged to the underside of the carapace. along its upper edge ('fig.7). 4 V A B FIGURE 7. Cross-sectional views of the posterior Darts of the shells of (A) G. crassiscutata and (B) G. incisa showing the diagnostic features of the caudal area. The caudal vertebrae are closely appressed to the underside of the buckler, where the dompressed and elongate transverse processes s'ometimes fit into shallow grooves. A supracaudal buckler of fused scutes, with similarly modified ver- tebrae has been described in G. orthopygia and G. osborniand (Hay, 1908), and in G. turgida ( Oelrich, 1957), as well as in some less well known forms. In all these species the unit is composed of fewer and more simple elements. The tail of G. incisa was obviously short and probably did not show externally in life ( Hg. 8). The last few caudal vertebrae are much 1963 AUFFENBERG: FOSSIL TESTUDININES 69 A· B FIGURE 8. Presumed appearance, in life, of the tail of (A) G. crassiscutata and ( B) C. incisa. compressed caudocranially and do not extend beyond the posterior or ventral edge of the buckler. Evidently two groups Of tortoises of the genus Geochelone were present in Florida ddring the Pleistocene. Each group is represented by a sihgle species, G. incisa and G. crassiscutata. These,groups seem to correspond to the two lines recognized in western Nprth America- G. orthopygia-G. osborniana and G. ligonia-G. rexroadensis. These two groups are so distinct that it seems best to recognize them as two separate.subgenera. G. osborniana (Hay) (figs. 27,28) is the type species ·of the subgenus Hesperotestudo, including G . orthopygia ( fig . 29 ), G. turgida ( fig. 9), G. riggsi (fig. 31), G. incisa (figs. 22,28,24), and all of their relatives. G. uintensis, G. tedwhitei (fig. 21), G. rex- roadensis, G. crassiscutata (fgs. 10, 11, 12, 13), and all of their rela- tives seem t6 form a natural group here designated as a new subgenus. SYSTEMATICS -GENUS GEOCHELONE Caudochelys, new subgenus TYPE SPECIES. Geochelone crassiscutata ( Leidy) DEFINITiON. A Nearctic subdivision of the genus Geochelone with a narrow nuchal scute, entoplastron about as wide as long, pec- toral scutes reduced along the midline, and limbs and tail heavily armored with dermal ossicles. Ossicles above the tail are never fused to form a supracaudal buckler, and the vertebrae are norm'al, not 70 BULLETIN FLORIDA STATE MUSEUM Vol. 7 - uMMP 33= 040'9 33482 V ,15..689 .......f f FIGURE 9. A series of plastra of G. turgida, 'showing thE projected and keeled epiplastral beak dlso found in G. incisa. compressed or fused, without greatly elongated transverse processes. Eocene to Pleistocene of North America. Geochelone ( Caudochelys) crassiscutata Figures 10, 11, 12, 13 Testudo crassiscutata Leidy, 1889. - Testudo sellardsi Hay, 1916. ·Loomis, 1927; Williams, 1950. Testudo Zuciae Hay, 1916. Loomis, 1927. Testudo ocalana Hay, 1916. Testudo distans Hay, 1916. Gopherus Ocalana ( Hay) Williams, 1950. Geochelone settardsi ( Hay) Ray, 1957. Geochetone luciae (Hay) Ray, 1957 Geochelone ocatana ( Hay) Ray, 1957. TypE. USNM 988, the greater portion of the right epiplastron, a part of the right hyoplastron, parts of the posterior lobe of the plas- tron, the left femur and the left tibia (fig. 10). TYPE LOCALITY AND HORIZON. Peace Creek, near Arcadia, Hardee County, Florida; Pleistocene, Recent, or both, but probably only the fornner. 1968 AUFFENBERG: FOSSIL TESTUDININES 71 A *1 -5 B FIGURE 10. Type of Geochetone crassiscutata (USNM 983),(A) right epipl as- tron, (B) xiphiplastron. B A i f 7 1la / j f C D FIGURE 11. 5ubadult G. crassiscutata ( UF 8151) from Haile VIII A, A_lachua County, Florida. * t 72 BULLETIN FLORIDA STATE MUSEUM Vol. 7 A t. \(<-j ·- , 15 1/ MA- 1.- 11dll 2 -1 - A-4 1 % . J -4 C \*09j / -01-7 FIGURE 12. Juvenile G. crassiscutata (UF 8189) -from Haile VIII A, Alachua County, Florida. : i t i 4m \A A j '# 01*S/ j B FIGURE 18. Plastra of subadult specimens of G. crassiscutata from Reddick I A, Marion County, Florida. (A)UF 2517,(B)UF 2480. 1968 AUFFENBERG: FOSSIL TESTUDININES 78 DIAGNOSIS. A species of giant Pleistocene testudinine, differing from G. incisa in. being larger, having. a thinner, less rugose shell, a shallower xiphiplastral notch, a less well developed and simpler epi- plastral beak, a longer tail, and, in large specimens, a small fused patch of dermal ossicles on the posterior surface of the thigh. It also differs in lacking a ca-udal buckler. The major differences between the fwo species of Florida Pleisto- cene tortoises are: G. incisa G. crassiscutata 1)_ Carapacial lerigth to .231 mm Carapacial length to at least 1150 mm 2) Supracaudal ossicles fused into Supracaudal ossie:les not fused into a a rounded caudal-buckler caudal buckler ( figs. 5,61 ( figs. 3; 6) 3) Caudal vertebrae fused, Caudal vertebrae not fused, not compressed antefoposteriorly, noticeably compressed, transverse transverse processes attenuate processes not as attenuate (fig. 3) 4) Shell, and particularly the Shell thinner proportionally, smoother peripherals, thicker prOportionally, rugose ( particularly in males ) 5) Xiphiplastral notch deep, Xiphiplastral notch· less deep, included included angle lesg obtuse angle.more obtuse ( fig. 1) (Rgl) 6) Epiplastral beak thicker, Epiplastral beak smaller, thinner, frequently strongly keeled rarely keeled, and then only slightly 7) Xiphiplastron proportionally Xiphiplastron proportionally lower higher at outer edgeof at outer edge of hypohyoplastral hypohyoplastral suture ( fig. 2) suture (fig: 2) 8) Width of first.suprapygal Width 6f first suprapygal less greater (4.14) (fig. 14) 9) Entoplastron narrower (fig. 15) Entoplastron wider ( fig. 15) 1 1 1 1 inciso crossisculata 1 I turgida I11lIIlI 3.5 -4.0 4.5 50 5.5 6.0 6.5 10. 8 80 Median length plastron . Widlh first supropygol FIGURE 14. Proportional width of the first suprapygal in G. incisa, G. crassis- cutata, and G. turgida. 74 BUI,LETIN FLORIDA STATE MUSEUM Vol. 7 incisa crassiscutata turgida 1 1 1 1 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Plastral length Width entoplastron FICURE 15. Proportional width of the entoplastron in G. incisa, G. crassiscutata, and G. turgida. Both species exhibit considerable variation in many of these charac- teristics, usually associated with sex or size. However, some evidence in G. crassiscutata ( in which small series are available from several localities) suggests that minor chronological or geographical differ- ences occur in certain shell ratios (fg. 16). The two available speci- mens of G. incisa from Haile X, Alachua County, have caudal bucklers with ossicles considerably more flattened than in any specimen from Haile VIII A, Alachua County. If valid, such differences are prob- ably indicative of chronological rather than geographical differentia- 50 - 45- / 40 - --------O 0 . Le ng th ,o f ep ip la st ra l lip m m ) 25- 20- 255/ Reddlck IA . 0 9 Halle ZIIA 15- 1„1,1,11 0 ' 30 35 40 45 50 55 60 65 70 1-ength of epiplastion (in mm ) FIGURE 16. An example of minor differences in shell ratios in two populations of G. crassiscutata. 1968 AUFFENBERG: FOSSIL TESTUDININES 75 Mon, as the localities are close to one another. Until larger series are available from additional localities such presumably minor differences cannot be critically evaluated. Geochelone crassiscutata (Leidy, 1889) is described on the basis of fragments of three individuals collected from the shoals of Peace Creek. The original collection of specimens made.by J. Francis Le- Baron was numbered 988 to 986 in the United States National Musdum. As Hay ( 1908') restricted the type material to that figured by Leidy, the type number should be USNM 983 (the number of the figured xipiplastron, etc.), rather than USNM 985, as stated by Hay. Most of the nontype material of the same lot ii catalogued as USNM 6780. - The type material includes a large part of the right epiplastron ( fig. 10 ), a part 6f the right hypoplastron, parts of the posterior lobe of the plastron, the left femur, and the left tibia. These remains suggest the complete turtle probably had a carapacial measurement of ap- proximately 1520 mm. Hay referred to the same species a number of other fragments of large testudinine turtles in the United States National Museum and the American Museum of Natural History. The preserved elements, particularly the xiphiplastron, suggest that this species represents the second major phyletic line of Geochelone in the Pleistocene of Florida, being closest to the Miocene G. ted- whitei (. Williams). f-- FIGURE 17. Type of Geochetone set- lardsi (USNM 8817), a left xiphiplastron. Geochetone sellardsi ( Hay, 1916 ), is based on part of a xiphiplastron and parts of a carapace 6f a large testudinine.turtle ( Bg. 17) collected at Vero, Indian River County, Florida ( USNM 8817, formerly FGS V-1881). The diagnosis indicates that G. sellardsi is similar to G. 76 BULLETIN FLORIDA STATE MUSEUM Vol. 7 crassiscutata ( Leidy), but differs from it ( Hay, 1908) in having the outer face of the anterior part of the thickened xiphiplastral border Hat or concave instead of convex. The thickness of the anterior end of fhe border is contained in the distance to the bottom of the xiphi- plastral notch 8.6 times, instead of 8 times as in G. crassiscutata. Other characters mentioned by Hay as important in separating the two species are the extent of horn covering the upper surface of the xiphiplastral lobes, and the sculpturing of the lower surface of the lobes. Hay thought that the degree of sculpturing might vary between individuals. An almost complete specimen presumably representing this species was described by Loomis ( 1927). The characters chosen as diagnostic 6f the species are now all known to be variable within a sample of one species from one local- ity. The name G. sellardsi is here referred to the synonymy of Geo- chelone crassiscutata Leidy. Geochelone luciae (Hay, 1916) has as the type a part of the right hypoplastral element (USNM 8818, formerly FGS V-1807) of a large tortoise from Vero, Indian River County, Florida. Hay states that this species possibly attained the size of G. crassiscutata. He says it dif- fers from the latter in having a thinner wall along the border of the base of the posterior lobe. However, on the type the articular sur- face between the hypoplastron _and the xiphiplastron is broken away. Therefore, Hay's comparison of the shape of this area in G. luciae and G. sellardsi is not warranted. Furthermore, the element is one B FIGURE 18. Type of Geochelone ocalana ( USNM 8822), a right- epiplastron. (A) dorsal,-(B) ventral. 1963 AUFFENBERG: FOSSIL TESTUDININES 77 which shows considerable variation in thickness. Evidence shows that this variation is both individual and ontogenetic within a single species from a single locality. As no feature clearly separates G. luciae from G. crassiscutata, the name is here placed in the synonymy of the latter. Geochelone ocalana ( Hay, 1916) is the right half of an epiplastron ( fig. 18) collected from a fissure in a quarry of the Florida Lime Com- pany, Ocala, Marion County, Florida. The type is USNM 8822 ( for- merly FGS V-4299) . According to Hay, the available material places this species closest to G. crassiscutata but differs from it in having a thicker epiplastral beak. The thickness of this element is extremely variable and almost useless in identiBcatidn of Geochelone remains from Florida. A hypoplastral and Brst pleural from the same locality were provisionally placed in this species by Hay. Williams ( 1950) incorrectly placed the species in the genus Gopherus, but indicated that the material was so fragmentary that positive identification was impossible. The present study indicates that G. ocalana is best placed in the synonymy of G. crassiscutata. FIGURE 19. Type of Geochelone distans ( USNM 8819), an entoplastron. Geochelone distans ( Hay, 1916 ). A complete entoplastron ( USNM 8819, formerly FGS V-4289) is the type (f® 19). It was collected at pit 5 of the Florida Lime Company at Ocala, Marion County, Florida. The distinguishing feature of the element is that the pectoral scutes extend anteriorly onto the entoplastron, whereas in nearly all other species these scutes have their anterior border just behind this ele- ment. The xiphiplastron is clearly from a large specimen, presumably possessing a thinner plastron than that in the type of G. crassiscutata. Hay was unable to compare the element with that of any other Pleis- 78 BULLETIN FLORIDA STATE MUSEUM Vol. 7 tocene species from Florida, except G. ocalana, from which it differs considerably. The species G. distans is here tentatively placed in the synonymy of G. crassiscutata on the basis that the presumed diagnostic character occurs·as a variant in several species of tortoises. Geochelone ( ? Caudochelgs) ha!/i Tedudo hoyt Sellards, 1916. Testudo louisekressmani Wark, 1929. Gopherus hal/i ( Seilards) Williams, 1950, Ray, 1957. TypE. USNM 8815 C formerly FGS V-5001) a large portion of the posterior parts of both the carapace and plastron. TYPE LOCALITY AND HORIZON. Amalgamated Phosphate Company pit, Brewster County, Florida; Pliocene, Bone Valley Gravel formation. DIAGNOSIS. A large species of Pliocene testudinine turtle differing from the dwarf Florideings nanus in being much larger, and in having the gular sulcus at an angle with the midline of the plastron, rather than perpendicular to it. The xiphiplastral notch is deeper and more acute than in G. crassiscutata. The exterior wall- of the hypoplastron of G. hayi is vertical, while in G. crassiscutata it slopes inward. The carapace of the type of G. hayi, though larger than that of G . crassis- cutata, is much thinner. Geochelone hagi represents a fairly large testudinine, estimated to have been about 1500 mm in length. The second neural is octagonal, FIGURE 20. Rear portion of the cara- pace of the type of Geochetone ha!/i ( USNM 8815) showing the correct shape of the second suprapygaI. 1963 AUFFENBERG: FOSSIL TESTUDININES 79 and neurals 4 to 8 are hexagohal. The proximal end of the second pleural is slightly reduced in width and is in contact with the second neural only, while the third pleural touches neurals 2,8, and 4. The first suprapygal is large and. rests on the eleventh maPginal and the pygal. The second suprapygal is considerably reduced ( Sellards's [1916] figure of the type showing the posterioF Portion of the cara- pace is in error, as the second. suprapygal is wider than originally illustrated [fig. 20] ). The species G. hayi was recently placed in the genus Gopherus (Williams, 1952). This interpretation is probably incorrect and G. ha!/i is here referred to the genus Geochelong. and tentatively to the sub- genus Caudochelys. Geochelone louisekiessmani ( Wark, 1929). This species is described on the basis of a disarticulated partial shell, including ~arts of the right and left halves of the epiplastron, fragments of the hypoplas- tron, a humerus, a pygals several fragments of the xjphiplastron, marginals, pleurals, and vertebral scutes. All this material was col- lected from pit 5 of the Amalgamated Phosphate Mining Company near Brewster, Florida. Stratigraphic data are lacking, though the spe- cimen is thought to have come from the Bone Valley Gravel for- mation. Many opinions on the age and manner of deposition of this forma- tion are published. This apparently is because of the complexity of the lithologic units. Most workers now agree that both Miocene and · PIiocene vertebrates are represented in its fauna; almost all the marine forms are considered Miocene, and the terrestrial species Pliocene: Of particular importance is the fact that the same mine has also produced many remains of Mastodon and Chlamgtherium ( Wark, 1929), both Pleistocene genera. Local deposits of Pleistocene age apparently occur throughout this entire region; normally as shallow lenses near the top of the section and probably representing stream or pond deposits. G. lot,isekressmani conceivably could have origi- nated from these superficial beds. It is here placed in the Pliocene, but with reservations. The type is a poor reconstruction of a poor specimen. Wark states that the carapace is represented by the "right forepart." Unfortu- natejy the type could not be located, but examination of the pub- lislied figure suggests that it is the rear part, embracing parts of both the right and left sides. The reconstructed sulci in Wark's Bgure are completely in error. The 4,pe was collected at the same mine and at the same time as 80 BULLETIN FLORIDA STATE MUSEUM Vol. 7 the, material described by Sellards as Testudo hayi. Sellards refers several other fragments collected from this mine to T. hayi. Wark makes no reference to the type or referred material of T. haYi. The original number of the type of G. louisekressmani and also that of T. hayi was FGS V-5001. Hence Wark's material was either part of the type material of T. hagi, or at least material referred to T. hayi by Sellards. Unfortunately, Sellards does not describe this additional material adequately enough to be positive. The only diagnostic characters mentioned by Wark are great Size, and lack of symmetry of the peripherals, pleurals, and neurals. The type material of G. hayi does not represent a small testudinine, though one considerably smaller than the type of G. louisekressmani Because nothing is known of the maximum size attained by G . hayi, the character is worthless. The lack of symmetry in various elements of G. louisekressmani can be explained by Wark's misinterpretation of the part of the Shell represented in the type material. In proper positi6n these elements not only seem quite symmetrical, as far as can be determined, but are similar to the same elements in the type of G. hayi. Therefore, the name Testudo louisekressmani Wark is here referred to the synonymy of Geochelone hagi C Sellards ). The University of Florida Collections has additional material referred to this speeies from several localities. This includes several isolated elements and fragments of the shell of an individual tollected at Haile VI A, near Haile, Alachua County, Florida; a few peripherals from the Camp 12 Mine, near Holder, Citrus County, Florida; and a few fragments from the MeGehee property, a few miles north of Newberry, Alachua County, Florida. None of these fragments gives any additional information on variability in the specific characters of · G. hagi. Geochelone ( Caudochelys) tedwhitei Figure 21 Testudo tedwhitei Williams, 1958. Geochelone tedwhitei C Williams) Ray, 1957. TypE. MCZ 2020, a complete plastron. TYPE LOCALITY AND HORIZON. Thomas Farm, Gilchrist County, Florida; Arikareean, Lower Miocene; Hawthorne formation. DIAGNOSIS. A medium-sized Miocene species of testudinine with gulars more triangular and the ends smaller than in G. crassiscutata. It differs from G. hayi in being smaller and in the shape of the supra- pygal. 1968 AUFFENBERG: FOSSIL TESTUDININES 81 REMARKs. Most of the following data comes from Williams ( 1953). As far as known, G. tedwhitei does not exceed 400 mm in plastral length. The length of the median sulcus 6f the abdominals is 6 to -7 3 B A FIGURE 21. Geochelone tedwhitei (A) Carapace. (B) Type plastron ( MCZ 2020). Both from Lower Miocene, Thomas Farm, Gilchrist County, Florida. times that of the pectoral scutes. The gulars are more nearly trian- gular and the anals smaller than G. crassisctitata. The Well developed nuchal scute reaches the anterior margin, and the costovertebral sulci are not deeply incised. The free margins are less reverted than in G. crassiscutata. Several plastra and an almost complete carapace · are available. G. tedwhitei is apparently most closely related to G. fan·i and G. ducatelli. Subgenus Hesperotestudo Williams TYPE SPECIES. Geochelone osborniana ( Hay). DEFINITION. A Nearctic subdivision of the genus Geochelone with a narrow nuehal scute, an entoplastron about as wide as long, reduced pectoral scutes, and limbs and ·tail heavily armored with dermal ossi- cles. Those above the tail are fused to f6rm a supracaudal buckler; the tail vertebrae are compressed, sometimes fused. The transverse processes of the caudal vertebrae are greatly elongated. (?) Eocene to Pleistocene of North America, and "Tertiary" of Asia. 82 BULLETIN FLORIDA STATE MUSEUM Vol. 7 Geochelone ( Hesperotestudo) incisa Figures 22,28,24 Testudo incisa Hay, 1916. Gopherus incisa (Hay) Williams, 1950. Geochetone incisa (Hay) Ray, 1957. TypE. USNM 8821, a complete left xiphiplastron ( fig. 22). TYPE LOCALITY AND HORIZON. Pit 5,. Florida Lime Company, Mar- ion County, Florida; Pleistocene. FIGURE 22. Type of Geochelone incisa ( USNM 882_1 ), a left xiphiplastron. DIAGNOSIS. A species of Pleistocene testudinine differing from G. crassiscutata in being smaller, more rugose, with a thicker shell, some- times with bosses on some of the peripherals and on the bridge; the , epiplastral beak is proportionately thicker, frequently with a keel above and below, and with acute tips that are more divergent than in G. crassiscutata,· the xiphiplastral notch is deeper, the enclosed angle being less than in G. crassiscutata; the entoplastron is proportionally longer in G. incisa; the interanal sulcus is shorter; the buttress at the outer anterior edge of the xiphiplastron is proportionally much higher. In life the tail ·may not have been noticeable externally; the last few caudal vertebrae are shortened and fused; the transverse processes are larger than in G. crassiscutata., Ossicles of the posterior surface of the thigh are not fused into a single unit, and a number of ossicles above the tail are fused into a plate that is absent in G. crassiscutata. 1968 AUFFENBERG: FOSSIL TESTUDININES 83 4.lif ·· 1 .1-2 1.1 \E-\ k:-6 \ i XKX\6/ /- - 1"- 3-. .5.-U,:fi-43\ / CT.VA1- 1 9,]sy\»4 -'~544\\V . r» 47-~83 + - -* - + 1 es. ~ #' C '. '/i A B .T i / 14&,/2.14,/.--- /\U.4/ T- \ rK»~ 141 A \ - \ 1/ f / 4 S li/< 3 9/6 /'It.1/ /\ f \ f«,7 7$«~'~~~ 1 )2; t( D FIGURE 28. Shell of adult d mcisa ( UF 8029 ) from Haile VIII A, Alachua County, Florida. (A) lateral, (B) posterior, (C) ventral, (D) anterior. The type is from the same locality as the type of G. ocalana. The diagnosis emphasizes the thick, heavy nature of the xiphiplastron. It is provided with a deep notch posteriorly. The terminal processes are acute, and the anal scute is very short at the midline. This species is distinct and valid, and can now be completely diag- nosed. A number of complete shells, some with limbs and even limb armor, are available from several localities in Florida. It obviously belongs to the Osborniana-Orthopygia-Turgida line, and represents the previously missing Pleistocene representative postulated by Williams in 1958. GENUS FLORIDEMYS Floridemys nanus Figure 25 Bystra naha Hay, 1916. Fbridemys nanus ( Hay) Williams, 1950. Williams, 1952, Loveridge and Wil- liams, 1957. TypE. USNM 10247, a complete shell of an adult specimen of a distinctive dwarf species of Pliocene testudinine, crushed on the left side of the carapace, and pushed dorsally so that in side view it looks B 84 B U LLE T IN F LO R ID A S TATE M U S E U M Vol. 7 O F 3 0 7 7 2, i ., 1 A C FIGURE 24. Selected views of an adult specimen of G. incisa ( UF 3077) from Haile VIII A. The posterior view of the carapace (C) illustrates an abnormality in the scutellation ( a pair of supracaudal scutes) which. is apparently fairly common in this species. 1963 AUFFENBERG: FOSSIL TESTUDININES 85 considerably higher than it probably was in life. Originally the type was in the collection of Dr. Henry G. Bystra of Brooksville, Florida. TYPE LOCALITY AND HORIZON. Collected from a hard-rock phosphate mine, near Holder, Citrus County, Florida; Pliocene ( or possibly Mio- cene), Alachua formation. DIAGNOSIS, A small species of tortoise distinguished from all other described forms by a transverse gulohumeral suicus. Other important characters are the truncated epiplastral beak and the straight humero- abdominal sulcus that is directed posteriorly, from its median origin. DESCRIPTION. The type of Floridemys nanus has been illustrated by Hay ( 1916), though the Egure leaves something to be desired for tracing articulations and sulcal patterns. An examination of the speci- men under different types of light, and both wet and dry, disclosed a number of characters not determinable from Hay's illustration, and not mentioned in the description. The specimen is therefore redescribed. A dwarf species of testudinine turtlp known from a single shell of an adult female turtle 105 mm carapace length. The sulcal growth pat- terns are moderately developed. The pleurals are alternately wide and narrow at their distal ends. The frst pleural is iIi contact with only the first neural, which is subrectangular in shape. The second neural is hexagonal with the short lateral sides diredted posteriorly; the third ==46*4===b- iii· hi 2----5 '»,- .11 ») ..1'. C :1 4,(: \3 :..it\_ f·Al - \:/ r* 3 " VJ2 Lilllli...1-,ZI":Illl-2521 Nr--r \ -.1-- 1.Em.*f, Jr k B 'ZZ' f t iup=ri--