Paleogeography of the Caribbean Region: Implications for Cenozoic Biogeography

Home , Anegada Passage, Caribbean Basin, Caribbean Sea, Geddes (crater), Geology of Cuba, Geology of Puerto Rico

PALEOGEOGRAPHY OF THE CARIBBEAN REGION: IMPLICATIONS FOR CENOZOIC BIOGEOGRAPHY

MANUEL A. ITURRALDE-VINENT Research Associate, Department of Mammalogy American Museum of Natural History Curator, Geology and Paleontology Group Museo Nacional de Historia Natural Obispo #61, Plaza de Armas, CH-10100, Cuba

R.D.E. MA~PHEE Chairman and Curator, Department of Mammalogy American Museum of Natural History

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 238, 95 pages, 22 figures, 2 appendices Issued April 28, 1999 Price: $10.60 a copy

Copyright O American Museum of Natural History 1999 ISSN 0003-0090 CONTENTS Abstract ...... 3 Resumen ...... 4 Resumo ...... 5 Introduction ...... 6 Acknowledgments ...... 8 Abbreviations ...... 9 Statement of Problem and Methods ...... 9 Paleogeography of the Caribbean Region: Evidence and Analysis ...... 18 Early Middle Jurassic to Late Eocene Paleogeography ...... 18 Latest Eocene to Middle Miocene Paleogeography ...... 27 Eocene-Oligocene Transition (35±33 Ma) ...... 27 Late Oligocene (27±25 Ma) ...... 31 Early Middle Miocene (16±14 Ma) ...... 31 Biogeographical Hypotheses and Caribbean Paleogeography ...... 35 Continent-Island Vicariance: Model of Rosen ...... 35 Passive Overwater Dispersal: Model of Hedges and Co-workers ...... 40 Preliminary Issues ...... 40 Sources of Error in Estimating Times of Lineage Origins ...... 43 Passive Transport and Cenozoic Surface-Current Patterns ...... 45 Surface-Current Patterns and Flotsam Dispersal ...... 45 Surface-Current Patterns and Paleogeography ...... 45 Surface-Current Patterns and Proxy Data ...... 48 Other Constraints ...... 50 GAARlandia Landspan and Island±Island Vicariance: Model of MacPhee and Iturralde-Vinent ...... 52 Landspans, Vicariance, and Diversity Scenarios ...... 52 Discussion ...... 56 Conclusions ...... 58 References ...... 59 Appendix 1: Reconstructing Caribbean Paleogeography: An Analytical Guide ...... 72 Yucatan Peninsula ...... 72 Northern Central America, Nicaragua Rise, and Western Jamaica ...... 73 Southern Central America ...... 75 Northwestern South America ...... 75 Aruba/Tobago Belt ...... 75 Greater Antilles ...... 77 Blue Mountains Block ...... 80 Aves Ridge, Lesser Antilles, and Grenada Basin ...... 84 Beata Ridge ...... 87 Cayman Islands and Cayman Ridge ...... 87 Appendix 2: A Plate Tectonic Model of the Caribbean from Latest Eocene to Middle Miocene ...... 87

2 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 3

ABSTRACT This paper* presents a series of detailed paleo- nected by a ``landspan'' (i.e., a subaerial connec- geographical analyses of the Caribbean region, tion between a continent and one or more off- beginning with the opening of the Caribbean ba- shelf islands) centered on the emergent Aves sin in the Middle Jurassic and running to the end Ridge. This structure (Greater Antilles ϩ Aves of the Middle Miocene. Three intervals within the Ridge) is dubbed GAARlandia. The massive up- Cenozoic are given special treatment: Eocene±Ol- lift event that apparently permitted these connec- igocene transition (35±33 Ma), Late Oligocene tions was spent by 32 Ma; a general subsidence (27±25 Ma), and early Middle Miocene (16±14 followed, ending the GAARlandia landspan Ma). While land mammals and other terrestrial phase. Thereafter, Caribbean neotectonism result- vertebrates may have occupied landmasses in the ed in the subdivision of existing land areas. Caribbean basin at any time, according to the in- The GAARlandia hypothesis has great signi®- terpretation presented here the existing Greater cance for understanding the history of the Antil- Antillean islands, as islands, are no older than lean biota. Typically, the historical biogeography Middle Eocene. Earlier islands must have existed, of the Greater Antilles is discussed in terms of but it is not likely that they remained as such (i.e., whether the fauna was largely shaped by strict as subaerial entities) due to repeated transgres- dispersal or strict continent±island vicariance. The sions, subsidence, and (not incidentally) the K/T GAARlandia hypothesis involves elements of bolide impact and associated mega-tsunamis. Ac- both. Continent±island vicariance sensu Rosen ap- cordingly, we infer that the on-island lineages pears to be excludable for any time period since forming the existing (i.e., Quaternary) Antillean the mid-Jurassic. Even if vicariance occurred at fauna must all be younger than Middle Eocene. that time, its relevance for understanding the ori- The fossil record, although still very poor, is con- gin of the modern Antillean biota is minimal. sistent with the observation that most land mam- Hedges and co-workers have strongly espoused mal lineages entered the Greater Antilles around over-water dispersal as the major and perhaps the Eocene±Oligocene transition. only method of vertebrate faunal formation in the Western Laurasia (North America) and western Caribbean region. However, surface-current dis- Gondwana (South America) were physically con- persal of propagules is inadequate as an expla- nected as continental areas until the mid-Jurassic, nation of observed distribution patterns of terres- ca. 170 Ma. Terrestrial connections between these trial faunas in the Greater Antilles. Even though continental areas since then can only have oc- there is a general tendency for Caribbean surface curred via landbridges. In the Cretaceous, three currents to ¯ow northward with respect to the major uplift events, recorded as regional uncon- South American coastline, experimental evidence formities, may have produced intercontinental indicates that the ®nal depositional sites of pas- landbridges involving the Cretaceous Antillean is- sively ¯oating objects is highly unpredictable. land arc. The Late Campanian/Early Maastrich- Crucially, prior to the Pliocene, regional pale- tian uplift event is the one most likely to have oceanography was such that current-¯ow patterns resulted in a landbridge, as it would have been from major rivers would have delivered South coeval with uplift of the dying Cretaceous arc. American waifs to the Central American coast, However, evidence is too limited for any certainty not to the Greater or Lesser Antilles. Since at least on this point. The existing landbridge (Panaman- three (capromyid rodents, pitheciine primates, and ian isthmus) was completed in the Pliocene; evi- megalonychid sloths) and possibly four (neso- dence for a precursor bridge late in the Middle phontid insectivores) lineages of Antillean mam- Miocene is ambiguous. mals were already on one or more of the Greater We marshal extensive geological evidence to Antilles by the Early Miocene, Hedges' inference show that, during the Eocene±Oligocene transi- as to the primacy of over-water dispersal appears tion, the developing northern Greater Antilles and to be at odds with the facts. By contrast, the land- northwestern South America were brie¯y con- span model is consistent with most aspects of An- tillean land-mammal biogeography as currently * Contribution 2 to the series ``Origin of the Greater known; whether it is consistent with the bioge- Antillean Land Mammal Fauna.'' ography of other groups remains to be seen. 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

RESUMEN El propoÂsito de este trabajo es presentar una vieron brevemenmte conectadas por una ``land- serie de anaÂlisis paleogeograÂ®cos detallados de la span'' (proyeccioÂn de terreno) (es decir, por una regioÂn del Caribe, comenzando con la apertura de coneccioÂn subaeÂrea entre un continente y una o la cuenca del Caribe en el JuraÂsico Medio y ex- maÂs islas situadas fuera del lõÂmite de la plataforma tendieÂndose hasta el Mioceno Medio. Tres intercontinental), coneccioÂn que estuvo centrada en la valos del Cenozoico reciben un tratamiento es- entonces emergida Cresta de Aves. Esta estructura pecial: la transicioÂn Eoceno-Oligoceno (35±33 (Crestas de las Antillas Mayores y de Aves) se Ma), el Oligoceno TardõÂo (27±25 Ma), y el Mio- denominoÂ GAARlandia. El evento de levanta- ceno Medio temprano (16±14 Ma). Aunque los miento masivo que aparentemente permitioÂ esta mamõÂferos terrestres pudieron haber ocupado ma- coneccioÂn terminoÂ hace unos 32 millones de anÄos; sas de tierra en la cuenca del Caribe en cualquier debido a una subsidencia general que terminoÂ con momento de su historia, de acuerdo con la inter- la fase de ``landspan'' de GAARlandia. Posterior- pretacioÂn que se presenta en este trabajo, las ac- mente la etapa neotectoÂnica caribenÄa resultoÂenla tuales Antillas Mayores, como islas, son no maÂs subdivisioÂn de las tierras existentes. antiguas que Eoceno Medio. Islas maÂs antiguas La hipoÂtesis GAARlandia tiene un gran signi- deben haber existido, pero no es probable que el- ®cado para comprender la historia de la biota An- las hayan permanecido como tales (es decir, como tillana. TõÂpicamente, la biogeografõÂa histoÂrica de entidades subaeÂreas) debido a las repetidas trans- las Antillas Mayores se discute en teÂrminos de si gresiones, subsidencia, y (no incidentalmente) al la fauna fue principalmente formada por disper- impacto del lõÂmite K/T y el megatsunami asociado sioÂn estricta o por estricta vicariancia continente± al mismo. De acuerdo con esto nosotros inferimos isla. La hipoÂtesis GAARlandia comprende ele- que los linajes insulares que forman la fauna an- mentos de ambas. Pero la vicariancia continente± tillana actual (y cuaternaria en general) deben ser isla al estilo de Rosen puede ser excluida para maÂs joÂvenes que el Eoceno Medio. El registro foÂs- cualquier momento desde el JuraÂsico Medio. In- il, a pesar de ser muy pobre auÂn, es consistente cluso si la vicariancia hubiese ocurrido en aquella con la observacioÂn de que la mayorõÂa de los li- eÂpoca, su relevancia para comprender el origen de najes de mamõÂferos llegaron a las Antillas Ma- la biota antillana moderna es mõÂnima. Hedges y yores alrededor del lõÂmite Eoceno±Oligoceno. sus colaboradores han propuesto con eÂnfasis la El oeste de Laurasia (America del Norte) y el dispersioÂn por agua como el principal, sino el uÂn- de Gondwana (AmeÂrica del Sur) estuvieron fõÂsi- ico, meÂtodo de formacioÂn de la fauna de verte- camente conectados como aÂreas continentales brados en la regioÂn del Caribe. Sin embargo, la hasta el intervalo Bajociano al Oxfordiano (178± 160 Ma) cuando se comenzoÂ a formar la cuenca dispersioÂn de propaÂgulos mediante las corrientes oceaÂnica del Caribe. Coneccciones terrestres entre marinas super®ciales es inefectiva para explicar dichas aÂreas continentales a partir de entonces los patrones de distribucioÂn actual de la fauna ter- soÂlo pudieron ocurrir mediante puentes naturales restre en las Antillas Mayores. Incluso aunque ex- de terreno. En el CretaÂcico tres eventos principa- iste una tendencia general de las corrientes super- les de levantamiento, coincidentes con inconfor- ®ciales del Caribe a ¯uir hacia el norte con res- midades regionales, pudieran haber producido pecto a la costa sudamericana, las evidencias ex- puentes intercontinentales que involucraron al perimentales indican que es praÂcticamente arco de islas volcaÂnicas de las Antillas. El lev- impredecible doÂnde seraÂn ®nalmente depositados antamiento ocurrido en el Campaniano tardõÂo a los objetos ¯otantes acarreados por dichas cor- Maastrichtiano temprano es el que tiene las ma- rientes. Al respecto, es crucial el hecho de que, yores posibilidades de haber producido un puente antes del Plioceno, la paleoceanografõÂa regional natural, ya que eÂste coincidioÂ en el tiempo con la fue tal que los patrones de corrientes de los rõÂos extincioÂn del vulcanismo cretaÂcico. No obstante, sudamericanos debieron acarrear los objetos a la la evidencia es muy limitada para tener alguna deriva provenientes de AmeÂrica del sur hacia las seguridad sobre este asunto. El puente natural que costas de AmeÂrica Central, o hacia el PacõÂ®co, no existe actualmente (itsmo de PanamaÂ) se completoÂ hacia las Antillas Mayores. Dado que al menos en el Plioceno; pero la evidencia para un puente tres linajes (roedores caproÂmidos, primates pite- anterior en el Mioceno Medio tardõÂo es ambigua. cinos y perezosos megalonõÂchidos) y posible- AquõÂ se presenta extensa evidencia geoloÂgica mente cuatro (insectõÂvoros nesofoÂntidos) de los para mostrar que, durante la transicioÂn entre el mamõÂferos antillanos se encontraban ya en las An- Eoceno y el Oligoceno, las tierras antillanas y la tillas Mayores a comienzos del Mioceno, las in- porcioÂn noroccidental de AmeÂrica del Sur estu- ferencias de Hedges respecto al dominio de la 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 5 dispersioÂn por agua como el modo de migracioÂn grafõÂa histoÂrica de los mamõÂferos terrestres antil- de esta fauna estaÂ en desacuerdo con los hechos. lanos tal como se conoce hoy dõÂa; aunque si tam- En contraste, el modelo de ``landspan'' es consis- bieÂn es consistente con la biogeografõÂa de otros tente con muchos de los aspectos de la biogeo- grupos es algo que estaÂauÂn por de®nir.

RESUMO Este trabalho apresenta uma seÂrie de anaÂlises entre um continente e uma ou mais ilhas oceaÃni- paleogeograÂ®cas detalhadas da regiaÄo do Caribe, cas). Esta ``landspan'' estaria centrada na, entaÄo comecËando com a abertura da bacia do Caribe no emergente, Cadeia de Aves. Denominou-se esta JuraÂssico MeÂdio e seguindo ateÂ o ®m do Mioceno estrutura (Cadeia das Grandes Antilhas ϩ Cadeia MeÂdio. TreÃs intervalos do CenozoÂico receberam de Aves) como GAARlaÃndia. O soerguimento especial atencËaÄo: a transicËaÄo Eoceno-Oligoceno massivo que aparentemente permitiu estas conex- (35±33 Ma), o ®nal do Oligoceno (27±25 Ma) e oÄes ocorreu a cerca de 32 milhoÄes de anos, sendo o inõÂcio do Mioceno MeÂdio (16±14 Ma). Ainda seguido por uma subsideÃncia geral que terminou que mamõÂferos e outros vertebrados terrestres pos- com a fase de ``landspan'' da GAARlaÃndia. Pos- sam ter ocupado massas de terra na bacia do Ca- teriormente, o neotectonismo caribenho resultou ribe em qualquer momento de sua histoÂria, segun- na subdivisaÄo das terras existentes. do o presente estudo, a existeÃncia das Grandes Tradicionalmente, a histoÂria biogeograÂ®ca das Antilhas, como ilhas, naÄo eÂ mais antiga do que o Grandes Antilhas eÂ discutida em termos de for- Mioceno MeÂdio. Ilhas mais antigas podem ter ex- macËaÄo da fauna estritamente por dispersaÄo ou por istido, no entanto, eÂ pouco provaÂvel que tenham vicariaÃncia continente-ilha. A hipoÂtese da GAAR- permanecido como tais por longos perõÂodos. As- laÃndia envolve elementos de ambas as correntes, sim sendo, inferimos que todas as linhagens que ainda que os modelos de vicariaÃncia continente- formam a fauna antilhana atual (ou seja, Quater- ilha sensu Rosen possam ser excluõÂdos para qu- naÂria) devam ser mais recentes que o Eoceno MeÂ- alquer perõÂodo desde o JuraÂssico MeÂdio. Hedges dio. O registro foÂssil, apesar de ser bastante pobre, e colaboradores teÃm veementemente sugerido a eÂ consistente com a observacËaÄo de que a maioria dispersaÄo atraveÂs d'aÂgua como o principal, senaÄo das linhagens de mamõÂferos terrestres chegaram ouÂnico, meio pelo qual teria ocorrido a formacËaÄo aÁs Grandes Antilhas por volta da transicËaÄo da fauna de vertebrados do Caribe. Entretanto, a Eoceno-Oligoceno. proposta de dispersaÄo de propaÂgulos por correntes O oeste da LauraÂsia (AmeÂrica do Norte) e o da marinhas super®ciais eÂ inadequada para explicar Gondwana (AmeÂrica do Sul) estiveram ®sica- os padroÄes de distribuicËaÄo de fauna terrestre ob- mente conectados como aÂreas continentais ateÂo servados nas Grandes Antilhas. Enfatiza-se que, JuraÂssico MeÂdio. TreÃs principais eventos de soer- antes do Plioceno, a paleoceanogra®a da regiaÄo guimento no CretaÂceo, registrados por inconfor- midades regionais, podem ter produzido pontes era tal que o padraÄo de ¯uxo de correntes dos intercontinentais envolvendo o arco de ilhas an- principais rios sulamericanos deviam carrear ob- tilhanas do CretaÂceo. O soerguimento ocorrido no jetos para a costa da AmeÂrica Central e naÄo para Campaniano superior/Maastrichtiano inferior pa- as Grandes e Pequenas Antilhas. No mõÂnimo treÃs rece ser o mais relacionado com a formacËaÄo de (roedores capromiõÂdeos, primatas piteciõÂneos e uma ponte de conexaÄo, uma vez que este coincide preguicËas megaloniquideas) e talvez quatro (in- com o soerguimento ocorrido durante o desapa- setõÂvoros nesofondideos) linhagens de mamõÂferos recimento do arco do CretaÂceo. A conexaÄo atual antilhanos jaÂ ocorriam em uma ou mais ilhas das (istmo do PanamaÂ) completou-se no Plioceno; Grandes Antilhas no inõÂcio do Mioceno, indican- evideÃncias de uma ponte anterior no ®m do Mio- do que as propostas de Hedge quanto a primazia ceno MeÂdio saÄo ambõÂguas. da dispersaÄo aquaÂtica da fauna naÄo estaÄo de acor- NoÂs buscamos extensivas evideÃncias geoloÂgicas do com os fatos. Neste sentido, o modelo de para demonstrar que durante a transicËaÄo Eoceno- ``landspan'' eÂ consistente com a maioria das pro- Oligoceno, a parte norte das Grandes Antilhas postas atualmente aceitas para a biogeogra®a dos (entaÄo em desenvolvimento) e o noroeste da mamõÂferos terrestres antilhanos. Permanece em AmeÂrica do Sul estiveram brevemente conectados aberto se este modelo esta de acordo com as pro- por uma ``landspan'' (i.e., uma conexaÄo sub-aeÂrea postas biogeograÂ®cas para outros grupos. 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

It is this independence of biological from geological data that makes the comparison of the two so interesting because it is hard to imagine how congruence between the two could be the result of anything but a causal history in which geology acts as the independent variable providing op- portunities for change in the dependent biological world. Ð Donn E. Rosen (1985: 637)

INTRODUCTION During the past century, a number of hy- seems inescapable that all three were in- potheses have been offered as partial or com- volved in the formation of the Antillean land- plete explanations for the origins of Antillean mammal fauna, although not necessarily in terrestrial vertebrate faunas.1 Three mecha- either the manner or the degree envisaged by nisms have been discussed extensively in the other authors. literature: (1) dispersal over water barriers Our present purpose is to present a fresh (e.g., Matthew, 1918; Darlington, 1938; perspective on the ``geography'' part of bio- Woods, 1989; Hedges et al., 1992, 1994; geography, as it relates to the Caribbean re- Hedges, 1996a, 1996b); (2) dispersal over gion, and to examine how this may offer short-lived landbridges and landspans2 (e.g., novel insights into historical processes of Fernandez de Castro, 1884; De La Torre, faunal formation. (For a listing of most of 1910; Gayet et al., 1992; MacPhee and Itur- the features and localities mentioned in the ralde-Vinent, 1994, 1995); and (3) vicari- text, see ®gure 1). Although our speci®c con- ance, i.e., splitting or division of a biota or cept of Antillean paleogeographical history taxon through the development of a natural differs in various ways from those of other barrier (e.g., Rosen, 1975, 1985; Guyer and authors (see Biogeographical Hypotheses Savage, 1987; MacPhee and Wyss, 1990). and Caribbean Paleogeography), we have Although these mechanisms are sometimes made a particular effort to document and presented as though they were discrete, mu- evaluate other views. tually exclusive alternatives, depending on It is widely recognized that hypotheses the time, place, and taxon under discussion, concerning Antillean historical biogeography any or all of them may have been involved are critically dependent on speci®c recon- in Antillean faunal formation. In fact, to pre- structions of regional paleogeography, paleo- view the chief conclusion of this paper, it ceanography, tectonics, and other bodies of data. However, most discussions of this sub- 1 There are several current biogeographical de®nitions ject by life scientists have tended to empha- of the `Àntilles,'' ``West Indies,'' ``Caribbean Islands,'' size biological evidence over geological ev- `ìnsular Neotropics'' and their various subdivisions. In idence. This bias should not be viewed as this paper, Greater and Lesser Antilles will have their being merely re¯ective of biologists' under- usual meanings; `Àntillean'' as an adjective refers to anything having to do with these islands, and is used in standable preference for their own kinds of preference to ``West Indian'' (which, as generally used, data, because several issues are involved. covers other, non-Antillean islands such as Bahamas). First, although much of the general geo- The Caribbean region, which we newly de®ne as a pa- logical and tectonic literature is signi®cant leogeographical concept, consists (at any stage of its defor understanding the biogeographical histo- velopment) of the Caribbean Sea and all of its contents, ry of the Caribbean region, virtually none of plus the facing continental margins of North, South, and it was written with the needs of biologists in Central America. Thus the Caribbean region is larger mind. Accordingly, biologists hoping to in- than the Caribbean Plate, although the structures on that tegrate geological information into their plate comprise most of the entities of interest here. It is also larger than ``West Indies'' as de®ned by Hedges work are faced with the daunting tasks of (1996a, 1996b). having to compile evidence from many dif- 2 For de®nition of ``landspan'' see section entitled ferent sources, judge as best they can the ac- GAARlandia Landspan and Island±Island Vicariance. curacy of age assignments and other primary 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 7 Fig. 1. Reference map of Caribbean region illustrating most of the geographical and geological features referred to in text. 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 data, and, frankly, recognize when geological land masses existed in the Caribbean, or theory outstrips fact. Lack of familiarity with when or how many times these land masses the subject matter and the methods of geo- were connected to nearby continents, or the logical argumentation may lead to simple nature of the relief they exhibited. Thus, factual errors, or, possibly worse, encourage Hedges et al. (1992) explored structural re- the uncritical acceptance of insuf®ciently lationships among Caribbean land masses tested geological scenarios primarily because and nearby continents on the basis of Pindell they appear to support certain biological hy- and Barrett's (1990) tectonic reconstruction potheses. It is partly for this reason that some which, in fact, contains no information on authors validly question whether biologists such relationships. (The latter authors discuss should place themselves in the very vulner- only the position of geological units, which able position of relying on the revealed truths is not paleogeography as we de®ne it.) of geologists to explain biogeographical pat- In this paper we offer the ®rst comprehen- ternings (Henderson, 1991: 61; see also Craw sive paleogeographical and paleoceanograph- and Weston, 1984). Knowledge, we suggest, ical reconstructions of the Caribbean basin, is the best antidote to vulnerability. from latest Eocene to Middle Miocene, an Second, despite their apparent elegance, interval selected for reasons explained in de- plate tectonic models (e.g., Malfait and Din- tail in succeeding sections. We also brie¯y kelmann, 1972; Duncan and Hargraves, review paleogeographical scenarios for Ju- 1984; Leclere and Stephan, 1985; Ross and rassic through Late Eocene time, for the pur- Scotese, 1988; Donnelly, 1989a; Pindell and pose of evaluating evidence for early land Barrett, 1990; Mann et al., 1995; Hay and connections and island permanency. Re¯ect- Wold, 1996; Iturralde-Vinent, 1996a, 1997b) ing our own interests, we concentrate on the vary widely in their comprehensiveness and paleogeography of the central part of the Ca- testability (Rull and Schubert, 1989; Per®t ribbean basin and portions of Central Amer- and Williams, 1989). For example, agree- ica and northwestern South America. In the ment is still lacking regarding the number main, the biogeographical implications that and ®t of plates and microplates in the Ca- we have pursued in this investigation are ribbean RegionÐa basic issue of fact (cf. those most closely tied in with geography. Donnelly, 1985; Ross and Scotese, 1988; Understanding the phylogeny of the Antil- Pindell, 1994; Hay and Wold, 1996). Fur- lean biota is equally important and interest- thermore, plate tectonic models do not nec- ing; however, this topic is reserved for a sub- essarily provide the kinds of information that sequent paper in this series (MacPhee and biologists are most interested in. Typically, Iturralde-Vinent, in prep.). such models focus on reconstructing histor- Because of the large quantity of ancillary ical positions of speci®c geologic units com- documentation required to support a study of monly denoted as plates, terranes, blocks, this sort, for ef®ciency in presentation much volcanic arcs, and ridges (e.g., Malfait and of the basic geological, paleogeographical, Dinkelmann, 1972; Duncan and Hargrave, and paleontological information is presented 1984; Leclere and Stephan, 1985; Donnelly, in the form of appendices, tables, and ®gures. 1985; Ross and Scotese, 1988; Pindell and The text summarizes this information in dis- Barrett, 1990; Pindell, 1994; Mann et al., cursive form, but its main function is to dis- 1995; Hay and Wold, 1996). They are not at cuss problems of interpretation and expla- all, or are only incidentally, concerned with nation. Readers wishing to utilize the chief creating well-constrained paleogeographical results of our investigations may pro®tably maps that portray the physical geography of consult the main text, but those requiring a such units (or parts thereof) through time. greater level of detail should refer to the ap- (For further discussion of these and other pendices throughout. concepts, see Paleogeography of the Carib- bean Region: Evidence and Analysis.) With ACKNOWLEDGMENTS the purely tectonic literature as the sole guide, one cannot derive any consistent pic- The manuscript of the present paper was ture concerning how many times subaerial awarded the 1997 Premio Anual de la Cien- 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 9

cia by the Academia de Ciencias de Cuba for many contributions to our ®eld program. Re- outstanding scienti®c investigations. We are search related to this paper was partly sup- grateful to the Academy, and thus the people ported by a Kalb¯eisch postdoctoral fellowship of Cuba, for this honor. at the AMNH (to MIV) and NSF 902002 (to We extend special thanks to Gregory Mayer RDEM). Figures were drafted by MIV, with (University of Wisconsin±Parkside), Peter the exception of ®gures 10 and 11 (by Patricia Mattson (Queens College, City University of J. Wynne) and ®gures 9 and 13 (by RDEM). New York), Gary Morgan (New Mexico Mu- Fernando Sicuro kindly supplied the Portu- seum of Natural History), and Marcelo San- guese version of the abstract. chez Villagra (UniversitaÈt TuÈbingen) for re- viewing the manuscript, to Lisa Gahagan (In- ABBREVIATIONS stitute of Geophysics, University of Texas at Austin) for programming assistance in devel- AMNH American Museum of Natural History ID immunological distance oping the tectonic model presented in appendix Ma millions of years (ago) 2, and to Clare Flemming (AMNH) for help MNHNH Museo Nacional de Historia Natural, with editing the text. We also acknowledge our La Habana great debt to Clare Flemming and IneÂs Horo- NWSA Northwestern South America (micro- vitz (AMNH) and to Stephen DõÂaz Franco and continent or microplate) Reinaldo Rojas Consuegra (MNHNH) for their Fm formation

STATEMENT OF PROBLEM AND METHODS We agree with Hedges (1996b: 166) that Eva and MacFarlane's (1985) investigation the aspect of Caribbean geological history of did not require this kind of reconstruction, greatest interest to biogeographers, the rela- and they simply portrayed the different geo- tionships of emergent land areas, is unfortu- logic units forming the Jamaican basement nately the one that is most poorly under- (ophiolites, metamorphic rocks, Cretaceous/ stood. This point is best explored by noting Paleogene island arc suites, and late Cam- some illustrative examples: panian to Holocene sedimentary formations) Not all geological maps contain recoverable as though they have had the same relative paleogeographical information. positions and areal dispositions since the Pa- Eva and MacFarlane's (1985: ®gs. 4±12) leocene. The message is that nonpalinspastic study of carbonate development in Jamaica reconstructionsÐby far the most common includes a series of nine illustrations that de- ``paleogeographical'' representations in the pict the development of various features geological literature (e.g., Khudoley and (e.g., subaerial land surfaces, subaerial and Meyerhoff, 1971; Maurrasse, 1982; Salvador, marine volcanoes, shallow and deep sea) 1987; Smith et al., 1994)Ðmay be of negli- from Paleocene to Pliocene times. However, gible value for biogeographical investigations they are not paleogeographical maps in any because they are not intended to be paleogeo- literal sense, because all features are shown graphically accurate. A simpli®ed example of as evolving within the present-day perimeter palinspastic reconstruction for the Greater An- of the island, with no attempt to restore fold- tilles is illustrated in ®gures 2 and 3. ed and faulted rock units to their original rel- Paleogeographical information is dif®cult to ative geographical positions. Because some treat in a uniform way. of Jamaica's basement rocks were strongly Just as there is no such thing as a perfect folded and faulted during the Middle Eocene biological classi®cation, there is no such and late Neogene (Lewis et al., 1990; Rob- thing as a perfect map. Often, maps are de- inson, 1994), any effort to capture paleogeo- signed to depict only one body of data ac- graphical reality would require the use of curately; thus, even if they are intentionally palinspastic methods to reconstruct displace- paleogeographical in nature, they may not ments of blocks along faults and restore orig- show different categories of information with inal surface areas of deformed formations. equal degrees of precision. For example, as 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 2. Simpli®ed palinspastic reconstruction of Greater Antilles Foldbelt along a cross section pass- ing through eastern Cuba and western Hispaniola, to illustrate methodological points. Note that the foldbelt consists of a series of tectonically superimposed units that have been foreshortened by deformation. As a result, the current width of the set of geological units transected by the section amounts to only a fraction of the units' original width (for additional explanation, see text). See ®gure 3 for location of cross section. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 11 are reference points for cross section depicted in ®gure 2. (Information Ј Ân. A and A Fig. 3. Palinspastic reconstruction of eastern part of Greater Antilles Foldbelt, for period corresponding to Late Eocene through mid-Oligocene compiled from many sources; see appendix 1.) (ca. 37±30 Ma).the This northern framework margin has of been the greatly Caribbean disrupted plate, since east the of mid-Oligocene Holguõ by sinistral strike-slip movements and deformations along 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 its title implies, the Atlas of Mesozoic and used for that purpose (e.g., Hedges, 1996a, Cenozoic Coastlines (Smith et al., 1994) is 1996b). concerned with depicting coastline informa- Island geology and island paleogeography tion, on a global scale, for the past 245 Ma. are not isomorphous. Because its intended scope is the entire plan- Conventionally, a geographical island is et, regional details are often lost or portrayed de®ned by its shoreline (i.e., its subaerial pe- inaccurately. As the authors point out, small rimeter), although other criteria are of course discrepancies do not affect the big picture, possible (e.g., 100 m isobath). Shorelines but they do complicate the use of the maps Ϫ can be affected by rise or fall in sea level, for other purposes. Of interest here is the fact that the Greater Antilles are depicted in their deposition or erosion, and uplift or subsi- current sizes and positions relative to North dence of the geological unit that constitutes America during all relevant time periods. Be- the islands basement. Shorelines are there- cause the dimensions of the Caribbean Sea fore exceptionally dynamic at virtually all di- have changed over time, the Greater Antilles mensional and temporal scales. Geological are forced by the mapping program into po- units, by contrast, are only indirectly affected sitions they could never have occupied. by sur®cial processes such as prograding or Thus, these islands overlap northern South degrading shorelines; they re¯ect a deeper America at 170 Ma, project east of Trinidad structure, and deep structure is rarely coter- into the South Atlantic (!) at 155 Ma, and minous with conventional geography. The ®nally end up at their present-day position at usual objective of geological study is to in- 80±0 Ma. Furthermore, subaerial exposures terpret the history of tectonic elements in on these islands are depicted on Oligocene terms of their formation, evolution and sub- and later maps, but not on earlier ones. This sequent transformation into other elements, suggests that land areas did not exist as such utilizing the imprints that such processes prior to the mid-Cenozoic, which is not ac- leave in the rocks themselves. Shoreline re- curate. Similarly, the Chortis Block is de- construction is therefore a dif®cult task, be- picted as overlapping southern Central cause the evidence needed to make paleo- America at 170 Ma, then acting as a bridge geographical reconstructions is almost inev- between southern Central America and North itably destroyed or modi®ed substantially America at 155 Ma, then coming into contact over geologically long periods of time. with the Maya (Yucatan) Block as both were Smith et al. (1994) pointed out that coast- uplifted around 148 Ma, and ®nally achiev- line reconstruction is additionally complica- ing its present-day position with respect to ted by the fact that different datasets, osten- North America by 80±0 Ma. The Yucatan sibly for the same interval, may yield quite Peninsula is successively depicted as (1) different paleogeographical results. This can overlapping South America at 170 Ma, (2) happen when stratigraphic data are collected uplifted and in contact with northern South for speci®c purposes, such as documenting a America by 155 Ma, and (3) situated near its general transgression or sealevel drop. For present-day position with respect to North example, reconstructions of Maastrichtian America in the Tithonian (148 Ma). Al- coastlines for the same area may look quite though these implied motions can be dis- different from one another, depending on missed as minor artifacts of mapping pro- what part of the interval and which events grams designed to show large segments of individual authors intended to depict. the geode, the point is that they seriously Another point about island geology versus con¯ict with all available models of Carib- island paleogeography can be made by ref- bean plate tectonics (Malfait and Dinkel- erence to Hedges' (1996b) claim that some mann, 1972; Donnelly 1985; Pindell and terrestrial environments in the Caribbean Sea Barrett, 1990; Pindell, 1994; Mann et al., may have been in existence since the end of 1995). Fine-scale interpretation of the paleo- the Cretaceous. Hedges (1996b: 166) stated geography of small areas is untenable with that there is ``no place in the West Indies that maps of this sort, although they have been is known by the presence of a continuous 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 13

sequence of sediments to have been emer- shallow-water promontory of the Chortis gent since the late Cretaceous, although some Block from Middle Eocene to Late Miocene areas of Cuba, northern Hispaniola, and pos- time, contra Sigurdsson et al. (1997) who de- sibly Puerto Rico, may have been.''3 It is un- pict it more correctly as a series of isolated clear what Hedges meant to convey by the carbonate banks. Cuba is shown as uplifted phrase ``continuous sequence of sediments,'' at 59, 21, and 10 Ma, but as completely sub- since the only conceivable contexts in which merged at 49 and 35 Ma, in con¯ict with the unbroken sedimentary accumulations might data analyzed by Iturralde-Vinent (1969, have occurred throughout the last 65 Ma are 1972, 1988a). deep oceanic basins situated on comparative- Pindell's (1994) maps are particularly ly ancient sea ¯oor (e.g., eastern Paci®c). problematic in their depiction of island arc Furthermore, his paleogeographical obser- connections. North and South America are vation has signi®cance only if it is addition- shown as being completely linked by island ally inferred that the hypothesized emergent arcs in the Valanginian, late Albian, Cam- lands of the late Mesozoic and early Ceno- panian, Maastrichtian, and Middle Eocene, zoic were incorporated while still subaerial and as nearly linked in the Barremian, Tu- into the developing Greater Antilles. We ronian, and Paleocene. Although these vol- know of no geological evidence that supports canic arcs certainly existed, reconstructing this inference; the little evidence that does them accurately as paleogeographical entities exist indicates that no terrestrial contexts requires detailed scrutiny of relevant evi- from these earlier periods survived as such dence. It is obvious from present-day geog- into the Late Eocene (see Paleogeography of raphy that volcanic arcs may form continu- the Caribbean Region: Evidence and Analy- ous subaerial entities (e.g., Kamchatka Pen- sis). insula, presently sutured to Chukotka; isth- Tectonic modeling and paleogeographical mus region of Central America) as well as reconstruction are not the same. island chains (e.g., Lesser Antilles, Kuriles). This point can be conveniently illustrated The life-span of an island qua island cannot by reference to Pindell's (1994) frequently be predicted from ®rst principles: islands at cited work. As part of a general tectonic re- the position of Krakatau and Aldabra, to cite construction of the Caribbean region, Pindell two quite different examples, have appeared (1994: ®g. 2.6a±n; see also Pindell and Bar- and disappeared more than once in the late rett, 1990) attempted to depict the paleogeo- Quaternary (Stoddart et al., 1971; Nunn, graphical history of certain physical features 1994). Without careful appraisal, contradic- (subaerial land, deep and shallow water, vol- tory conclusions may be reached on the basis canic arcs). Although Pindell's model utilizes of the same evidence. For example, Gayet et an extensive database, it contains a few inal. (1992: ®g. 1) argued on the basis of Pin- accuracies and unveri®able conjectures that dell's model (see Pindell and Barrett, 1990) have both tectonic and paleogeographical im- that ``the terrestrial bridge that linked North plications. For example, Florida and Baha- and South America by latest Cretaceous and mas are shown as comprising a single car- Paleocene times probably comprised the bonate platform from late Jurassic through Greater Antilles and the Aves Ridge which Late Miocene, which contradicts recent evi- consisted of a magmatic [arc] submitted to dence for their long-term separation (Austin uplift and deformation....'' By contrast, et al., 1988; Droxler et al., 1989; Hine, 1997; Hedges (1996a), also citing Pindell and Bar- Denny et al., 1994; Iturralde-Vinent et al., rett (1990), claimed that any possible con- 1996a). The Nicaragua Rise is drawn as a nection between North and South America via the developing ``proto-Antilles'' (or ``proto-Greater Antilles'') was sundered in 3 ``West Indies'' is de®ned by Hedges (1996b) to include the Greater and Lesser Antilles, Bahamas, and a the Late Cretaceous (70±80 Ma). However, number of small islands that lie immediately off the con- in actuality Pindell and Barrett (1990; see tinental shelves of Central and South America. It does also Pindell, 1994) took no position on the not include the southern Netherlands Antilles or Trinidad existence of land connections, as this issue and Tobago (cf. footnote 1). was ancillary to the topics they were consid- 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 ering. In any case, if numerous long-term last? (5) What were the surface areas of in- connections between North and South Amer- dividual lands, and how did their sizes alter ica had indeed existed (as might be inferred over time? As has already been made clear, from a literal reading of Pindell's [1994] such questions cannot be answered merely maps), evidence of this fact would surely by examining geological maps or tectonic re- have been found in the vertebrate fossil rec- constructions. ord; but it has not (Gingerich, 1985; Alva- Our paleogeographical maps provide in- rado, 1988; Webb, 1985; see also Biogeo- formation on four contexts: high-elevation graphical Hypotheses and Caribbean Paleo- and low-elevation terrestrial environments geography). (hereafter, ``highlands'' and ``lowlands''), Similar dif®culties attend the use of Per®t and shallow-water and deep-water marine and Williams' (1989) tectonic reconstruc- environments (``shallow marine'' and ``deep tions as paleogeographical evidence. In this marine''). These environments are distin- useful and incisive paperÐin several ways guished by certain diagnostic features, the intellectual precursor of the present among which positive or negative elevation monographÐmaps were intended to support relative to ambient sea level is the most sig- a critical discussion of Antillean biogeogra- ni®cant (see also appendix 1). Although pre- phy and paleogeography. However, their re- cise measures of elevation are not possible, constructions actually present little in the plausible benchmark estimates can be made way of physical geography, as land and sea within Ϯ1 order of magnitude. We de®ne are the only items discriminated. More prob- highlands as environments that existed at lematically, their maps (but not their text) positive elevations greater than ca. 200 m; give the impression that most of the Greater lowlands were less than 200 m. Although Antillean islands originated as such in the with good faunal evidence it is possible to late Cretaceous, and merely grew larger as distinguish marine environments very ®nely, they were tectonically transported to their we reconstruct only twoÐshallow marine, current relative plate position. In reality, the covering shelf conditions to Ϫ100 m; and Greater Antilles in their current guises are deep marine, embracing all sea-¯oor settings relatively young geographical features (see deeper than Ϫ100 m. appendix 1; Iturralde-Vinent, 1978, 1982, We emphasize that the contact line be- 1988a, 1994a). tween terrestrial and marine environments on From this brief review it is evident that the any given map should be thought of as a me- recent Caribbean geological literature is not dian value for coastline position during the (and was never intended to be) a source of interval being depicted. The accuracy of any ready-made, easily interpreted paleogeo- paleocoastline delimitation is, in any case, a graphical mapsÐthat is, maps speci®cally function of sedimentary exposure: in general, designed to trace the physical and positional paleocoastline positions can be traced more history of particular geographical entities. accurately within uplifted areas than in ones Appropriate design features for such maps that are currently below sea level. vary with the nature of the entities being Three quasi-independent parameters were traced. In the case of terrestrial environ- used in map construction: (1) geological con- ments, which constitute our special concern, stitution, (2) geographical positioning, and reliable paleogeographical maps of the Ca- (3) physical paleogeography. Geological ribbean region would help to both constrain constitution is the sum of those attributes of and enrich discussion of a range of signi®- a particular geological unit that are de®ned cant inquiries, among which are: (1) When by its composition, boundaries, and position did speci®c terrestrial environments with respect to other such units in the Carib- (``lands'') exist in the Caribbean region, and bean area or elsewhere. A geological unit can for how long? (2) Where were these lands be thought of as a time-bounded suite of geodesically located, at any given time pe- rocks that were formed by a particular set of riod? (3) What was the nature of physical geodynamic processes operating at a desig- connections between and among different nated location in the lithosphere. Typical ex- lands? (4) How long did such connections amples of such units might include a speci®c 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 15 section of oceanic or continental crust, a vol- a new stack, one having the distinctive com- canic arc, a set of genetically related sedi- posite nature of a foldbelt (a geological unit mentary basins, a foldbelt, or a block-terrane. in its own right, consisting of several amal- Each geological unit has its own ontogeny, gamated ``fossil'' precursors). Thus a new spanning its origin, evolution, and possible geography may arise ontogenetically from transformation into other units. Units are the old. named for convenience and book-keeping In creating reliable paleogeographical purposes, but the reader should be aware that maps, it has long been recognized that to de- names do not necessarily imply identity with termine the successive sizes of a geological geographical elements. For example, the Ca- unit, the effects of movement and deforma- ribbean Mountains of northern South Amer- tion have to be ®guratively undone. How ica, which now form an area of high relief, ``reverse ontogeny'' can be understood and contain fragments of oceanic crust and island worked out in a particular case of interest is arc. Both Beata and Aves are identi®ed as illustrated in ®gure 2. The two cross sections ``ridges,'' but geologically they are quite dif- in ®gure 2A represent the structure of the ferent. The Beata Ridge is a thick oceanic foldbelt in eastern Cuba and western Hispan- crustal unit, while the Aves Ridge is part of iola as seen today. In ®gure 2B, these cross an extinct volcanic arc (Holcombe et al., sections are restored to their relative posi- 1990). In the Caribbean region, many larger tions before separation caused by sinistral features, such as terranes, blocks, ridges, and movements along fault AЈ±A (steps 1 and 2). arcs, are minutely subdivided by faults. Fault This requires the deletion of entities that boundaries permit the delimitation of smaller have been intercalated as the result of move- entities that can be considered to have had ment along AЈ±A (Cayman Trench and semi-independent histories since faulting oc- southern and northern Hispaniolan blocks). curred (appendix 1, ®g. 1). With these omitted, it can be seen that the Geological units can be superimposed in a two cross sections can be precisely lined up single stack (i.e., in the same crustal posi- along their volcanic arc sequences (step 2). tion), as the following simple example illus- In step 3, we simplify the present-day rela- trates. A section of early Mesozoic oceanic tive position and width of the geological crust might evolve into a Cretaceous volca- units found in the foldbelt (carbonate conti- nic arc; later, both of these units might be nental margin, ophiolites, volcanic arcs, and deformed into a series of mountains and ba- sedimentary basins), depicting them as a se- sins. The fate of each unit implies a different ries of superimposed bars. In steps 4 and 5 paleogeographical setting (in this example we sequentially remove the effects of over- there are three: ocean ¯oor, volcanic island- thrusting and shortening due to internal de- arc, and fully terrestrial conditions with com- formation within the units themselves, there- plex relief). In this example only the last geo- by resolving the original width of the fold- logic unit can be said to be ``active'' or still belt. Although this example is schematic, it in its original geomorphological form. The makes the point that the geography of the other two units are no longer resolvable as present may differ radically from the geog- either ocean ¯oor or volcanic arc; they are raphy of the past, even for the ``same'' land part of the basement of the last unit. mass. A more dynamic example would illustrate For each geological unit under discussion, the point that, over time, segments of the positional coordinates in time (in Ma) and earth's crust can change their relationship, space (in degrees of latitude and longitude) form, and position, thereby affecting paleo- are provided in appendix 2. Unit positions geography. Along compressional plate were constrained per time-slice using various boundaries, for example, oceanic crust and sources of information, including maximum ridges may undergo subduction, thereby los- possible amplitude of strike-slip movements, ing their original geological and geomorpho- continuity of geological structures across dif- logical character. Similarly, ocean crust or ferent block-terranes or arc segments, pres- volcanic arc suites overthrusting continental ence of correlatable rock complexes of margins along a collisional suture will create known age in two or more distinct units, and 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 so forth (®g. 3). Plate motions provide an- position. These may include (1) sediments other, more general form of constraint (ap- deposited in terrestrial environments (e.g., pendix 2). red beds, alluvium), (2) weathering surfaces The third variable used to build paleogeo- or weathering products (e.g., paleosols, graphic maps is physical geography, which ``coated'' pebbles), (3) land-derived sedi- we de®ne as information (indicators) regard- ments in contiguous marine basins (e.g., con- ing positive or negative relief of contiguous glomerates, sandstones), (4) lagoonal depos- geological units (or their subcomponents) its representing fresh- or brackish-water near- across time. Relevant information is widely shore environments, (5) coastal sediments but thinly scattered in the geological litera- (e.g., beach sands, dunes), and (6) remains of ture. For effective use, this information had terrestrial organisms preserved in marine to be re®ned in various ways, including re- sediments (e.g., lignites, terrestrial plants, interpretation of the age of late Tertiary sec- pollen, spores). tions according to current paleontological Among marine indicators, different rock and stratigraphical criteria, reanalysis (if re- types and their fossil inclusions are of great quired) of depositional environment, deter- value because they are highly correlated with mination of topographic indicators, and so water depth at the time of original deposi- forth. tion. Lack of marine sediments in a particular A land indicator provides evidence of the section may be the result of erosion (within existence of subaerial conditions within a a hiatus), so whether a transgression actually geological unit at a speci®c time in its on- occurred has to be resolved by examining the togeny. However, land indicators have to be composition and environment of deposition interpreted carefully, as some are much better of rocks in surrounding basins. This is why than others. Thus, although unconformities some geological units are represented as sub- and hiatuses both represent gaps in the geo- marine environments in the paleogeographi- logical record, their signi®cance is not the cal maps, even though marine rocks of ap- same. An unconformity is a surface that sep- propriate age are not known within them. For arates two superimposed sets of strata. Un- example, although several hiatuses are re- conformities are erosional surfaces and may corded on the Beata Ridge, unequivocal in- provide clear evidence of land emergence if dicators of subaerial conditions have not associated with long-lasting hiatuses (but see been found. Accordingly, we assume that below). Although also a gap in the rock re- these hiatuses are due to submarine erosion cord, a hiatus is de®ned as time not repre- and nondeposition, and portray the ridge as sented by strata (i.e., an interval of nonde- a submarine feature from Late Oligocene to position, erosion, or both). Uplift and non- Recent (see ®g. 4). deposition of sediments during n1 Ma can Other phenomena may have paleogeo- produce a hiatus, but uplift also causes ero- graphical signi®cance if they provide insights sion of preexisting rocks, thereby addition- into speci®c conditions or occurrences in the ally enlarging the gap in the record by n2 past. For present purposes, the most impor- Ma. The total gap (n1 ϩ n2) is therefore pro- tant of these is termination of volcanic activ- duced by erosion as well as nondeposition. ity in island arcs. This phenomenon, usually Owing to the effects of bottom currents or due to arc±arc, arc±ridge or arc±continental rising sea ¯oor, hiatuses can also be produced collision (Hamilton, 1988), is known to have in the absence of subaerial exposure by sub- a profound effect on uplift. The mechanism marine erosion, nondeposition, or both of of uplift is related to the emplacement of these processes. Thus, a hiatus by itself does huge intrusive bodies coincidental with arc not imply land emergence, nor does the ac- extinction, causing widespread isostatic ad- tual gap (in millions of years) necessarily justment (uplift) along the arc axis (Iturralde- mirror the time during which the area was Vinent, 1988a, 1994a). Uplift is then fol- uplifted as land. lowed by subsidence within a period of only Usually, indicators other than gaps per se a few million years. Although postmagmatic- are required in order to reach reliable con- phase uplift is well substantiated, its perti- clusions about land emergence and coastline nence to Caribbean paleogeography has not 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 17

Fig. 4. Beata Ridge: top left, Current topography and approximate thickness of late Tertiary sediments (deep to basement); top right, stratigraphic columns (in the case of core sites, gaps in shading signify presence of hiatuses); and bottom, cross section (compiled from many sources; see appendices 1 and 2). Chronostratigraphic reference column in this and all other section diagrams after Berggren et al. (1995), at scale. For details on biozones and chronology consult ®gure 20. 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 been widely appreciated (but see Early Mid- and early Middle Eocene consistently pro- dle Jurassic to Late Eocene Paleogeography). duced substantial uplift followed by deep As discussed here in relation to the arcs as- erosion of the extinct volcanic arc edi®ces sociated with the evolution of the Caribbean (Iturralde-Vinent, 1988a, 1994a; MacPhee Plate, termination of arc magmatism in the and Iturralde-Vinent, 1994, 1995). Aptian, Late Campanian/Early Maastrichtian,

PALEOGEOGRAPHY OF THE CARIBBEAN REGION: EVIDENCE AND ANALYSIS This section provides an analysis of the type vegetation at several localities of ?Early/ several categories of basic geological infor- Middle Jurassic to early Late Jurassic age in mation presented in the tables, ®gures, and Mexico and western Cuba (Areces-Mallea, appendices. Appendices 1 and 2 and tables 1990). 1±4 should be consulted throughout for sup- In the North American portion of Laurasia porting evidence and additional literature not an important marine transgression took place directly referenced in the following para- during early to middle Oxfordian time. Vo- graphs. lant terrestrial and shallow-water marine vertebrates, indicative of proximate land envi- EARLY MIDDLE JURASSIC TO LATE ronments, make an appearance at this time at EOCENE PALEOGEOGRAPHY localities on the Guaniguanico terrane that forms westernmost Cuba (Iturralde-Vinent This long stage in the evolution of the Ca- and Norell, 1996). The occurrence of this ribbean region may be considered to have be- Oxfordian faunule in what is now Cuba con- gun with the creation of a basin, the embry- stitutes an example of ``Viking funeral ship'' onic Caribbean Sea, coincident with the emplacement (McKenna, 1973), because break-up of Pangaea and the separation of these taxa were extinct before Guaniguanico Laurasia from Gondwana (®g. 5; table 1). reached western Cuba long after detaching During the late Triassic/middle Jurassic, an from the Yucatan borderland early in the Ter- epicontinental siliciclastic basin developed tiary (Iturralde-Vinent, 1994a, 1996a; Bra- between the cratonic areas of South and North America in reaction to the eastward lower and Iturralde-Vinent, 1997). In the migration of the Tethys (Anderson and South American portion of Gondwana the Schmidt, 1983; Burke et al., 1984; Bartok, paleogeographical context was different, be- 1993). This epicontinental sea should not be cause the main transgression across the thought of as the Caribbean basin as pres- northern continental margin took place later, ently con®gured, but as a precursor situated during the Early Cretaceous. With the Ox- within western Pangaea (see Pindell, 1994: fordian transgression and widening marine ®g. 2.6a). Its epicontinental nature is dem- gap between Laurasia (North America) and onstrated conclusively by associated marine Gondwana (South America), any possibility invertebrate faunas and sediment composi- of direct, overland dispersal between these tion (Salvador, 1987, 1991; Pszczolkowski, continental areas ended (®g. 5). 1987). The developing Caribbean seaway (``His- The early Caribbean basin began as a nar- panic Corridor'' of Bartok et al., 1985) un- row seaway between the Paci®c and Tethys, derwent widening from the Middle Jurassic probably during the Bajocian/Bathonian to the Early Cretaceous as a consequence of (Bartok et al., 1985) as oceanic crust was be- sea ¯oor spreading (Pindell, 1994: ®g. 2.6c, ing formed between western Laurasia (North d). Oceanic crust and sediments formed at America) and western Gondwana (South that time are now represented in part by de- America) (®g. 5; Pindell, 1994: ®g. 2.6b, c). formed ophiolite bodies and thrust belts The existence of lands closely bordering this around the margins of the Caribbean region seaway is indicated by evidence of coastal- (Guatemala, Greater Antilles, Aruba/Tobago 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 19

Belt, Caribbean Mountains, and Colombian± Fm, a volcanic arc section in Hispaniola Venezuelan Andes) (Dengo and Case, 1990). (Smiley, MS, cited by Kesler et al., 1991b). Oceanic basalts associated with radiolarian This assemblage (including Gleichenites, cherts and carbonate rocks of Jurassic Zamites, Phoenicopsis, Yuccites, Podozami- through Cretaceous age have been found tes, and other taxa) is thought to have grown within these allochthonous crustal bodies in a warm, open, seasonally dry habitat ad- (Bartok et al., 1985; Montgomery et al., jacent to the shallow-water marine environ- 1994; Iturralde-Vinent, 1996a). In some plac- ment in which the remains were deposited es these vulcano-sedimentary rocks consti- (Kesler et al., 1991b). Other geological in- tute segments of a continuous section (e.g., dicators of emergence during the evolution northwestern Cuba, southwestern Puerto of the Cretaceous arc include terrestrially de- Rico), but most frequently they consist of posited volcanic rocks and several major un- isolated cobble- to boulder-sized bodies conformities occurring within volcanic arc within highly deformed belts. Although it sections, usually in association with hiatuses cannot be determined from existing infor- and basal conglomerates. mation whether the age gaps record actual Unfortunately, the paleogeographical in- hiatuses, the absence of any evidence of ter- formation content of these indicators is lim- restrial environments in these contexts ited and cannot be reliably used to provide a strongly suggests prevailing deep-water con- detailed assessment of areal extensiveness ditions from their origin until their incorpo- and orographic relief in the Cretaceous arc. ration into the foldbelts fringing the Carib- On the other hand, such information does bean region. provide some indication of the temporal suc- Indications of lands or shallow seas (or cession of environments in speci®c geologi- both) within the con®nes of the early Carib- cal units. In Hispaniola, the Neocomian bean sea are found in rocks of the Cretaceous plant-bearing rocks are overlain by marine volcanic arc. As a geological unit, the Cre- limestones of the Albian Rio Husillo Fm taceous arc is de®ned by a particular set of (Kesler et al., 1991b; Iturralde-Vinent, igneous, sedimentary, and metamorphic 1997a), implying that a transgression oblit- rocks of Neocomian through late Campanian/ erated previously existing terrestrial environ- early Maastrichtian age (Dengo and Case, ments. This sequence of events is rather 1990; Iturralde-Vinent, 1994a, 1994c, 1996a, 1996b). Today, elements of this arc are wide- common in the Cretaceous arc sections, and ly distributed in the foldbelts found within our interpretation is substantiated by two the Caribbean region (®g. 5). The paleogeo- pieces of evidence. First, volcanic and non- graphical position of the Cretaceous arc in volcanic marine rocks drape all of the un- relation to North and South America remains conformities recorded in the Cretaceous arc, the subject of debate (see Leclere and Ste- indicating that hiatuses were succeeded by a phan, 1985; Ross and Scotese, 1988; Don- new phase of marine deposition (Nagy et al., nelly, 1989a; Pindell, 1994, Mann et al., 1983; Lewis et al., 1991; Rojas et al., 1995; 1995; Hay and Wold, 1996; Iturralde-Vinent, Iturralde-Vinent, 1995, 1997a; Beccaluva et 1996a, 1997b). al., 1996). Second, the rudist limestones oc- Shallow marine environments are marked cur in the form of lenses intercalated within by the occurrence of rudist limestones of dif- other marine sediments and laterally transi- ferent ages, occurring as isolated, lenticular tional with isochronous deeper water beds intercalations within marine-deposited vol- (Rojas et al., 1995). This indicates that any canic sediments. Although the presence of intra-Caribbean land environments existing rudists is not diagnostic of nearby emergent on the volcanic arc at that time would have land, it is consistent with the existence of been limited in size, and therefore suscepti- atoll-like islands similar to those seen today ble to rapid obliteration during transgressive on shallowly submerged volcanoes. More phases. We conclude from this that, while substantive indications of land development land environments certainly existed in the in the Cretaceous island arc are Neocomian Caribbean Basin during the Cretaceous, they plant remains reported from the Los Ranchos were short-lived, probably winking in and 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 21 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 23 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

out of existence within periods of only a few Donnelly, 1989a; Iturralde-Vinent, 1994a, million years. 1997b). However, whatever its starting po- A good example of such evanescence can sition might have been, there is little evi- be seen in the section along the Canal Paso dence to support the view that the continents Bonito a Cruces, northwest of Sierra de Es- were physically united by the arc. In princi- cambray, Cuba (unpubl. obs.). Here, Creta- ple, connection might have occurred during ceous volcanic breccias and agglomerates are major uplift events recorded in association patchily overlain by slope and alluvial sands with various unconformities (between the and gravels which have been weathered to Neocomian and early Albian, between the paleosols that exhibit caliches, root casts, and Coniacian and Santonian, and in the early other indications of subaerial exposure. Campanian). However, information is lack- These rocks are in turn succeeded by un- ing concerning whether these erosional sur- weathered, well-bedded tuffs and rare marine faces were continuous or synchronous along limestones. Thus, however long this island the trend of the arc (Iturralde-Vinent 1994a, may have existed, it did not last as land into 1996b, 1997b). During the late Campanian later epochs. and early Maastrichtian (ca. 70±80 Ma) sub- Tracking the paleoposition (i.e., changes in stantial subaerial exposure existed along the latitude and longitude through time) of the Cretaceous arc and adjacent continental mar- Cretaceous volcanic arc is another issue of gins, as indicated by evidence of deforma- great importance. There appears to be a con- tion, angular uncomformities, hiatuses, deep- sensus, at least among authors of the most seated erosion, mountain building, and ter- widely cited plate tectonic models, that this restrial sedimentation (including conglomer- arc originated in the Paci®c (Malfait and ate and paleosol development) (Khudoley Dinkelmann, 1972; Burke et al., 1984; Le- and Meyerhoff, 1971; Mattson, 1984; Push- clere and Stephan, 1985; Pindell and Barrett, charovski et al., 1989; Maurrasse, 1990; 1990; Pindell, 1994), although other inter- Lewis et al., 1990; Iturralde-Vinent, 1994a, pretations also exist (®g. 5; Turner, 1972; c, 1995, 1996b, 1997b; Beccaluva et al., 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 25

Fig. 5. Plate tectonic model of the Caribbean region for Jurassic through Eocene (after Iturralde- Vinent, 1997b, slightly modi®ed). Maps are designed to display tectonic information; they are not paleogeographically accurate. 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

1996). At this time magmatic activity ter- the formation of the Paleogene volcanic arc. minated along large segments of the volcanic However, magmatism was short-lived in this arc, including its western and eastern extrem- new arc, effectively ending by the Middle ities (i.e., modern central and western Cuba Eocene. Within the Paleogene arc, marine in the north and west, and the Netherlands sediments uninterruptedly ®lled some basins, and Venezuelan Antilles and Caribbean indicating that land contacts between arc Mountains in the south and east). Cessation components and the continents were not (or of the magmatic phase is coincidental on a were no longer) in existence (Lewis and wider scale with the Subhercinian orogeny Straczek, 1955; BresznyaÂnszky and Iturralde- (Schwan, 1980; Leonov and Khain, 1987), Vinent, 1985). Uninterrupted sedimentation an important global tectonic event, and sev- from Paleocene through Middle Eocene is re- eral drops in eustatic sea level (Haq et al., corded in basins located on the Caribbean 1987). It is reasonable to infer that if any sea¯oor (Edgar et al., 1973; Sigurdsson et al., land connection existed between North and 1997), as well as in foldbelt areas, along the South America in the last part of the Creta- trend of the inactive segments of the Creta- ceous, this connection most probably would ceous arc (e.g., deep marine sediments out- have occurred during the late Campanian and cropping through much of La Habana and early Maastrichtian. However, if in fact con- Matanzas provinces in west central Cuba) tact occurred it would have been brief, be- (Bronnimann and Rigassi, 1963; Pszczol- cause transgressive late Maastrichtian marine kowski, 1987; BresznyaÂnszky and Iturralde- sediments are recorded in the Cretaceous Vinent, 1985; Bralower and Iturralde-Vinent, volcanic arc as well as in North and South 1997). Furthermore, a major transgression America (table 1; appendix 1). occurring between the late Early and Middle There does not appear to be any reliable Eocene produced extensive deposits of shal- evidence of permanent islands or island±con- low- and deep-water marine carbonate rocks tinent connections in the Caribbean region on previously positive areas throughout the during the early part of the Paleogene, al- Caribbean (Lewis and Straczek, 1955; Bron- though we cannot reject this possibility al- nimann and Rigassi, 1963; Iturralde-Vinent, together (see Biogeographical Hypotheses 1982, 1994a; Holcombe et al., 1990; Lewis and Caribbean Paleogeography). An uncon- et al., 1990; Maurasse, 1990; Edgar et al., formity dated to the K/T boundary is known 1973; Sigurdsson et al., 1997). The occur- to exist in some parts of the Caribbean sea rence of this transgression militates against ¯oor, but uninterrupted sedimentation evi- there having been any considerable exposure dently continued elsewhere in the Caribbean of land during the early Middle Eocene. Basin, even within the area of the inactive Another global tectonic event, the Illyrian Cretaceous arc. This unconformity could phase of tectogenesis (Leonov and Khain, have been caused by submarine erosion rath- 1987), produced a profound modi®cation of er than by cessation of deposition if, for ex- the Caribbean tectonic regime between Mid- ample, it was induced by tsunamis triggered dle and Late Eocene (®g. 5). Extensive uplift by the K/T impactor landing in Chicxulub and deformation occurred not only in the Ca- (Pszczolkowski, 1986; Maurrasse and Sen, ribbean Basin per se but also in the surround- 1991) and/or in the Yucatan Basin (Iturralde- ing continental margins and oceanic domains Vinent, 1992). (Khudoley and Meyerhoff, 1971; GonzaÂlez After a period of relative quiescence, vol- de Juana et al., 1980; Mattson, 1984; Lewis canic activity began again in the Caribbean et al., 1990; Iturralde-Vinent 1981, 1994a, region in the Paleocene (Iturralde-Vinent, 1994b, 1994c, 1996a; Maurrasse, 1990). 1994a, 1997b), as indicated by the occur- With reorganization of the geodynamic re- rence of a set of magmatic, sedimentary, and gime, the relative motion of the Caribbean metamorphic rocks that widely outcrop in the Plate shifted eastward, and relative motion region (eastern Cuba, Jamaica, Hispaniola, between the North and South American Puerto Rico, Virgin Islands, Cayman Ridge, plates decreased markedly (Pindell, 1994). In Nicaragua Rise, and Aves Ridge/Lesser An- concert with these developments, several de- tilles). This resumption of activity ushered in formed belts were consolidated and accreted 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 27 against continental areas, and magmatic ac- vision and separation of block-terranes pre- tivity shifted to new locations (Central viously acting as continuous landmasses. America and Lesser Antilles). The north- This subdivision may have been signi®cant western edge of the Caribbean Plate (NW biogeographically if it caused island±island Greater Antilles Belt) collided with the Yu- vicariance (as opposed to continent±island catan and Bahamas margins, while the south- vicariance; see Biogeographical Hypotheses eastern edge (Aruba/Tobago Belt) interacted and Caribbean Paleogeography). with the South American margin. Also, beginning in the latest Eocene, new tectonic el- EOCENE±OLIGOCENE TRANSITION (35±33 Ma) ements were de®ned within the Caribbean re- The transition between the end of the Eo- gion. Transverse faulting divided the plate cene and the beginning of the Oligocene into several microplates and block-terranes, (zones P16 to P18 of Berggren et al., 1995) and their subsequent displacement disrupted coincides with the Pyrenean phase of tecto- the original structure (see appendix 2). genesis (Schwan, 1980; Leonov and Khain, In summary, existing data indicate that 1987), the effects of which are well repre- subaerial entities were formed along the Cre- sented in the Caribbean region (MacPhee and taceous and Paleogene volcanic arcs and Iturralde-Vinent, 1995). In this phase, gen- nearby continental margins from time to time eral tectonic uplift coincided with a major from the Jurassic into the Eocene. However, eustatic sea level drop at ca. 35 Ma (Miller there is no evidence that any of these entities et al., 1996). As a result, subaerial exposure lasted for long periods; indeed, none seems within the Caribbean basin was probably to have survived as emergent land into the more extensive then than at any other time subsequent interval (latest Eocene to Middle in the Cenozoic, including the late Quater- Miocene). Nevertheless, if the Cretaceous arc nary. The map in ®gure 6 re¯ects this fact ever connected North and South America, (see also table 2). However, it is important to this most likely occurred during the late compare this map with other ``Oligocene'' Campanian to early Maastrichtian (ca. 70±80 reconstructions which represent this period Ma ago), just after extinction of the Creta- as one of overall minimum land exposure ceous volcanic arc. (e.g., GonzaÂlez de Juana et al., 1980; Gallo- way et al., 1991; Macellari, 1995). The ex- LATEST EOCENE TO MIDDLE planation for the difference in treatment lies MIOCENE PALEOGEOGRAPHY in the fact that most Oligocene reconstruc- In this section, three ``snapshot'' intervals tions depict the paleogeography of the mid- are discussed in detail: Eocene±Oligocene to later Oligocene, by which time the Pyre- transition (35±33 Ma), Late Oligocene (27± nean orogenic phase had terminated (see dis- 25 Ma), and early Middle Miocene (16±14 cussion of next map). Ma). Basic data are presented in appendix 1 Evidence for Pyrenean uplift can be seen (see also ®gs. 6±8 and tables 2±4). These in- in stratigraphic sections as well as submarine tervals were chosen in order to contrast pe- dredge samples, drill cores, and seismic lines riods of maximum and minimum land de- recovered from many parts of the Caribbean velopment. The Eocene±Oligocene transition and surrounding continental borderlands (ta- was a time of general uplift; therefore, the ble 2; appendix 1). Stratigraphic sections amount of subaerial land in the Caribbean consistently lack marine sediments of latest should have been at a maximum. The Late Eocene±Early Oligocene age, presenting in- Oligocene was a time of high sea level, and stead hiatuses, red beds, and other kinds of therefore of mimimum exposure (and, prob- terrestrial deposits (table 2; appendix 1). In ably, interconnectedness) of emergent areas many sedimentary basins located near former in the early Cenozoic. In the early Middle topographic highs, latest Eocene/Early Oli- Miocene, further isolation of land areas took gocene sediments carry abundant land-de- place as a consequence of active tectonic dis- rived debris, chie¯y very coarse conglomer- ruption of the northern and southern Carib- ates (matrix or clast-supported) and sand- bean Plate boundaries. In the case of the stones. This type of sedimentary unit is so Greater Antilles, this resulted in the subdi- common that it may usefully be dubbed the 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 29

``Eocene±Oligocene transition conglomerate rate is in agreement with the combined ef- event.'' In places far removed from terrestrial fects of lowered sea level and restricted sea- sediment sources, such as basins in the Ca- water circulation (Haq et al., 1987). ribbean sea ¯oor, deep-marine deposition The Aves Ridge deserves special mention was condensed (i.e., pelagic sedimentation because it has been proposed as the site of a occurred at a low rate). Low sedimentation potential landspan between the Greater An- 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Fig. 6. Paleogeography of Caribbean region as reconstructed for latest Eocene/Early Oligocene (35±33 Ma). The reconstruction is designed to depict conditionssealevel obtaining and in widespread the regionalthis Caribbean uplift period, region (Pyrenean depicted during orogenesis). landspanbetween the Although shorelines the period subaerial and Southern conditions of contact Hispaniolan existed maximum point and along Cenozoic with Blue at land South Mountains least America exposure, Blocks part should occasioned of (for be by the explanation, regarded Aves very see as Ridge low text conjectural, during eustatic and as appendix is 1). the illustrated connection 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 31 tilles and northern South America (Borhidi, imentary basins of the Caribbean sea ¯oor, 1985; MacPhee and Iturralde-Vinent, 1994, the rate of sediment accumulation increases 1995). This structureÐpresently almost com- at this time and deeper water environments pletely submergedÐwas originally continu- upwardly dominate Oligocene sections. The ous with the Greater Antilles Ridge and is amount of land-derived debris in such sec- considered to have constituted a single entity tions is substantially reduced, with ®ne-grain in latest Eocene/Early Oligocene time. We and biogenic deposits dominating. Neverthe- argue that exposure of the ridgecrest created, less, heights-of-land remained persistently for a short time ca. 33±35 Ma, a series of subaerial along the axis of GAARlandia, as large, closely spaced islands or possibly a shown by the existence of rocks formed in continuous peninsula stretching from north- nonmarine environments, presence of land- ern South America to the Puerto Rico/Virgin derived sediments and plant remains in prox- Islands Block (see GAARlandia Landspan imate marine basins, and depositional hiatus- and Island±Island Vicariance). es in sections. These results suggest that, in Among the points in the paleogeographi- an area such as the Caribbean region, in cal reconstruction that require further re®ne- which vertical motions have been marked, ment and explanation are the close position- the amount of terrestrial exposure or inun- ing of southwestern Hispaniola and the Blue dation cannot be simply read out from sea- Mountains Block, as well as inferred per- level curves (e.g., Haq et al., 1987). manent exposure of parts of Jamaica as early as 33±35 Ma (see appendix 1). Also note- EARLY MIDDLE MIOCENE (16±14 Ma) worthy is the fact that western Cuba would The early Middle Miocene paleogeogra- have been separated by deep-water environ- phy (zones M5±M7 of Berggren et al., 1995) ments (Havana±Matanzas Channel) from of the Caribbean region shows the effects of central and eastern Cuba at this time. disruption of the deformed foldbelt bounding Other features characteristic of the present the Caribbean Plate (®g. 1; table 4). The pro- Caribbean sea ¯oor that did not exist in Eo- cess of Neogene disruption, the tectonic cene±Oligocene times (and are therefore not characteristics of which have been extensive- depicted) include the Cayman Trench and ly investigated (Mann et al., 1990; Pindell Anegada Trough, among others (see appen- and Barrett, 1990; Pindell, 1994; Macellari, dix 1) (Calais et al., 1989, 1992; see discus- 1995), is recorded east of Cuba along the sion by MacPhee and Iturralde-Vinent, 1994, plate's northern boundary, and within the 1995). Netherlands and Venezuelan Antilles to Trin- idad and Tobago along its southern bound- LATE OLIGOCENE (27±25 Ma) ary. Localized extension occurred along both boundaries as grabens, pull-apart basins, and The Late Oligocene (zones P21b±P22 of trenches began to form. This, combined with Berggren et al., 1995) was a time of exten- continuing marine transgression, served to sive marine invasions, probably due to a further isolate fault-bounded block-terranes combination of tectonic subsidence and high from one another along plate boundaries. Ex- sea level stands (table 3). Marine sediments amples of extensional features formed or ac- of this age are common in North and South tivated at this time in the central Caribbean America and the Greater Antilles (®g. 7). In- Basin include the Cayman Trench between undation of terrestrial environments began as Cuba and Hispaniola, the Anegada Trough early as zone P19 (Berggren et al., 1995) and between the northern Virgin Islands and St. continued into zone P22. Evidence of Late Croix/Aves Ridge, the Puerto Rico Trench Oligocene transgression is seen in strati- (Mann and Burke, 1984; Mann et al., 1990), graphic sections, dredge samples, drill cores, and the Tortuga Basin (®g. 1). Signi®cantly, and seismic lines recovered from many parts extension also took place along the axis of of the Caribbean region and the Florida the Nicaragua Rise (Droxler et al., 1989) Block. Stratigraphic sections consistently re- Figure 8 shows Puerto Rico and Hispan- cord mid-Late Oligocene marine deposits iola as being connected by a neck of land overlying older rocks. In the more distal sed- into the Miocene. The existence of this con- 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 33 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Fig. 7. Paleogeography of Caribbean region as reconstructed for Late Oligocene (27±25 Ma). During this period, general subsidence and higher sealevels greatlynorthwestern diminished South America land (for area explanation, within see the text and Caribbean appendix Sea, 1). sundering the landspan connection between the Greater Antilles and 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 35

nection is indicated, although not proven, by the amount of land in the central part of the evidence that (1) the formation of the Mona Caribbean Basin was approximately the same Passage was a neotectonic event (Larue et in the Late Oligocene and the early Middle al., 1990; Jany et al., 1990; Masson and Miocene. However, by the Middle Miocene, Scanlon, 1991), and (2) the ®rst late Ceno- block-terranes fringing the Caribbean Plate zoic marine sediments to onlap eastern His- were already widely separated; many of paniola and western Puerto Rico were Pleis- these would never be reunited again, even tocene limestones, implying the existence of during the glaciations (low-stand events) of a barrier of some kind. Additional data from the late Quaternary. Late in the Middle Mio- the ¯oor of the Mona Passage would help to cene, western Cuba ®nally achieved dryland clarify the history of this connection. contact with central Cuba after the disap- Comparison of ®gures 7 and 8 reveals that pearance of the Havana±Matanzas Channel.

BIOGEOGRAPHICAL HYPOTHESES AND CARIBBEAN PALEOGEOGRAPHY This section is not intended as a review of vation was to wed modern ideas concerning the recent systematically oriented literature the relationships of whole faunas and areas on Caribbean biogeography. Such an under- (see Croizat, 1964; Nelson and Platnick, taking, if properly comprehensive, would be 1981; Humphries, 1992; Morrone and Crisci, a major undertaking unto itself and therefore 1995) to the emerging theory of plate tecton- requires separate treatment (MacPhee and ics, for one of the world's most complicated Iturralde-Vinent, in prep.). Our narrower pur- biological and geological regions. Vicariance pose in this paper is to examine speci®c theory has been critically explored with spe- problems in light of the new paleogeograph- cial reference to Antillean faunas in several ical reconstructions developed in preceding recent works (e.g., Guyer and Savage, 1987, sections, with an emphasis on ``how they did 1992; Kluge, 1988, 1989; Page and Lydeard, it'' as opposed to ``who did it.'' We pursue 1994; Roughgarden, 1995). In the main, this by considering, in turn, three quite dif- however, cladistic biogeographers have not ferent models proposed by Rosen (1975, concerned themselves with updating or re- 1985), Hedges and co-workers (1992, 1994; vising Rosen's (1975: 453-454) paleogeo- Hedges, 1996a, 1996b), and MacPhee and Iturralde-Vinent (1994, 1995). graphical inferences, despite his exhortations that they do so. CONTINENT±ISLAND VICARIANCE: In Rosen's model, the ``Antillean archi- MODEL OF ROSEN pelago'' or ``proto-Antilles''Ðhere under- Rosen's (1975, 1985) continent±island vi- stood as the Cretaceous and Paleocene±Eo- cariance model was the earliest attempt to cene volcanic arcs, although he did not make create a cladistically oriented biogeography this distinctionÐwere assumed to have orig- of the Caribbean region with an emphasis on inated as a series of closely spaced islands vertebrates.4 Rosen's (1975) principal inno- on the leading edge of the Caribbean crustal plate in roughly the position occupied by 4 As a matter of historical record, the ®rst intuitive present-day Central America. As these is- model of Antillean vicariance was proposed by J. Issac lands were tectonically transported eastward, del Corral (1940) in order to explain the origin of the they interacted with adjacent continental Cuban mammal fauna by Wegenerian ``continental margins in such a manner that they were able drift.'' According to this author, before the Late Miocene to receive the greater part of their biota in the Greater Antilles were attached to northern South essentially one event (Rosen, 1975: ®g. 8). America (Colombia and Venezuela), whence they re- As Per®t and Williams (1989) pointed out, ceived their mammalian fauna. Later on, the islands drifted northward to their current positions. Vandel although Rosen (1975) described this com- (1973) used Issac del Corral's (1940) model to account mon-cause event as vicariant in nature, he for the relationship of troglobytic faunas in Cuba and was ultimately noncommittal as to how im- northern South America. migration actually occurred. (Rosen's maps 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 37 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Fig. 8. Paleogeography of Caribbean region as reconstructed for early Middle Miocene (16±14 Ma). Neogene tectonic movements and deformations along thetectonic margins of blocks the andand Caribbean terranes South Plate separated subdivided America byappendices former not 1 deep-water structural in and gaps ridges contact 2). (Greater (e.g., across Antilles Cayman and Panamanian Trench, Aves region Mona Ridges), (Duque-Caro, Passage, creating 1990; Anegada isolated Kolarsky Trough). et Southern al., Central 1995a, America 1995b; for explanation, see text and 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 39 imply that some short-distance, overwater gin terranes (Iturralde-Vinent et al., 1996a). transport would have been required, which of As we have emphasized, any islands then ex- course muddies a primary distinction be- isting along the axis of the developing fold- tween vicariance and dispersal.) Subsequent belt would have been intensely modi®ed by plate motions, Rosen argued, carried the is- thrusting, folding, subsidence, and burial be- lands and their resident faunas to their cur- neath thick tectonic nappes. Therefore, the rent locations. During this postvicariant in- geography of the ``proto-Antilles'' (i.e., the terval, amounting to all of the Cenozoic, the Cretaceous and Paleogene volcanic arcs) faunas of the islands were further shaped by probably has little or nothing to do with the extinction, local radiation, and, in a small geography of the existing islands (Iturralde- number of cases, by later overwater dispers- Vinent, 1982). als (Rosen, 1975: ®gs. 9±12). He ended this This point was also missed by Guyer and section of his paper (pp. 453±454) with a list Savage (1987, 1992), who in some respects of nine ``general remarks'' on his ``combined went even further than Rosen did in assum- vicariant and geophysical model,'' remarking ing the permanency of islands. Guyer and that he incorporated ``many provisional ideas that urgently need independent testing.'' Savage (1987: 526) proposed that, with the In assessing Rosen's (1975) views in light exception of a very few recent dispersals, the of modern tectonic theory, it should be ap- ancestors of the anole faunas of the large is- preciated that the tectonic model that Rosen lands were emplaced all at once, and diver- utilizedÐbasically that of Malfait and Din- si®ed there on ``the remnants of the once kelmann (1972)Ðis not very different in its more-or-less continuous land connection (the conceptual framework from current models proto-Greater Antillean block) that originally of the evolution of the Caribbean Plate (e.g., lay between North and South America during Pindell, 1994). Instead, the essential problem late Cretaceous to early Tertiary.'' lies with Rosen's paleogeographical scenar- Although Rosen (1975) thought that fossil io: he simply assumed an identity relation- evidence would prove to be of great impor- ship between geological units and geograph- tance for testing vicariance, at that time the ical entities in his discussion of the origin Antillean paleontological record was of little and early history of the ``proto-Antilles,'' as value for testing hypotheses of vicariance be- becomes obvious when his maps are exam- cause it was almost exclusively Quaternary ined (on this point see also Rosen [1985] and (MacPhee and Wyss, 1990). In the years comments by Per®t and Williams [1989] and since Rosen wrote, the vertebrate fossil re- Hedges [1996a]). In short, he viewed the pa- cord has improved marginally, but not to the leoislands which existed in the position of extent that it can provide a critical test. As Central America in the Cretaceous as some- noted by MacPhee and Wyss (1990), a strong how the ``same'' as the ones in existence to- test of vicariance would require the discov- day, as the ``transposed, original archipelago, ery of (1) numerous representatives of con- the Antilles'' (Rosen, 1975: 432; see also tinental lineages (2) of the same or similar Rosen, 1985: 652). age that are (3) not simply early members of This view of paleogeographical continuity clades represented in the Antillean Quater- across 80 Ma or more of earth history is fun- nary. damentally ¯awed, because it does not take into consideration the effects of tectonic pro- As to (1) and (2), the discovery that a rhin- cesses on Antillean biogeography between ocerotoid perissodactyl (Hyrachyus sp.) lived the late Campanian and Recent (see Iturral- on an Early Eocene landmass now incorpo- de-Vinent, 1982). As noted above, the fold- rated into present-day western Jamaica belt which constitutes the geological base- (Domning et al., 1997) is certainly important ment of the present-day Greater Antilles was because it establishes that Rosen's mecha- created by complex phases of amalgamation nism (continent±island vicariance) may in- and deformation of rock units comprising the deed occur. However, to date this discovery Cretaceous and Paleogene arcs, together with remains the only fossil-based evidence for associated oceanic crust and continental mar- Rosen-style vicariance in the Antillean ver- 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 tebrate record.5 However fascinating its dis- important mechanism of faunal formation in covery may be on other grounds, one Eocene the West Indies. A centerpiece of their ar- rhinocerotoid from Jamaica is of limited ex- gument is that ``all groups examined had planatory signi®cance. If this species lived lower estimates of divergence than would be on a terrane that was part of Central America predicted by proto-Antillean vicariance, sug- prior to the latter's separation and amalgam- gesting an origin by over-water dispersal in ation with other terranes represented in mod- mid- to late Cenozoic'' (Hedges, 1996a: 97). ern JamaicaÐas is indeed possible, given its Although hypotheses other than Rosen-style, substantial ageÐthen it quali®es as a vicar- mid-Cretaceous vicariance are brie¯y refer- iantly emplaced taxon in the very sense that enced (e.g., that of MacPhee and Iturralde- Rosen (1975) intended. There may be more Vinent, 1994, 1995), discussion is otherwise such taxa; the only way to ®nd out is to pros- essentially bipolar: if classic continent±island pect in the correct contexts on other islands. vicariance can be rejected, it seems, over- Other mammalian fossil discoveries made water dispersal must be correct. With respect in recent years include a ?megalonychid to historical arguments no such certainty is sloth femur from an Early Oligocene context ever possible, and this is decidedly the case in southwestern Puerto Rico (MacPhee and with Antillean vertebrate colonizations. The Iturralde-Vinent, 1995), a possible insecti- later papers in the series by Hedges and co- vore in Early Miocene Dominican amber workers introduce minor updates of the al- (MacPhee and Grimaldi, 1996; see also Itur- bumin data and their interpretation. Graphic ralde-Vinent and MacPhee, 1996), and vari- representations of times of origin also differ ous remains attributable to a platyrrhine pri- in various, sometimes subtle ways (cf. Hed- mate, a capromyid rodent, and another me- ges et al., 1992: ®g. 1; Hedges et al., 1994: galonychid from Early Miocene Domo de ®g. 2; Hedges, 1996a: ®g. 2). Zaza, central Cuba (MacPhee and Iturralde- Page and Lydeard (1994) have compre- Vinent, 1994, 1995). Although these discov- hensively discussed a number of systematic eries signi®cantly extend the insular records and interpretative issues raised by Hedges et of several higher-level mammalian taxa to al. (1992); Hedges et al. (1994) should be the early Neogene/late Paleogene, all of them consulted for replies to their criticisms. Much lie within clades that survived into the An- of the discussion in these papers is beyond tillean Quaternary and are therefore not un- the scope of the present investigation and ambiguously representative of a ``continen- will not be summarized here. However, other tal'' faunal aliquot. Reptile fossils have been points are clearly pertinent. These may be found in Tertiary contexts on the islands, but grouped under three headings: (1) prelimi- few have been adequately published. nary issues, (2) sources of error in estimating times of lineage origins, and (3) passive PASSIVE OVER-WATER DISPERSAL: transport and Cenozoic surface±current pat- MODEL OF HEDGES AND terns. CO-WORKERS In several recent papers, Blair Hedges and PRELIMINARY ISSUES his colleagues (Hedges et al., 1992, 1994; Hedges' basic database is impressive, but Hedges 1996a, 1996b) have argued that anal- partitioning it in different ways brings out ysis of immunological distances among a va- some of the underlying uncertainties and am- riety of reptiles and amphibians reveals that biguities in his presentation of the informa- overwater dispersal has been by far the most tion. Some of these are detailed in the following paragraphs, but others would require 5 Additional, albeit nonvertebrate, fossil evidence for a larger investigation than we have the com- vicariance is provided by the diverse ant fauna recovered from Early Miocene Dominican amber. This fauna has petence to pursue. a more continental character than the extant Hispaniolan (1) Number of lineages analyzed. Hedges fauna and includes forms never known to occur on oce- (1996a: 113) concluded that the ``major ®nd- anic islands (see commentary by Mayer and Lazell, ing of this analysis is that all but one or two 1988: 1477). of the 77 independent lineages of amphibians 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 41

and reptiles . . . [sampled for] the West Indies (Pindell and Barrett, 1990)'' (Hedges, 1996a: apparently originated by dispersal in the Ce- 104). An analogous induction is made for the nozoic.'' In fact, age-of-origin estimates are teiid Cnemidophorus vanzoi of the Maria Is- speci®cally made for only 72 lineages, not lands (St. Lucia). Origin bandwidths of this 77; we use the corrected ®gure in computing magnitudeÐessentially equivalent to all of proportions in subsequent paragraphs. Line- the post-Paleocene CenozoicÐcould be con- ages lacking such estimates are Hyla heilpri- sistent with practically any biogeographical ni, Phyllodactylus wirshingi, Mabuya lineo- hypothesis, not just dispersal. lata, the Leptotyphlops bilineata group, and (3) Taxa are not discriminated in terms of Geochelone sp. (described as `ùnknown'' in interpretative signi®cance. Even if it were the text although his table 3 presents an es- accepted that the leading cause of faunal for- timate of ``0±2?'' Ma). Lineages that cannot mation in the West Indies was overwater dis- be dated as to origin cannot contribute to the persal, this phrase as usually understood cov- argument that their emplacement must have ers several quite different mechanisms. The been essentially random with respect to majority of lizards, for example, would pre- time.6 sumably not be capable of dispersing by (2) Mixture of morphological and immu- long-distance swimming (i.e., self-powered nodiffusion data. Although Hedges et al. overwater dispersal, unassisted by rafting, (1992) presented ID evidence for divergence palm ``boot'' transport, or other classically time for a number of taxa, for many otherÐ invoked means). By contrast, species of croc- more than 40 (56%) in Hedges' (1996a) most odilians (Crocodylus) and chelonians (Geo- recent analysisÐID data are not provided chelone, some pelomedusids) that are able to and morphological divergence is used instead tolerate saltwater conditions could have at- as a proxy measure. Although these latter tained their known distributions under their data are separately analyzed, their value own power, whether their swimming was di- seems to us to be incidental to demonstrating rected or not. Their propagules could have the validity of the main argument (i.e., that dispersed by rafting as well, but including Rosen-style vicariance is falsi®ed by the lack taxa that are inherently ambiguous as to of temporal patterning in divergence times as probable method of dispersal adds nothing to estimated by the ID data). Since there is no resolution of the debate. The pertinence of linear clock that can be applied to rates of this point becomes obvious when viewed in morphological divergence, estimated times light of the problems with Hedges' (1996a) of lineage origins can only be expressed in evaluation of surface-current ¯ow in the Ca- the broadest terms. Thus Gymnophthalmus ribbean Sea (see below). pleei, an endemic Lesser Antillean teiid, is (4) Overrepresentation and ambiguous viewed on morphological grounds as a close signi®cance of nonendemics. The bulk of relative of northern South American G. li- Hedges' (1996a) taxon list (37/72 ϭ 51%) neatus. Time of origin (by dispersal) is listed consists of lineages that are de®ned as ``non- as 0±45 Ma, the lower limit being based ex- endemics'' (species having populations on clusively on an estimate of the ``geologic or- the mainland as well as one or more West igin of the Lesser Antilles in the Eocene Indian islands). Hedges assumed that the existence of allopatrically distributed popula- 6 Gregory C. Mayer (personal commun.) has called tions of single species is good evidence that our attention to three errors in enumerated entries in some island colonizations occurred so re- Hedges' (1996a) table 3: (1) 11, Crocodylus intermedi- cently that source populations have not dif- us, known from only one or two vagrant individuals, ferentiated from colonizing ones. From a di- cannot be considered to be established in the West In- versity standpoint, the number of recent (0± dies; (2) 26, Iguana iguana does not occur on the Cay- 2 Ma) lineages is certainly overstated relative man Islands; and (3) 41, Mabuya bistriata is presumably a lapsus for Mabuya mabuya; M. bistriata is a Brazilian to their importance. However, from a mech- species that is unlikely to be conspeci®c with anything anism standpoint, if each colonization were in the West Indies. Dropping these taxa from the list will an event separate from every other, the rate also affect the count, but we have not made this correc- of dispersal must have been higher in the re- tion here. cent past (assuming that extinctions can be 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 ignored; see below). Even so, the mechanism their method of transport to the Greater An- of emplacement need not have been always tilles (see above). However, in terms of his- the same. For example, although continent± torical biogeography, the real issue is the island vicariance can be excluded from con- comparative lack of nonendemic populations sideration, for very recent introductions de- in the Greater Antilles in contrast to the Less- termining whether range extensions were due er Antilles and islands on the fringe of the to natural dispersal or to anthropogenic trans- Caribbean Sea. How can this pattern be ex- port cannot be settled merely by observing plained? that the latter is assumed to be less likely. Not knowing whether the six Greater An- Hedges (1996a: 99) noted that lineages ``that tillean herp lineages noted by Hedges were clearly are the result of human introduction representative of many more that he could are mentioned but not treated in the analy- have cited, we examined all distributions list- ses.'' However, taxa that are ambiguous in ed by Schwartz and Henderson (1991) in this regard, such as Gonatodes albogularis, their exhaustive catalog. In addition to those Hemidactylus haitianus (ϭ H. brookii haiti- already discussed, and resolving cases of tax- anus of other authors), and H. mabouia, are onomic doubtfulness in Hedges' favor, there nevertheless included by implication in var- are only eight other apparent nonendemics ious statements to the effect that ``nearly (all saurians) whose distribution includes a all,'' ``99%,'' and ``virtually all'' lineages mainland and one or more of the Greater An- originated by natural overwater dispersal. tilles (®ve species of Anolis and three of The degree of human involvement in pro- Sphaerodactylus). We cannot comment on ducing some herp distributions will probably whether any of these may qualify as addi- never be known except by determining from tional `ìndependent lineages''; judging from fossil evidence that the taxon existed in the Hedges' comments under individual genera, West Indies well before the earliest presumed he seems uncertain as well. However, it date of human arrival (ca. 7000 yrbp; Burney seems reasonable to conclude that the num- et al., 1994). Currently, however, the pale- ber of ``Quaternary dispersals'' to the Great- ontological record is of little assistance in er Antilles counted according to Hedges' this regard as few of the herp taxa listed by methodology amounts to only a fraction of Hedges as having a Quaternary origin have those that can be enumerated for the other independent, associated radiometric dates islands, even though the Greater Antilles from sites in the West Indies (cf. Morgan and comprise 90% of the land area in the West Woods, 1986; MacPhee et al., 1989). Al- Indies. It may be that the herpetofaunas of though it can be argued, trivially, that in- small islands, with their low endemism, instances of anthropogenic dispersal still count dicate that dispersal is possible; but the fau- as dispersals, in a paper designed to test sce- nas of the larger islands, with their very high narios of faunal formation it is surely prob- rates of endemism, show that colonization is lematic to overrepresent such taxa in the da- hard. But the cays of Cuba do not seem to tabase. have gained unusual numbers of nonendem- (5) Low number of nonendemic lineages ics by comparison to the main island, despite in the Greater Antilles. According to the data the fact that some of the cays are also close presented by Hedges, the great majority of to a continental margin (e.g., Archipielago de nonendemic herp lineages live on islands Sabana-CamaguÈey; ®g. 1). other than the Greater Antilles. Setting aside (6) Unknown shaping in¯uence of extinc- crocodiles and tortoises, only 6 of 33 Qua- tion. Finally, if, as Hedges' interpretation of ternary nonendemic lineages listed by Hed- his data implies, it were possible for many ges (1996a) have distributions that include a continental taxa to emit successful propa- continental mainland and one or more of the gules during the past 2 Ma, then it should Greater Antilles (Gonatodes albogularis, have been possible throughout the Cenozoic. Hemidactylus haitianus, H. mabouia, Iguana That is, it should have been possible if, at iguana, Mabuya ``bistriata,'' and Nerodia any and all times during the last 65 Ma, the clarki). Of these, at least three are of ques- likelihood of passive dispersal was equipo- tionable pertinence given the ambiguity of tential. Hedges (1996a: 116) addressed this 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 43 point indirectly, noting only that, for the ly near relatives. To restate their main points, West Indies in general, the ``large number of and assuming the valency of their molecular [Quaternary] lineages almost certainly is an clock model, we are told that (1) ``systematic artifact due to fewer extinctions expected for error'' of this kind must always be unidirec- recent lineages.'' Once again, we suggest, tional (i.e., will always result in overestima- fossil evidence will turn out to be critically tion of time since divergence), and (2) any important for determining whether massive corrections resulting from the discovery and faunal turnovers or taxon cycles occurred on utilization of exact sisters would only shorten these islands. divergence times. We explore the implications of these points by means of simple phy- SOURCES OF ERROR IN ESTIMATING logenetic scenarios superimposed on equally TIMES OF LINEAGE ORIGINS simple biogeographical ones (®g. 9A, B). In ®gure 9A, imaginary taxa A±F are Hedges et al. (1992, 1994; see also Hedg- known to be phylogenetically related in the es, 1996a, 1996b) allowed that their estima- manner indicated in the small cladogram in- tions of divergence times for investigated lin- set on the lower right. The groups of mono- eages could be subject to three sources of phyletic terminal taxa are holophyletic; there error. The ®rst two are concerned with the are no unknown (undiscovered) taxa above interpretation of the immunological data the lowermost node. Taxa A, C, and F are themselves; they concern the effect of dif- extant (solid stems). Taxa B, D, and E are ferences in reciprocal estimations of IDs be- extinct (outline stems). The main cladogram tween taxa, and interclade variablity in rates is superimposed on two landmasses (``Main- of albumin evolution. Hedges and co-work- land,'' `Ìsland'') separated by an expanse of ers discounted the ®rst on the ground that water (suggested by double-headed arrow). errors (if any) would not lead to consistent The same conventions are followed in ®gure (unidirectional) underestimation of IDs 9B, except that the distribution of taxa on the across many groups. In discounting the sec- landmasses is different. ond, they acknowledged that rate variability In ®gure 9A, one major clade, of which A, exists, but at such a low level (they cite 10± B, and C are the terminal taxa, was always 15% as a reptile maximum) that it would resident on Mainland. (The sense of time is have no effect on the kinds of interpretation toward top of page, as indicated by vertical they pursued in their paper. Their certainty time line on right.) At a speci®c point in that these potential sources of error affected time, the other major clade, of which D, E, their conclusions only marginally is open to and F are the terminal taxa, came to be iso- challenge (Page and Lydeard, 1994), but we lated on Island. The diagram is purposely are particularly interested in the underpin- ambiguous as to whether the basal split oc- nings of their third source of potential error. curred though dispersal or vicariance. All Hedges et al. (l992: 1910-1911) acknowl- that is assumed is that a barrier separated edged that populations of the last common ancestor of . . . the species used here may not be representative all terminal taxa, interrupting gene ¯ow. As- of the most recent divergence event between the lin- sume now that albumins of A and F (taxa in eages examined (i.e., a member of the mainland taxon boxes outlined in heavy black) are compared closest to the island taxon examined inadvertently within the same analytical framework as the was not used); this type of error always will result in an overestimation of the time of lineage divergence one followed by Hedges et al. (1992) and for the taxa from different land masses. If this sys- yield an ID of 100, for the sake of this ex- tematic error could be corrected, some distances re- ample. In this case, the geological age of the ported here could only be lower, further suggesting basal split between the branches ending in A dispersal as the primary mechanism for vertebrate and F, as estimated by ID (open circle on colonization of the West Indies. time line), is for certain the same as the ap- We take this passage to mean that it is not pearance of the stipulated water barrier be- crucial that actual sister taxa (hereafter, `èx- tween populations of the last common an- act sisters'') be used, so long as (by impli- cestor (closed circle). Such a situation, in cation) the taxa compared are indeterminate- which ID time, cladogeny, and the event of 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 9. Splitting lineages vs. splitting lands: Limitations on interpretation (see text).

interest all intersect on the same graph of the barrier between F and its closest exact (dashed line on inset cladogram), may be de- sister, the clade terminating in taxa D and E scribed as one of congruence. (hence ``?1D''). The method is blind to later In ®gure 9B, all initial factors are the splits (including the one that installed the an- same, except that extinct taxa D and E were cestor of F on Island) because the critical Mainland residents. Only the branch repre- taxa are extinct and cannot therefore be sam- sented by terminal taxon F was isolated on pled. The result is one of incongruence. Island. If the extinct taxa were not known, it In answer to criticisms by Page and Ly- would be assumed that taxa A and C consti- deard (1994), Hedges et al. (1994: 48) stated tute the sister group of F. Measured ID that errors in taxonomic sampling do not would still be 100 (between A and F or C matter, because overestimates are always uni- and F), but it would not date the appearance directional, and furthermore ``whether the 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 45

IDs coincide or not would not affect that PASSIVE TRANSPORT AND CENOZOIC conclusion.'' However, it actually does mat- SURFACE-CURRENT PATTERNS ter, because ®lling a matrix with overestimates can obscure whatever patternÐinclud- SURFACE-CURRENT PATTERNS AND FLOTSAM ing any concentration of splitsÐthat may ex- DISPERSAL: Hedges (1996b: 186) contended ist within the phylogeny (or sets of compared that `àn overwhelming majority (99%)'' of phylogenies). In their original ®gure illus- all `ìndependent lineages'' of Antillean ver- trating times of origin (Hedges et al., 1992: tebrates originated from dispersants arriving ®g. 1), for example, the ``West Indies±Main- as passively transported ``¯otsam,''and that, land'' row is ®lled with a ¯ock of ID mea- furthermore, this ``dispersal pattern can be surements distributed across the range 55±32 explained by the nearly unidirectional current Ma. Some cluster fairly tightly around 28± ¯ow [in the Caribbean Sea] from the southeast to the northwest, bringing ¯otsam from 24 Ma, while others group at approximately the mouths of South American rivers (e.g., 18 and 11 Ma. In a later version of this ®gure Amazon, Orinoco) to the islands of the West (e.g., Hedges et al., 1994: ®g. 2), which in- Indies.'' Within the Caribbean itself, taxic cludes errors of the estimates, no pattern is similarity among nonvolant faunas of differ- evident, and the lack thereof (i.e., apparent ent islands is to be explained by subsequent randomness) is taken to be evidence for over- interisland dispersals occurring in the same water dispersal of relevant taxa/lineages. manner, i.e., passive dispersal by surface cur- However, this conclusion is also unjusti®ed, rents. because they do not know (as they acknowl- It is critical to note that Hedges' (1996a, edge) whether they are judging distances be- 1996b) argument implicitly requires that the tween exact sisters, i.e., whether congruence modern surface-current pattern, in which av- obtains. In the context of the present paper, erage ¯ow is indeed `àlmost unidirectionally it would be especially important to know if from southeast to northwest,'' obtained for there was a concentration of splits during the the whole period in which the Antillean fau- earlier part of the Oligocene, when we think na was being formed (i.e., virtually the entire paleogeographical conditions were especially Cenozoic acording to Hedges' own analysis). favorable for overland colonization (see The only causative mechanism for the `àl- GAARlandia Landspan and Island±Island most unidirectional'' ¯ow that Hedges Vicariance). And indeed there could be, if (1996a, 1996b) refers to, however, is the Cor- the pre-28 Ma splits represent overestimates iolis effect, i.e., the tendency for the trajec- of just the sort Hedges et al. (1992) deem tories of moving objects on the earth's sur- unimportant. face to be de¯ected either to the right (north- We take the point that validly identi®ed ern hemisphere) or the left (southern hemi- post-Oligocene splits cannot be explained by sphere). He gives little attention to the either continent±island vicariance or the in¯uence of varying paleogeographical con- landspan hypothesis (see below), although ®gurations of the Caribbean region on cur- some intra-Caribbean splits could be related rent ¯ow. As we show in this section, present to island±island vicariance events in the Ear- evidence supports the conclusion that the ex- ly and Middle Miocene. Some post-Oligo- isting pattern of surface currents within the cene lineage origins were probably occa- Caribbean Sea is characteristic only of the sioned by overwater dispersals, although for last 4 million years. Before that, other pat- their identi®cation using Hedges' methodol- terns predominated. Some of these, we argue, ogy, much would depend on whether all of are incompatible with the history of faunal their molecular clocks ticked true all the emplacement in the Caribbean region as en- time. Nevertheless, the timing of splits esti- visaged by Hedges (1996a, 1996b). mated from albumin divergences cannot be SURFACE-CURRENT PATTERNS AND PALEO- taken as proxies for dating actual coloniza- GEOGRAPHY: Figure 10 compares reconstruct- tion events unless one knows the true branch- ed patterns of marine surface currents for the ing sequence of a phylogeny (cf. Hedges et Late Eocene±Oligocene (35±32 Ma), Late al., 1994: ®g. 1). Oligocene/Middle Miocene (30±14 Ma), 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 10. Paleoceanography of Caribbean region for selected intervals between latest Eocene and Recent, modi®ed from Mullins et al. (1987), Duque-Caro (1990), and Droxler et al. (in press). The purpose of these reconstructions is to depict possible effects of GAARlandia and other features on surface-current patterns during the latter part of the Cenozoic (MacPhee and Iturralde-Vinent, 1995; Iturralde-Vinent et al., 1996a). Since the latest Eocene the course of westward-¯owing currents in the mid-Atlantic and Caribbean area has been greatly affected by the appearance and disappearance of various land barriers. Thus, the Circumtropical Current, which originally passed into the western Paci®c, was temporarily disrupted in the latest Eocene and Early Oligocene by the emergence of GAARlandia, and was permanently disrupted by the completion of the Isthmus of Panama in the Late Pliocene. This current has played a leading role in mediating climate in the Caribbean region during the Cenozoic. Gulf coast of North America not intended to be paleogeographically accurate. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 47

Late Miocene (6±5 Ma), and Pliocene±Ho- evidence of erosional surfaces. Also, the rate locene (4±0 Ma) (see also ®gs. 6±8). These of pelagic sedimentation in this period was reconstructions, slightly modi®ed from those extremely low. (See description of Holes 998 developed by Mullins et al. (1987), Duque- and 999, Leg 165, by Sigurdsson et al. Caro (1990), and Droxler et al. (1998), at- [1997]; Oligocene sediments are absent in tempt to consider the effects of GAARlandia some DSDP cores [Edgar et al., 1973].) and other recently described paleogeograph- Late Oligocene to Middle Miocene (30-14 ical features on surface currents (MacPhee Ma): Water circulation patterns were sub- and Iturralde-Vinent, 1995; Iturralde-Vinent stantially altered by paleogeographical et al., 1996a). changes during mid-Oligocene/Middle Mio- Latest Eocene to Early Oligocene (35±32 cene highstands. The seaways to the Paci®c Ma): During this interval, the surface-current and northern Atlantic persisted, and the Ori- pattern in the Caribbean region would have noco and Magdalena still drained into the differed radically from patterns that obtained Caribbean Sea (Hoorn et al., 1995). At the before or after this time. There are two pa- beginning of this period, communication be- leogeographical reasons for this: the exis- tween the Paci®c and equatorial Atlantic tence of the Panamanian Seaway, connecting would have been greatly improved following the Caribbean Sea with the Paci®c Ocean; the subsidence and inundation of much of the and the conformation of GAARlandia, then Aves Ridge, as indicated by the existence of at its subaerial maximum. The effect of these shared foraminiferal assemblages on the Pa- two features would have been to enhance ci®c and Caribbean sides of the Panamanian communication with the Paci®c while im- seaway (Duque-Caro, 1990). Warm water de- peding it with the equatorial Atlantic (Don- rived from the equatorial Atlantic would nelly, 1989b). Connection with the Atlantic have been more widely distributed by the would have continued at higher latitudes, via Circumtropical Current (Mullins et al., 1987; the Yucatan and Havana±Matanzas Chan- Droxler et al., 1998), now no longer pre- nels. However, throughput from the Carib- vented by southern GAARlandia from fully bean Sea to the Atlantic via the Strait of Flor- communicating with the Caribbean Sea. In- ida seems to have been rather low (Iturralde- fusion of warm Atlantic water may have in- Vinent et al., 1996a). Because of the con®g- ¯uenced regional climate, as it did elsewere uration of land masses, the Gulf Stream (Tsuchi, 1993). Within the Caribbean Sea it- would have been fed mostly by the north- self, a minor current headed northwest to westerly current ¯owing parallel to Bahamas. push warm water through the Havana±Ma- In South America, the early Magdalena and tanzas Channel and Yucatan Channel, there- Orinoco Rivers would have ¯owed directly by contributing to the early Loop surface into the the Caribbean Sea (Hoorn et al., current in the Gulf of Mexico (Iturralde-Vi- 1995). nent et al., 1996a). However, ¯ow in the ear- Under such conditions of at least partial ly Loop was not as strong as that of the mod- isolation, there was probably a very low rate ern system (Mullins et al., 1987; Hine, 1997). of water circulation across different portions These new water masses had a pronounced of the Caribbean Sea. Sub-basins such as the sedimentological effect in the southern part Gulf of Mexico may have behaved as large, of the Strait of Florida: the West Florida semi-independent embayments, in which cy- ramp began to prograde from the east, as il- clonic circulation patterns dominated, with lustrated by west-dipping clinoforms. seasonal changes in direction of ¯ow. These Later, in the Middle Miocene, rejuvenation interpretations are supported by two major of tectonism modi®ed the pattern of water lines of evidence. Seismostratigraphic studies circulation by redistributing the ¯ow carried (Mullins et al., 1987; Denny, 1992; Hine, by the Circumtropical Current (Iturralde-Vi- 1997; Denny et al., 1994) carried out in the nent et al., 1996a). In the Late Miocene, up- southern part of the Strait of Florida indicate lift of the Andes drastically altered the ¯uvial that the West Florida ramp was in an aggra- pattern of northern South America, eventu- dational regime from the Late Eocene to the ally causing the Orinoco to empty into the Late Oligocene, with little winnowing and no Atlantic (Hoorn et al., 1995). Foraminiferal 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

assemblages in sediments of late Middle to SURFACE-CURRENT PATTERNS AND PROXY Late Miocene age in the Atrato Basin on the DATA: A current's average vector is not the Paci®c side of Colombia exhibit an over- only in¯uence operating on material passive- whelmingly ``Californian'' aspect (Duque- ly carried along the sea surface, as experi- Caro, 1990), suggesting that at this time there mental data from free-¯oating objects illus- was little throughput from the Caribbean side trate. Molinari et al. (1979) and Kinder of the Panamanian seaway. Nevertheless, im- (1983) employed satellite telemetry to track pairment of circulation across the seaway 23 free-drifting buoys released from stations was evidently short-lived, because Late Mio- in the Lesser Antilles (®g. 11A). After wan- cene to Pliocene rocks in the Atrato basin dering widely, often for several months, contain assemblages having a mixed Paci®c/ bouys ended up not only along or near the Caribbean aspect (Duque-Caro, 1990). coasts of Jamaica, Cuba, Hispaniola, Puerto Late Miocene (6±5 Ma): In the Late Mio- Rico, and several other Caribbean islands, cene (®g. 10), water circulation within the but also those of Central America, Yucatan, Caribbean Sea may have been diverted to the and the Gulf of Mexico. The point here is northwest as a result of (1) reduction in the not that the free-¯oating buoys achieved a width of the Panamanian seaway as the isth- certain distribution, but instead that the av- mus advanced southwards; (2) extension and erage gross direction of surface currents pro- subsidence of the Nicaragua Rise, which had vided only the most general guide to the previously impeded ¯ow into the Yucatan probable movement of any given buoy. Fur- Channel (Mullins et al., 1987; Droxler, ther, the path followed by each buoy was 1995); and (3) closure of the Havana±Matan- substantially affected by local eddies, deep zas Channel, thereby forcing all northwest- cyclonic water circulation, storms, and other bound ¯ow into the Yucatan Channel (Itur- events (see Molinari et al., 1979; Kinder, ralde-Vinent et al., 1996a). The increased 1983; Kinder et al., 1985; Sou et al., 1996). volume of water diverted into the Loop Cur- Signi®cantly, such forces generally acted to rent modi®ed depositional conditions in the increase rather than decrease trip length, an Strait of Florida. Erosion took place on the important consideration in evaluating the Pourtales Terrace (Gomberg, 1974) and the likelihood of successful overwater dispersal Miami Terrace (Mullins and Newmann, for animals that are not physiologically 1979), and there was a 50% decrease in the adapted for temporally lengthy sea journeys rate of sediment accumulation in the south- (such as land mammals). ern Strait of Florida (Austin et al., 1988). The Brucks' (1971) drift bottle study yielded modern Loop Current/Gulf Stream circula- broadly similar results. Brucks released bot- tion was initiated and carbonate deposition tles at various locations off the Windward Is- changed drastically, becoming a pelagic lands (southern Lesser Antilles), Gulf of slope-front-®ll system (Mullins et al., 1987). Honduras, and southwestern Caribbean Pliocene to Holocene (4±0 Ma): The last (northern coast of Panama). In the Windward map in this series illustrates the modern pat- Islands release area, bottle recovery records tern of water circulation (Atlas Nacional de indicated that surface currents are cyclonic, Cuba, 1970; Emery and Uchupi, 1972). This but with a de®nite northward trend (instead pattern was initiated around the Miocene± of a due westward trend as stipulated in pre- Pliocene boundary, following the complete vious literature). Bottles came ashore in the closure of the Panamanian waterway and the northern Windward Islands and a wide vari- termination of the Circumtropical Current. ety of other locations (®g. 11B, table 5) dis- Its successor, the Caribbean Current, is fed tributed throughout the Caribbean Sea and directly by the Atlantic equatorial current; Gulf of Mexico. Of interest is the fact that surface currents are now mostly directed to- the greatest number of reports (38% of total) ward the northwest. Like other changes dis- were from Central America; only 15% were cussed in this section, this ®nal series of cur- from the Greater Antilles (as much as Yu- rent reorganizations must have had profound catan alone). Calculated minimum speeds of effects on terrestrial climates as well (Frakes, bottles varied from 0.1 to 2.0 knots, with 1979). faster rates of movement in summer and mid- 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 49

Fig. 11. Surface-current patterns in the Caribbean Sea and nearby waters, as revealed by bouy and drift bottle experiments. A. Tracks of bouys released by Molinari et al. (1979) between October 1975 and June 1976 (redrawn from original). Arrows have been added to pinpoint last recorded positions of selected bouys. Results of this experiment demonstrate that ``average'' marine-current patterns, as rou- tinely depicted on oceanographic maps, fail to capture substantial variance in surface-water movement in the Caribbean Sea. B. Surface-current pattern derived from drift bottle experiments conducted by Brucks (1971, modi®ed from original). Bottles were released from sites in the eastern Caribbean. The majority of returned bottles transported to points west of the Lesser Antilles were deposited along the east coast of Central America, Gulf of Mexico, and Florida. Relatively few bottles were returned from the Greater Antilles. These results strongly imply that, given existing surface-current patterns, ¯otsam emitted from the Orinoco and Amazon Rivers is much more likely to end up in southeastern North America or Central America than in the Greater Antilles. 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

ricane transport, and there is at least one recent case in which this is the only feasible explanation for a dispersal event. A number of individuals of Iguana iguana, a species which does not occur in Anguilla, were conclusively identi®ed on that island in the month following the passage of Hurricane Luis across the northern Lesser Antilles in August, 1995 (Censky et al., 1998). It is probable but not demonstrated that the dispersants came from Guadeloupe or one of the other islands south of Anguilla that support Iguana iguana. Although northward transport of propagules by hurricanes might explain some aspects of Lesser Antillean sau- rian biogeography, modal hurricane tracks may well have differed in previous epochs. Before the completion of the Panamanian isthmus, for example, storm tracks may have been de¯ected relatively southward by the warm Circumtropical Current ¯owing into the Paci®c. Whether this occurred or not has not been properly modelled, but any southward de¯ection of high-energy storms would winter than in spring and fall. These results have reduced the chances of successful dis- are in good agreement with work on the dy- persal to the Greater Antilles. namic sea-surface topography of the Carib- Tracking studies provide another body of bean Current, which subdivides into a faster- relevant proxy data. Mean sediment concen- moving southern current and a slower north- trations in sea water can be tracked and mea- ern one (Duncan et al., 1982). sured using satellite imagery. In one such The relatively slow speed at which pas- study (Richardson, 1996), it was found that sively transported objects move under aver- water discharged by the Amazon does not age conditions in the West Indies is further always move northwestward into the Lesser illustrated by recent work on passively dis- Antilles, but is sometimes forced to ¯ow persed pelagic larvae (e.g., Roberts, 1997). east±southeast. This ®nding is in agreement These investigations indicate that distances with the fact that the North Equatorial Coun- actually or potentially travelled by such lar- tercurrent carries much of the Amazon's out- vae during one- and two-month ``transport ¯ow eastward into the central Atlantic (Rich- envelopes'' are relatively small. For exam- ardson, 1996). Interestingly, sediment track- ple, the two-month envelope for larvae dis- ing studies have also shown that out¯ow persing in the northeastern Caribbean spans from the Orinoco is normally directed to the only the distance from Anguilla to eastern northwest, where it can be detected as far Puerto Rico. A route crossing the entire Ca- north as Puerto Rico. By contrast, out¯ows ribbean Sea from southeast to northwest, as from the Magdalena and Lake Maracaibo envisaged in Hedges' model, would presum- quickly become disorganized after entering ably take much longer. Except under unusual the Caribbean Sea: pigment can be traced for circumstances, therefore, it would appear that only a short distance before becoming dilut- modern surface currents in the Caribbean Sea ed (Richardson, 1996: ®g. 21-12). do not ¯ow rapidly enough to ensure the sur- OTHER CONSTRAINTS: Hedges' (1996a) ID vival of terrestrial amniotes dispersing by data indicate that two surviving herp lineages rafting (with the possible exception of some (Eleutherodactylus and Cricosaura typica) reptiles). could have been established on landmasses This limitation might be overcome by hur- in the Caribbean Sea as early as the latest 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 51

Cretaceous or early Paleogene (see also which may explain the acknowledged recen- Hedges et al., 1992, 1994). In his view, em- cy of so many relatively undifferentiated placement at this early date could have oc- South American herps in the southern Lesser curred through either continent±island vicar- Antilles. Yet if these ®gures were truly rep- iance or overwater dispersal. However, dur- resentative of the vertebrate fauna as a ing the latest Eocene and Early Oligocene, whole, then it would be necessary to conclude circulation within the Caribbean Sea was that either (1) the islands had a taxically mi- limited, and, as discussed above, in any case nuscule vertebrate fauna prior to the Plio- the patterns then existing would not have Pleistocene, or (2) there was a virtually com- uniformly favored the movement of water plete faunal turnover in the latter part of the masses toward the northwest (®gs. 6, 10). Neogene. At least for mammals, the ®rst point Further, given the paleoposition of the is ¯atly contradicted by fossil ®nds made in mouths of the Magdalena and Orinoco, ¯ot- the Greater Antilles in the last decade: every sam issuing from these rivers would have major taxon of land mammals known from been much more likely to end up on the is- the Antillean Quaternary now has a minimum lands that then comprised much of southern origin date of Early Miocene or earlier Central America and the Nicaraguan Rise (MacPhee and Iturralde-Vinent, 1994, 1995; than, say, the Cuban/Hispaniolan end of MacPhee and Grimaldi, 1996). The possibility GAARlandia. of high faunal turnover, however, is not di- The same constraints would have contin- rectly tested by this evidence as no `ùnex- ued into the subsequent period (30±14 Ma) pected'' taxa other than Hyrachyus in Jamaica that we have modelled. In our view, it is like- have been recovered (see MacPhee and Wyss, ly that any natural rafts coming out of large 1990). For lizards the data are also unhelpful, northwestern South American rivers would inasmuch as recent dispersals (as evaluated by have been sent into the Paci®c (®gs. 8, 10; Hedges) have added very little to the faunally see also Frakes, 1979; Kennett, 1985; Mul- rich Greater Antilles, even though there may lins et al., 1987; Droxler et al., 1998). At that have been many independent events (G. May- time part of the present Amazon Basin was er, personal commun.). occupied by a large marine embayment The proxy data for the effect of surface- (Hoorn et al., 1995). Drainage off the Guy- current ¯ow as a dispersal agent suggest that, ana Shield would have been the only signif- on the basis of existing current patterns, the icant source of ¯otsam directed into the At- Orinoco is more likely to release ¯otsam that lantic; however, from this source any pas- ®nds its way into the central Caribbean Sea sively transported material would have been than is the Amazon. Whether this pattern was as likely to drift toward Africa as the West always the rule in earlier times is moot, but Indies (cf. Richardson, 1996). Similar con- it is surely relevant that the Amazon has been ditions would have existed during the sub- an Atlantic coastal river only since the Late sequent 10 Ma. However, it is necessary to Miocene (Hoorn et al., 1995). If rivers acted observe that increasing amounts of water as potential sources of animal-bearing ¯ot- would have been directed toward the north- sam in the manner contemplated by Hedges west as Central America grew southward in for geologically long periods of time, then it the Late Miocene. From this it follows that is the basins of northwestern South America, Hedges' (1996a) emplacement scenario rather than the Amazon itself, that should be would increase in likelihood the closer one under consideration. comes to the present. The trouble is that, if A ®nal occasion of lack of ®t between emplaced by overwater dispersal, the speci- Hedges' model and the paleogeographical re- ®ed `òverwhelming majority'' of indepen- constructions offered here concerns the avail- dent insular lineages would have to have ability of lands in the Caribbean Sea at pur- originated 4 million years ago or less. Hedg- ported times of land vertebrate colonization. es' (1996a) table 3 may be said to underline The data of Hedges and co-workers suggest this very possibility, since no fewer than 55 that 6 to 11 herp lineages originated earlier investigated origins (76% of total) are dated than the Eocene±Oligocene transition. Al- to, or overlap with, the period 4±0 Ma, though the number of such `èarly'' lineages 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 is obviously small and therefore does not and the Southern Ryukyu landspan extending bear much analysis, the fact that any lineages from Taiwan at least as far as Okinawa (and may go back to the earlier Paleogene con- therefore joining these islands with continen- ¯icts with our view that the islands that now tal eastern Asia; Ota, 1998). This usage exist in the Caribbean are post-Middle Eo- stands in contrast with that favored for cene in origin (Iturralde-Vinent, 1982; ``landbridge,'' here restricted to mean land MacPhee and Iturralde-Vinent, 1994, 1995). linkages between continental regions (e.g., Hedges (1996b: 166) resolved this con¯ict DeGeer, Panamanian, Beringian landbrid- by stating that ``the recent suggestion that ges). It should also be noted that on-shelf (or there were no permanently subaerial land- ``continental'') islands are simply extensions masses in the Greater Antilles prior to [the of continents that may function as submerged end of the Middle Eocene] is speculative; it margins, islands, or peninsulas (or often as can neither be refuted nor supported with all three over time), and they are always current evidence.'' We interpret the nature of broadly affected by the same tectonic and the current evidence differently, but, notwith- isostatic regimes as their mainlands. Off- standing that, we pose the following conun- shelf islands are not by any useful de®nition drum. If evanescent islands existed in the Ca- extensions of continents, even if they incor- ribbean Sea before the Middle EoceneÐand porate continental fragments (e.g., Madagas- it is highly probable that they didÐthen any car). Off-shelf islands may be affected by resident faunas must have either perished at major tectonic events that also affect nearby the time of inundation or subsidence, or they continents, but they may otherwise evolve found a means to transfer to newly risen quite independently. landmasses elsewhere. Hedges (1996a) men- We introduce this terminological re®ne- tioned the former possibility in relation to the ment in order to differentiate between two sterilizing effects of giant tsunamis generated quite dissimilar contexts in which faunal dis- by the K/T bolide impact, but speculated that tributions can be in¯uenced by the appear- there were, nevertheless, some places where ance of a land connection. In general, trans- faunal elements could have persisted into the fers from one faunal area to the other across Cenozoic. As we have repeatedly asked in a newly developed land connection will be this paper, if such places existed, where were controlled by a host of factors (i.e., ®lters) they, and what is their paleogeographical his- that affect the likelihood that any given spe- tory? cies will complete the journey successfully. However, the outstanding feature of land- GAARLANDIA LANDSPAN AND bridges is that they connect areas having con- ISLAND±ISLAND VICARIANCE: tinental-scale faunal diversity. At least in the- MODEL OF MACPHEE AND ory, the entire diversity of each area is avail- ITURRALDE-VINENT able for interchange (although in practice the actual number of transfers in either direction is usually much less than the theoretical max- LANDSPANS,VICARIANCE, AND imum). That ¯ow actually occurs bidirection- DIVERSITY SCENARIOS ally is amply demonstrated by relevant pa- A ``landspan'' is here de®ned as a subaer- leontological records (cf. Webb, 1976), even ial connection (whether continuous or punc- if it occurs predominantly in one direction. tuated by short water gaps) between a con- Landspans are markedly different because tinent and an off-shelf island (or island arc). one terminus lacks continental-scale diversi- Although few continuous landspans exist at ty (and, indeed, may initially have no fauna present (e.g., Kamchatka), during Pleistocene at all). Although in principle any number of glaciations several signi®cant ones were cre- continental faunal elements might cross a ated during phases of lowered sea level. newly created landspan to colonize an island Among these were the Greater Palawan land- or island chain, long-term survival after ini- span connecting Bulabac-Palawan-Calami- tial colonization will be highly correlated ans with Borneo (and therefore with south- with the availability of appropriate habitat in eastern Asia; Heaney and Regalado, 1998) what are, after all, absolutely small places. In 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 53

Fig. 12. Schematic paleogeographical scenario for Caribbean region, middle Mesozoic to late Ceno- zoic, emphasizing permanency of terrestrial enviroments and ¯uctuations in land±land connections. Gen- erally speaking, light shading implies temporary lands or connections between landmasses. ``Evanescent islands'' are purely conjectural, and their number and position are not be taken literally. Late Mesozoic and Miocene landbridges between the Americas are not con®rmed. Black arrows signify direction of propagule dispersal over land connections, as established by fossil evidence. Thus, dispersal was bidirec- tional in case of Plio-Pleistocene Panamanian landbridge (Great American Biotic Interchange), which permitted exchange between two large continental faunas. By contrast, in case of GAARlandia and Chortis/ western Jamaica landspans, propagule movement (at least for vertebrates) was neccesarily unidirectional, as only the continental terminus possessed a fauna when the landspan was formed. Timing of major tectonic events also indicated. Chronometric scale purposely nonlinear. Interconnections existing at any given time can be easily visualized by placing a ruler horizontally across page. contrast, continent-sized areas are much ment, sustained at length in this paper, that more likely to support a great diversity of the Cenozoic paleogeography of the Carib- habitat types, increasing the chances of suc- bean region strongly favored emplacement cess of a substantial variety of immigrants. over land (as opposed to over water) only The two-part landspan/vicariance model of once in the past 65 Ma. (Details of Cenozic MacPhee and Iturralde-Vinent (1994, 1995, connections among lands is schematically this paper) attempts to infer mechanisms of presented in ®gure 12.) faunal formation in the Greater Antilles from The ®rst component of the model seeks to detailed paleogeographical reconstructions, explain how land mammals might have fossil evidence, and species/area relation- reached the northern Greater Antilles from ships. Central to the hypothesis is the argu- northwestern South America by dispersing 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 across a short-lived landspan during a re- persal. The ``orderly,'' multi-island distribu- stricted period in the mid-Cenozoic (ca. 33± tion of lower level monophyletic units of 35 Ma according to our current estimate). land mammals (particularly sloths and insec- Speci®cally, it is hypothesized that, during tivores) possibly supports this inference of the Eocene-Oligocene transition, the Aves island±island vicariance and ought to be test- Ridge became subaerial for a short interval, able cladistically (MacPhee and Iturralde, in possibly as short as one or two million years. prep.; White and MacPhee, in prep.). At that time the islands on the northern part This model has major implications for An- of the Greater Antillean Ridge (central and tillean vertebrate paleontology. As noted ear- eastern Cuba, north-central Hispaniola, lier, Hedges' (1996a, 1996b) ID data appear Puerto Rico, Virgin Islands) were in a close- to show that a large proportion of the West packed array; they either constituted a single, Indian herpetofauna originated extremely re- large island, or a series of islands separated cently (although much of its diversity is due by very narrow water gaps. By connecting to Tertiary colonizations, as his data also proximally with the eastern end of the Great- show). By contrast, Rosen's (1975) and er Antilles Ridge and distally with north- MacPhee and Iturralde-Vinent's (1994, 1995) western South American microcontinent, the models require that the Antillean fauna was subaerial Aves Ridge completed the GAAR- formed much earlier (Late Cretaceous or landia landspan. As in the case of the north- Late Eocene/Early Oligocene, respectively). ern connection, the linkage between GAAR- If the overwater dispersal model is a gener- landia and northwestern South America ally accurate narrative of faunal formation in would have existed for only a short time (i.e., the West Indies, it follows that the fauna between major occurrences of postmagmatic must have formed in an episodic, accretion- arc uplift in latest Eocene and general sub- ary manner. Accordingly, one would not ex- sidence in Oligocene). pect to ®nd a faunal assemblage that was The second component seeks to explain markedly different or systematically more di- how certain distributions of faunal elements verse than the modern one at any point in the might have been produced via island±island past (low diversity scenario). On the other vicariance, due to the subdivision of the is- hand, if either Rosen-style vicariance or the lands themselves. Mid-Oligocene and Mio- landspan hypothesis were correct, it would cene marine transgressions and neotectonics be expected that diversity would be much (Mann and Burke, 1984; Mann et al., 1990) greater at the crucial times during which the substantially affected the disposition and pa- fauna was being emplaced (high diversity leogeography of the Greater Antilles Ridge. scenario). Figure 13 explores how features of Several pull-apart basins and related features a good paleontological record might be able opened or expanded along the northern Ca- to help distinguish between high-diversity ribbean plate boundary (Cayman Trough, and low-diversity scenarios. Mona Canyon, Sombrero Basin/Anegada In the ®gure, taxa a±p (left column) are Passage), creating deep-water channels and the initiators or founding species of a series basins between tectonic units. Paleogeo- of different imaginary clades. T1, T2, and graphically, the end result was the gradual T3 are three different time transects (for the subdivision of the subaerial parts of the sake of this example, T1 represents Late Eo- Greater Antilles Ridge. For example, eastern cene; T2, Early Oligocene, T3, Pleistocene/ Cuba and northern Hispaniola, physically Holocene boundary). The cartouches, repre- connected during the Early Oligocene, were senting the separate clades, are each por- sundered by the expansion of the Windward trayed as having a ``lifespan'' in relation to Passage later in that epoch (Iturralde-Vinent the time axis. and MacPhee, 1996); by contrast, the con- This example permits models of supposed nection between central Hispaniola and initial faunal diversity to be compared (see Puerto Rico probably lasted until late in the box in upper right for shading conventions). Miocene (®gs. 6±8, 12). These subdivisions The light gray shading represents possible would have divided the ranges of terrestrial expectations under the high initial diversity faunal elements previously emplaced by dis- model (taxa ϭ 16). Dark gray represents ex- 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 55

Fig. 13. Speculating on the faunal history of the Greater Antilles: Signi®cance of fossils, ``ghosts,'' and incomplete evidence (see text). pectations under the low initial diversity of some common cause (e.g., vicariance, model (taxa ϭ 4). Gradient-shaded dots rep- landspan). (Prior to T1, it is assumed that resent fossil discoveries; single dots repre- there was no fauna at all.) Later on, fauna is sent isolated ®nds, numerous dots many reduced by extinctions. By time T2, taxa a± ®nds. To make comparisons realistic, taxa c are already extinct; at various times there- surviving beyond T3 are shown as having after, taxa d±l also disappear. Taxa m±p sur- very good late Quaternary fossil records (m, vive into the late Quaternary as the much- n, o, and p; n and o now extinct). diminished remnants of a once-diverse fauna. Scenario 1 (high initial diversity, few or Scenario 2 (low initial diversity, fauna no dispersals to augment original composi- gradually augmented by new dispersals stag- tion): At about T1 a fauna of high diversity gered over a long period): In this example, (taxa a±p, light grey cartouches) is emplaced taxa a±l were never part of the fauna; the essentially simultaneously through operation entire complement at any time derives from 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 taxa represented by dark gray cartouches within a narrow time slice (e.g., ®rst-appear- (m±p), whose initiators arrive at different ance fossils of taxa m±p). Multiple ®rst ap- times (asterisks). Faunal diversity (by clade) pearances at a speci®c time could be evi- was never more extensive than it was in the dence of the operation of a common cause. Quaternary. (For speci®c examples of the use of these Scenarios 1 and 2 cannot be distinguished criteria in analyzing the Antillean mamma- merely by inspection of the neontological lian colonization record, see MacPhee and fauna. However, they can be critically eval- Iturralde-Vinent, 1995). uated with fossil evidence. Recovery of substantial numbers of ``unexpected'' fossils DISCUSSION (i.e., evidence of clades not represented in the Quaternary) would tend to favor the high Most of the topics that require discussion initial diversity model. Failure to ®nd such in relation to the geology and paleogeogra- fossils would imply that the known, latest phy of the landspan hypothesis have been Cenozoic level of diversity is all there ever dealt with at length in earlier sections of this was, and that the low initial diversity sce- work. Our purpose here is to summarize nario is correct. what we consider to be the oustanding prob- As noted above, recent paleomammalogi- lems with this hypothesis, as a guide to fur- cal discoveries in the northern Greater An- ther work. tilles (Cuba, Hispaniola, and Puerto Rico) (1) As noted in earlier sections, we cannot have signi®cantly extended the records of yet offer detailed geological and paleogeo- several clades (MacPhee and Iturralde-Vi- graphical reconstructions of the Caribbean nent, 1994, 1995; MacPhee and Grimaldi, area during the Mesozoic/early Paleogene, 1996). Nevertheless, all Tertiary taxa recov- and therefore we are unable to settle whether ered to date from these islands appear to be landspans/landbridges existed in this region closely related to clades known from the before the mid-Cenozoic (but see ®g. 12 and Quaternary, which favors the low initial di- point 3 below). versity model. At present any ®rm conclu- (2) Although geographically a member of sion would be premature, as there are as yet the Greater Antilles, Jamaica has had a tectonic no fossil vertebrate sites that date to the cru- history quite different from that of the other cial period in the Late Eocene when perma- islands in the group. The only tectonic unit nent land environments were ®rst established currently incorporated into Jamaica that might in the northern Greater Antilles. Since the have had some relationship to evolving paleogeographical history of Jamaica has GAARlandia is the Blue Mountains Block. If, been markedly different from that of the oth- as some evidence indicates, the Blue Moun- er Greater Antilles, it would be equally pre- tains Block lay relatively close to the northern mature to read anything into the remarkable Greater Antilles during the Cenozoic, it may discovery of Hyrachyus on that island have received immigrants directly from (Domning et al., 1997). There may be more GAARlandia, either over water or over a land such taxa, in Jamaica and elsewhere; the only connection with southern Hispaniola (®g. 12). way to ®nd out is through intensive pros- Collision of Western Jamaica with the Blue pecting. Mountains BlockÐif such an event actually Speci®c colonization timetables are al- took placeÐwould not have occurred earlier ways subject to refutation paleontologically, than the Miocene. because fossils provide minimum dates of (3) In principle, some propagules could occupation. Thus discovery of an ``early'' have reached Caribbean landmasses from the fossil attributable to taxon m outside the tem- mainlands either before or after the landspan poral limits of its cartouche (dark gray shad- period, if colonizations occurred by over± ing) is evidence that the colonization time water dispersal or continent±island vicari- originally inferred for m is incorrect. Fossils ance. However, in our view, earlier coloni- dating ®rst appearances are also of great sig- zations (if they occurred) were ultimately ni®cance if they represent numerous clades doomed to failure (no permanent landmasses (whether expected or unexpected) and occur before Late Eocene), and later colonizations 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 57

would have become increasingly dif®cult by this or any other model, will have to be (owing to the disappearance of landspans and elucidated by additional geological, paleogeo- tectonic dismemberment of GAARlandia). graphical, and paleontological research. Nevertheless, two herp lineages identi®ed by (5) Detailed paleogeographical reconstruc- Hedges (1996a, 1996b) may be examples of tions of the now-submerged Aves Ridge are early emplacement, although there is no fos- not currently possible. However, as noted in sil evidence relating to these taxa. Jamaican appendix 1, there is evidence that portions of Hyrachyus does not constitute a counterex- this ridge must have been subaerially elevat- ample: perched on its Viking funeral ship, it ed at one time (e.g., regional pattern of post- remains a Central American/North American magmatic phase uplift in the late Paleogene, taxon, despite its allochthonous presence in thin Cenozoic sedimentary cover, presence of the Greater Antilles (Domning et al., 1997). Oligocene and Early Miocene land-derived Very late originations, as apparently oc- conglomerates on ridge). The critical issue is curred in the Lesser Antilles among many whether the uplifted, emergent Aves Ridge herp groups (Hedges, 1996a, 1996b) and per- could have formed a corridor of some sort haps some mammals (e.g., sigmodontine ra- between the Greater Antilles and northwest- diation of northern Lesser Antilles and Ja- ern South America. Wells drilled along the maica) cannot be explained by the Greater ridge's structural highs might provide the late Antilles landspan model. They are the result Paleogene record still needed to meaningful- of the operation of some other mechanism, ly constrain the corridor hypothesis, although evidently natural dispersal (or, in some cases, it cannot be predicted whether such data will possibly human transport). be detailed enough to determine if GAAR- (4) Although island±island vicariance may landia constituted a single subaerial entity at provide a neat solution for the distributions of any given time. Transitory water gaps, for several tightly related Quaternary taxa, we example, may have intervened for short pe- cannot account for all cases. For example, the riods of time between land masses situated presence of apparently endemic species of on the Aves Ridge or between segments of capromyid rodents and Nesophontes in the the Greater Antilles. Importantly, however, Quaternary of the Cayman Islands (Morgan, none of them would have had the duration 1994) cannot currently be explained as a con- or depth of the Havana±Matanzas Channel sequence of island±island vicariance, inas- between western and central Cuba. Alterna- much as the Cayman Islands are probably tively, faunal elements might have dispersed very recent geographical entities that are un- in a pulsed manner within the con®nes of a likely to have had any land connection with single event, if uplift and subsidence affected either Cuba or Jamaica during the Eocene± different parts of GAARlandia at different Oligocene transition (®gs. 6±8). A very late times. Episodic movements of this kind, if land connection between the Cayman Islands they occurred, might have increased the and Cuba might have occurred during the up- chance of extinction of faunal elements un- lift of the Sierra Maestra during the Plio-Pleis- able to transfer to the next landmass on the tocene, as these mountains are located in the chain, further diminishing overall diversity. same trend and geological unit as the Cay- (6) The degree to which the NWSA mi- mans (Per®t and Heezen, 1978; Case et al., crocontinent was physically separated from 1984; Sigurdsson et al., 1997). However, this the rest of the continent by marine barriers possibility would require a substantial amount has not been fully clari®ed. From the Eo- of subsidence during the last 5 Ma, as sea cene±Oligocene transition, the Middle Mio- ¯oor depths between Cayman Brac and Cabo cene, and especially during high sea-level Cruz (western end of Sierra Maestra) are in stands in the Late Miocene, Pliocene, and excess of 1000 m (G. Morgan, personal com- Quaternary a large marine embayment in the mun.). It may also be noted that colonizations location of the present-day Orinoco River ba- by lineages that were evidently not proxi- sin isolated the microcontinent from the mately South American (e.g., solenodontids, Guyana highlands to the east (®gs. 2±4; Nu- with proximate ancestry in North America or tall, 1990; Webb, 1995; RaÈsaÈnen et al., 1995; possibly the Old World), not well constrained Cooper et al., 1995; Kay and Madden, 1997). 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Prior to the rise of the Andes in southern American groups that ``should'' have made Colombia and Ecuador, the microcontinent the journey (Simpson, 1956) may have may or may not have been bounded by an- something to do with the relative paleogeo- other water gap or lowland to the southwest graphical isolation and faunal diversity of the at the location of the Guayaquil Portal source area as well as the nature and duration (Domning, 1982; Hoorn et al., 1995; Webb, of the landspan. 1995). Thus, to a greater or lesser degree, the (7) The Late Miocene occurrence of two northwestern and eastern parts of the conti- sloth taxa in North America and the equally nent were at least partially isolated from one early presence of procyonids (Cyonasua) and another for long periods of time, until com- now a ?cuvieroniine gomphothere in South paratively recently. These points raise an in- America (Webb, 1985; Frailey et al., 1996) teresting issue. If the northwestern microcon- may indicate that these continents experi- tinent was substantially isolated from the rest enced limited biotic interchange prior to the of South America during the late Paleogene formation of the (last) Panamanian bridge in to early Neogene by marine and orogenic the late Neogene. The mechanism that per- barriers, then only that fraction of the South mitted this limited interchangeÐif that is American biota occupying the northwestern what it wasÐis obscure; a possible dryland corner of the continent would have been in connection is signalled by the 12.9±11.8 Ma a position to cross over the evanescent land- hiatus in the Atrato Basin during the late span into GAARlandia. Unfortunately, we Middle Miocene (®g. 12; Duque-Caro, know very little about the faunal composi- 1990), but its existence is uncon®rmed. We tion of northwestern South America at the doubt that the sloths swam the distance, al- end of the Paleogene (Marshall, 1985; Kay though we note that at least one extinct phyl- and Madden, 1997). Nevertheless, the fact lophage (Thalassocnus) is thought to have that the Antillean fauna has apparently al- been highly aquatic (cf. de Muizon and ways lacked representatives of many South McDonald, 1995).

CONCLUSIONS (1) Number of intercontinental landbrid- likely that there were islands in the Carib- ges. The last time that western Laurasia bean Sea from the time of its opening in the (North America) and western Gondwana Jurassic onward. On the other hand, it is very (South America) were physically connected unlikely that any of the early islands contin- as continental areas was during the Middle uously remained as such (i.e., as subaerial Jurassic, ca. 170 Ma. Terrestrial connections geographical entities) into later times, due to between these continental areas since then repeated transgressions, subsidence, and, not can only have occurred via landbriges. In the incidentally, the K/T bolide impact and as- Cretaceous, three major uplift events, record- sociated mega-tsunamis (cf. Hedges et al., ed as regional unconformities, may have pro- 1992). duced intercontinental landbridges involving Since the close of the Mesozoic, any land- the Cretaceous Greater Antillean island arc. bridge between North and South America The late Campanian/early Maastrichtian up- would have to have involved Central Amer- lift event is the one most likely to have re- ica. The existing bridge (Panamanian isth- sulted in a landbridge, as it would have been mus) was completed only in the Plio-Pleis- coeval with uplift of the dying Cretaceous tocene. Evidence for a precursor bridge late arc. However, the evidence is too limited for in the Middle Miocene is ambiguous at this any certainty on this point. time. Whether the Cretaceous island arc was in- (2) Role of GAARlandia landspan. There volved in the formation of a late Mesozoic is evidence that northwestern South America landbridge between North and South Amer- was brie¯y connected during the Eocene±Ol- ica carries no necessary biogeographical im- igocene transition with large landmasses plications. As we have stated, it is not un- emergent on the Greater Antilles Ridge and 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 59

Aves Ridge. The massive uplift event that rent dispersal of passively transported prop- apparently permitted these connections was agules seems to us to be an ineffective ex- spent by 32 Ma; a general subsidence fol- planation of observed patterns of faunal dis- lowed, ending the landspan phase. Thereaf- tribution in the Greater Antilles. Even though ter, Caribbean neotectonism resulted in the a general tendency exists for Caribbean sur- subdivision of remaining land areas, possibly face currents to ¯ow northward with respect causing multiple instances of true vicariance to the South American coastline, experimen- among vertebrate species then present. tal evidence indicates that it is highly unpre- (3) Modes of faunal formation in the dictable where passively ¯oating objects Greater Antilles. Currently, there are three caught in these currents will be deposited. main models of faunal formation in the West Prior to the Pliocene, regional paleogeogra- Indies of interest to vertebrate biogeography was such that current-¯ow patterns from phers: strict dispersal, strict continent±island major rivers should have delivered most vicariance, and one that combines dispersal South American waifs to the Central Amer- ican coast, not to the Greater or Lesser An- and vicariance in a two-phase process. This tilles. Since at least three (capromyid rodents, paper reviews recent contributions to theory pitheciine primates, and megalonychid relevant to each of these major modalities. sloths) and possibly four (solenodontid in- Continent±island vicariance in the classic sectivores) lineages of Antillean mammals sense of Rosen (1975, 1985) appears to be were already on one or more of the Greater excludable for any period since the mid-Ju- Antilles by the Early Miocene (MacPhee and rassic; even if vicariance occurred at that Iturralde-Vinent, 1995), Hedges' inference as time, its relevance for understanding the or- to the primacy of overwater dispersal appears igin of modern Antillean faunas is minimal. to be at odds with the facts. Hedges and co-workers (Hedges et al., 1992, The landspan model is consistent with 1994; Hedges, 1996a, 1996b) have strongly most aspects of Antillean land-mammal bio- espoused overwater dispersal as the major geography as now known (MacPhee and and perhaps only method of vertebrate faunal Iturralde-Vinent, 1995); whether it is consis- formation in the Caribbean region. Notwith- tent with the biogeography of other groups standing their well-argued case, surface-cur- remains to be seen.

REFERENCES Albear, J., and M. A. Iturralde-Vinent Areces-Mallea, A. 1985. EstratigrafõÂa de las provincias de La 1990. Piazopteris branneri (White) Lorch, Habana. In M. A. Iturralde-Vinent helecho del JuraÂsico Inferior-Medio de (ed.), ContribucioÂn a la geologõÂa de las Cuba. Rev. Soc. Mex. Paleontol. 3(1): provincias de La Habana y Ciudad de 25-40. La Habana, pp. 12±54. La Habana: Ed- Atlas Nacional de Cuba itorial CientõÂ®co-TeÂcnica. 1970. Atlas Nacional de Cuba. La Habana Algar, S., and J. Erikson and Moscow: Academies of Sciences 1995. Correlation of the Jurassic through Ol- of Cuba and USSR. igocene stratigraphic unit of Trinidad Austin, J. A., W. Schlager, A. A. Palmer, and ODP and northeastern Venezuela. Int. Geol. Leg 101 Scienti®c Party Rev. 37: 313±334. 1988. Proceedings of the Ocean Drilling Pro- Alvarado, G. E. 1988. CentroameÂrica y las Antillas: Puente, gram, Initial Reports (Pt. A), Vol. 101. barrera y ®ltro bioloÂgico entre Norte y College Station, TX: Ocean Drilling SudameÂrica (CretaÂcico al Presente): Program. GEOISTMO, vol. 2(1): 9±25. Balkwill, H. R., G. Rodrigue, F. I. Paredes, and J. Anderson, T., and V. Schmidt P. Almeida 1983. The evolution of Middle America and 1995. Northern part of Oriente basin, Ecua- the Gulf of Mexico-Caribbean region dor: re¯ection seismic expression of during Mesozoic time. Geol. Soc. Am. structures. In A. J. Tankard, R. Suarez Bull. 94: 941±966. Soruco, and H. J. Welsink (eds.), Petro- 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

leum basins of South America. Am. Conference, Caracas: 439±440. Cara- Assoc. Pet. Geol. Mem. 62: 559±571 cas: Cromotik. Bartok, P. E. Bonini, W. E., R. B. Hardgrave, and R. Shagam 1993. Prebreakup geology of the Gulf of (eds.) Mexico-Caribbean: its relations to Tri- 1984. The Caribbean-South American plate assic and Jurassic rift systems of the boundary and regional tectonics. Geol. region. Tectonics 12: 441±459. Soc. Am. Mem. 162. Bartok, P. E., O. Renz, and G. E. G. Westermann Borhidi, A. 1985. The Siquisique ophiolites, northern 1985. Phytogeographic survey of Cuba. 1. Lara State, Venezuela: a discussion of The phytogeographic characteristics their Middle Jurassic ammonites and and evolution of the ¯ora of Cuba. Acta tectonic implications. Geol. Soc. Am. Bot. Hungarica 31: 3±34. Bull. 96: 1050±1055. Bouysse, P., P. Andreief, M. Richard, J. C. Baub- Beccaluva, L., M. Coltorti, G. Giunta, M. Iturral- ron, A. Mascle, R. C. Maury, and D. de-Vinent, E. Navarro, F. Siena, and F. Westercamp Urbani 1985. Aves swell and northern Lesser Antil- 1996. Cross sections through the ophiolitic les ridge: rock-dredging results from units of the southern and northern mar- Arcade 3 cruise. In A. Mascle (ed.), gins of the Caribbean plate in Venezue- Caribbean geodynamics: 65±76. Paris: la (Northern Cordilleras) and Central Edit. Technip. Cuba. O®oliti 21(2): 85±103 Bowin, C. Beets, D. J., W. V. Maresh, G. T. Klaver, A. Mot- 1975. The geology of Hispaniola. In A. E. M. tana, R. Bocchio, F. F. Beunk, and H. P. Nairn and F. G. Stehli (eds.), The ocean Monen basins and margins, vol. 3: Gulf of 1984. Magmatic rock series and high-pres- Mexico and Caribbean Sea: 501±552. sure metamorphism as constraints on New York: Plenum Press. the tectonic history of the southern Ca- Bralower, T., and M. A. Iturralde-Vinent. ribbean. In W. E. Bonini, R. B. Hard- 1997. Micropaleontological dating of the col- grave, and R. Shagam (eds.), The Ca- lision between the North American ribbean±South American plate bound- plate and the Greater Antilles Arc in ary and regional tectonics. Geol. Soc. western Cuba. Palaios 12: 133±150. Am. Mem. 162: 95±129. BresznyaÂnszky, K., and M. A. Iturralde-Vinent Bellizia, A., and G. Dengo 1978. PaleogeografõÂa del PaleoÂgeno de Cuba 1990. The Caribbean mountain system, north- oriental. Geol. Mijnb. 57: 123±133. ern South America; a summary. In G. 1985. PaleogeografõÂa del PaleoÂgeno de las Dengo and J. E. Case (eds.), The ge- provincias de La Habana. In M. A. Itur- ology of North America, vol. H, The Caribbean region: 167±177. Boulder, ralde-Vinent (ed.), ContribucioÂnala CO: Geological Society of America. geologõÂa de las provincias de La Ha- Berggren, W. A., D. V. Kent, C. C. Swisher, and bana y Ciudad de La Habana: 100±115. M.-P. Aubry La Habana: Editorial CientõÂ®co-TeÂcni- 1995. A revised Cenozoic geochronology and ca. chronostratigraphy. Soc. Econ. Paleon- Bronnimann, P., and D. Rigassi tol. Mineral. Spec. Publ. 54: 129±212. 1963. Contribution to the geology and pale- Bird, D. E. ontology of the area of the city of La 1991. An integrated geophysical interpreta- Habana, Cuba, and its surroundings. tion of the Grenada basin. Unpub. M.S. Eclog. Geol. Helvetiae 56: 193±480. thesis, Dep. Geosciences, Univ. Hous- Brouwer, S. B., and P. A. Brouwer ton, TX. 1982. GeologõÂa de la regioÂn ambarõÂfera ori- Bird, D. E., S. A. Hall, and J. F. Casey ental de la RepuÂblica Dominicana. In 1993. Interpretation of magnetic anomalies W. Snow et al. (eds.), Trans. 9th Carib- over the Grenada basin. Tectonics 12: bean Geol. Conf., Santo Domingo, Do- 1267±1279. minican Republic 1: 305±322. Bock, W. D. Brucks, J. T. 1972. The use of foraminifera as indicators of 1971. Currents of the Caribbean and adjacent subsidence in the Caribbean. In Trans- regions as deduced from drift-bottle actions of the 6th Caribbean Geological studies. Bull. Mar. Sci. 21: 455±465. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 61

Buf¯er, R. T., W. Alvarez, G. Suarez, A. Camargo, (eds.), The ocean basins and margins, A. Ocampo, and K. Pope vol. 3: Gulf of Mexico and Caribbean 1995. Chicxulub crater, Tertiary carbonate Sea: 107±180. New York: Plenum margins, and the cenote ring. Ann. Press. Meet. Geol. Soc. Am., abstracts with Case, J. E., T. L. Holcombe, and R. G. Martin program, p. A-348. 1984. Map of the geologic provinces in the Buf¯er, R. T. and J. M. Hurst Caribbean region. In W. E. Bonini, R. 1995. Disintegration of the Jurassic-Lower B. Hardgrave, and R. Shagam (eds.), Cretaceous megabank, Cuba-Florida- The Caribbean±South American plate Bahamas. 1st SEPM Congr. Sedimen- boundary and regional tectonics. Geol. tary Geology, program and abstracts, p. Soc. Am. Mem. 162: 1±30. 35. St. Pete Beach, FL. Case, J. E., R. Shagam, and R. F. Giegenganck Bunce, E. T., J. D. Philips, R. L. Chase, and C. 1990. Geology of the northern Andes; an O. Bowin overview. In G. Dengo and J. E. Case 1970. The Lesser Antilles arc and the eastern (eds.), The geology of North America, margin of the Caribbean Sea. In A. E. vol. H, The Caribbean region: 177± Maxwell, (ed.), The sea 4: 359±385. 200. Boulder, CO: Geological Society New York: Wiley-Interscience. of America. Burke, K., C. Cooper, J. Dewey, P. Mann, and J. Censky, E. J., K. Hodge, and J. Dudley Pindell 1998. Over-water dispersal of lizards due to 1984. Caribbean tectonics and relative plate hurricanes. Nature 395: 556. motions. In W. E. Bonini, R. B. Hard- Cooper, M. A., F. T. Allison, R. Alvarez, M. Coral, grave, and R. Shagam (eds.), The Ca- R. H. Graham, A. B. Hayward, S., J. ribbean±South American plate bound- Howe, J. MartõÂnez, R. Naar, R. PenÄas, ary and regional tectonics. Geol. Soc. A. J. Pulham, and A. Taborda Am. Mem. 162: 31±64. 1995. Basin development and tectonic history Burney, D. A., L. P. Burney, and R. D. E. of the Llanos basin, Eastern Cordillera, MacPhee and Middle Magdalena Valley, Colom- 1994. Holocene charcoal stratigraphy from bia. Am. Assoc. Pet. Geol. Bull. 79: Laguna Tortuguero, Puerto Rico, and 1421±1443. the timing of human arrival on the is- Craw, R. C., and P. Weston land. J. Archaeol. Sci. 21: 273±281. 1984. Panbiogeography: a progressive re- Butterlin, J. search program. Syst. Zool. 33: 1±13. 1960. GeÂologie geÂneÂrale et reÂgionale de la Croizat, L. ReÂpublique d'Haiti. UniversiteÂ de Par- 1964. Space, time, form; the biological syn- is, Trav. MeÂm. Inst. Hautes Etud. Am. thesis. Caracas (published by the au- Latine, vol. 6: 194 pp. thor). Butterlin, J. and F. Bonet Darlington, P. J. 1966. Les formations ceÂnozoõÈques de la partie 1938. The origin of the fauna of the Greater mexicaine de la presqu'õÃle de Yucatan. Antilles, with discussion of dispersal of Trans. 3rd Caribbean Geol. Conf., animals over water and through the air. Kingston, Jamaica: 75±90. Q. Rev. Biol. 13: 274±300. Calais, E., B. M. de Lepinay, V. Renard, and M. De La Torre, C. Tardy 1910. Osamentas foÂsiles encontradas en las 1989. GeÂometrie et regime tectonique le long casimbas de la Sierra de Jatibonico. d'une limite de plaques en coulissage: ComprobacioÂn de la naturaleza conti- la frontieÁre nord-CaraõÈbes de Cuba aÁ nental de Cuba a principios de la era Hispaniola, Grandes Antilles. C. R. Cuaternaria. Rev. Fac. Cienc. Letras Acad. Sci., Paris 308: 131±135. Univ. La Habana 10: 78±88. Calais, E., B. M. de Lepinay, P. Saint-Marc, J. De Zoeten, R. Butterlin, and A. Schaaf 1988. Structure and stratigraphy of the Cen- 1992. La limite de plaques deÂcrochante nord tral Cordillera Septentrional, Domini- caraõÈbe en Hispaniola: eÂvolution paleÂo- can Republic. Unpubl. Master's thesis, geÂographique et structurale ceÂnozoõÈque. Univ. Texas at Austin. Soc. geÂol. France Bull. 163: 309±324. De Zoeten, R., and P. Mann Case, J. E. 1991. Structural geology and Cenozoic tec- 1975. Geophysical studies in the Caribbean tonic history of the Central Cordillera Sea. In A. E. M. Nairn and F. G. Stehli Septentrional, Dominican Republic. In 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

P. Mann, G. Draper, and J. F. Lewis 118. Gainesville, FL: Sandhill Crane (eds.), Geologic and tectonic develop- Press. ment of the North America±Caribbean Donnelly, T. W., G. S. Horne, R. C. Finch, and E. plate boundary in Hispaniola. Geol. LoÂpez-Ramos Soc. Am. Spec. Pap. 262: 265±281. 1990. Northern Central America; the Maya Dengo, G., and J. E. Case (eds.) and Chortis Blocks. In G. Dengo and J. 1990. The geology of North America, vol. H, E. Case (eds.), The geology of North The Caribbean region. Boulder, CO: America, vol. H, The Caribbean region: Geological Society of America. 37±76. Boulder, CO: Geological Soci- Denny, W. A. ety of America. 1992. Seismic stratigraphy and geologic his- Donovan, S. K. tory of Mid-Cretaceous through Ceno- 1994. Trinidad. In S. K. Donovan and T. A. zoic rocks, southern Straits of Florida. Jackson (eds.), Caribbean geology, an Unpubl. Master's thesis, Univ. Texas at introduction: 209±228. Kingston, Ja- Austin. maica: The University of the West In- Denny, W. A., R. T. Buf¯er, and J. A. Austin dies Publishers' Association. 1994. Seismic stratigraphy and geologic his- Donovan, S. K. and T. A. Jackson (eds.) tory of Mid-Cretaceous through Ceno- 1994. Caribbean geology, an introduction. zoic rocks, southern Straits of Florida. Kingston, Jamaica: The University of Am. Assoc. Pet. Geol. Bull.78: 461± the West Indies Publishers Association. 487. Draper, G. Dolan, J., P. Mann, R. de Zoeten, C. Heubeck, J. 1989. Comparison of the Cretaceous-Paleo- Shiroma, and S. Monechi cene age rocks of northern Hispaniola 1991. Sedimentology, stratigraphy, and tec- and southeastern Cuba: implication for tonic synthesis of Eocene-Miocene sed- the tectonic evolution of the Greater imentary basins, Hispaniola and Puerto Rico. In P. Mann, G. Draper, and J. F. Antilles. 1st Congr. Cubano Geol., re- Lewis (eds.), Geologic and tectonic de- suÂmenes y programa, p. 211. velopment of the North America±Ca- Draper, G., G. Gutierrez, and J. F. Lewis ribbean plate boundary in Hispaniola. 1997. Thrust emplacement of the Hispaniola Geol. Soc. Am. Spec. Pap. 262: 217± peridotite belt: orogenic expression of 241. the mid-Cretaceous Caribbean arc po- Domning, D. P. larity reversal? Geology 24: 1143± 1982. Evolution of manatees: a speculative 1146. history. J. Paleontol. 56: 599±619. Droxler, A. W. Domning, D. P., R. J. Emry, R. W. Portell, S. K. 1995. Control of the last 35 M.Y. Caribbean Donovan, and K. S. Schindler tectonic evolution on interoceanic wa- 1997. Oldest West Indian land mammal: rhin- ter exchange and heat/salt latitudinal ocerotoid ungulate from the Eocene of transfer. Ann. Meet. Geol. Soc. Am., Jamaica. J. Vertebr. Paleontol. 17: 638± abstracts with program, p. A-153. 641. Droxler, A. W., A. A. Staples, E. Rosencrantz, R. Donnelly, T. W. T. Buf¯er, and P. A. Baker 1985. Mesozoic and Cenozoic plate evolution 1989. Neogene tectonic disintegration of a of the Caribbean region. In F. G. Stehli carbonate Eocene-Early Miocene me- and S. D. Webb (eds.), The great Amer- gabank along the northern Nicaragua ican biotic interchange: 89±121. New Rise, Caribbean Sea. 12th Caribbean York: Plenum Press. Geol. Conf., St. Croix, U. S. Virgin Is- 1989a. Geologic history of the Caribbean and lands, abstracts with program, p. 41. Central America. In A. W. Bally and A. Droxler, A. W., K. Burke, A.D. Cunningham, A. R. Palmer (eds.), The geology of North C. Hine, E. Rosencrantz, D. S. Duncan, America, vol. A, An overview: 299± P. Hallock, and E. Robinson 321. Boulder, CO: Geological Society 1998. Caribbean constraints on circulation of America. between Atlantic and Paci®c oceans 1989b. History of marine barriers and terres- over the past 40 million years. In T. trial connections: Caribbean paleogeo- Crowley and K. Burke (eds.), Tectonic graphic inference from pelagic sedi- boundary conditions for climate recon- ment analysis. In C. A. Woods (ed.), struction: 160±191. Oxford: Oxford Biogeography of the West Indies: 103± Univ. Press. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 63

Duncan, R. A., and R. B. Hargraves FernaÂndez de Castro, M. 1984. Plate tectonic evolution of the Carib- 1884. Prueba paleontoloÂgica de que la isla de bean region in the mantle reference Cuba ha estado unida al continente frame. In W. E. Bonini, R. B. Hard- americano y breve idea de su constitu- grave, and R. Shagam (eds.), The Ca- cioÂn geoloÂgica. An. Acad. Cienc. Ha- ribbean±South American plate bound- bana 21: 146±165. ary and regional tectonics. Geol. Soc. Fox, P. J., W. F. Ruddiman, W. B. F. Ryan, and B. Am. Mem. 162: 81±93. C. Heezen Duncan, C. P., S. G. Schladow, and W. G. Wil- 1970. The geology of the Caribbean crust, I: liams Beata Ridge. Tectonophysics 10: 495± 1982. Surface currents near the Greater and 513. Lesser Antilles. Int. Hydrogr. Rev. 59: Fox, P. J., E. Schreiber, and B. C. Heezen 67±78 1971. The geology of the Caribbean crust: Duque-Caro, H. Tertiary sediments, granitic and basic 1990. Major Neogene events in Panamic rocks from the Aves Ridge. Tectono- South America. In R. Tsuchi (ed.), Pa- physics 12: 89±109. ci®c Neogene events, their timing, na- Frailey, C. D., K. E. Campbell, and L. Romero- ture and interrelationships: 101±114. Pittman Tokyo: Univ. Tokyo Press. 1996. Early proboscideans in the Amazon Ba- sin of South America. J. Vertebr. Pa- Eberle, W., W. Hirdes, R. Muff, and M. Pelaez leontol. 16 (Suppl.): 34A. 1982. The geology of the Cordillera Septen- Frakes, L. A. trional (Dominican Republic). In W. 1979. Climates throughout geological times. Snow et al. (eds.), Trans. 9th Caribbean New York: Elsevier. Geol. Conf., Santo Domingo, Domini- Frost, S. H., J. L. Harbour, D. K. Beach, M. J. can Republic, 1: 619±632. Realini, and P. M. Harris Edgar, N. T., J. B. Saunders, H. M. Bolli, R. E. 1983. Oligocene reef trace developments, Boyce, W. S. Broecker, T. W. Donnelly, southwestern Puerto Rico. Sedimenta J. M. Gieskes, W. W. Hay, R. M. Ho- 9: 1±144. rowitz, F. Maurrasse, H. P. Nieto, W. Galloway, W. E., D. G. Bebout, W. L. Fisher, J. Prell, I. P. Silva, W. R. Riedel, N. B. Dunlap, R. Cabrera-Castro, J. E. Schneidermann, and L. S. Waterman Lugo Rivera, and T. M. Scott 1973. Initial Reports of the Deep Sea Drilling 1991. Cenozoic. In A. Salvador (ed.), The ge- Project, vol. 15. Washington, D.C.: ology of North America, vol. J, The U.S. Gov. Printing Of®ce. Gulf of Mexico basin: 245±324. Boul- Emery, K. O., and E. Uchupi der, CO: Geological Society of Ameri- 1972. Western North Atlantic Ocean: Topog- ca. raphy, rocks, structure, water, life, and GarcõÂa, E., and F. Harms sediments. Am. Ass. Pet. Geol. Mem. 1988. Informe del mapa geologico de la Re- 17: 1±532. puÂblica Dominicana, escala 1:100 000: Erikson, J. P., and J. L. Pindell Hoja San Juan. Santo Domingo: Direc- 1993. Analysis of subsidence in northeastern cioÂn General de MinerõÂa. Venezuela as a discriminator of tecton- Gayet, M., J. C. Rage, T. Sampere, and P. Y. Gag- ic models for northern South America. nier Geology 21: 945±948. 1992. ModaliteÂ des exchanges de vertebreÂs Escalante, G. continentaux entre l'AmeÂrique du Nord 1990. The geology of southern Central Amer- et l'AmeÂrique du Sud au CretaceÂ su- ica and western Colombia. In G. Dengo peÂrior et au PaleÂoceÁne. Soc. geÂol. and J. E. Case (eds.), The Geology of France Bull. 163: 781±791. North America, vol. H, The Caribbean Geddes, A. J. region: 201±230. Boulder, CO: Geolog- 1994. Jamaica±Geology, 1:250 000 map. ical Society of America. Kingston, Jamaica: Mines and Geology Eva, A. N., and N. A. McFarlane Division, Ministry of Mining and Nat- 1985. Tertiary to early Quaternary carbonate ural Resources. facies relationships in Jamaica. In Gingerich, P. D. Trans. 4th Latin American Geol. Conf., 1985. South American mammals in the Paleo- Port-of-Spain, Trinidad, 1979, 1: 210± cene of North America. In F. G. Stehli 219. and S. D. Webb (eds.), The great Amer- 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

ican biotic interchange: 123±137. New tonic history of the Southeastern ter- York: Plenum Press. mination of the Cordillera Central, Do- Gomberg, D. minican Republic. In P. Mann, G. Drap- 1974. Geology of the Pourtales Terrace. Flor- er, and J. F. Lewis (eds.), Geologic and ida Sci. 37, suppl. 1: 1±15. tectonic development of the North GonzaÂlez de Juana, C., J. M. Iturralde de Arozena, America±Caribbean plate boundary in and X. Picard Cadillat Hispaniola. Geol. Soc. Am. Spec. Pap. 1980. GeologõÂa de Venezuela y de sus cuen- 262: 315±336. cas petrolõÂferas, 2 vols. Caracas: Edi- Heubeck, C., P. Mann, J. Dolan, and S. Monechi ciones Foninves. 1991. Diachronous uplift and recycling of Graham, A. sedimentary basins during Cenozoic 1990. Late Tertiary microfossil ¯ora from the tectonic transpression, northeastern Ca- Republic of Haiti. Am. J. Bot. 77: 911± ribbean plate margin. Sediment. Geol. 926. 70: 1±32. Guyer, D., and J. Savage Hine, A. C. 1987. Cladistic relationships among anoles 1997. Structural, stratigraphic, paleoceano- (Sauria: Iguanidae). Syst. Zool. 35: graphic development of the margins of 509±531. the Florida Platform. In A. F. Randazzo 1992. Anole systematics revisited. Ibid. 41: and D. S. Jones (eds.), The geology of 89±110. Florida: 169±194. Gainesville, FL: Hamilton, W. B., Univ. Florida Press. 1988. Plate tectonics and island arcs. Geol. Holcombe, T. L., and N. T. Edgar Soc. Amer. Bull. 100: 1503±1527. 1990. Late Cretaceous and Cenozoic evolu- Haq, B. U., J. Hardenbol, and P. R. Vail tion of Caribbean ridges and rises with 1987. Chronology of ¯uctuating sea levels special references to paleogeography. since the Triassic. Science 235: 1156± In A. Azzaroli (ed.), Biogeographical 1167. Hay, W. W., and N. Wold aspects of insularity, Atti Convegni 1996. A simpler plate-tectonic history for the Lincei 85: 611±626. Rome: Accademia Caribbean. Zentbl. Geol. PalaÈontol. Teil Nazionale dei Lincei. 1 (7/8): 917±934. Holcombe, T. L., J. W. Ladd, G. Westbrook, N. T. Heaney, L. R., and J. C. Regalado Algar, and C. L. Bowland 1998. Vanishing treasures of the Philippine 1990. Caribbean marine geology; ridges and rain forest. Chicago: The Field Muse- basins of the plate interior. In G. Dengo um. and J. E. Case (eds.), The geology of Hedges, S. B. North America, vol. H, The Caribbean 1996a. The origin of West Indian amphibians region: 231±260. Boulder, CO: Geolog- and reptiles. In R. Powell and R. W. ical Society of America. Henderson (eds.), Contributions to Hoorn, C., J. Guerrero, G. A. Sarmientos, and M. West Indian herpetology: a tribute to A. Lorente Albert Schwartz: 95±128. Ithaca, NY: 1995. Andean tectonics as a cause for chang- Society for the Study of Amphibians ing drainage patterns in Miocene north- and Reptiles. ern South America. Geology 23: 237± 1996b. Historical biogeography of West Indian 240. vertebrates. Annu. Rev. Ecol. Syst. 27: Humphries, C. J. 163±196. 1992. Cladistic biogeography. In P. L. Forey Hedges, S. B., C. Hass, and L. Maxson (ed.), Cladistics: A practical course in 1992. Caribbean biogeography: molecular ev- systematics: 137±159. Oxford: Claren- idence for dispersal in West Indian ter- don Press. restrial vertebrates. Proc. Natl. Acad. Hunter, V. F. Sci. 89: 1909±1913. 1978. Notes on the Tertiary stratigraphy of 1994. Reply: Toward a biogeography of the Margarita Island. Geol. Minjb. 57: Caribbean. Cladistics 10: 43±55. 189±192. Henderson, I. M. Issac del Corral, J. 1991. Biogeography without area? Aust. Syst. 1940. El geosinclinal cubano. Rev. Soc. Cu- Bot. 4: 59±71. bana Ing. 34(4): 485±623. Heubeck, C., and P. Mann Iturralde-Vinent, M. A. 1991. Structural geology and Cenozoic tec- 1969. Principal characteristics of Cuban Neo- 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 65

gene stratigraphy. Amer. Assoc. Pet. 1996d. Cuba: El archipieÂlago volcaÂnico Paleo- Geol. Bull. 53: 1938±1955. ceno-Eoceno. Ibid. 231±246. 1972. Principal characteristics of Oligocene 1996e. Magmatismo de margen continental de and Lower Miocene stratigraphy of Cuba. Ibid. 121±130. Cuba. Ibid. 56: 2369±2379. 1997a. Meeting Reports. 1. Stratigraphy and 1975. Problems in the application of two correlation of Cretaceous volcanic arc modern hypothesis to Cuba and the Ca- rocks, Dominican Republic (IGCP ribbean region. Ibid. 59: 838±855. 364): July 1997. J. Pet. Geol. 20: 489± 1976±77. EstratigrafõÂa de la zona Calabazas- 491 Achotal, MayarõÂ Arriba, Oriente. Rev. 1997b. IntroduccioÂn a la geologõÂa de Cuba. In MinerõÂa en Cuba 5: 9±23 (pt. 1), 6: 32± G. Furrazola BermuÂdez and K. N. 40 (pt. 2). Cambra (eds.), Estudios sobre geologõÂa 1978. Los movimientos tectoÂnicos de la etapa de Cuba: 35±68. La Habana: Centro de desarrollo platafoÂrmico de Cuba. Nacional de InformacioÂn GeoloÂgica. Geol. Mijnb. 57: 205±212. Iturralde-Vinent, M. A., and R. D. E. MacPhee 1981. Nuevo modelo interpretativo de la ev- 1996. Age and paleogeographical origin of olucioÂn geoloÂgica de Cuba. Rev. Cienc. Dominican amber. Science 273: 1850± Tierra Espacio 3: 51±90. 1852. 1982. Aspectos geoloÂgicos de la biogeografõÂa Iturralde-Vinent, M. A., and M. Norell de Cuba. Ibid. 5: 85±100. 1996. Synopsis of Late Jurassic marine rep- 1988a. Naturaleza geoloÂgica de Cuba. La Ha- tiles from Cuba. Am. Mus. Novitates bana: Editorial CientõÂ®co-TeÂcnica. 3164: 17 pp. 1988b. Consideraciones generales sobre el Iturralde-Vinent, M. A., G. Hubbell, and R. Rojas magmatismo de margen continental de 1996a. Catalog of Cuban fossil Elasmobran- Cuba. Rev. TecnoloÂgica 18(4):17±24. chii (Paleocene-Pliocene) and paleo- 1991. Deslizamientos y descensos del terreno geographic implications of their Lower en el ¯anco meridional de la Sierra to Middle Miocene occurrence. J. Geol. Maestra, Cuba sudoriental. In Morfo- Soc. Jamaica 31: 7±22. tectoÂnica de Cuba Oriental: 24±27. La Iturralde-Vinent, M. A., G. MillaÂn, L. Korpas, E. Habana: Inst. GeografõÂa, Academia de Nagy, and J. PajoÂn Ciencias de Cuba. 1996b. Geological interpretation of the Cuban 1992. A short note on the Cuban late Maas- K-Ar database. In M. A. Iturralde-Vi- trichtian megaturbidite (an impact de- nent (ed.), Cuban ophiolites and vol- rived deposit?). Earth Planet. Sci. Lett. canic arcs. UNESCO International 109: 225±228. Geological Correlation Program Project 1994a. Cuban geology: a new plate tectonic 364, Special Contribution 1: 48±69. synthesis. J. Pet. Geol. 17: 39±70. Miami, FL. 1994b. Interrelationships of the terranes in Jackson, T., and S. Donovan western and central Cuba: comment. 1994. Tobago. In S. K. Donovan and T. A. Tectonophysics 234: 345±348. Jackson (eds.), Caribbean geology, an 1994c. Meetings Reports: tectonostratographic introduction: 193±208. Kingston, Ja- correlation of the NW Caribbean: Do- maica: The University of the West In- minican Republic. J. Pet. Geol. 17: dies Publishers' Association. 243±245. Jackson, T. A., and E. Robinson 1995. Meetings Reports: Caribbean ophiolites 1994. The Netherlands and Venezuelan Antil- and volcanic arcs: Jamaica, October les. In S. K. Donovan and T. A. Jackson 1994: J. Pet. Geol. 18: 234±235. (eds.), Caribbean geology, an introduc- 1996a. Introduction to Cuban geology and tion: 249±264. Kingston, Jamaica: The geophysics. In M. A. Iturralde-Vinent University of the West Indies Publish- (ed.), Cuban ophiolites and volcanic ers Association. arcs. UNESCO International Geologi- Jany, I., K. M. Scanlon, and A. Mauffret cal Correlation Program Project 364, 1990. Geological interpretation of combined Special Contribution 1: 3±35. Miami, seabeam Gloria and seismic data from FL. Anegada Passage (Virgin Islands, 1996b. Cuba: El arco de islas volcaÂnicas del North Caribbean). Mar. Geophys. Res. CretaÂcico. Ibid. 179±189. 12: 173±196. 1996c. EstratigrafõÂa del arco volcaÂnico en Jones, B. Cuba. Ibid. 190±227. 1994. The Cayman Islands. In S. K. Donovan 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

and T. A. Jackson (eds.), Caribbean ge- Kluge, A. ology, an introduction: 87±110. King- 1988. Parsimony in vicariance biogeography: ston, Jamaica: The University of the a quantitative method and a Greater West Indies Publishers Association. Antillean example. Syst. Zool. 37: Kay, R. F., and R. H. Madden 315±328. 1997. Paleogeography and paleoecology. In 1989. A concern for evidence and a phylo- R. F. Kay, R. H. Madden, R. L. Cifelli, genetic hypothesis of relationships and J. J. Flynn (eds.), Vertebrate pale- among Epicrates (Boidae, Serpentes). ontology in the Neotropics: the Mio- Ibid. 38: 7±25. cene fauna of La Venta, Colombia: Kolarsky, R. A., P. Mann., and S. Monechi 520±550. Washington, D. C.: Smith- 1995a. Stratigraphic development of south- sonian Institution Press. western Panama as determined from in- Kennett, J. P. tegration of marine seismic data and 1985. The Miocene oceans: paleoceanogra- onshore geology. In P. Mann (ed.), phy and biogeography. Geol. Soc. Am. Geologic and tectonic development of Mem. 163: 1±340. the Caribbean plate boundary in south- Kesler, S. E., N. Russell, C. Reyes, L. Santos, A. ern Central America. Geol. Soc. Am. RodrõÂguez, and L. Fondeur Spec. Pap. 295: 159±200. 1991a. Geology of the Maimon Formation, Kolarsky, R. A., P. Mann., and W. Montero Dominican Republic. In P. Mann, G. 1995b. Island arc response to shallow subduc- Draper, and J. F. Lewis (eds.), Geologic tion of the Cocos ridge, Costa Rica. In and tectonic development of the North P. Mann (ed.), Geologic and tectonic America±Caribbean plate boundary in development of the Caribbean plate Hispaniola. Geol. Soc. Am. Spec. Pap. boundary in southern Central America. 262: 173±186. Geol. Soc. Am. Spec. Pap. 295: 235± Kesler, S. E., N. Russell, J. Polanco, K. McCurdy, and G. L. Cumming 262. 1991b. Geology and geochemistry of the Early Ladd, J., T. Shih, and C. Tsai Cretaceous Los Ranchos Formation, 1981. Cenozoic tectonics of Central Hispan- Central Dominican Republic. In P. iola and adjacent Caribbean Sea. Am. Mann, G. Draper, and J. F. Lewis (eds.), Assoc. Pet. Geol. Bull. 65: 466±489. Geologic and tectonic development of Larue, D. K. the North America±Caribbean plate 1994. Puerto Rico and Virgin Islands. In S. boundary in Hispaniola. Geol. Soc. K. Donovan and T. A. Jackson (eds.), Am. Spec. Pap. 262: 187±202. Caribbean geology, an introduction: Khudoley, C. 151±165. Kingston, Jamaica: The Uni- 1967. Principal features of Cuban geology. versity of the West Indies Publishers Am. Assoc. Pet. Geol. Bull. 51: 668± Association. 677. Larue, D. K., and H. F. Ryan Khudoley, K., and A. A. Meyerhoff 1990. Extensional tectonics in Mona Passage, 1971. Paleogeography and geological history Puerto Rico and Hispaniola. In D. K. of Greater Antilles. Geol. Soc. Am. Larue and G. Draper (eds.), Trans. 12th Mem. 129: 1±199. Caribbean Geol. Conf., St. Croix, U.S. Kinder, T. H. Virgin Is.: 223±230. Miami, FL: Miami 1983. Shallow currents in the Caribbean Sea Geological Society. and the Gulf of Mexico as observed Larue, D. K., J. Joyce, and H. Ryan with satellite-tracked drifters. Bull. 1990. Neotectonics of the Puerto Rico Mar. Sci. 33: 239±246. Trench: extensional tectonism and Kinder, T. H., G. W. Heburn, and A. W. Green forearc subsidence. In D. K. Larue and 1985. Some aspects of Caribbean circulation. G. Draper (eds.), Trans. 12th Caribbean Mar. Geol. 68: 25±52. Geol. Conf., St. Croix, U.S. Virgin Is.: Klitgord, K. D., and H. Schouten 231±247. Miami, FL: Miami Geologi- 1986. Plate kinematics of the central Atlantic. cal Society. In P. R. Vogt and B. E. Tucholke (eds.), LebroÂn, M. C., and M. R. Per®t The geology of North America, vol. M, 1993. Stratigraphic and petrochemical data The western North Atlantic region: support subduction polarity reversal of 351±378. Boulder, CO: Geological So- the Cretaceous Caribbean island arc. J. ciety of America. Geol. 101: 389±396. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 67

Leclere, A., and J. F. Stephan sins of South America. Am. Assoc. Pet. 1985. Evolution geÂodynamique des points Geol. Mem. 62: 757±780. chauds. In A. Mascle (ed.), Caribbean MacLaughlin, P. P., I. P. Gill, and W. A. van der geodynamics: 21±34. Paris: Ed. Tech- Bold nip. 1995. Biostratigraphy, paleoenvironments Leonov, Yu. G., and V. E. Khain and stratigraphic evolution of the Neo- 1987. Problems of a systematic approach to gene of St. Croix, U.S. Virgin Islands. the study and correlation of tectonic Micropaleontology 41: 293±320. phases. In Yu. G. Leonov and V. E. MacPhee, R. D. E., and D. A. Grimaldi Khain (eds.), Global correlation of tec- 1996. Mammal bones in Dominican amber. tonic movements: 1±12. New York: Nature 380: 489±490. Wiley. MacPhee, R. D. E., and M. A. Iturralde-Vinent Lewis, G. E., and J. A. Straczek 1994. First Tertiary land mammal from Great- 1955. Geology of south-central Oriente Prov- er Antilles: an Early Miocene sloth ince, Cuba. U. S. Geol. Surv. Bull. 975- (Xenarthra, Megalonychidae) from D: 171±336. Cuba. Am. Mus. Novitates 3094: 13 Lewis, J. F., G. Draper, C. Bourdon, C. Bowin, P. pp. Mattson, F. Maurrasse, F. Nagle, and G. 1995. Origin of the Greater Antillean land Pardo mammal fauna, 1: New Tertiary fossils from Cuba and Puerto Rico. Ibid. 3141: 1990. Geology and tectonic evolution of 31 pp. northern Caribbean margin. In G. Den- MacPhee, R. D. E., and A. R. Wyss go and J. E. Case (eds.), The Geology 1990. Oligo-Miocene vertebrates from Puerto of North America, vol. H, The Carib- Rico, with a catalog of localities. Am. bean region: 77±140. Boulder, CO: Mus. Novitates 2965: 45 pp. Geological Society of America. MacPhee, R. D. E., D. C. Ford, and D. A. Mc- Lewis, J. F., A. Amarante, G. Bloise, J. G. Farlane JimeÂnez, and H. D. DomõÂnguez 1989. Pre-Wisconsinan mammals from Ja- 1991. Lithology and stratigraphy of Upper maica and models of late Quaternary Cretaceous volcanic and vulcaniclastic extinction in the Greater Antilles. Quat. rocks of the Tireo Group, Dominican Res. 31: 94±106. Republic and correlation with the Mas- Malfait, B., and M. Dinkelmann sif du Nord in Haiti. In P. Mann, G. 1972. Circum-Caribbean tectonics and igne- Draper, and J. F. Lewis (eds.), Geologic ous activity and the evolution of the and tectonic development of the North Caribbean plate. Geol. Soc. Am. Bull. America±Caribbean plate boundary in 83: 251±272. Hispaniola. Geol. Soc. Am. Spec. Pap. Mann, P., and K. Burke 262: 143±164. 1984. Neotectonics of the Caribbean. Rev. LoÂpez-Ramos, E. Geophys. Space Phys. 22: 309±362. 1975. Geological summary of the Yucatan Mann, P., and R. A. Kolarsky Peninsula. In A. E. M. Nairn and F. G. 1995. East Panama deformed belt: structure, Stehli (eds.), The ocean basins and age, and neotectonic signi®cance. In P. margins, vol. 3, The Gulf of Mexico Mann (ed.), Geologic and tectonic de- and Caribbean: 257±282. New York: velopment of the Caribbean plate Plenum Press. boundary in southern Central America. Lugo, J., and P. Mann Geol. Soc. Am. Spec. Pap. 295: 111± 1995. Jurassic-Eocene tectonic evolution of 130. Maracaibo basin, Venezuela. In A. J. R. Mann, P., C. Schubert, and K. Burke Tankard, R. Suarez Soruco, and H. J. 1990. Review of Caribbean neotectonics. In Welsink (eds.), Petroleum basins of G. Dengo and J. E. Case (eds.), The South America. Am. Assoc. Pet. Geol. geology of North America, vol. H, The Mem. 62: 699±725. Caribbean region: 307±338. Boulder, Macellari, C. E. CO: Geological Society of America. 1995. Cenozoic sedimentation and tectonics Mann, P., G. Draper, and J. F. Lewis of the southwestern Caribbean pull- 1991. An overview of the geologic and tec- apart basin, Venezuela and Colombia. tonic development of Hispaniola. In P. In A. J. R. Tankard, R. Suarez Soruco, Mann, G. Draper, and J. F. Lewis (eds.), and H. J. Welsink (eds.), Petroleum ba- Geologic and tectonic development of 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

the North America±Caribbean plate go and J. E. Case (eds.), The geology boundary in Hispaniola. Geol. Soc. of North America, vol. H, The Carib- Am. Spec. Pap. 262: 1±51. bean region: 141±166. Boulder, CO: Mann, P., F. W. Taylor, R. L. Edwards, and T-L. Geological Society of America. Ku Mayer, G. C., and J. D. Lazell 1995. Actively evolving microplate formation 1988. Signi®cance of frog in amber. Science by oblique collision and sideways mo- 239: 1477. tion along strike-slip faults: an example Maze, W. from the northeastern Caribbean plate 1984. Jurassic La Quinta Formation in the Si- margin. Tectonophysics 246: 1±69. erra de PerijaÂ, northwestern Venezuela. Marshall, L. G. Geology and tectonic environment of 1985. Geochronology and land-mammal red beds and volcanic rocks. In W. E. biochronology of the Transamerican Bonini, R. B. Hardgrave, and R. Sha- Faunal Interchange. In F. G. Stehli and gam (eds.), The Caribbean±South S. D. Webb (eds.), The great American American plate boundary and regional biotic interchange, pp. 49±85. New tectonics. Geol. Soc. Am. Mem. 162: York: Plenum Press. 263±282. Marton, G., and R. T. Buf¯er 1993. Application of simple-shear model to McFarlane, E., and L. S. Menes the evolution of passive continental 1991. Lower Cretaceous. In A. Salvador margins of the Gulf of Mexico Basin. (ed.), The geology of North America, Geology 21: 495±498. vol. J, The Gulf of Mexico basin: 181± Mascle, A., M. Cazes, and P. Le Quellec 204. Boulder, CO: Geological Society 1985. Structure des marges et bassins caraõÈ- of America. bes: une revue. In A. Mascle (ed.), Ca- McKenna, M. C. ribbean geodynamics: 1±20. Paris: Ed. 1973. Sweepstakes, ®lters, corridors, Noah's Technip. arks, and beached Viking funeral ships Masson, D. G., and K. M. Scanlon in paleogeography. In D. H. Tarling and 1991. The neotectonic setting of Puerto Rico. S. K. Runcorn (eds.), Implications of Geol. Soc. Am. Bull. 103: 144±154. continental drift to the earth sciences, Matthew, W. D. 1: 293±306. New York: Academic 1918. Af®nities and origins of the Antillean Press. mammals. Geol. Soc. Am. Bull. 29: Meyerhoff, A. A., and C. W. Hatten 657±666. 1968. Diapiric structure in Central Cuba. Am. Mattson, R. Assoc. Pet. Geol. Mem. 8: 315±357. 1984. Caribbean structural breaks and plate 1974. Bahamas salient of North America: tec- movements. In W. E. Bonini, R. B. tonic framework, stratigraphy and pe- Hardgrave, and R. Shagam (eds.), The troleum potential. Am. Assoc. Pet. Caribbean±South American plate Geol. Bull. 58: 1201±1239. boundary and regional tectonics. Geol. Meyerhoff, H. A. Soc. Am. Mem. 162: 131±152. 1933. The geology of Puerto Rico. Univ. Maurrasse, F. Puerto Rico Monograph Series B, no. 1982. Survey of the geology of Haiti; guide 1: 1±360. to ®eld excursions in Haiti. Florida Geol. Soc. Spec. Pub.: 1±103. Michalzik, D. 1990. Stratigraphic correlation of the Circum- 1987. SedimentacioÂn y sucesioÂn de facies en Caribbean region. In G. Dengo and J. un margen continental pasivo del TriaÂs- E. Case (eds.), The geology of North ico al CretaÂcico temprano del noreste America, vol. H, The Caribbean region: de la Sierra Madre Oriental, MeÂxico. pls. 4, 5A, and 5B. Boulder, CO: Geo- Univ. Nac. Auton. Linares, Fac. Cienc. logical Society of America. Tierra Actas 2: 27±31. Maurrasse, F., and G. Sen Miller, K. G., G. S. Mountain, Leg 150 Shipboard 1991. Impacts, tsunamis, and the Haitian Cre- Party, and members of New Jersey taceous-Tertiary boundary layer. Sci- Coastal Plain Drilling Project ence: 252: 1690±1693 1996. Drilling and dating New Jersey Oligo- Maury, R. C., G. K. Westbrook, P. E. Baker, P. cene-Miocene sequences: ice volume, Bouysse, and D. Westcamp global sea level, Exxon record. Science 1990. Geology of Lesser Antilles. In G. Den- 271: 1092±1095. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 69

Mills, B. Mullins, H. T., Gardulski, A. F., S. W. Wise, and 1994. Venezuela surface geology, scale 1: J. Applegate 1 000 000. Caracas: CXY International. 1987. Middle Miocene oceanographic event Molinari, R. L., D. K. Atwood, C. Duckett, M. in the eastern Gulf of Mexico: impli- Spillane, and I. Brooks cations for seismic stratigraphic succes- 1979. Surface currents in the Caribbean Sea sion and Loop Current/Gulf Stream cir- as deduced from satellite tracked drift- culation. Geol. Soc. Am. Bull. 98: 702± ing buoys. Proc. Gulf Caribb. Fish. 713. Inst. 1979: 106±113. Nagle, F. Monroe, W. H. 1972. Rocks from the sediments of escarp- 1980. Geology of the middle Tertiary forma- ments on the Aves ridge. In Trans. 6th tions of Puerto Rico. U. S. Geol. Surv. Caribbean Geol. Conf., Caracas: 409± Prof. Pap. 954: 1±93. 423. Caracas: Cromotik. Montadert, L., J. Sigal., B. Biju-Duval, A. Mascle, Nagy, E., K. BrezsnyaÂnszky, A. Brito, D. CoutõÂn, and A. Eva F. Formell, G. Franco, P. Gyarmati, G. 1985. PreÂcisions sur la stratigraphie des seÂries Radocz, and P. Jakus creÂtaceÂes et la tectonique neÂogeÁne de la 1983. ContribucioÂn a la geologõÂa de Cuba JamaõÈque. In A. Mascle (ed.), Carib- Oriental. La Habana: Editorial CientõÂ- bean geodynamics: 419±426. Paris: Ed. ®co Technip. Nelson, G., and N. I. Platnick Montgomery, H., E. A. Pessagno, and J. L. Pin- 1981. Systematics and biogeography: cladis- dell tics and vicariance. New York: Colum- 1994. A 195 Ma terrane in a 165 Ma Sea: bia Univ. Press. Paci®c origin of the Caribbean plate. Nemec, M. C., GSA Today 4(1): 1±7 1980. A two phase model for the tectonic MoraÂn-Zenteno, D. J., P. Corona-ChaÂvez, and G. evolution of the Caribbean. In W. Snow Tolson et al. (eds.), Trans. 9th Caribbean Geol. 1996. Uplift and subduction erosion in south- Conf., Santo Domingo, Dominican Re- western Mexico since the Oligocene; public, 1: pp. 24±34. pluton geometry constraints. Earth Nunn, P. D. Planet. Sci. Lett. 141: 51±65. 1994. Oceanic islands. Oxford: Blackwell Morgan, G. S. Nutall, C. P. 1994. Late Quaternary fossil vertebrates from 1990. A review of the Tertiary non-marine the Cayman Islands. In M. A. Brunt molluscan faunas of the Pebasian and and J. E. Davies (eds), The Cayman Is- other inland basins of northwestern lands: natural history and biogeogra- South America. Bull. Brit. Mus. (Nat. phy: 465±508. Boston: Kluwer Aca- Hist.), Geol. 45:165±371. demic. Ota, H. Morgan, G., and C. A. Woods 1998. Geographic patterns of endemism and 1986. Extinction and the zoogeography of speciation in amphibians and reptiles of West Indian land mammals. Biol. J. the Ryukyu archipelago, Japan, with Linn. Soc., 28: 167±203. special reference to their paleogeo- Morrone, J. J., and J. V. Crisci graphical implications. Res. Popul. 1995. Historical biogeography: introduction Ecol. 40: 189-204. to methods. Annu. Rev. Ecol. Syst. 26: Page, R. D. M., and C. Lydeard 373±401. 1994. Toward a cladistic biogeography of the Mueller, R. D., J.-Y. Royer, and L. A. Lawver Caribbean. Cladistics 10: 21±41. 1993. Revised plate motions relative to the Pardo, G. hotspots from combined Atlantic and 1975. Geology of Cuba. In A. E. M. Nairn Indian Ocean hotspot tracks. Geology and F. G. Stehli (eds.), The ocean ba- 21: 275±278. sins and margins, vol. 3, The Gulf of Muizon, C. de, and H. G. McDonald Mexico and Caribbean: 553±616. New 1995. An aquatic sloth from the Pliocene of York: Plenum Press. Peru. Nature 375: 224-227. Parnaud, F., Y. Gou, J.-C. Pascual, M. A. Capello, Mullins, H. T., and A. C. Newmann I. Truskowski, and H. Passalacqua 1979. Geology of the Miami Terrace and its 1995. Stratigraphic synthesis of northern Ven- paleoceanographic implications. Mar. ezuela. In A. J. R. Tankard, R. Suarez Geol. 30: 205±232. Soruco, and H. J. Welsink (eds.), Petro- 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

leum basins of South America. Am. planatory note to the tectonic map of Assoc. Pet. Geol. Mem. 62: 681±698. Cuba scale 1:500 000 [in Russian]. Per®t, M. R., and B. C. Heezen Moscow: Geological Institute, USSR 1978. The geology and evolution of the Cay- Academy of Sciences, Academia Nau- man Trench. Geol. Soc. Am. Bull. 89: ka. 1155±1174. Randazzo, A. F., and D. S. Jones Per®t, M. R., and E. E. Williams 1997. The geology of Florida. Gainesville, 1989. Geological constraints and biological FL: Univ Florida Press. retrodictions in the evolution of the Ca- RaÈsaÈnen, M. E., A. M. Linna, J. C. R. Santos, and ribbean sea and its surroundings. In C. F. R. Negri A. Woods (ed.), Biogeography of the 1995. Late Miocene tidal deposits in the Am- West Indies: 47±102. Gainesville, FL: azonian foreland basin. Science 269: Sandhill Crane Press. 386±389. Pessagno, E. Ravenne, C., P. Le Quellec, and P. Valery 1963. Planktonic foraminifera from the Juana 1985. DepoÃts carbonateÂs profonds des Baha- DõÂaz Formation, Puerto Rico. Micro- mas. In A. Mascle (ed.), Caribbean paleontology 9: 53±60. geodynamics: 255±271. Paris: Ed. Pindell, J. Technip. 1994. Evolution of the Gulf of Mexico and Richardson, P. L. the Caribbean. In S. K. Donovan and 1996. Tracking ocean eddies. In R. Gordon T. A. Jackson (eds.), Caribbean geolo- Pirie (ed.), Oceanography, contempo- gy, an introduction: 13±40. Kingston, rary readings in ocean science, 3d ed.: Jamaica: The University of the West In- 88±104. Oxford Univ. Press. dies Publishers Association. Roberts, C. M. Pindell, J. L., and S. Barrett 1997. Connectivity and management of Ca- 1990. Geological evolution of the Caribbean ribbean coral reefs. Science 278: 1454± region: a plate tectonic perspective. In 1457. G. Dengo and J. E. Case (eds.), The Robinson, E. geology of North America, vol. H, The 1965. Tertiary rocks of the Yallahs area. J. Caribbean region: 405±432. Boulder, Geol. Soc. Jamaica 7: 18±27. CO: Geological Society of America. 1994. Jamaica. In S. K. Donovan and T. A. Pinet, B., D. Lajat, P. Le Quellec, and P. Bouysse Jackson (eds.), Caribbean geology, an 1985. Structure of Aves Ridge and Grenada introduction: 111±127. Kingston, Ja- Basin from multichannel seismic data. maica: The University of the West In- In A. Mascle (ed.), Caribbean geodyn- dies Publishers Association. amics: 53±64. Paris: Ed. Technip. Rojas, R., M. A. Iturralde-Vinent, and P. W. Skel- Pszczolkowski, A. ton 1978. Geosynclinal sequences of the Cordil- 1995. Stratigraphy, composition and age of lera de Guaniguanico in western Cuba; rudist-bearing deposits. Revista Mexi- their lithostratigraphy, facies develop- cana de Ciencias GeoloÂgicas. 12: 272± ment and paleogeography. Acta Geol. 291 Polonica 28: 1±96. Rosen, D. E. 1986. Megacapas del Maestrichtiano de Cuba 1975. A vicariance model of Caribbean bio- occidental y central. Bull. Polish Acad. geography. Syst. Zool. 24: 431±464. Sci., Earth Sci. 34: 81±87. 1985. Geological hierarchies and biogeo- 1987. Paleogeography and paleotectonic evo- graphical congruence in the Caribbean. lution of Cuba and adjoining areas dur- Ann. Missouri Bot. Garden 72: 636± ing the Jurassic-Early Cretaceous. Ann. 659. Soc. Geol. Poloniae 57: 127±142. Rosencrantz, E. Pszczolkowski, A., and R. Flores 1990. Structure and tectonics of the Yucatan 1986. Fases tectoÂnicas del CretaÂcico y del Pa- basin, Caribbean sea, as determined leoÂgeno en Cuba occidental y central. from seismic re¯ection studies: Tecton- Bull. Polish Acad. Sci., Earth Sci. 34: ics 9: 1037±1059. 95±111. 1995. Opening of the Cayman trough and Pushcharovsky, Yu., A. Mossakovsky, G. Nekra- evolution of the northern Caribbean sov, S. Sokolov, and M. A. Iturralde- plate boundary. Ann. Meet. Geol. Soc. Vinent Am., abstracts with program, p. A-153. 1989. Tectonics of the Republic of Cuba. Ex- 1996. Basement structure and tectonics in the 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 71

Yucatan basin. In M. A. Iturralde-Vi- Tertiary time. Am. Assoc. Pet. Geol. nent (ed.), Cuban ophiolites and vol- Bull. 64: 359±373. canic arcs. UNESCO International Schwartz, A., and R. W. Henderson Geological Correlation Program Project 1991. Amphibians and reptiles of the West In- 364, Special Contribution 1: 36±47. dies: descriptions, distributions, and Miami, FL. natural history. Gainesville, FL: Univ. Ross, M. I., and C. R. Scotese Florida Press. 1988. A hierarchical tectonic model of the Shaw, P. R., and S. C. Cande Gulf of Mexico and Caribbean region. 1990. High-resolution inversion for South At- Tectonophysics 155: 139±168. lantic plate kinematics using joint al- Ross, M. I. timeter and magnetic anomaly data. J. 1992. PaleoGeographic Information System/ Geophys. Res. 95: 2625±2644. Mac. Earth in Motion Technologies, Sigurdsson, H., M. Leckie, G. Acton, and ODP Users Manual, 1±32. Leg 165 Scienti®c Party Roughgarden, J. 1997. Proceedings of the Ocean Drilling Pro- 1995. Anolis lizards of the Caribbean. Oxford gram, Initial Reports, Leg 165. College Univ. Press. Station, TX: Ocean Drilling Program. Rull, V. and C. Schubert Simpson, G. G. 1989. EvolucioÂn de la hipoÂtesis sobre el ori- 1956. Zoogeography of West Indian land mammals. Am. Mus. Novitates 1759: gen del Caribe. Interciencia 14(2): 74± 28 pp. 85. Smith, A. G., D. G. Smith, and B. M. Funnell Russell, N., and S. E. Kesler 1994. Atlas of Mesozoic and Cenozoic coast- 1991. Geology of the maar-diatreme complex lines. Cambridge: Cambridge Univ. hosting precious metal mineralization Press. at Pueblo Viejo, Dominican Republic. Sohl, N. F., E. MartõÂnez, P. SalmeroÂn-UrenÄa, and In P. Mann, G. Draper, and J. F. Lewis F. Soto-Jaramillo (eds.), Geologic and tectonic develop- 1991. Upper Cretaceous. In A. Salvador (ed.), ment of the North America±Caribbean The geology of North America, vol. J, plate boundary in Hispaniola. Geol. The Gulf of Mexico basin: 205±244. Soc. Am. Spec. Pap. 262: 203±216. Boulder, CO: Geological Society of Salvador, A. America. 1987. Late Triassic-Jurassic paleogeography Sou, T., G. Halloway, and M. Eby and origin of the Gulf of Mexico Basin. 1996. Effect of topographic stress on Carib- Am. Assoc. Pet. Geol. Bull. 71: 419± bean Sea circulation. J. Geophys. Res. 451. 101C7: 16,449±16,453. 1991. Origin and development of the Gulf of Stephan, J.-F., B. Mercier de Lepinay, E. Calais, Mexico Basin. In A. Salvador (ed.), M. Tardy, C. Beck, J.-C. Olivet, J.-M. The geology of North America, vol. J, Vila, P. Bouysse, A. Mauffret, J. Bour- The Gulf of Mexico basin: 389±444. gois, J.M. Thery, J. Tournon, R. Blan- Boulder, CO: Geological Society of chet, and J. Decourt America. 1990. Paleodynamic maps of the Caribbean: Saunders, J. B., P. Jung, and B. Biju-Duval 14 steps from Lias to present. Soc. 1986. Neogene paleontology in the northern geÂol. France Bull. (8) 6: 915±919. Dominican Republic, 1. Field surveys, Stockhert, B., W. V. Maresch, M. Brix, C. Kaiser, lithology, environment, and age. Am. A. Toetz, R. Kluge, and G. Kruckhans- Paleontol. Bull. 89: 1±79. Lueder Schlager, W., R. T. Buf¯er, D. Angstadt, and R. 1995. Crustal history of Margarita Islands L. Phair (Venezuela) in detail: constraint on the 1984. Geologic history of the southeastern Caribbean plate-tectonic scenario. Ge- Gulf of Mexico. In Initial Reports of ology 23: 787±790. the Deep Sea Drilling Project 77: 715± Stoddart, D. R., J. D. Taylor, F. R. Fosberg, and 738. Washington D.C.: U.S. Gov. Print- G. E. Farrow ing Of®ce. 1971. Geomorphology of Aldabra atoll. Phi- Schwan, W. los. Trans. R. Soc. London 260B: 31± 1980. Geodynamic peaks in alpinotype orog- 65. enies and changes in ocean ¯oor Tankard, A. J., R. SuaÂrez and H. J. Welsink (eds.) spreading during Late Jurassic±Late 1995. Petroleum basins of South America. 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Am. Assoc. Pet. Geol. Mem. 62: 1± gene of Central Haiti. Am. Paleontol- 792. ogy Bull. 79:1±136. Toloczyki, M., and I. RamõÂrez Viniegra, O. F. 1991. Mapa geoloÂgico de la RepuÂblica Dom- 1981. Great carbonate bank of Yucatan, inicana, escala 1:250 000, comp. by Di- southern Mexico. J. Pet. Geol. 3: 247± reccioÂn de MinerõÂa, D.R., and Bundes- 278. anstalt fuÈr Geowissenschaften und Webb, S. D. Rohstoffe, Germany. 1976. Mammalian faunal dynamics of the Tsuchi, R. great American interchange. Paleobi- 1993. Neogene evolution of Paci®c Ocean ology 2: 220±234. gateways with reference to the event- 1985. Late Cenozoic mammal dispersals be- biostratigraphy of Southwest Japan. In tween the Americas. In F. G. Stehli and F. Hehuwat, E. P. Utomo, and A. Dhar- S. D. Webb (eds.), The great American ma (eds.), Proceedings of an Interna- biotic interchange: 357±386. New tional Workshop on the Neogene Evo- York: Plenum Press. lution of Paci®c Ocean Gateways 1995. Biological implications of the Middle (UNESCO IGCP±355), Bandarlam- Miocene Amazon seaway. Science 269: pung, Indonesia, 3-5 August, 1993: 361±362. 2±9. Weidie, A. E., W. C. Ward, and R. H. Marshall 1979. Geology of Yucatan Platform. In Field Turner, M. D. trip guidebook: geology of Cancun, 1972. Geology of the San Sebastian area of Quintana Roo, Mexico: 50±68. Mid- northwestern Puerto Rico and the tec- land, TX: West Texas Geological So- tonic development of Puerto Rico and ciety. surrounding areas. Unpubl. Ph.D. diss., Westercamp, D., P. Andreieff, P. Bouysse, A. Mas- Univ. Kansas, Lawrence, KS. cle, and J. C. Baubron Vandel, A. 1985. The Grenadines, southern Lesser Antil- 1973. Les isopodes terrestres et cavernicoles les: I. Stratigraphy and vulcano-struc- de l'Ile de Cuba. In T. Orghidan et al. tural evolution. In A. Mascle (ed.), Ca- (eds.), ReÂsultats des expeÂditions bio- ribbean geodynamics: 109±118. Paris: speÂleologiques cubano-roumaines aÁ Ed. Technip. Cuba, vol. 1: 153±190. Bucarest: Edi- Woods, C. A. torial Academiei Republicii Socialiste 1989. The biogeography of West Indian ro- Romania. dents. In C. A. Woods (ed.), Biogeog- Van den Bold, W. raphy of the West Indies: 741±797. 1981. Distribution of Ostracoda in the Neo- Gainesville, FL: Sandhill Crane Press.

APPENDIX 1: RECONSTRUCTING CARIBBEAN PALEOGEOGRAPHY: AN ANALYTICAL GUIDE This appendix presents additional maps, tables, stratigraphic position and fossil context (where stratigraphic columns, and other information rel- applicable) in light of the new chronostratigraphic evant to interpreting the geological history of the framework provided by Berggren et al. (1995). Caribbean region and the Late Tertiary paleogeo- Use of this new scale necessitated modi®cation of graphical reconstructions presented elsewhere in published time-stratigraphic positions of several this paper. Information in the appendix is linked formations. Paleoenvironmental interpretations of to tables 1±4, which analytically summarize rel- each unit were assessed using data and interpre- evant geological data and literature. Each table tations in the literature as well as the results of covers a particular temporal interval, organized by our own ®eldwork. geological units (®rst column). These units, de- Yucatan Peninsula ®ned in terms of their current location on the ge- The Yucatan Peninsula (Maya Block or MB; ode (second column), are the entities traced on the ®g. 14) has been part of the North American plate paleogeographic maps. The third and fourth col- since the late Cretaceous (Marton and Buf¯er, umns synthesize the evidence available for spe- 1993). Tectonic activity has been largely concen- ci®c environmental indicators, as preserved in trated in the southwestern part of MB, far from rock-stratigraphic records. the areas immediately relevant to this study. Nev- Age of each entity was veri®ed according to ertheless, there is considerable information on this 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 73 unit and its Caribbean borderland (Butterlin and fore the Late Eocene (probably since the Maas- Bonet, 1966; Case, 1975; Lopez-Ramos, 1975; trichtian; Case, 1975; Mascle et al., 1985; Rosen- Weidie et al., 1979; Viniegra, 1981; Mascle et al., crantz, 1990, 1996). Therefore, for most of the 1985; see also the paleogeographical maps com- Cenozoic, if not longer, the Cuban terranes (wher- piled in papers edited by Salvador, 1991). ever they were then stationed) and MB have been Stratigraphic data indicate that MB was prob- separated by a signi®cant water barrier (see also ably uplifted from Jurassic through Barremian discussion in MacPhee and Iturralde-Vinent, time (early Cretaceous). The block was covered 1995). by shallow seas from late Aptian through Late Eocene, although small islands probably existed Northern Central America, Nicaragua Rise, and on it from time to time (Salvador, 1991; Mc- Western Jamaica Farlane and Menes, 1991). Brief emergence of Geologically, northern Central America (com- MB may have occurred around 88±90 Ma, as a prising Nicaragua, Honduras, Guatemala, and hiatus has been reported within the Turonian (Lo- southern Mexico) consists of a single tectonic ter- pez Ramos, 1975; Weidie et al., 1979; Viniegra, rane, the Chortis Block (CB; ®g. 15). In most cur- 1981). This correlates well with a Turonian hiatus rent plate tectonic models, CB is assumed to have represented in rock sequences in the northern Ca- originated off Mexico on the Paci®c margin of ribbean and Gulf of Mexico (Meyerhoff and Hat- North America, and to have been rotated into its ten, 1968; Schlager et al., 1984; Pszczolkowski present position late in the Cenozoic (Malfait and and Flores, 1986; Iturralde-Vinent, 1994a). Dinkelmann, 1972; Donnelly et al., 1990). The Seismic refraction studies of the Chicxulub bo- Monagua±Polochic deformed system, the hinge lide crater on the NW corner of the Yucatan Pen- zone between CB and terranes farther north in insula indicate that islands produced by impact Mexico, was particularly active from Middle Eo- debris might have had a brief existence there dur- cene through Middle Miocene (Pindell, 1994; ing the early Paleocene, before they vanished and MoraÂn-Zenteno et al., 1996). Emplacement of CB were covered by younger marine carbonate de- is generally correlated with the evolution of the posits (Buf¯er et al., 1995). Unfortunately, there Cayman Trench system (Pindell, 1994; Rosen- is not enough information available on these crantz, 1995; MoraÂn-Zenteno et al., 1996). Ap- structures to offer a plausible reconstruction of preciable parts of CB have been uplifted for most their position or duration. of the Cenozoic (Maurrasse, 1990; Donnelly et Several geologic units in western and central al., 1990). Cuba (Guaniguanico, Pinos, and Escambray ter- Geographically, the Nicaragua Rise (NR) is the ranes) are allochthonous (Iturralde-Vinent, 1994a, prolongation of CB into the Caribbean Sea. Geo- 1994b). These terranes were detached from their logically, however, these units are quite different original location along the Yucatan borderland (Holcombe et al., 1990; Donnelly et al., 1990). (Pszczolkowski, 1987; Rosencrantz, 1990) be- Stratigraphic data for NR are spotty, but isolated tween the Late Paleocene and the Middle Eocene, wells, dredge hauls, and seismic stratigraphy are during the formation of the Greater Antilles Fold- suf®cient to create a broad-brush picture. Most belt (®g. 5; Bralower and Iturralde-Vinent, 1997). importantly, these data con®rm that NR and the However, the tectonic processes involved (intense Western Jamaican Block (WJ) have shared a con- folding, thrust faulting, metamorphism) took place siderable amount of geological history. at signi®cant depth, and it is quite improbable that Assuming that the Cretaceous basement of WJ any of these terranes include units from MB that can be correlated with basement rocks of NR (also were actually emergent at the time of their incor- known to be Cretaceous [Holcombe et al., 1990]), poration into Cuba. it appears that this terrane (i.e., Western Jamaica According to Butterlin and Bonet (1966) and and Nicaragua Rise together) was the site of vol- Galloway et al. (1991), parts of MB have been canic arc activity in the last part of the Mesozoic, permanently subaerial since Late Eocene. How- under mostly submarine conditions. Local hiatus- ever, latest Eocene to Early Miocene deposits are es in the Albian and Turonian suggest bouts of rare and have been found only in the northern and temporary uplift of the terrane in the Late Creta- northwestern parts of the Yucatan Peninsula. In ceous (Case, 1975; Mascle et al., 1985; Per®t and the Oligocene and Miocene, the northeastern sec- Heezen, 1978; Holcombe et al., 1990; Maurrasse, tion of the peninsula was covered by water, as was 1990). Although different marine environments the westernmost extremity of Cuba and Isla de la dominate the Paleocene and Eocene section (Hol- Juventud during the Miocene (Iturralde-Vinent, combe et al., 1990; Robinson, 1994), occurrence 1969). Furthermore, data from offshore seismic of at least one terrestrial vertebrate in Early Eo- lines clearly indicate that the Yucatan Channel is cene rocks of WJ indicates the local occurrence an ancient feature that was in existence long be- of land (Domning et al., 1997). Perhaps, therefore, 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 14. Yucatan Peninsula (Maya Block): top left, Simpli®ed pre-Miocene terrestrial geology; top right, late Tertiary stratigraphic columns (compiled from many sources; see appendix 1); bottom, cross- section along offshore seismic line (courtesy Institute for Geophysics, University of Texas at Austin). Cenozoic land connections between Maya and Western Cuban Blocks are ruled out by the shape of the section underlying the Yucatan channel and the presence of sedimentary rocks of Mesozoic and Cenozoic age. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 75

CB, NR, and WJ were terrestrially connected to Northwestern South America North America for a short time in the Early Eo- The geology of this region has been investiga- cene. (Eastern Jamaica, comprising the Blue ted by many authors (®g. 17); key references in- Mountains Block, has had a different and more clude GonzaÂlez de Juana et al. (1980) Bonini et complex history; see below.) NR was nearly at al. (1984), Duque-Caro (1990), Dengo and Case ambient sealevel during the latest Eocene, and (1990), and Tankard et al. (1995). The regional positive areas were partly or completely uplifted geological picture is too complicated for useful during most of the Oligocene. Miocene to Recent summary here, but some general points need to marine rocks are found throughout, suggesting be brie¯y canvassed for the purposes of this paper. that the NR became once again a submarine fea- Detailed paleogeographic reconstruction of north- ture by the Late Miocene. western South America (NWSA Microcontinent) Two types of late Tertiary sedimentary environ- prior to the latest Eocene is beyond the scope of ments can be de®ned for WJ (Eva and McFarlane, this report (but see Tankard et al., 1995). 1985; Robinson, 1994). One environment is in- During most of the late Tertiary, the NWSA dicated by shallow marine carbonate rocks, re- Microcontinent has moved predominantly east- sembling those of NR. The other environment is ward with respect to the motion of the rest of indisputably deep water, suggesting that Jamaica South America (GonzaÂlez de Juana et al., 1980; was located at the eastern edge of the Rise. The Case et al., 1990; Bartok, 1993; Pindell, 1994; possible existence of a short-lived hiatus at the Balkwill et al., 1995). Also, this region has been base of the shallow-water Oligocene section in strongly affected by vertical movements since the WJ (Eva and McFarlane, 1985; Robinson, 1994) latest Eocene. Marine sediments corresponding to correlates well with Oligocene uplift postulated the P17±18 zones (35±33 Ma) of Berggren et al. for NR. Final uplift of WJ, to create most of the (1995) are not known on the microcontinent prop- island as it is known today, probably occurred er, indicating that most of the area was uplifted during that time (tables 1, 2). However, mid-Oli- during the important tectonic deformation that gocene (P19 zone) and younger marine deposits took place during the Middle Miocene (Eva and have been found in several basins (GonzaÂlez de McFarlane, 1985; Mann et al., 1990, 1995; Rob- Juana et al., 1980; Duque-Caro, 1990; Cooper et inson, 1994). al., 1995), indicating that uplift was followed fair- Southern Central America ly rapidly by subsidence or higher sea stands (or According to data compiled by Escalante both). This must have been accompanied by con- (1990) and Kolarsky et al. (1995a, 1995b), South- siderable modi®cation of the pattern of river ern Central America (SCA) is underlain by Me- drainage (Hoorn et al., 1995). The epicontinental sozoic oceanic crustal rocks and late Campanian± seaway that apparently converted northwestern Eocene oceanic crust and volcanic arc suites. The South America into a large island was in existence Late Eocene to Recent section is summarized in from the latest Eocene (®g. 17; Cooper et al., ®gure 16 (after Escalante, 1990; see also Kolarsky 1995; Kay and Madden, 1997). et al., 1995a, 1995b). The Eocene±Oligocene Aruba/Tobago Belt boundary interval (35±33 Ma) is marked by local The structure and history of the Aruba/Tobago uplift and deposition of angular, poorly sorted Belt (ATB) has to be considered separately (®g. conglomerates, derived from a local source. As 18). The basement of these islands consists of elsewere in the Caribbean region, the Late Oli- Mesozoic oceanic crust and Cretaceous volcanic gocene of SCA is transgressive and features ma- arc units. Equivalent rocks are also found allo- rine rocks indicative of shallow- and deep-water chthonously in the Caribbean Mountains, above environments. An important event occurred late continental margin sediments, all partially meta- in the Middle Miocene, as elsewhere in the Ca- morphosed (GonzaÂlez de Juana et al., 1980; Beets ribbean region, when strong volcanic activity and et al., 1984; Mascle et al., 1985; Jackson and general uplift were initiated (see Dengo and Case, Donovan, 1994). The deformation of ATB took 1990; Duque-Caro, 1990). According to Duque- place when the Caribbean Plate interacted with Caro (1990), late in the Middle Miocene, SCA the South American continental margin (Beets et may have been suf®ciently uplifted to serve as a al., 1984; Erikson and Pindell, 1993; Macellari, connector between Central and South America. 1995). Between 35 and 33 Ma this interaction The existence of late Neogene deformation and produced general uplift of the Belt (see above), as thrust faulting recorded at the hinge zone between well as progressive subsidence of local basins Panama and South America (Mann and Kolarsky, along the continental margin (Macellari, 1995; Er- 1995) tends to support this inference, although it ikson and Pindell 1993; Stockhert et al., 1995). In is far from demonstrated. Trinidad, sedimentation in deep-water conditions 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 15. Chortis/Nicaraguan Rise and Western Jamaican Blocks: top, Location map, drilling sites, and important geological features; middle, late Tertiary stratigraphic data for Nicaraguan Rise and Western Jamaica (dredge samples from wall of Cayman Trench); bottom, simpli®ed cross-section, Ni- caraguan Rise to Colombian Basin (compiled from many sources; see appendix 1). Stratigraphic data suggest that Nicaraguan Rise was at or near sealevel in latest Eocene, and remained positive into Early and Middle Oligocene. Off the Rise, Late Oligocene subsidence is indicated by deep-water sediments in ODP 999 hole. 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 77

Fig. 16. Southern Central America and northern South America (Atrato Basin area): right, Location map, major features, and late Tertiary rocks; left, stratigraphic columns (compiled from Escalante, 1990; Duque-Caro, 1990; Kolarsky et al., 1995a, 1995b). began in the Late Oligocene and persisted until Late Eocene and Late Miocene seen in sections late in the Miocene (Algar and Erikson, 1995). from Aruba and Margarita. This area is provision- Similarly, in the Orinoco River basin, marine sed- ally regarded as the zone in which the inferred iments of Eocene to Recent age occur as outcrop- land connection between the Aves Ridge and con- ping or subsurface deposits (GonzaÂlez de Juana et tinent was formed during the Eocene±Oligocene al., 1980; Cooper et al., 1995). Stratigraphic se- transition. quences in ATB and several basins in the NWSA Microcontinent present a record of the same crit- Greater Antilles ical events in late Tertiary geological history that The Greater Antilles have been the subject of have already been described for other parts of the detailed geological research for most of this cen- Caribbean region, such as the 35±33 Ma hiatus tury. Key modern sources include Khudoley and and subsequent Oligocene transgression (GonzaÂ- Meyerhoff (1971), Dengo and Case (1990), Mann lez de Juana et al., 1980). In some areas, the hi- et al. (1991), Donovan and Jackson (1994), and atus seems to have involved long-lasting subaerial Iturralde-Vinent (1996a). The unusual complexi- exposure, with shorter intervals of marine inun- ties of this region militate against easy paleogeo- dation in the Late Oligocene and the late Early graphical reconstruction, and it is recognized that Miocene. For example, this interpretation seems much more ®eld and laboratory work will have to to apply to the exceptional hiatus between the be undertaken before paleomapping projects will 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Fig. 17. Northwestern South America: Schematic late Tertiary geological map, showing location of Orinoco and eastern Venezuelan basins separating NWSA Microcontinent fromwhile Guyana marine shield. deposits Noteappendix that 1). of rocks earliest conforming in Oligocene age age to Eocene±Oligocene are boundary restricted are to absent throughout, Orinoco and eastern Venezuelan basins (compiled from many sources; see 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 79

phosed, deformed, and jumbled together under the pervasive effects of tectonic forces (Fig. 5; Dengo and Case, 1990; Iturralde-Vinent, 1994a). Widely varying opinions concerning the plate tectonic evolution of these elements can be found in the literature (cf. Pindell, 1994; Iturralde-Vinent, 1994a; Hay and Wold, 1996). Most aspects of the tectonic history of GAF prior to the latest Eocene lie outside the special concerns of this paper (but see ®g. 5 and main text for background). From the literature we have compiled stratigraphic data (®gs. 19, 20) pertinent to interpreting the post-Eocene paleogeography of the Greater Antilles (Cuba: Bronnimann and Rigassi, 1963; Nagy et al., 1983; Albear and Iturralde-Vinent, 1985; Iturralde-Vinent, 1969, 1972, 1996a, MacPhee and Iturralde-Vinent 1994, 1995; His- paniola: Butterlin, 1960; Van den Bold, 1981; Eberle et al., 1982; Maurrasse, 1982, 1990; Saun- ders et al., 1986; GarcõÂa and Harms, 1988; Mann et al., 1991; Heubeck et al., 1991; Toloczyki and RamõÂrez, 1991; Iturralde-Vinent and MacPhee, 1996; Blue Mountains Block: Robinson, 1965, 1994; Eva and McFarlane, 1985; Maurrasse, 1990; Geddes, 1994; Puerto Rico: Meyerhoff, 1933; Pessagno, 1963; Monroe, 1980; Frost et al., 1983; MacPhee and Iturralde-Vinent, 1995; Vir- gin Islands: Larue, 1994; MacLaughlin et al., 1995). Eastward shift of tectonic activity has been a marked trend in the evolution of the Greater An- tilles since the end of the Eocene. Thus, in western and central Cuba, vertical movements, with limited sinistral strike-slip faulting, has been the rule since latest Eocene (Iturralde-Vinent, 1978). By contrast, east of the Guacanayabo±Nipe fault Fig. 18. Late Tertiary stratigraphic columns in eastern Cuba, sinistral faulting and transpres- for selected basins on NWSA Microcontinent sional tectonics have been dominant. This has re- (northwestern South America) and Aruba/Tobago sulted in strong deformation and subdivision of Belt (compiled from many sources; appendix 1). GAF into a series of block-terranes correlated Batuco Fm (Aruba±Bonaire) was originally dated with the opening of trenches, grabens, and pull- as Eocene±Early Oligocene, but age markers are apart basins (Ladd et al., 1981; Larue et al., 1990; ambiguous and suggest Late Oligocene to Early Larue and Ryan, 1990; Jany et al., 1990; MacPhee Miocene age (Lepidocyclina, Heterostegina, Par- and Iturralde-Vinent 1995; Mann et al., 1990; Ca- arotalia, Antiguastrea cellulosa [sensu GonzaÂlez lais et al., 1992). These disrupted block-terranes de Juana et al., 1980]). Lithology as in ®gure 20A. have to be returned to their original, latest Eocene position in order to reconstruct Greater Antillean achieve the needed level of detail and authorita- paleogeography (®gs. 3, 6). tiveness. However, thanks to recent investigations Figure 3 is a palinspastic reconstruction of a some headway has been possible (see Paleoge- critical area in the northern Greater Antilles that ography of the Caribbean Region: Evidence and encompasses present-day eastern Cuba, northern Analysis). Hispaniola, and Puerto Rico. Offsets along major The basement of the Greater Antilles Foldbelt strike-slip faults have been calculated for some (GAF) consists of old continental-margin suites, faults (Iturralde-Vinent, 1981; De Zoeten and Mesozoic oceanic crustal units, Cretaceous/Paleo- Mann, 1991); these ®gures are used where avail- gene volcanic arcs, and latest Cretaceous to Re- able. Our reconstruction is based strictly on con- cent sedimentary basinsÐall partly metamor- cordances between identical or strongly correlat- 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 able geological units (particularly those of Late strongly disrupted the assemblage of block-ter- Eocene age or older), as follows: ranes (cf. ®gs. 1±3). (1) ?Neocomian±Campanian Cretaceous vol- Blue Mountains Block canic-arc complexes: These rocks outcrop from A possible land connection between the southwest-central Cuba across Hispaniola into Puerto ern peninsula of Hispaniola (SH) and the Blue Rico and the Virgin Islands. Mountains Block (BM) of eastern Jamaica is il- (2) Maastrichtian massive conglomerates, dom- lustrated for the Eocene±Oligocene transition (®g. inated by ophiolite pebbles and overlain by Pa- 6). The evidence for this connection is circum- leocene/Early Eocene white tuffaceous rocks: stantial at present, and therefore its inclusion in This suite of rocks only outcrops east of HolguõÂn paleogeographical reconstructions of the Greater in eastern Cuba and in a small area in northwest- Antilles requires some elaboration. ern Hispaniola. This suite is of unique importance It is generally accepted that Jamaica originated for correlating terranes (Iturralde-Vinent, 1994b). as a single crustal unit in the Cretaceous arc (Pin- (3) Ophiolite trend: Outcropping ophiolites in dell, 1994). However, as previously noted, there Cuba follow the same trend as those in Hispan- is evidence that Jamaica is structurally and litho- iola, especially when their paleoposition is recon- logically divisible into two major terranes, constructed palinspastically. sisting of a large western block (Clarendon and (4) Distinctive metamorphic rock units: This Hanover Blocks of Lewis et al., 1990) and a series consists of four distinctive rock suitesÐ smaller Blue Mountains Block. These two ter- marble and schists of the Bahamas margin com- ranes differ radically in crustal composition, de- plex, amphibolites (metaophiolites), serpentinites gree of metamorphism, and stratigraphy (includ- with blocks of eclogite, and metamorphosed Cre- ing the span of temporally correlated units), as is taceous volcanic arc rocks. This combination of evident from several recent papers and mapping metamorphics outcrops in easternmost Cuba and projects (Geddes, 1994; Montadert et al., 1985; northwestern Hispaniola (Puerto Plata±SamanaÂ). Lewis et al., 1990; Robinson, 1994). It is true that, (5) Paleogene volcanic arc rocks: Rocks de- after the Middle Eocene, resemblances between rived from the Paleogene volcanic arc outcrop in coeval formations in different parts of Jamaica eastern Cuba as well as the northern peninsula of greatly increase (e.g., Bonnie Gate Fm; Robinson, Haiti, central Hispaniola, and Puerto Rico. 1965, 1994). However, lithology by itself has lim- (6) Latest Eocene/Oligocene sedimentary ited correlation value in this case, as composition- rocks: Units of this age in the GuantaÂnamo Basin ally similar formations of late Tertiary age out- correlate precisely with those found in the the Ci- crop widely in the Greater Antilles. bao-Altamira Basin of Hispaniola (Calais et al. According to Pindell's (1994) model of the or- 1992; Iturralde-Vinent and MacPhee, 1996). igin of Jamaica, the island's basement as a whole The paleoposition of Puerto Rico/Hispaniola is was originally part of the Cretaceous volcanic not so well constrained as that of Cuba/Hispan- arc located on the leading edge of the Caribbean iola, although it is known that the Cretaceous vol- Plate. As the plate moved east during the late canic arc complex outcropping in eastern Hispan- Cretaceous, the basement rocks of Jamaica re- iola also forms a large portion of the basement of mained attached to northern Central America. It Puerto Rico. The most important correlatable is assumed by necessity that these rocks were units are the Duarte complex of Hispaniola and carried to their present-day position when the the Bermeja complex of Puerto Rico, which lie in Nicaragua Rise (which originated in the Paci®c) the same trend. Also, outcrops of Paleogene rocks was inserted into the Caribbean (Pindell, 1994: on the eastern side of Hispaniola lie in the same ®g. 2.6). If this were so, one would expect to trend as their equivalents in Puerto Rico. ®nd strong similarities between the ophiolitic The close match between the main structural and Cretaceous-arc suites of western Cuba and fabric and compositional elements of eastern Jamaica, since they were located side by side in Cuba, Hispaniola, and Puerto Rico/Virgin Islands the original arc (sensu Pindell, 1994). Yet there as illustrated in ®gure 3 is valid only for the in- is virtually no similarity between relevant geo- terval between the latest Eocene and the mid-Ol- logical sections (cf. Iturralde-Vinent, 1996a, igocene (35±30 Ma). Before the latest Eocene, ov- 1996b, 1996c and Robinson, 1994). By contrast, erthrusting and extensive superposition of terranes there are evident resemblances in the ophiolitic took place in this area (Meyerhoff and Hatten, and metavolcanic sequences of the Eastern Cu- 1968; Pardo, 1975), indicating that a different pal- ban Block and BM, suggesting that these ter- eogeographic organization prevailed at that time ranes belong to the same geological province (see main text). Subsequent to the Late Oligocene, (Iturralde-Vinent, 1995). Most importantly, the movement along several sinistral faults has geological composition of the Mesozoic rocks of 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 81 Fig. 19. Greater Antilles: Simpli®ed geological maps showing major terranes and faults active during late Tertiary. Numbers refer to geographical location of late Tertiary stratigraphic columns depicted in ®gure 20A, B. 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Fig. 20. Greater Antilles: Late Tertiary stratigraphic columns for selected basins: A, Blue Mountains, Cuba (this page); B, Hispaniola, Puerto Rico/Virgin Islands (compiled from many sources; see appendix 1; opposite page).

SH and BM are also remarkably similar. (Cf. de- one important piece of the puzzle is the identity scriptions of southern Hispaniola by Butterlin of the 33±35 Ma hiatus (corresponding to zones [1960] and Maurrasse [1982] with descriptions P17±18) occuring in BM and SH. In SH, the of Blue Mountains by Robinson [1994] and deep-water G. ampliapertura to G. kugleri zones Montadert et al. [1985].) of the Jeremie Fm unconformably overlie older These observations can be made concordant if rocks (®g. 20B). Rocks deposited in the time-slice it is accepted that BM originated as part of the corresponding to zones P17±18 have not been re- northern Greater Antilles, while WJ evolved from ported; this may indicate that SH was uplifted at the leading edge of the Nicaragua Rise (sensu Pin- that time, as were many other terranes in the Ca- dell, 1994). In this interpretation, these terranes ribbean (including blocks having similar oceanic maintained a separate existence until the Middle crustal structure, e.g., Beata Ridge). Miocene, when they were conjoined during tec- Likewise, well-dated sediments of 35±33 Ma tonic deformations recorded in the island (Mon- age have not been recognized in the Blue Moun- tadert et al. 1985). We acknowledge that this dual- tains either. Eva and McFarlane (1985) identi®ed origin hypothesis represents a break with the or- the Early Oligocene in Jamaica on the basis of an thodox view, and that further substantiation is re- assemblage including Dictyoconus cookei, Ar- quired (see also Stephan et al. 1990: pls. 8±10). chaias angulatus, and several small species of Pe- With regard to the possible presence of land on neroplis, but these taxa are ambiguous indicators the Blue Mountains unit during the early Tertiary, of this age, because A. angulatus and Peneroplis 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 83

Fig. 20. Continued. are extant taxa, while D. cookei is not restricted contexts like the Bonny Gate Fm in which allo- to the Oligocene (see Bronnimann and Rigassi, chthonous biodetritus is frequently encountered 1963; Albear and Iturralde-Vinent, 1985). Simi- (Eva and McFarlane, 1985; Robinson, 1965, larly, Robinson (1965, 1994) identi®ed the hemi- 1994). pelagic Bonny Gate Fm from localities surround- The possibility that uplift occurred within BM ing the central Blue Mountains as being Eocene and SH 35±33 Ma is underscored by the ubiqui- to Late Oligocene in age. However, he does not tous presence of a hiatus of that precise age in list fossils consistent with a P17±18 zone alloca- existing positive structures almost everywhere in tion (sensu Berggren et al., 1995). In any case, the Caribbean, including the submarine Beata the use of benthic forams as index fossils to date Ridge. Absence of this hiatus in eastern Jamaica narrow time intervals is problematic, especially in would be completely anomalous. 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

The nonpalinspastic maps of Eva and Mc- dredge hauls from the basin's margins consist of Farlane (1985) indicate the occurrence of land in mostly Eocene and younger sedimentary and vol- the Blue Mountains during the Paleocene and Eo- caniclastic rocks (®g. 21). cene, and, much later, after the Middle Miocene. According to Pindell (1994), GB opened be- Although this scenario, involving repeated emer- tween the Paleocene and Late Eocene, but we gence and submergence of the Blue Mountains, postulate a somewhat younger date (Late Eocene cannot be directly challenged on the basis of the or younger) for the following reasons. If GB is few facts available, we argue that it is more par- interpreted as a back-arc basin, the disjunction of simonious to hypothesize that the Blue Mountains the AR±LA arc into two independent geological have been permanently subaerial since the latest units (Aves Ridge remnant arc and Lesser Antilles Eocene. This view is consistent with (1) the ab- active arc) would have probably been caused by sence of Eocene and younger rocks from the core a local change in the subduction regime (e.g., al- of BM; (2) the lithological character of the Bonny teration of angle of dip of lower slab, or migration Gate Fm, in which the presence of coarse clastic of position of subduction zone). We hypothesize debris (olistostromes, sandstones, silts) in the bas- that this event was correlated with Late Eocene al Lloyd member (Maurasse, 1990) indicates the cessation of volcanic activity in AR (and a con- existence of land during the late Paleogene; and committantly great increase in activity in LA) and (3) the occurrence of shallow-water biodetritus increased thickness of Oligocene and younger throughout the Bonny Gate section (Eva and sediments in GB (see seismic sections in Nemec McFarlane, 1985), suggesting shelf or coastal en- [1980] and Pinet et al. [1985]). vironments prevailing at the time of deposition. Figure 21 depicts islands and other features of Better evidence of the emergence of the Blue the eastern Caribbean as they appear today, to- Mountains and the nature of their connection with gether with simpli®ed late Tertiary stratigraphic Hispaniola is sorely needed. columns for AR, GB, and LA. The thickness of Tertiary sediments (Pinet et al., 1985) indicates Aves Ridge, Lesser Antilles, and Grenada Basin that AR and LA have been positive for most of In the last three decades, a large amount of geo- the Cenozoic. Positive topography is also indicat- logical and geophysical data has been collected ed by the occurrence of shallow-water limestones concerning the Aves Ridge (AR), Lesser Antilles of Eocene to Lower Miocene age dredged from (LA), and Grenada Basin (GB). For an overview ridge walls of these features. In addition, Early and additional references, the reader is referred to Oligocene slope deposits have been recovered papers by Fox et al. (1971), Bouysse et al. (1985), from cores from Saba Well (Nemec, 1980); their Pinet et al. (1985), Holcombe et al. (1990), and compositional character suggests that they were Maury et al. (1990). Here we concentrate on derived from some nearby area (presumably AR) physical paleogeography. that had been block-faulted and signi®cantly up- Cretaceous and Paleogene volcanic and pluton- lifted (Pinet et al. 1985). ic rocks of island arc af®nities occur in AR (Bun- Calculations by Holcombe and Edgar (1990) ce et al., 1970; Fox et al., 1971; Nagle, 1972; establish that, given certain assumptions, before Bouysse et al., 1985; Westercamp et al., 1985; the Miocene most of the topographic highs on AR Holcombe et al., 1990), as do Mesozoic and Eo- may have been subaerial, or nearly so, and would cene volcanic rocks in LA (®g. 15). This basic have formed a string of emergent lands along a compositional similarity suggests that, from Cre- north±south axis. This inference is corroborated taceous through Eocene time, AR and LA were a by the frequent occurrence of conglomerates in single entity: the AR±LA Volcanic Arc (Pinet et samples dredged from the Ridge (Nagle 1972; al., 1985; Bouysse et al., 1985). This arc was pre- Bouysse et al., 1985). These poorly sorted con- sumably linked geologically to the Aruba/Tobago glomerates contain rounded pebbles and cobbles Belt in the south and the eastern Greater Antilles of andesite, up to 10 cm long and showing altered in the north, because all of these landmasses pos- (weathered?) cortices in a calcareous±tuffaceous sess a similar Cretaceous volcanic arc-ophiolite matrix. The matrix has yielded large benthic fo- basement. rams and algae, which Bock (1972) identi®ed as If AR and LA once comprised a single arc, it Oligocene, Early Miocene, or older (Nagle, 1972). can be concluded that, at some time in the past, The existence of these conglomerates strongly the GB that now separates these two entities did implies the existence of subaerial conditions just not exist. However, the age of this basin has not before and at the time that they were being been well constrained. Inconclusive seismic evi- formed. Observations on the effects of weathering dence suggests that GB is ®lled by sedimentary on granular igneous and sedimentary rocks in the rocks of Paleocene(?) to Recent age (Pinet et al., Greater Antilles (so-called ``big boulder bed'' of 1985; Bouysse et al., 1985; Bird, 1991), while Bronnimann and Rigassi [1963]; M. A. Iturralde- 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 85

Fig. 21. Aves Ridge, Grenada Basin, and Lesser Antilles: left, Present-day topography and thickness of Tertiary sediments; right, stratigraphic columns (compiled from many sources; see appendix 1). Aves Ridge and Grenada Basin columns based on combined seismic interpretation and dredge-haul samples. Relative thinness of Paleocene±Oligocene sediments contrasts markedly with thick Miocene±Recent deposits (Holcombe et al., 1990), and is consistent with interpetation that Aves Ridge has been positive for much of the Cenozoic. Existence of terrestrially derived conglomerates establishes that portions of the ridge were subaerial in Oligocene. 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238

Vinent [unpubl. data]) indicate that, under tropical conditions, surface weathering of the cortices of granular rocks causes material to be concentrical- ly lost, eventually resulting in the formation of rounded cobbles. Most rivers in the Greater An- tilles are far too short to produce roundedness by abrasion alone, which suggests that conglomerates containing large, rounded cobbles may be diagnostic of highland conditions prevailing at the time of deposition. Incorporation of rounded cobbles into a calcareous matrix is obviously second- ary, the result of transport to marine environments. In our view, AR conglomerates are readily correlatable with the Eocene±Oligocene conglomerate event recorded elsewere in the Greater An- tilles and South America (MacPhee and Iturralde- Vinent, 1995). These facts suggest that AR was a topographic high (Donnelly, 1989b) from the Eocene to the Lower Miocene, and was actually emergent for some inde®nite (but probably short) period within the latest Eocene/Early Oligocene, after the termination of arc magmatism and related uplift in the Greater Antilles and Aves Ridge. AR rapidly subsided thereafter, and was already deeply submerged by the Middle Miocene, as indicated by seismic pro®les and the occurrence of Middle Miocene and younger deep-water sediments in wells (DSDP 30, hole 148; Edgar et al., 1973) and dredge hauls (Nagle 1972; Bock, 1972; Bouysse et al., 1985). Structurally, the southern portion of AR±LA is part of the deformed and partly obducted volcanic arc complex that also forms the basement of the Caribbean Mountains and islands along the Aru- ba/Tobago Belt (GonzaÂlez de Juana et al., 1980; Jackson and Robinson, 1994). That the NWSA Microcontinent (Gonzlez de Juana et al., 1980; Balkwill et al., 1995; Parnaud et al., 1995) shares a signi®cant portion of its Cenozoic geological history with the southern AR±LA is also indicated by the fact that they were extensively and coter- minously uplifted around the Eocene±Oligocene boundary (35±33 Ma). The possibility that they were physically connected by emergent land during this time is strongly suggested by the absence of marine sediments of this age in northwestern South America, as well as in the Aruba/Tobago Belt and AR±LA arc. Fig. 22. Cayman Rise, Ridge, and Islands: The stratigraphic record of LA (Maury et al., Late Tertiary stratigraphic columns (compiled 1990) shows a history of activity in marine as from Per®t and Heezen, 1978; Jones, 1994; Si- well as subaerial contexts since the Eocene. The gurdsson et al., 1997). ODP hole 998 drilled on presence of late Tertiary marine sedimentary Cayman Rise shows very low sedimentation rates rocks intercalated within volcanic sequences is occurred during Eocene±Oligocene transition and evidence that the islands have not been perma- late Middle/early Late Miocene. Low rates cor- nently uplifted since the Eocene, but have had a relate well with hiatuses and uplift events in Cay- complex history involving emergence, subsi- man Islands and elsewhere in Caribbean region. dence, and migration of topographic highs. In 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 87 their present form the majority of them are cer- to latest Eocene/Early Oligocene uplift is impor- tainly young (Pliocene to Recent). tant, because it establishes that areas composed of Beata Ridge thick oceanic crust were affected in the same The geological history of Beata Ridge (BR) has manner as geological units composed of continen- been recently reviewed by Fox et al. (1970), Case tal or island-arc crust, underlining the all-embrac- (1975), Mascle et al. (1985), Holcombe et al. ing character of this event (MacPhee and Iturral- (1990), and Maurrasse (1990). According to Case de-Vinent, 1995). (1975) and Holcombe et al. (1990), BR was an Cayman Islands and Cayman Ridge undifferentiated part of the Caribbean oceanic Because of the small size of outcrops on the crust until the onset of orogenic movements in the Cayman Islands and dif®culties in recovering suf- Late Cretaceous. Organizationally, BR consists of ®cient dredge samples from the walls of the Cay- a set of tectonic blocks (®g. 4, cross section, el- man Trench (CT), information is limited on these evations considerably exaggerated) that have been units (Per®t and Heezen, 1978). The most recent positive since the end of the Maastrichtian(?). BR review of the geology of the Cayman Islands is was markedly affected by uplift (up to 1000 m) by Jones (1994); additional topics of interest are in both Middle Eocene/Oligocene and Middle covered by Case (1975), Holcombe et al. (1990), Miocene time, but there is no direct evidence that and Rosencrantz (1990, 1995). it was actually emergent at any stage. In particu- Stratigraphic columns (®g. 22) indicate that be- lar, the lack of conglomerates in wells and dredge tween 35 and 33 Ma the Cayman Ridge was cov- hauls speaks against the existence of high, dry ered by shallow water. Evidence of deeper water areas on BR (cf. discussion of Aves Ridge). Hi- environments occurs only in Early Miocene (and atuses do occur in the BR section, but none of later) rocks dredged from the walls of CT, indi- them shows an unequivocal signature of emer- cating that this is when the trench system began gence. For example, the Middle Miocene hiatus to open (Per®t and Heezen, 1978). On the islands observed in cores may be due to an interruption themselves, shallow-water limestones of Oligo- in deposition or to the effects of submarine ero- cene±Miocene and Middle Miocene age have sion on BR (Mullins et al., 1987; Iturralde-Vinent been documented, as has a hiatus within the Early et al., 1996a). Beata Island, the only emergent part Miocene. These data underline the very recent of the ridge, probably became subaerial in the character of the islands on the Cayman Ridge. Quaternary. Therefore, Beata Ridge is of no sig- However, they do not necessarily preclude the ni®cance for interpreting the history of land con- possibility of a recent land connection between nections in the Caribbean region, as it was evi- the Cayman Islands and eastern Cuba, as both are dently never a structural or paleogeographical link located along the same structural trend, i.e., the between a mainland and the Greater Antilles Cayman Ridge (but see point [4] in discussion (Heubeck and Mann, 1991) (®gs. 6±8; tables 1±4). section under GAARlandia Landspan and Island± Nevertheless, evidence that BR was subjected Island Vicariance).

APPENDIX 2: A PLATE TECTONIC MODEL OF THE CARIBBEAN FROM LATEST EOCENE TO MIDDLE MIOCENE Models of the plate tectonic evolution of the 1. Crustal plates are deformable Caribbean region tend to agree on the major is- Interactions between plates commonly result in sues (Malfait and Dinkelmann, 1972; Ross and profound deformations of crustal blocks and ter- Scotese, 1988; Pindell and Barrett, 1990; Pindell, ranes, not only along plate margins but also with- 1994), but many details remain uncertain. Most in intraplate domains. Typical deformations in- discrepancies among models concern tectonic de- clude crustal shortening and superimposition of velopments prior to the latest Eocene, although units as a consequence of folding and thrust fault- controversy attends several aspects of plate move- ing (see ®gs. 2 and 5) as well as the partial or complete destruction of microplates, blocks, and ment in the crucial interval between the end of terranes at subduction zones. These processes op- the Eocene and the Middle Miocene. In order to erate at all scales, resulting in modi®cation of the present new data and interpretations bearing on size and con®guration of individual blocks as well late Tertiary tectonics, we constructed a tectonic as entire plates. From this it follows that tectonic model for the interval 35±14 Ma (see ®gs. 23±26, models that purport to be realistic must take some table 6). Basic assumptions underlying our recon- account of these processes; if not, results will be struction are presented below and in the caption interpretatively problematic. The recent tectonic of ®gure 23. model for the Caribbean published by Hay and 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 89 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Â dextral fault. Ârida±Bocono 35 Ma) paleopositions using the following four constraints: (1) Cayman Trench system movements ϳ Fig. 23. Caribbean tectonic framework: Current positions of geological units active from latest Eocene to Late Miocene. Units have been sequentially absorbed byJamaica acts Guacanayabo±Nipe as and twobut Oriente distinct structurally terranes, related sinistral Western toaccounting Jamaica faults for Eastern (originally sinistral and Cuban associated movements andplate local with associated Southern for with Nicaraguan underthrusting Hispaniolan alkaline Rise) most vulcanism Blocks); within and of along (3) Blue late Hispaniola Hess southern Mountains escarpment; Tertiary, (Mann Central (part and due America (4) of et to NWSA originates Caribbean activity al., Microplate crust, southwest along acts of 1995); as Me Chortis part (2) Block, of Caribbean palinspastically restored to Late Eocene ( 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 91

Wold (1996) may be cited as an example of the dencia and Muertos Troughs. In the Greater An- latter. In their model, tectonic blocks and terranes tilles, the Middle Eocene orogeny was associated move, but they do not deform, even after the with cessation of magmatic activity and uplift of elapse of many millions of years (Hay and Wold, volcanic structures formed in the Paleocene 1996: ®gs. 2±7). The resulting lack of realism is though early Middle Eocene. As magmatism end- evident in the evolution of Hispaniola (our Central ed in the Greater Antilles, it also terminated on and Northern Hispaniolan Blocks, ®g. 23). His- the Cayman and Aves Ridges, thereafter shifting paniola is depicted by these authors as suffering permanently to the Lesser Antilles arc (cf. ®gs. 5 no deformations or alterations from the late Me- and 6). This orogeny additionally led to deacti- sozoic onward; to ®t within the space available vation of the Yucatan Basin spreading center and per time slice, it has to be sequentially moved the shifting of ocean crust production to the Cay- from a position within the Paci®c realm (150±130 man center (Rosencrantz, 1990). Due to subse- Ma) to the margin of the Chortis Block (100 Ma), quent movements, crustal segments that were thence to the margin of the Maya Block (67.5 originally part of the foldbelt created by this event Ma), thence south of western Cuba (58.5 Ma), are now distributed in a broad circum-Caribbean thence south of eastern Cuba (49.5 Ma), ®nally swath, involving southern Mexico, Greater Antil- ending up east of Cuba at 24.7 Ma (Hay and les, Aves Ridge, Aruba/Tobago Belt, Caribbean Wold, 1996: ®gs. 2±7). Further contortions are in- Mountains, Columbian/Venezuelan Andes, and troduced by unconstrained rotation of the terrane Central America (®g. 23). along its major axis from N±S at 130 Ma to ENE± WSW at 49.5 Ma. These proposed lateral dis- 3. Island arc magmatic activity on the Caribbean placements and rotations ®nd no support in the plate occurred in discrete stages and was not a geological composition or structure of central and continuous process northern Hispaniola (in addition to main text, see The most important magmatic events in the his- ®gs. 3 and 5 and appendix 1; Draper, 1989; Mann tory of the Caribbean area were (1) continental et al., 1991). margin magmatism (170±110 Ma) in association with the break up of Pangaea (Maze, 1984; Bar- 2. Global phases of orogeny have had important tok, 1993; Iturralde-Vinent, 1994a), (2) oceanic implications for Caribbean plate evolution magmatism (170±110 Ma) related to the forma- Many different bouts of orogeny, from regional tion of the proto-Caribbean oceanic crust between to global, have affected the Caribbean region and North and South America (Pindell, 1994), (3) surrounding areas during the past 170 Ma. The eruption of alkaline volcanoes related to intraplate most signi®cant of these took place in the late tectonic activity along major faults (Dengo and Aptian (120±110 Ma), late Campanian±early Case, 1990), and (4) the evolution of the volcanic Maastrichtian (75±70 Ma), and Middle Eocene arcs. (45±40 Ma) and had worldwide effects. In the Ca- With respect to the last of these phenomena, arc ribbean, these effects included (1) modi®cation of magmatic activity on the Caribbean plate, several the rates of plate movement, (2) rotation of major discrete stages are evident: (1) ?Neocomian to stress axes, (3) modi®cation of the orientation and Aptian (120±110 Ma), (2) Albian to Coniacian± extension of volcanic arcs, (4) alteration of arc Santonian (100±87 Ma), (3) Santonian to ?early magmatic geochemistry, and (5) consolidation of Maastrichtian (87±70 Ma), (4) mid-Paleocene to foldbelts (®g. 5; Schwan, 1980; Mattson, 1984; early Middle Eocene (60±55 Ma), and (5) latest Pszczolkowski and Flores, 1986; Iturralde-Vinent, Eocene to Recent (37±0 Ma). Each of these re- 1994c; Iturralde-Vinent et al., 1996b; Bralower gionally discrete magmatic stages exhibited a spe- and Iturralde-Vinent, 1997). The orogeny which ci®c geological signature marked by structural un- occurred in the Middle Eocene is especially note- conformities due to tectonic deformation and up- worthy. Correlated with this orogeny were (1) re- lift, hiatus formation related to erosion and non- duction in the relative motion of the North and deposition, and deposition of coarse clastic and South American plates (Pindell, 1994: ®g. 2.3), carbonate sedimentary rocks (see Paleogeography (2) reorientation of the Caribbean Plate stress ®eld of the Caribbean Region: Evidence and Analysis, from mainly NE±SW to dominantly E±W, and (3) Early Middle Jurassic to Late Eocene Paleoge- formation of numerous microplates, blocks, and ography). This conception of periodic arc mag- terranes along plate margins (Case et al., 1984). matism, punctuated by nonvolcanic intervals, Many of the critical geological units discussed in contradicts the widely held view originally for- this paper (e.g., various Cuban blocks) were mulated by Malfait and Dinkelmann (1972), who formed after this event, as were many major fault- envisaged a single ``Great Arc'' continuously de- bounded structures such as the Windward and veloping on the leading edge of the Caribbean Mona grabens, Cayman Trench, and the Provi- plate from the Jurassic onward (see also Burke et 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Â transform fault trend. Ârida-Bocono Fig. 24. Plate tectonic reconstruction, Caribbean Region, Eocene±Oligocene transition (35±33 Ma). Figures 24±26 were generated using the program PLATES (Institute forover Geophysics, a University set of intervalused Texas (here, for at 35±14 orientation Austin), Ma)accuracy. in but Contact relative which between do to North consequences not a AmericanHispaniolan of ®xed plate represent transform speci®c (NOAM) master paleogeographical faults, displacements and reference and Caribbean reality. can unit plate between be Some (here, (CARIB) Caribbean investigated occurs structural North along plate American elements Motagua/Swan/Nipe±Guacanayabo/northern and plate; subdivided South see into American table smaller plate 6). units Present-day (SOAM) coastlines along to are Me preserve tectonic 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 93 Â transform faults, but Oca±Pilar fault trend became Ârida-Bocono Fig. 25. Plate tectonic reconstruction, Caribbean region, Late Oligocene (27±25 Ma). For abbreviations and other details, see ®gure 24. In this stage, Cayman Ridgetransform and fault. Eastern CARIB±SOAM Cuban contact Block continued became attached to to lie NOAM, along and the contact with trend CARIB of in Me eastern Cuba jumped to the Oriente active in associationAmerica. with Hess alkaline Escarpment volcanic faultmargins activity. trend in Active was eastern plate active Caribbean. convergence in continued association in with Lesser alkaline volcanoes. Antilles Strong and deformation Paci®c and Ocean accretion margin took of place Central along plate 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 238 Fig. 26. Plate tectonic reconstruction, Caribbean Region, Middle Miocene (16±14 Ma). For abbreviations, see ®gure 24. In this reconstruction, Northern Hispaniola Block iscontact now is attached located toand as NOAM, before, Paci®c and but Ocean contact Oca±Pilarwere margin with fault active, CARIB of trend producing has Central wastime, extension jumped America, highly strong along to active deformation with trend Septentrional in and fault of extension in this sediment Chortis/Nicaraguan of Hispaniola. period. accretion Rise volcanic CARIB±SOAM Active along (for activity plate underthrusting sites into convergence fronts of continued Panama continued Oligo-Miocene along at area. volcanic Lesser plate Pedro activity, Antilles border see and in ®g. Hess eastern 15). Escarpment Caribbean. At faults the trends same 1999 ITURRALDE-VINENT AND MACPHEE: CARIBBEAN PALEOGEOGRAPHY 95 al., 1984; Pindell, 1994). Recent supporters of the boundaries along major faults (in the north, from continuous-development model include Mann et Nipe±Guacanayabo to Oriente to Septentrional; in al. (1995: ®g. 36A±C), who argued that the con- the south, from the MeÂrida/BoconoÂ suture toward vergent front of the Caribbean Plate was active the Oca±Pilar fault; ®g. 23±26 and appendix 1). with subduction deepening to the south from Migration of volcanic activity and the other Maastrichtian until Middle Eocene times. How- phenomena noted above have been interpreted as ever, no magmatic activity subsequent to the late a consequence of the oblique collision and result- Campanian is recorded in western and central ing ``escape to the east'' (or ``escape to the Cuba (Iturralde-Vinent, 1994a), Aruba/Tobago ocean'') of the Caribbean plate as its leading edge Belt (Hunter, 1978; Jackson and Robinson, 1994), progressively collided with the Bahamas platform or the Caribbean Mountains (Bonini et al., 1984; (e.g., Mann et al., 1995). However, Bralower and Macellari, 1995). Additionally, volcanic arc rocks Iturralde-Vinent (1997) have rejected this inter- of mid-Paleocene to early Middle Eocene age in pretation as it concerns Cuba, on the ground that eastern Cuba are structurally unconformable to the Cuba±Bahamas collision is conventionally pre-Maastrichtian Cretaceous arc rocks (®g. 5). dated to Early Eocene but arc extinction actually The subduction zone of this arc was located to the occurred much earlier (15 Ma previously, in the south and deepened to the north, rather than vice Late Cretaceous; see also Iturralde-Vinent, 1994a, versa (®g. 5; Iturralde-Vinent, 1994a, 1996d; Si- 1994c). Earlier extinction of the Cretaceous arc is gurdsson et al., 1997). also seen in the Caribbean Mountains (Bonini et 4. Stress ®elds have rotated eastward within the al., 1984; Macellari, 1995; Beccaluva et al., 1996) Caribbean region and the Aruba/Tobago belt (Jackson and Robin- Stress-®eld rotation during the formation and son, 1994), indicating that the Cuban case is not evolution of the Caribbean was ®rst proposed by anomalous (®g. 5). Iturralde-Vinent (1975). This phenomenon is ev- The mechanism of stress-®eld rotation is not ident in the present-day N±S orientation of the understood. Speculatively, it might be assumed convergence front (island-arc subduction zone) of that the phenomenon is driven by the same pro- the Lesser Antillean and Central American arcs, cess that also affects the movement of tectonic and in the extension of arc magmatism southward plates. From this perspective, tectonic events re- in Central America during the last 25 Ma. It is corded in the lithosphere may be thought of as a also evident in the location of post-Eocene trans- consequence of interactions between individual form faults and associated deformations along the plates as they accommodate reorientations (rota- northern and southern margins of the Caribbean tions) of the deep-seated source (mantle-core) of Plate, and in the sequential shifting of plate the stress ®eld.