BULLETIN of the FLORIDA STATE MUSEUM Biological Sciences Volume 27 1982 Number 3 THE SQUAMATE REPTILES OF THE INGLIS IA FAUNA (IRVINGTONIAN: CITRUS COUNTY, FLORIDA) PETER A. MEYLAN UNIVERSITY OF FLORADA GAINESVILLE Numbers of the BULLETIN OF THE FLORIDA STATE MUSEUM, BIOLOGICAL SCIENCES, are published at irregular intervals. Volumes contain about 300 pages and are not necessarily completed in any one calendar year. OLIVER L. AUSTIN, JR., Editor RHODA J. BRYANT, Managing Editor Consultants for this issue: J. ALAN HOLMAN ROBERT M. SULLIVAN S. DAVID WEBB Communications concerning purchase or exchange of the publications and all manu- scripts should be addressed to: Managing Editor, Bulletin; Florida State Museum; University of Florida; Gainesville, Fl. 32611, U.S.A. Copyright © by the Florida State Museum of the University of Florida This public document was promulgated at an annual cost of $5000.00 or$5.00 per copy. Itmakes available to libraries, scholars, and all interested persons the results of researches in the natural sciences, emphasizing the circum-Caribbean region. Publication date: 30 June 1982 Price: $5.00 THE SQUAMATE REPTILES OF THE INGLIS IA FAUNA (IRVINGTONIAN: CITRUS COUNTY, FLORIDA) PETER A. MEYLANI ABSTRACT: The early Pleistocene,Inglis IA site has yielded the largest and most diverse sample of fossil squamates in eastern North America. About 4000 specimens, including 250 skull elements, represent 26 species of 5nakes, 4 species oflizards, and l amphisbae- nian. The herpetofauna is essentially modern; 21 of 26 snake species, 2 of 4 lizard species, andthe amphisbaenian survive in Floridatoday. Three Inglis IA snakes now extinct are Diadophis elinorae, Xenodontinae (cf. Dryinoides), and Regina intermedia (n. sp.). The extant species Opheodrys vernatis and Heterodon nasicus are present in the Inglis IA fauna and have apparently been replaced in Florida by Opheodrys aestivus and Hete- rodonsimus. The lizards includeOphisaunts ventratis, Sceloporusundulatw, Gerrhono- tus, and an extinct new species, Eumeces carri. The amphisbaenian is Rhineura cf. R. jtoridana The ecological requirements of modern counterparts of the fauna suggest that Inglis IA represents a community froma region of high pine with xeric hammock interspersed. This open high pine country was part of a savanna that extended around the Gulf of Mexico during the late Cenozoic. The savanna apparently served as a corridor for mammals going to and from South America, but it did not act in the same capacity for the squamate fauna, which remained essentially autochthonous. The composition of the squamate faunasuggeststhatthiseorridor maintained aconnection tothe west thathad previously existed farther north. SU MARIO : E 1 s itio denominado Inglis IA correspondiente al temprano Pleistoceno ha dejado la mis grande y mas diversificada muestra de fdsiles de reptiles en el Este de Norte Amarica. Cerca de 4000 muestras, incluyendo 250 elementos de craneo, represen- tan 26 especies de serpientes, 4 especies de lagartijas y 1 espeeie de anfisbaenido. La herpetofauna es esencialmente moderna: 21 de las 26 especies de serpientes, 2 de las 4 especies de lagartijas y el unico anfisbaenido son los sobrevivientes actualmente pre- sentes en Florida. Las tres especies de serpientes extintas pertenecientes al sitio Inglis IA son Diadophis etinorae, Xenodontinae (cf. Dryinoides), y Regina intermedia (n. sp.). Opheodrys vernalis yHeterodonnasicus son especies existentes que estan presentes en la fauna del Inglis IA pero que aparentemente, .en Florida, han sido reemplazadas por Opheodrys aestivus y Heterodon simus. Las especies de lagartijas incluyen Ophisaurus ventratis, SceloporUB undidatw, Gerrhonotw sp„ y una nueva especie extinta, Elimeces carrl. Los requerimientos ecoldgicos de la presente faunasuguiere que sus antepasados en el Inglis IA representaron una comunidad de regi6n de pino alto intercalada con formaci- ones xerofiticas de "hammock." Esta drea abierta con pinos altos constituya parte de savana que se extendi6 alrededor del Golfo de Mexico durante el tardfo Cenozoico. Esta savanaaparentementesirvidcomocorredor parael paso demamiferoshaciay desde Sur Am*rica, pero no actud con la misma eficacia para los reptiles, los cuales permanecieron esencialmente como aut6ctonos. La composici6n de la fauna de reptiles sugiere que este corredor, que previamente habia existido mhs hacia el norte, mantuvo conexidn con el Oeste. 'The author is a Laboratory Technologist in Herpetblogy, Florida State Museum, and a graduate student:in the Debattment df Zoology, both atthe Univdrsity of Florida, Gainesyille 32611. This paper was originally submitted in partial fulfillment for the degree of Master of Science at the University of Flbrida. MEYLAN, P. A. 1981. The Squamate Reptiles of the Inglis IA Fauna (Irvingtonian: Citrus County, Florida). Bull. Florida State Mus., Biol. Sci. 27 (3): 111- TABLE OF CONTENTS INTRODUCTION .3 ACKNOWLEDGEMENTS . 5 METHODS AND ABBREVIATIONS ........................ . 5 SYSTEMATIC PALEONTOLOGY ............................ . 6 Reptilia: Squamata........................................... .. 6 Suborder Lacertilia......................................... ... 6 Family Anguidae........................................ .. 7 Family Iguanidae .. 15 Family Scincidae............................................ ...18 Suborder Amphisbaenia ... 19 Family Rhineuridae ... 19 Suborder Serpentes........................................ ..20 Family Colubridae............................................. ....20 Subfamily Colubrinae................................... ...22 Subfamily Lampropeltinae.............................. ...29 Subfamily Natricinae..................................... ...39 Subfamily Xenodontinae............................... ...45 Subfamily Incertae Sedis......................................... ..55 Family Elapidae......................................... ...55 Family Viperidae .. 56 PALEOECOLOGY.......................................... ..61 Taphonomy................................................ ..61 Diagenetic Factors ......................................... .. 63 Collection Bias and Sample Size............................ ..63 Faunal Requirements...................................... ..64 Ecological Interpretation ................................... .. 64 ZOOGEOGRAPHY .. 67 LITERATURE CITED...................................... ..72 APPENDIX 1 LIZARD SKELETONS EXAMINED .......... .. 77 APPENDIX 2 SNAKE SKELETONS EXAMINED........... .. 78 APPENDIX 3 SELECTED COMPARATIVE MATERIALS... ...79 1982 MEYLAN: INGLIS IA SQUAMATES 3 INTRODUCTION Studies of fossil squamates in North America have focused largely on two general geographic regions, theHigh Plains and Florida. Work by Estes, Holman, and others has documented the rich fossil history for squamates in the High Plains states. Although Florida's fossil squamates received much attention during the 1950's and 1960's (Brattstrom 1953; Auffenberg 1955, 1956, 1963; Holman 1958, 1959a, 1959b, 1962; Gut and Ray 1963; Estes 1963), most of the material excavated in the last 15 years has not been described. Unreported fossil squamates are present in at least 8 localities: 1 Oligocene, 6 Miocene, and 1 early Pleistocene (Inglis IA). Of these, Inglis IA (a sinkhole in Citrus County, Florida) has the largest squamate fauna. Biostratigraphic correlations place it in the early Pleistocene (earliest Irvingtonian) (Webb 1974). When Auffenberg (1963) made his major survey of fossil snakes in Florida, no early Pleistocene sites were known. He recognized the potential value of an early Pleistocene site to the study of Florida's paleoherpetology, and his work showed that the modernization of Florida's snake fauna occurred between the middle Pliocene and mid- dle Pleistocene. The Inglis fauna documents these final steps in the evolution of Florida's snakes and lizards. Inglis IA is a rich fauna, with over 4000 elements that represent 31 squamate species. This is the largest sample of Squamata known from Florida. The fauna's large size makes it suitable for paleoecological interpretation and allows determination of the zoogeographical affini- ties of southeastern squamates in the earliest Pleistocene. Further- more, the Inglis IA fauna contributes to our knowledge of the evolu- tionary history of 31 taxa. Inglis IA was discovered by Jean Klein and Robert Martin in 1967, when excavations for the Cross Florida Barge Canal exposed it. After several small collections were made in intervening years, the site was collected in its entirety in 1974. The following description is based primarily on Klein's (1971) account. The site was exposed on the north bank of the now defunct Cross Florida Barge Canal in Section 8, R.16 E., T.17 S., Citrus County, Florida. The depositoccurred in a sinkhole in the Inglis member of the Ocala Group (or more simply called Ocala limestones) (late Eocene). The fauna accumulated in locally derived elastic sediments, which occupied a volume about 10 m by 20 m and 3 to 4 m deep. In January 1974, approximately 300 cubic meters of fossiliferous sands were removed. The sands were washed tbrough screens at the site and the recovered fossils taken to the Florida State Museum for study. Klein (1971) recognized six stratigraphic units within the sinkhole. 4 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 The lowest, termed the basal conglomerate, included a large quantity of bone fragments, many of which were waterworn and polished. He attributed the wear to the action of waters of a spring boil (see Paleo- ecology). The next four units varied from 1 to 2 m thick. They filled the bulk of thesinkhole and contained the majority ofthe fossils. Klein said the sediments were deposited during periods of high and low water tables. During high water low energy deposition within the sinkhole produced the deposition of clays, and during a lowered water table sand rapidly accumulated. The entire sequence was capped by a cemented silica sandstone. Problems in determining the age of Florida's sinkhole faunas are discussed by Webb (1974). The absence of physical stratigraphic con- trol is the major drawback. Because fossils accumulate as isolated deposits in sinkholes and caves, deposits of differing ages can occupy adjacent sinkholes in the same formation. Thus, superpositional data are not available and correlations of their contained faunas must be based wholly on biostratigraphic methods. In his review of the chronological framework for the Florida Pleis- tocene, Webb (1974) cited 11 important biostratigraphic indicator lineages, seven of which assist in determining the age of the Inglis IA fauna. The record of the cotton rat (Sigmodon) is perhaps the most valuable. Sig?nodon curtisi, the most common cricetine rodent in the fauna, is known from late Blancan or early Irvingtonian sites in the American West, including the Curtis Ranch Fauna of Arizona, the Kentuck Fauna of Kansas, and the Vallecito sequence in California (Martin 1979). Smilodon gracilis, a sabercat present in the Inglis fauna, first appears in the late Blancan or early Irvingtonian of North America. By the late Irvingtonian it is replaced by Smilodon fatalis (Webb 1974). Members of the armadillo lineageKraglievichia to Chlamytheri~m increase in size from the Biancan through the Rancholabrean (Robert- son 1976). The Inglis IA chlamythere is intermediate in size between late Blancan Kaglievichia and Irvingtonian and Rancholabrean Chlamytherium septentrionalis. The presence of Hemiauchenia rather than Palaeolama and ofPla- tygonus bicalcaratus rather than Platygon*s cumberlandensis in the Inglis fauna suggests a late Blancan or early Irvingtonian age (Webb 1974). The absence of Bison is further evidence that Inglis IA is pre-Rancholabrean. The freshwater turtle.Pseudemys is the only reptile that helped date Inglis IA. This genus was examined in detail by Klein (1971), who 1982 MEYLAN: INGLIS IA SQUAMATES 5 determined that the Inglis IA Pseudemys was a morphological inter- grade between the Blancan Pseudemys platymarginata and Irvingto- nian Pseudemys seripta petrolef Fossil squamates are too poorly known to be of any help in determining the age of Inglis IA. Based on the lineages discussed above, Webb(1974) placed the Inglis IA fauna in the earliest Irvingtonian. In particular the fauna from Inglis IA closely resembles that from Curtis Ranch in Arizona, which has been dated at 1.9 million years old. ACKNOWLEDGEMENTS The motivation to study fossil squamates and the guidance to see the study through have been provided by Walter Auffenberg. David Webb and Ron Wolff have also advised me, especially in the preparation of this thesis. To them I extend my thanks. Bruce J. MacFadden, Robert Sullivan, J. Alan Holman, and Rhoda J. Bryant also provided comments that improved this contribution. A number of people helped with specific aspects of this study and deserve recognition and thanks. Tom Van Devender, Richard Zweifel (AMNH), and George Zug (USNM) provided comparative material. Pat Srygley and Howard Kochman helped me with the use of packaged computer programs. Computing was done at the Northeast Regional Data Center of the State University System of Florida, located on the campus of the University of Florida in Gainesville. Ron Wolff and Donna Born Drakeprovided equip- ment and advice for the preparation of the included photographs. Esta Belcher pre- pared Figures 1, 3, 6, 9, 13, 14, and 15. Angela O'Brien and Kelly Howard cheerfully typed the final draft. Special thanks go to all of the people who have helped to assemble the Florida State Museum Herpetological Skeleton Collection, without which this study would have taken much longer. My wife, Anne, provided endlessencouragementand instruction in good English. She diverted time from her own studies to type the first draft of this thesis. METHODS AND ABBREVIATIONS IDENTIFICATION. - Most of the material reported here was identified by direct com- parison or univariate statistics (student's t-test). Diagnostic characters reported in the literature were used when possible. Additional distinguishing features were deter- mined from alarge sample of comparativeosteological material of lizards(Appendix 1) and snakes (Appendix 2). Multivariate statistical programs were found to be appropriate for identifying iso- lated vertebrae. Discriminate analysis was used to distinguish the vertebrae of snake species that show extreme similarity in vertebral form. Ratios of pairs of linear measurements from single vertebrae were used as the dis- criminating variables in the analyses. The use of ratios in multivariate statistical analysis has been questioned by Atchley et al. (1976) and requires justification. Ratios are used to remove the effect of individual vertebra size, and their employment has become standard in studies of snake vertebrae (Johnson 1955; Auffenberg 1963). How- ever, there are some statistical consequences of combining two variables to produce a third stemming largely from changes in coefficients of variation (CV) Of the denomina- tor variable (Atchley et al. 1976). In the present study such effects were minimized by using variables with approximately equal coefficients of variation. Heyer (1978) justified the use of ratios in a study of leptodactylid frogs by comparing results from analyses using ratios and linear measurements. In the pre5ent study, test comparisons of analyses using ratios and lihear measurements produced similar results for two apalyses, and better results using ratios in a third (Table 1). 6 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 Table 1. - Results of discriminantanalysis using measurements and ratios as varia- bles. (Poor discrimination is indicated by a higher percentage of reclassifi- cation.) Percentage of Reclassification Analysis With Ratios With Measurements Elaphe (2 species) 10.6 9.6 "Racers" (3 species) 8.8 7.4 V ipers (4 species) 10.1 24.3 Two types of analyses were used to maximize the information obtained from the discriminant analysis. The Statistical Analysis System (Barr et al. 1976) was used to identify the fossils because it includes a simple testdata statement that automatically classifies unknowns. It also produced less reclassification than a similar stepwise BMDP analysis (Dixon and Brown 1979). The BMDP analysis was used to provide information on the relative importance of each of the variables used in the analysis. TERMINOLOGY. - Terminology of snake bones follows: vertebrae, Auffenberg (1963); skulls, Bullock and Tanner (1966); compound and basiparasphenoid, Estes et al. (1970); basiparasphenoid, Underwood (1967); and compound and palatine, Marx and Rabb (1972). Lizard bone terminology follows: vertebrae, Auffenberg (1963); and skull, Oel- rich (1956) and McDowell and Bogert (1954). Measurements of snake vertebrae follow the methods of Auffenberg (1963). DETERMINATION OF MINIMUM NUMBER OF INDIVIDUALS (MNI). - For lizards, MNI can be determined using the same technique used for mammals, i.e. the most common right, left, or midline element is counted. Estimation accuracy can be increased by considering the size range of elements. The same technique can be used for snakes and amphisbaenians if skull material is common. If skull material is rare, which is typically the case, an estimate of MNI can still be made if the maximum difference in centrum length (or any other measure) from a single column of a given species is known. By subtractingthe centrum length of the smallestknown fossil of aspecies from the largest, a range in size for the fossil vertebrae is determined. The MNI is found by dividing the range_in sizeof the fossils by the maximum range in the column of thatsamespecies(and rounding up to the next whole number). Some precaution must be taken, because large gaps in the rangeof centrum lengthsof the fossils can producespurious results. Nosuch gaps are present in data gathered from Inglis material. This method probably greatly underestimates the number of individuals in large samples. The following abbreviations are used in the text: AMNH, American Museum of Natural History; UF, University of Florida/Florida State Museum; USNM, National Museum ofNatural History; CL, centrum length; CTW, cotyle width; NAW, neural arch width; NH, neural spine height; NLU, neural spine length at upper edge; NSB, neural spine length at base; POPR, postzygaphophyseal to prezygapophyseal length; PRPR, width across the prezygap6physes; ZW, zygosphene width; R, right; L,· left: MNI, minimum number of individuals; N, sample size; OR, observed range; X, mean; SD, standard deviation. SYSTEMATIC PALEONTOLOGY ORDER SQUAMATA OPPELL 1811 SUBORDER LACERTILIA GUNTHER 1867 Four species of lizards, representing three families, are present in 1982 MEYLAN: INGLIS IA SQUAMATES 7 the Inglis IA fauna. Comparisons of the fossils were made only to the lizard families now found in the New World (Anelytropsidae, Angui- dae, Gekkonidae, Helodermatidae, Iguanidae, Scincidae, Teiidae, Xantusidae, and Xenosauridae). Statements concerning lizard mor- phology for which no citation is provided are based on examination of the comparative material listed in Appendix 1. FAMILY ANGUIDAE COPE 1864 SUBFAMILY GERRHONOTINAE COPE 1900 cf. Gerrhonotus WEIGMANN 1828 FIGURE 1 A REFERRED MATERIAL. - UF 26409, 1 vertebra. DESCRIPTION. - The single vertebra has a subtriangular centrum 3.35 mm long and 2.1 mm wide. The centrum is gently rounded ven- trally with no haemal keel or subcentral ridges. The neural spine is low and rises from the neural arch at an angle of 25°. It projects posteriorly beyond the edge of the neural arch. Rib articulations are not strongly differentiated into parapophyses and diapophyses. COMPARISONS. - The absence of subcentral ridges indicates that the referred vertebra is not that of a teiid or iguanid. The centrum is longer and more triangular than thatof any gekkonids examined. It is more rounded ventrally than in Ophisaurus or Heloderma. The absence of the haemal keel in the fossil suggests that it is not a scincid or diploglossine anguid. The simple rib articulations eliminate the possibility thatXenosaltrus is represented. The fossil is very similar to Recent Gen*onotus multicari,zatus, but due to a lack of material is not assigned to species. DISTRIBUTION. - The genus Gen"honotus is found throughout west- ern North America from northern Mexico to southern British Colum- bia. This genus may be represented by material from upper Cretace- ous, upper Paleocene, and lower Eocene of the Midwest (Meszoely 1970). Later records include the Mio-Pliocene of Nebraska (Meszoely 1970), lower Pliocene of Kansas (Wilson 1968), and upper Pliocene (Blancan) of Texas (Rogers 1976). REMARKS. - Verification of Gerrhonotus in the early Pleistocene of Florida is required before any zoogeographical conclusions can be drawn. Its presence in Florida could be a result of the great radiation of anguids that apparently occurred in the Tertiary ofNorth America. 8 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 A v* B tammas~ee5-11~p-»*ar/'/£/r~1//'///- ·+ *915*/-" %~ 4 1 11 C D 1- 1 1 E F 436.lill' 1-I-1 6. r: ' G 1 1 H 11 11 1 J 11 11 Figure 1. - Limbed lizards and an amphisbaenian from Inglis IA: Gerrhonotu.q (A) vertebra, ventral view, X7. Rhineura ef. R. ./Zondana (B) vertebra, dorsal view, X10. SCdoporUS Undulatus (C) dentary, lingual view, X4; (D) maxilla, lingual view, X5; (E) frontal, dorsal view, X5; (F) vertebra, ventral view, X 10. Eumeces carri (n. sp.) dentary, (HOLOTYPE), X 10; (G)lingual and (H)labial views; (I) vertebra, ventral view, X 10; (J) maxilla, (PARATYPE), lingual view, X 10 (each scale = 2 mm). 1982 MEYLAN: INGLIS IA SQUAMATES 9 GENUS Ophisaetrus DAUDIN 1803 Ophisaurus ventralis (LINNAEUS) 1766 FIGURE 2 REFERRED MATERIAL. - UF 26410, 10 R and 11 L maxillae; UF 26411, 12 R and 11 L dentaries; UF 26412, 3 R and 1 L pterygoids; UF 26413, 3 R and 3 L compounds; UF 26414, 1 R and 1 L frontals; UF 26415, 2 parietals; UF 26416, 3 occiputs; UF 26417, 1 L exoccipital; UF 26418, 2 basisphenoids; UF 26419, 1 quadrate; UF 26420, 1 ilium; and 926 vertebrae including UF 26421, 24 cervicals; UF 26422,526 presacrals; UF 26423,22 sacrals; UF 26424,48 nonautotomic caudals; and UF 26425,306 autotomic caudals. DESCRIPTION. - The maxillae have a broad facial flange that con- tributes to the posterior border of the nares. Premaxillary processes are bifurcate; posterior processes are single. There are 13-17 teeth (X = 15.0 i 1.04) on 12 complete maxillae. The teeth of both the maxillae and dentaries are closely spaced and unicuspid; they have well deve- loped striations on the lingual surface. Thirteen complete dentaries possess 16-21 teeth (X = 17.7 * 1.44); all have a surangular notch that extends anteriorly well beyond the posterior end of the tooth row. A single patch of 30-37 teeth is present on the pterygoids. The compounds consist of an angular and supra-angular anteriorly, and a splenial on the lingual surface. An articular is fused to the supra-angular. There is a posteroventrally directed retroarticular process. A reduced mandibular foramen is located well anterior to the articular condyle. The frontals are paired and long, narrow, and pointed anteriorly. The parietals are longer than wide and lack anterolateral processes. The supratemporal processes are robust and nearly meet at their bases. The occiputs are elongate with the prootic bones extending ante- riorly. Crista prootica are poorly developed. The disassociated exoe- cipital is identical to those of the intact occiputs. The two fossil basisphenoids have short, robust basipterygoid pro- cesses which extend anterolaterally. The processes originate from the ventral surface of the basisphenoid. The quadrate is triangular in lateral view, and taller than it is thick. The ilium is reduced, consisting of a short, curved shaft with a circular distal expansion. The thoracie vertebrae are short and wide and have flat centra. The neural spines are broad and low. The sacral vertebrae have flat centra and irregular distal extremities on the transverse processes. The hae- mal arches of the caudal vertebrae are fused without,atrace of suture. Autotomy septa are present in all but the largest (most proximal) caudal vertebrae. 10 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 C D 1 1 11 E F Gl~ H r Ill I Figure 2. - Ophizaunis ventrati.9 from Inglis IA: (A) dentary, lingual view, X3; (B) maxilla, lingual view, X3; (C) pterygoid, ventral view, X3; (D) compound, lingual view, X3; (E) parietal, dorsal view, X3; (F) frontal, dorsal view, X3; (G) occiput, dorsal view, X3; (H) basisphenoid, fronto-dorsal view, X4; (I) quadrate, lateral view, X4; (J) ilium, lateral view, X4 (each scale = 5 mm). 1982 MEYLAN: INGLIS IA SQUAMATES 11 COMPARISONS. - The fossil dentaries and maxillae can be distin- guished from North American iguanids (except Phrynoso?na), helo- dermatids, teiids,Anniella, and xantusids (exceptKialtberina) by the closely spaced unicuspid teeth. They differ from Klauberina and gek- konids in having well developed striations on the lingual surface ofthe teeth. These striations and a more symmetrical tooth shape distin- guish the fossil dentaries and maxillae from Gerrhonotus. The dentar- ies differ fromPhrynosoma and all scincids in having the surangular notch extended anteriorly beneath the tooth row. Bifurcate premaxil- lary processes distinguish the fossil maxillae from Phrynosoma. The fossil maxillae differ from those of scincids in having a single posterior process. The fossil dentaries and maxillae have significantly fewer teeth than modern samples of Ophisaurus attenuatus and 0. COmpressus (t-test, P< .01) (Fig. 3). They have more teeth than modern O. ventra- liB, but the difference is not significant. The pterygoids of 0. ventralis, including the fossils reported here, are apparently unique among North American lizards in having a single patch of 15 or more teeth. O. attenuatuB has 10-20 teeth arranged DENTARIES OPHISAUROS COM'IRVSUS E-'Il) OPHISAURUS Hill-#-4ATTENUATUS n.8 OPHISAURUS ~ .fll/--IlVENTRALIS n=19 INGLIS IA 1FOSSILS n=13 MAXILLAE OPHISAURUS COMKR;SSUS --4.-) OPHISAURUS I -ATTENUATUS n=8 OPHISAURUS , n=17 INGLIS IA FOSSILS - | « n=12 I1ll11I11l11Il1 13 15 17 19 21 23 25 27 NUMBER OF TEETH Figure 3. - Number of teeth in the dentaries and maxillae of three Recent North American Oph€SaltrUS and the Inglis IA fossils. Mean, 95% confidence limits, and range are indicated. 12 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No..3 in one or two rows. Gerrhonotus and O. compressus have 10 or fewer pterygoid teeth. The fossil compounds differ from those of all North American lizards except Ophisaurus in having a small mandibular foramen located well anterior of the artieular condyle. The mandibular fora- men is reduced in Gerrhonotus but it is also more posteriorly located. There are no apparent differences in the compounds of the modern Ophisaurus species examined. Among extant North American lizards, only the Scincidae, Xantu- sidae, Helodermatidae, some diploglossine anguids, and Ophisaurus have divided frontals. The divided fossil frontals are as long and narrow as those of Eumeces and Ophisa1trus. As in these two genera, the frontals are more than three times longer than the width at their widest point. The frontals of Eumeces are not pointed anteriorly as they are in the fossils and in Ophisaurus. The frontal of Ophisaurus attentuatus makes up a larger part of the dorsal edge of the orbit than in 0. compressus or O. ventralis (Wilson 1968). In the Inglis fossil, the proximity of the prefrontal and postfron- tal sutures indicates that the frontals made up only a small part of their respective orbits, so the material could represent either 0. ven- tralis or 0. compressus. The referred parietals are longer and more narrow than those of all North American lizards examined, except Ophisaurus. The parietals could not be identi fied to species, but are referred to 0. ventralis on the basis of associated material. A characteristic feature of the occiput of all anguimorph lizards is the extended nature of the prootic bone, which extends anterior to the semicircular canal (McDowell and Bogert 1954). This condition, as well as the near absence of the crista prootica, distinguishes the fossil occiputs from those of all extant North American lizards, except Ophisaurus and Anniellc Anniella differs from the fossils in having reduced spheno-occipital tubercles on the basioccipital. There is no consistent interspecific variation in the occiputs of modern North American Ophisaunks. The fossils are referred to Ophisaunts ventra- liB on the basis of associated material. The fossil basisphenoids were only compared to material of species thought to occur in the Inglis fauna. They differ from Eumeces, in which the lateral wall of the abducens canal is continuous with the anterior edge of the basipterygoid process. They differ from Scelopo- rus, which has relatively longer, thinner basipterygoid processes. In Gerrhonotus these processes are directed anteriorly rather than anterolaterally. The quadrates of most North American lizards are D-shaped in 1982 MEYLAN: INGLIS IA SQUAMATES 13 lateral view. Only in scincids, helodermatids, and Ophisaurus are they triangular, as they are in the Inglis material. The quadrates of helo- dermatids are unlike the fossils in having a distinctive T-shape from above. Those of Eumeces are thicker than those of Ophisaarus. The quadrates of Ophisaurus are not diagnostic at the species level. The reduced size of the ilium of Ophisaltrus is distinctive. The fossil differs from 0. compressus, which has a triangular rather than round distal expansion. The expansion is round in 0. attenuatus and 0. ventralis. The body vertebrae of Ophisaurus differ from those of all other North American lizards in being short and wide and having a flat centrum and a broad, low neural spine (Etheridge 1961). The absence of articulating surfaces on the transverse processes of the sacral verte- brae indicates that these vertebrae belong to the limbless genus Ophi- saurus. OnlyAnniella is similar, but it has lower neural spines. The caudal vertebrae differ from all limbed forms in having haemal arches fused to the centrum withoutatrace of suture. Again, Anniella is similar, but it has a much lower neural spine (Etheridge 1961). In the present study the body vertebrae of modern Ophisaurus species were found to show more variation than Etheridge (1961) reported. He separated the three well known modern species in North America by differences in the length to width ratio of the centrum (CL/NAW) and/or by differences in the angle of the posterior border of theneuralspine. TwoMexicanspeciesareknownfromatotalofthree specimens, none of which is prepared as a skeleton. 0. compressus has distinctly narrower vertebrae than eitherO. ventralis or 0. attenuatus, OPHISAURUS COMPRESSUS F-3n=40 OPHISAURUS ATTENUATUS ~ n=66 OPHISAURUSVENTRALIS n=99 1 ---*--1 1 FOSSILS n=99 Il l lllltlllll11lIl1lllll 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 CENTROM LENGTH/NEURAL ARCH WIDTH Figure 4. - Vertebrae shape (centrum length/neural arch width) for three Recent North American Ophisaurus and Inglis IA fossils. Mean, standard deviation, two standard errors of the mean, and range are indicated. 14 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 but there is little difference between the latter two (Fig. 4). A similar figure for separating 0. compressus and 0. ventralis in Auffenberg (1955) includes a misleading error - the scale of 0.1 to 1.0 should read 1.1 to 2.0. Etheridge (1961) used differences in the angle of the poste- rior edge ofthe neural spine to separateO. attemiatus (45°-65°) from 0. ventralis (65°-84°). In the present study, this angle was found to be 65-85° in O. attenuatus and 70-90° in O. ventralis. The difference between Etheridge's data and those reported here may be explained in part by the geographical origin ofthe samples. Etheridge'sO. attenua- tus were from Kansas and Texas and represented only 0. attenuaks attenuatus (Etheridge, pers. comm.). The sample used in the present study included both 0. a. atten~atw from Texas and Kansas and 0. a. longicaudu.s from Florida. Although no clear difference exists between the vertebrae of 0. attemtatus, 0. ventralis, and the fossils, Figure 4 shows that the mean and the 95% confidence limits for the fossil body vertebrae are more similar to 0. ventralis than to O. attenuatus. The sacral and caudal vertebrae of Ophisaurus are not diagnostic at the species level except in one case. The caudals of 0. compressus are unique in lacking fracture planes. The presence of these planes in the fossil caudals indicates that 0. compressuB is not represented. Not all of the material referred to Ophisaurus ventralis is diagnos- tic. However, statistical variation observed in the fossils (see Table 2) is Table 2. - Diagnostic osteological characters for United States Ophisaurus species and Inglis IA fossils. Fossil 0. attenitatils 0. compressus O. ventratis Maxillary Teeth 15.0 i 1.04 16.8 * 1.14 21.0 i 1.58 14.4 * 0.91 (mean + one S. D.) Dentary Teeth 17.7 i 1.44 19.3 f 1.87 24.9 i 1.46 16.7 i 1.10 (mean i one S. D.) Pterygoid Teeth 30-37 10-20 10 10-25 (patch) (in 1 or 2 rows) (in 1 row) (patch) CL/NAW (body vertebrae) 1.33 i .123 1.51 =t .151 1,74 i .097 1.37 i .121 (mean E one S. D. ) Frontals in Orbit Small Large Small Small Distal Expansion of Ilium Round Round Triangular Round Fracture plane in Caudal Vertebrae Present Present Absent Present 1982 MEYLAN: INGLIS IA SQUAMATES 15 not greater than that in any one of the modern species examined, suggesting thatonly one species is present in the Inglis fauna. Analysis of seven characters in combination (Table 2) indicates that O. ventralis is the species present in the Inglis fauna. DISTRIBUTION. - Ophisaunts ventralis is found in the southeastern coastal plain and Piedmont from Louisiana to North Carolina. It is known as a fossil from the late Mioeene of Nebraska (Holman 1975), the earlyPliocene of Kansas(Wilson 1968), middlePlioceneof Florida (Auffenberg 1955),late Irvingtonian of Florida(Holman 1959a,b), and Rancholabrean of Florida (Auffenberg 1955; Holman 1958) and Mis- souri (Holman 1965). Ophisa~rus attenuatus has been reported from late Plio-Pleistocene localities in Kansas and Oklahoma (Etheridge 1961). Holman (1970) described Ophisaltrus canadensis from the upperMiocene of Saskatchewan and suggested that this species might be ancestral to O. ventralis and O. attemtatus. The Buda Local Fauna (Miocene: Arikareean) and Love Bone Bed(Miocene: Clarendonian) of Florida contain unstudied Ophisaurus material. REMARKS. - The Ophisaurus material from Inglis is the most extensive reported for the genus in the New World. It provides pre- viously unavailable morphological data on Ophisaurusventralis in the early Pleistocene. Minor differences between the Inglis 0. ventralis and a modern sample include higher average numbers of dentary and maxillary teeth, and substantially higher numbers of pterygoid teeth in the fossils. No changes were detected in the cranium or dermal roofing bones. An important consistency is seen in the condition of the ilium. It is evident that 0. ventralis had reached its present limbless condition by the earliestPleistocene and has not subsequently reduced the pelvic girdle. Speciation in North American Ophisattrus has been attributed to isolation of various stocks duringthe Pleistocene(MeCon- key 1954). The fossil record desetibed above shows that this is not the case for 0. ventralis and 0. attenuatus, which are both present before the Pleistocene. FAMILY IGUANIDAE GRAY, 1827 GENUS Sceloporus WEIGMANN 1828 Sceloporus undulatus (LATRIELLE) 1802 FIGURE 1 C-F REFERRED MATERIAL. - UF 26426,3 frontals; UF 26427,33 R and 36 L dentaries; UF 26428, 11 R and 10 L maxillae; UF 26432,2 humeri; UF 26433, 1 femur; and 53 vertebrae including: UF 26429,42 cervi- cals and thoracics; UF 26430,2 sacrals; and UF 26431, 9 caudals. 16 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 DESCRIPTION. - The frontals are single, unsculptured, narrow between the orbits, and have poorly developed crista cranii. The tooth-bearing elements have tricuspate teeth posteriorly. The cusps are not strongly developed. The dentaries are long and slender with Meckle's groove open lingually. They have an average of 3.57 teeth per mm. Maxillae have an average of 3.14 teeth per mm. The vertebrae have trapezoidal centra with moderately developed haemal keels and weakly developed subcentral ridges. Both sacral vertebrae are second members of disarticulated sacral pairs. They have posterolaterally directed tubercles on the posterior margins of the transverse processes. COMPARISONS.- AmongextantNorth American lizards, only igua- nids, teiids, gekkonids, xenosaurids, and Gerrhonotus have single fron- tals (Camp 1923). Unlike the fossils, crista cranii are well developed in the frontals of both gekkonids and xenosaurids. The fossil frontals are relativelyshorter than those of the teiid generaAmeiva and Cnemido- phoru# They are narrower between the orbits than the frontals of the iguanid Anolis carolimensis, butarevery similar to Sceloporus woodi and S. undulat148. The dentigerous elements possess typically iguanid tricuspate teeth. The three cusps are less developed than those of Le€ocephalus, and the central one is not as pronounced as in Anolis carolinensis. The teeth and the long slender shape of the dentary suggest a small Sceloporus. The referred vertebrae have centra sim- ilar in shape to most iguanid lizards. Subcentral ridges are less well DENTARIES SCELOPORUS 1--I.../....F-IIIHUNDULATUS n:8 INGLISIA FOSSIL S I n =IO SCELOPORUS I ~ I WOODI n=18 MAXILLAE SCELOPORUS UNDULATUS - | n:8 INGLISIA FOSSILS n=2 SCELOPORUS , --4/- IWOODI n=18 l lilli l lilli 1 3.0 3.2 3.4 3.6 3.8 40 4.2 TOOTH DENSITY IN TEETH PER mm. Figure 5. - Tooth density in the dentaries and maxillaeof RecentSceloporusitndutatus, Recent Sceloponts woodi, and fossil Scetoponis from Inglis IA. Mean, 95% confidence limits, and range are indicated. 1982 MEYLAN: INGLIS IA SQUAMATES 17 developed than in the teiidsAmeiva and CnemidophorUs, but they and the haemal keels are better developed than in any scincids, gekkonids, or anguids examined. The vertebrae of many small iguanids are sim- ilar, and the reference of these fossils to Sceloporus is based on the associated skull elements . All of the elements referred to Sceloporus undldatus closely resemble S. undulatus and S. woodi These two species can be distinguished on the basis of tooth density (Fig. 5). Sceloporus woodi has a significantly larger number of teeth on the dentary and maxilla than does S. undulatus or the fossils (t test, P < 0.05). The fossils do not differ in tooth density from S. undulatus. Therefore the frontals and postcranial material, along with the den- tigerous elements, are assigned to S. undulatus. DISTRIBUTION. - Sceloporus und*latlts is found throughout the southern two-thirds of the United States from extreme southeastern Nevada to New Jersey and Florida. It is known as a fossil from the Blancan of Texas (Rogers 1976), Irvingtonian of Maryland (Holman 1977a) and Texas (Holman 1969a), and Rancholabrean of Arizona (Van Devender et al. 1977), Florida (Brattstrom 1953), Georgia (Hol- man 1967), Missouri(Holman 1974), andTexas(Holman 1968b, 1969a; Gehlbach and Holman 1974). Theoldestknown sceloporine is from the upper Miocene of Saskatchewan (Holman 1970). REMARKS. - In their discussion of the evolution of the genus Scelo- porus, Larsen and Tanner (1975) reported S. undulatus to be a member of the virgatus group, the most complex and recently derived of three species groups. They suggested that the initial division of this group occurred when its progenitor was isolated in various refugia during the first glacial advance, and that with the first interglacial virgatus group stock migrated east and west to produce occidentalis, undulatus, and woodi. The appearance of S. undulatus by the Pleisto- cene indicates that somedivisionofthev*watus group had occurred by the end of the Pliocene. Jackson (1973) found Sceloporus woodi, a Florida endemic, to be more similar to southwestern members of the virgatus group than to Sceloporus undulatus undulatus. He cited the possible late arrival (Sangamonian = late Irvingtonian) of S. undulatus as additional evi- dence that it was a southwestern form which was isolated in Florida during the Pleistocene and gave rise to S. woodi. This new evidence of S. undulatus in Florida by the earliest Pleistocene renews the possibil- ity that it is ancestral to Sceloporus woodi 18 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 FAMILY SCINCIDAE GRAY 1825 GENUS E~meces WE:IGMANN 1834 Elimeces carri new species FIGURE 1 G-J DIAGNOSIS. - Eumeces carri is similar to several small Eumeces species (e.g. E inexpectatus, E. faBciatus). However, it differs from these and all Recent species examined in features of the teeth and the dentary. The dorsal margin of the dentary in all availableEvmeces (14 species) is straight. The anterior dilation of the fossil dentary(Fig. 1 G) could be a result of wear, but the presence of a labial ridge, which apparently occurs in no other skink, suggests that the anterior end of the dentary is modified. The teeth are also modified, being rounded or even blunt and unstriated rather than striated and weakly cusped as in most skinks. The high degree of lingual tooth erosion seen in the fossils could be a result ofdepositional wear, but the identical pattern of wear on both toothed elements suggests a predepositional feature. Such erosion may indicate very rapid tooth replacement. HOLOTYPE. - UF 26435 (Fig. 1 G, H), a left dentary from Inglis IA, Irvingtonian of Citrus County, Florida. PARATYPE. - UF 26432, (Fig. 1 J) a right maxilla. REFERRED MATERIAL. - UF 26434,11 thoracic vertebrae; 2 pairs of fused sacral vertebrae (tentative). DESCRIPTION. - The dentary is nearly complete and contains space for 17 teeth. The 10 preserved teeth are blunt, unstriated, and eroded lingually to about two-thirds their height. Meckle's groove is open along the entire length of the bone. In labial view the dentary narrows and then expands anteriorly. A well developed ridge occupies the middle of the labial surface at the same level as this expansion. The maxilla is nearly complete. The most anterior portion of the tooth row and the facial wing are missing. The maxilla has 15 or 16 tooth positions. The teeth are eroded lingually to about two-thirds of their height. The referred vertebrae are small (< 2 mm CL) and about twice as long as wide. Centra have nearly parallel sides and are rounded ven- trally. Haemal keels are broad and weakly developed. Subcentral ridges are absent. Neural spines rise. from the neural arch at an angle of 15-20°. The sacral vertebrae are about as wide as long and have long transverse processes. Each pair is fused at the centrum and at the ends of the transverse processes. COMPARISONS. - The thoracic vertebrae compare best to those of the scincid genusE*meces. The haemal keels are weaker than in iguanids, but stronger than in Gerrhonotus or Ophisaurus. The fossil vertebrae 1982 MEYLAN: INGLIS IA SQUAMATES 19 differ from those of the teiids, Cnemidophona and Ameiva, which have well developed subcentral ridges. They differ from gekkonid vertebrae, which tend to be relatively shorter and have neural spines that rise more vertically from the neural arch. Sacral vertebrae are fused only in scincids, xantusids, gekkonids, and some teiids (not fused in Cnemidophorus)..Because no other material was found that could be referred to the Teiidae, Gekkonidae, or Xantusidae, the sacral verte- brae are tentatively referred to this skink. The lingual erosion of the teeth in the maxilla and dentary suggests non-anguimorph, vertical tooth replacement. Among lizards with ver- tical tooth replacement, iguanids and teiids have some multicuspid teeth; Meckle's groove is closed in gekkonids and xantusids. The only remaining non-anguimorph New World families are the Scincidae and the Anelytropsidae. It is unlikely that the latter, a small mono- typic form known only from central Mexico, is represented. REMARKS. - The Inglis material represents a new and specialized species of skink. The study of its relationships will require additional comparative material. At least two individuals of this skink are represented in the fauna. SUBORDER AMPHISBAENIA GRAY 1825 FAMILY RHINEURIDAE VANZOLINI 1951 GENUS Rhineura COPE 1861 ef. Rhineuraforidana (BAIRD) 1858 FIGURE 1 B REFERRED MATERIAL. - UF 26438, 32 thoracic and 1 eaudal vertebra. DESCRIPTION. - The thoracic vertebrae(Fig. 1B) are dorsoventrally compressed and as long as or slightly longer than wide. The large pre- and postzygapophyses are adjacent to the neural arch and connected by an interzygapophyseal ridge. Zygosphenes and zygantra are absent. Neural spines project posteriorly when enlarged, but typically are reduced to a ridge. On either side of the neural spine is a series of raised striations. The vertebral centra are rounded below and have no haemal keel or subcentral ridges. The paradiapophysis is single. The single caudal vertebra is as above, but, in addition, has anterolaterally directed transverse processes and ventrally fused hemapophyses. COMPARISONS. - The vertebrae of rhineurids donotdiffer markedly between species. They are apparently unique among amphisbaenians in bearing numerous longitudinal striations on the neural arch (Ber- man 1976). 20 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 DISTRIBUTION. - Rhineurafloridana is at present restricted to the northern half of peninsular Florida. Rhineurid vertebrae from the late Irvingtonian (Holman 1962, 1959a) and Rancholabrean (Holman 1958) of Florida are referred to R. foridana. The Inglis material extends the fossil record for rhineurids in Florida to the beginning of the Pleistocene, but unstudied material in the Buda Local Fauna (Arikareean) indicates an early Miocene arrival of rhineurids in Florida. REMARKS. - The identification of fossil rhineurids in the literature has been based completely on skull characters. The absence of skull material from Inglis makes a positive identification of this rhineurid impossible, but because these fossils occur within the present range of Rhineura floridana and are identical to recent examples, the Inglis material is tentatively referred to this species. SUBORDER SERPENTES LINNEAUS 1758 Living members of the suborder Serpentes are divided into three infraorders: the Scolecophidia, the Henophidia, and the Caenophidia (Underwood 1967). Each of these has a diagnostic vertebral form (Holman 1979). The Scolecophidia are primitive burr6wing snakes. Their vertebrae are depressed and lack neural spines. The cotyle is oval in shape and the haemal keel and subeentral ridges are absent or poorly developed. The henophidians include boas and pythons. The vertebrae of this group have neural spines that tend to be short and thick, rarely as long as the neural arch. Paradiapophyses tend to be undifferentiated, and accessory processes are poorly developed or absent. The advanced snakes, or Caenophidia, have vertebrae with thin neural spines that are typically longer than those of henophidians; they are often nearly as long as the neural arch. Paradiapophyses are divided into parapophyses and diapophyses. Accessory processes tend to be well developed. On the basis of these characters, all of the snake vertebrae from Inglis IA are referable to the Caenophidia. Smith et al. (1977) applied the name Colubroidea to this group and recognized four families within it. Members of three of these families, Colubridae, Elapidae, and Viperidae, constitute the snakes of the Inglis fauna. FAMILY COLUBRIDAE OPPEL 1811 This study was facilitated by dividing American colubrid snake vertebrae into five artificial groups. Four of these morphological groups approximate natural phylogenetic assemblages of species rec- ognized by various authors. These are the Natricinae, Xenodontinae, Lampropeltinae, and Colubrinae. 1982 MEYLAN: INGLIS IA SQUAMATES 21 The Natricinae are easily distinguished by the presence of laterally compressed hypapophyses throughout the column (Underwood 1967). A second group has dorsoventrally flattened (depressed) neural arches. It approximates the Xenodontinae (Ophiinae of Dunn 1928) and includesHeterodon, which is clearly a xenodontine, andliarancia, which was placed in this subfamily by Dunn (1928) and Neill (1964). Underwood(1967) placed the genus in the Lycodontinae. Theextent of vaulting or depression of the neural canal is difficult to determine in very small snakes. They are assembled in a third, clearly unnatural group which includes Carphophis, Diadophis, Rhadinaeg Tantilla, Stilosoma, and Sonora. The remaining larger colubrid vertebrae are of twokinds(Fig. 6, Table3). Onegroup isshort and wide, with poorly developed epizygapophyseal spines and short laterally directed acces- sory processes. The other group is longer and more narrow, with epizygapophyseal spines that are moderately to well developed and long anterolaterally directed accessory processes. The former group includes Pituophis, Lampropeltis, Elaphe, Arizona, Rhinoceil~s, and Cemophora. This assemblage approaches Dowling and Duellman's (1974) Lampropeltiinae, a tribe of their subfamily Colubrinae. Smith et al. (1977) raised this tribe to subfamilial status on the basis that knowledge of snake relationships is too uncertain to recognize tribes. The group with long vertebrae remains in the Colubrinae and includes Coluber, Drymarchon, Masticophis, and Opheodrys. Table 3. - Vertebral characters for the separation of the Colubrinae and Lampro- peltinae (as recognized in this study). Lampropeltinae Colubrinae (8 species, 29 (4 species, 10 Vertebral columns, 223 columns, 79 characters vertebrae) vertebrae) CL/NAW less than 1.27 greater than 1.37 (except in (except in Druma,·chon) Centophora) Epizygapophyseal weak to absent moderately to well Spines (96.9%) developed (83.5%) Accessory shorter than longer than Process prezygapophyseal prezygapophyseal Length width (85.4%) width (64.6%) Accessory laterally anterolaterally Process. directed (65.1%) directed (87.4%) Direction 22 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 OPHEODRYS AESTIVUS 1 n=18 MASTICOPHIS , FLAGELLUM n=26 DRYMARCHON --4,-CORAIS n=13 CONSTRICTOR n=26 MEILTAUNOLeCUS |~| n= 26 LAMPROPELTIS ITRIANGULUM | n=20 LAMPROPEL.JIS ~» GETULUS | 1 n = 32 LAMPROPELTIS 141CALLIGASTER n=39 ELAPHE 2/4.ItVULPINA n=28 ELAPHE OBSOLETA I .1 1 n=32 GEULTAT~TA 1 1 n= 32 CEMOPHORA COCCINEA | -- 1 n=18 ARIZONA ELEGANS n=16 lllllllllllll 0.8 1.0 1.2 1.4 1.6 1.8 2.0 CENTRUM LENGTH/NEURAL ARCH WIDTH Figure 6. - Centrum length/neural arch width (CL/NAW) for lampropeltine and , colubrine snakes. Mean, 95% confidence limits, and range are indicated. The present study does not address the validity of these natural groups. The subfamilial names Colubrinae, Lampropeltinae, Natrici- nae, and Xenodontinae are used only to assist in the organization and identification of the fossils. SUBFAMILY COLUBRINAE OPPEL 1811 For the purpose of this paper, the Colubrinae is considered to include the racer-like snakes Coluber, Masticophis, Drymarchon, Opheodrys, and Salvadora. The vertebral centra of these snakes are nearly half again as longas they are wide, except inDrymarchon. The 1982 MEYLAN: INGLIS IA SQUAMATES 23 accessory processes are long and straight. Epizygapophyseal spines are better developed in this group than in other subfamilies (less so in Opheodrys and Salvadora). GENUS Coluber LINNAEUS 1758 Coluber COMBtrictor LINNAEUS 1758 FIGURE 7 A-D, TABLE 4 REFERRED MATERIAL. - UF 26360,334 vertebrae; UF 26361, 1 L pterygoid; UF 26362,2 Rcompounds; UF 26363, 1 L and 2 R dentaries. DESCRIPTION. - The referred vertebrae have long narrow centra that are nearly square across the zygapophyses (Table 4). Accessory processes are long, narrow and anterolaterally directed. Haemal keels are variable, but are typically narrow and straight with a slight lateral expansion just anterior to the condyle. Epizygapophyseal spines are well developed. The fossil left pterygoid is moderately large. It is broken at the posterior end of the toothrow. The pterygoid flange has a poorly deve- loped, laterally directed ectopterygoid process. The flange is slightly constricted between the ectopterygoid process and the base of the quadrate process. A ridge is present on the dorsal surface; it extends anteriorly beyond the ectopterygoid process. In the fossil compound bones the labial and lingual flanges are subequal in height, the lingual flange being slightly taller than the labial flange. All three fossil dentaries are broken posterior to the 12th tooth. Meckle's groove closes completely by the 8th or 9th tooth. The anterior end of these dentaries is very slightly curved. COMPARISONS. - The referred vertebrae are similar to those of Masticophis and, to a lesser extent,Drymarchon. In recent individuals of D. corais from Florida the anterior edge of the neural spine is bevelled. In Masticophis, Coluber, and some Mexican and Central AmericanDrymarchon, the anterior edge ofthe neural spine issquare. Thus the bevel, used as a standard diagnostic character for Drymar- chon, does not always hold. This problem, and the fact that neural spines from many of the fossils were broken, encouraged meto include Drymarcho?% in a discriminant analysis with Coluber andMasticophis. The analysis developed is independentof neural spine characters. It is based on 26 vertebrae from four Coluber constrictor, 16 vertebrae from two Drymarchon corais and 26 vertebrae from four Masticophis Bagellum. Ofthe68 vertebrae, 6 (8.8%) are reclassified by the analysis. The most important characters for the separation of these groups (ZW/NAW, PRPR/POPR, PRPR/NAW, PRPR/ZW) are shown in Table 4. 24 BULLETIN FL()RIDA STATE MUSEUM Vol. 27. No. 3 A B C f 1 Ill - D ~E F4 * *f - 5 m 1 1 111 11 Figure 7. - Fossil Co/uber and Mci.Nticophix from Inglis IA: Co/,(her constrictor (A) pterygoid, dorsal view, X5: (B) pterygoid, ventral view, X5: (C) compound, labial view. X2: (D) dentary, lingual view, X 3. Mcixtic·oph iNfInge//um (E) maxilla. occlusal view. X3: (F) compound. labial view, X2 (each scale = 5 mm). 1982 MEYLAN: INGLIS IA SQUAMATES 25 Table 4. - The four most important ratios for the discrimination of Coluber constric- ton Drymarchon corais, and Masticophidagetium (mean zE one stand- ard deviation). (ZW/NAW)1 (PRPR/POPR)2 (PRPR/NAW)3 (PRPR/ZWy Coluber constrictor Recent (N = 27) .98 i .04 1.04 + .04 1.79 i .10 1.83 i .11 Fossils (N = 334) .96 i .98 .99 zE .05 1.74 i .11 1.90 i .15 Masticophis fagellitm Recent (N = 26) .98 + .05 .98 + .05 1.81 + .05 1.84 zE .11 Fossils (N = 331) .95 i .08 .97 rt .05 1.80 E .11 1.89 i .13 Drymarchon corais Recent (N = 14) .81 i .02 1.12 i .06 1.65 i .07 2.03 i .06 Fossils (N = 29) .84 i .03 1.03 i .06 1.72 i .06 2.04 zE .06 'Zygasphene Width/Neural Arch Width 2Prezygapophysis to Prezygapophysis Width/Prezygapophysis to Postzygapophysis Length :'Prezygapophysis to Prezygapophysis Width/Neural Arch Width 'Prezygapophysis to Prezygapophysis W idth/Zygasphene W idth The fossil pterygoid is broken posteriorly. The presence of an ecto- pterygoid process (Fig. 7) indicates that it is not that of a viperid (see Figs. 5, 29, 30; Brattstrom 1964). The process in the fossil is less developed than that of the larger North American natricines and Farancia (App. 3-4). It differs from Heterodon in being laterally, rather than anteriorly, directed(Fig. 12B). From the large lampropel- tines, the fossil differs in having the dorsal ridge extended anteriorly beyond the ectopterygoid process. It is most similar to Colqiben Masti- cophis and Drymarchon, differing from the latter two only in having the ectopterygoid flange rise from the tooth row at a high angle and in having the ectopterygoid process more anteriorly located. Among larger North American snakes very few genera have sub- equal flanges on the compound bone (Marx and Rabb 1972). The lingual and labial flanges are subequal in the referred fossils, as they are in Masticophis and Coluber. In both Masticophis and Coluber the lingual flange is slightly the larger, and in both the extension of these two flanges increases with age. However, the lingual flange of C. constrictor is always more dorsally expanded than in individuals ofM. jZagellum of the same size. Based on this criterion, the fossils are referred to C constrictor rather than Masticophis. The fossil dentaries are most similar to Colltber and Masticophis, which have Meckle's groove closed completely between the 8th and 11th teeth (App. 3-6). They are assigned to the former because they do not show the more distinct curve at the anterior end of the dentary of Mazticophis. DISTRIBUTION. - Coluber conStrictor is found throughout the United States, except in the desert southwest. Its extensive fossil record must 26 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 be regarded with caution because of the difficulty involved in distin- guishing it fromMasticophidagellum. The oldest referred material is from the early Pliocene of Nebraska. Blancan fossils are known from Idaho (Holman 1968a), Kansas (Brattstrom 1967), and Texas (Rogers 1976). Irvingtonian specimens are reported from Arizona(Brattstrom 1955a; Lindsay and Tessman 1974), Florida (Holman 1959a), Kansas (Brattstrom 1967), and Maryland (Holman 1977a). Rancholabrean records occur in at least 10 states from coast to coast (see Holman 1981). Extinct, but possibly related iorms includeParaoxybelis,/Zoridalius from the early Miocene of Florida (Auffenberg 1963), Paracoluber storeri from the lateMioeene ofWyoming and Saskatchewan(Holman 1970), and Coluberplioagellu.s from the late Plioceneof Kansas (Wilson 1968). REMARKS. - It is apparent from the fossil record that Coluber constrictor is widely distributed by the Irvingtonian. The referred skull material from Inglis shows that the skull of this species has not changed over the last two million years. GENUS Drymarcho,% FITZINGER 1843 Drymarchon corais (BOIE) 1827 REFERRED MATERIAL. - UF 26367,29 vertebrae. DESCRIPTION. - The fossil vertebrae differ from those of Coluber constrictor only in being relatively wider. Neural spines, if present, are squared off anteriorly. COMPARISONS. - Long accessory processes and well developed epi- Zygapophyseal spines identify these as colubrine vertebrae. The ver- tebrae were identified as D. corais by the discriminant analysis des- cribed under Coluber constrictor. They differ from modern D. corais from Florida, in which the anterior edge of the neural spine is bevelled. Specimens from other parts of the range often have unbe- velled neural spines (UF 11467, 11782, 11784). DISTRIBUTION. - At presentD. corais is found from southern Texas to northern Argentina, with a disjunct subspecies in Florida, Ala- bama, and Georgia. It is known as a fossil from 10 middle or late Pleistocene and 6 late Pleistocene localities in Florida (see Holman 1981). Material from the Pleistocene of Texas that may represent this species has been lost (Holman 1969a). REMARKS. - Florida's Recent D. corais is clearly a relict of a once more widely distributed species. The Inglis fossils referred toD. corais show greater similarity to Recent populations from Mexico and Cen- tral America than to those in Florida. Apparently differentiation of Florida's D. corais populations has occurred since the early Pleisto- 1982 MEYLAN: INGLIS IA SQUAMATES 27 cene. This suggests thatDrymarchon became isolated in Florida after Inglis time. GENUS Masticophis BAIRD AND GIRARD 1853 Masticophis Bagellum (SHAW) 1802 FIGURE 7 E,F, TABLE 4 REFERRED MATERIAL. - UF 26364, 331 vertebrae; UF 26265,2 R, 1 L maxillae; UF 26366, 1 R, 2 L compounds. DESCRIPTION. - In most respects the Masticophis vertebrae fit the description of Coluber constrictor above, but they are slightly nar- rower across the zygapophyses and the centrum (Table 4). The complete left maxilla is from an adult snake; the two right maxillae are from subadults and are missing their anterior halves. The complete maxilla has space for 15 teeth with no diastema. The prefrontal process is located at the 5th and 6th teeth; the ectopterygoid process is located at the 12th tooth. There are three tooth sockets posterior to the latter process. A dorsal constriction of the maxillary ramus is present at the level of the ectopterygoid process. The constric- tion of the maxillary ramus is more pronounced in the smaller speci- mens than in the complete adult maxilla. The three compound bones are complete. Labial and lingual flanges are subequal in height; the lingual flange is slightly higher than the labial flange. COMPARISONS. - The fossil vertebrae were identified using discrim- inant analysis (see Coluber constrictor). The most important charac- ters for distinguishing M. ~Ragellum from Coluber constrictor and Drymarchon corais appear in Table 4. Lack of adequate comparative material prevented the inclusion of Masticophis taeniatw in this anal- ysis. However, the largest referred vertebrae are much larger than those of Recent M. taeniatus. The number and placement of teeth and the position of the two maxillary processes distinguish the referred maxillae from most other colubrids (App. 3-1). In the lampropeltine genera these processes are closer to one another and are located more posteriorly than in the fossils. North American natricines have more teeth; the minimum observed is 19 in Seminatrix. Heterodon has a diastema in the tooth row. The maxillae of Farancia are more robust and have the two processes more widely separated. Forms similar to the fossils include Drymarchon, whichhas more teeth(ca22) in adults and C. constrictor, which differsonly in lacking the constriction ofthe maxillary ramus at the level oftheectopterygoid process. BothM.,/Zagellum andM. taenia- tus have the constriction described in the fossils. Four maxillae of the latter differ from the fossils in having more (17-18) teeth. 28 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 The three fossil compound bones are referable to Ma*icophis based on the characters discussed under Coluber constrictor. DISTRIBUTION. - M.,/Zagellum is found throughout the southern half of the United States and in northern Mexico. The fossil record of M. ,/Zagellum must be regarded with caution because of the difficulty involved in distinguishing this species from C co?6strictor and other species of Masticophis. Vertebrae of this species have been reported from the Blancan of Arizona (Lindsay and Tessman 1974) and Texas (Rogers 1976), the Irvingtonian of Florida (Holman 1959a; Auffen- berg 1963) and Arkansas (Dowling 1958), and the Rancholabrean of Arizona, Arkansas, California, Florida, New Mexico, Nevada, and Virginia (see Holman 1981). REMARKS. - The Inglis material is the earliest reported from Flor- ida and includes the oldest known cranial material. It substantiates the appearance of Masticophis jZagellum by the earliest Pleistocene. GENUS Opheodrys FITZINGER 1843 Opheodrys vernalis (HARLAN) 1827 REFERRED MATERIAL. - UF 26368,29 vertebrae. DESCRIPTION. - These vertebrae have centra which are longer than wide (CL/NAW X = 1.54 + .145). They are nearly square across the zygapophyseal faces (POPR/PRPRX= 1.03 * .061).The cotyle is oval and epizygapophyseal spines are present. The accessory processes are short and often laterally directed. The haemal keel and subcentral ridges are poorly developed. POPR/PRPR CL/NAW OPHEODRYS VERNALIS n=18 INGLIS FOSSILS - 1 n=16 OPHEODRYS AESTIVUS ,-1.Aill-Iit n=18 lllllll 0.8 1.0 1.2 1.4 1.6 1.8 2.0 Figure'8.- Diagnostic ratios for the vertebrae of North American Opheodrys. Mean, 95% confidence limits, and range are indicated. POPR/PRPR = postzygapophyseal to prezygapophyseal length/width across the prezygapophyses, CL/NAW = centrum length/neural arch width. 1982 MEYLAN: INGLIS IA SQUAMATES 29 COMPARISONS. - The lack of development of the accessory processes and subcentral features serves to distinguish Opheodrys from other North American colubrids with long narrow vertebrae. The two extant species of Opheodrys can be distinguished by differ- ences in their centrum and neural arch shapes (CL/NAW and POPR/PRPR) (Auffenberg 1963). The fossils differ significantly from O. aestivvs in both these ratios (Fig. 8). They do not differ from O. vernalis and thus are assigned to this species. DISTRIBUTION. - 0. vernalis is at present restricted to the north- eastern portion of the United States from northeastern Kansas to Maine. Isolated populations occur at medium to high elevations in several western states, North Carolina, the eastern coastal plain of Texas, and northern Mexico. 0. aestivus occupies a nearly complimen- tary range from Florida to Texas, north to the Ohio River Valley and southern New Jersey. 0. aestivus, and not 0. vernalis, is present in the vicinityofInglis IA today. 0. vernalis is known as a fossil only from the Irvingtonian of Maryland (Holman 1977a). 0. aestivus is known from Rancholabrean localities in Florida (Auffenberg 1963) and Texas (Holman 1969a). REMARKS. - The Inglis Opheodrys vernalis, together with the large number of Recent isolated populations, is evidence of a once greater range for this species. The Inglis material indicates that in the early Pleistocene this species occurred as far south as Florida. The compli- mentary range of modern Opheodrys aestivus suggests that it has replaced 0. vernalis over much of its former range. In light of the present fossil record, this would have had to occur in the middle or late Irvingtonian. SUBFAMILY LAMPROPELTINAE DOWLING 1975 The subfamilial name Lampropeltinae is the best available for the North American constricting colubrids. In the present study, Arizona, Elaphe, and Pituophis have been added, on the basis of vertebral morphology, to the genera considered by Dowling and Duellman (1974) to bemembers ofthis group. Possibly thesimilarity in vertebral form of these snakes is due to convergence. GENUS Cemophora COPE 1860 Cemophora coccinea (BLUMENBACH) 1788 APPENDIX 3-5, TABLE 5 REFERRED MATERIAL. - UF 26269,28 vertebrae. DESCRIPTION. - The vertebrae are moderately small (CL = 1.75- 2.35 mm) and slightly longer than wide. The cotyles are as large or larger than the neural canal. Epizygapophyseal spines are absent. 30 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 Neural spines are almost as high as long, and are not undercut ante- riorly. The haemal keels are narrow and not expanded posteriorly. The accessory processes are as long or longer than the prezygapophyseal width, and are directed laterally. COMPARISONS. - The vertebrae are from a moderately smalllam- propeltine snake. The relative size of the neural canal indicates that the fossils are not those of juveniles of a larger species. The fossils differ from Rhinoceilus, which has neural spines undercut anteriorly (App. 3-5). Arizona is similar, but has wider vertebrae (CL/NAW, Table 4). The vertebrae of Cemophora coccinea can be separated from those of Lampropeltis triangul*m only with great difficulty. The best character for separation is the ratio of the height to the length of the neural spine (NLU/NH). The neural spine of modern L. triangulum is not as high as that of C. coccinea (Table 5). Vertebrae thought to represent one or the other of these forms are referred to C coccinea if NLU/NH is:52.6 and toL. triangulum ifNLU/NH is> 3.0. Vertebrae falling in the region of overlap (2.6 < NLU/NH > 3.0) were not referred to species unless they were very narrow (CL/NAW 2 1.40), as in modern C coccinem DISTRIBUTION. - Recent populations of C coccinea occur through- out the southeastern United States from southern New Jersey and central Missouri to Louisiana and Florida. An isolated population occurs in southern coastal Texas. This species has been reported from a single Rancholabrean locality in Florida (Auffenberg 1963). A form with similar vertebral morphology, PseudocemophoTa antiqua, is known from the Hemingsfordian of Florida (Auffenberg 1963) and Wyoming (Holman 1976). It is similar to both Cemophora and Lam- propeltis, but its relationship to these two genera remains unclear. Table 5. - Diagnostic ratios for the vertebrae of Cemophora coccinea, Lampropettis trianglthlm, and Arizona etegans. (NLU/NH)1 (CL/NAWY - N OR X SD N OR X SD Cemophora coccinea Recent 21 1.8-3.0 2.44 .42 23 1.05-1.55 1.33 .13 fossil 27 1.4-2.8 2.00 .33 27 1.05-1.55 1.18 .13 Lampropettis triangulum Recent 26 2.6-5.0 3.46 .77 19 1.10-1.35 1.24 .07 fossil 10 3.0-4.0 3.34 .46 10 1.0-1.29 1.19 .08 Arizona elegalls Recent 16 .87-1.08 .96 .11 1Neural Sbine Ldngth/Neural Spine Height 2CentrumLength/Neural Akh Width 1982 MEYLAN: INGLIS IA SQUAMATES 31 REMARKS. - Williams and Wilson(1967) suggested thatCemophora is aspecialized derivative ofsome member oftheL. triangulum group, thoughL. trial%gulum has the lowestneural spine ofanyLampropeltis. This apparent specialization for burrowing would have had to be reversed in order to arrive at the higher neural spines of C. coccinea. Thus it is more likely that Cemophora diverged from some earlier form of Lampropeltis. The Inglis fossils constitute the oldest record of Cemophora coccinea. The observed range in centrum lengths of the referred vertebrae indicates that at least two individuals ofC. coccinea were preserved in the Inglis IA fauna. GENUS Elaphe FITZINGER 1833 Elaphe guttata (LINNEAUS) 1766 FIGURE 9 A-C, TABLE 6 REFERRED MATERIAL. - UF 26370, 71 vertebrae; UF 26371, 1 L pterygoid; UF 26372, 1 L compound. DESCRIPTION. - The referred vertebrae lack hypapophyses. The centra are as wide as long (Table 6). Haemal keels are moderately to well developed and have posterior expansions. Subcentral ridges are moderately developed and straight. Accessory processes are not longer than the prezygapophyseal width. Neural spines are as tall as long. Epizygapophyseal spines are very weak or absent. Zygosphenes are usually flat or convex from above. Only the anterior third of the large left pterygoid is preserved. The anterior margin of the ectopterygoid flange rises from the toothbear- ing ramus at an angle of 25°. The ectopterygoid process is not well developed. The tooth row is straight. No ridge is evident on the dorsal surface of this fragment. The referred compound is large and nearly complete. The lingual flange is highly arched. Its dorsal edge is twice as high as the dorsal edge of the labial flange. COMPARISONS. - The pterygoid fragment has a pterygoid flange and thus cannot represent a viperid; it is too large to be Micrums. Unlike the pterygoids of the colubrine group and Lampropeltis getu- lus, there is no ridge on the dorsal surface at the level of the ectoptery- goid process. The weak development of this process indicates that the fossil is not a xenodontine or natricine. The extension of the pterygoid flange is more anterior in the fossil than in Elaphe obsoleta, less anterior than inPituophismelanoleucus and identical toE. guttata (see App. 3-3). The compound differs from colubrines in having only one flange developed. The lingual flange is unlike E. obsolete, Pituophis mela,zo- 32 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 A B C k - - 111 111 D E F l K 71 - g V . -r 1 = f-- 111 111 Figure 9. - Fossil E/aphe from Inglis IA: Elaphe guttata (A) pterygoid, ventral view, X5; (B) pterygoid, dorsal view, X5; (C) compound, labial view, X4. Elaphe obsolete (D) maxilla, occlusal view, X2; (E) palatine, dorsal view, X5: (F) parasphenoid, dorsal view, X4 (each scale = 5 mm). 1982 MEYLAN: INGLIS IA SQUAMATES 33 leucus, and Lampropeltis calligaster in being highly arched. This condition is present in Farancia abacura, Lampropeltis get*lus, Elaphegattata, and viperids. The two former have taller labial flanges than the fossil. Viperids have a less symmetrically arched labial flange (see Figs. 31-32, Brattstrom 1964). The referred vertebrae are of the lampropeltine type. Lampropeltis getulus differs from the fossils in having better developed subcentral ridges (strong in 83% of L. get,Ll,13 and 47% of E. guttata) that curve inward near the condyle (Brattstrom 1955a; Auffenberg 1963). Pituo- ph'is is unlike the fossils in having higher neural spines and zygo- sphenes that are usually concave from above (Auffenberg 1963). The fossils have better developed subcentral ridges thankrizona. They are larger than the vertebrae of Lampropeltis triangulum and Cemophora coccinea (1.6-2.45 mm CL in L. triangulam, 1.75-2.35 mm CL in C. coccinea, and 3.2-6.6 mm in the fossils). The fossil vertebrae are similar to several species of Elaphe. The vertebrae of variousElap/ie species have been separated on the basis of the height of the neural spine (Auffenberg 1963; Holman 1968a). E. vulpina has a lower neural spine than E. guttata or E obsoleta. It is not present in the Inglis fauna. Discriminant analysis was used to assign the vertebrae to E. obsoleta or E. guttata. The analysis is based on 28 vertebrae from four individuals of Elaphe guttata and 38 vertebrae from five individuals ofElaphe obsoleta. Of the 66 original observations, 7 (10.6%) were reclassified. Values for the most important measurements and ratios are found in Table 6. DISTRIBUTION. - Elaphe guttata is found throughout the southern half of the United States from New Mexico, Colorado, and northeast- ern Mexico to New Jersey and Florida. Fossils of this species are known from the Blancan of Texas (Rogers 1976) and Kansas (Bratt- strom 1967), the Irvingtonian of Kansas (Brattstrom 1967), and the Table 6. - The four most important ratios for the separation of Elaphe obsoleta and Elaphe guttata by discriminant analysis. (CL/NAW)1 (CL/CTWY (ZW/NAWp (CL/ZWy Elaphe guttata Recent (N = 32) 1.10 i .11 1.74 i .11 .81 zE .06 1.35 1 .14 fossils (N = 71) 1.08 i .09 1.49 + .19 .81 i .06 1.34 i .13 Elaphe obsoleta Recent (N = 33) 1.01 i .06 1.71 i .15 .81 + .07 1.25 + .09 fossils (N = 98) .98 i .06 1.41 i .18 .82 * .06 1.19 i .09 'Centrum Length/Neural Akh Width 2Centrum Length/Cotyle Width {Zygasphene Width/Neural Arch Width 'Centrum Length/Zygasphene Width 34 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 Rancholabrean of Kansas, Missouri, Texas, and Florida (see Holman 1981). Four extinct species of Elaphe have been described from the Mio- cene and Pliocene of North America (Holman 1979). There is no evi- dence that any of these are closely related to Elaphe guttate REMARKS. -The Inglis IA fossils establish the presence ofE. guttata in Florida by the earliest Pleistocene. The observed range in centrum lengths of the fossils shows that at least three individuals are present. Elaphe obsoleta (SAY) 1823 FIGURE 9 D-F, TABLE 6 REFERRED MATERIAL. - UF 26373,98 vertebrae; UF 26374, 1 L maxilla; UF 26375, 1 R palatine; UF 26376, 1 parasphenoid. DESCRIPTION. - The vertebrae fit the description ofElaphe guttata. The only qualitative difference is that the neural spines tend to be higher than long. The maxilla is nearly complete(Fig. 9 D). There is adiagonal break at the 13th tooth; probably a 14th tooth is lost. There is no diasterria. The maxilla is large, but not robust. The shaft is curved medially anterior to the 4th tooth, posterior to that point it is straight. The prefrontal process is at the level of the 7th and 8th teeth. The ectopte- rygoid process is at the level of the 12th and 13th teeth. Both processes are subrectangular. The palatine is moderately large. It is worn and broken posterior to the medial process. The posterior edge of the lateral process and the anterior edge of the medial process are opposite. There are four tooth positions anterior to the lateral process. The maxillary nerve foramen passes through the lateral process. The posterior opening of this fora- men lies below the dorsal surface of the lateral process. The base of the parasphenoid is missing posterior to the middle of the pituitary fossa. The cultriform process is stout, widens anteriorly, and extends beyond the suborbital flanges, which are well developed. There is a frontal step at the midpoint of the cultriform process. The suborbital flange has a smooth margin. COMPARISONS. - The referred vertebrae have been identified by the discriminant analysis discussed under Elaphe guttata. The fossil maxilla is that of a large colubrid (App. 3-1). It has fewer teeth than any natricine. The absence of a diastema indicates that it is not Heterodon (Fig. 12 A). Either one or both of the maxillary pro- cesses are more posteriorly located in the fossil than in Farancia, Coluber, Drymarchon, or Masticophis. Most of the maxillary ramus of the fossil is straight, unlike that in Lampropeltis getulus or Pituophis melanoleucus, which is gently curved. The maxilla ofE guttata andE 1982 MEYLAN: INGLIS IA SQUAMATES 35 obsoleta are similar to the fossil. That of the former differs slightly from both the fossil and E. obsoleta in having a posteriorly projecting prefrontal process. The palatine differs from that of most largeNorth American snakes in its basic form (App. 3-2). The two processes are absent in all viperids (Brattstrom 1964, Fig. 10). InHeterodon the base of this lateral process extends to the anterior end of the bone. The medial process of alllarge Nero(lia species has a shorter base than the fossil . Drymarchon, Coluber, and Masticophis have medial and lateral processes that arise from the same section of the toothed ramus. In the fossil the baseof the lateral process lies completely anterior to the base of the medial pro- cess. The fossil differs from the remaining large lampropeltines, except E. obsoleta and E. vulpina, in the placement of the maxillary nerve foramen in the lateral process. In the fossil the posterioropening of the foramen is below the dorsal surface of this process. In Elaphe guttata, Pitdophis melanoleucus, Lampropeltis getulus, and Lampro- peltis calligaster, this opening is in the dorsal surface of the lateral process. The final assignment of the fossil toE. obsoleta rather thanK vulpina is based on the presence of vertebrae of the former. The fossil parasphenoid differs from those of viperids in being nar- rower and in lacking the ventral keel found in that group (see Figs. 25-26, Brattstrom 1964). Unlike the fossil, larger natricines have no frontal steps on the cultriform process. The fossil differs from Heter- odon, in which orbital flanges are absent or restricted to the basal part of the bone. The cultriform process is narrower than in Farancia. but wider than inDrymarchon, Coluber, andMcaticophis. The latter three also have suborbital flanges reaching the anterior end of the cultri- form process. The suborbital flanges in Lampropeltis get*lus and Elaphe guttata are poorly developed, which results in a narrower cultriform process than in the fossil. The fossil is very similar to Pituophis melanoleums and E. obsoleta, but it lacks the small notches present in the suborbital flanges on either side of the parasphenoid in Pi&tophis, and thus is assigned to E. obsoleta (see App. 3-4). DISTRIBUTION. - Elaphe obsoleta is found today throughout the eastern half of the United States. It is known as a fossil from the Blancan and Irvingtonian of Kansas(Brattstrom 1967) and Blancan of Texas (Rogers 1976j. It is also present in seven middle to late Pleisto- cene sites in Florida (see Holman 1981). Holman (1973) provided a tentative phylogeny for Elaphe in which he designated E. buisi Hol- man from the middle Pliocene of Oklahoma and E. kansensis (Gil- more) from the lower Pliocene of Kansas as ancestors to E. obsoleta. REMARKS. - The Inglis material documents the presence of this species in Florida at the beginning of the Irvingtonian. E obsoleta is 36 BULLETIN FLORIDA STATE MUSEUM Vol. 27, No. 3 thus widely distributed by the early Pleistocene. At least two individ- uals are represented in the fauna. GENUS Lampropeltis FITZINGER 1843 Lampropeltis getulus (LINNEAus) 1766 REFERRED MATERIAL. - UF 26377,58 vertebrae. DESCRIPTION. - The referred vertebrae are short and wide with moderate to low neural spines. The neural arches are slightly depres- sed. Thesubcentral ridgesare very well developed(in most they are so strong that the ventral surface of the centrum appears to be excavated between the subeentral ridges and haemal keel). Subcentral ridges curve inward near the cotyle. Haemal keels are usually widened posteriorly. COMPARISONS. - The extreme development of the subeentral ridges of Lampropeltis getulus distinguishes it from most other large lam- propeltines (Auffenberg 1963). Lampropeltis calligaster is similar, but has haemal keels that remain uniformly narrow to the condyle (100% vs. 23% in L. getulus). It also has less well developed subeentral ridges (88% moderate to weak vs. 83% strong in L. getulus). Farancia also has similar vertebrae, but has more depressed neural arches and wider, blunter accessory processes (Auffenberg 1963). DISTRIBUTION. - L. getulus ranges from coast to coast across the southern United States and northern Mexico. This species is known as a fossil from Blancan localities in Kansas (Brattstrom 1967) and Texas (Rogers 1976), Irvingtonian localities in Kansas (Brattstrom 1967), Texas (Holman 1969b), and Florida (Holman 1959a), and Ranchola- brean localities in Arizona, California, Florida, Georgia, Kansas, New Mexico, Nevada, Tennessee, and Texas (see Holman 1981). Two extinct species ofLampropelt~ have been described, but they are more closely related to the tricolored king snakes, includingL. trial%gullim (Holman 1970), and are discussed under that heading. REMARKS. - Blanchard(1921) and Blaney(1977) suggested that the Recent subspecies, Lampropeltis getulus splendida, from Texas and Mexico best approximates the ancestor of now widely dispersed and highly variable Lampropeltis getulus. Blaney (1977) considered the peninsular Florida form, L. g. foridana, to be a direct and closely related derivative of L. g. splendida. Thus a splendida-like form was probably once widespread. The Inglis record now indicates that this had occurred by the earliest Pleistocene. At least two individuals of L. getulas, one large and one of medium size,