Sheed Mpeg? eR CEE RES | baletst ta tds mea oo ies Teas) Bitte SOPRA CRESS CEBEEL ESTE MERE ESS: HARVARD UNIVERSITY we Library of the Museum of Comparative Zoology Gilbert Dennison Harris (1864 - 1952) Founder of the Bulletins of American Paleontology (1895) ISBN 0-87710-4: tases 27 1992 aes ae OLUME 101, NUMBER 338 DECEMBER 31, 1991 Neogene Paleontology in the northern Dominican Republic 11. The Family Faviidae (Anthozoa: Scleractinia) Part I. The Genera Montastraea and Solenastrea by Ann F. Budd Paleontological Research Institution 1259 Trumansburg Road Ithaca, New York, 14850 U.S.A. , - 7 Se PALEONTOLOGICAL RESEARCH INSTITUTION — Officers IBREESTDENT Rs terre crea vettys ae Leas ata ascent EES HARRY A. LEFFINGWELL WAGE-PRESIDENE yar ok er eh Pr a Ae J. THomMAS DuTRO, JR. SEGREDARIWA ene ee senate ted ROE ars eect eons ae pee tees HENRY W. THEISEN SDREASWIRERS ene eee aetna Or re crete Re eke tena JAMES C. SHOWACRE ASSISTANT TREASURER ..... Sir sient ogee a Oe RRA Nee RoGeER J. HOWLEY TTR COR ys eos Fe Eee ee eS TERIA PETER R. HOOVER JUEG AT: (GOUNSED c Gatien Rats ae ose eee HENRY W. THEISEN Trustees BRUCE M. BELL (to 6/30/93) EDWARD B. Picou, JR. (to 6/30/92) CARLTON E. BRETT (to 6/30/92) CONSTANCE A. SANCETTA (to 6/30/94) WILLIAM L. CREPET (to 6/30/94) JAMES C. SHOWACRE (to 6/30/93) J. THOMAS DuTRO, JR. (to 6/30/93) JAMES E. SORAUF (to 6/30/94) HARRY A. LEFFINGWELL (to 6/30/93) JOHN STEINMETZ (to 6/30/94) Rosert M. LINSLEY (to 6/30/92) HENRY W. THEISEN (to 6/30/92) SAMUEL T. PEES (to 6/30/92) RAYMOND VAN HouTtTE (to 6/30/94) WILLIAM P. S. VENTRESS (to 6/30/93) BULLETINS OF AMERICAN PALEONTOLOGY and PALAEONTOGRAPHICA AMERICANA IBETERD Res bLOOVE Rie tae esses eee ee eee ei eee Cea eter EDITOR Reviewers for this issue S. D. CAIRNS P. R. HOOVER T. A. STEMANN A list of titles in both series, and available numbers and volumes may be had on request. Volumes 1-23 of Bulletins of American Paleontology have been reprinted by Kraus Reprint Corporation, Route 100, Millwood, New York 10546 USA. Volume | of Palaeontographica Americana has been reprinted by Johnson Reprint Corporation, 111 Fifth Ave., New York, NY 10003 USA. Subscriptions to Bulletins of American Paleontology may be started at any time, by volume or year. Current price is US $45.00 per volume. Numbers of Palaeontographica Americana are priced individually, and are invoiced separately on request. for additional information, write or call: Paleontological Research Institution 1259 Trumansburg Road Ithaca, NY 14850 USA (607) 273-6623 MCZ . LIBRARY? JAN 2.7 1992 ARVARD The Paleontological Research Institution acknowledges with special thanks the contributions of the following individuals and institutions ($1000 or more at the discretion of the contributor) JAMES E. ALLEN (1967) AMERICAN O1L COMPANY (1976) ATLANTIC RICHFIELD COMPANY (1978) CHRISTINA L. BALK (1970, 1982, 1983) Hans M. Bo i (1984) RuTH G. BROWNE (1986) Mr. & Mrs. KENNETH E. CASTER (1967) CHEVRON OIL COMPANY (1978, 1982) EXXON COMPANY (1977 TO DATE) Lois S. FOGELSANGER (1966) GuLF O1L CORPORATION (1978) MERRILL W. Haas (1975) RoBERT C. HOERLE (1974-1977) RICHARD I. JOHNSON (1967, 1986) J. M. MCDONALD FOUNDATION (1972, 1978) MobsiL Ort CORPORATION (1977 TO DATE) SAMUEL T. PEEs (1981) RICHARD E. Petit (1983) Rosert A. PoHowsky (1982) Texaco, INc. (1987 TO DATE) UNION OIL OF CALIFORNIA (1982 TO DATE) UNITED STATES STEEL FOUNDATION (1976) CHARLES G. VENTRESS (1983 TO DATE) CHRISTINE C. WAKELEY (1976-1984) R. TUCKER ABBOTT JAMES E. ALLEN ELIZABETH A. BALCELLS-BALDWIN CHRISTINA L. BALK Bruce M. BELL Rosert A. BLACK RICHARD S. BOARDMAN Hans BOLLI Davip JOHN BOTTJER RutH G. BROWNE J. DAvip BUKRY SysBiL B. BURGER Lyte D. CAMPBELL JOHN L. CARTER ANNELIESE S. CASTER KENNETH E. CASTER JOHN E. DUPONT J. THOMAS DuTRO, JR. J. MARK ERICKSON RICHARD J. ERICKSON Lots S. FOGELSANGER A. EUGENE FRITSCHE CHRISTOPHER L. GARVIE ERNEST H. GILMOUR MERRILL W. HAAS ANITA G. HARRIS STEVEN M. HERRICK RoBERT C. HOERLE F. D. HOLLAND, JR. FREDERICK H. C. HOTCHKISS DAvip JABLONSKI RICHARD I. JOHNSON Davip B. JONES PETER JUNG TOMOKI KASE PATRICIA H. KELLEY Davip GARRETT KERR Ceci, H. KINDLE WILLIAM F. Kose, II Jiri Kriz LIFE MEMBERS RALPH L. LANGENHEIM, JR. Harry A. LEFFINGWELL EGBerT G. LEIGH, JR. GERARD A. LENHARD Loute N. MARINCOVICH, JR. DONALD R. MOORE SHuyt Niko HrrosH1 NODA SAKAE O’HARA WILLIAM A. OLIVER, JR. SAMUEL T. PEES RICHARD E. PETIT EDWARD B. Picou, JR. Ropert A. POHOWSKY JOHN PojeTA, JR. JOHN K. POPE ANTHONY RESO ARTHUR W. ROCKER ARNOLD Ross WALTER E. SAGE, III JOHN B. SAUNDERS JUDITH SCHIEBOUT MIRIAM W. SCHRINER EDWARD S. SLAGLE RoserT E. SLOAN Davip H. STANSBERY JORGE P. VALDES CHARLES G. VENTRESS WILLIAM P. S. VENTRESS EmiLy H. VOKES HAROLD E. VOKES CHRISTINE C. WAKELEY THOMAS R. WALLER ALBERT D. WARREN, JR. Gary D. WEBSTER RALPH H. WILLOUGHBY ARMOUR C. WINSLOW THOMAS E. YANCEY Victor A. ZULLO TOLUME 101, NUMBER 338 DECEMBER 31, 1991 Neogene Paleontology in the northern Dominican Republic 11. The Family Faviidae (Anthozoa: Scleractinia) Part I. The Genera Montastraea and Solenastrea by Ann F. Budd Paleontological Research Institution 1259 Trumansburg Road Ithaca, New York, 14850 U.S.A. Library of Congress Card Number: 85-63715 Printed in the United States of America Allen Press, Inc. Lawrence, KS 66044 U.S.A. CONTENTS APNE RRC] 6 icy cuemrencheter thai cS ete CERT SEE SAR ney ey eS PCR PS So kar eer Preeti ry eI ee ie ie Pa eR re ey thrahe Pere en Ad sk ae OA & IRGC os ake atts area ee atc t ae SRC CER RRC Renee ce ce Rae Irae eo ne ete as eee cee ene et eM i OPO Rh Ra er Be Sonat Bf I RORW RO TG NYC WTC) WL cya ee nee Ory Bar a5 ee es a A ieee de PIR ORDA ea ders eR cert be, ee ER EE SS al ING SNe hi tere harvey ace ie Eee Ones eet near iter see itor ints avai Re poGidctouc ope aoe das cuba O86 MMS tetutionalkA bbe Via tLOMS hay csosclereececte ye tetas eee Sona aot OE er RMS POEs Sn ec en NCE rar) AUSREPSIG Neuve ToT eoeT ey cis SU Pe neces ere Biostragraphyandt PaleGecOlopy:: 2:5 aes, Aerstessicce ses cis/oe aoe. Pie i au oe eerie ea ES ee Te eer ee EO ee a re Taxonomic Method PLO DISMUM rye eee Rs Se ons e eieedie Siac is ja eta erceis MEA ars ane eter bed canta esTashe ions delatoee AIS Te Eh ent ne AE aC ee OE een eee MES CE Sc Se a ena ecg SOE eR ee cine Renn Raeaeicicas Brace te tice Saar eRe Ss eferine Se yal EW A at A ares ot arses Ae EOE Bane ce (CHET CIG Tc BPs nS Sint era eicecic RA Ie te in he Meer ey ie Re ae Se Eo RRR EAE are. Ceara icd IaOwe eee SCAN eh OR ia eS oe ene StatisticalWBrocedures en sevice eects a Tee ce TALE POU Ce Ae Tey MORI EPS VEO POTENT rea ee OR OE one ResultsiandsinterpretattonSteac cnae ena tstch aration eh ates eae ey aes tree er IC eee ee on eee Womparisonsiwithiothen Cari bbeanwatinasyeeceisseee cera ne eee een ee ee ee eee eee eee Systematic Paleontology ) RaNioys EYe( 010) opciones Re eee aan Caner tile Met RMI. onic aD Re Pea OO etC Ob amilyprhaviidae Gregory U9 OO) eres aicii vse se hare ners es dnePuveyn Svavevicas cov seve eres se sytes races eeerevere, sensatniay a ety aces svaucyaveeevs aA seeiausen evel erencee ste enenerere tere Genus MontastracaBlainvilles i830) “sect detscr.s spice dere eect nadie niehene occ eeee ee eieEe eee e ee eee eee IMONLASITACQDFEVISDUNCATs US OAT he sree crteiors Ae ede iene he ee SE eee RE eer ne eee Montastraearcanalis\(Wiaughantell9ilG)i ee ees tenner terest cine cece ee ee eee ee MMONLASTLACAICAVErNOSAIEANNACUS LTO). ce orcce crchere ie ete crerere ete ere ee re ee eee ee er eee iMontastraeacylindrica(Duncans (863) ia oaccat etree ee ee ee ee es ee ee eee Montastraeaendothecatai@Duncans i863)" sons ic ece cette oo ie ree et a cae ere MONAT ACANITNUALGI(MOUN GAN SIS OS) wets sceseccc erararaee ees aor ee ene ee eee ne ee eee WMontastraeatrinitatis; (Vaughanwn Vaughan-and Hofimeisters 1926)) seescec sec cee oe cies ecterisicioeie si cieieieieoiereraeae cites Genuspsolenastrea MilnesEGwardsiandHMatme |S 4 Oi erecr -ncrcyctsess cee arc acre eee reeset ete ee Peete oi sietencielch ie teicietal ctessicticteletsta te aigisis SolenastreaibournoniMualnevEdwards andehaimewl849) paesera nase ence eens ey erie ete ere tie eines SOLENASTFeanYAades (Manassa) eps Sek pracy rec ate ee Een TE ee eee eee ete Appendix Ia. Means and standard deviations of all characters in the seven species of Montastraea herein described ............. Appendix Ib. Means and standard deviations of all calical characters in the two species of So/enastrea herein described .......... ReferencesiGitedig te te teeta A Sen cy. fac on Ae edo hh oashlesy sect connate tu cgce yesh utes Fede REPS ee EOL TER Cer var aE oO OG, sr GE RT ey Sane Re ALALOS Uae ete ros aXe vo cree Sere cu © SecA SAPS Sylce foo Ve i ba eerteg oR Te ghee) one aloha One E NG ROS HepB MGS RSEIor SO eT oF ELSE CASON SERS C Les eae AT LST ST hehe co Ra ener LIST OF ILLUSTRATIONS Text-figure Page 1. Scanning electron microscope photographs showing septal structure in three families within the suborder Faviina .............. 7 2 \Maprndicating the locationiofitheixivenrsections:sam pled) fice ec eysisie cs cscieiate cieisistese eicie ctiesinis © tical icisiee ele eieieieete eae eee 9 35 Bancharts summarizing theiquantity ofunaterialicollected) jaa4--- see seel- eee eects eierers cies cae cisions iene eee 10 4. Diagrams ‘showing the'distributions of'species' within'selected! river:sectioms: (22... <== -ee sere cic senses) sere sini wield eterno 11 5. Montastraea. Variation within species in two corallite character complexes through a composite stratigraphic section ........... 12 6. Solenastrea. Variation within species in the corallite character complex distinguishing species through a composite stratigraphic EXC) 010) acer REE ae Rewer terrae eR (en nce ee ede ea re SE ere PRE SG TSe Leesencr ae ROMO Re GhtO DEO RO Ob bdodenooceece 13 7. Scanning electron microscope photographs of modern Montastraea annularis from different reef habitats near Discovery Bay, LET 6 (hc ae oi gener ean aoe ee ns een ire ee em ee ae eee SMA O DLAC act ccoooeose< 15 8. Longitudinal thin-sections showing the structure of the coenosteum in Solenastrea and Montastraea ....................-+00-- 17 9. Drawings showing some of the characters measured and points digitized on thin-sections ......................---..0-2000005- 21 10: .Glusteranalysis'of: colonies of Montastraea:in the NMB)collections: <2. <= sensi c eietisicie an aelemie «-rareiele yates «octets eee 23 iil”, WWontastraeaGanonical/ discriminant analysis’ ofthe NNMBicollections’ 22. ..-.-0.--2-csee- ce soe eels cee eee eee ene 24 125 1 Gluster‘analysis‘oficolonies'of Solenastreainithe NMB' collections: sn... as <2 sepeneicl sieve ce ciereie = eat ee sive eee 24 135 Solenastrea: Canonical discriminant analysis'ofithe NMBicollections! <0. .....m.-- 7.6 ses 2 cose er eenien see ee nee eee 25 14. Means and standard deviations for eight characters in the seven Montastraea specieS .............. 0. sce e eee cece eee e eee eeee 27 15. Means and standard deviations for six characters in the two Solenastrea specieS .............00000cce cece eee tees eset eeeee 28 16. Montastraea. Canonical discriminant analyses distinguishing three Oligocene and ten Neogene Caribbean species .............. 30 17. Montastraea. Network of shortest Mahalanobis’ distances between Caribbean species... 2.2... 6. ee eee eee ee 31 18. Montastraea. Variation within Caribbean species in corallite characters through the Cenozoic ................... 0.002 e eee e eee 32 19. Solenastrea. Comparisons with the Tamiami Formation of south Florida, formations in the Dominican Republic, and the Imperial Formation ‘of:south-central'Galifornial %c/ecvc 25 weve eis sis evans wie. Stores aate a 6 Wis tae NO eM teases Seale oteee ayo Tne ne eee 33 20. Drawing on which the original description of Montastraea cavernosa was based .............. 00 cc cece cece eect eee nese seen 38 LIST OF TABLES Table Page 1. List of specimens of Montastraea collected by E. and H. Vokes, and measured and used in the statistical analyses .............. 13 2. Chi-square approximations resulting from the Kruskal-Wallis test and Spearman correlation coefficients between stratigraphic position within the Dominican Republic sequence, and the first two canonical variables (CV1, CV2) distinguishing species in each genus .... 14 3. List of all formally described species of Agathiphyllia, Montastraea, and Solenastrea from the Miocene through lower Pliocene of the’Garibbean region: showing their currentitaxonomic status) | -/s5- sccse eos ae eile eines cictieieiel toes ee aisle ee ere eee 17 4. List of Neogene types identified by T. W. Vaughan and used in statistical analyses of Montastraead . 2... cee ee 18 5; List:and' description of corallite:characters‘analyzed'in Montastraed 2)... 2 scccee cases see siosie ciislemie aieleierceeiee eects 19 6: List:and‘descriptionoficorallitecharacters:analyzedian Solenastrea) So n.sseoees oie ieee eine eae reat 20 7. Weighting of characters\in the Montastraea‘stepwise\discriminant analysis) 222. 2..2-. y- eee severe i reiele eee eese ee eee eee 22 8. Weighting of characters in the Solenastrea stepwise discriminantianalysis 22... 5 scene neces cee cece remains « cecueinete 24 9. Montastraea. F-statistics for Mahalanobis’ distances between the seven NMB clusters, groups based on Neogene types, and populations Ofitwo; Modern species: a..0y.. 6 5:2, cysseredsnacacs.srevoveie, slasctateiencicseuorevese shane rate spake ovasetttele etek Siti e ooo TEE Tee Oe Eee ee 25 10. Solenastrea. Differences in canonical discriminant scores between means of the two NMB clusters and holotypes for Caribbean NEOBENE SPECIES. oak Ssieis Pies Hees Gee ewe ale idee SEO GIS OR eer EEE EE nee ne ee en CROC Oceans 26 NEOGENE PALEONTOLOGY IN THE NORTHERN DOMINICAN REPUBLIC 11. The family Faviidae (Anthozoa: Scleractinia) Part I. The Genera Montastraea and Solenastrea by ANN F. BUDD Department of Geology The University of lowa Iowa City, IA 52242 U.S.A. ABSTRACT Multivariate statistical analyses are used to distinguish species in the genera Montastraea and Solenastrea through a continuous Neogene sequence (five Ma time interval) in the Cibao Valley of the northern Dominican Republic. Some older (by approximately 10 Ma) material from the same region also is included in the analyses. The material consists of approximately 280 colonies of Montastraea (74 of which are measured) from a total of 59 localities, and 66 colonies of So/enastrea (15 of which are measured) from a total of 37 localities. Twelve additional colonies of Montastraea from the Vokes’ collections of the same localities are also measured, and added to the data set. The material is first sorted into the two genera on the basis of qualitative examination of septal structure, the structure of the columella and associated paliform lobes, and the texture of the coenosteum. Sixteen characters consisting of linear distances and counts are measured in transverse thin-sections of ten corallites per colony in Montastraea, ten similar characters are measured on the upper surface of ten calices per colony in Solenastrea. The data are analyzed using cluster and canonical discriminant analysis to group the colonies into clusters representing species. Seven species are so defined in Montastraea and two in Solenastrea. These groupings are then used statistically to reclassify type specimens for 12 of the 17 described species of Montastraea and four of the seven described species of So/enastrea. Three of the 12 species are synonymized in Montastraea, and two of the four species are synonymized in Solenastrea. Further qualitative study of the remaining types suggests that nine species of Montastraea and two species of So/enastrea existed altogether in the Caribbean during the Neogene. The stratigraphic range of two of the seven Dominican Republic species of Montastraea is shown to extend back to the Oligocene. Another of the Dominican Republic species is found to exist today, and is widely distributed throughout the Caribbean. Of the nine Neogene Caribbean species, only this species survived the Plio-Pleistocene extinction event. Only one species of Montastraea is found to be endemic to the Dominican Republic. One of the remaining three species of Montastraea also has a limited stratigraphic distribution and appears confined to the southern Caribbean. Both species of So/enastrea appear to range from the Early Neogene to the Recent, and are widely distributed throughout the Caribbean. Trends within each species of Montastraea are analyzed through the sequence using nonparametric statistical procedures. Significant changes are detected upsection for at least four of the seven species in character complexes related to corallite size, septal development, and coenosteum development; however, significant correlations with species diversity suggest that these trends may be environmental in origin. Occurrence data suggest that two of the seven species of Montastraea may be indicative of shallow, nearshore conditions, whereas another two may be confined to muddy, and presumably deeper, patch reef localities. When data spanning the Oligocene to Recent are analyzed, significant directional trends are detected in one of the three longer- ranging Dominican Republic species; however, the amount of change does not exceed that observed within modern species. This suggests that, despite an apparent zigzag pattern, net stasis may be the rule in Montastraea. This study represents part of a multidisciplinary project on the paleontology and stratigraphy of the northern Dominican Republic, coordinated by P. Jung and J. B. Saunders of the Naturhistorisches Museum in Basel, Switzerland. RESUMEN Se utilizan analises estadisticos para distinguir especies en los géneros Montastraea y Solenastrea a travéz de una secuencia Neogena continua (intervalo de tiempo de cinco milliones de anos) en el Valle Cibao en el norte de la Republica Dominicana. Se incluyen también en los analises algunos materiales mas antiguos (de aproximadamente 10 milliones de anos) de la misma region. Los materiales consisten en aproximadamente 280 colonias de Montastraea (74 de las cuales se miden) de un total de 59 localidades, y 66 colonias de Solenastrea (15 de las cuales se miden) de un total de 37 localidades. También se miden y se agregan al conjunto de datos 12 colonias adicionales de Montastraea de las colecciones Vokes de las mismas localidades de la Universidad de Tulane. Primero se separa el material de dos géneros en base a examenes Cualitativos de la estructura del septo, la estructura del eje central y de los lobulos paliformes asociados, y de la textura del coenosteum. Luego se miden 16 caracteres consistentes de distancias y cuentas lineares en secciones finas transversas de 10 coralitas por colonia en Montastraea; se miden 10 caracteres similares en la superficie superior de 10 calices por colonia en Sol/enastrea. Se analizan los datos utilizando analises discriminativos canonicos y de grupos para agrupar las colonias en colecciones representativas de las especies. Se definen asi siete especies en Montastraea y dos en Solenastrea. Luego se usan estadisticamente estas agrupaciones para reclasificar especimenes tipos de 12 de las 17 especies descriptas de Montastraea y cuatro de las siete especies descriptas de So/enastrea. Tres de las 12 especies en Montastraea y dos de las cuatro de las especies de Solenastrea son sinonimas. Mas estudios cualitativos de los tipos BULLETIN 338 restantes sugieren que nueve especies de Montastraea y dos de Solenastrea existieron en el Caribe durante el Neogeno. Se ha demostrado que la zona estratigrafica de dos de las siete especies de Montastraea de la Republica Dominicana se remonta al Oligoceno. Un otro de las especies de la Republica Dominicana existe hoy y esta ampliamente distribuida a travéz del Caribe. De las nueve especies Neogenas del Caribe, solo esta especie sobrevive la extincion del Plio-Pleistoceno. Se ha encontrado que solo una especie de Montastraea es endemica de la Republica Dominicana. Uno de las tres otras especies restantes de Montastraea tambien tiene una distribucion estratigrafica limitada, y parece estar confinada al sur del Caribe. Ambas especies de Solenastrea aparentemente extenden desde el Neogeno temprano al Reciente, y estan ampliamente distribuidas a travéz del Caribe. Se analizan tendencias dentro de cada especie de Montastraea a travéz de la secuencia usando procedimientos estadisticos no paramétricos. Se detectan cambios importantes en una direccion arriba en la seccion en a lo menos cuatro de las siete complejidades de caracteres relacionados con el tamano de las coralitas, el desarrollo del septo, y el desarrollo del coenosteum; sin embargo, correlaciones importantes con la diversidad de especies sugieren que estas tendencias pueden ser debidas, en origen, al medio ambiente. Los datos de ocurrencia sugieren que dos de las siete especies de Montastraea pueden ser indicativa de la existencia de condiciones someras, y cerca de la costa; mientras que dos pueden estar confinadas a localidades barrosas de arrecifes isoladas, que estan presumiblemente mas hondas. Cuando se analizan datos que abarcan del Oligoceno al Reciente, se detectan tendencias direccionales significativas en solo uno de las especies Dominicanas de gran extension temporal; sin embargo la cantidad de cambio no excede lo que esta observada en especies Recientes. Este sugiere que, a pesar de un modelo que parece zigzag, estasis neta puede ser la regla en las especies de Montastraea. Este estudio representa parte de un proyecto multidisciplinario de la paleontologia y estratigrafia del norte de la Republica Dominicana, coordinado por P. Jung y J. B. Saunders del Naturhistorisches Museum en Basel, Suiza. INTRODUCTION This paper is the third in a series on the systematics and evolutionary history of the reef-corals from the middle Miocene to middle Pliocene of the northern Dominican Republic. It is the first of two papers on the family Faviidae Gregory, 1900, one of the most taxonomically diverse and abundant groups of corals throughout the sequence. Excluding the once-synon- ymized family Trachyphylliidae Verrill, 1901 (follow- ing Veron, Pichon, and Wijsman-Best, 1977), the fam- ily Faviidae is represented in the sequence by as many as nine genera and 20 species. Of these genera, two are currently extinct and five are currently restricted to the Caribbean. Similarly, only 12 species of the family Faviidae occur today in the Caribbean. Thus, the fam- ily was significantly more diverse in the Caribbean during the Neogene than it is today, and presumably experienced considerable extinction between late Plio- cene and modern time. The purpose of the present study is morphometrically to redefine and formally to describe the taxa represented in the Caribbean Neo- gene using a well-documented sequence of fossil pop- ulations. The results are interpreted to ascertain which species became extinct and which have survived until modern time. The systematic revisions that constitute the basis of the present study will be used in the future to reconstruct the phylogeny of the family globally at the species level. In general, the family Faviidae is characterized by septa composed of simple trabeculae, arranged in one or two laminar fan systems, which form smooth, acute teeth along the upper septal margins (Wells, 1956; Text- fig. 1). Within the family, genera are distinguished by colony form or, in other words, by degree of integration of corallites within colonies, a trait controlled by asex- ual budding of corallites during colony growth. Species are distinguished by the architecture of the individual corallites and, in particular, by features related to their size (Vaughan, 1901, 1907). Thus, formation of genera appears the result of changes in growth and develop- ment of colonies; whereas, formation of species (herein termed “‘speciation’’) involves changes in growth and development of individual corallites. Because of the large amount of material involved, the present treat- ment of the family has been subdivided into two parts. This first part focuses on species recognition within the two most abundant and presumably most speciose gen- era, Montastraea Blainville, 1830 and Solenastrea Milne Edwards and Haime, 1848. The second part, to follow later in the series, focuses on the recognition of seven less abundant and less diverse genera. The two genera in the present paper are strikingly similar mor- phologically. They both form massive, plocoid colo- nies by extratentacular budding; therefore, corallites within their colonies are relatively less well-integrated. Species within each of the two genera differ primarily in corallite size. As in the two previous papers (Foster, 1986, 1987), the material on which this study is based was collected between 1978 and 1980 by J. Geister, P. Jung, J. B. Saunders, and co-workers as part of their large-scale, multidisciplinary project on the paleontology and stra- tigraphy of the Neogene of the Cibao Valley region. All collecting localities are keyed into their detailed stratigraphic sections (Saunders et a/., 1982; Saunders, Text-figure 1.—Scanning electron microscope photographs show- ing upper septal margins characteristic of three families within the suborder Faviina. (A, B) the family Faviidae, characterized by reg- ularly well-developed septal teeth, SUI 54923, Favia fragum (Esper, 1795), Recent, La Parguera, Puerto Rico; (C, D) the family Mean- drinidae, characterized by minute septal teeth, SUI 54925, Dicho- coenia stokesi (Milne Edwards and Haime, 1848) Recent, Discovery Bay, Jamaica; (E, F) the family Mussidae, characterized by extremely long, wide teeth, SUI 54924, Jsophyllia sinuosa (Ellis and Solander, 1786), Recent, Discovery Bay, Jamaica. A,C, E, x10; B, D, F, x39. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD Ps os 3 2 ee — es 7 8 BULLETIN 338 Jung, and Biju-Duval, 1986). The sequence is notable in that it is one of the longest, most continuous, and best-studied sections through Neogene coral deposits in the Caribbean. It is also distinctively well-preserved. The collections studied include all macrofossils ex- tracted from the surface of the outcrop and all asso- ciated microfossils picked from bulk samples. The samples were taken at carefully selected, closely spaced stratigraphic intervals, and have been dated using mi- crofossil occurrences. In total, the project involves as many as fifty specialists on a wide variety of taxonomic groups. The eventual aim is to assemble a data set of occurrences of different taxonomic groups through the sequence and to use this data set in interpreting en- vironmental as well as evolutionary change. The first major study of the Faviidae from the north- ern Dominican Republic was made by Duncan (1863, 1864, 1868) on the Heneken collection (Heneken, 1853), now deposited at the British Museum (Natural History) [BM(NH)]. In these publications, Duncan de- scribed 16 species (12 of which were new) belonging to the family Faviidae. Of these, nine (eight of which were new) belong to the genera Montastraea and So- lenastrea. Most of Duncan’s descriptions, however, were based on single specimens or fragments of spec- imens, which Vaughan (1919) later re-interpreted as representing a total of six species of Montastraea and Solenastrea, only four of which were new. Shortly after Duncan, Pourtalés (1875) listed ten species (three un- identified) of the family Faviidae in his list of corals, collected by W. B. Gabb (Gabb, 1873), and now de- posited at the Museum of Comparative Zoology of Harvard University (MCZ) and at the Academy of Natural Sciences of Philadelphia (ANSP). None, how- ever, were described as new. The ten species included four species of the genera Montastraea and Solenas- trea. Almost fifty years later, Vaughan (1919) described five species of the family Faviidae (including four in the genera Montastraea and Solenastrea) in the Maury collection [Maury, 1917; deposited at the United States National Museum (USNM)] from the Neogene of the Dominican Republic. None were described as new. Vaughan and Woodring (1921, pp. 134, 135) later add- ed nine more faviids (including five Montastraea and Solenastrea) to the number, as part of a faunal list on their large, well-documented collections also deposited at the USNM. Again, however, no new species were formally described. Finally, Vaughan and Hoffmeister (1925) formally described two new species belonging to the family Faviidae, based on material in the Gabb collection. Neither new species belonged to Montas- traea or Solenastrea. No further work has been done on the family Faviidae from the Neogene of the Do- minican Republic. ACKNOWLEDGMENTS I am grateful to J. Geister (Bern, Switzerland), P. Jung [Naturhistorisches Museum Basel (NMB)], and J. B. Saunders (NMB) for collecting the material, pro- viding locality information, and assisting in sorting and curating specimens. Emily and Harold Vokes, Tulane University (TU), also generously provided additional material. H. Klein of the University of lowa (SUI) and R. Brickson (SUI) provided specimens of Solenastrea from the Pliocene of Florida. K. Muller (NMB) and T. Bahns (SUI) prepared the thin-sections; U. A. Dogan (SUI) assisted with scanning electron microscopy; and B. Fouke (SUI) measured the NMB Solenastrea colony surfaces. Points were digitized from Solenastrea thin- sections using image-analyzing equipment made avail- able at the NSF-sponsored workshop on ‘“‘Morpho- metrics in Systematic Biology” during May, 1988, at the University of Michigan. J. Geister provided x-ra- diographic equipment. Many of the whole colony pho- tographs were prepared by the photography staff at the British Museum of Natural History [BM(NH)]. Pho- tographs were also provided by W. Suter (NMB) and M. Serrete, Muséum national d’Histoire naturelle, Par- is (MNHNP). H. Greenberg (SUI) and C. Brochu (SUI) assisted with preparation of plates; J. Kralick (SUI) with computer graphics; and R. Petrick (SUI) and G. Greiner (SUI) with typing. I thank the following individuals and institutions for loans and assistance with museum material: R. Pan- chaud (NMB); J. Golden (SUI); S. D. Cairns and T. Coffer, United States National Museum of Natural History (USNM); B. R. Rosenand S. Naylor [BM(NH)]; R. Portell, Florida State Museum, University of Flor- ida (UF); W. D. Hartman, Yale Peabody Museum (YPM); A. Johnston, Museum of Comparative Zool- ogy, Harvard University (MCZ); M. Grasshoff, Natur- museum Senckenberg (NMS); J. Maréchal (MNHNP); and D. J. Nelson, Wagner Free Institute of Science (WFIS). I am grateful to S. D. Cairns, P. R. Hoover, and T. A. Stemann for reviewing the manuscript, and to F. Rogers and J. Golden for commenting on it. B. R. Rosen [BM(NH)] and J. W. Wells (Ithaca, NY) pro- vided invaluable advice on faviid morphology and tax- onomy. This research was supported by grants from the U. S. National Science Foundation (BSR 83-07109, BSR 86- 05277). INSTITUTIONAL ABBREVIATIONS AMNH: American Museum of Natural History, New York; NYo UsS2A: ANSP: Academy of Natural Sciences of Philadelphia, Philadelphia, PA, U.S. A. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 9 BM(NH): The Natural History Museum, London, En- gland, U. K. MCZ: Museum of Comparative Zoology, Harvard University, Cambridge, MA, U.S. A. MNHNP: Muséum national d’Histoire naturelle Paris, Paris, France NF: Nancy Foster coral collection specimen numbers (specimens reposited at USNM) NMB: Naturhistorisches Museum Basel, Basel, Swit- zerland NMS: Natur-museum Senckenberg, Frankfurt, Ger- many SUI: University of Iowa (formerly the State University of lowa), Iowa City, IA, U.S. A. TU: Tulane University, New Orleans, LA, U.S. A. UCMP: University of California, Museum of Pale- ontology, Berkeley, CA, U.S. A. UF: Florida State Museum, University of Florida, Gainesville, FL, U.S. A. UI: University of Illinois, Department of Geology, Ur- bana, IL, U.S. A. USGS: United States Geological Survey, Washington, DC, U.S.A. USNM: United States National Museum of Natural History, Washington, DC, U.S.A. WFIS: Wagner Free Institute of Science, Philadelphia, PA, U.S.A. YPM: Yale Peabody Museum, New Haven, CT, WE SHA: BIOSTRATIGRAPHY AND PALEOECOLOGY Montastraea Blainville, 1830 and Solenastrea Milne Edwards and Haime, 1848 are abundant in four of the river sections (Rio Cana, Rio Gurabo, Rio Mao, and Rio Yaque del Norte) collected by Saunders, Jung, and Biju-Duval (1986) through the Neogene of the Cibao Valley (Text-fig. 2). They were not found elsewhere in the study area. Specimens of Montastraea were col- lected at a total of 59 localities, ranging in age from middle Miocene to middle Pliocene. Specimens of So- lenastrea were collected at a total of 37 localities, rang- ing in age from late Oligocene to middle Pliocene. Two species, M. limbata (Duncan, 1863) and S. bournoni Milne Edwards and Haime, 1849 were es- pecially common, occurring at more than 30 localities each (Text-fig. 3). S. bournoni was found throughout all four river sections, whereas M. limbata was re- stricted to late Miocene and younger portions of the four river sections (Text-fig. 4). Two species, M. trinitatis (Vaughan in Vaughan and Hoffmeister, 1926) and S. hyades (Dana, 1846) were more common lower in the sequence, especially in the lower to middle Miocene Lopez section of the Rio Yaque del Norte. In general, however, S. hyades was notably rare throughout the studied sequence. Three species, M. brevis (Duncan, 1864), M. cylindrica (Dun- can, 1863), and M. endothecata (Duncan, 1863) were not found in the sections along the Rio Yaque del Norte. They were common, instead, in exposures of 9 10 20km 4 Rio Cana 2 Rio Gurabo 3 Rio Mao 4 Rio Amina 5 Cafada Zalaya 6 Rio Yaque del Norte 7 City of Santiago 8 Arroyo Punal 9 Rio Verde | Upper Cenozoic fe. Oligocene - Early Miocene ? Text-figure 2.— Map indicating the location of the river sections sampled. Montastraea and Solenastrea were found in only four sections: (1) Rio Cana, (2) Rio Gurabo, (3) Rio Mao, and (6) Rio Yaque del Norte (map from Saunders, Jung, and Biju-Duval, 1986). 10 BULLETIN 338 the late Miocene to earliest Pliocene Gurabo Forma- tion along the Rio Gurabo and the Rio Cana. Of these three species, only M. cylindrica was found higher in the section in the early Pliocene Mao Formation reefs along Rio Cana. The remaining two species, M. canalis (Vaughan, 1919) and M. cavernosa (Linnaeus, 1767), were found in moderate abundances throughout the four river sections. > PERCENT OCCURRENCE IN ALL MONTASTRAEA LOCALITIES ies] PERCENT OCCURRENCE IN ALL SOLENASTREA LOCALITIES 200 + 180 + 160+ 140+ 120+ 100 + 80+ 60 + 40+ TOTAL NUMBER OF COLONIES >| BE a bre can cav cyl end lim To trace patterns of evolutionary and environmental change through the sequence within the nine species, morphologic variation was analyzed quantitatively across a composite of the Rio Cana and Rio Gurabo sections [constructed by correlating the two sections as in Foster (1986; 1987)], which spans a time interval of approximately five million years (Text-figs. 5, 6). The characters analyzed consisted of the most impor- tant character complexes (the so-called canonical vari- ables calculated in the next section) distinguishing the Dominican Republic species within each genus. In Montastraea, three complexes were analyzed: (1) cor- allite size (canonical variable 1); (2) septal develop- ment (canonical variable 2); and (3) coenosteum de- velopment (canonical variable 3). In So/enastrea, only one character complex was studied, a complex related to the length of the tertiary septa and the development of the columella. To determine if any changes in these complexes oc- curred upsection, the composite section was subdivid- ed into twelve 100 m-thick intervals, and a Kruskal- Wallis rank-sum oneway analysis of variance (PROC NPARIWAY of SAS) was performed to determine if any differences existed in means of colonies collected within each 100 m interval. In Montastraea, no sig- nificant differences could be detected in the three char- acter complexes between stratigraphic intervals within any of the seven species. The one possible exception occurred in M. cylindrica in canonical variable 2 (Text- fig. 5). In Solenastrea, too few stratigraphic intervals were represented within each species to permit com- parison (Text-fig. 6). Further statistical tests performed by calculating Spearman rank correlation coefficients between ca- nonical variable scores and stratigraphic elevation on the composite section for each colony (PROC CORR = Text-figure 3.—Bar charts summarizing the quantity of material collected. (A) percentage of all Montastraea localities containing each species; (B) percentage of all Solenastrea localities containing each species; (C) total number of colonies collected of each species. “‘n” = total number of localities, “bre” = M. brevis, “can” = M. canalis, “cav” = M. cavernosa, “cyl” = M. cylindrica, “end” = M. endoth- ecata, “lim” = M. limbata, “tri” = M. trinitatis, “bou” = S. bour- noni, “hya” = S. hyades. —_s Text-figure 4.—Diagrams showing the distributions of species within selected river sections collected by Saunders, Jung, and Biju- Duval, 1986. Each vertical line within each plot represents one spe- cies. Ticks along each line represent stratigraphic positions (meters from the datum as measured by Saunders, Jung, and Biju-Duval, 1986) at which the species was found to occur. Numbers to the right of each tick mark indicate the number of localities represented by each point. “‘n” = total number of colonies containing each species, “bre” = M. brevis, “can” = M. canalis, “‘cav’”’ = M. cavernosa, “cyl” M. cylindrica, “end” = M. endothecata, “lim” = M. limbata, “tri” = M. trinitatis, ‘““bou”” = S. bournoni, “hya” = S. hyades. (A) Rio Gurabo, (B) Rio Cana, (C) Rio Yaque del Norte. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD il of SAS) suggest that slight directional change may have occurred upsection in M. cylindrica in canonical vari- able 2, and in M. brevis and M. cavernosa in canonical variable 3. Because of the small sample sizes involved in these analyses, 12 colonies of Montastraea collected by Em- ily and Harold Vokes of Tulane University (Table 1) and 13 colonies collected by T. W. Vaughan [USNM 62728 (NF448), 66829 (NF458, 460, 461), 66831 (NF489), 66832 (NF485), 66833 (NF420, 421), 66867 (NF424, 425), 66899 (NF284), 66902 (NF289), 66904 (NF292)] were added to the data set, and the analyses were rerun using data for the first two canonical vari- ables on each corallite. In this case, Kruskal-Wallis tests indicate that highly significant differences occur between statigraphic levels in all six species (Table 2), A Rio Gurabo bre icav Gy en) Iiin boul hya Rio Cana n=5 n=37 n=2 =e 4 9 x 1 2 1 1 ' ' 2 n=6 5 x 9 2 1 1 n=4 ? 2 x bire> can cav cyl Tend iim tri bow ihya C R Lon eaiaiue = alell Norte SN n= S77; can cav lIm tri bou hya 12 BULLETIN 338 with the exception of canonical variable | in M. can- alis. Spearman rank correlation coefficients suggest that change within two of the five species (4. brevis and M. cavernosa) is directional in canonical variable 1, and that change within four of the six species (M. /im- bata, M. brevis, M. canalis, and M. cavernosa) is di- rectional in canonical variable 2. In other words, cor- allite size is increasing upsection in M. cavernosa, and is decreasing upsection in M. brevis. Septal develop- ment is decreasing overall upsection in M. cavernosa, M. canalis, M. brevis, and M. limbata. These results suggest that significant change may be occurring within A MONTASTRAEA CANONICAL VARIABLE 1 100 300 500 700 900 1100 COMPOSITE SECTION 88) CANONICAL VARIABLE 2 100 300 500 MONTASTRAEA 700 900 1100 COMPOSITE SECTION Text-figure 5.—Montastraea. Variation within species in two corallite character complexes through a composite stratigraphic section (con- structed by correlating the two sections as in Foster, 1986, 1987). The points (labelled 1-7) represent means for every 100 m interval along the composite section. 1 = M. limbata, 2 = M. trinitatis, 3 = M. brevis, 4 = M. canalis, 5 = M. cylindrica, 6 = M. cavernosa, 7=M. endothecata. Vertical lines on either side of each point are one-half standard deviation in length. (A) Canonical variable | of the Montastraea canonical discriminant analysis, which is most strongly related to corallite size. (B) Canonical variable 2 of the Montastraea canonical discriminant analysis, which is most strongly related to septal development. In canonical variable 1, significant directional change was detected upsection in species 3 (decrease) and species 6 (increase). In canonical variable 2, slight decreases were detected overall in species 1, 3, 4, and 6. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 13 Table 1.—List of specimens of Montastraea collected by E. and H. Vokes, and measured and used in the statistical analyses. See Saunders, Jung, and Biju-Duval (1986) for detailed descriptions of localities. catalogue number locality number section species USNM 86898 TU 1231 Rio Gurabo M. endothecata USNM 86899 TU 1422 Rio Cana (Arroyo Bellaco) M. canalis USNM 86900 TU 1354 Rio Cana (Canada de Zamba) M. limbata USNM 86901 TU 1344 Rio Gurabo M. cylindrica USNM 86902 TU 1405 Rio Yaque del Norte (Arroyo Babosico) M. endothecata USNM 86903 TU 1405 Rio Yaque del Norte (Arroyo Babosico) M. limbata USNM 86904 TU 1215 Rio Gurabo M. brevis USNM 86905 TU 1246 Rio Gurabo M. brevis USNM 86906 TU 1278 Rio Gurabo M. endothecata USNM 86907 TU 1208 Roadcut 4 km east of Los Quemados M. limbata USNM 86908 TU 1208 Roadcut 4 km east of Los Quemados M. trinitatis USNM 86909 TU 1208 Roadcut 4 km east of Los Quemados M. cavernosa lineages that appear static based on only the NMB With the data at hand, it is difficult to determine if material. To evaluate such change adequately, many the changes observed in the larger data set are envi- more specimens of each species need to be collected ronmental in origin. Although the Cibao Valley Neo- and measured from the sequence using the NMB lo- gene sequence is believed to have been deposited under cality scheme. gradually deepening conditions, some of the material SOLENASTREA 15 12 @ DISCRIMINANT FUNCTION ng OD —6 100 200 300 400 500 600 700 800 - 900 COMPOSITE SECTION Text-figure 6.—Solenastrea. Variation within species in the corallite character complex distinguishing species through a composite strati- graphic section (constructed by correlating the two sections as in Foster, 1986, 1987). The points (labelled | and 2) represent means for every 100 m interval along the composite section. 1 = S. hyades, 2 = S. bournoni. In each case, standard deviations are smaller than the size of the dot representing the mean for each sample. Too few stratigraphic intervals are represented to permit analysis of directional change through the section. 14 BULLETIN 338 Table 2.—Chi-square approximations resulting from the Kruskal-Wallis test and Spearman correlation coefficients between stratigraphic position within the Dominican Republic sequence, and the first two canonical variables (CV, CV2) distinguishing species in each genus. strati- graphic number of number of range in 100m ae Oe species corallites DR (m) intervals chi? I, chi? Te M. limbata (1) 304 847 6 49.02 .022 29.65 —.2228 M. brevis (3) 134 272 5 23.223 —.361° 15.18 = TE M. canalis (4) 55 382 3 5.36 = SID 20.01 = 43/15 M. cylindrica (5) 83 365 4 352252 =.109 Bleors -O15 M. cavernosa (6) 57 622 4 14.56% .3708 27.102 —.684 M. endothecata (7) 74 393 3 13.72 142 26.98 —.071 ap) = O!01- in the uppermost portions of the sequence (e.g., the Mao Formation reefs along Rio Cana) appears to have been transported from shallower water (Saunders, Jung, and Biju-Duval, 1986; Evans, 1986). Nevertheless, sig- nificant positive Spearman rank correlations between canonical variable | (corallite size) and counts of num- ber of poritid species at each locality suggest that some of the trends may indeed be environmental in six spe- cies of Montastraea (all except M. endothecata). Sig- nificant negative correlations with number of poritid species also occur in canonical variable 2 (septal de- velopment) in M. trinitatis and M. brevis. Assuming that numbers of poritid species increase in shallow reefal environments, similar increases in corallite di- ameter and decreases in septal development have been observed on modern reefs extending from deeper fore- reef to shallower backreef environments (Foster, 1980; 1985). Thus, morphology may be responding to an increase in water energy and light. The distribution patterns of occurrences of each spe- cies also appear to be environmentally controlled. For example, two of the three species of Montastraea that occur lower in the section (M. cavernosa and M. trin- itatis) appear to be confined to the upper portions of the sequence (the Mao Formation) along Rio Cana (Text-fig. 4), and may be indicative of more nearshore conditions. M. brevis and M. endothecata, on the other hand, appear to be confined to the muddy and pre- sumably deeper patch reef localities in lower portions ofthe Gurabo Formation (Text-fig. 4). Occurrence data for more reef-coral species are needed to substantiate these hypotheses. In contrast to the generally wide biogeographic dis- tributions of both modern (Veron, 1986) and fossil (Foster, 1986) corals, two of the seven Montastraea species described herein (M. brevis and M. cylindrica) are restricted to a few isolated localities in the southern and central Caribbean, and may eventually prove in- dicative of a southern reef-coral biogeographic prov- ince or subprovince within the Caribbean during the Neogene (Budd, 1989). Data from more localities are needed to substantiate this hypothesis. In conclusion, data on both morphological variation within species of Montastraea and associations of all reef-coral species through the sequence offer enormous potential in paleoenvironmental interpretation. How- ever, in order to analyze such morphological variation within species, more samples need to be collected and measured from selected localities. /. cavernosa and to a lesser extent, M. Jimbata, appear the most useful in this regard, due to their high abundances and the more pronounced changes in morphology stratigraph- ically up the sequence. In order to analyze coral as- sociations, more of the corals in the NMB collections need to be identified and tallied (see discussion in Fos- ter, 1986). TAXONOMIC METHOD PROBLEM The major problem presented by the material stud- ied in the present monograph is that of species rec- ognition, and not recognition of higher categories. Al- though Montastraea Blainville, 1830 and Solenastrea Milne Edwards and Haime, 1848 have similar corallite sizes and spacing, and similar patterns of septal ar- rangement, they can be readily distinguished on the basis of texture of the coenosteum. Moreover, despite high levels of phenotypic plasticity and genetic vari- ability within species, the two living Caribbean species of Montastraea also appear to form discrete morpho- logic units separated by a wide morphologic gap. This gap, however, is more likely the result of extinction of species with intermediate morphologies than of mor- phologic divergence during speciation or of phyletic evolution within lineages (Budd, 1990). Preliminary studies suggest that intermediate species of Montas- traea may have been so numerous in the Caribbean during Neogene time as to form morphologic continua or species complexes (Budd, 1989; 1990). The lack of morphologically discrete species within these com- plexes appears the result of both high morphological variation within species and low morphological di- vergence during speciation (Budd, 1990). The problem DOMINICAN REPUBLIC NEOGENE. | 1: BUDD may further be confounded by iterative evolution (Bell, 1988), in which similar morphologies evolve repeat- edly over time from the same ancestral stock. Although little work has been done to examine the influence of natural selection on morphological or ge- netic differentiation between living populations, mor- phological variation caused by phenotypic plasticity has been described quantitatively within the two living Caribbean species of Montastraea by Foster (1979, 1980, 1985). The results of this work suggest that no simple patterns of variation or relationships between skeletal morphology and the environment exist. The two living species differ not only in magnitude but also in pattern of morphological variation; therefore, results computed for one species cannot unequivocally be used to predict those in another species. Variation within colonies and populations is higher in MM. cavernosa (Linnaeus, 1767); whereas variation between popula- tions is higher in M. annularis (Ellis and Solander, 1786) (Foster, 1985; Budd, 1990). Thus, genetic vari- ation is believed higher in M. cavernosa, and pheno- typic plasticity higher in M. annularis. In both species, corallite diameter and corallite spacing increase, and theca thickness decreases in more protected, muddy habitats. However, in M. annularis, coenosteum den- sity and septum thickness decrease in muddy habitats; whereas, in M. cavernosa, they remain the same or increase. Trends from shallow to deep water do not correspond with those from clear to muddy water. In M. annularis, deeper water colonies have smaller cor- allite diameters and more widely spaced corallites. Their coenosteum is denser, and theca thicker (Text-fig. 7). Similarly, trends across individual colonies from col- ony top to bottom do not always reflect those from clear to muddy water. In contrast to patterns between clear and muddy environments, coenosteum density and septum and theca thickness increase from colony top to bottom in M. annularis, and the theca thickness increases from colony top to bottom in M. cavernosa (Foster, 1985). Equally complex and unique patterns of morphologic variation have been described within four of the five living Australian species of Montastraea (Veron, Pichon, and Wijsman-Best, 1977). The first comprehensive attempt to describe the full range of variation within species complexes of Tertiary Caribbean Montastraea was made by Vaughan (1919) who distinguished two major groups of species, one (four species) with three cycles of septa and the other Text-figure 7.—Scanning electron microscope photographs of modern Montastraea annularis from different reef habitats near Dis- covery Bay, Jamaica. All photos, = 10. (a) SUI 45448, mid-forereef (20 m), (b) SUI 45794, deep forereef (50 m), (c) SUI 47056, backreef (1-2 m). Deeper-water colonies have smaller corallite diameters, more widely-spaced corallites, denser coenostea, and thicker thecae and septa. 16 BULLETIN 338 (10 species) with four cycles. Within each group, he distinguished species primarily on the basis of (1) cor- allite size, (2) development of the costae, and (3) rel- ative thicknesses and lengths of the septa. Within two species [M. cavernosa and M. tampaensis (Vaughan, 1919)], Vaughan (1919) named varieties, again based on corallite size and costae development. He described each variety as distinctive, but within the ‘“‘ordinary” range of variation of the species. Nevertheless, Vaughan (1919) did change the status of some of his earlier varieties (e.g., Orbicella cavernosa brevis Vaughan, 1901) by raising them to species status as new material became available for study. In general, Vaughan’s treatment is particularly noteworthy in that he ranked the characters he used to distinguish species by study- ing variation within the living Caribbean species. Another significant attempt at describing such Neo- gene species complexes in Montastraea was made by Chevalier (1954, 1961). In the Mediterranean Miocene alone, he distinguished five subgenera of Montastraea on the basis of wall structure and the development of the coenosteum. Within each subgenus, he distin- guished a number of species (total: 28) on the basis of (1) development of the costae and (2) corallite size. Many of Chevalier’s species, unlike Vaughan’s, were based on fewer than three specimens, and Chevalier was unable to compare variability within living species with that he observed in the fossils. Although Solenastrea is not reported to form such extensive species complexes, the problem of recogniz- ing species is equally difficult, also due to widespread morphologic variation within species and morphologic overlap between species. This is especially the case in the two living Caribbean species, S. hyades (Dana, 1846) and S. bournoni Milne Edwards and Haime, 1849, which are distinguished primarily on the basis of highly variable characters such as fusion of the ter- tiary and secondary septa and corallite diameter (Vaughan, 1919). Although no attempts have been made to describe the variability among living repre- sentatives within these two species quantitatively, or to relate variability to specific environmental param- eters, Vaughan (1919) does note extensive variation in corallite size and spacing and in coenosteum density. Due to the above noted problems in species recog- nition, species in the present study have been discrim- inated phenetically using a morphometric approach similar to that of Foster (1984). The specimens were first qualitatively sorted into genera, and then quan- titatively grouped into clusters using two multivariate statistical procedures: (1) cluster analysis based on dis- tances between colonies; and (2) a series of canonical discriminant analyses in which the original clusters were combined and modified until the clusters proved maximally discrete. To facilitate final cluster defini- tion, patterns of variation within each cluster were compared with variation observed among nearby, en- vironmentally-distinct Jamaican populations of the two living Caribbean species. The final clusters, therefore, represent morphologic concentrations of specimens, herein recognized as “‘species’’, which can be traced through time and theoretically may overlap at the mar- gins. Because of the limited amount of material in the NMB museum collections, the present study is only preliminary in nature. As more material is collected and analyzed, some specimens at the margins of clus- ters may prove misclassified. Thus, the present con- tribution serves mainly to identify the number of clus- ters and their centroids, and to estimate the variability within each cluster. Within this framework, all avail- able type material from the Caribbean Neogene has been re-evaluated, and the evolutionary history of each cluster traced through Neogene time. MATERIAL The material studied consists of all specimens of Montastraea (280 colonies total) and Solenastrea (66 colonies total) collected in the Dominican Republic by J. Geister, P. Jung, J. B. Saunders, and other coworkers between 1978 and 1980 (Saunders, Jung, and Biu- Duval, 1986), and is currently deposited at the Na- turhistorisches Museum in Basel. These coral collec- tions from the Dominican Republic are termed ““NMB” collections in the following discussion in order to dis- tinguish them from other type and comparative ma- terial used in the analyses. Colonies of massive plocoid faviid corals with predominantly cylindrical corallites were first separated from the rest of the NMB coral collections and then sorted by genus. None of the ma- terial was found to have the synapticular wall structure characteristic of Agathiphyllia Reuss, 1864, the la- mellar columella characteristic of Tarbellastraea Al- loiteau, 1952 and Antiguastrea Vaughan, 1919, or well- developed, prominent pali characteristic of Plesiastrea Milne Edwards and Haime, 1848. So/enastrea was dis- tinguished from Montastraea on the basis of the ve- sicular texture of the coenosteum, the reduced costae, and occasional reduced paliform lobes (Text-fig. 8). Of the 346 NMB specimens, 74 well-preserved colonies of Montastraea and 15 well-preserved colonies of So- lenastrea were selected for measurement. These were chosen to represent a wide range of localities and cor- allite sizes. To increase the sample size, 11 colonies of Montastraea collected from the same stratigraphic se- quence by Emily and Harold Vokes of Tulane Uni- versity were also thin-sectioned, measured, and in- cluded in all statistical analyses (Table 1). Measurements were also made on 33 type specimens (including topotypes, some primary types, and some nontype specimens identified by T. W. Vaughan) of 12 DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 17 Table 3.—List of all formally described species of Agathiphyllia, Montastraea, and Solenastrea from the Miocene through lower Plio- cene of the Caribbean region, showing their current taxonomic status. Agathiphyllia: Astraea antiguensis Duncan, 1863! Astraea tenuis Duncan, 1863 Cyathomorpha anguillensis Vaughan, 1919? Montastraea: Astraea brevis Duncan, 1864 Astraea costata Duncan, 1863 [= Agathiphyllia antiguensis (Dun- can)] Text-figure 8.—Longitudinal thin-sections showing the structure of the coenosteum in Solenastrea and Montastraea. (a) NMB D5794, S. bournoni, lower Pliocene, locality NMB 15822, Rio Gurabo, Mao Formation, Dominican Republic, x10; (b) NMB D5701, M. en- dothecata, upper Miocene, locality NMB 16911, Rio Mao, ?Gurabo Formation, Dominican Republic, x10. The coenosteum is more vesicular in Solenastrea due to the lack of costae extending across the coenosteum. Table 3.—Continued. Astraea cylindrica Duncan, 1863 Astraea endothecata Duncan, 1863 Astraea radiata var. intermedia Duncan, 1863 [= ? Montastraea imperatoris (Vaughan)] Cyathomorpha roxboroughi Vaughan, 1919 [= Montastraea en- dothecata (Duncan)] Heliastraea altissima Duncan, 1868 [= ? Montastraea trinitatis (Vaughan)] Heliastraea insignis Duncan, 1868 [=? Montastraea canalis (Vaughan)] Madrepora annularis Ellis and Solander, 1786 Madrepora cavernosa Linnaeus, 1767 Montastrea davisina Weisbord, 1973 [= Solenastrea hyades (Dana)] Montastrea peninsularis Weisbord, 1973 [= Solenastrea hyades (Dana)] Orbicella bainbridgensis Vaughan, 1919 [= Montastraea endoth- ecata (Duncan)] Orbicella canalis Vaughan, 1919 Orbicella cavernosa var. cylindrica Vaughan, 1919 [= Montastraea cylindrica (Duncan)] Orbicella cavernosa var. endothecata Vaughan, 1919 [= Montas- traea endothecata (Duncan)] Orbicella cumutensis Hoffmeister in Vaughan and Hoffmeister, 1926 [= ? Montastraea trinitatis (Vaughan)] Orbicella gabbi Vaughan, 1919 [= ? Diploastrea] Orbicella imperatoris Vaughan, 1919 Orbicella limbata var. pennyi Vaughan in Vaughan and Hoff- meister, 1926 [= Montastraea limbata (Duncan)] Orbicella tampdensis Vaughan, 1919 Orbicella tampdensis var. silecensis Vaughan, 1919 [= Montas- traea canalis (Vaughan)] Orbicella trinitatis Vaughan in Vaughan and Hoffmeister, 1926 Plesiastraea ramea Duncan, 1864 [= Montastraea limbata (Dun- can)] Phyllocoenia limbata Duncan, 1863 Phyllocoenia sculpta var. tegula Duncan, 1863 [= Montastraea limbata (Duncan)]} Solenastrea: Astraea hyades Dana, 1846 Cyphastrea tampae Weisbord, 1973 [= Solenastrea bournoni Milne Edwards and Haime] Plesiastraea distans Duncan, 1864 [= Solenastrea bournoni Milne Edwards and Haime] Plesiastraea globosa Duncan, 1864 [= Solenastrea bournoni Milne Edwards and Haime] Solenastrea bournoni Milne Edwards and Haime, 1849 Solenastrea fairbanksi var. minor Vaughan, 1917 [= ? Solenastrea bournoni Milne Edwards and Haime] Solenastrea fairbanksi var. normalis Vaughan, 1917 [=? Solen- astrea bournoni Milne Edwards and Haime] Solenastraea verhelsti Milne Edwards and Haime, 1857, of Duncan (1864) [ =Solenastrea bournoni Milne Edwards and Haime] Stephanocoenia fairbanksi Vaughan, 1900 [= ? Solenastrea bour- noni Milne Edwards and Haime] Stephanocoenia fairbanksi var. columnaris Vaughan, 1900 [= ? Solenastrea bournoni Milne Edwards and Haime] ' Miocene specimens reported of this species (Vaughan, 1919) prob- ably belong to Montastraea endothecata (Duncan) * One specimen (USNM 325214; Pl. 1, fig. 5) originally assigned to this species belongs to Agathiphyllia hilli (Vaughan); all others are questionably assigned to Montastraea. 18 BULLETIN 338 Table 4.—List of Neogene types identified by T. W. Vaughan and used in statistical analyses of Montastraea. Specimens in groups 16 and 17 consist of colonies of the two living species mentioned in the text. Group 11: 1. USNM 66833 (NF 420), topotype, Astraea brevis Duncan, Dominican Republic 2. USNM 66883 (NF 421), topotype, Astraea brevis Duncan, Dominican Republic 3. USNM 66829 (NF 458), topotype, Astraea brevis Duncan, Dominican Republic 4. USNM 66829 (NF 460), topotype, Astraea brevis Duncan, Dominican Republic 5. USNM 66829 (NF 461), topotype, Astraea brevis Duncan, Dominican Republic 6. USNM 66831 (NF 489), topotype, Astraea brevis Duncan, Dominican Republic Group 12: 7. USNM 66906 (NF 385), topotype, Astraea cylindrica Duncan, Dominican Republic 8. USNM 66880 (NF 68), Vaughan nontype, Astraea cylindrica Duncan, loc. USGS 8297, Trinidad Group 13: 9. USNM 66867 (NF 424), topotype, Astraea endothecata Dun- can, Dominican Republic of the 17 described species of Montastraea from the Neogene of Caribbean region (Tables 3, 4). Two of the remaining six described species, M. annularis (Ellis and Solander, 1786) and M. cavernosa (Linnaeus, 1767), occur today in a range of reef environments across the Caribbean region, and were represented in the current analyses by measurements taken on 40 colonies of liv- ing M. annularis (group 16: SUI 45425-45464) and 32 colonies of living M. cavernosa (group 17: SUI 48748-48779), both collected from four environmen- tally-distinct reef habitats near Discovery Bay, Ja- maica. Of the remaining four species, 4. cumutensis (Hoffmeister in Vaughan and Hoffmeister, 1926) and M. ramea (Duncan, 1864) were each described on the basis of only one small specimen, the holotype, which could not be thin-sectioned due to museum restric- tions. M. radiata var. intermedia (Duncan, 1863) also consists of one specimen, which could not be found. In Solenastrea, measurements were made on the sur- face of six holotypes listed in Table 3 (Plesiastraea dis- tans Duncan, 1864, Plesiastraea globosa Duncan, 1864, Solenastrea bournoni Milne Edwards and Haime, 1849, Stephanocoenia fairbanksi Vaughan, 1900, Solenas- trea fairbanksi minor Vaughan, 1917, and Solenastrea Jairbanksi normalis Vaughan, 1917). Holotypes for As- traea hyades Dana, 1846 and Stephanocoenia fair- banksi columnaris Vaughan, 1900 could not be found, and the surface of the holotype for Cyphastrea tampae Weisbord, 1973 is too poorly preserved for measure- ment. Therefore, to represent Solenastrea hyades in the statistical analyses, the holotype of Astraea excelsa Table 4.—Continued. 10. USNM 66867 (NF425), topotype, Astraea endothecata Dun- can, Dominican Republic Group 14: 11. USNM 353656 (NF 65), hypotype, Heliastraea altissima Duncan, loc. USGS 8297, Trinidad 12. USNM 66832 (NF 485), Vaughan nontype, Heliastraea al- tissima Duncan, Dominican Republic Group 15: 13. USNM 63432 (NF 276), Vaughan nontype, Heliastraea in- signis Duncan, loc. USGS 8713, Dominican Republic Group 18: 14. USNM 324881 (NF 192), holotype, Orbicella bainbridgensis Vaughan, loc. USGS 3383, Georgia 15. USNM 324882 (NF 209), topotype, Orbicella bainbridgensis Vaughan, loc. USGS 3383, Georgia Group 19: 16. USNM 324867 (NF 261), topotype, Orbicella canalis Vaughan, loc. USGS 6016, Panama 17. USNM 324867 (NF 258), topotype, Orbicella canalis Vaughan, loc. USGS 6016, Panama 18. USNM 324867 (NF 260), topotype, Orbicella canalis Vaughan, loc. USGS 6016, Panama 19. USNM 324867 (NF 263), topotype, Orbicella canalis Vaughan, loc. USGS 6016, Panama Group 21: 20. USNM 324890 (NF 238), topotype, Orbicella imperatoris Vaughan, loc. USGS 6015, Panama 21. USNM 324872 (NF 244), topotype, Orbicella imperatoris Vaughan, loc. USGS 6016, Panama 22. USNM 324875 (NF 246), topotype, Orbicella imperatoris Vaughan, loc. USGS 6016, Panama 23. USNM 324875 (NF 247), topotype, Orbicella imperatoris Vaughan, loc. USGS 6016, Panama Group 22: 24. USNM 66878 (NF 61), Vaughan nontype, Orbicella limbata var. pennyi Vaughan, loc. USGS 9198, Trinidad Group 23: 25. USNM 324891 (NF 172), topotpye, Orbicella tampaensis Vaughan, loc. USGS 4999, Florida 26. USNM 324890 (NF 176), topotype, Orbicella tampaensis Vaughan, loc. USGS 4999, Florida 27. USNM 324890 (NFI177), topotype, Orbicella tampaensis Vaughan, loc. USGS 4999, Florida 28. USNM 324890 (NF 178), topotype, Orbicella tampaensis Vaughan, loc. USGS 4999, Florida Group 24: 29. USNM 66852 (NF 492), Vaughan nontype, Orbicella trinitatis Vaughan, Dominican Republic Group 25: 30. USNM 66899 (NF 284), topotype, Phyllocoenia limbata Dun- can, loc. USGS 8545, Dominican Republic 31. USNM 66902 (NF 289), topotype, Phyllocoenia limbata Dun- can, loc. USGS 8541, Dominican Republic 32. USNM 66904 (NF 292), topotype, Phyllocoenia limbata Dun- can, loc. USGS 8738, Dominican Republic 33. USNM 62728 (NF 448), topotype, Phyllocoenia limbata Dun- can, Dominican Republic DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 19 Table 5.—List and description of characters analyzed in Montastraea. Measurements in characters 1-6 were made at maximum to the nearest 0.10 mm, those in characters 7-12 to the nearest 0.05 mm, and those in characters 13-16 to the nearest 0.025 mm. abbre- character viation description 1. corallite diameter CD Linear measure between theca/corallite cavity margins; average of longest and shortest lengths (CD lines) 2. total number of septa NS Count 3. corallite spacing NND Linear measure between theca/corallite cavity margins of nearest neighboring corallites 4. coenosteum diameter CND Linear measure between theca/coenosteum margins of nearest neighboring corallites 5. coenosteum density CNNV Linear measure, parallel to the CND line, of non-void material across the coenosteum 6. coenosteum density CNP Linear measure, perpendicular to the CND line, of non-void material crossing a 1 cm line 7. columella width CLW Linear measure between outer columella/corallite cavity margins, average of longest and shortest lengths 8. columella density CLNV Linear measure, parallel to two CD lines, of non-void material across the columella; average 9. theca thickness TT Linear measure between theca/coenosteum margins; average at two CD lines 10. septum length (first cycle) SLP Linear measure between columella and theca margins; average at two CD lines 11. septum length (second cycle) SLS Linear measure similar to SLP on major septa adjacent to SLP; average 12. septum length (highest cycle) SLT Linear measure between septum tip and theca margin of septum between SLS and SLP; average 13. septum thickness (first cycle) STP Linear measure of thickness of septa at SLP at septum midpoint; average 14. septum thickness (second cycle) STS Linear measure of thickness of septa at SLS at septum midpoint; average 15. septum thickness (highest cycle) STT Linear measure of thickness of septa at SLT at septum midpoint; average 16. costa thickness (first cycle) CST Linear measure of costa thickness at SLP; ~0.15 mm from corallite cavity; average Dana, 1846 [= S. hyades in Vaughan (1919)] was mea- sured. In addition, for comparison with other Neogene Caribbean faunas, five colonies of S. hyades and five colonies of S. bournoni from the lower Pliocene Tam- iami Formation of south Florida (SUI 60785-60794), and eight colonies (topotypes) of S. fairbanksi from the lower Pliocene Imperial Formation of south-central California (SUI 45614, 45616-45618, 45625, 45627, 45628, 45631) were measured in thin-section. Where possible, 10 mature corallites were measured in each colony. Maturity was judged by examination of the development of the highest septal cycle. These 10 consisted of two to five corallites in each of two to three transverse thin-sections, cut from approximately the top, middle, and base of each colony. Previous work on living Caribbean Montastraea (Foster, 1985) has indicated that this sampling scheme is adequate for estimating colony means and variances needed to discriminate species, and to make preliminary esti- mates of variation within species. CHARACTERS The characters analyzed consist of linear measure- ments and counts on 16 corallite features in transverse thin-sections of Montastraea Blainville, 1830 (Table 5; Text-fig. 9a) and on 10 calice features on colony surfaces of Solenastrea Milne Edwards and Haime, 1848 (Table 6). In addition, for comparison with other Caribbean faunas, six features were measured in trans- verse thin-sections of Solenastrea. Linear distances calculated in this third data set were based on points digitized in two dimensions (Table 6; Text-fig. 9b). In general, measurements made on thin-sections are pre- ferred over those made on calical surfaces in studies of fossil massive colonial corals, because colony sur- faces are often worn. Therefore, many more characters can be measured with greater accuracy and consistency in thin-section. However, in Solenastrea, measure- ments were made on colony surfaces, due to the im- portance of surficial paliform lobes in distinguishing species. Throughout the present study, all thin-sections were prepared from chips cut within 5 mm of the col- ony surface and ground to a thickness of 40 um. The characters were selected to include all diagnostic features traditionally used to distinguish species of Montastraea and Solenastrea (Vaughan, 1919). Al- though colony shape is described qualitatively herein in the section on systematic paleontology, no attempt has been made to quantify colony shape, or to use it as a character in discriminating species, because of the fragmentary nature of much of the material. In general, the characters analyzed can be grouped into five in- terrelated categories: (1) corallite size and spacing; (2) septal number and length; (3) columella (and associ- ated paliform lobes) width and porosity; (4) septum, theca, and costa thickness; and (5) development of the coenosteum. 1. Corallite size and spacing.— Because of their plo- coid colony form and strong hexagonal symmetry, cor- allites are almost invariably circular in both Montas- traea and Solenastrea; therefore, an average of the longest and shortest corallite diameter (CD) was used to describe the size of each corallite. However, in the 20 BULLETIN 338 Table 6.—List and description of characters analyzed in Solenastrea. Measurements in characters 1, 2, 4, 5,9, and 10 were made at maximum to the nearest 0.05 mm; those in characters 6-8 to the nearest 0.033 mm. description Linear measure across corallite center between theca/corallite cavity margins; mean of Linear measure between theca/corallite cavity margins of nearest neighboring corallites Linear measure through corallite center across paliform crown; mean of the greatest Linear measure through corallite center across columellar complex; mean of the greatest Linear measure across theca between theca/corallite cavity and theca/coenosteum mar- Linear measure across primary costa along theca Linear measure across major septum at septum midpoint; mean of thickest and thin- Linear measure from septum tip to inner thecal margin; mean of longest and shortest Vertical distance from columella to uppermost thecal margin Linear measure between margins of every other primary septa; mean of six lengths Linear measure through corallite center across columellar complex; mean of three Linear measure from septum tip to inner thecal margin; mean of three lengths Linear measure from septum tip to inner thecal margin; mean of six lengths Cosine of angle formed by points at the outer and inner tertiary septum margins and at the outer secondary septum margin; mean of three angles abbre- character viation at colony surface: 1. corallite diameter CD greatest and smallest lengths 2. corallite spacing NND 3. total number of septa NS Count of all major and minor septa 4. paliform crown width PA and shortest distances 5. columella width CLW and smallest lengths 6. theca thickness TT gins 7. costa thickness CST 8. septum thickness ST nest septum 9. tertiary septum length SLT septa 10. calice elevation CA in transverse thin-section: 1. chord length CH 2. columella width CLW lengths 3. secondary septum length SLS 4. tertiary septum length SLT 5. angle between secondary and AT tertiary septa 6. theca thickness Ihe Linear measure across theca from outer secondary septum margin to coenosteal cavity; mean of three lengths Solenastrea thin-sections, the average distance be- tween points at the thecal margins of every other pri- mary septum (CH) was used as a measure of corallite size. Calice elevation (CA) was estimated only in So- lenastrea, by measuring surface relief within calices. In both genera, each mature corallite is surrounded by five to seven mature neighboring corallites, which are separated by highly variable amounts of coenosteum. Spacing between corallites was therefore estimated in Montastraea and on Solenastrea surfaces by measuring the distance to the nearest neighboring mature corallite (NND). 2. Septal number and length.—Septal number (NS) was determined by counting all major and minor septa within each corallite. In general, in Montastraea and Solenastrea, the six primary and usually the six sec- ondary septa extend completely from the theca to the columella. If present, the 12 tertiary and 24 quaternary septa usually extend only partially from the theca to the columella, and often are directed toward and may even fuse with the next higher cycle. In the present study, the lengths of the primary (SLP), secondary (SLS), and highest septal cycle (SLT) were measured in Montastraea, and the lengths of the secondary (SLS) and tertiary (SLT) septa were measured in Solenastrea. In the So/enastrea thin-sections, the angle between the secondary and tertiary septa (AT) was also approxi- mated. 3. Columella width and porosity.—In both Montas- traea and Solenastrea, the columella consists of a spongy, upward-spiralling network of trabeculae ex- tending from the inner edges of the septa. In the present study, the width of this columellar complex (CLW) and its porosity (CLNV) were estimated. Although sep- arate, discrete structures formed by septal substitution, termed “pali” (Vaughan and Wells, 1943; Wells, 1956), do not occur, paliform lobes do develop along the up- per surface of the septa, forming a crown that encircles the columella, especially in So/enastrea. On Solenas- trea surfaces, the width of this palar crown (PA) was also measured. 4. Septum, theca, and costa thickness.—In Montas- traea, the theca is septothecal (i.e., formed by thick- ening of the septa); whereas, in So/enastrea, it is sep- tothecal or parathecal (i.e., formed by dissepiments). The costa represent prolongations of the septa beyond the theca, which form strong planar vertical partitions within the coenosteum. In the present study, the thick- nesses of three of these elements were measured and compared [i.e., the theca (TT), the septa (ST, STP, STS, STT), and the costa (CST)]. 5. Development of the coenosteum.—The volume DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 21 (CND) and density (CNNV, CNP) of the coenosteum were measured only in Montastraea. Here, the coe- nosteal voids are rectangular in shape due to the in- fluence of the costae in coenosteal construction. In con- trast, the costae are weak in Solenastrea, and the coenosteum is vesicular in structure. STATISTICAL PROCEDURES Discrimination of species using the NMB material.— Species were distinguished within each of the two gen- era in the present study following a three-step proce- dure (cf. Foster, 1984; Budd, 1988): (1) Mahalanobis’ distances were calculated between all NMB and TU colonies within each genus (PROC CANDISC of SAS); (2) these distances were used to group colonies into clusters by average linkage cluster analysis (UPGMA) Text-figure 9.—Drawings showing some of the characters mea- sured and points digitized on thin-sections. (a) Montastraea: CD, corallite diameter; CLW, columella width; CST, costa thickness; NND, corallite spacing; STP, septum thickness; TT, theca thickness. (b) Solenastrea: CH, chord length = average of distances from points 11 to 20, 20 to 29, 11 to 29, 15 to 24, 24 to 33, and 15 to 33; CLW, columella width = average of distances from points 5 to 8, 6 to 9, and 7 to 10; SLS, secondary septum length = average of distances from points 13 to 17, 22 to 26, and 31 to 35; SLT, tertiary septum length = average of distances from points 12 to 16, 14 to 18, 21 to 25, 23 to 27, 30 to 34, and 32 to 36; TT, theca thickness = average of distances from points 13 to 19, 22 to 28, and 31 to 37. (PROC CLUSTER of SAS); and (3) a series of ca- nonical discriminant analyses (SPSS-X discriminant procedure) were run on the clusters, and on combi- nations and modifications thereof, until the clusters were distinct at p<0.0001. The resulting clusters or “‘sroups”’ represent the species described in this paper. The use of Mahalanobis’ distances in cluster analysis (step 1) most heavily weights those characters that best distinguish among colonies. Canonical discriminant analysis (step 3) further refines the clusters, and thereby alleviates some of the ambiguities associated with choice of clustering level. Due to the small sample size and short length of geologic time represented (<10 Ma), no attempt was made to sort the colonies by time interval before running the analyses (cf. Budd, 1988) in order to prevent artifically inducing stasis. As found by Cheetham (1986) in trial runs on hypothetical data sets, such use of discriminant analysis in species rec- ognition does not mask gradual changes within char- acters. All analyses in the present study have been run using SAS version 5 (SAS Institute, Cary, NC) and SPSS-X release 3.0 (SPSS Inc., Chicago, IL) on the University of lowa IBM 4381 mainframe computer. In Montastraea, all 16 characters were used to cal- culate the Mahalanobis’ distances between all pairwise combinations of the 84 measured NMB and TU col- onies. The results of cluster analysis performed on these distances are shown in the dendrogram in Text-figure 10. Using a cutpoint of approximately 0.38 for colonies with smaller corallites and 0.55 for colonies with larger corallites, 14 groups were formed based on qualitative study of the dendrogram (Text-fig. 10). Cutpoints are higher for groups with larger corallites due to the fact that amounts of variability within species are strongly correlated with size (Foster, 1985). These groups were then re-analyzed by performing a series of stepwise discriminant analyses on means of the 16 characters for each colony. In this procedure, groups were com- bined if F-values derived from Mahalanobis’ distances had significance levels of greater than 0.0001. Group assignments for misclassified colonies were modified by trial and error to obtain the highest percentage of correctly classified corallites. The final results yielded seven groups among the NMB specimens in Montas- traea (Text-fig. 11; Table 7). The validity of these groups was further tested by subdividing each colony into two halves and performing average linkage cluster analysis as above using the colony half data. The positions of the two halves for each colony were then examined on the dendrogram to ensure that no two halves of the same colony were assigned to different groups. In Solenastrea, all ten colony surface characters were used to calculate the Mahalanobis’ distances between all pairwise combinations of 12 of the 15 measured colonies. The three remaining colonies had to be de- i) tO BULLETIN 338 Table 7.—Weighting of characters in the Montastraea stepwise discriminant analysis. Total-sample correlations between the canonical variables (CV1-CV3) and the original variables (COR), and standardized canonical coefficients (SCC). Only values with high magnitudes are given. Abbreviations for characters are explained in Table 5. CVI (86.1%) original variable COR SCC COR CD .703* - alle)e) NS = = .329* NND - 0.31 — CND — — - CNNV - —0.88* - CNP — — — CLW 649 0.26 .182 CLNV = 0.50 - TT — 0.25 _ SLP .666 — “153 SLS .640 0.67 _ SLT _ - S50" STP _ _ a STS = _ .166 STT _ — —.125 CST 243 —0.31 2 Cv2 Cv3 (5.2%) (4.4%) scc COR scc = — 122, —3.06 =-631 —3.18* 3.97 =a Ofae 2.07 — =-393 = - 318 =— - - Slails) 1.52 _ 1.67 1.16 _ - * Most important variables. leted at this stage of the analysis because of missing values caused by poor preservation. The cluster anal- ysis results are shown in Text-figure 12. Using a cut- point of 0.85, two groups were formed based on qual- itative study of the dendrogram. Canonical discriminant analysis performed on means of the ten characters for each colony confirmed that the two groups were dis- crete, and was used to assign the remaining colonies to one of the two groups (Text-fig. 13; Table 8). Finally, to identify the NMB species, two further canonical discriminant analyses were performed using the NMB clusters and the measured types: (1) Mahal- anobis’ distances were calculated between every pair- wise combination of the NMB Montastraea clusters and types in each of the groups representing the pre- viously-described Montastraea species in Table 4; and (2) the final canonical discriminant analyses distin- guishing the Montastraea and the Solenastrea NMB species were rerun adding the types and leaving them unclassified. Results for Montastraea are given in Ta- ble 9 and Text-figure 1 1b; and for Solenastrea in Table 10. Both sets of results reveal a very complicated re- lationship between the types and the NMB species. Therefore, to interpret these relationships further, comparisons with other Caribbean faunas reported in the next section were also used in determining final syYynonymics. Means and standard deviations of all single char- acters are given for each species in Appendix I. Char- acters revealing highly significant differences between species are plotted for Montastraea in Text-figure 14, and for So/enastrea in Text-figure 15. In all text-figures, tables, and appendices, the data used in Montastraea represent colony means, whereas in Solenastrea, they represent individual corallites. RESULTS AND INTERPRETATIONS In the final canonical discriminant analysis of Mon- tastraea, seven groups were distinguished with 100% of all colonies correctly classified. Although minor overlap occurs between groups (Text-fig. 11a), F-sta- tistics derived from pairwise Mahalanobis’ distances between groups reveal highly significant differences (in all cases, p<0.0001). Four significant canonical dis- criminant functions (CV1—CV4) were calculated, with CV1 accounting for 86.1% of the variation, CV2 for 5.2%, CV3 for 4.4%, and CV4 for 2.7%. The stepwise variable selection procedure showed that all 16 mea- sured characters were needed to distinguish maximally among groups in the final analysis. In descending order, the four most important characters were: CD, CND, SLT, and CNNV. The first canonical variable (CV1) distinguished six of the seven groups, and appears to be most strongly related to corallite size. It weighted coenosteum density (CNNV) most heavily, and it is most strongly correlated with corallite diameter (CD) (Table 7). The second canonical variable (CV2) pri- Text-figure 10.—Cluster analysis of colonies of Montastraea in the NMB collections. Dendrogram calculated by average linkage cluster analysis on Mahalanobis’ distances among colonies. Each terminal branch represents one colony and is labelled using museum catalog numbers (numbers preceded by ““D’? = NMB; others = USNM). Parentheses indicate the species to which the colony was eventually assigned: | = M. limbata, 2 = M. trinitatis, 3 = M. brevis, 4 = M. canalis, 5 = M. cylindrica, 6 = M. cavernosa, 7 = M. endothecata. Shaded areas on the dendrogram delineate the initial 14 groups described in the text. DOMINICAN REPUBLIC NEOGENE. 11: BUDD 86902(7) D5820(7) a DS 7/Nate7,) D5752(6) D5709 (6) 86909(6) D5578(6) D56498(5) 058657 (5) 05626 (|) - ps D5614(4) a D5553( 4) D5738(4) D5561(5) D5557 (5) D5647(5) D5751( 4) D5652(2) D5748(1) D5560 (3) 86804(3) * 86905(3) D5555(3) D5742(2) DETTE) 86908(2) . 05754, 2) . D5735( 2) 05732(2) D5644(1) D5643(1) D5722(1) D5563 (1) D5822 (1) 86800 (1) - D5602(1) 05593. 1) D5747(1) 05714(1) 05654( 1) 057121) ~ 05655(1) D5546( 1) AVERAGE (Di STANCE BETWEEN CLUSTIERS 86906(7) D5821(7) D5723(7) D5637( 4) D5750(6) D5600(6) D5648 (6) D5547(6) 86901(5) D5710(1) D5638 (4) D5608( 4) 86899( 4) D5619( 4) 05568 (5) D5744(2) D5638(4) D5551(4) D5646(5) D5740(2) D5726 (3) D5568 (3) D5567(3) D5743 (2) D5656(1) D5834(1) D5753(2) D5737(2) D5734(2) D5707(1) 86903(1) 05715(1) D5582(1) 0556211) 86907(1) D5617 (1) D5706(1) D5749(2) D5746(1) DSF) D5853(1) D5626( 1) D5556(1) 23 24 BULLETIN 338 marily distinguished groups 3 and 5, and appears to be most strongly related to septal development. It weighted coenosteum density (CND) most heavily, and itis most strongly correlated with number of septa (NS) and the inverse of the length of the highest septal cycle (—SLT). CND was also most heavily weighted on CV3, A MONTASTRAEA losP1 oSP5 leSP2 mSP6 x |ASP3 vSP7 Wd =| < & > z ) 75 [o) z oO -8 a4 0 4 8 12 16 CANONICAL VARIABLE 1 B MONTASTRAEA TYPES N S 2 g z Oo z [o) z << Oo CANONICAL VARIABLE 1 c MONTASTRAEA TYPES 10 ToGRP19 | @GRP21 a |aGRP22 us |aGRP23 a sl |OGRP24 @ ImGRP25 E < > all << cS) = Oo z 4 oO 0) 4 8 12 16 20 CANONICAL VARIABLE 1 Table 8.— Weighting of characters in the So/enastrea stepwise dis- criminant analysis. Total-sample correlations between the canonical variable and the original variables (COR), and standardized canon- ical coefficients (SCC). Abbreviations for characters are explained in Table 6. NU = characters not used by stepwise procedure. original variable COR scc CD 189 NU NND —.023 0.87 NS —.154 NU PA 195 2.53} CLW .214 4.66* Ame .242 NU CST 116 329 ST —.054 = iow SLT .289* 2.78 CA —.194 NU * Most important variables. D5728 (2) D5629 (2) D5627 (2) D5641 (1) D5594 (1) D5589 (1) D5753 (1) D5597 (1) D5587 (1) D5607 (1) D5572 (1) D5570 (1) a J a en o o SIN =) 2 o wv N °o o oO °o - =- - = o AVERAGE DISTANCE BETWEEN CLUSTERS Text-figure 12.—Cluster analysis of colonies of So/enastrea in the NMB collections. Dendrogram calculated by average linkage cluster analysis on Mahalanobis’ distances among colonies. Each terminal branch represents one colony and is labelled using NMB catalog numbers. Parentheses indicate the two groups or clusters evaluated using canonical discriminant analysis. — Text-figure 1 1.—Montastraea. Canonical discriminant analysis of the NMB collections. (a) Plot of scores on the first two canonical variables showing polygons outlining the range of variation between colonies in the seven species. Species | = M. limbata; species 2 = M. trinitatis; species 3 = M. brevis; species 4 = M. canalis; species 5 = M. cylindrica; species 6 = M. cavernosa; species 7 = M. en- dothecata. (b), (c) Plots of scores showing polygons around the NMB species [labelled with circled numbers as given in (a)] and points indicating the measured types listed in Table 4. Based on these results, groups 22 and 25 were synonymized with species 1; groups 14 and 24 with species 2; group 11 with species 3; groups 15 and 19 with species 4; group 12 with species 5; group 17 with species 6; and group 13 with species 7. Group 23 was later synonymized with species 4 and group 18 with species 7. Groups 16 and 21 were left unsynonymized at this stage. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 25 Table 9.—Montastraea. F-statistics (and their significance levels) for Mahalanobis’ distances between the seven NMB clusters (columns labelled 1-7), groups based on Neogene types (rows labelled 11-25) as listed in Table 4, and populations of two modern species. Because of differences in sample sizes, values were compared only between columns within rows. number of 1 2 3 4 5 6 7 species (group) colonies n= 32 n= 13 n=7 n=I11 n=8 n=8 n=7 A brevis (11) 6 7.06 (.00) 3.46 (.00) 1.75 (.04)* 5.16 (.00) 5.99 (.00) 11.45 (.00) 22.26 (.00) A. cylindrica (12) 2 6.07 (.00) 3.76 (.00) 2.76 (.00) 2.76 (.00) 2.11 (.01)* 3.10 (.00) 8.68 (.00) A. endothecata (13) 2 19.53 (.00) 14.74 (.00) 10.51 (.00) 6.79 (.00) 10.27 (.00) 3.41 (.00) 2.48 (.00)* H. altissima (14) 2 1.71 (.05) 0.98 (.48)* 1.83 (.03) 2.17 (.00) 2.05 (.01) 5.36 (.00) 12.37 (.00) H. insignis (15) 1 4.52 (.00) 4.49 (.00) 3.95 (.00) 1.98 (.02)* 3.01 (.00) 2.98 (.00) 4.75 (.00) O. bainbridgensis (18) 2 19.94 (.00) 15.91 (.00) 12.86 (.00) 11.22 (.00) 20.76 (.00) 8.82 (.00) 9.16 (.00)* O. canalis (19) 4 13.34 (.00) 7.99 (.00) 8.11 (.00) 2.30 (.01)* 5.59 (.00) 3.99 (.00) 9.42 (.00) O. imperatoris (21) 4 7.90 (.00) 8.86 (.00) 9.60 (.00) 11.66 (.00) 10.68 (.00) 19.19 (.00) 31.21 (.00) O. limbata var. pennyi (22) 1 0.45 (.96)* 1.11 (34) 1.40 (.15) 2.92 (.00) 2.82 (.00) 5.79 (.00) 10.67 (.00) O. tampaensis (23) 4 30.76 (.00) 22.39 (.00) 19.07 (.00) 12.26 (.00)* 14.70 (.00) 9.79 (.00) 12.82 (.00) O. trinitatis (24) 1 0.45 (.96) 0.45 (.96)* 1.50 (.11) 2.48 (.00) 2.43 (.00) 5.39 (.00) 10.08 (.00) P. limbata (25) 4 0.34 (.99)* 2.31 (.00) 2.70 (.00) 6.20 (.00) 4.69 (.00) 13.81 (.00) 26.04 (.00) M. annularis (16) 40 18.14 (.00) 12.14 (.00) 13.22 (.00) 25.82 (.00) 31.02 (.00) 40.68 (.00) 63.24 (.00) 32 M. cavernosa (17) 127.34 (.00) 55.08 (.00) 38.05 (.00) 24.86 (.00) 11.13 (.00) 18.72 (.00)* 19.38 (.00) * Synonymized in species descriptions. SOLENASTREA FREQUENCY =o rls ORS 2 ae a ee On aes DISCRIMINANT FUNCTION Text-figure 13.—So/enastrea. Canonical discriminant analysis of the NMB collections. Histogram of frequencies of mean colony scores on the discriminant function. Left angle stripes = S. bournoni, crosshatch = S.hyades. 26 BULLETIN 338 Table 10.—Solenastrea. Differences in canonical discriminant scores between means of the two NMB clusters (columns labelled 1, 2) and holotypes for Caribbean Neogene species. i ——_a—_—_—_—_ species Astraea excelsa Dana (= hyades) Plesiastraea distans Duncan Plesiastraea globosa Duncan Solenastrea bournoni Milne Edwards and Haime Stephanocoenia fairbanksi Vaughan Solenastrea fairbanksi var. minor Vaughan number of museum catalogue variables number used in analysis 1 2 YPM 1727 7 4.367 3.218* BMNH R28758 1/ 7.026* 14.612 BMNH R28871 7 1.425* 9.011 MNHNP 794 6 O557* 5.628 USNM 157516 3 3.019 0.077 USNM 68284 3 3.520 0.577 USNM 68283 3 3.520 LSy// Solenastrea fairbanksi var. normalis Vaughan * Synonymized in species descriptions. which appears most strongly related to corallite spacing and the structure of the coenosteum. A number of characters including CD, CND, CLW, and STS were most heavily weighted on CV4, which appears to be most strongly related to septal thickness. Univariate analyses of variance further show differ- ences among the seven clusters in all 16 measured characters. The greatest differences among clusters oc- cur in four characters related to corallite size: corallite diameter (CD); primary septum length (SLP); colu- mella width (CLW); and secondary septum length (SLS) (Text-fig. 14). Comparisons between the NMB Montastraea clus- ters and groups based on the Montastraea types (Table 4) reveal a highly complicated relationship between the two data sets. The results of the canonical discrim- inant analysis in which the types were left unclassified (Text-figs. 11b, c) show that none of the 14 groups of types falls clearly within any of the NMB clusters. In- stead, the groups lie at the margins of clusters or equally near two adjacent clusters. However, Mahalanobis’ distances between the NMB clusters and groups based on the types (Table 9) do clearly suggest some syn- onymies. Using a significance level of 0.01, groups 25 and 22 appear to be the same as NMB cluster |, groups 14 and 24 appear to be the same as NMB cluster 2, group 11 appears to be the same as NMB cluster 3, groups 15 and 19 appear to be the same as NMB cluster 4, and group 12 appears to be the same as NMB cluster 5. A relatively short distance also occurs between group 13 and NMB cluster 7. In addition, as discussed in the next section, modern populations of M. cavernosa ap- pear temporally continuous with NMB cluster 6. This reduces the 12 groups listed in Table 4 to a total of 10 pecies, one of which belongs to one of the two modern ies. As noted in the section on systematic pale- ontology, together with results of comparisons with other Caribbean faunas and qualitative study of the primary types of 14 of the 17 Montastraea species [i.e., not including primary types of M. radiata var. inter- media, M. altissima, or M. insignis that could not be found (Table 3)], these results suggest that the total number of Neogene species of Montastraea can be re- duced to nine. In the final canonical discriminant analysis of So- lenastrea, two groups were distinguished with 100% of all colonies correctly classified. The F-statistic (F=10.417 with six and four degrees of freedom) de- rived from the Mahalanobis’ distance between the two groups has a significance of p=0.02, which further in- dicates that the two groups are discrete. One canonical discriminant function was calculated which had a Wilks’ lambda value of 0.06015 and a corresponding chi- square value of 16.866, with four degrees of freedom and a p-value of 0.0098. The stepwise variable selec- tion procedure showed that only six of the 10 measured characters were needed in the final analysis. They in- cluded in descending order of importance: SLT, ST, CLW, CST, NND, and PA. The canonical discrimi- nant function weighted columella width most heavily, and is most strongly correlated with length of the ter- tiary septa (Table 8). Univariate analyses of variance show that seven of the 10 measured characters differ between the two So- /enastrea clusters. In descending order of significance, these include: tertiary septum length (SLT); corallite diameter (CD); columella width (CLW); paliform crown width (PA); and total number of septa (NS) (Text-fig. 15). Differences between clusters could not be detected in theca thickness (TT), septum thickness (ST), or cos- ta thickness (CST). The results of canonical discriminant analyses in which the six So/enastrea holotypes were left unclas- sified (Table 10) suggest that holotypes for three species (Plesiastraea distans Duncan, Plesiastraea globosa Dun- Text-figure 14.—Means and standard deviations for eight char- acters in the seven Montastraea species (Appendix Ia). The midpoint of each vertical line represents the mean, and the length of the line on either side of the midpoint is one standard deviation. Nonpara- metric analysis of variance (PROC NPARIWAY of SAS) shows that the species differ significantly in each case. “tn” = number of colonies measured. Species | = M. limbata; species 2 = M. trinitatis; species 3 = M. brevis; species 4 = M. canalis; species 5 = M. cylin- drica; species 6 = M. cavernosa; species 7 = M. endothecata. CORALLITE DIAMETER (mm) COLUMELLA WIDTH (mm) SECONDARY SEPTUM THICKNESS (mm) NUMBER OF SEPTA DOMINICAN REPUBLIC NEOGENE. |1: BUDD SPECIES SPECIES SPECIES SPECIES PRIMARY SEPTUM THICKNESS (mm) HIGHEST CYCLE SEPTUM LENGTH (mm) CORALLITE SPACING (mm) THECA THICKNESS (mm) 27 | fe) | (a7) n=105 n=73 ms SPECIES SPECIES SPECIES l \ lL n=70 SPECIES 28 BULLETIN 338 can, and Solenastrea bournoni Milne Edwards and Haime) clearly belong to NMB cluster 1. The holotype for Astraea excelsa (= S. hyades) appears closer to NMB cluster 2. Holotypes for Stephanocoenia fair- banksi Vaughan and its varieties appear closer to NMB cluster 2; however, as discussed in the next section, CORALLITE DIAMETER (mm) SPECIES 2.000 1.800 1.600 1.400 1.200 1.000 0.800 0.600 0.400 0.200 0.000 COLUMELLA WIDTH (mm) SPECIES 0.500 0.400 0.300 0.200 THECA THICKNESS (mm) 0.100 0.000 SPECIES study of thin-section measurements made on topo- types and comparisons with other Neogene Caribbean faunas suggest that S. fairbanksi actually may be in- termediate between the two NMB clusters and may therefore represent a separate species. More material belonging to NMB cluster 2 is needed to evaluate this 2.000 1.800 CORALLITE SPACING (mm) SPECIES 2.000 1.800 1.600 1.400 1.200 1.000 0.800 0.600 0.400 0.200 0.000 PALIFORM CROWN WIDTH (mm) SPECIES 2.000 1.800 1.600 1.400 1.200 1.000 0.800 0.600 0.400 0.200 0.000 TERTIARY SEPTUM LENGTH (mm) SPECIES Text-figure 15.—Means and standard deviations for six characters in the two Solenastrea species (Appendix Ib). The midpoint of each vertical line represents the mean, and the length of the line on either side of the midpoint is one standard deviation. Nonparametric analysis of variance (PROC NPARIWAY of SAS) shows that the species differ significantly in each case. ‘‘n’”’ = number of corallites measured. Species 1 = S. bournoni; species 2 = S. hyades. DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 29 Table 11.—Localities and number of colonies measured in each of five time intervals. number of age locality colonies repository 1. late Oligocene Anahuac Formation, Texas 9 SUI Juana Diaz Formation, Puerto Rico 12 SUI Brownstown Formation, Jamaica 5 SUI 2. early Miocene Tampa Formation, Florida 8 USNM Chattahoochee formation, Georgia 11 USNM Larés Formation, Puerto Rico y SUI Santa Ana and Rio Lajas formations, Chiapas, Mexico 24 UI 3. middle Miocene Anguilla Formation, Anguilla 23 USNM La Boca Formation, Panama 27 USNM Baitoa Formation, Dominican Republic 9 NMB 4. late Miocene/early Pliocene Cercado, Gurabo, and Mao formations, Dominican Republic 90 NMB 5. Recent Discovery Bay, Jamaica 72 SUI possibility. curred in the Caribbean during the Neogene. In summary, the results for Montastraea are as fol- lows. (1) Seven slightly overlapping groups exist. The most overlap occurs between species 4 and 6 and between species 3 and 5. The amount of overlap, however, is far less than that found in species of Porites from the same sequence in the Dominican Republic. (2) All 16 characters contribute to distinguishing the seven groups. The most important character complex is related to corallite size. Other character complexes are related to corallite spacing and the density of the coenosteum and to the thickness and length of the highest septal cycle. (3) Analyses of type material for 12 Neogene species and of populations of the two modern Caribbean spe- cies suggest that two described species belong to NMB cluster 1 (M. limbata), two belong to NMB cluster 2 (M. trinitatis), and possibly two belong to NMB cluster 4 (M. canalis). One described species belongs to NMB cluster 3 (M. brevis), one belongs to NMB cluster 5 (M. cylindrica), and one belongs to NMB cluster 7 (M. endothecata). This results in a reduction in the number of Neogene species from 12 to 10. Together with the results presented in the next section, this suggests that a total of nine species of Montastraea existed in the central Caribbean during the Neogene. In summary, the results for Solenastrea are as fol- lows. (1) Two groups exist. Too few specimens were col- lected of the second group, however, to examine pos- sible overlap between groups. (2) Only six of the 10 measured characters were need- ed to distinguish the two groups. The characters most important in distinguishing species are the length of the tertiary septa (SLT) and several features, such as columella width, which were found to be directly cor- related with corallite size. (3) Comparisons with primary types suggest that ei- ther two or three species of Solenastrea may have oc- COMPARISONS WITH OTHER CARIBBEAN FAUNAS To determine the total duration and geographic range of the species described herein, morphological mea- surements for each species have been compared with measurements made on colonies from a number of other Tertiary and modern Caribbean localities. In Montastraea Blainville, 1830, these localities occur in 10 widely scattered formations of late Oligocene and early to middle Miocene age (Table 11), in addition to the previously described modern reef environments near Discovery Bay, Jamaica. The measurements an- alyzed are identical to those described earlier for the NMB material (Table 5; Text-fig. 9). Measurements for all the NMB material were included in the analyses. In Solenastrea Milne Edwards and Haime, 1848, due to the general paucity of abundant older material, the material analyzed consisted of 10 colonies from the lower Pliocene Tamiami Formation of south Florida and eight colonies from the lower Pliocene Imperial Formation of south-central California, as described previously in the section on material. Unlike the pre- vious statistical analyses, the measurements were de- rived from points digitized on thin-sections, as de- scribed previously in the section describing the characters (Table 6; Text-fig. 9). Measurements for only eight NMB colonies were included in the analyses: four from Rio Gurabo (D5584, D5603, D5724, D5794), one from Rio Cana (D5627), and three from the Mio- cene section along Rio Yaque del Sur (D5670, D5675, D5690). To analyze the Montastraea data, the material was sorted into five time intervals as shown on Table 11, and species were statistically discriminated within each time interval using cluster analysis and canonical dis- criminant analysis as described in the previous section on statistical techniques. In addition to the two modern Caribbean species, a total of 20 fossil species were 30 BULLETIN 338 distinguished altogether (13 of which are shown on Text-fig. 16). These species were then linked into lin- eages representing single species lines using Mahala- nobis’ distances among all 21 species (following the methods of Budd, 1988). In this procedure, distances between species in successive time intervals were com- pared with distances between species within time in- tervals (Text-fig. 17). Species in successive time inter- vals were linked if the distance between them was less than the minimum distance between species in each of the two intervals. In three cases, linkages appear ambiguous: (1) in M. imperatoris (Vaughan, 1919) from the Oligocene to early Miocene; (2) in M. trinitatis (Vaughan in Vaughan and Hoffmeister, 1926) from the middle Miocene to late Miocene/early Pliocene; and (3) in M. cavernosa (Linnaeus, 1767) from the late Miocene/early Pliocene to Recent. In such cases where more than one possible linkage appears viable, Ma- halanobis’ distances calculated using only the adjacent pair of intervals were used to establish linkage. Linkage reduced the 22 species to eleven (Text-fig. 17). These results suggest that two of the early Miocene types listed in Table 4 (i.e., groups 18 and 23) may be synonymous with Dominican Republic species. Group 18, which includes the holotype of Orbicella bainbridg- ensis Vaughan, 1919, appears linked with M. endoth- ecata (Duncan, 1863). Group 23, which includes top- otypes but not the holotype of Orbicella tampaensis Vaughan, 1919, appears linked with M. canalis (Vaughan, 1919). The results provide preliminary estimates of pat- terns of speciation and extinction within Caribbean Montastraea as well as amounts of directional phyletic evolution within each Montastraea species. However, due to small sample sizes, some estimates of species durations may be reduced. The patterns found suggest that all three Oligocene species (M. endothecata, M. imperatoris, and M. canalis) survived the late Oligo- cene extinction event (Frost, 1977), and two (M. en- dothecata and M. canalis) may have been abundant for more than twenty million years. During the early Miocene, two new species (M. tampaensis and M. trin- itatis) arose, one of which (M. tampaensis) became extinct shortly thereafter. During the middle Miocene, only one new species [M. ? anguillensis (Vaughan, 1919)] arose, but it soon became extinct. Within the late Miocene to early Pliocene (covered herein), four new species [M. limbata, M. brevis, M. cylindrica (Duncan, 1863), and M. cavernosa] arose, and three CANONICAL VARIABLE 2 CANONICAL VARIABLE 2 CANONICAL VARIABLE 2 LATE OLIGOCENE CARIBBEAN MONTASTRAEA | OSP1=imp | @SP2=can CANONICAL VARIABLE 1 EARLY MIOCENE CARIBBEAN MONTASTRAEA | OSP1=?tam | @SP2=imp | ASP3=tri | aSP4=end ; OSPS5=can -4 —12 -8 —4 (0) 4 8 WZ CANONICAL VARIABLE 1 MIDDLE MIOCENE CARIBBEAN MONTASTRAEA | OSP1=imp CANONICAL VARIABLE 1 Text-figure 16.—Montastraea. Canonical discriminant analyses distinguishing three Oligocene and nine Neogene Caribbean species. Plots of scores on the first two canonical variables showing polygons outlining the range of variation among colonies within each species. Oligocene species 3, early Miocene species 4, and middle Miocene species 2 were subsequently linked with NMB species 7 (= M. endothecata). Oligocene species 2, early Miocene species 5, and middle Miocene species 3 were subsequently linked with NMB species 4 (= M. canalis). Early Miocene species 3 and middle Miocene species 5 were subsequently linked with NMB species 2 (= M. trinitatis). “imp” = M. imperatoris; =canwe— M. canalis; “end” =M. endothecata; “tam” = M. tampaensis, “tri” = M. trinitatis; “ang” = M. anguillensis. DOMINICAN REPUBLIC NEOGENE. 11: BUDD 31 (M. limbata, M. brevis, and M. cylindrica) became ex- tinct shortly thereafter. Furthermore, two longer-rang- ing species (M. canalis and M. trinitatis) also became extinct at the end of the Neogene time interval rep- resented in the northern Dominican Republic. Only one new species [M. annularis (Ellis and Solander, 1786)] arose after the Pliocene. Thus, rapid periods of species diversification (adap- tive radiations) appear to have been rare in Caribbean Montastraea. Unlike Porites (see Foster, 1986), a few species were capable of surviving extinction episodes during the late Oligocene and Plio—Pleistocene. Some species had long durations (maximum = 20 Ma) with limited speciation and extinction within each interval. Maximum diversity occurred during the late Miocene/ early Pliocene. To test whether gradual directional change occurred within any of the longer-ranging species lines, canon- ical discriminant analysis was performed among the 11 unlinked species. The scores for the first two ca- nonical variables were then analyzed for non-random change through time in the four longer ranging lineages (Text-fig. 18). In this case, characters related to corallite size (e.g., CD, CLW, SLP, SLS) are most strongly CARIBBEAN (4) MIlo/PI Io correlated with canonical variable 1, and characters related to skeletal texture (e.g., CNP, STS, CST) are most strongly correlated with canonical variable 2. Re- sults of Duncan’s multiple range test indicate that di- rectional changes occur in these two variables through the sequence in M. imperatoris and in M. canalis, but not in M. trinitatis or M. endothecata. Mahalanobis’ distances between pairs of nearby populations of the two living Caribbean species were compared between earliest and latest occurrences for these two species, following the methods of Stanley and Yang (1987). These results indicate that the observed temporal change is less than or equal to environmental variation within the two species. Therefore, stasis prevailed. In Solenastrea, the material was sorted by locality, and species were first discriminated within each lo- cality region (Dominican Republic; Tamiami For- mation of Florida; Imperial Formation of California) using Cluster analysis and canonical discriminant anal- ysis as described in previous statistical techniques. The clusters from each locality were combined with those from other localities using Mahalanobis’ distances, and a final series of canonical discriminant analyses were run on the combined clusters. The final results are MONTASTRAEA Text-figure 17.—Montastraea. Network of shortest Mahalanobis’ distances between Caribbean species within stratigraphic levels and between adjacent levels. Double lines indicate species linked between levels. Numbers for each species are identification numbers previously used for that interval (see Text-fig. 16). 32 BULLETIN 338 shown in Text-figure 19. Two discrete species were found, and the characters SLS and —SLT were most heavily weighted on the discriminant function distin- guishing the two species. One species (S. bournoni) occurred in all three locality regions. The other was found only in the Tamiami Formation. Mahalanobis’ distances were then calculated be- tween localities in which the species co-occurred. The results show that distances were insignificant between populations from the Imperial Formation and Florida. A CARIBBEAN MONTASTRAEA 6 5 a 4 Himead joa) < 2 & St eal Zz ) S) Z = Zz —2 a | % BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 PLATE 22 red TS tee tal f DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 71 EXPLANATION OF PLATE 22 PION LUSLFACAITFENTL ALISA N{AUL IATA) bos hessicccree ae Oe oe eee en Ce See is i Rasen Cores, Seifeeslone fylSeFel lel een es SEOToU eH ss page 42 Whole colonies, colony fragments, and x-radiograph of a colony. Colony morphology ranges from relatively small knobs to large hemispherical mounds. Corallites are closely spaced, and are variably small to intermediate in size, sometimes elliptical in shape. Figure 1. Figured specimen. NMB D5754. Lower Miocene, locality NMB 17283, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Upper colony surface, x1. 2. Figured specimen. NMB D5737. Lower Miocene, locality NMB 16943, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Upper colony surface, x1. 3. Holotype, USNM 353657. Middle Miocene, locality USGS 8299, Manzanilla Formation, Trinidad. Surface of colony fragment, x1. 4. Figured specimen. NMB D5741. Lower Miocene, locality NMB 16945, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Upper colony surface, x 1/2. 5. Holotype of Madrepora annularis Ellis and Solander, a species closely related to M. trinitatis. Hunterian Museum, Glasgow. Recent, locality unknown. Upper colony surface, x1 (photo by Trevor Graham of the Hunterian Museum). 6. Figured specimen. NMB D5732. Lower Miocene, locality NMB 16943, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. X-radiograph, x 1. 72 BULLETIN 338 EXPLANATION OF PLATE 23 Montastraea trinitatis, (Vaughan) ..6.. ic. gs ciacrk oo os es ei PO ee ee Eee page 42 Close-ups of calical surfaces and transverse thin-sections. The septa are arranged in three to four septal cycles with the fourth cycle ranging from absent to completely developed. The costae are weakly developed, and are roughly equal in thickness. Figure 1. Holotype. USNM 353657. Same specimen as Plate 22, figure 3. Calical surface, x 10. Figured specimen. USNM 66852 (NF492). Miocene, locality USGS 8668, Baitoa, Dominican Republic. Calical surface, x 10. Figured specimen. NMB D5741. Same specimen as Plate 22, figure 4. Calical surface, x 10. Figured specimen. NMB D5732. Same specimen as Plate 22, figure 6. Calical surface, x 10. Figured specimen. NMB D5741. Same specimen as Plate 23, figure 3. Transverse thin-section, x 10. Possible synonym. USNM 353656 [hypotype of Heliastraea altissima Duncan of Vaughan in Vaughan and Hoffmeister, 1926]. Middle Miocene, locality USGS 8297, Tamana Formation, Trinidad. Transverse thin-section, x 10. Dw wt BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 PLATE 23 PLATE 24 BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 73 EXPLANATION OF PLATE 24 RI OMLASLFACAILFINILALISI NV AUS Tal) tans cio ac. ces) se Peas as AE A oe ots Os nee ee page 42 SEM photographs and transverse and longitudinal thin-sections. The columella generally is weak; however, strong, erect paliform teeth are developed in front of the first and second septal cycles. The endothecal and exothecal dissepiments and corallite wall are thick and well-developed. Figure 1. Figured specimen. NMB D5735. Lower Miocene, locality NMB 16943, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. SEM photograph of calical surface, x 10. 2. Figured specimen. NMB D5735. Same specimen as Plate 24, figure 1. SEM photograph of calical surface, x 20. 3. Figured specimen. NMB D5730. Lower Miocene, locality NMB 16943, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. SEM photograph of calical surface, x 10. 4. Possible synonym of Montastraea limbata (Duncan), which resembles M. trinitatis. USNM 353654 [holotype of Orbicella limbata var. pennyi Vaughan in Vaughan and Hoffmeister, 1926]. Lower Miocene, locality USGS 8298, Nariva Formation, Trinidad. Transverse thin-section, x 10. 5. Figured specimen. NMB D5730. Same specimen as Plate 24, figure 3. SEM photograph of longitudinal break, x 20. 6. Figured specimen. NMB D5729. Lower Miocene, locality NMB 16943, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Longitudinal thin-section, x15. 74 BULLETIN 338 EXPLANATION OF PLATE 25 Solenastrea bournoni Milne Edwards and Haime............---- +--+. page 44 Whole colonies and x-radiograph of a colony. Colony morphology is extremely variable, ranging from small, irregular knobs and spheres to larger, massive, hemispherical, and columnar mounds. Upward colony growth is highly regular. Corallites are relatively small and widely spaced. Figure 1. Figured specimen. NMB D5589. Upper Miocene, locality NMB 15851, Rio Gurabo, Gurabo Formation, Dominican Republic. Colony side, <1. . Figured specimen. NMB D5753. Lower Miocene, locality NMB 17283, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Side of a small irregular knob-shaped colony, * 1. 3. Synonym. BM(NH) R28871 [holotype of Plesiastraea globosa Duncan]. Neogene, “Silt of the Sandstone Plain”, Dominican Republic. Longitudinal cut through the growth axis of a colony, x1. 4. Figured specimen. NMB D5603. Lower Pliocene, locality NMB 16811, Rio Gurabo, Gurabo Formation, Dominican Republic. X-radiograph, x1. 5. Figured specimen. NMB D5641. Upper Miocene, locality NMB 16883, Rio Cana, Gurabo Formation, Dominican Republic. Upper colony surface, <1. 6. Figured specimen. USNM 66843 (NF488). ?Lower Pliocene, locality USGS 8734, Rio Mao, Gurabo Formation, Dominican Republic. Upper colony surface, x1. 7. Synonym. BM(NH) R28758 [holotype of Plesiastraea distans Duncan]. Neogene, “‘Nivajé Shale”, Dominican Republic. Upper colony surface, 1. 8. Figured specimen. NMB D5587. Upper Miocene, locality NMB 15850, Rio Gurabo, Gurabo Formation, Dominican Republic. Upper colony surface, x 1. Ne BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 PLATE 25 BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 PLATE 26 a e° aul * . a". oe WEA =) DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 75 EXPLANATION OF PLATE 26 Solenastrea bournoni Milne Edwards and Haime..................... 0000 ccce cece vce c eee e ev ecseces page 44 Close-ups of calical surfaces. The septa are arranged in three complete cycles with the first and second cycles extending to the columella. The third cycle is short, and usually is free, and the corallite wall is elevated. Figure il. 2. Figured specimen. BM(NH) R28871. Same specimen as Plate 25, figure 3. Calical surface, x 10. 3) 4. Figured specimen. NMB D5572. Upper Miocene, locality NMB 15845, Rio Gurabo, Gurabo Formation, Dominican Republic. wn Holotype. MNHNP 794. Recent, locality unknown. Calical surface, x 5. Figured specimen. BM(NH) R28758. Same specimen as Plate 25, figure 7. Calical surface, x 10. Calical surface, = 10. . Figured specimen. USNM 66843 (NF488). Same specimen as Plate 25, figure 6. Calical surface, x 10. Figured specimen. NMB D5589. Same specimen as Plate 25, figure 1. Calical surface, = 10. 76 BULLETIN 338 EXPLANATION OF PLATE 27 Solenastrea bournoni Milne Edwards and Haime...........-....---. 222s eee eee eect e eee ee tee page 44 SEM photographs and thin-sections. The columella is porous and wide with often well-developed paliform lobes in front of the first and second cycles. The corallite wall and exothecal dissepiments are thin. Figure Ie N Figured specimen. USNM 86910. Miocene, locality TU 1442, Rio Yaque del Norte, Dominican Republic. SEM photograph of calical surface, x 10. Figured specimen. NMB D5584. Upper Miocene, locality NMB 15850, Rio Gurabo, Gurabo Formation, Dominican Republic. Transverse thin-section, x10. Figured specimen. NMB D5641. Same specimen as Plate 25, figure 5. SEM photograph of calical surface, x 15. . Figured specimen. NMB D5794. Lower Pliocene, locality NMB 15822, Rio Gurabo, Mao Formation, Dominican Republic. Transverse thin-section, x10. . Figured specimen. USNM 86911. Lower Pliocene, locality TU 1344, Rio Gurabo, Mao Formation, Dominican Republic. SEM photograph of a longitudinal break, x 25. . Figured specimen. NMB D5584. Same specimen as Plate 27, figure 2. Longitudinal thin-section, x 10. PLATE 27 BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 PLATE 28 = DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 77 EXPLANATION OF PLATE 28 ISDICHASTFEALNV AGES (ID ANA rs acres ore ees ee Eee Pa aeons HEP TE eats as cia Peg eray esmeve oe eueayeneicee Seneutyre page 45 Whole colonies, x-radiograph, and close-ups of calical surfaces. Colonies generally are encrusting knobs or moderate-sized hemispherical mounds. Upward colony growth is highly irregular. Corallites are relatively large, and are variably spaced. Figure 1. Figured specimen. YPM 1586 [Verrill’s (1901) specimen of S. hyades (Dana)]. Recent, St. Thomas. Colony side, x '/2. 2. Figured specimen. NMB D5727. Upper Miocene, locality NMB 16939, Rio Yaque del Norte, Baitoa Formation, Dominican Republic. Upper surface of a knob-shaped colony, x1. 3. Synonym. YPM 1727 [holotype of Astraea excelsa Dana]. Recent, West Indies. Side of a columnar-shaped colony, * | 4. Figured specimen. NMB D5627. Upper Miocene, locality NMB 16853, Rio Cana, Cercado Formation, Dominican Republic. X- radiograph, x1. 5. Figured specimen. YPM 1586. Same specimen as Plate 28, figure 1. Calical surface, x5. 6. Synonym. UF 8291 [holotype of Montastrea peninsularis Weisbord]. Lower Miocene, Tampa Formation, Ballast Point, Florida. Calical surface, x 10. 78 BULLETIN 338 EXPLANATION OF PLATE 29 Solenastrea:hyades: (Dana) «a. ossy-sg ser Wiser a Oe a Se page 45 SEM photographs and thin-sections. The columella is thin and less porous, with few or no paliform lobes. The third septal cycle is relatively long, and it fuses with the second septal cycle. The corallite wall and exothecal dissepiments are relatively thick. Figure 1. Figured specimen. USNM 36662 (NF499). Recent, 3-7 m depth, Cedar Key, Florida. SEM photograph of a calice, x 20. Figured specimen. USNM 36662 (NF499). Same specimen as Plate 29, figure 1. Transverse thin-section, x 10. Figured specimen. NMB D5627. Same specimen as Plate 28, figure 4. SEM photograph of a calice, x 20. Figured specimen. NMB D5627. Same specimen as Plate 29, figure 3. Transverse thin-section, x 10. Figured specimen. NMB D5627. Same specimen as Plate 29, figure 3. SEM photograph of a longitudinal break, x 20. Figured specimen. USNM 36662 (NF499). Same specimen as Plate 29, figure 1. Longitudinal thin-section, x 10. AwRwWN PLATE 29 BULLETINS OF AMERICAN PALEONTOLOGY, VOLUME 101 iY, a “-_ a2 7 om x hin : — eet ‘ wy a, pe Gt hom pio a £2. “a DOMINICAN REPUBLIC NEOGENE. | 1: BUDD 719 INDEX Note: Page numbers are in light face; plate numbers are in bold face type; the page numbers on which principal discussions occur are in italics. PRT CAN eas cre coacs sasbeuca av ecws sabedeswsteves cuccusrccdeeesteeemereseenes 39 Agathiphyllia Reuss, 1864 ...............-..00000eeeee ees 16,17,34,35,50 antiguensis (Duncan, 1863) .............2:.:+sese00+ 17,35,40,41,50 MELA Waighan wl 99) it sccescreccesscees- cece vse so-scenne=nce @ 1 a PREPARATION OF MANUSCRIPTS Bulletins of American Paleontology usually comprises two or more sep- arate monographs in two volumes each year. This series is a publication outlet for significant longer paleontological monographs for which high quality photo- graphic illustrations and the large quarto format are a requisite. Manuscripts submitted for publication in this monograph series must be typewritten, and double-spaced throughout (including direct quotations and ref- erences). All manuscripts should contain a table of contents, lists of text-figures and (or) tables, and a short, informative abstract that includes names of all new taxa. Format should follow that of recent numbers in the series. All measurements must be stated in the metric system, alone or in addition to the English system equivalent. The maximum dimensions for photographic plates are 178 mm x 229 mm (7” x 9”: outlined on this page). Single-page text-figures should be drafted for reproduction as single column (82 mm; 3'4”) or full page (178 mm; 7”) width, but arrangements can be made to publish text-figures that must be larger. Any lettering in illustrations should follow the recommendations of Collinson (1962). Authors must provide three (3) copies of the text and accompanying illus- trative material. The text and line-drawings may be reproduced xerographically, but glossy prints at publication scale must be supplied for all half-tone illustrations and photographic plates. These prints should be identified clearly on the back. All dated text-citations must be referenced. Additional references may be listed separately if their importance can be demonstrated by a short general com- ment, or individual annotations. Referenced publication titles must be spelled out in their entirety. Citations of illustrations within the monograph bear initial capitals (e.g., Plate, Text-figure), but citations of illustrations in other articles appear in lower-case letters (e.g., plate, text-figure). Original plate photomounts should have oversize cardboard backing and strong tracing paper overlays. These photomounts should be retained by the author until the manuscript has been formally accepted for publication. Explanations of text-figures should be interleaved on separate numbered pages within the text, and the approximate position of the text-figure in the text should be indicated. Explanations of plates follow the Bibliography. Authors are requested to enclose $10 with each manuscript submitted, to cover costs of postage during the review process. Collinson, J. 1962. Size of lettering for text-figures. Journal of Paleontology, vol. 36, p. 1402. paATE DUE pemco INC 38-2031 HUAN eine ddaire RHE Wa ar ii tiAd renbiats bate eat a itagstabelh aah Fee jeans chute puri Wphagteey rere eH a edt Hays vires : i eset 7 3 ¥ rat pine y Psat V Rests sereee aaa ae vgte wnatlatiit d er hy eaten ee rtsereee eat tee Cini Aaa