ye Pre et 4 as ; * Rearerertlic gt F 4 Fed a Thonn Aal PTY * Ae ESS Aon Fo a ors : Skeet sin atee : , Wate. Re CNS A Wi Weer at ¥, if RE i z ath Brae ere Angin 88 M8 Hitt wb 4s Pie Le acer earned ee wr cteor . Y Share Wesel Ste AO eS aie 8278s oe Noa areatigas tee tye Safi Wad eet Oe ee er Aerie eens tas ’ eek ah, Ros ah ea RR Are ha A808, Sea On yall Aan aS AK he tes: yA ADS Andee afte ew Bye Lakes crm 4 7 Pa see Ri ees, we phea ee! per senett SHUT ye arse ee aroptprenpeie. etree ean rasa . ey a! if . ~ Le gta iy tee wees pea ; Seema ei eg ed Bete on" pecan one 5 ied Be pepe 5 : aT DP ia das seeate nasi vera eyes Sern eat fet Ee Mad ae we er fibres! sate pry eats siytcea Ligeti a ma yt ay Deatitedtt Pore Si ep 7 when ee eer tt rr a ae Pit oe re oR tee fete eprqeeare Ce hee ae (ears en Fe oT eh) ’ : he on at vet vd o a red ey 4 Mt Siete eNO i Nata eta 6 Ur oy au My og atte a ow eet ‘i ‘ yes + wey an ea Lag) Bat D DO k, on) NVINOSHL SMITHSON S321NVNUGIT LIBRARIES SMITHSONIAN RARIES INSTITUTION NOILNLILSNI be Ea ENV EN Fer OWIVIET ETT OOU/IY NOILNLILSNI INSTITUTION NOILNLILSNI we t Ww Uw GE INSTITUTION SAIYVUGIT_LIBRARIES NVINOSHLINS S3iYVYUSITLIB SMITHSONIAN Oe POC WE iat aN. INSTITUTION NOILNLILSNI LIBRARIES ‘SZIUVYEIT LIBRARIES INSTITUTION NOILNLILSNI PEOVaPes ewes owewv INSTITUTION, NOILNLILSNI NVINOSHLINS S31YVUgaly SMITHSONIAN SJIYVYGIT LIBRARIES SMITHSONIAN wre Ss. RARIES “Z Y/ NVINOSHLIWS » NOILNLILSNI INSTITUTION NOILNLILSNI INSTITUTION LIBRARIES SMITHSONIAN NVINOSHLINS S3IYVYSIT LIB SMITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLIWS LIBRARIES NOILNLILSNI SJINVUGIT LIBRARIES SMITHSONIAN 4 i = 30) SOLE SELON ION a. NVINOSHLI WS SMITHSONIAN NVINOSHLINS SMITHSONIAN NVINOSHLINS S3IUVUGII LIBRARIE 817 LIBRARIES SMITHSONIAN NVINOSH1 SMITHSON NVINOSH1 NVINOSHL SMITHSON e INSTITUTION NOILNLILSNI NVINOSHLINS S31YVYdIT LIBRARIES SI SMITHSONIAN or RARIES LIBRARIES SaJiuvugit LIBRARIES INSTITUTION NOILNLILSNI_N' 817 LIBRARIES SMITHSON 2 saiuvugi INSTITUTION NOILNLILSNI S3IyVvy INSTITUTION NOILNLILSNI S3IYVYUSITLIB NOILNLILSNI_ NVINOSHLINS SI Ssaluvugit_LIBRARIES INSTITUTION x NS NY saluvugiq ee ee SMITHSONIAN NVINOSHLIWS SMITHSONIAN NVINOSHLIWS SMITHSONIAN NVINOSHLINS LIBRARIES SMITHSONIAN INSTITUTION NOILN N' S NOILMLILSNI NOILALILSNI INSTITUTION NOILNLILSNI NVINOSHLINS S3IYVUEIT LIBRARIES SI INSTITUTION NOILNLILSNI INSTITUTION INSTITUTION saiuvyal SMITHSONIAN NOILNLILSNI_ NI tg INSTITUTION LIB a ne, NOILNLILSNI VINOSHLINS S3JIYVYUEIT LIBRARIES NVINOSHIINS S3Iuvudi? LIBRARIES SMITHSONIAN N NVINOSHLINS S3IYVYGIT LIBRARIES SN ES SMITHSONIAN LIBRARIES SMITHSONIAN we LIBRARIES NOILNLILSNI Saluvugil LIBRARIES INSTITUTION NOILNLILSNI SMITHSONIAN NX 17 a we lyvug INSTITUTION NOILNLILSNI INSTITUTION NOILNLILSNI SAIYvVYsITLIBRARI Saiuvugi INSTITUTION S $3 INSTITUTION NOILNLILSNI_ NVINOSHLINS S31uvugi7_LIBRARIES Sh ey ~ e ‘’ > INSTITUTION NOILNLILSNI NVINOSHLINS SMITHSON IAN NVINOSHLIW SMITHSONIAN NYINOSHLIWS LIBRARIES SMITHSONIAN Nt NOILNLILSNI NOILNLILSNI INSTITUTION NOILNLILSNI PDLI NVINOSHLINS S3I1Y¥VYUSIT LIBRARIES SLaOSOND. J17 LIBRARIES A171 LIBRARIES ION = NOILALILSNI ; 1ON JON ~ViSON; (Saran 44 = 3 \ ci = < = < = as Ks g : 4 = = f ood ; \\ ay EQ 2 3 2 5 2 ff Ug 5 WX w y a : iN 8 ”n ine 7) 72) hy ” WK . z = z = a = g Wy = WS 2 a » 2 a ae ’ a = a ‘ MIAN _INSTITUTION NOILNLILSNI_NVINOSHLIWS Sa IyYVYdiT_ LIBRAR! ES) SMITHSONIAN INSTITUTION NOILMLILSN 40 uw @ WX » é = +o e n = ba fa ul a = & = & co = oo ee ox = oc a << = < en < = are = Woe 3 o = o a 2 = Y mw = (0 . = ire = Pee) = & * = ~ e Be” ts 2 WAN INSTITUTION NOILNLILSNI NVINOSHLINS S3SIYVYSEIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSN a = - z iS z - z \. o2 _ o = ow = Y 20 = | 2 eo) YE % > 3] "3 : = > : LY = : \ a = m at m = Vif m = \ — ” 5 = o — 1o7) = IWS Sa 1uvyua Mul BRARI See SO NIAN 3 ca NOUMEALITSNI NVINOSHLINS (54 lyuvud rout BRARIES : z : 2 My Si iN a4 NOE Ne a aN : : g : Sy, © 1 aN VIAN __INSTITUTION NOILNLILSNI NVINOSHLINS S3IYVYSGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSN | ee S ” = 7) = ” a” eo ” s n” fe aW wi = = = o = oc. = re e = = < a <3; = < 4 S = = 4 = = re = S = rs) a ,; ro) ze ma at Zz ees 2 =; if Zz =f IWS Sa IyYvug Pee BRARI Boe SMITHSONIAN INSTITUTION _NOLLMLILSNINVINOSHLINS _ $3 luVvud bans LIBRARIE ~~ S = o ys o = ° = 5 ra = ra 5 2 5 » = > = 4, > fe > ke > Ee 3) = a?) = a = p°) no = * = = w” res 2 is Z a Z Z Z a MAN INSTITUTION NOILALILSNI NVINOSHLINS S31YVYaIT_LIBRARIES SMITHSONIAN INSTITUTION NOILALILSN if < 2 ‘ Ke AN 7) z 7) z n z wo = = =< = = = < = =z X ar = 4 z ‘ + Zo N\ A a Ai 0: : gs Na: 5 Nw: =e NS 2 = Zz, = YS 2 fe NS =e oS : : Se ene ae aes INS S3IYVYEIT_LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI_ NVINOSHLINS S3IYVYeEIT LIBRARIE: w Zz sd Zz Xt Z ef z —_ yn ” ” = ” oc = = ys = o = a a 4 K = = < a < pi = c y \Y = e oO = cm za fee} = RASS ra) = - ° aa re) = °o SS = re) - za 24 a a Ta 4) z WNIAN INSTITUTION NOILNLILSNI NVINOSHLINS S3!1YVYUEIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSN MmAUTiLUS Supplement 2 (Issued with Volume 108) Molecular Techniques and Molluscan Phylogeny Proceedings of a Symposium held at the Eleventh International Malacological Congress Siena, Italy 31 August-5 September, 1992 Edited by M. G. Harasewych Department of Invertebrate Zoology National Museum of Natural History Smithsonian Institution Washington, DC 20560 USA and Simon Tillier Laboratoire de Malacologie Muséum national d’Histoire naturelle 55, rue Buffon F-75005 Paris, FRANCE 1994 ISSN 0028-1344 Copyright 1994 by THE NAUTILUS All rights reserved. No part of this publication may be reproduced by any means without prior permission of the copyright owner. Available from: THE NAUTILUS P.O. Box 7279 Silver Spring, MD 20907 USA per > aan Supplement 2 (Issued with Volume 108) August 11, 1994 ISSN 0028-1344 CONTENTS MOLECULAR TECHNIQUES AND MOLLUSCAN PHYLOGENY PREUACE, 5 ooh ss Saale SO a ey Ce ee MS ea RO eee a ne Fie Fa an ma 1 PLENARY LECTURE George M. Davis Molecular Genetics and Taxonomic Discrimination....................... 3 D. J. Colgan The Evolutionary Consequences of Restrictions on Gene W. F. Ponder low-E xamplesmnomprly drobiidesnailsia aes nen ne een eee 25 Kenneth C. Emberton Allozyme Cladistics in Malacology: Why and How?..................... 44 S. Laura Adamkewicz Use of Random Amplified Polymorphic DNA (RAPD) M. G. Harasewych Markers to Assess Relationships Among Beach Clams of the (CemUSOD ONC ciara asap eas ree a a aay, Pel GUE a A At tsb dl Jeffrey L. Boore Mitochondrial Genomes and the Phylogeny of Mollusks.................. 61 Wesley M. Brown Jonathan Terrett The Mitochondrial Genome of Cepaea nemoralis Sue Miles (Gastropoda: Stylommatophora): Gene Order, Base Richard H. Thomas Composition and Heteroplammy .................... ee ee 79 Douglas I. Cook The Highly Variable and Highly Mutable Mitochondrial Eleftherios Zouros DNA Molecule of the Deep Sea Scallop Placopecten MEP CULOTULCUS POT tater AE oa. oh oc 85 Elaine Rumbak David G. Reid Richard H. Thomas Birgitta Winnepenninckx Thierry Backeljau Rupert De Wachter Gary Rosenberg George M. Davis Gerald S. Kuncio M. G. Harasewych Simon Tillier Monique Masselot Jean Guerdoux Annie Tillier Jonathan B. Geller Dennis A. Powers Maria Lazaridou- Dimitriadou Y. Karakousis A. Staikou Thierry Backeljau Karin Breugelmans Herwig Leirs Teresa Rodriguez Dimitri Sherbakov Tatyana Sitnikova Jean-Marie Timmermans Jackie L. Van Goethem Erik Verheyen Reconstruction of Phylogeny of 11 Species of Littorina (Gastropoda: Littorinidae) using Mitochondrial DNA Sequence Daltaw \.o WA 5 Bene ee ee ee re Small Ribosomal Subunit RNA and the Phylogeny of the Moolltsea:c- ier cto sen een EN eh ae nec a pee ld oe oe TR Preliminary Ribosomal RNA Phylogeny of Gastropod and UnionoideantBivalve NMollusksiey eee ie naennnnne Monophyly of Major Gastropod Taxa Tested from Partial 28S rRNA Sequences, with Emphasis on Euthyneura and Hot-Vent Limpets Peltospiroidea............................. Site-Directed Mutagenesis with the Polymerase Chain Reaction for Identification of Sibling Species of Mytilus.......... PRESENTED AS A POSTER Morphological and Genetic Variation in Greek Populations of the Edible Snail Helix aspersa Miller, 1774 (Gastropoda: Pulmonata): A Preliminary Survey ................ PRESENTED AS A POSTER Application of Isoelectric Focusing in Molluscan Systeinaties’\: steer eters ohne sew ce ae metas hore ete Rene ree THE NAUTILUS, Supplement 2:1-2, 1994 Preface Page 1 With its tremendous diversity and excellent fossil rec- ord extending from the earliest Cambrian to the present, the phylum Mollusca presents extraordinary opportu- nities and challenges to students of all aspects of evolu- tionary biology. The confluence of new techniques of data acquisition (among them scanning and transmission electron microscopy and nucleic acid sequencing) and data analysis (most notably cladistic methodology) has prompted many advances in studies of molluscan phy- logeny at all levels over the past decade. Yet despite the use of varied data sets and techniques, many of the most basic questions at the highest taxonomic ranks remain unresolved. While mollusks, particularly Cepaea, Ceri- on, and Partula, have played important roles in the stud- ies of population genetics and speciation, the phylum remains underrepresented in studies at higher taxonomic levels. Happily, this situation is changing, and increased research in molecular malacology is evidenced by the appearance of the “Mollusc Molecular News’ and bul- letin board on the internet. The papers comprising this volume were presented at the Symposium on Molecular Techniques and Molluscan Phylogeny that was convened during the Eleventh In- ternational Malacological Congress, held in Siena, Italy, in August-September of 1992. This collection of works contains the majority of the papers and posters presented during this Symposium, and represents a overview of contemporary research in this rapidly growing area of systematic malacology. Although a symposium on a sim- ilar topic was held during the American Malacological Union meetings in 1987, the results presented at that time were judged by the participants to be too prelim- inary to warrant publication. We are heartened that the majority of the contributors to the present symposium felt that their work was sufficiently advanced to be pub- lished. We thank Prof. Folco Giusti for the invitation to or- ganize this symposium, and for his assistance in coping with the many attendant details. We extend our appre- ciation to all participants in the Symposium, and rec- ognize the special efforts that several colleagues have made to attend these meetings. The contributions of the numerous co-authors, many of whom were not present at the Symposium but nevertheless contributed to its success, are gratefully acknowledged. In the Plenary Lecture, which opened the Congress and set the stage for the Symposium, George Davis set forth his views on the utility of molecular genetics for molluscan systematics. Presenting case studies primarily from his own work, he documented the value of molec- ular data as well as the potential pitfalls in incautious interpretation of molecular data. The organization of the Symposium into three sessions, each covering a broad topic, is reflected in the arrange- ment of papers in this volume. The first group of papers documents the use of population genetics techniques for studies of speciation or phylogenetic inference. Colgan and Ponder present a sophisticated analysis of the varied effects of restrictions on gene flow on speciation using allozyme data. Emberton reviews and compares methods for constructing phylogenies using allozyme data, while Adamkewicz and Harasewych explore the utility of RAPD techniques, recently developed for differentiating pop- ulations and strains, for the inference of phylogenetic relationships among closely related species. The second group of papers focuses on the mitochon- drial genome and explores the utility of mitochondrial DNA (mtDNA) for investigating a broad range of genetic relationships. Boore and Brown review the varying and contradictory hypotheses regarding the relationships of the phylum Mollusca and its classes, and advocate the use of the arrangement of genes (gene order) of the mitochondrial genome as phylogenetic characters. Ter- rett and colleagues present new data on the gene order of Cepaea nemoralis and compare it with those of other metazoans. At the other extreme, Cook and Zouros an- alyze inheritance patterns of variations in the size of the mitochondrial genome among sibling scallops, and con- clude that because of its rapid turnover, such size vari- ation does not provide information useful for taxonomic studies. In the last paper of this group, Rumbak and colleagues use the sequence of a portion of the gene for the small ribosomal RNA to study the relationships of 13 species of littorinids. The third group contains papers on the use of ribo- somal sequence data to resolve phylogenetic relationships among mollusks. Winnepeninckx and associates present the complete sequence and structure of the 18S rRNA of Onchidella celtica, a pulmonate snail, and compare it to 25 other known metazoan sequences to assess the monophyly and relationships of three molluscan classes as well as the relationships of Mollusca among the Meta- zoa. Preliminary analyses of sequences derived from the D6 loop of the 28S rRNA of 43 gastropod and bivalve species reported by Rosenberg and collaborators reveal variation in the rates of sequence divergence but do not refute morphology-based phylogenies. A group led by S. Tillier present the most broadly represented sequence- Page 2 based phylogeny of Gastropoda to date. Geller and Pow- ers use site-directed mutagenesis to descriminate be- tween sibling species of Mytilus based on a single base difference in a region of their 16S ribosomal gene. Two additional papers, originally presented as posters, are included. One (Lazaridou-Dimitriadou and col- leagues) reports on allozyme variation in Greek popu- lations of Helix aspersa, the other (Backeljau and nu- merous collaborators) reviews the use of isoelectric focusing in molluscan systematics. Initial attempts to investigate the origins and early evolution of the Mollusca by the use of molecular data have met with only limited success, and in the process have questioned some of the most basic precepts of mor- THE NAUTILUS, Supplement 2 phology-based classification. Perhaps most notable of these is the growing body of evidence from both nuclear and mitochondrial genomes that places Bivalvia as the out- group to a clade containing the Gastropoda and Poly- placophora. Clearly, the addition of taxa, especially from presently unrepresented molluscan classes, would be of great value. The growing body of molecular data, together with increasingly sophisticated and better reasoned methods of analysis, portend great advances in our understanding of molluscan evolution in the coming years. M. G. Harasewych Simon Tillier THE NAUTILUS, Supplement 2:3-23, 1994 Page 3 Molecular Genetics and Taxonomic Discrimination George M. Davis Pilsbry Chair of Malacology The Academy of Natural Sciences 1900 Benjamin Franklin Parkway Philadelphia, PA 19108, USA a a a PROLOGUE The following paper is based on the Plenary address I gave before the 11th UNITAS Congress, the International Congress of Malacologists, held in Siena, Italy 28 August-7 September 1992. The address is dedicated to Professor Foleo Giusti, a friend, colleague, and scholar dedicated to excellence in sys- tematics who, with G. Manganelli (1992), wrote concerning the discrimination of species: “The good systematist is not one without doubts or the one who always succeeds in defining a phenomenon or recognizing a ‘species’. It is rather the one who studies the phenomenon trying to understand it in all its facets and who is not above admitting that its exact nature escapes him ...Let us stress again that only a little humility and a little consciousness are required. ” INTRODUCTION If we could call back to the present some of the early fathers of morphology-based malacology, for example Cuvier, Bouvier, Troschel, Stimpson, Pelseneer, Thiele, Johansson, and Pilsbry to name a few, and bring them up to speed on the vast accumulation of literature since their time, they would readily understand and be en- thusiastic about the modern day potential for sophisti- cation in taxonomic discrimination. They would say that it is about the recognition of, and the definition of species, genera, and higher taxa. They would be in agreement, and I with them, that the fundamental basis for taxo- nomic discrimination was then, and is today, the com- parative anatomical data set. Unfortunately, over the past several decades, detailed comparative anatomy has been the most under-used tool in molluscan systematics. However, given the recent ac- ceptance of cladistic methodologies (in malacology, only in the past 6 to 7 years), one sees a return to comparative anatomy. The hunt for unique anatomical characters and character-states in order to nest taxa in sets based on synapomorphies, is gaining increased respectability. Growing awareness that there is a need for well defined qualitative anatomical characters and their states, which serve to demonstrate differences among taxa in order to construct hypotheses of evolved relationships (phyloge- nies), should stimulate modern anatomical work on all groups of mollusks. Molecular techniques have long been used as an aid for discriminating among taxa. However, as with cla- distic tools, malacologists have lagged far behind micro- biologists, mammalogists, and herpetologists in applying them. The use of immunology in systematics is over four decades old; the use of allozymes, three decades. The now-generation is scrambling to sequence RNA and DNA aided by PCR and cloning. The use of allozymes in molluscan systematics is now well established and will not be supplanted by genomic techniques for years to come. The quantities of useful information that can be gained through allozyme elec- trophoresis are enormous and can be obtained at rela- tively little cost compared to the considerable expenses involved in pursuing sequence work. Allozymes are es- pecially useful in comparing closely related genera, spe- cies within a genus, and sorting out species-level prob- lems. Allozyme electrophoresis is an ideal tool for population genetics as applied to delineating species. DNA-RNA sequencing is in its infancy, literally explod- ing in dimensions of use, problems, and surprises. Given 600 million years of spectacular molluscan evolution, one can be sure that molluscan DNA from more than 100,000 living species from seven classes will yield numerous surprises and cause researchers to have many a migraine headache. A major concern with molecular data is the analysis of data. It is generally agreed today that there is as yet no truly satisfactory way to analyze such data to ade- quately portray relationships (exhaustively reviewed by Buth, 1984 for allozymes and by Swofford & Olsen, 1990 for sequence data). There is a dichotomy of approach: cladistic and phenetic. In a phenetic mode, the usual approach is a UPGMA treatment of distance data based either on allele frequencies (allozymes) or sequence dif- ferences. A phenogram is the standard presentation. Cla- distic analysis requires a unit character. With allozymes, using the locus as a character seems to be the best at present, with different allele combinations scored as char- acter-states. With sequence data, the unit character sug- Page 4 gested is the gene. Further, genes must be calibrated for the taxa studied relative to rates of evolution if credible phylogenies are to be structured. Phylogenetic hypoth- eses are based on cladograms derived from computer programs such as PAUP (Swofford, 1983) or HENNIG- 86 (Farris, 1988) that use parsimony criteria to obtain the shortest possible tree. However, there is a universal call by those involved in the evolution of bacteria, viruses, and protists as well as other groups of taxa for research to provide better modes of analyses of molecular data to make better trees (Davis, 1994). One rarely sees both phenetic and cladistic analyses together in the same paper, especially when the database is an anatomical one. I strongly urge that both be used and the results examined for congruence. In the phenetic mode, combining principal component analysis (PCA) and multidimensional scaling (MDS) with ordination di- agrams usually yields considerably better results than simply using similarity or distance coefficients to do a UPGMA structured phenogram. The benefits are: 1) As one moves from the initial phenogram to PCA to MDS, a tree of relationships between each taxon (spe- cies, individual, etc.) such as a Prim Network, usually becomes shorter and the cophenetic correlation with the original matrix increases indicating that the result better portrays relationships among the taxa (phenetic parsi- mony, Davis et al., 1994a). 2) Ordination diagrams of individuals or species or genera in 1 X 2, 1 X 8, 2 X 8 dimensional space with taxa connected by a Prim Network provides a consid- erably improved understanding of taxon interrelation- ships both as to scale of divergence and direction of divergence of the taxa. Ordination of taxa in n-dimen- sional space removes the constraints of the one dimen- sional phenogram. 3) The PCA allows one to assess character correlations that are the basis for the distribution of taxa along each dimension of n-dimensional space. The utility of these multivariate techniques will be highlighted later in this paper. It has been a decade and a half since I reviewed ex- perimental methods in molluscan taxonomy (Davis, 1979b). While it was possible then to review nearly the entire literature pertaining to molluscan molecular sys- tematics from amino acid work, through immunology and protein electrophoresis, it would be counter-pro- ductive to attempt to do so now. Instead, I will give my views on the utility of molecular genetics today for mol- luscan systematics. In outline form below are presented the topics I will discuss as to the prominent uses of mo- lecular genetics: I: To uncover cryptic species. II: For population genetics: A: To detect and study hybridization. B: To examine special selective pressures. C: To study breeding structure. III: To study patterns of evolution: A: Speciation. B: Phylogeny. THE NAUTILUS, Supplement 2 IV: To uncover unique aspects in the evolutionary process. In presenting case studies, primarily from my own experiences with these issues, there are several key and central issues to focus on and always keep in mind: 1: There is no universal molecular clock, and taxa therefore must be calibrated relative to genetic distance. Different clades may have evolved at different rates. 2: Genetic distances do not define species or higher taxa. 3: Species concepts are important relative to inter- preting genetic distance data. 4: Definitions of taxa and discrimination among taxa fundamentally require anatomical ground-plan data in- cluding developmental and cytological data, not genetic distances. 5: When morphological data yield few characters and character-states that enable discrimination among taxa, the need for molecular genetic data increases. CLOCKS AND CALIBRATION It is not the purpose of this paper to exhaustively review all that has been written on this topic. It is useful to abstract five key points derived from the reviews of Tem- pleton (1980, 1981), Barton (1989), Harrison (1991) and the diverse publications of Gillespie (1984, 1986a,b; 1987) that focus on the issues of molecular clocks. 1) Rates of molecular evolution are more variable than expectations based on a simple Poisson mutation process. 2) Protein - DNA evolution rates are not consistent with neutral theory. 3) Molecular evolution is rate-variable and epi- sodic, and, in the view of some, best explained by in- voking natural selection ( Nevo & Beiles, 1988; Skibinski & Ward, 1982; Murray et al. 1991). 4) Any attempt to apply a molecular clock to comparisons of even closely related species is hazardous. 5) There is no simple pattern between mode of speciation and genetic distance. With due respect, one must consider the serious ar- guments of those supporting neutral mutation theory (refer to numerous papers by Nei and Kimura reviewed in Nei & Graur,1984) employing statistical tests of data derived from allozymic data providing heterozygosity or gene diversity where effective population size and mu- tation rates are known. But there is so much more in- volved in considering rates of evolution; one example is provided. Consider the rapid duplication and loss of genes coding for the alpha chains of hemoglobin where dif- ferences in rate are apparently associated with differ- ences in lengths of non-coding regions (Zimmer et dl., 1980). They consider that adaptive evolution may de- pend more on the type of genetic variability than on various point mutations that affect protein structure. That there is no universal clock, and that different clades may evolve at different rates based on molecular genetic data is clearly demonstrated in Figure 1, where regressions 1, 3, and 6 pertain to different groups of teleost fish ( from Hillis & Moritz,1990 based on data from Avise & Aquardo, 1982). For further elucidation G. M. Davis, 1994 18 1 2 wo = —> 12 @ (5) Cc ® oO ® 2 ke} @ (5) & (7) C1) 6 E = 3 4 5 6 0 7 0 0.5 1.0 Nei’s genetic distance D Figure 1. Regressions of changes of Nei’s genetic distance D through time for different clades. Clades 1,3,6 are for different teleost fishes. Redrawn from Hillis and Moritz (1990). on this topic, refer to Hillis and Moritz (1990) for similar plots for DNA, RNA, etc. To provide an example, a statistical analysis of DNA sequences from artiodactyls, rodents and primates show that there is no global mo- lecular clock in mammals. Rates of nucleotide substitu- tions in rodents are approximately four to eight times higher than in higher primates( Li et al., 1987) Regres- sions for different molluscan clades are similar to those shown in Figure 1, as demonstrated below. Genetic Distances, Their Utility and Calibration: Different measures of genetic distance have different utilities. When referring to genetic distance based on allozyme “alleles”, one must state whose formula was used. The traditional D is that of Nei (1972) although the Nei (1978) modification is now widely used; it rep- resents the accumulated number of codon substitutions per locus since time of divergence; it is a squared distance that rises with time. There is no upper limit, constraints being only the number of loci used and alleles found. It is non-metric. One of the beauties of this measure is that as taxa diverge more and more, the difference is not squeezed between 0 and 100%. Yes, there are genetic interpretation and statistical problems with values ex- ceeding 1.0, but aside from the problems of metricity, the cause for the rise above 1.0 is clear from the data. Sewell Wright (1978) considered Arc distance (Cavalli- Page 5 ARC-D Se ooo oo oe ie a NEI'S-D Figure 2. A comparison of Nei’s D with Arc distance based on data from my various papers. As Arc distance becomes com- pacted, Nei’s D continues to rise. As Nei’s D is compacted at lower values, Arc distance gives a better approximation of dif- ferentiation among closely related taxa. Sforza & Edwards, 1967) to be the best as it is a metric distance where the “coordinates in the hyperspace are the square root of the allele frequencies, a procedure that locates all populations on the surface of a hyper sphere with respect to a locus”. Best or not in the pure mathematical sense, both distances have their use. A plot of Nei vs. Arc D values is given in Figure 2 based on a variety of data I have published or accumulated. It is clear that Nei’s D values are compacted at the lower values (to the left) where closely related populations are compared. In this situation, Arc D gives a better under- standing of differentiation. At the other end of the plot, with the greater divergence of distantly related taxa, the Arc D values become compacted and less informative while Nei’s D values continue to rise thus providing a better understanding of divergence even though the val- ues do not rise linearly and are not metric. Nei’s D can be used to calibrate within-group variance, and gaps between nested sets of taxa that represent different tax- onomic levels. Nei’s D can be used for multidimensional scaling and with a Prim network to obtain a three di- mensional visual sculpture of relationships, patterns of divergence, and gaps between sets of taxa as long as one understands the constraints and limitations due to the non-metricity of the measure. When publishing results, I advocate providing both Nei’s and Arc distances. Cer- tainly the Nei’s D is needed because of the vast literature now built up using this measure. It has become a uni- versal standard for comparison. Beware of those who grind-run gels-publish without studying the patterns and processes of morphological diversification, life history diversification and ecological diversification throughout the clade of concern and in sister clades. There is so much more involved in under- standing species differentiation then examining a genetic distance and extrapolating taxonomic rank. This is well understood in vertebrate literature; see the review of Avise and Aquardo (1982). However, I have heard it Page 6 THE NAUTILUS, Supplement 2 Table 1. Calibrating clades for Nei’s D: selected examples. Mean + standard deviation. Population level Hydrobiidae (Davis et al., 1989) 6 populations Truncatellidae (Rosenberg, 1989) 4 populations 2 populations Pomatiopsidae (Woodruff et al., 1988) 7 populations Planorbidae (Mulvey et al., 1988) 6 populations Planorbidae (Bandoni et al., 1990) 12 populations Unionidae (Davis et al., 1981) 5 populations 11 populations Hydrobia truncata Truncatella pulchella Anodonta cataracta Elliptio complanata Truncatella caribaeensis Oncomelania hupensis quadrasi Biomphalaria glabrata Biomphalaria pfeifferi 0.008 + 0.005 0.029 + 0.020 0.067 0.039 + 0.038 0.103 + 0.068 0.053 + 0.066 0.034 + 0.038 0.065 + 0.039 Species: differences among species of a genus Truncatellidae (Rosenberg, 1989) 5 species Truncatella Succineidae (Hoagland & Davis, 1987) 2 species Novisuccinea 2 species Oxyloma Unionidae 7 species (Davis, 1981) Elliptio 3 species (Davis, 1983) Uniomerus 3 species (Kat, 1983a) Anodonata 6 species (Kat, 1983b) Lampsilis Sphaeriidae (Hornbach, 1980) 4 species Sphaerium 2.226 + 0.862 0.004 0.269 0.210 + 0.017 0.308 + 0.165 0.457 + 0.073 0.609 + 0.478 0.568 + 0.310 Genera: differences among genera of the same tribe Unionidae (Davis, 1981; Davis et al., 1981) Amblemini Pleurobemini 4 genera 3 genera 0.651 + 0.275 0.243 + 0.086 Subfamilies: differences among subfamilies of Unionidae (Davis, 1981; Davis et al., 1981) 3 subfamilies remarked by some malacologists newly come to the use of allozymes that the average Nei genetic distance(D) among populations of a species is < 0.09( or some such value), or that D = 0.20 or 0.30 indicates a species difference, especially if the populations involved are sep- arated by considerable distance (widely allopatric). Woodruff et al. (1988) have done this while applying an evolutionary species concept (e.g. Wiley, 1981) [or one could equally use a phylogenetic species concept (for a review, see Cracraft, 1989)] to argue that populations with a D of 0.60 are distinct species as they have their own evolutionary history and can be diagnosed on unique qualitative characters (e.g. unique alleles). Are they cor- rect? No, not on this evidence alone. There are cases of perfectly good species with D < 0.06 (Sene & Carson, 1977; Kirkpatrick & Selander, 1979; Davis et al., 1981) and cases with populations with a large D value not being 1.903 + 0.186 considered different species. Consider the complexity of assessing taxonomic meaning of D for mole rats of the superspecies complex of Spalax ehrenbergi, in Israel, where members differ by a D of 0.039 compared with a D of 0.234 between the two superspecies S. leucodon and S. ehrenbergi, indicating the recency of divergence in the later and considerable age of divergence from the former (Nevo, 1991). These issues will be revisited in this paper. Nei’s D is used to calibrate clades because one can compare among taxa from populations to subfamilies or families for the very reason that there is no upper limit for Nei’s D. In this context, refer to Table 1. At the population level, it is instructive to know that allopatric populations of Hydrobia truncata (Hydrobiidae), spread out along the coast of North America from New England to Maryland have a mean D of 0.008 while in species of G. M. Davis, 1994 a sister family (Truncatellidae), populations of the same species distributed from Florida through the West Indies differ by a mean D of 0.029 or more, i.e. four times as much difference as in the former case. But, in the pul- monate Planorbidae, widely allopatric populations of Biomphalaria glabrata differ by a mean D of 0.103 (Mul- vey et al., 1988)! Studies of 12 populations of Biom- phalaria pfeifferi widely distributed throughout Kenya revealed that two pairwise comparisons had Nei’s D > 0.240 (Bandoni et al., 1992) while the average D was 0.053. Genetic distances can point out anomalies, but do not explain them. Three uses in Table 1 illustrate: 1) the D among five species of “Truncatella” is 2.226, consider- ably more than between three subfamilies of Unionidae (1.903). The data suggest that more than one genus is involved. Is this correct? One needs to go back to the detailed comparative anatomy to look for marked changes in groundplan among species groups. Rosenberg (per- sonal communication) showed me that his anatomical data would indeed justify different generic rank given the types of differences that do occur among the taxa he studied relative to the types of differences that justify generic discrimination among taxa of sister rissoacean families (Hydrobiidae, Pomatiopsidae, etc.). He awaits data for more species before naming genera. 2). In the pulmonate family Succineidae, in one instance perfectly distinct species differ by a D of 0.004, while in another the difference is 0.269. The former should be populations according to some who follow a D formula to assign taxonomic rank, while the latter data do suggest different species. I will revisit this case later. 3) In the freshwater bivalve Elliptio, the difference among species is low, 0.210 while in another clade (different tribe) the differ- ence among Lampsilis species is three times greater, i.e. 0.609. At higher taxonomic levels, it is clear that differences seen among subfamilies of Unionidae that go back at least to the Cretaceous accumulated slowly while much more recently evolved “genera” of Truncatellidae ( given the ages of the West Indian islands for these terrestrial taxa) differentiated genetically at a considerably greater rate. THE UTILITY OF MOLECULAR GENETICS I: CRYPTIC SPECIES AND CONVERGENCE The following example abstracted from Davis (1983) serves four purposes: 1) It demonstrates the power of molecular genetics to uncover cryptic species. 2) It clear- ly demonstrates nature’s perversity in obscuring genetic diversity under the cover of convergent morphologies. 3) It shows that when too few morphological character- states serve to distinguish generic groupings of species, molecular genetic data may be decisive. 4) It underscores the fact that shells provide the least valuable data for elucidating relationships among higher taxa (genera, tribes, etc.). Page 7 The North American freshwater clam genus Uniom- erus was long considered to have only one species, U. tetralasmus (Johnson, 1970). The genus is barely dis- cernible from Elliptio. The shells of both genera are highly convergent for most species (differing only in beak sculpture that is mostly eroded away at an early stage of growth), and the only anatomical discriminant found thus far involves the complexity, or lack of it, in the branching of the branchial papillae. William Heard and I collected what appeared to be a single population of Uniomerus from an area of some 100m? in a stream of the panhandle of Florida. Electro- phoretic results ( 14 loci, 24 alleles) revealed two inter- mixed but distinctly different species (the shells were numbered, thus allowing separation of shells based on results of unique alleles). After the fact, I could separate the species on the basis of shell for about half of the shells. I subsequently obtained a population of Uniom- erus from Georgia with a distinctive shell phenotype and did a three-way electrophoretic comparison. To my con- siderable surprise, the greatest D in the three pairwise comparisons was 0.498 (from the two Floridian taxa), a value greater than between any of the 13 pairwise com- parisons for species of non-lanceolate Elliptio (greatest D of 0.446; Davis et al., 1981; Davis, 1984). Three species of Uniomerus were involved, not one! Parallelisms in shell characters disguised the two sympatric Floridian species. The myth that ecologically induced variation in shell size and other shell characters was generally con- siderable within unionid species, and thus there was only one species of Uniomerus, was exploded. The question arose: Could one clearly demonstrate the relative value of shell data, morphological data, and mo- lecular genetics data for discriminating evolved rela- tionships among species of Unionidae? To answer this question, I compared the following taxa: The three spe- cies of Uniomerus, Elliptio complanata, Fusconaia fla- va, Lampsilis teres, Quadrula quadrula, Quincuncina infucata. These taxa were classified by Davis and Fuller (1981) into three tribes on the basis of comparative im- munology and anatomy as follows: Pleurobemini: Ellip- tio, Fusconaia, Uniomerus; Amblemini: Quadrula, Quincuncina,; Lampsilini: Lampsilis. Too few anatom- ical characters have been found to allow a definitive cladistic analysis for unionid genera. Davis and Fuller (1981) found eight, of which two were shell characters. The larger clades (tribe, subfamily levels) are readily separated, but not the differences among some genera such as Uniomerus vs. Elliptio. Thus, anatomical data are limited. I compared the above taxa using shell morphometric data and molecular genetics. I used the same multivariate procedures with both data sets, ie. multidimensional scaling (MDS) with ordination of taxa on the first two principal components. The taxa were then connected with a Prim network. In the morphometric analysis, 39 characters were scored and treated by Principal Com- ponent Analysis to assess character correlations prior to MDS. Results are shown in Figures 3 and 4. Results based Page 8 QI EC U1 = THE NAUTILUS, Supplement 2 FF foe U3 Figure 3. Ordination of unionid taxa on first two principal components following multidimensional scaling and use of Prim network; based on shell morphometric data. The shapes approximate the shapes of the taxa. Three taxa could not be discriminated (EC,U1,U2). EC, Elliptio complanata; FF,Fusconaia flava; LT, Lampsilis teres; QI, Quincuncina infucata, QQ, Quadrula quadrula; U1,2,3, Uniomerus species 1,2,3. Adapted from Davis (1983). on shell data indicate relationships based on correlates of shell shape, not those based on anatomy and immu- nology. One cannot separate individuals of Elliptio com- planata and two species of Uniomerus. The data based on allozymes (Figure 4) reflect the tribal relationships based on anatomy and immunology. Clearly there are shell shape convergences (Figure 3). In Figure 4, Fus- conaia flava and Elliptio complanata group close to- gether in the same computer calculated set (D = 0.208) with a D less than that between the latter species and Uniomerus species 1 (D = 0.303), in spite of the fact that the shape of Fusconaia is the same as that of Quad- rula and distinctly different from the shape of Elliptio and Uniomerus. Lampsilis is genetically highly diver- gent from the other taxa as it should be, based on anatomy and immunology. The advantages of using MDS and the Prim network are clear. One can see the pattern of di- vergence of species of Uniomerus from species of Ellip- tio. One can see the direction of divergence of species from each other. It is readily observed that the greatest G. M. Davis, 1994 LT Page 9 U2 U1 EC ete FF eS eS QQ! Figure 4. Same taxa and ordination techniques as in Figure 3 but using genetic distances from allozyme electrophoresis. All taxa are clearly separated. Fusconaia and Quadrula are not closely allied as in Figure 3. Fusconaia is closely allied genetically to Elliptio (contrast Figure 3). There is greater distance between U2 and U8 than there is between Elliptio and Fusconaia! Adapted from Davis (1983). distance between two species of Uniomerus exceeds the distance between two genera, i.e. Elliptio complanata and Fusconaia flava; however, the direction of diver- gence of Fusconaia is away from Uniomerus relative to Elliptio. Take Away Messages Several points can be made: 1) There is a cryptic radiation of species of Uniomerus hidden by parallel evolution of shell shape and by con- chological variation that was previously interpreted as ecophenotypic or random variation. 2) It is predicted that other species of Uniomerus will be found. Consider the type species, U. tetralasmus from Texas that has a shell phenotype different from the phe- notypes of the three species discussed here. From the evidence thus far, populations of Uniomerus with dif- ferent shell phenotypes have a high probability of being different species. 3) Shells that look similar may belong to genera of diverging clades. Shell data provide the least valuable data for taxonomic discrimination above the species lev- el, and especially above the generic level. 4) When the anatomical database is weak, one must rely more on molecular databases. 5) Genetic distances must be calibrated within and between clades in order to understand the magnitude of distance values that indicate specific status. Gaps be- tween species groups and the direction of taxon diver- gence are clarified using multidimensional scaling and ordination with a network. II; POPULATION GENETICS Hybridization The following example demonstrates the use of molec- ular genetics to uncover hybridization. Bianchi et al. (1994) electrophoretically examined the allozymes of two North American freshwater gastropod pleurocerid spe- cies, Elimia virginica, an eastern slope species, and E. Page 10 livescens, an interior basin species. These two species are readily distinguished by differences in shell sculpture. What drew our attention was the presence of both species in the Erie Canal, which connects the interior basin Lake Erie to the Hudson River, an eastern slope drainage system. The canal was built in 1825. Near the center of the canal at Mudlock, Strayer noted a blurring of shell phenotypes and asked the question: Could hybridization be occurring? The electrophoretic study yielded data from 22 loci involving 39 alleles. Nei’s genetic distance in the two species comparison was 0.332; Arc distance was 0.577, values indicative of species differences for North Amer- ican Pleuroceridae (again,calibration) as evidenced by the work of Dillon (1984,1988), Dillon and Davis (1980), and Chambers (1978, 1980). We divided snails from the Mudlock location into two groups: one most resembling the E. livescens shell phenotype, with some shells ap- pearing to be pure E. livescens, the other most resem- bling the E. virginica type with some shells appearing to be pure E. virginica. There were 11 polymorphic loci, of which nine were diagnostic. A sampling of results is given in Table 2. It is clear that hybridization and in- trogression have occurred. There are no F, generation snails involved as there is fixation for alternative alleles at the LAP locus. However, evidence for introgression was found at seven (78%) of the diagnostic loci. In our study and those referenced above, mean het- erozygosities are very low for populations and species (0.002-0.03 most common). Thus, the situation along the Erie Canal stands out prominently with H values of 0.035, 0.044. As is characteristic of hybrid zones, polymorphism increases, there is an increase in rare alleles, and hetero- zygote deficiency is frequently found (Hewitt, 1988; also note the presence of new electromorphs called hybri- zymes by Woodruff, 1989). Vainola and Huilsom (1991) effectively described the occurrence of a hybrid zone for populations of Mytilus, but their diagnostic loci were not as distinct as those in our study. Special Selective Pressures In a study of populations of marine marshland snails, Hydrobia truncata, from North America, New England to Maryland, there were no significant anatomical dif- ferences except for size (Davis et al. 1988; 1989). Elec- trophoretic studies involved 80 loci, 49 alleles. There were 18 invariant loci. One population in a salt pond (Flax Pond) on Long Island, New York was of particular interest. This pond was 185 years old. The population differed from the others by Nei’s D of 0.12 + 0.01 while the other populations differed among themselves by a D of 0.07 + 0.01. This difference is due, in part, to eight unique alleles. The F; statistic shows population differ- entiation due to the unique genetics of the Flax Pond population; there are no significant regional differ- ence(Fs; = 0.004). The Flax Pond population is different from the other populations for two additional reasons: 1) the snails are of gigantic size; 2) the snails are excep- THE NAUTILUS, Supplement 2 Table 2. Sampling of allele frequencies from the pleurocerid hybridization study. virginica __ livescens virginica _ livescens Locus control hybrid-mix hybrid-mix control AAT-1 100 1.00 0.097 0.711 — 97 — 0.903 0.289 1.00 GPI 100 1.00 0.016 0.974 = — 0.984 0.025 1.00 NADD 100 1.00 0.210 0.763 = 105 — 0.790 0.237 1.00 LAP 100 1.00 — 1.00 — 97 — 1.00 — 1.00 tionally heavily parasitized. The extreme parasite burden causes the gigantism (Davis et al., 1988). The density of snails is also very high, some 25,000/m?. The most plau- sible explanation for the maintenance of unique alleles in a population of only 185 generations is density de- pendent selection where any mutation yielding a unique allele yields some benefit to a population extraordinarily stressed by parasitism (an idea attributable to Haldane). Breeding Structure Allozyme electrophoretic studies play a central role in the modern study of genetic structure of populations and especially breeding structure. The topic has been re- viewed as it relates to mollusks (Selander & Ochman, 1983). An example of such a study is provided by McCracken and Selander (1980), in which they studied the breeding systems of 14 species of three families of terrestrial slugs in the eastern United States. They dem- onstrated that for six species, the normal breeding system was either facultative or obligatory self-fertilization. One of the species had three monogenic strains. One of the most studied self-fertilizing species is the land snail Rum- inia decollata with more than 30 monogenic or very weakly polymorphic strains (Selander & Kaufman, 1978). Take Away Message Allozyme electrophoresis is, and will continue to be, an essential tool for studying population structure, patterns of reproduction, and uncovering different effects of se- lective pressures. III: PATTERNS OF EVOLUTION A: Speciation Species Definitions: One cannot speak of discriminating among species and engage in describing new species without committing to a species concept and being pre- G. M. Davis, 1994 pared to defend the concept. This commitment is es- pecially important if one is to apply molecular data in justifying species status. This is not the place, nor is there enough time in this presentation to argue at length all pertinent aspects for defending a particular concept. I will be brief. There are essentially five concepts worth arguing about of which I accept the fourth in the list. These are exten- sively reviewed by Templeton (1989). (1) The Biological Species Concept defended by Mayr (1963) has also been called the isolation species concept (Patterson, 1985; Templeton, 1989). It isnot acceptable as it stands because the emphasis is on isolating mechanisms. As most of us know from experience, the problem is not one of distin- guishing species that occur in sympatry, but those that are allopatric, e.g. in Korea vs. Japan. Intrinsic isolating mechanisms are irrelevant as isolating barriers during allopatric speciation. This is not to reject the relevance of isolating mechanisms when they occur and are part of the process of speciation, such as an instantaneous cytological event within a population that causes repro- ductive isolation. However, the process of speciation in allopatry has nothing to do with “isolating barriers”. Because Mayr associated polytypic species with con- venience in pigeon-holing taxa, various authors have re- jected both the biological species concept and the pol- ytypic species concept (e.g. Cracraft, 1989). However, these are two separate issues. There is value in under- standing isolating mechanisms where they do apply dur- ing the process of speciation; and I agree with Templeton (1989) that polytypic taxa are relevant as a concept in theory and practice. It is a straw-man issue to dismiss both concepts just because some thought to use polytypic taxa as an excuse to reduce the number of names one would have to deal with. The polytypic species issue will be revisited later. Gareth Nelson (1989) rejects the biological species con- cept because it assumes that species are the basic units of evolution. Since he rejects the idea that species are the basic units of evolution, he rejects the concept of species! While I can agree that species are not the basic units of evolution, I still can accept the reality of species. Populations are the fundamental units of evolution in as much that mutations or new gene combinations in in- dividuals of a population are basic to the process of spe- ciation, and excepting clonal and/or selfing individuals, population structure is necessary for the spread of new genes. I reject both the evolutionary (model 2) and phylo- genetic (model 3) species concepts for the same reasons. Wiley’s (1981) evolutionary species concept involves “a lineage which maintains its integrity from other such lineages and has its own evolutionary tendencies and historical fate”. It requires parental system of ancestry and descent. It requires large disjunctions in allopatry where one or more unique diagnostic qualitative char- acters are needed. The phylogenetic species concept of Rosen (1978) and Cracraft (1989) is the same but the analysis must be done by cladistic methods. “A phylo- Page 11] genetic species is an irreducible (basal) cluster of organ- isms, diagnosably distinct from other such clusters, and within which there is a parental pattern of ancestry and descent”. As Nelson and Platnick (1981) stated earlier, “species are simply the smallest detected samples of self perpetuating organisms that have unique sets of char- acters’. At this point one should note that all species concepts discussed here involve taxa that are monophy- letic, ie. there is a parental pattern of ancestry and de- scent; even the biological species concept. The problems with the above concepts (models 2 and 3) are that they provide no guidance as to what characters are important for defining species. By their criteria, two allopatric populations that differ in an electrophoretic study of allozymes, by two or three alleles unique to each population, should be called different species. Hardly! While both concepts admit to cohesion, i.e. the integrity of a morphological database, they do not allow for vari- ance, or how much variance can be associated with a species. They deal only with the concept of cohesion, not with the mechanisms responsible for cohesion. Ignorance of population genetics as it relates to expression of mul- tiple alleles at a single locus would easily lead to splitting populations into discrete species when they really should be considered one species. I prefer the cohesion species of Templeton (1989) (model 4) that includes the recognition concept (model 5) of Patterson (1985). According to Patterson, a species is a population of biparental organisms sharing a common fertilization system. This is the “flip side” of the biological species concept of Mayr. For Templeton, speciation is the evolution of cohesive mechanisms. A species is a group of populations of a monophyletic lineage. There is phenotypic-reproductive cohesion. These populations share the same fundamen- tal niche, i.e. the populations are interchangeable = demographic exchangeability. Natural selection pro- motes cohesion both through favoring reproductive co- hesion, genetic relatedness and affecting the limits of demographic exchangeability. The concept integrates population genetics and ecology with the standard stud- ies of morphology. The concept can be applied to all organisms from outbreeders to syngameons or partho- genetic organisms. Using this concept requires more work and more data, but who ever said that understanding the process of spe- ciation was simple and uncomplicated? The process of distinguishing species is not simply akin to picking up marbles and assigning names to them on the basis of size, surface patterns, and colors. With the cohesion concept, two populations, one in Japan and the other in Korea, might have their own evolutionary fates ahead of them, yet still belong to the same species. Patterns of Speciation: The relative ease of discrimi- nating among species very much depends on the pattern of speciation one encounters. There are two major modes that I have encountered again and again: adaptive ra- diation and morphostatic radiation. Osborn (1918) first Page 12 used the term adaptive radiation: “ Adaptive radiation is, descriptively, this extreme diversification of a group [ e.g. mammalian, reptilian] as it evolves in all the dif- ferent directions permitted by its own potentialities and the environments it encounters.” For Stanley (1979), adaptive radiation is the rapid progression of new taxa from a single ancestral group. If one takes Stanley’s def- inition, one might be dealing with the concept that Os- born had in mind, where there is a considerable diver- sification in morphological ground plans, or one might be involved with a morphostatic radiation as defined by Davis (1992). Benton (1988) very well captured the essence of the Osborn concept. The “key phases of an adaptive radia- tion such as that of the placental mammals 65 million years ago... involves (1), “an initial phase of rapid diversification from a single ancestor... (2) the estab- lishment of a diversity of new body plans in this early phase... (3) early extinctions amongst the initial ele- ments of the radiation. . .(4) a final phase of stabilization of the lineages. ...’ It is important to assess the time scale of an adaptive radiation. The mammalian radiation exploded between 65 to 55 million years ago. All the orders from bats to whales were established in this 10 million year period. There is also a taxon scale to consider. There has been extraordinary morphological diversifi- cation within the freshwater gastropod family Poma- tiopsidae, but especially in the subfamily Triculinae that is wholly southeast Asian and southern Chinese. In rel- atively short time, some 12 million years to the present, these small snails have diversified into three tribes and over 23 genera (Davis,1992). The rapid diversification probably took place in the first few million years (Davis et al., 1984). There is a splendid adaptive radiation that centers in the Mekong River (Davis, 1979) that involves some ten of the 23+ genera. There is also a large mor- phostatic radiation in southern China. I use the term morphostatic radiation to include those monophyletic taxa that have indeed radiated, but in al- lopatry and where there are little or no discernible niche differences. Likewise there is comparatively little mor- phological differentiation compared to what one sees in an adaptive radiation where morphology may vary in a number of dimensions that reflect adaptations to differ- ent environments. I use the term to replace Gitten- berger’s (1991) “non-adaptive radiation” that is: “.. non-adaptive radiation should denote evolutionary diversification from a single clade, not accompanied by relevant niche diversification. . the various species re- sulting from the process would,in principle, not be able to be sympatric.” As it is reasonable to consider that all species in nature are adapted to their environments, the term non-adaptive seems unsuitable. Further, one can discern between adaptive radiations sensu Osborn, where morphological adaptations to differing environments are considerable, and morphostatic radiations as defined above. Gittenberger maintains that there are numerous intermediate situations between the two types I describe, and that the concepts of higher taxa and genera are too THE NAUTILUS, Supplement 2 Table 3. Scoring genera and the radiation of species of each genus for qualitative morphological and ecological differences involving eight characters that includes ecology. See text for details. Ay = coefficient of radiation diversity. Adaptive radiation: Triculinae Huben- Lacun- Pachy- Julli- dickia opsis drobia enia Shell 5) 10 10 10 Radula 5 5 0 0 Mantle cavity 0 5 0 Head 0 5 0 0 Female reprod. system 5 5 5 5 Male reprod. system 5 0 0 5 Nervous system 0 0 0 0 Ecology 5 10 10 5 Sum scores = 25 40 25 25 Ay = 8.68 Morphostatic radiation: Triculinae Neotric- Gamma- Wucon- Tricula ula tricula chona Shell 0 0 0 5) Radula 0 0 0 Mantle cavity 0 0 0 0 Head 0 0 0 0 Female reprod. system 5) i) 5 5 Male reprod. system 0 0 0 0 Nervous system 0 0 0 0 Ecology 0 0 0 0 Sum scores = bY i) 5 10 ill defined. I reject these notions. There are sufficient quantitative and cladistic procedures to clearly define taxa except in certain instances of morphostatic radiation where the mosaic of few characters confounds under- standing the limits of species in allopatry. In a morphostatic radiation one may be able to discern among species or subspecies because of shell sculptural differences such as seen in Albinaria of Gittenberger’s (1991) example. There may be a mosaic of quantitative differences that separate species. But these are, in such radiations, usually small differences. To make the point, I compare four genera of the Triculinae adaptive radi- ation with four genera of the Triculinae morphostatic radiation in Table 8. Each genus is scored for qualitative differences in morphology and ecology compared with other genera, where the species occupy different ecolo- gies. There are eight characteristics scored. A genus is scored 5 when it clearly differs from other genera in a change in ground plan; species in the genus are scored 5 when they diverge in a character. For example, La- cunopsis differs from other genera in ecology (5) and the species have radiated into different niches (5); ecol- ogy thus scores 10. In the same genus the radula differs from others in the tribe but the radula is the same in all of the species, therefore it scores 5 for radula. In a mor- phostatic radiation one cannot tell the shells of Neotricula G. M. Davis, 1994 from those of Tricula; shells score 0 as the species do not radiate with different shell shapes or sculpture; etc. To quantify the differences between the types of ra- diation, I use a coefficient of adaptive differentiation, Ag, that is calculated using the formula: It is a standardized sum of mean squared differences where n = number of characters scored; $,S5. . .S, =in- dividual score for a character for generic species group 1, etc.; N = number of generic species groups. From the example given in Table 3, there is 4.5 times the amount of diversification morphologically and ecologically in the adaptive radiation of triculine taxa compared to the mor- phostatic radiation of triculines. I have digressed considerably on species definitions and patterns of speciation because it becomes clear that discriminating among species in an adaptive radiation, and where several congeneric species may be located in a habitat, may be a relatively easy task. The difficulties reside with allopatric taxa of a morphostatic radiation where molecular genetics may be extremely useful or may confound the issue. The situation I presented above for the three species of Uniomerus is an example of usefulness of molecular genetics. I will present two ex- amples where the interpretation of molecular genetics must be made in light of all other data. Oncomelania hupensis polytypic species: Oncomelania is a member of the Pomatiopsidae: Pomatiopsinae, with what I currently consider to be two species: Oncomelania hupensis and O. minima. This genus is a member of a morphostatic radiation. The latter species occurs in Ja- pan; the former has subspecies distributed in China (1), Taiwan (2), Japan (1), Sulawesi (1), and the Philippines (1). The systematics of this genus was reviewed by Davis (1980,1981). Briefly, O. hupensis was considered to be a polytypic species because: (1) the anatomy, except for differences in size, was identical for all allopatric pop- ulations; (2) the main shell differences were the occur- rence of ribs on some populations in southern China; however ribbing is controlled by a single gene and this gene has been manipulated by hybridization experi- ments; (3) the populations can be hybridized with no loss of viability of F, or subsequent generations; (4) as anyone who has simultaneously raised these snails in culture will attest to, there is ecological exchangeability between pop- ulations. There are indeed small differences in suscep- tibility to different allopatric populations of Schistosoma japonicum; in size; in degree of shell varix formation; and the degree of gland formation about the medial aspects of the eyes. Recently, Woodruff et al. (1988) did an electropho- retic allozymic study of several populations of Onco- Page 13 melania hupensis quadrasi from the Philippines and compared them with O. h. hupensis from China. The Nei’s D among the Philippine populations averaged 0.036 (greatest value = 0.134); The Chinese and Philippine populations differed by a mean of 0.62 + 0.04. They concluded that the Chinese and Philippine snails be- longed to different species because of the large genetic distance. They justified this conclusion by (1) invoking the evolutionary species concept; (2) they reject the bi- ological species concept with its emphasis on reproduc- tive isolation and consider polytypic species to be an essential element of the biological species concept; (3) they argue that subspecies is a category of convenience, a way of pigeon-holing taxa and reducing the number of names one has to deal with; (4) they argue that the use of subspecies causes confusion by underestimating the number of independently evolving lineages. I disagree! The polytypic species concept with its sub- species is indeed useful; it certainly does not have to be married to the biological species concept. Also, the rec- ognition of subspecies does not reduce the number of names used nor does it imply that the allopatric popu- lations involved are not independent evolving lineages. The use of subspecies serves a very useful purpose and is not used as a matter of convenience. Further, the Oncomelania hupensis polytypic species complex does meet the major criteria of the evolutionary species con- cept: reproductive recognition and genetic integrity. Large genetic distances by themselves do not serve to define species. For example, Johnson et al. (1984) found that a population of Cepaea nemoralis introduced from Europe to Lexington, Virginia (southern U.S.A.) differed from a population from Florence, Italy by Nei’s D of 0.631; a population from Santa Croce near Pavia, Italy compared with the Florence population differed by D of 0.391. This species is well known for geographic vari- ation for both shell polymorphism and allozymes. The species is well studied throughout its range. To quote Johnson et al. (1984), “The decoupling of genetic di- vergence from speciation emphasizes the limitations of viewing the process of speciation solely in genetic terms.” In questioning the origin of the Lexington population, Stine (1989) used restriction enzyme analysis of mito- chondrial DNA to demonstrate it to be more closely related to populations from England rather than from Italy (Nei’s D of 0.409). What Woodruff et al. (1988) are doing is ignoring the great genetic cohesion that unites the subspecies of On- comelania hupensis, a cohesion that is associated with demographic exchangeability. One powerful example of this genetic cohesiveness is the invariability of the re- productive systems. There is indeed cohesion in repro- ductive recognition. While I certainly agree that the ability to hybridize is not a criterion for merging per- fectly good species (examples of syngameons are nu- merous), it is instructive that the large Nei’s D does not interfere with the hybridization of these subspecies with no loss of viability in the offspring or through successive generations. There is indeed genetic cohesiveness. Be- Page 14 ROGERS’ D .90 .80 .70 .60 .50 1.7 1.6 1.5 1.4 1.3 THE NAUTILUS, Supplement 2 -OVALIS .CHIT .MINN.-A . MINN.-B 74 FA C4 CS Belen (CS SP.aiG SP.- B . SP-A nnn nan OXYLOMA .40 .30 -20 - OVALIS . CHIT . MINN.-A . MINN.-B Behe (S / SPB 2 SAS [D) Dn NnNnNHnDn 222 2 6 SPAS /A\ 1.2 1.1 1.0 0.0 UPGMA- MORPHOLOGICAL DISTANCE Figure 5. A comparison of phenograms comparing some of the succineid taxa studied by Hoagland and Davis (1987). Allozymic electrophoretic results are compared with morphological results. Modified from Hoagland and Davis (1987). Sp. D was not studied electrophoretically. S= Succinea; N=Novisuccinea. See text for details. yond the cohesiveness of the reproductive organs, there is the cohesiveness in all the other details of anatomy, reproductive habit, responses to environmental manip- ulation. The usefulness of the subspecies designation in this example is to bring attention to the great cohesion throughout this complex and understand what this im- plies for many aspects of the biology of the species through time. It would be useful if Woodruff et al. would examine the disruption of cohesion (i.e. morphological diversifi- cation) in sister taxa to Oncomelania that are likewise considered part of a morphostatic radiation, i.e. species of Tricula or Neotricula. One finds numerous characters of use to distinguish among species. Examples of these are most frequently found in slight modifications of the reproductive systems; e.g. penis with papilla in one spe- cies, without papilla in another; penis with pronounced ejaculatory duct in one species, without ejaculatory duct in another; penis mounted center on the head vs. right of center; seminal receptacle arising at position “a” in one species, or in position “b” in another. Also,the shell G. M. Davis, 1994 Page 15 A O.HUPENSIS N.LILU G.CHINENSIS G.SONGI —— SS -80 32 -64 -56 -48 -40 32 24 16 .08 -00 B O.HUPENSIS N.LIL G.CHINENSIS G.SONGI —S 1.00 -90 -80 70 -60 -50 -40 30 -20 -10 -00 Cc O.HUPENSIS N.LILW G.CHINENSIS G.SONGI 1.00 -90 -80 70 -60 50 -40 -30 20 -10 -00 Figure 6. Phenogram based on UPGMA treatment of genetic distances. A. Nei’s D; B. Wright's modified Rogers’ D; €. Arc D. O. = Oncomelania, N.= Neotricula; G.= Gammatricula. Adapted from Davis et al. (1994b). may have an internal tooth on the columella vs. no tooth; and so on. These small differences are the types found in a morphostatic radiation in contrast to major ground- plan changes found in an adaptive radiation. The contrast with polytypic Oncomelania hupensis is striking. To achieve full species status, some occurrence must cause disruption of the cohesion seen. This is evidenced in Oncomelania minima of Japan where the shell shape departs from that seen in Oncomelania hupensis; there Oncomelania hupensis Neotricula lilii Gammatricula songi Gammatricula chinensis Figure 7. Cladogram based on a Hennig86 treatment of the same allozyme data used in Figure 6, but scoring each locus as a character. are several shifts in the morphologies of the reproductive systems (Davis, 1969), character-state changes similar to those seen in the sister subfamily Triculinae. The point is that one must know what occurs in sister taxa relative to the taxon under study; what are the patterns of char- acter change relative to recognition of species. Unfortunately, the use of subspecies in malacology is generally farcical! No wonder the term subspecies is little respected when subspecific status is awarded to popu- lations that differ by so slight a character-state as an extra bump or node or rib in one population that is not seen in another. Numerous subspecies have been based on conchology alone where the basic definition of a species has not been worked out, let alone any understanding of what the extra bump means. However, while most mal- acological subspecies currently named in the literature have no biological validity, there are indeed substanti- ated cases of polytypic species, and Oncomelania hu- pensis is one of them. The land snail genus Succinea: In this example, com- parative anatomy, ecology, and molecular genetics were used to assess species status. As will be shown, molecular genetic data were useful in some cases, not useful in other cases. The question was, what was the true identity Page 16 2) 2 G 2 i . = 2 2 > = x= () = = o Zs) (2) = o o 10(1) 7(0) 1-6(1) © —_ AUTAPOMORPHIES ea) SYNAPOMORPHIES PLESIOMORPHIES Figure 8. Cladogram based on morphological data for the same taxa shown in Figures 6, 7. of a rare and endangered species, Succinea chittenan- goensis Pilsbry, and how could it be distinguished from other sympatric species on or at the Chittenango Falls in upper New York State in the northeastern USA? The study included topotypical S. ovalis and populations of Succinea from Pennsylvania and Minnesota. The out- groups for the electrophoretic study were Oxyloma re- Gammatricula chinensis hilii Gammatricula songi Neotricula THE NAUTILUS, Supplement 2 tusa Lea, and O. decampi gouldi Pilsbry of Chittenango Falls. Through electrophoresis and shell morphometrics, another species was found at the falls in addition to S. ovalis and S. chittenangoensis, a species with a shell shape similar to S. putris (Linnaeus) of Europe. In the anatomical studies, 51 characters were scored using bi- nary coding. In the allozyme studies, 31 loci involving 87 alleles were found. The data of both sets were ana- lyzed using multivariate analysis yielding UPGMA de- rived phenograms as shown in Figure 5 [modified and simplified from Hoagland & Davis (1987)]. The findings were: (1) Genetic and morphological data support the conclusion that three genera are involved; Succinea, Novisuccinea, and Oxyloma. (2) Oxyloma is more closely related to Succinea than it is to Novisuccinea. (8) N. ovalis and N. chittenangoensis at the falls cannot be distinguished electrophoretically while they are clearly distinct in terms of anatomy and ecology (as well as on shell differences). They are distinct species. (4) The two populations of Novisuccinea from Minnesota are clearly not N. ovalis. They are not morphologically distinct yet they have diverged genetically (Nei’s D = 0.104). Fur- ther studies would be necessary to assess whether or not they are specifically distinct. (5) In the remainder of the comparisons results based on morphology paralleled those based on molecular genetics. Why are the falls Novisuccinea species morphologi- cally divergent yet not electrophoretically so? The prob- able answer is that the area was glaciated until 10 to 12 thousand years ago. With retreat of the glaciers and the uncovering of the falls, N. chittenangoensis evolved from an ancestor of regional N. ovalis by colonizing the falls with concomitant shifts in morphology in adapting to new ecological space. There has not been enough time to diverge in terms of allozymes. Oncomelania hupensis r = 0.84 Figure 9. Prim network based on multivariate analysis of morphological data. The taxa are those treated in Figures 6-8. G. M. Davis, 1994 B: Phylogeny Molecular genetics are certainly useful in assessing phy- logeny. I will provide an example that builds on the morphostatic radiations given above involving the Po- matiopsidae: Pomatiopsinae and Triculinae (Davis et al., 1994b). The questions asked were: Are the Pomatiopsinae and Triculinae monophyletic? Is Oncomelania closely related genetically to the more generalized triculine taxa that are part of the morphostatic radiation? What genetic distances might one expect between genera of the Tri- culinae? Is a cladogram based on genetic data congruent with a cladogram based on anatomy? Are these clado- grams congruent with biogeographical data? The electrophoretic analysis involved 28 loci and 78 alleles. The morphological analysis involved 17 charac- ters. Phenograms based on UPGMA treatment of three genetic distances are given in Figure 6. A cladogram based on using each enzyme locus as a character and applying Hennig86 version 1.5 (Farris, 1989) is given in Figure 7. Oncomelania is the outgroup; there was no differential weighting or polarities assigned. Only one tree resulted, with a consistency index of 0.62. The clado- gram based on the morphological data is given in Figure 8. The phenograms and the two cladograms are congru- ent. The cladograms are congruent with biogeography and the hypothesis on the direction of evolution from northern Burma-western Yunnan, China with dispersion and divergence down evolving river systems (Davis, 1980, 1992). These congruencies give confidence about the phylogenetic results published earlier based solely on comparative anatomy (e.g. Davis & Kang, 1990; Davis, 1992) The question about monophyly is also answered. There is no great divergence of Oncomelania from the triculine taxa. The question was justified for the following reason. In the Pomatiopsinae, the spermathecal duct runs from the bursa copulatrix to the anterior end of the mantle cavity. In the Triculinae the spermathecal duct runs from the bursa to the pericardium or to the posterior end of the mantle cavity. Are the spermathecal ducts homolo- gous? It has been a hypothesis that the spermathecal duct in the Triculinae derived from the primitive gonoperi- cardial duct that connects the oviduct to the pericardium in some rissoacean taxa. However, in the Pomatiopsinae there is a vestigial gonopericardial duct and the sper- mathecal duct! Two families might be involved. The average Nei’s D between Oncomelania and the triculine taxa is 1.29 + 0.41. This is not a large distance considering what one might expect of different families. It is especially not large when one calibrates the system. The distance between two of the triculine taxa is 1.26, a D value greater than between Oncomelania and Gam- matricula songi where D = 1.00. In describing G. songi, Davis et al.(1994) stated that the anatomical innovations found in this species warranted generic status, but that a new genus would not be named until more species of Gammatricula were found and studied. As shown in Figure 9, G. songi and G. chinensis diverge equally from Page 17 Neotricula along the Prim Network. Considering there to be three triculine genera involved, the average Nei’s D among them is 0.890 with a range of 0.689 to 1.236. Thus, Oncomelania seems more to be a genus closely allied within a triculine generic grouping rather than a member of a different subfamily. Once again the point is made: Measures of genetic distance do not serve to define taxon levels! The subfamilies Pomatiopsinae and Triculinae are firmly based on qualitative anatomical data that in either a phylogenetic/ cladistic or multi- variate analysis support those diverging sets of genera at a hierarchical level deserving subfamilial status. The ge- netic data do serve to confirm close genetic relationship, not a highly disjunct pattern indicating polyphyly. Take Away Message Discriminating taxa at the species level is most difficult when one is dealing with allopatric populations of a mor- phostatic radiation. There are indeed different processes of speciation. Speciation may proceed uncoupled from genetic differentiation seen in structural genes such as demonstrated using allozymes. Considerable genetic dis- tances do not necessarily mean that the overall genetic cohesiveness among populations is disrupted to the extent that species status is attained. Rapid morphological change in adapting to new environmental space may outpace molecular genetic change. In examining a large radiation spread over great distances, one would expect that in perhaps 70% or more of the species, morphological and molecular genetic change would diverge in parallel. Un- tangling species-level problems can be a most challenging task as pointed out by Giusti and Manganelli (1992; see Prologue). For those engaged in this task, one needs as much data as one can obtain, certainly building on a firm platform of detailed comparative anatomy. Ecolog- ical data are essential. Molecular data are always useful, but do not add to the solution of a problem in a rote formulated way. Above all, molecular data must be cal- ibrated for the radiation under study. Concerning phylogeny: Molecular genetic tools are es- sential to test phylogenies based on comparative anato- my. Together, both data sets provide insight into the rate of evolution. Together, both data sets serve to test hy- potheses about biogeography. IV: UNCOVERING UNIQUE ASPECTS IN EVOLUTIONARY PROCESS It has been known for a long time now that different molecular data sets may yield different results. Also, one set of tools is better suited for assessing relationships at one taxonomic level, while other tools are better suited for a different taxonomic level. For example, restriction enzyme analysis of mitochondrial DNA is most suited for determining relationships at the population level, or among closely related species. Allozymes are superb for studies of population genetics and to assess relationships Page 18 THE NAUTILUS, Supplement 2 QUADRULA, MEGALONAIAS, UNIOMERUS ANODONTA, LAMPSILIS, OBLIQUARIA ELLIPTIO, AMBLEMA, FUSCONAIA, PLECTOMERUS, UNIO GONIDEA CUMBERLANDIA MONODONTA MARGARITIFERA MARGARITIFERA MARGARITIFERA FALCATA Figure 10. Phenogram following UPGMA treatment of distance coefficients based on LrRNA sequence differences among species of freshwater clams of the family Unionidae (for further details, see Rosenberg et al., 1994). among species, monophyletic genera and tribes to sub- families. Immunology has been used with success from the species to family level. However, as discussed above, there is no universal molecular clock. Different data sets may yield different results. For example, Murray et al. (1991), as part of a series of excellent studies of evolution and speciation within the Pacific islands land snail Par- tula, compared the results of morphological investiga- tions, protein electrophoresis, and mtDNA and defended the following: “...the different data sets evolve inde- pendently and at variable rates. This mosaic pattern of evolution can only occur if natural selection plays a role in the genetic differentiation of Partula”’. We are now in a new age, one of sequencing. Cloning genes and nucleic acid sequence analysis began to ex- plode in the decade of the 80's. Exciting developments were made possible with DNA amplification by the poly- merase chain reaction (PCR) where a DNA segment of some 6000 base pairs may be amplified starting with as little as a single gene copy (reviewed by Landergren et al., 1988). As with the emergence of any new technology, one should expect some surprises. Paradigms based on studies of mammals may be shattered when studying mollusks that have evolved over 600 million years with amazing diversification of anatomical groundplans, physiologies, and genetics. I present one such surprise encountered when studying large-ribosomal-RNA se- quences of a series of freshwater clams (Unionidae) as part of a larger study that included land snails and the prosobranch Oncomelania (Emberton et al., 1990). The taxa studied are listed in Table 4 in the classifi- cation scheme of Davis and Fuller (1981). We examined some 150 base sequences that included the highly con- served 5’ end and the D-6 divergent domain plus flank- ing regions. We scored 26 differences among taxa and subjected these to a simple standard UPGMA treatment with the resulting phenogram shown in Figure 10. I wish to make only a few remarks about these results; a more detailed treatment of these sequence data in relationship to sequence data from diverse mollusks is presented later in this issue (Rosenberg et al., 1994). Morphological, immunological, and allozyme data support the concept that there are three equal and di- vergent clades (Figures 11, 12, adapted from Davis & Fuller,1981; Davis et al., 1981). I would prefer to call them subfamilies, while others have split off the group of Margaritifera as a separate family. I point out, how- ever, that given the weight of evidence, the group of Anodonta is equally divergent from other non-Margar- itifera unionids and thus should be accorded equal rank either at the family level or subfamily level. While there is congruence of the morphologi- cal,allozymic, and immunological data, the LrRNA se- Table 4. Unionid species used to study LrRNA sequences classified in the scheme of Davis and Fuller (1981) based on immunological and morphological data. Margaritiferinae Cumberlandia monodonta Margaritifera margaritifera Margaritifera falcata Anodontinae Anodonta cataracta Anodonta imbecilis Anodonta grandis Ambleminae Gonideini Gonidea angulata Pleurobemini [should be Unionini] Elliptio complanata Pleurobema cordatum Fusconaia cerina Unio pictorum Uniomerus “tetralasmus” Amblemini Amblema plicata Quadrula quadrula Quadrula cylindrica Megalonaias boykiniana Plectomerus dombeyianus Lampsilini Lampsilis claibornensis Lampsilis teres Obliquaria reflexa G. M. Davis, 1994 Page 19 GONIDEINI = ce) AN 7 AMBLEMINAE MARGARITIFERINAE I ANODONTINAE Figure 11. Ordination diagram following multidimensional scaling using immunological distances from freshwater clams of the family Unionidae. Computer-derived sets and subsets are enclosed in the dashed lines. As with the allozymic data-set (Fig.10), there are three discrete clusters: Margaritiferinae, Anodontinae, and Ambleminae( Amblemini,Lampsilini, Unionini, Gonideini). Adapted from Davis and Fuller (1981). quence data offer some surprises! As seen in Figure 10, (1) Anodonta cannot be distinguished from Lampsilini genera Lampsilis and Obliquaria. (2) One cannot distin- guish among species or genera in the groupings of An- odonta etc., Elliptio etc., and Quadrula etc., yet there are distinct differences between the two species of Mar- garitifera. (3) Cumberlandia monodonta, is widely sep- arated from the species of Margaritifera. The differences among the Maragitiferinae taxa and with other unionids involve sequence changes at 25 positions, while the dif- ference between the Anodonta group of genera and the Quadrula group of genera involves only one difference. The point to be made here is that Anodonta seems firmly nested with other genera of the tribes Unionini ,Amble- mini, and Lampsilini while on all other data, both mor- phological and molecular (immunological and allozym- ic), Anodonta is highly divergent from genera of those tribes (Figures 11,12). The three Margaritiferinae taxa Page 20 ANODONTA S -1.0 MARGARITIFERA © THE NAUTILUS, Supplement 2 0.2 = EUSCONAIA ae J (0 . \ \ - \ | 2 ; /ELLIPTIO Y, -0.9 Figure 12. Ordination diagram following multidimensional scaling using allozymic electrophoretic data from freshwater clams of the family Unionidae. Computer-derived sets and subsets are enclosed in dashed lines. A Prim network is used to connect taxa. Note the direction of divergence of Anodonta away from Margaritifera, apecommately equidistant from the set of the Ambleminae ( Lampsilis, Fusconaia, Elliptio). Adapted from Davis et al. (1981). show considerable divergence among themselves, with considerable changes in the variable and 3’ flanking re- gion, not seen in the other unionids. This is indeed a surprise. One interpretation that warrants further testing, is that the Margaritiferinae diverged from all other unionids at an early date and uniquely departed from other unionids in this pattern of sequence changes. An- odonta, while maintaining the rather conservative se- quence structure, diverged from the non-Margaritiferi- nae clade, also at an early date, and rapidly diversified morphologically with concomitant immunological and allozymic changes. Take Away Message It is clear that a single measure of genetic distance cannot be used to discriminate among taxa. Speciation indeed G. M. Davis, 1994 progresses by different patterns and processes. Speciation may proceed unhinged from genetics (as evidenced by current molecular techniques); some allopatric popula- tions may retain great cohesion in breeding system, mor- phology and demographic exchangeability (all geneti- cally controlled) yet accumulate considerable structural gene changes that, by themselves, do not justify giving the populations species rank. Generally, morphological and genetic data diverge in parallel. Two points are especially clear: (1) a species concept is necessary that does not go to the absurdity that one or two qualitative differences among allopatric populations justifies species status, especially on the basis that the separated popu- lations, being thus isolated, have their own unique tra- jectory in time and space; (2) one needs all the possible data one can obtain to sort out some species problems, starting with detailed anatomical data and ecological observations. Indeed, the new generation of molecular tools are yielding surprises. Sequence data join the other tools in providing powerful insights into patterns and processes of evolution. However, as mountains of data accumulate, it will become increasingly clear that all the problems with other molecular data sets will become evident with sequence data: convergences, sequences of one molecule ( e.g. LrRNA) being uninformative for some groupings of taxa, while showing wild divergences for other taxa, etc. These problems will settle down with the sequencing of whole genes and using genes as characters in phylo- genetic analysis. We are a long way from this, as yet, costly and time consuming task. Increased automation of procedures will ease the task. CONCLUSION Clearly nature is both capricious and pernicious in how she spins off species and promotes patterns and processes of evolution. It certainly appears this way to a seasoned systematist. The work of discriminating among taxa is clearly complex and multidimensional. There are no rote rules to apply such as stating that species status is achieved when Nei’s D equals some artificial value. What is splen- did today are the variety of tools that can be applied to solving taxonomic problems. The battery of new molec- ular tools are especially appreciated and provide the basis for much rigor in testing hypotheses about taxonomic relationships. I have discussed in this lecture the utility of molecular tools for uncovering cryptic species, for studying population genetics, and for studying patterns of speciation and phylogeny. I hope that I have made clear the point that the fun- damental basis of taxonomic discrimination is based on detailed comparative anatomy and cytology. Genetic dis- tances, by themselves, do not serve to define species or higher taxa. There is no universal molecular clock! Fur- ther, morphological, allozymic, MtDNA, and DNA se- quences may diverge at different rates within the same taxon. Taxa within a clade must be calibrated relative to genetic distance. In studying a situation involving the Page 21 species-level, it is useful to know if one is involved with an adaptive radiation or a morphostatic radiation; it is useful to know the characters and character-state changes that serve to distinguish species and genera in sister taxa. To add to the want list, a systematist would like to de- termine the ecological correlates of morphology, the time of taxonomic divergence, and the direction of evolution. Such data require knowing a group on a global basis. Timing and direction may come from paleontological evidence or from geological events. And still, as Giusti and Manganelli (1992) stated so well, a good systematist “..is not above admitting that its exact nature [what is or has occurred] escapes him. . .”. Understanding a complex situation in speciation may take years of study. ACKNOWLEDGMENTS This presentation, and much of the work presented here was supported by an NIH grant AI 11873 TMP. The graphics were prepared by Susan Trammell. I am in- debted to Gary Rosenberg for reading and making useful criticisms of this manuscript. LITERATURE CITED Avise, J.C. and C. F. Aquardo. 1982. A comparative summary of genetic distances in the vertebrates. Evolutionary Bi- ology 15:151-158. Bandoni, S., M. Mulvey, D.Koech and E.Loker. 1990. Genetic structure of Kenyan populations of Biomphalaria pfeifferi (Gastropoda: Planorbidae). Journal of Molluscan Studies 56: 383-391. Barton, N. H. 1989. Founder effect speciation. In: D. Otte and J. Endler (eds.) Speciation and Its Consequences. Sin- auer Press, Sunderland, MA. pp.256-296. Benton, M. J. 1988. The nature of an adaptive radiation. Tree 3:127-128. Bianchi, T. S.,G. M. Davis and D. Strayer. 1994. An apparent hybrid zone between freshwater gastropod species Elimia livescens and Elimia virginica (Mesogastropoda: Pleuro- ceridae). American Malacological Bulletin (in press). Buth, D. G. 1984. The application of electrophoretic data in systematics studies. Annual Review of Ecology and Sys- tematics 15: 501-522. Cavalli-Sforza, L. L. and A. W. Edwards. 1967. Phylogenetic analysis: models and estimation procedures. Evolution 21: 550-570. Chambers, S. M. 1978. An electrophoretically detected sibling species of “Goniobasis floridensis”. Malacologia 17: 157- 162. Chambers, S. M. 1980. Genetic divergence between popu- lations of Goniobasis (Pleuroceridae) occupying different drainage systems. Malacologia 20:63-81. Cracraft, J. 1989. Speciation and its ontology: The empirical consequences of alternative species concepts for under- standing patterns and processes of differentiation. In: D. Otte and J. Endler, (eds.) Speciation and its consequences. Sinauer Associates,Sunderland, MA. pp. 28-59. Davis,G. M. 1969. Reproductive, neural and other anatomical aspects of Oncomelania minima (Prosobranchia: Hydro- biidae). Venus 28:1-36. Davis, G. M. 1979a. The origin and evolution of the gastropod ~ Page 22 family Pomatiopsidae, with emphasis on the Mekong River Triculinae. Monograph of The Academy of Natural Sci- ences No.20:1-120. Davis, G. M. 1979b. Experimental methods in molluscan sys- tematics. In: Fretter, V. and J. Peake (eds.) Pulmonates Vol. 2. Academic Press, New York pp. 99-169. Davis, G. M. 1980. Snail hosts of Asian Schistosoma infecting man: origin and coevolution. In: Bruce, J. et al. (eds.) The Mekong Schistosome. Malacological Review, Supplement 2:195-238. Davis, G. M. 1981. Different modes of evolution and adaptive radiation in the Pomatiopsidae (Prosobranchia: Mesogas- tropoda). Malacologia 21:209-262. Davis, G.M. 1983. Relative roles of molecular genetics, anat- omy, morphometrics, and ecology in assessing relation- ships among North American Unionidae (Bivalvia). In: Oxford, G. S. and D. Rollinson (eds.) Protein Polymor- phism: Adaptive and Taxonomic Significance. Systematics Association Special Volume No.24, Academic Press, New York pp. 193-221. Davis, G. M. 1984. Genetic relationships among some North American Unionidae (Bivalvia): Sibling species, conver- gence, and cladistic relationships. Malacologia 25:629-648. Davis, G. M. 1992. Evolution of prosobranch snails trans- mitting Asian Schistosoma; coevolution with Schistosoma: A review. Progress in Clinical Parasitology 3:145-204. Davis, G. M. (ed.) Systematics and Public Health:A contri- bution to the Systematics Agenda 2000 (in press). Davis, G. M., C. E. Chen and S. H. Yu. 1994a. Unique mor- phological innovation and population variation in Gam- matricula songi, a new species of Triculinae from China (Gastropoda: Rissoacea). Proceedings of the Academy of Natural Sciences of Philadelphia 145:107-145. Davis, G. M., C. E. Chen, X. P. Zeng, S. H. Yu and L.Li. 1994b. Molecular genetic and anatomical relationships among, pomatiopsid (Gastropoda: Prosobranchia) genera from southern China. Proceedings of The Academy of Natural Sciences of Philadelphia 145:191-207. Davis, G. M., V. Forbes and G. Lopez. 1988. Species status of northeastern American Hydrobia (Gastropoda:Proso- branchia): Ecology, morphology and molecular genetics. Proceedings of The Academy of Natural Sciences of Phil- adelphia 140:191-246. Davis, G. M. and S. L. H. Fuller. 1981. Genetic relationships among recent Unionacea (Bivalvia) of North America. Malacologia 20:217-253. Davis, G. M., W. H. Heard, S. L. H. Fuller and C. Hesterman. 1981. Molecular genetics and speciation in Elliptio and its relationships to other taxa of North American Unionidae (Bivalvia). Biological Journal of the Linnean Society 15: 131-150. Davis, G. M. and Z. B. Kang. 1990. The genus Wuconchona of China (Gastropoda: Pomatiopsidae: Triculinae): Anat- omy, systematics, cladistics and transmission of Schisto- soma. Proceedings of the Academy of Natural Sciences of Philadelphia 142:119-142. Davis, G. M., Y. H. Kuo, K. E. Hoagland, P. L. Chen, H. M. - Yang and D. J. Chen. 1984. Kunmingia, a new genus of Triculinae (Gastropoda: Pomatiopsidae) from China: Phenetic and cladistic relationships. Proceedings of the Academy of Natural Sciences of Philadelphia 136:165- 193. Davis, G. M., M. McKee and G. Lopez. 1989. The identity of Hydrobia truncata (Gastropoda: Hydrobiinae): com- parative anatomy, molecular genetics, ecology. Proceed- THE NAUTILUS, Supplement 2 ings of The Academy of Natural Sciences of Philadelphia 141:333-359. Dillon, R. T. 1984. Geographic distance, environmental dif- ference, and divergence between isolated populations. Sys- tematic Zoology 33:69-82. Dillon, R. T. 1988. Evolution from transplants between ge- netically distinct populations of freshwater snails. Genetica 76:111-119. Dillon, R. T. and G. M. Davis. 1980. The Goniobasis of southern Virginia and northern North Carolina: genetic and shell morphometric relationships. Malacologia 20:83- 98. Emberton, K. C., G. S. Kuncio, G. M. Davis, S. M. Phillips, K. M. Monderwicz and Y. H. Guo. 1990. Comparison of recent classifications of stylommatophoran land-snail fam- ilies, and evaluation of large-ribosomal-RNA sequencing for their phylogenetics. Malacologia 31:327-352. Farris, J. 1988. HENNIG 86, Version 1.5. Computer Software. Stonybrook, NY. Gillespie, J. H. 1984. The molecular clock may be an episodic clock. Proceedings of the National Academy of Science. U.S.A. 81:8009-8013. Gillespie, J. H. 1986a. Rates of molecular evolution. Annual Review of Ecology and Systematics 17:637-665. Gillespie, J. H. 1986b. Variability of evolutionary rates of DNA. Genetics 113:1077-1091. Gillespie, J. H. 1987. Molecular evolution and the neutral allele theory. Oxford Surveys Evolutionary Biology 4:10- 37. Gittenberger, E. 1991. What about non-adaptive radiation? Biological Journal of the Linnean Society 43:263-272. Giusti, F. and G. Manganelli. 1992. The problem of the spe- cies in malacology after clear evidence of the limits of morphological systematics. In: E. Gittenberger, E. and J. Goud (eds.). Proceedings of the Ninth International Mal- acological Congress, Edinburgh, 1986. Unitas Malacolo- gica, Leiden. pp. 153-172. Harrison, R.G. 1991. Molecular changes at speciation. Annual Review of Ecology and Systematics 22:281—308. Hewitt, G. 1988. Hybrid zones - Natural laboratories for evolutionary studies. Trends in Ecology and Evolution 3:158-167. Hillis, D. M. and C. Moritz. 1990. An overview of applications of molecular systematics. In: Hillis, D. and C. Moritz (eds.) Molecular Systematics. Sinauer Associates, Inc. Sunder- land, MA pp. 502- 515. Hoagland, K. E. and G. M. Davis. 1987. The succineid snail fauna of Chittenango Falls, New York: Taxonomic status with comparisons to other relevant taxa. Proceedings of The Academy of Natural Sciences of Philadelphia 139: 465-526. Hornbach, D. J., M. J. McLeod, S. I. Guttman and S. K. Seilkop. 1980. Genetic and morphological variation in the fresh- water clam, Sphaerium (Bivalvia: Sphaeriidae). Journal of Molluscan Studies 46:158-170. Johnson, M. S., O. C. Stine and J. Murray. 1984. Reproductive compatibility despite large-scale genetic divergence in Ce- paea nemoralis. Heredity 53:655-665. Johnson, R. I. 1970. The systematics and zoogeography of the Unionidae (Mollusca: Bivalvia) of the southern Atlantic slope region. Bulletin of the Museum of Comparative Zo- ology at Harvard. 140:263-450. Kat, P. W. 1983a. Genetic and morphological divergence among nominal species of North American Anodonta (Bi- valvia; Unionidae). Malacologia 23:361-374. G. M. Davis, 1994 Kat, P. W. 1983b. Genetics, morphological divergence, and speciation among Lampsilis. Journal of Molluscan Studies 49:133-145. Kirpatrick, M. and R. K. Selander. 1979. Genetics of speci- ation in lake whitefish in the Allegash Basin. Evolution 33: 478-485. Landergren, U., R. Kaiser, C. T. Caskey and L. Hood. 1988. DNA diagnostics - molecular techniques and automation. Science 242:229-237. Li, W. H., M. Tanimura and P. M. Sharp. 1987. An evaluation of the molecular clock hypothesis using mammalian DNA sequences. Journal of Molecular Evolution 25:330-342. Mayr, E. 1963. Animal Species and Evolution. Harvard Uni- versity Press, Cambridge, MA. McCracken, G. F. and R. K. Selander. 1980. Self-fertilization and monogenic strains in natural populations of terrestrial slugs. Proceedings of the National Academy of Sciences 77:684-688 Mulvey, M., M. C. Newman and D. S. Woodruff. 1988. Ge- netic differentiation among West Indian populations of the schistosome-transmitting snail Biomphalaria glabrata. Malacologia 29:309-317. Murray, J., O. C. Stine and M. S. Johnson. 1991. The evolution of mitochondrial DNA in Partula. Heredity 66:93-104. Nei, M. 1972. Genetic distance between populations. Amer- ican Naturalist 106:283-292. Nei, M. and D. Graur. 1984. Extent of protein polymorphism and the neutral mutation theory. In: Hecht, M., B. Wallace and G. Prance (eds.) Evolutionary Biology 17:73-118. Nelson, G. 1989. Species and taxa: Systematics and Evolution. In: Otte, D. and J. Endler (eds.) Speciation and its Con- sequences. Sinauer Associates, Inc. Sunderland, MA pp: 60-81. Nelson, G. and N. Platnick. 1981. Systematics and Biogeog- raphy: Cladistics and Vicariance. Columbia University Press, New York. 567 pp. Nevo, E. 1991. Evolutionary theory and process of speciation and adaptive radiation in subterranean mole rats, Spalax ehrenbergi superspecies, in Israel. In: M. Hecht, B. Wal- lace and R. Macintyre (eds) Evolutionary Biology 25:1- 125. Nevo, E. and A. Beiles. 1988. Genetic parallelism of protein polymorphism in nature: ecological test of neutral theory of molecular evolution. Biological Journal of the Linnean Society 35:229-245. Osborn, H. F. 1918. The Origin and Evolution of Life. Scrib- ner’s Sons, New York. 322 pp. Patterson, H. E.H. 1985. The recognition concept of species. In: Vrba, E. S. (ed.) Species and Speciation. Transvaal Museum Monograph No.4:21-29. Rosen, D. 1978. Vicariant patterns and historical explanation in biogeography. Systematic Zoology 27:159-188. Rosenberg, G. 1989. Phylogeny and evolution of terrestriality of the Atlantic Truncatellidae. Ph.D. Dissertation. Har- vard University. Rosenberg, G., G. S. Kuncio, G. M. Davis and M. G. Harasew- ych. 1993. Ribosomal RNA phylogeny of selected gas- tropod and unionacean bivalve mollusks. In: Harasewych, M. G. and S. Tillier (eds.) Molecular Techniques and Mol- Page 23 luscan Phylogeny: Proceedings of a symposium held at the Eleventh International Malacological Congress, Siena, 1992. The Nautilus Supplement 2:111-121 (This volume). Selander, R. K. and D. W. Kaufman. 1973. Self-fertilization and genetic population structure in a colonizing landsnail. Proceedings of the National Academy of Sciences. 70:1186- 1190. Selander, R. K. and H. Ochman. 1983. The genetic structure of Pulmonates as illustrated by Molluscs. In: Isozymes: Current Topics in Biological and Medical Research 10:93- 123. Alan R. Liss Inc.,N.Y. Sene, F. M. and H. L. Carson. 1977. Genetic variation in Hawaiian Drosophila. 1V Allozymic similarity between D.silvestris and D. heteroneura from the island of Hawaii. Genetics 86:187-198. Skibinski, D. and R. D. Ward. 1982. Correlations between heterozygosity and evolutionary rate of proteins. Nature 298:490-492. Stanley, S. M. 1979. Macroevolution. W. H. Freeman, San Francisco, CA 364 pp. Stine, O. C. 1989. Cepaea nemoralis from Lexington, Vir- ginia: The isolation and characterization of their mito- chondrial DNA, the implications for their origin and cli- matic selection. Malacologia 30:305-315. Swofford, D. L. 1983. PAUP: Phylogenetic Analysis Using Parsimony. Computer Software. Illinois Natural History Survey, Champaign, Illinois. Swofford, D. L. and G. J. Olsen. 1990. Phylogeny Reconstruc- tion. In:Hillis, D. and C. Moritz (eds.) Molecular System- atics. Sinauer Ass. Sunderland, MA pp. 411-501. Templeton, A. R. 1989. The meaning of species and speci- ation: a genetic perspective. In: D. Otte and J. Endler (eds.) Speciation and its Consequences. Sinauer Associates, Inc., Sunderland, MA pp. 3-27. Templeton, A. R. 1980. Modes of speciation and inferences based on genetic distances. Evolution 34:719-729. Templeton, A. R. 1981. Mechanisms of speciation - a popu- lation genetic approach. Annual Review of Ecology and Systematics 12:23-84. Vainola, R. and M. M. Huilsom. 1991. Genetic divergence and a hybrid zone between Baltic and North Sea Mytilus populations (Mytilidae: Mollusca). Biological Journal of the Linnean Society 43: 127-148. Wiley, E. O. 1981. Phylogenetics. J. Wiley, New York. Woodruff, D.S. 1989. Genetic anomalies associated with Cer- ion hybrid zones: the origin and maintenance of new elec- tromorphic variants called hybrizymes. Biological Journal of the Linnean Society 36:281-294. Woodruff, D., K. C. Staub, E. S. Upatham, V. Viyant and H. C. Yuan. 1988. Genetic variation in Oncomelania hu- pensis: Schistosoma japonicum transmitting snails in Chi- na and the Philippines are distinct species. Malacologia 29:347-361. Wright, S. 1978. Evolution and the Genetics of Populations. Vol.4. The University of Chicago Press. 580 pp. Zimmer, E. A., S. L. Martin, S. M. Beverly, Y. W. Kan and A. C. Wilson. 1980. Rapid duplication and loss of genes coding for the chains of hemoglobin. Proceedings of the National Academy of Science, USA. 77:2158-2162. 7 = ‘t inieinl, Mey ees THE NAUTILUS, Supplement 2:25-43, 1994 Page 25 The Evolutionary Consequences of Restrictions on Gene Flow: Examples from Hydrobiid Snails D. J. Colgan W. F. Ponder The Australian Museum P.O. Box A285 Sydney South, Australia 2000 ABSTRACT The evolutionary consequences of restrictions on gene flow are discussed in relation to the population genetic structure and speciation of four Australian hydrobiid snail faunas. The stud- ied faunas comprise: (1) the brackish water genus Tatea; (2) species of Fluvidona in freshwater streams at Wilsons Prom- ontory; (3) species of Fonscochlea and Trochidrobia in artesian springs near Lake Eyre in South Australia; and (4) species of an undescribed genus at Dalhousie Springs in northern South Australia, another arid zone artesian spring complex. Gene flow in these hydrobiids is very variable. It is high in Tatea, relatively high at Dalhousie Springs and extremely low in Wilsons Prom- ontory and the Lake Eyre springs. Levels in the latter two faunas are similar despite a great disparity in geographic area. In these four faunas, the detail of gene flow patterns is complex, emphasising the dependence of population structure on the interaction of current and historical factors. This is illustrated by speciation patterns, the numbers of species and their dis- tributions usually correlating well with observed levels of gene flow. However, there are examples, among species groups with comparable, but very low gene flow, in which some taxa have undergone speciation yet others have not. The data were analysed using F-statistics and the private allele frequency approaches. Whilst the qualitative conclusions from the two approaches were generally similar, the exceptions usually indicated (on biogeographic grounds), that the F-sta- tistics approach is the more reliable estimator of gene flow. The private alleles approach is dependent on a coincidence of the scale of sampling with the biological scale of population sub- division. Intensive sampling schemes, as utilised in our studies, tend to find even rare alleles in more than one population even though they may be quite restricted in geographical distribu- tion. An analytical method for treating conditional allelic fre- quencies would not be as sensitive to this problem as the private alleles approach. Key words: Gene flow, snails, freshwater, isozymes, Hydro- biidae, evolution, population structure. INTRODUCTION The evolutionary fate of populations is largely deter- mined by two sets of factors. The first may be charac- terised as “coping factors’ —those determining the sur- vival of a population. This set includes external pressures, such as predation, parasitism and disease, extremes of, or changes in, climate, competition from other species or resource diminution. It also includes endogenous prop- erties such as the amount of local inbreeding or the ca- pacity of the breeding system to engender genetic re- combination. The second set may be characterised as “isolating factors’ —those factors which affect the evo- lutionary history of a population relative to other, orig- inally con-specific, populations. In general, this history depends on how effective gene flow is in overcoming differentiation inevitably arising from genetic drift or responses to local selection. Conversely, speciation pro- cesses are contingent on an “evolutionarily sufficient” restriction of gene flow between populations that have successfully accommodated the first set of factors. These two sets of factors are, however, also inter-related. For instance, it may be only through the introduction of a novel gene from another area that a population is enabled to withstand a climatic change. Many studies have shown that speciation is directly related to reductions in gene flow (reviewed by Grant, 1980; Porter, 1990). It is difficult, however, if not im- possible, to predict what restrictions on gene flow over what period of time constitute evolutionarily significant barriers. The importance of the evolutionary conse- quences of restriction on gene flow is such that its esti- mation remains a goal of many experimental studies (e.g. Skibinski et al., 1983; Waples, 1987; Johnson et al., 1988; Mitton et al., 1989; Arter, 1990; Porter, 1990; Preziosi & Fairbairn, 1992). There is also continuing interest in the development of mathematical models for the analysis of gene flow (e.g. Slatkin, 1985a; Barton & Slatkin, 1986; Slatkin & Barton, 1989) and/or the the genetical sub- division of species (Cavalli-Sforza & Feldman, 1990). We have been investigating the hydrobiid gastropod faunas of a variety of habitats to characterise their tax- onomic and population genetic structure. We were par- ticularly interested in how biological and environmental differences between these faunas are reflected in the Page 26 degree of genetic divergence and the amount of genetic exchange between populations, and in the evolutionary consequences of those differences. A priori, factors such as the biological ability to disperse, the geographic scale of the system, weather patterns, topography and the ac- cessibility and suitability of the habitat for potential bi- ological dispersal agents might all be expected to play a part in determining levels of gene flow. Gene Flow In this paper we define gene flow as the genetically- effective transmission of alleles between extant discrete populations and between various parts of the range with- in species in which population boundaries cannot be dis- cerned or do not occur. We do not regard re-colonization after local extinction as an example of gene flow (agree- ing with Endler, 1973; Grant, 1980;—but contrast Slatkin 1985b). Re-colonization, or original colonization simply increases the number of populations of a species, other processes being required for phylogenetic consequences. Generally, continued gene exchange will homogenize original and derived populations. Significant differenti- ation requires persistent marked reductions in gene flow. Any estimate of gene flow made on the basis of allelic frequencies confounds factors operating during two dis- tinct phases of the differentiation process. Firstly, the establishment of a population implies a sampling process which may cause differences in frequency arrays (Carson & Templeton, 1984; Wool, 1987). Secondly, the differ- ences reflect subsequent patterns of gene flow and dif- ferential selection as well as any current trends. In the discussion below, we usually make the assumption that the comparisons between estimates of gene flow in dif- ferent biological situations reflect differences in only one of these two confounded factors in any given case. This may not always be an accurate description of biological reality. High allelic frequency differences may result from a divergent initial founder effect or from a sub- sequent reduction in gene flow, differential selection or any combination of these three factors. Direct methods of estimating gene flow, requiring observation of the mating success and/or fertility of known immigrant in- dividuals, largely overcome these problems. The labour and practical difficulies involved in making such esti- mates is such, however, that indirect methods are usually pursued (Slatkin, 1985b; Johnson et al., 1988). The Measurement of Gene Flow Both methods of estimating gene flow which are used here measure the parameter Nm, where N is the (effec- tive) size of each sub-population and m is the probability that a gamete in the offspring generation is an immigrant to the sub-population where it occurs. Three principal models of population structure have been developed as mathematical abstractions to provide a theoretical framework for measuring Nm: THE NAUTILUS, Supplement 2 (A) The Island model (Wright, 1931), in which each of an infinite number of discrete sub-populations receives migrants at random from other sub-pop- ulations. The geographic distance between pop- ulations does not affect the rate of gene flow be- tween them. The Stepping Stone model (Kimura & Weiss, 1964), in which gene flow occurs between a population and its immediate neighbours in one or two di- mensional geographic arrays. Gene flow does not occur directly between two populations which are not immediate neighbours. Models in which the population is considered to be continuously distributed in one or two dimen- sions, with the degree of genetic differentiation of individuals separated by a given distance de- termined by the levels of gene flow. The accuracy of these models’ approximation to pop- ulation structure will vary. If the studied species has high vagility, the requirement of the Island model that each sub-population exchanges migrants with all others will be a more accurate approximation than if the vagility is low. Conversely, the Stepping Stone model’s restriction on migration between populations which are not near neighbours is more likely to be accurate if the studied species has low vagility. & 9 F-Statistics The overall inbreeding coefficient can be partitioned into components reflecting non-random breeding (F\,) and the effects of between sub-population differentiation (F<) (Wright, 1951). Under the infinite island model of pop- ulation subdivision, if the migration rate is small, then (Wright, 1951): Fer = (1 + 4Nmy? A variety, indeed almost a plethora, of alternative meth- ods for the calculation of quantities very similar or iden- tical to Fg; have been suggested (Wright, 1951, 1978; Nei, 1973; Nei & Chesser, 1983; Cockerham, 1969; Weir & Cockerham, 1984). Many of these were developed in response to complications of the original two-allele per locus situation studied by Wright. It can be shown that these are usually encompassed by natural extensions of Wright’s approach. Others attempt to take account of relaxation of the simplifying assumptions (negligible se- lection, mutation, etc.) made in Wright's analyses. The various methods have been widely reviewed (e.g. Chak- raborty & Leimar, 1987; Weir, 1990) and the differences between them shown, generally, to be of second-order significance. Moreover, both analytical (Slatkin, 1985b) and simulation (Slatkin & Barton, 1989) studies tend to emphasise the qualitative similarities between F; vari- ables defined under either the Island or Stepping Stone Models. Where estimates of gene flow given below are based on Fez, this will be indicated by Nm(F¢,). D. J. Colgan and W. F. Ponder, 1994 Conditional Allelic Frequencies There are two main methods of analysing gene flow using the approaches of Slatkin (1981, 1985a). The first requires that the “occupancy rate’ for an allele be determined. This is the number of sub-populations in which the allele is found, divided by the total number of sub-populations. The conditional average frequency of the allele is the average of its frequencies in those sub-populations where it actually occurs. Levels of gene flow between sub-pop- ulations are visualised by graphing the conditional av- erage frequency of each allele against occupancy rate. Such representations can be useful in comparison of the levels of gene flow in different taxa (e.g. Govindaraju, 1989) but, in the absence of an analytical theory, can be used to provide numerical estimates only by making analogies with the results of computer simulations (John- son et al., 1988). In the second method, attention is re- stricted to “private alleles”, i.e. those found in only one sub-population. Slatkin’s simulations (1985a) found that, in both Island and Stepping Stone Models, the conditional average frequency of private alleles (p(1)) is approxi- mately linearly related to the migration rate by the ex- pression: logio(P(1)) = alogi(Nm) + b where a and b take values dependent on the number of individuals sampled from each sub-population. One claimed advantage of this approach is that its estimates of migration rates are theoretically only slightly depen- dent on mutation or the many types of selection which might operate (Barton & Slatkin, 1986; Slatkin & Barton, 1989). Estimates of gene flow based on the frequency of private alleles given below are designated as Nm(p(1)). The Family Hydrobiidae Small prosobranch snails of the world-wide family Hy- drobiidae are the most diverse freshwater gastropods, with nearly 400 generic names currently in use (Kabat & Hershler, 1993). Commonly, in Australia, freshwater Hydrobiidae occupy small streams or springs. The pop- ulations in these isolated or semi-isolated habitats show varying degrees of differentiation because of an apparent inability to disperse readily. A low level of dispersal may be possible, for example by birds or even flying insects (Rees, 1965; Boeters, 1979, 1982). Other genera inhabit brackish or estuarine waters. Ge- netic data have been used to test hypotheses based on morphological criteria in such taxa (Lassen, 1979; Davis et al., 1988, 1989; Ponder & Clark, 1988; Ponder et al., 1991). They tend to have large geographic ranges, partly because some have a planktonic marine larval phase, but also because they live in tidal marshland habitats where they are potentially readily transported by birds. In the remainder of this paper, we will concentrate on-our recent investigations of three freshwater hydro- biid radiations at Wilsons Promontory, in the Lake Eyre supergroup of the South Australian Mound Springs and Page 27 in the Dalhousie Springs complex at the north of South Australia. For comparative purposes, we will often refer to our studies of the brackish water (usually estuarine) genus Tatea (Ponder et al., 1991). The fauna of the Lake Eyre spring supergroup has been formally described (Ponder et al., 1989) and that of Dalhousie Springs has been briefly reported on with respect to shell morphology (Ponder, 1989). The Wilsons Promontory study will be described in detail in a forthcoming publication (Ponder et al., 1994). MATERIALS AND METHODS Summary information regarding collecting sites, etc. can be found in Appendix 2. More detail is provided for Tatea in Ponder et al. (1991), the Wilsons Promontory Fluvidona in Ponder et al. (1994), the Lake Eyre springs in Ponder et al. (1989) and on Dalhousie Springs in Zeidler and Ponder (1989). Genotypic data are available from the senior author. Data for Tatea and Fluvidona on allozymic frequencies, observed heterozygosities and the various environmental parameters which were mea- sured are given in Ponder et al. (1991, 1994). Similar data for the other faunas will be presented separately for each system. Standard methods for cellulose acetate electrophoresis were used (Hebert & Beaton, 1989, Ponder et al., 1991). Individual snails were homogenized with 10-30 ul (mean 20 ul) of buffer, providing enough sample for up to 12 gels. Because of their small size, it was not possible to examine each snail for all enzymes. Where more than one locus is shown below as encoding the same enzyme, each was designated numerically in order of decreasing mobility. Allozymes identified for each locus are desig- nated in the same way. The enzymes scored for each species, or species grouping, together with abbreviations used, Enzyme Commission Numbers, and number of presumptive loci are listed in Appendix 1. There were some differences between taxa in the number of loci that were electrophoretically interpretable. These are speci- fied in Appendix 1. The computer packages BIOSYS-1 (Swofford & Selander, 1981), NTSYS (Rohlf, 1990) and PHYLIP, version 3-4 (Felsenstein, 1989) were used to assist analysis. All taxonomic groupings treated here were initially analysed without assuming any hierarchical structure of the populations. F,;, conditional allozymic frequencies and private allelic frequencies were calculated. Popu- lations were then clustered in hierarchies, as described below and as detailed in Appendix 2. Components of overall genetic differentiation were obtained for this clus- tering using the WRIGHT78 step of BIOSYS. Allozymic frequencies in taxonomic units at each intermediate level of the clustering were calculated after pooling the data from the sub-units included in the same group. The pooled data were used to estimate Fs; values and conditional (and private) allelic frequencies for units at this inter- mediate level. This approach has a statistical tendency to reduce the variance in gene frequencies (and hence Page 28 Figure 1. Map of Wilsons Promontory showing major drainages and the distribution of Fluvidona species. The area covered is shown by the dot at the head of the arrow on the inset map of Australia. The other inset shows a detail of Whisky Creek. Locations of straight-sided snails are shown by O, the MPI 3 convex by Ml, MPI 4 by @ and MPI 5 by Q. Drainages: (1) Darby River; (2) Whisky Creek; (3) Squeaky Creek; (4) Tidal River; (5) Titania Creek; (6) Growler Creek; (7) Frasers Creek; (8) Roaring Meg; (9) Picnic Creek; (10) First Bridge Creek; (11) Freshwater Creek; (12) Blackfish Creek; and (13) China- mans Creek. increase the estimate of Nm) to a degree dependent on levels of variation between samples pooled into the same unit. If the variation is merely a sampling artefact then the procedure will increase accuracy of Nm estimation. But if the variation is due to biological subdivision of the populations, then the estimate should be regarded more as an upper limit on the degree of gene flow. A priori, it is not possible to decide which of these two alternatives is correct as we do not know what constitutes an effec- tively panmictic unit in these hydrobiids. A second effect of the pooling procedure is that alleles which are found in more than one population and hence not “private” in the original subdivision, may be regarded as private at higher clustering levels if they are there restricted to only one unit. Again, this reflects the uncertainty about the biological structure of the population. Pooling data may not always resolve this uncertainty but patterns in such analyses will usually be informative about popu- lation structure to at least some extent. The sample sizes used in the studies varied from locus to locus and from population to population. The average sample per locus is shown in Appendix 2. When Nm was estimated from the conditional frequencies of private alleles, the parameter values for a sample size of 25 were taken from Barton and Slatkin (1986). Alternative pa- THE NAUTILUS, Supplement 2 PN Kilometres Figure 2. Map of the Lake Eyre mound springs showing the distribution of Fonscochlea and Trochidrobia species. The area covered is indicated on the inset of Australia. The site of a spring group is indicated by an “x”. Presence of a species in the group is indicated by @ for F. accepta, © for F. aquatica, @ for F. zeidleri, a for F. billakalina, W for F. variabilis, @ for T. punicea, © for T. minuta and O for T. smithi. The spring groups are: (1) Freeling; (2) Outside; (3) Twelve Mile; (4) Strangways; (5) Billakalina; (6) Beresford /Warburton; (7) Coward/Jersey /Elisabeth/Kewson; (8) Blanche Cup; (9) Her- mit Hill; (10) Davenport; and (11) Welcome. rameter values did not, however, significantly affect nu- merical estimates of gene flow. THE STUDY AREAS Wilsons Promontory Wilsons Promontory (Figure 1), the southern-most part of the Australian mainland (39°S, 146°28’W), consists of granite hills up to 754 m with many permanent streams and rivers fed by high rainfall (>1000 mm per year). Its geological and climatological setting are summarised by Wallis (1988) and Schmidt and Thornton (1992). The hydrobiid fauna (genus Fluvidona) of the Promontory comprises two endemic morphologically-recognisable species, the shells of one with straighter whorl! outlines and one with more convex whorls. The latter morpho- species is divisible into three genetic species by very nearly fixed sympatric differences in the MPI phenotype, referred to as the MPI 3, MPI 4 and MPI5 genetic species. For hierarchical analyses of population structure, sites were grouped within streams, streams within catchments and catchments within species, providing two interme- diate levels (streams, catchments) in an analysis. Popu- lations in which hybrids were seen were ignored in the D. J. Colgan and W. F. Ponder, 1994 analyses examining gene flow within genetic species (see Appendix 2). South Australian Mound Springs The springs, fed from the Great Artesian Basin of Aus- tralia, are of considerable limnological and conservation significance to the very arid area in which they occur (Ponder, 1986; Harris, 1993). The springs generally lie on the fringes of the Basin, where the aquifers abut impervious rock or lie near the surface. The Lake Eyre Supergroup, the most extensive group of springs associ- ated with the Great Artesian Basin, extends about 400 km between Marree and Oodnadatta and provides vir- tually the only permanent water in the area. The area with springs is only rarely more than 20 km wide, so that the supergroup conforms quite well to a one di- mensional, discontinuous model. The geological history of these springs is not well known. Estimates of the ages of some large (extinct) mounds range from late Miocene to Recent. These are probably at least Pleistocene (Wopf- ner & Twidale, 1976; Williams & Holmes, 1978; Thomp- son & Barnett, 1985) but the springs have probably been in the area much longer. The taxa presently inhabiting the springs may be relicts of more widespread forms from a generally wetter period in the Neogene or may represent faunas associated with these artesian springs through much of the Tertiary. The predominant drainage pattern in the Lake Eyre basin is at right angles to the line of springs. Hence the transport of snails between spring groups by floods would be unlikely—although such transport could occur within spring groups. Spring nomenclature and grouping used in this paper follows Ponder et al. (1989). The hydrobiid fauna consists of two endemic genera, Fonscochlea (five species) and Trochidrobia (four species) (Ponder et al., 1989) For hierarchical analyses, springs were compared within spring groups. The groups were then collected into “clusters” (see Figure 2, Appendix 2), the Southern cluster comprising springs between Welcome Springs and Hermit Hill, the Middle cluster comprising springs be- tween the Blanche Cup complex and Strangways and the Northern comprising Outside, Twelve Mile and Freeling Springs. If species were found in more than one of these clusters, a second intermediate level was in- cluded in the hierarchy. This could not, however, be done for all species, Fonscochlea accepta, for instance being found only in the Southern cluster. The Southern cluster was divided into three groups: (1) Welcome Springs; (2) Davenport Springs; and (3) the Hermit Hill springs. The Middle cluster was divided into five groups: (1) Blanche Cup springs, (2) the group consisting of Cow- ard, Kewson, Elizabeth and Jersey springs, (3) Billakal- ina; (4) Beresford and Warburton Springs; and (5) Strang- ways Springs. The Northern cluster was divided into two groups: (1) Twelve Mile and Outside springs; and (2) Freeling Springs. Page 29 Kilometres Figure 3. Map of the Dalhousie Springs showing the distri- bution of the globular (4), pupiform (@) and Fluvidona-like snails (%). Spring groups are identified by letters near dashed boundaries. Dalhousie Springs This complex is another large group of arid zone springs in northern South Australia associated with the Great Artesian Basin (Figure 3). Aspects of its geology and biology are surveyed in a number of papers in Zeidler and Ponder (1989). The many springs in the complex occupy an area of about 70 km?. They range in size from small nascent or senescent seeps, to actively flowing and, at the upper end of the size scale, to outlets (of about 140 L/sec) which feed pools of 50 m or more in width with outflow channels supporting wetland vegetation for up to 15 km (Smith, 1989). Their combined discharge accounts for 90-95% of the total produced by all South Australian artesian springs (Smith, 1989), and 41% of the overall output from Great Artesian Basin springs (Ha- bermehl, 1982). They are well separated from springs of the Lake Eyre supergroup, the northernmost population of species from that region (F. zeidleri) being 140 km away. These springs are likely to be early Pleistocene in age (Krieg, 1989). Minor local overflow due to rare heavy rain may facilitate interspring transport. Major flooding is unlikely (Kotwicki, 1989). Spring nomenclature and groupings used in this paper follow Zeidler and Ponder (1989), except as specified below. Eight main groups of springs, designated A to H are recognised (Figure 3, Appendix 2). C is divided into four sub-groups, and D into two. Herein, group H will be treated as comprising two groups, because H3 is well separated from H1. We have also split E into two sub- groups containing, respectively, (1) E5 and El and (2) Page 30 THE NAUTILUS, Supplement 2 Table 1. Estimates of Nm derived from the average frequency of private alleles or from the average F,, at various clustering levels. The overall estimates assume no population hierarchy. The next two columns are estimates from data pooling within the first intermediate hierarchical level and the final two columns are for pooling within the second hierarchical level (where applicable). The upper figure in each cell is the observed value of the variable. The lower figure is the value of Nm calculated from the observation. Level 1 pooling Level 2 pooling Overall Species For pi) Fluvidona (straight-sided) 0.130 0.444 0.776 0.3138 Fluvidona (convex) MPI 3 0.086 0.535 0.872 0.217 MPI 4 0.075 0.681 1.104 0.117 MPI 5 0.161 0.240 0.295 0.791 Fonscochlea accepta 0.116 0.207 0.525 0.958 F. aquatica 0.110 0.781 0.570 0.070 F. zeidleri 0.360 0.728 0.074 0.093 F. variabilis 0.196 0.792 0.211 0.066 F. billakalina 0.044 0.263 2.769 0.700 Trochidrobia punicea 0.143 0.600 0.363 0.167 T. smithi 0.426 0.730 0.055 0.092 T. minuta 0.086 0.375 0.872 0.417 Dalhousie (globular) 0.159 0.321 0.302 0.529 Dalhousie (pupiform) 0.180 0.413 0.244 0.355 Dalhousie (Fluvidona-like) 0.296 0.670 0.104 0.123 E2, E7 and E8. Generally, for hierarchical analyses, springs were clustered into sub-groups, sub-groups into groups and groups into the species. Three hydrobiid spe- cies were recognised in our genetic studies. The globular and pupiform species belong to an undescribed endemic genus, the third species (also endemic) being tentatively included in the widespread genus Fluvidona. Separate analyses were performed for each species. RESULTS The levels of gene flow in the four Fluvidona taxa from Wilsons Promontory are extremely low as shown by the overall F-statistics in Table 1. The smallest F,; value is for the convex MPI 5 genetic species. At 0.240, this is, F sr p(1) F sr p(1) 0.120 0.375 0.108 0.348 0.491 0.417 0.589 0.468 0.110 0.523 0.570 0.228 0.199 0.628 0.081 0.462 0.205 0.148 0.967 0.291 0.091 0.282 0.073 0.254 0.791 0.637 1.157 0.734 0.028 0.110 6.036 2.023 0.128 0.756 0.149 0.412 0.439 0.081 0.338 0.357 0.289 0.750 0.155 0.625 0.108 0.083 0.316 0.150 0.300 0.783 0.173 0.688 0.101 0.069 0.261 0.113 0.226 0.300 0.165 0.583 0.098 0.580 0.160 0.470 0.696 0.175 0.299 0.282 0.265 0.584 0.224 0.346 0.125 0.178 0.167 0.472 0.244 0.429 0.144 0.381 0.038 0.105 0.028 0.082 3.565 2.131 6.036 2.799 0.015 0.296 0.014 0.228 17.706 0.595 19.942 0.846 0.300 0.492 0.101 0.258 however, near the higher end of the range previously found for gastropods over comparable geographic scales (Gould & Woodruff, 1986, 1990; Johnson et al., 1988). The overall Fs; values for the straight-sided Fluvidona species and the convex MPI 3 and MPI 4 genetic species are very high. The levels of migration suggested by these values range down to 0.117 for the MPI 4 genetic species, implying that the fraction of a deme which is replaced by immigrants each generation (m = 0.117/N) is very small. The estimates of Nm(F,,) values based on the pooling of data from individual samples may be com- plicated by the likelihood that the pooled data do not represent single populations. The trends in the estimates are, however, very similar to those based on single pop- ulations. Those for the MPI 4 genetic species are higher than those for other Fluvidona taxa, but still suggest that D. J. Colgan and W. F. Ponder, 1994 Page 31 Table 2. Variance components in the hierarchical F-statistic analyses. The Fy figures indicate that variance ascribable to variation in the specified X variable (e.g., population) within the specified Y variable (e.g., spring group). Where two intermediate levels are used for a species in the hierarchy, all six cells are filled. Where one level is used only three cells are filled. The top figure in each cell is the calculated Fy, value and the bottom, the percentage of the total variance comprised by this. X variable: Population Population Y variable: Level 1 Level 2 Species Fluvidona (straight-sided) 0.226 0.434 13 27 Fluvidona (convex) MPI 3 0.005 0.374 0 7 MPI 4 0.152 0.539 6 20 MPI 5 0.052 0.190 6 20 Fonscochlea accepta 0.122 35 F. aquatica 0.217 0.744 7 23 F. zeidleri 0.346 0.677 12 23 F. variabilis 0.382 0.599 1 19 F. billakalina 0.101 22 Trochidrobia punicea 0.118 0.262 6 13 T. smithi 0.745 48 T. minuta 0.059 8 Dalhousie (globular) 0.288 0.303 33 35 Dalhousie (pupiform) 0.304 0.363 24 28 Dalhousie (Fluvidona-like) 0.538 43 Variance components Population Level 1 Level 1 Level 2 Total Level 2 Total Total 0.426 0.269 0.257 0.015 26 17 16 ] 0.570 0.371 0.567 0.312 26 7 26 14 0.679 0.456 0.622 0.305 25 16 23 ll 0.255 0.145 0.214 0.081 Qi 15 23 9 0.173 0.058 49 16 0.771 0.673 0.707 0.105 24 21 22, 3 0.718 0.506 0.568 0.127 24 We 19 4 0.781 0.352 0.646 0.454 24 11 20 14 0.220 0.1383 48 29 0.574 0.164 0.518 0.423 28 8 25 21 0.719 0.100 46 6 0.358 0.318 49 43 0.284 0.021 — 0.006 —0.027 33 2 =Il =8) 0.384 0.085 0.115 0.032 30 U 9 2 0.656 0.049 53 4 Nm is less than one. Values for MPI 4 are all higher than for the individual sample estimation, marginally so for the pooling of samples within tributaries and notably for the pooling into catchment based units. Even so, the data suggest that a catchment receives less than one migrant from another catchment in every three generations. Restrictions on gene flow are also suggested by analyses of the conditional frequency of private allozymes. Es- timates of migration rates based on these data are much greater than those based on Fe; and differ in the relative rates ascribed to the different taxa. The latter situation is particularly notable in MPI 4 which apparently has the highest rate of inter-population migration among all four species, whereas its Nm(F¢,) is the lowest. The components of variance due to differentiation be- tween taxonomic units at different hierarchical levels are presented in Table 2. Although comparison of these val- ues is complicated by varying proportions of the popu- lations being pooled at each level, some trends can be observed. Particularly striking is the concordance be- tween the three MPI genetic species, where in each case almost half of the variation is explained by differences between tributaries or between catchments. This con- trasts with the straight-sided Fluvidona where only one third of the variability is explained by such differences. We have investigated gene flow in eight of the nine Lake Eyre mound springs hydrobiids, Trochidrobia in- flata being found in only two of our sample sites. As can be seen in Figure 2 and Appendix 2, the distributions of these species vary markedly in size. Fonscochlea accepta is restricted to the Southern cluster of springs, F. billak- alina to the central cluster and T. minuta to the northern. The range of the other species extends into more than one spring cluster, with F. zeidleri and F. variabilis being found in all three clusters. The apparent levels of gene flow between the populations of the species reflect this variability in range. F. accepta has high gene flow, with Nm(F.,) between spring groups being more than two. Conversely, gene flow in F. aquatica, the snail which is an ecological replacement for F. accepta in the central Page 32 THE NAUTILUS, Supplement 2 Table 3. Estimates of F,,; or Gs; in gastropods. Measures using Gs; (Nei, 1973) are indicated by an asterisk. Geographic scale is the distance between the extremes of the sampled range. References are: (1) Johnson and Black (1984a,b); (2) Brown (1991); (3) Mitton et al. (1989); (4) Campton et al. (1992); (5) Grant and Utter (1988); (6) Day (1990); (7) Chambers (1980); (8) Jarne and Delay (1990); (9) Mulvey et al. (1988); (10) Bandoni et al. (1990); (11) Johnson et al. (1988); (12) Gould and Woodruff (1986); and (13) McCracken and Brussard (1980). Species and reference No. of loci Marine species (1) Siphonaria jeanae 4 4 (2) Haliotis rubra 12 (3) Strombus gigas 7 (4) Strombus gigas 4 4 (5) Nucella lamellosa 2 2 (6) Nucella lapillus 8 8 8 Freshwater species (7) Goniobasis (2 species) 14 (8) Lymnaea peregra 6 (9) Biomphalaria glabrata 13 (10) Biomphalaria pfeifferi W Terrestrial species (11) Partula taeniata 17 P. suturalis 16 (12) Cerion (New Providence) 8 (18) Triodopsis albolabris 2 and northern clusters, is quite low, Nm being 25 times less between spring groups. In these large aquatic Fon- scochlea, approximately the same relative levels of gene flow are indicated by the estimates derived from con- ditional allelic frequencies. Using Fs, for estimation, a similar pattern is shown in comparisons of the smaller aquatic species F. billakalina and F. variabilis, with gene flow between spring groups in the former being nine times the level in the latter. In Trochidrobia, gene flow inferred from Fy statistics between spring groups in T. minuta is twice as high as it is in the other two species of the genus. In these latter two sets of comparisons, however, the estimates derived from conditional allelic frequencies do not show the same pattern as the F.; estimates. F. billakalina has a similar Nm(F,) to F. var- iabilis and the value for T. punicea is almost five times as great as that for T. minuta. The components of variation due to different hierar- chical levels are strikingly similar in the aquatic F. aqua- tica and the amphibious F. zeidleri. There is some dis- agreement as to the level of Nm(Fs;) between spring clusters for these species, but otherwise estimates of inter- population migration in these species are remarkably concordant. The concordance is significantly less for Nm(p(1)). The components of variation are also similar in two other species (F. variabilis and T. punicea) from the Lake Eyre mound springs. Interestingly, the pattern No. of samples Scale (km) Fg, mean 1 10 0.002 28 2,500 0.004 18 5,000 0.022 21 5,000 0.076 4 500 0.011* 14 5,000 0.023* 12 0.1 0.021 30 1,000 0.286 6 0.5 0.015 10 10 0.092 15 20 0.195 12 1,000 0.408 4 50 0.018 6 1,000 0.805 12 500 0.589 22 20 0.279 23 20 0.168 36 30 0.143 i 500 0.255 of variation of this pair differs from that of the two large aquatic Fonscochlea, showing much greater between- spring divergence. Estimated Nm(F¢,) between populations at Dalhousie Springs is in the higher reaches of the ranges observed in these studies, in both the pupiform and globular spe- cies. This trend is even more marked for inter-group migration as assayed at higher hierarchical levels. Inter- spring subgroup migration is higher in the globular snails than in any other level-one pooling, and the estimate for the pupiform snails is exceeded only by F. accepta and F. billakalina. The calculated migration rates betwen spring groups are higher for both the globular and pup- iform Dalhousie radiations than for any other taxon in our studies. As expected, the proportion of variation ex- plained by differences at the lower hierarchical levels is very high in comparison to our other studies. Indeed, virtually all of the variation in the globular snails is due to differentiation of populations within spring-subgroups, within spring groups, or within the overall spring com- plex (and not of spring-subgroups within groups, etc.). The levels of gene flow estimated from Nm/(p(1)) for the two higher hierarchical levels are extremely high in the context of the present results. To the extent that these estimates are credible, they reinforce the suggestion that gene flow is high in the Dalhousie Springs complex, at least as compared to the other study sites. D. J. Colgan and W. F. Ponder, 1994 DISCUSSION Our hydrobiid studies emphasise the dependence of pop- ulation structure on a wide range of interacting biological and environmental factors which must be considered in historical terms. The three hydrobiid faunas we have treated extensively in this paper, and the previously stud- ied Tatea (Ponder et al., 1991) differ (either certainly or probably) in such biological characteristics as size, thermal and salinity tolerances and desiccation resis- tance. These various hydrobiids also occupy different types of habitat. Tatea occupies essentially continuous habitat. Wilsons Promontory Fluvidona and the Dal- housie Springs snails have habitats which are discontin- uous, with relatively small distances between suitable areas. The Lake Eyre fauna is in discontinuous, widely separated habitat. Before discussing our results in detail, some prelimi- nary comparisons may be made to reinforce that gene flows actually do differ between the faunas. The Wilsons Promontory radiation occupies a much smaller geo- graphic area than the Lake Eyre fauna, with the excep- tion of Fonscochlea accepta. The markedly higher levels of gene flow in F. accepta might suggest that the un- derlying processes are more efficient in this taxon. Con- versely, such relativity emphasises the very restricted levels of gene flow in the Wilsons Promontory snails. This is also suggested by comparisons of these Fluvidona spe- cies with the Lake Eyre F. aquatica, F. variabilis and F. zeidleri. The MPI 4 and MPI 3 genetic species do have a slightly higher Nm(F,,;) than the Fonscochlea species but this is a minor difference when the disparity between their ranges is considered. These Fonscochlea species each have linear extents well over an order of magnitude larger than the MPI 3 and MPI 4 Fluvidona (over 200km as opposed to less than 20km). Population Structure in Quasi-Continuous Aquatic Habitats Although the vagaries of ocean currents may reduce gene flow to or from a particular area (Todd et al., 1988; Mitton et al., 1989), marine snails with planktonic (and especially planktotrophic) larvae have wide natural dis- tributions in which variation between local populations is only a minor fraction of overall genetic diversity (Table 3). This pattern is found in Nassarius. obsoletus (Gooch et al., 1972), Littorina littorea (Berger, 1973; Janson, 1987), Siphonaria jeanae (Johnson & Black, 1984a, 1984b), the species with planktotrophic larvae among the Crepidula studied by Hoagland (1984), Strombus gigas (Mitton et al., 1989; Campton et al., 1992) and Haliotis rubra (Brown, 1991) which has a lecithotrophic larva. Species distributions in these marine gastropods tend to be either widely separated, often in conjunction with geographic barriers to gene flow, or to be broadly sympatric (e.g. Littorina - Berger, 1973; Janson, 1987; Crepidula - Hoagland, 1984) presumably reflecting past allopatric speciation and subsequent dispersal into sym- Page 33 patry. The latter pattern was observed in our studies of Tatea (Ponder et al., 1991). This genus, predominantly estuarine with an assumed free-swimming larval stage, has a very wide distribution, being found throughout temperate Australia. Its two species T. rufilabris and T. huonensis are sympatric over virtually all of this range. There are exceptions to these patterns of speciation in some groups with specialised feeding patterns which have high species densities (Vermeij, 1987). Groups with direct larval development or brooding should have reduced gene flow, greater differentiation and, conceivably, higher likelihood of speciation. The first two predictions have been borne out in studies of Littorina saxatilis (Snyder & Gooch, 1973; Janson, 1987) and Nucella lamellosa (Grant & Utter, 1988). That a short planktonic phase increases rates of speciation is less certain. Littorina saxatilis does have a number of closely- related sibling species (Janson, 1987; Johannesson, 1988; Sundberg et al., 1990) and N. lamellosa may represent a species complex (Grant & Utter, 1988). However, N. lamellosa, as presently recognised, has one of the largest geographic ranges of North Pacific gastropods. Population Structure in Discontinuous Aquatic Habitats There have been major investigations of gene flow and the genetic structure of two groups of the freshwater gastropods in the caenogastropod genus Goniobasis. Chambers (1978, 1980) investigated species from Florida and Dillon and Davis (Dillon & Davis, 1980; Dillon, 1984) those from the border regions of Virginia-North Caro- lina. Three main results are relevant. (1) There is a high degree of genetic divergence between populations within the same drainage system indicating low levels of gene flow. (2) There are larger differences between drainage systems, reflecting an even smaller likelihood of inter- drainage gene flow. (3) Identified taxa tend to remain allopatric or parapatric, with geographically-restricted ranges, suggesting that dispersal after speciation is lim- ited. Dillon (1988) provides direct information on rates of gene flow in transplanted G. proxima populations. These are about 15-20 m upstream and 5-10 m down- stream (per year), the discrepancy in movement rates being caused by the behavioural tendency of freshwater (Dillon, 1988) (and even riparian—Arter, 1990) snails to crawl upstream in compensation for down current drift. The findings for Goniobasis are not true of all fresh- water snails, as shown by studies of genetic variation in basommatophoran pulmonates. Dispersal in these snails is often assisted by self-fertilisation (Mimpfoundi & Greer, 1989; Bandoni et al., 1990) and species distributions are generally wide-ranging. Measurement of gene flow is often hampered because of extremely low levels of ge- netic variation (Mimpfoundi & Greer, 1989; Jarne & Delay, 1991). Where flow can be assessed, as in Biom- phalaria straminea (Woodruff et al., 1985), B. came- runensis (Mimpfoundi & Greer, 1990), B. pfeifferi (Ban- doni et al., 1990) or Bulinus cernicus (Rollinson et al., Page 34 1990), evidence of substantial population differentiation is usually found, albeit at a geographic scale much larger than in Goniobasis. Genetic variation in Biomphalaria glabrata from the Caribbean is, however, primarily due to inter-island differentiation (78%), with only 2% being due to intra-island differences (Mulvey et al., 1988). Each of the three main observations on Goniobasis is applicable to the Wilsons Promontory Fluvidona but they are less accurate descriptions of the two artesian spring faunas, indicating that they are not generally character- istic of freshwater dioecious gastropods. At Dalhousie Srings, levels of inter-spring and inter-spring group dif- ferentiation are not high and the globular and pupiform species are sympatric over a substantial range. There are instances in the Lake Eyre fauna where sister-species remain essentially allopatric (e.g. F. accepta and F. aqua- tica), but these are the exceptions. Habitat Stability and Area Effects in Wilsons Promontory Habitat on Wilsons Promontory has probably been re- duced during periods of aridity. The Fluvidona popu- lation structure may still be showing distortions caused by recent aridity-induced interruptions of migration, particularly in the more upland MPI 3 and MPI 4 genetic species. Modelling suggests that the re-attainment of structural equilibrium following disruption is of the order of hundreds of generations rather than thousands (or more) for both the Fs; (Crow & Aoki, 1984) and p(1) (Slatkin & Barton, 1989) approaches. The re-attainment can, of course, only begin after the disruption is halted. In the MPI 4 genetic species there are several geographic groupings broadly definable by catchment. Gene flow between these is low, with Nm(Fs,) values of 0.148 for populations pooled within tributaries and 0.291 within catchments. Following a period of aridity, increased lev- els of migration in wetter times may overcome incipient genetic divergence, unless the period were so prolonged that the isolates undergo speciation. Such considerations reinforce the importance of modelling the impact of non- equilibrium population structures on measures such as Fey (e.g. Whitlock (1992). They also suggest comparison with “area effects’ and their various evolutionary con- sequences. Area effects were initially recognised by their distinc- tive phenotypic frequency arrays sharply clinally de- marcated from neighbouring areas (Cain & Currey, 1963) and were subsequently observed for allozymic frequen- cies (Ochman et al., 1983; Johnson et al., 1984). Area effects have been correlated with the rapid expansion of relict populations (Cameron & Dillon, 1984; Ochman et al., 1983; Johnson et al., 1984). They were initially pre- sumed to be stable characteristics of essentially contin- uous populations, leading White (1978) to entertain the possibility that they might be involved in parapatric spe- ciation. This is doubtful for Cepaea, at least, given that the genus contains only four species (Gould & Woodruff, 1990). Conversely, Clarke and Murray (1969) considered THE NAUTILUS, Supplement 2 parapatric speciation associated with area effects was a likely cause of distribution patterns in local isolates and semispecies of the seven species of Partula. Although this remains a possible hypothesis (Murray & Clarke, 1980) confirmation has been hindered by the finding that, de- spite substantial intra-specific variability, electrophoretic divergence between reproductively-isolated taxa is low (Johnson, 1977; Johnson et al., 1986a). One critical ques- tion regarding population differentiation can, however, be addressed using electrophoretic data. The studies of Johnson et al. (1986b) on allozymic variation suggest that, rather than deriving from multiple invasions, the fauna on Moorea evolved as an endemic radiation. Partula is a recent invader of the Society Islands, prob- ably arriving no more than 2.5 million years ago (Johnson et al., 1986b). This contrasts with the apparent antiquity of C. nemoralis which can be distinguished from its sister C. hortensis in fossil beds at least 10 million years old (Lamotte, 1951). It is not surprising, then, that the causes of area effects, and their phylogenetic consequences, should differ between the two genera. The Wilsons Prom- ontory Fluvidona exhibit a mixture of characteristics. The fauna is speciose, at least for such a small area, of probable ancient origin and subject to range expansion and contraction. In contrast to Cepaea, these changes have encouraged speciation. In contrast to Partula, this has probably been allopatric. Speciation and Gene Flow in Lake Eyre and Dalhousie Springs It is very difficult to predict what amount of gene flow would permit speciation because of the interdependence of biological and current and historical abiotic factors. This can be illustrated by a number of examples. Firstly, the Dalhousie Springs globular and pupiform snails are distinct species, as judged by nearly-fixed differences in wide sympatry. They are very closely related to each other and have no known close relatives. In situ diver- gence would be paradoxical if the current high levels of gene flow within these species reflect those applying his- torically. Comparisons of F. zeidleri with the allopatric pair of sister species F. accepta and F. aquatica gives a second example of this type. As judged by Fg, values, gene flow levels in F. zeidleri and F. aquatica are cur- rently almost identical and very low. Why should the evolutionary fate of the ancestor of F. accepta and F. aquatica differ from that of F. zeidleri? Spatial or tem- poral local factors must have had a significant cladoge- netic impact in this case. The evolution of new species in Partula apparently reflects the successive west to east appearance of new land masses in the archipelago-wide basis, with no species being found on more than one island (Johnson et al., 1986a). We expected a contrasting pattern in the Lake Eyre hydrobiids, owing to the potential confounding of geographic proximity with sporadic spring appearance. This was not observed, however, with almost all major genetic groups being definable by an appropriate north- D. J. Colgan and W. F. Ponder, 1994 south division. Only in F. zeidleri and F. billakalina are there populations which are more closely related to dis- tant areas than to their neighbours. This suggests that springs are colonised from relatively local sources as they arise, minimising the role of long-distance gene flow caused by factors such as bird transport. Gene Flow Estimation by Fs, or p(1) Two recent studies have suggested that the private alleles model (Slatkin, 1985a) is not as useful an estimator of gene flow as Fs; (Waples, 1987; Johnson et al., 1988). Johnson et al. (1988) compared the two approaches by relating them to direct estimates of gene flow in Partula. The private allele approach tended to give a higher es- timate of interpopulation migration than did the F-sta- tistics approach, which gave values nearer to the direct estimates. This is not always the case in our investigations; Nm<(p(1)) is higher than Nm(F7) in eight of fifteen com- parisons using the overall data, eleven (of fifteen) at the first level hierarchical pooling, and eight (of ten) at the second hierarchical pooling. Waples’ (1987) arguments against the general utility of the private alleles approach rest on the inconsistency of its estimates of gene flow with known larval dispersal patterns in ten species of shore fishes. In the hydrobiid data, too, there are notable inconsistencies in the estimates. For instance, Nm(p(1)) for the population by population migration rate in the MPI 4 genetic species is the highest of the four Wilsons Promontory Fluvidona, yet in all but one other estimate, including both pooled Nm(p(1)) comparisons and all Nm(Fz), this species has the lowest estimated gene flow. The private alleles approach also gives divergent esti- mates of gene flow in F. aquatica and F. zeidleri while the estimates for these species from the Fs; approach are highly concordant. There are several possible reasons why the private alleles model may not reflect population structure as ac- curately as the Fs; approach. Firstly, the average fre- quency of private alleles (p(1)) may be determined from too few observations (Waples, 1987; Johnson et al., 1988). The number of private alleles in our observations mainly exceeds the figure of 20, sufficient to obviate sample-size effects (Slatkin, 1985a). This is so for all three calculations (population by population and both poolings) for the Wilsons Promontory MPI 4 genetic species, F. zeidleri, F. aquatica, F. variabilis and T. punicea. Sample size problems can also arise if there are too few populations and this may apply to our analyses in which data were variously pooled. Particularly, the very high estimates of Nm(p(1)) in Dalhousie Springs data pooled into less than eight samples may be due to such effects although Slatkin (1985a) considered five populations sufficient to provide a good estimate. The present data suggest that estimation problems may also be caused where there are too many samples. There are numerous instances in all three present studies of alleles which are found in only a small geographic range. Such a range, be it the length of a tributary or a small Page 35 spring group, may represent the true neighbourhood size for the population. Too intensive sampling might obscure this, inflating the estimate of Nm. The impact of such effects on p(1) indicates that for this approach to be successful, there must be a fortuitous match between the scale of sampling, the biological scale of effective pop- ulation boundaries and inter-population migration. Anal- ysis of gene flow using the conditional frequency of all alleles would not be affected to anything like the same extent by the intensity of sampling. Consequently we suggest that an analytic model allowing information from all alleles to provide a numerical estimate of Nm would be more useful than current approaches. Waples (1987) mentions some other difficulties with the private alleles approach, including the non-linearity of the estimator when Nm > 10 or Nm < 0.1 and the possibility that uniform selection may bias the estimates if Nm < 1] (Slatkin, 1985a). Estimation of gene flow based on Fs; may also be biased by the failure of the assumption that selection is negligible (Wright, 1951; Waples, 1987; Johnson et al., 1988; Porter, 1990). To date, however, we have not detected significant selective differentials in our studies of hydrobiids. Despite the apparently poorer performance of the pri- vate alleles estimates compared to the use of F-statistics, they are easy to compute and should be included in investigations of gene flow for comparative purposes. More weight should be given to Fs; values when grossly divergent estimates of migration rates are obtained from the two approaches. Yet such divergences should also be taken as a signal that there has been significant distur- bance in the evolutionarily recent past - or that sampling does not match the scale of population structure. ACKNOWLEDGMENTS We thank Gerard Clark, Peter Eggler and Themo Terzis for performing the electrophoresis and collection assis- tance, and Janet Waterhouse, David MacIntosh, Jimmy and Hazel Ronay, Des Beechey and Roger De Keyzer for collection assistance. Alison Miller prepared the fig- ures. We thank Professor M. Johnson, and two anony- mous reviewers for attentive reading and criticism of the manuscript. This work was largely supported by grants from the Australian Research Council and the NPWS Endangered Species Program. LITERATURE CITED Arter, H.E. 1990. Spatial relationships and gene flow paths between populations of the alpine snail Arianta arbusto- rum (Pulmonata: Helicidae). Evolution 44:966-980. Bandoni, $.M., M. Mulvey, D.K Koech and E.S Loker. 1990. Genetic structure of Kenyan populations of Biomphalaria pfeifferi (Gastropoda: Planorbidae). Journal of Molluscan Studies 56:383-391. Barton, N.H. and M. Slatkin. 1986. A quasi-equilibrium the- ory of the distribution of rare alleles in a subdivided pop- ulation. Heredity 56:409-415. Page 36 Berger, E. 1973. Gene-enzyme variation in three sympatric species of Littorina. Biological Bulletin 145:83-90. Boeters, H.D. 1979. Species concepts of prosobranch fresh- water molluscs in western Europe, 1. Malacologia 18:57— 60. Boeters, H.D. 1982. Species concepts of prosobranch fresh- water molluscs in western Europe, 2. Malacologia 22:499- 504. Brown, L.D. 1991. Genetic variation and population structure in the Blacklip Abalone, Haliotis rubra. Australian Journal of Marine and Freshwater Reseasrch 42:77-90. Cain, A.J. and J.D. Currey. 1963. Area effects in Cepaea. Philosophical Transactions of the Royal Society of London B 246: 1-181. Cameron, R.A. and P.J. Dillon. 1984. Habitat stability, pop- ulation histories and patterns of variation in Cepaea. Malacologia 25:271-290. Campton, D.E., C.J. Berg, Jr., L.M. Robinson and R.A. Glazer. 1992. Genetic patchiness among populations of queen conch Strombus gigas in the Florida Keys and Bimini. Fishery Bulletin, U.S. 90:250-259. Carson, H.L. and A.R. Templeton. 1984. Genetic revolutions in relation to speciation phenomena:the founding of new populations. Annual Review of Ecology and Systematics 15:97-131. Cavalli-Sforza, L.L. and M. Feldman. 1990. Spatial subdi- vision of populations and estimates of genetic variation. Theoretical Population Biology 37:3-25. Chakraborty, R. and O. Leimar. 1987. Genetic variation with- in a sub-divided population. In: Ryman, N. and F. Utter (eds.). Population genetics and fishery management. Uni- versity of Washington Press, Seattle, p. 90 - 120. Chambers, S.M. 1978. An electrophoretically detected sibling species of “Goniobasis floridensis’’ (Mesogastropoda: Pleu- roceridae). Malacologia 17:157-162. Chambers, S.M. 1980. Genetic divergence between popula- tions of Goniobasis (Pleuroceridae) occupying different drainage systems. Malacologia 20:63-81. Clarke, B. and J. Murray. 1969. Ecological genetics and spe- ciation in land snails of the genus Partula. Biological Jour- nal of the Linnean Society 1:31—42. Cockerham, C.C. 1969. Variance of gene frequencies. Evo- lution 23:72-84. Crow, J.F. and K. Aoki. 1984. Group selection for a polygenic trait: estimating the degree of population subdivision. Pro- ceedings of the National Academy of Sciences U.S.A. 81: 6073-6077. Davis, G. M., V. Forbes and G. Lopez. 1988. Species status of northeastern American Hydrobia (Gastropoda: Proso- branchia): Ecology, morphology and molecular genetics. Proceedings of the Academy of Natural Sciences of Phil- adelphia 140:191-246. Davis, G. M., M. McKee and G. Lopez. 1989. The identity of Hydrobia truncata (Gastropoda: Hydrobiinae): Com- parative anatomy, molecular genetics, ecology. Proceed- ings of the Academy of Natural Sciences of Philadelphia 141: 333-359. Day, A.J. 1990. Microgeographic variation in allozyme fre- quencies in relation to the degree of exposure to wave action in the dogwhelk Nucella lapillus (L.) (Prosobran- chia: Muricacea) Biological Journal of the Linnean Society 40:245-261. Dillon, R.T., Jr. 1984. Geographic distance, environmental difference, and divergence between isolated populations. Systematic Zoology 33:69-82. THE NAUTILUS, Supplement 2 Dillon, R.T., Jr. 1988. The influence of minor human distur- bance on biochemical variation in a population of fresh- water snails. Biological Conservation 43:137-144. Dillon, R.T., Jr. and G.M. Davis. 1980. The Goniobasis of Southern Virginia and Northwestern North Carolina: Ge- netic and shell morphometric relationships. Malacologia 20:83-98. Endler, J.A. 1973. Gene flow and population differentiation. Science 179:243-250. Felsenstein, J. 1989. PHYLIP - Phylogeny Inference Package (Version 3.2). Cladistics 5:164-166. Gooch, J.L., B.S. Smith and D. Knupp. 1972. Regional survey of gene frequencies in the mud snail Nassarius obsoletus. Biological Bulletin 142:36-48. Gould, S.J. and D.S. Woodruff. 1986. Evolution and system- atics of Cerion (Mollusca: Pulmonata) on New Providence Island: a radical revision. Bulletin of the American Mu- seum of Natural History 182:389-490. Gould, S.J. and D.S. Woodruff. 1990. History as a cause of area effects: an illustration from Cerion on Great Inagua, Bahamas. Biological Journal of the Linnean Society 40: 67-98. Govindaraju, D.R. 1989. Estimates of gene flow in forest trees. Biological Journal of the Linnean Society 37:345-357. Grant, V. 1980. Gene flow and the homogeneity of species populations. Biological Zhentralblatt 99:157-169. Grant, W.S. and F.M. Utter. 1988. Genetic heterogeneity on different geographic scales in Nucella lamellosa (Proso- branchia, Thaididae). Malacologia 28:275-287. Habermehl, M.A. 1982. Springs in the Great Artesian Basin, Australia - their nature and significance. Bureau of Mineral Resources, Canberra, Geology and Geophysics, Report No. 235. Harris, C. 1993. Mound springs: South Australian conserva- tion inititives. Rangelands Journal 14:157-173 Hebert, P.D.N. and M.J. Beaton. 1989. Methodologies for Allozyme analysis using Cellulose Acetate Electrophoresis. Helena Laboratories, Austin. 32 pp. Hoagland, K.E. 1984. Use of molecular genetics to distinguish species of the gastropod genus Crepidula (Prosobranchia: Calyptreidae). Malacologia 25:607-628. Janson, K. 1987. Allozyme and shell variation in two marine snails (Littorina, Prosobranchia) with different dispersal abilities. Biological Journal of the Linnean Society 30:245- 256. Jarne, P. and B. Delay. 1990. Population genetics of Lymnaea peregra (Miller) (Gastropoda: Pulmonata) in Lake Ge- neva. Journal of Molluscan Studies. 56:317-321. Jarne, P. and B. Delay. 1991. Population genetics of fresh- water snails. Trends in Ecology and Evolution 6:383-386. Johanessen, K. 1988. The paradox of Rockall: Why is a brood- ing gastropod (Littorina saxatilis) more widespread than one having a planktonic larval dispersal stage (L. littorea)? Marine Biology 99:507-518. Johnson, M.S. and R. Black. 1984a. The Wahlund effect and the geographic scale of variation in the intertidal limpet Siphonaria sp. Marine Biology 70:295-302. Johnson, M.S. and R. Black. 1984b. Pattern beneath the chaos: The effect of recruitment on genetic patchiness in an in- tertidal limpet. Evolution 38:1371-1383. Johnson, M.S., B. Clarke and J. Murray. 1977. Genetic vari- ation and reproductive isolation in Partula. Evolution 31: 116-126. Johnson, M.S., J. Murray and B. Clarke. 1986a. An electro- phoretic analysis of phylogeny and evolutionary rates in SS SSS SETS D. J. Colgan and W. F. Ponder, 1994 the genus Partula from the Society Islands. Proceedings of the Royal Society of London B 227:161-177. Johnson, M.S., J. Murray and B. Clarke. 1986b. High genetic similarities and low hetreozygosities in land snails of the genus Samoana from the Society Islands. Malacologia 27: 97-106. Johnson, M.S., B. Clarke and J. Murray. 1988. Discrepancies in the estimation of gene flow in Partula. Genetics 120: 233-238. Johnson, M.S., O.C. Stine and J. Murray. 1984. Reproductive compatibility despite large-scale genetic divergence in Ce- paea nemoralis. Heredity 53:655-665. Kabat, A.R. and R. Hershler. 1993. The prosobranch snail family Hydrobiidae (Gastropoda: Rissooidea): Review of classification and supraspecific taxa. Smithsonian Contri- butions to Zoology. Smithsonian Contributions to Zoology, 547:1-94. Kimura, M. and G.H. Wiess. 1964. The stepping stone model of population structure and the decrease of genetic cor- relation with distance. Genetics 49:561-576. Kotwicki, V. 1989. Floods in the western Lake Eyre basin and their impact on Dalhousie Springs. In: Zeidler, W. and W.F. Ponder (eds.). Natural history of Dalhousie Springs. South Australian Museum, Adelaide, p. 41-45. Krieg, G.W. 1989. Geology. In: Zeidler, W. and W.F. Ponder (eds.). Natural history of Dalhousie Springs. South Austra- lian Museum, Adelaide, p. 19-26. Lamotte, M. 1951. Recherches sur la structure génétique des populations naturelles de Cepaea nemoralis L. Bulletin Biologique de France, Supplement 35:1-239. Lassen, H.H. 1979. Electrophoretic enzyme patterns and breeding experiments in Danish mud snails (Hydrobiidae). Ophelia 318:38-87. McCracken, G.F. and P.F. Brussard. 1980. The population biology of the White-lipped land snail, Triodopsis albol- abris: genetic variability. Evolution 34:92-104. Mimpfoundi, R. and G.J. Greer. 1989. Allozyme comparisons among species of the Bulinus forskalii group (Gastropoda: Planorbidae) in Cameroon. Journal of Molluscan Studies 55:405-410. Mimpfoundi, R. and G.J. Greer. 1990. Allozyme variation among populations of Biomphalaria camerunensis (Boett- ger, 1941) (Gastropoda: Planorbidae) in Cameroon. Jour- nal of Molluscan Studies 56:373-381. Mitton, J.B., C.J. Berg, Jr., and K.S. Orr. 1989. Population structure, larval dispersal and gene flow in the Queen Conch, Strombus gigas of the Caribbean. Biological Bul- letin 177:356-362. Mulvey, M., M.C. Newman and D.S. Woodruff. 1988. Genetic differentiation among West Indian populations of the schistosome-transmitting snail Biomphalaria glabrata. Malacologia 29:309-317. Murray, J.and B. Clarke. 1980. The genus Partula on Moorea: speciation in progress. Proceedings of the Royal Society of London B 211:83-117. Nei, M. 1973. Analysis of gene diversity in subdivided pop- ulations. Proceedings of the National Academy of Sciences U.S.A. 70:3321-3323. Nei, M. and R.K. Chesser. 1983. Estimation of fixation indices and gene diversification. Annals of Human Genetics 47: 253-259. Ochman, H., J.S. Jones and R.K. Selander. 1983. Molecular - area effects in Cepaea. Proceedings of the National Acad- emy of Sciences U.S.A. 80:4189-4193. Ponder, W.F. 1986. Mound springs of the Great Artesian Page 37 Basin. In: DeDeckker, P. and W.D. Williams (eds.). Lim- nology in Australia CSIRO, Melbourne and W. Junk, The Hague, p. 403-420. Ponder, W.F. 1989. Mollusca. In: Zeidler, W. and W.F. Pon- der (eds.). Natural history of Dalhousie Springs. South Australian Museum, Adelaide, p. 71-80. Ponder, W. F. and G.A. Clark. 1988. A morphological and electrophoretic examination of “Hydrobia buccinoides’, a variable brackish-water gastropod from temperate Aus- tralia (Mollusca: Hydrobiidae). Australian Journal of Zo- ology 36:661—689. Ponder, W.F., D.J. Colgan and G.A. Clark. 1991. The mor- phology, taxonomy and genetic structure of Tatea (Mol- lusca: Gastropoda: Hydrobiidae), estuarine snails from temperate Australia. Australian Journal of Zoology 39:447- 497. Ponder, W. F., R. Hershler and B. Jenkins. 1989. An endemic radiation of hydrobiid snails from artesian springs in north- ern South Australia: their taxonomy, physiology, distri- bution and anatomy. Malacologia 31:1-140. Ponder W.F., D.J. Colgan, G.A. Clark, A.C. Miller and T. Terzis. 1994. Microgeographic genetic and morpholog- ical differentiation of freshwate snails - a study on the Hydrobiidae of Wilsons Promontory, Victoria, southeast- ern Australia. Australian Journal of Zoology, In press. Porter, A.H. 1990. Testing nominal species boundaries using gene flow statistics: the taxonomy of two hybridising Ad- miral butterflies (Limenitis: Nymphalidae). Systematic Zoology 39:131-147. Preziosi, R.F. and D.J. Fairbairn. 1992. Genetic population structure and levels of gene flow in the stream-dwelling waterstrider, Aquarius (= Gerris) remigis (Hemiptera: Gerridae). Evolution 46:430-444. Rees, W.J. 1965. The aerial dispersal of Mollusca. Proceedings of the Malacological Society of London 36:269-282. Rohlf, F. J. 1990. NTSYS-pe. Numerical taxonomy and mul- tivariate analysis system. Version 1.60. Exeter Software, New York. Rollinson, D., R.A. Kane, A. Warlow and V.R. Southgate. 1990. Observations on genetic diversity of Bulinus cernicus (Gas- tropoda: Planorbidae) from Mauritius. Journal of Zoology 222:19-26. Schmidt, E.R. and I.W.B. Thornton. 1992. The Psocoptera (Insecta) of Wilsons Promontory National Park, Victoria. Memoirs of the Museum of Victoria, 53, 137-220. Skibinski, D.O.F., J.A. Beardmore and T.F. Cross. 1983. As- pects of the population genetics of Mytilus (Mytilidae: Mollusca) in the British Isles. Biological Journal of the Linnean Society 19:137-183. Slatkin, M. 1981. Estimating levels of gene flow in natural populations. Genetics 99:323-335. Slatkin, M. 1985a. Rare alleles as indicators of gene flow. Evolution 39:53-65. Slatkin, M. 1985b. Gene flow in natural populations. Annual Review of Ecology and Systematics 16:393-430. Slatkin, M. and N.H. Barton. 1989. A comparison of three indirect methods for estimating average levels of gene flow. Evolution 43:1349-1368. Smith, P.C. 1989. Hydrogeology. In: Zeidler, W. and W.F. Ponder (eds.). Natural history of Dalhousie Springs. South Australian Museum, Adelaide, p. 27-39. Snyder, T.P. and J.L. Gooch. 1973. Genetic differentiation in Littorina saxatilis (Gastropoda). Marine Biology 22:177- 182. Sundberg, P., A.J. Knight, R.D. Ward and K. Johannesson. Page 38 1990. Estimating the phylogeny in mollusc Littorina sax- atilis (Olivi) from enzyme data: methodological consid- erations. Hydrobiologia 193:29-40. Swofford, D.L. and R.B. Selander. 1981. BIOSYS-1: a FOR- TRAN program for the comprehensive analysis of elec- trophoretic data in population genetics and systematics. Journal of Heredity. 72: 281-283. Thomson, R. and S. Barnett. 1985. Geology, geomorphology and hydrogeology. In: Greenslade, J., L. Joseph and A. Reeves (eds.). South Australia’s mound springs. Nature Conservation Society of South Australia, Adelaide, p. 3- 26. Todd, C.D., J.N. Havenhand and J.P. Thorpe. 1988. Genetic differentiation, pelagic larval transport and gene flow be- tween local populations of the intertidal marine mollusc Adalaria proxima (Alder and Hancock). Functional Ecol- ogy 2:441-451. Vermeij, G. 1987. The dispersal barrier in the tropical Pacific: implications for molluscan speciation and extinction. Evo- lution 41:1046-1058. Wallis, G.L. 1988. Wilsons Promontory. In: Clark, I. and B. Cook (eds.). Victorian Geology Excursion Guide. Austra- lian Academy of Science, Canberra, 353-372. Waples, R.S. 1987. A multispecies approach to the analysis of gene flow in marine shore fishes. Evolution 41:385-400. Weir, B. 1990. Intraspecific differentiation. In: Hillis, D.M. and C. Moritz (eds.). Molecular systematics. Sinauer As- sociates, Sunderland, p. 373-410. Weir, B. and C.C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38:1358— 1370. THE NAUTILUS, Supplement 2 White, M.J.D. 1978. Modes of speciation. W. H. Freeman and Co., San Francisco. pp. 455. Whitlock, M.C. 1992. Temporal fluctuations in demographic parameters and the genetic variance among populations. Evolution 46:608-615. Williams, A.F. and J.W. Holmes. 1978. A novel method of estimating the discharge of water from mound springs of the Great Artesian Basin, Central Australia. Journal of Limnology 38:262-272. Woodruff, D.S., M. Mulvey and M.W. Yipp. 1985. Population genetics of Biomphalaria straminea in Hong Kong. Jour- nal of Heredity 76:355-360. Wool, D. 1987. Differentiation of island populations: a lab- oratory model. American Naturalist 129:188-202. Wopfner, H. and C.R. Twidale. 1967. Geomorphological his- tory of the Lake Eyre Basin. In: Jennings, J.N. and J.A. Mabbutt (eds. ). Landform studies in Australia . Australian National University Press, Canberra, p. 118-143. Wright, S. 1931. Evolution in Mendelian populations. Ge- netics 16:97-159. Wright, S. 1951. The genetical structure of populations. An- nals of Eugenics 15:323-354. Wright, S. 1978. Evolution and the genetics of populations. Vol. 2. The theory of gene frequencies. University of Chi- cago Press, Chicago. Zeidler, W. and W.F. Ponder. 1989. Natural history of Dal- housie Springs. South Australian Museum, Adelaide. pp. 138. Page 39 D. J. Colgan and W. F. Ponder, 1994 T I I I I I I Z Z z I I I i I I Z Z I I I Z I Z I Z g I I I I Z I I Z Z Z Z Z Z I Z Z Z Z Z T I I i Z 4 Z I I I I I I I I I I i I z I @ € I € € I I I I I Z Z I Z Z I i I H 9 a ci 100] eS! et eS! el el eS 4 4 Nao AN aH NG oH ATA Sn Na AO Lon) CON Ta AO | AQ NAAN Lael = 6LLG TTS PIT TT IGVG PPI GGVS IlVsé IlVs IlVsé ILVs SII ST 8Tes LET TT ITV cr IIt OS TTT TT LG C19G 61S STITT SoC1LGE CICGP ITT ILTTCPV L1T9@ IS ls ou O'" Oddn Id HdSs dNd Hd9d-9 WOd D9T TA IVI Udd HdO IdW HAWN dv'l HdI HddH WH LdD IdD dd9 TIVO HA LSH ONG LVV dV UONeIAIIQGY ase,Aroydsoydoidd asoon[s qqn aseiauos! ayeydsoydasor y, aseuaso1pAyep [OWGIOS ase[A1oydsoyd apisoajonu suring ‘Boi1pAyap a}euoon|soydsoy g-9 aseynuoon|soydsoy g ayelsqns A[D-ney-neyT : ayelsqns noyJ-[eA : ayersqns epy-neT : a}eI\sqns O1g-9y4q ‘sesepndag aseuasoipAyep autdo}0Q aseiowos! ayeydsoydesouuryy aseusso1pAYyap 9}e[P asepyndedourwme autona'T aseudsOIpAYap 9}eI}00S] aseussoipAyep 9}e1AyngdxoipAyy ISCUTYOXI}Y aseuluesuel) ayeAniAd-ajeureyn{y) ase1oWOs! a}eydsoydasoonys) ‘Bo1pAyap ayeydsoyd-¢g-[o1904]5) asepisoyoryes)-g aseyeipAy o}e1eUIN J aSP19]Sq ase[ouy aseloysuPIOUIUe aje}Iedsy aseyeydsoyd ourlyeyx[ Vy awiAZzuy ‘saloads ayI]|-puopian)y aisnoyleq 94} ‘[ pur ‘sjreus wu4ojstdnd pue sejnqoys ssutidg aisnoyyeq ey} ‘H ‘pynumw “yf, pue 1yjwWs “fH ‘vaqvund nigospry901], ‘SyIQDLIDA “J A ‘Mappraz “yy q ‘vouonbp “gq cd ‘vidas vajysoosuoy st g ‘Duopian)y A10JUOWIOIg SUOS|IA\ SI Y UUINTOD ‘exe] payeusIsap ay] UT a[qejoidsayUT a19M YOIYM TOO] JO Joquinu ay) pue Jequinu ‘9 'q ‘UOTyeIAsIqqe syI ‘auIeU sUIAZUA 9Y} AAIZ SUUINIOO ay], ‘s[feus parpnys Jo sdnoxs o1wWOUOKe} ay] UT 190] a[qeje1d1aquI jo siaquinu ey], “[ xipueddy Page 40 Appendix 2. Hierarchical sample structure analysed for var- ious hydrobiid species. The highest hierarchical levels are writ- ten flush to the left margin of each column. Lower levels are successively indented. Spring groups and watersheds are iden- tified in figures 1 to 3. The average number of specimens scored for each locus is given after the sample designations. Wilsons Promontory Fluvidona Hierarchy: Catchment-Stream- Site. ‘MAIN’ indicates the principal stream of the catchment, ‘ONE’, ‘TWO’, etc., the tributaries. Sites within catchments are numbered, approximately, clockwise. Straight-sided WHISKY CREEK LOWER WCl 12.64 8.36 12.09 8.27 11.41 SQUEAKY CREEK MAIN SC2 FRASER CREEK MAIN FC1 GROWLER CREEK MAIN GC10 Convex MP13 GROWLER CREEK MAIN GCl12 FIVE GC13 FRASER CREEK MAIN FC3 ROARING MEG MAIN RM1 RM3 ONE RM2 TWO RM4 FIRST BRIDGE CREEK MAIN FBI PICNIC CREEK MAIN PCl 7.09 7.09 7.99 7.27 7.00 7.04 7.04 8.54 5.27 13.64 7.04 7.32 Appendix 2. Continued. FRESHWATER CREEK TWO FW3 WATERLOO BAY MAIN WBI WB2 Convex MP14 WHISKY CREEK MAIN TIDAL RIVER ONE TR4 TRS THREE TR7 TR8 FOUR TRO FIVE TRI TRIO TWO TR2 SIX TR6 GROWLER CREEK MAIN GC21 ONE GC16 TWO GC6 THREE GC4 GC5 BLACKFISH CREEK ONE BC4 TWO BC6 THREE BC8 THE NAUTILUS, Supplement 2 6.41 7.09 7.09 16.09 13.55 12.95 12.99 13.95 13.90 27.95 38.45 8.63 8.36 7.63 8.00 8.09 8.81 8.63 8.36 11.41 8.96 7.93 7.82 7.91 5.91 5.91 7.45 D. J. Colgan and W. F. Ponder, 1994 Page 41 Appendix 2. Continued. Appendix 2. Continued. SQUEAKY CREEK DAVENPORT SPRINGS MAIN DS4 13.92 SC2 7.36 DS5 10.38 D : ONE S6 13.62 SC3 7.36 eet. HILL 15.42 Convex MPI5 HH8 10.38 CHINAMANS CREEK HH9 15.38 MAIN HH10 10.46 CCl ; 15.72 HH11 10.46 DARBY RIVER Hd ee MAIN F. aquatica DR2 14.50 MIDDEE ONE ei hs 9.70 DR3 13.96 BC15 9.85 TWO BC16 9.85 DR4 14.64 COWARD/KEWSON WHISKY CREEK CS19 9.81 MAIN ES20 9.15 ES21 O70 WCl 12.55 ey 7.54 WC2 10.00 Wwc4 11.64 JS28 9.73 WC5 11.64 E29 oo TIDAL RIVER BERESFORD SPRINGS MAIN BS22 9.69 TRI 10.72 STRANGWAYS SPRINGS GROWLER CREEK ae ae MAIN : GC25 3.34 NORTE IDE SPRIN FRESHWATER CREEK OUPre.S i OS25 8.85 MAIN TM26 9.00 FW 7.0 : FREELING SPRINGS FRASER CREEK FR31 10.73 MAIN FR32 9.73 MNI a8 F. billakalina ONE BILLK FC1 7.09 BK18 7.92 SQUEAKY CREEK STRANGWAYS MAIN SS24 8.00 SCl 14.28 $S30 9.13 Lake Eyre Hydrobiidae Hierarchy: Region-Spring group-Site. F. variabilis Samples are identified by spring group initials and a number SOUTH indicating south-north order in the entire Lake Eyre collec- WELCOME SPRINGS tions. WS2 9.80 F. accepta DAVENPORT SPRINGS WELCOME SPRINGS DS6 8.60 WS1 15.38 WS? 10.38 MIDDLE WS3 15.38 BLANCHE CUP BC13 9.70 BC14 10.44 BC15 10.20 BC16 7.92 Page 42 Appendix 2. Continued. COWARD/KEWSON CSA 7.92 DS4 9.00 CS19 8.00 DS5 8.92 ES20 8.56 DS6 8.00 ES21 10.08 KH97 8.44 HERMIT HILL JS28 10.20 HH7 8.00 JE29 7.80 HH8 8.00 HH10 7.96 BERESFORD SPRINGS HH12 aD BS22 8.32 WA23 8.20 OEE BLANCHE CUP NOU BC13 8.00 OUTSIDE SPRINGS BCI5 AOS Soe ae COWARD/KEWSON / CS19 7.38 FREELING SPRINGS S20 788 FR31 7.79 ES21 7.92 FR382 7.96 KH27 7.88 F. zeidleri ae oe SOUTH es HERMIT SPRINGS T. smithi HH8 3.50 MIDDLE so BLANCHE OU a BLANCHE CUP r BC14 10.25 BK18 16.89 COWARD/KEWSON BERESFORD SPRINGS CS17 10.25 BS22 10.22 CS19 8.75 STRANGWAYS SPRINGS ES20 10.21 Seay 13.27 ES21 9.96 SS30 9.22 KH27 7.88 JS28 10.17 NORTH JE29 8.33 OUTSIDE SPRINGS BILLAKALINA OS25 15.33 BERESFORD SPRINGS IP, antineie BS22 iorae OUTSIDE SPRINGS WA23 10.25 OS25 8.44 STRANGWAYS SPRINGS FREELING SPRINGS $S24 10.09 FR31 13.61 NORTH FR32 11.56 OUTSIDE SPRINGS Dalhousie Springs Hydrobiidae Hierarchy: Spring group-Sub- TM26 8.88 group-Site. Samples marked with a “p” are from the pool of large springs and those with “o” from the outflow. FREELING SPRINGS @lawalan FR32 8.79 obular 5 A ONE Al 6.64 T. punicea A3 8.57 SOUTH A8 19.57 WELCOME SPRINGS B ONE Bl 11.48 WS! 8.00 C A Calp 49.50 WS3 9.46 Calo 37.07 Calo 28.39 Calo 33.54 CaQ 32.25 THE NAUTILUS, Supplement 2 Appendix 2. Continued. DAVENPORT SPRINGS D. J. Colgan and W. F. Ponder, 1994 Page 43 Appendix 2. Continued. Appendix 2. Continued. B Cb2 6.93 D A Dal 6.00 Cb2a 5.93 Da2 17.64 Cb2b 6.29 Da3 11.75 c Cel 6.07 B Dbl 6.00 Cc3 6.50 Db2 9.79 D Cdlp 7.40 Db4 5.96 Cdlp 7.00 E A E5 23.21 Cdlo 4.32 B El 19.85 Cdlo 7.00 E2 9.21 Cd2 14.25 ETa 18.07 Cd8 Sl E8 21.93 Pupiform F ONE Fl 20.04 A ONE Al 18.89 F2 17.89 AQ 21.29 G ONE Ga2 26.07 AS 6.11 Ga8 33.14 A6 17.82 Ga4 15.00 A8 34.11 Ga6a 6.39 B ONE Bl 8.86 Ga6b 14.07 B2 9.79 H ONE Hl 6.96 Cc A Ca2 29.79 TWO H3 17.75 Ca3 26.29 Fluvidona-like oe ie © Cdil 4.16 ala Ca7b 30.71 . oe oe Ca8 17.79 Cal2 30.18 F F9 4.16 Cal3 28.25 G Gal 4.16 B Cb4 8.21 Ga2o 4.16 Cb5o 6.00 Ga6 4.16 Cb5p 17.86 C Cel 6.43 Cc4 16.43 Cc8 8.32 D Cdlp 6.96 Cd3 5.61 Cd5 5.43 Cd9 6.36 THE NAUTILUS, Supplement 2:44-50, 1994 Page 44 Allozyme Cladistics in Malacology: Why and How? Kenneth C. Emberton Department of Malacology Academy of Natural Sciences of Philadelphia 1900 Benjamin Franklin Parkway Philadelphia, PA 19103-1195, USA ABSTRACT A hypothetical but plausible data set is introduced for which Mickevich and Mitter’s (1981, 1983) qualitative-coding, min- imum-turnover cladistic (= discrete parsimony) method ac- curately reconstructs phylogeny, whether rare alleles are de- tected or not, but for which both the UPGMA distance method and a hand-calculated application of Swofford and Berlocher’s (1987) frequency-parsimony method give incorrect phyloge- nies. Mickevich and Mitter’s (1981, 1983) method is outlined and demonstrated. When applying this method to 20 polygyrid genera using Hennig86 (Farris, 1988), two problems arose and were circumvented. First, allelic combinations occurred in complexly interrelated sets (interim solution: treat such sets as single character-states); and second, alternative character-state trees existed for each locus (solution: binary-code each alter- native, then weight by the reciprocal of the number of alter- natives). Cladistic analysis of the polygyrid-genera allozyme data (Emberton, 1994) ordered yielded the same topology as, but higher resolution than, when run unordered. Key words: allozymes, phylogenetics, cladistics, discrete par- simony, distance methods, frequency-parsimony, ordered vs. unordered data, Gastropoda Polygyridae. INTRODUCTION Allozyme data remain among the easiest and cheapest to obtain for molecular systematics (Richardson et al., 1986; Hillis & Moritz, 1990). Despite the current revo- lution in nucleic-acid sequencing (see other papers in this volume), allozymes may continue to play a major role in molluscan systematics in many laboratories throughout the world, for many years to come. Of the two discrete ways to use allozyme data for systematics (Sarich, 1977), population genetics has dom- inated in molluscan studies (Berger, 1983; Cain, 1983; Johnson et al., 1988; Hillis, 1989; Woodruff, 1989; Wood- ruff & Solem, 1990) while phylogenetic reconstruction has been less common. This paper deals only with phy- logenetic reconstruction. There are three main approaches to phylogenetic re- construction using allozymes (Buth, 1984; Swofford & Berlocher, 1987): distance methods (Sneath & Sokal, 1978; Farris, 1972), cladistics (= discrete parsimony = phy- logenetic but not taxonomic aspects of cladistics) (Hen- nig, 1966; Wiley, 1981; Brooks & McLennan, 1991; Har- vey & Pagel, 1991), and frequency-parsimony (Swofford & Berlocher, 1987). Of these, distance methods have been by far the most commonly used in molluscan systematics (e.g. Davis et al., 1981; Johnson et al., 1986; Emberton, 1988), whereas cladistics has been relatively uncommon (Emberton, 1988, 1991; Hoeh, 1990) and frequency-par- simony remains untried. Which of these three is the most appropriate application of allozyme data to phylogenetic reconstruction? What is the best method? Does this meth- od hold up even when rare alleles are undetected? Is this method practical? What are its pitfalls, and how can they be avoided? The purpose of this paper is to begin to address these questions by (1) introducing a hypothetical case of allo- zyme evolution for which only a cladistic method, and neither UPGMA (a distance method) nor frequency-par- simony (as understood and hand-calculated by the pres- ent author), accurately reconstructs phylogeny, both with and without detection of rare alleles; (2) explaining and demonstrating Mickevich & Mitter’s (1981, 1983) qual- itative-coding, minimum-turnover cladistic method; and (3) documenting how this method was applied, using Hennig86 programs (Farris, 1988), to a complex mala- cological data set (Emberton, 1994). MATERIALS AND METHODS The hypothetical phylogeny (Figure 1) consists of an outgroup (out) and three taxa (A, B, and C) with the tree topology: out(A(B,C)). The devised allozyme data from this phylogeny (Figure 1) consist of three loci (locus 1, locus 2 and locus 3), each of which has three alleles (a, b, and c). In all three loci the designated course of evo- lution was from allele a to allele b, passing through the intermediate stage of heterozygosity ab. In locus 1, the outgroup is fixed for a, taxon A is heterozygous for ab, and taxon B is fixed for b; taxon C has evolved further to acquire a third allele (c), for which it is heterozygous (bc). Locus 2 in this hypothetical case K. C. Emberton, 1994 Page 45 Allele Frequencies a |1.00 0.05 - - Locus 1 b = 0.95 1.00 0.05 c = = = 0.95 a |1.00 0.05 - - Locus 2 b - 0.95 0.05 1.00 c - = 0.95 = a {1.00 0.05 - 0.95 Locus 3 b — 0.05 1.00 0.05 c - 0.90 - - Figure 1. Hypothetical phylogeny and allozyme frequencies for three loci frequencies for three loci. See text for explanation. is identical to locus 1, except that allele c is acquired by taxon B rather than by taxon C. Locus 8 differs in that allele c is acquired only by taxon A, and taxon C partially reverts to allele a. The frequencies of alleles are shown in Figure 1, and range from 0.05 to 1.00. For a representative distance-method analysis of this hypothetical phylogeny and allozyme-evolution pattern, unweighted pair group, mathematical averaging (UPGMA) was used (Sneath & Sokal, 1973). This method can be applied to any genetic-distance matrix; the index of genetic distance chosen here was that of Prevosti (see Wright, 1978) because of its simplicity of calculation, its UPGMA from Frequency Cladistics: Prevosti Dist. Parsimony Min. Turnover out C A B out A B Cc out A B Cc ey kaa] [iss Locus 1 out B A G out A B Cc out A B Cc = fie hes Locus 2 out C A B out C A B out A B (e jel Locus 3 US out B A c out C A B out A B Loci 1+2+3 Figure 2. Performance tests of three different methods of phy- logenetic inference, analysing the hypothetical data set in Fig- ure 1. See text for details. UPGMA from Frequency Cladistics: Prevosti Dist. Parsimony Min. Turnover out A B c out A B Cc out A B Cc =) [ea Locus 1 out B A Cc out A B Cc out A B Cc Lj Lo LJ Locus 2 out C A B is © A B out A B Cc [ EA LI Locus 3 ] L | out C A B out A B Cc (three LJ Loci 1+2+3 equal al- ternatives) Figure 3. Same as Figure 2, but after deleting allelic occur- rences at frequencies of 0.05. meeting of the triangle-equality criterion, and its simi- larity in performance to the preferred—both theoreti- cally and empirically—Cavalli-Sforza and Edwards arc and arc-chord indices (see Wright, 1978). It must be emphasized that UPGMA, although commonly used, is well known to be highly prone to inaccuracies when evolutionary rates differ among lineages. A fuller and fairer test of distance methods—beyond the scope of this paper—would have to include neighbor joining and oth- er methods that do not assume constancy of evolution in all lineages. Application of frequency-parsimony attempted to fol- low Swofford and Berlocher (1987), analyzing the data by hand rather than using the FREQPARS program of Swofford (1988). According to the present author's un- derstanding, the frequency-parsimony method (Swof- ford & Berlocher, 1987) finds a set of hypothetical an- cestors that minimizes the total amount of change in all alleleic frequencies; there may be more than one set of hypothetical ancestors—in such cases, hypothetical an- cestors were chosen as identical to extant taxa whenever possible. Cladistic (= discrete-parsimony) analysis treated the locus as character, allelic combinations as character-states, and heterozygotes as evolutionarily intermediate be- tween homozygotes for the same alleles (qualitative cod- ing, minimum turnover model of Mickevich & Mitter, 1981, 1983; see below). This is a favored method of cladistic treatment of allozyme data (Buth, 1984), but a fuller evaluation would also have to consider transmodal theories as outlined by Mickevich and Weller (1990). Each of these three methods of phylogeny reconstruc- tion was applied four times: to locus 1, to locus 2, to locus 3, and to all three loci combined. The entire analysis was then repeated under the assumption that rare alleles were Page 46 out A B Cc a [2-00 0.05 - S Locus 1 b | = 0.95 1.00 0.05 (oe = = = 0.95 IL a |1.00 0.05 - - Locus 2 b = 0.95 0.05 1.00 @ 2 0.95 - a ab b b a --> ab --> b LJ Loci 1, 2 ab --> b out A a --> ab Figure 4. Application of cladistic steps 1-5 (see text) to Figure 1’s loci 1 and 2. See text for explanation. undetected, i.e. after deleting all allelic occurrences of frequency 0.05. All computations were by hand. To explain and demonstrate Mickevich and Mitter’s (1981, 1983) qualitative-coding, minimum-turnover cla- distic method, the method was broken down into easy- to-follow steps, with a worked-through example using the hypothetical data set (Figure 1). This method was applied, using Hennig86 programs (Farris, 1988), to a data set from 20 genera of polygyrid land snails (Gastropoda: Pulmonata: Stylommatophora), as part of a broad-based phylogenetic analysis that also incorporated behavior and reproductive anatomy (Em- berton, 1994). The data consisted of eight cladistically informative loci whose allelic variation was classified into 29 character states. Hennig86 programs were used be- cause Platnick’s (1989) empirical tests found them su- perior to any other programs then available in finding the most parsimonious cladograms from real data sets. Current PAUP programs, however, may perform more accurately than in the past, and they would probably be more flexible in dealing with the polygyrid data set (see below). RESULTS Figure 2 shows results of the three methods applied to the data of Figure 1. UPGMA gave the incorrect phy- logeny for each of the three loci, as well as for combined loci. Frequency-parsimony (as interpreted by the present author) gave the correct phylogeny for loci 1 and 2, but the incorrect phylogeny both for locus 8 and for all three loci combined. The cladistic method, on the other hand, always gave the correct phylogeny, although with in- complete resolution for locus 3, in which taxa A, B, and C appear in a trichotomy. Deleting rare alleles gave similar results, but with less resolution overall (Figure 3). Thus UPGMA produced THE NAUTILUS, Supplement 2 out A B (eo a |1.00 0.05 - 0.95 Locus 3 b - 0.05 1.00 0.05 Cc > 0.90 = = a ab b ab a --> ab --> b out A B Cc = | a --> ab Locus 3 Figure 5. Application of cladistic steps 1-5 (see text) to Figure 1’s locus 3. See text for explanation. the incorrect phylogeny for all three loci, and gave three equal alternative phylogenies for the combined loci. Fre- quency parsimony (as interpreted by the present author) was correct for locus 2, but gave incorrect results for loci 1 and 3 and for combined loci. The cladistic method gave topologies that were incorrect for locus 1, correct and resolved for locust 2, and correct but incompletely resolved (with a trichotomy for taxa A, B, and C) for both locus 3 and for the combined loci. Thus the cladistic method was the only one of the three to accurately re- construct phylogeny from the hypothetical data set, whether rare alleles were detected or not. The qualitative coding, minimum turnover model (Mickevich & Mitter, 1981, 1983; Buth, 1984) can be outlined in six steps: 1. Treat each locus as a character. 2. Delete alleles found in only one taxon (autapo- morphies) (although it is not clear whether Mick- evich and Mitter are strong advocates of this par- ticular practice). 3. Treat each allelic combination as a character state, ignoring frequencies (qualitative coding). 4, Order character states into character-state trees that minimize the total number of allelic changes (min- imum-turnover model). 5. Root character-state trees by outgroup comparison. 6. Combine all character-state trees using the prin- ciple of parsimony (a computer program is usually required). out A B | Ll 3b --> 3ab (reversal) = lab --> 1b Loci 1+2+3 mm 2ab --> 2b 3ab --> 3b mm ila --> lab mm 2a --> 2ab 3a --> 3ab Figure 6. Application of cladistic step 6 (see text) to character- state trees of Figures 4 and 5. See text for explanation. K. C. Emberton, 1994 MPI 100/100 3 100/102 99/100 99/99 4 99/102 a P MPI 102/104 a MPI 101/101 2 4 \ 4 MPI 102/102 jal Figure 7. Detected allelic character states in the MPI locus among 20 genera of polygyrid snails, with hypothesized char- acter-state assignments and transformations (from Emberton, in review). These steps are demonstrated for the hypothetical phy- logeny in Figures 4-6. Figure 4 shows loci 1 and 2, each of which yields the same cladogram, and each of which is treated separately (step 1). Since allele c occurs in only one taxon, it is deleted as a phylogenetically uninform- ative autapomorphy (step 2). This leaves three allelic combinations (a, b, and ab), each of which is treated as a single character state, regardless of allelic frequencies (step 3). Thus taxa B and C are scored equally for char- acter-state b, even though this allele occurs at frequency 1.00 in taxon B and at frequency 0.05 in taxon C (Figure 4). Ordering these three character states by the mini- mum-turnover model (step 4) results in the order b <- > ab <-> a. This order is more parsimonious regarding allelic changes (i.e., turnover) than either alternative (i.e., b <-> a <-> abora <-> b <-> ab). Rooting this ordered character-state tree (a linear transformation se- ries in this case) is done by outgroup comparison (step 5). The outgroup has character state a, which is therefore hypothesized as plesiomorphic. The rooted character- state tree, therefore, is a -> ab —> b for both locus 1 and locus 2. Each of these two loci has the same single- locus cladogram shown at the bottom of Figure 4. Cladistic treatment of hypothetical locus 3 is similar and is outlined in Figure 5. The locus is treated inde- pendently (step 1); the autapomorphy for allele c in taxa A is deleted (step 2); the character states are defined as allelic combinations a, b, and ab (step 3), which are ordered and rooted in the tree a -> ab —> b (steps 8 and 4). The resulting single-locus cladogram (bottom of Figure 5) also puts taxa A, B, and C in a monophyletic clade defined by the synapomorphic transition a —> ab. This cladogram differs from those for loci 1 and 2 (Figure 4), however, in that it gives no further resolution: the transition ab -> b shows up only as an autapomorphy of taxon B. Step 6 consists of combining the single-locus character- state trees or cladograms (Figures 4 and 5) in the most parsimonious way (i.e., involving the least amount of overall turnover among character states within their re- spective loci). This results in the cladogram of Figure 6, Page 47 which is a perfectly accurate reconstruction of the orig- inal hypothetical phylogeny (Figure 1). The preceding example was simple enough to be per- formed by hand. For complex data sets, the program Hennig86 (Farris, 1988) has been recommended (Plat- nick, 1989). For encoding branching character-state trees for Hennig86 analysis, binary coding is needed. The pres- ent author finds it useful to think of binary coding as a method of encoding not by character state but by trans- formation (between two character states). To encode the hypothetical data, each transformation is numbered as follows: Transformation Locus Transformation Number ] a— ab 0 1 ab > b 1 2 a — ab 2 2 ab > b 3 8 a — ab 4 3 ab > b 5 The following taxon-by-transformation matrix, then, is submitted to Hennig86. Each taxon is scored for the occurrence (1) or non-occurrence (0) of each transfor- mation in the lineage that produced that taxon. Transformation ie OC hla oe 2y es out 0 0 0 0 0 0 A ] 0 ] 0) 1 0 B 1 1 1 1 1 1 C 1 Il ] a a 0 Given this matrix, Hennig86 produces the correct clado- gram shown in Figure 6. Applying this six-step method of allozyme cladistics to 20 polygyrid genera (Emberton, 1994) resulted in problems in steps 4 and 6. In step 4 (ordering allelic combinations into character-state trees), there was often a problem of a confusingly large number of allelic com- binations, many of which occurred in complexly inter- related sets. For example, the mannose phosphate isom- erase locus (MPI) yielded five alleles (99, 100, 101, 102, and 104, referring to their relative positions in mm on the electrophoretic gels). These alleles occurred in eight allelic combinations, which are shown in Figure 7. Or- dering these eight combinations as individual character states would produce extreme complexity and, given the distributions of these combinations among taxa (Ember- ton, 1994), would yield apparently little additional phy- logenetic information. The ultimate solution to this problem seems to require writing a computer program to perform the minimum- turnover algorithm, regardless of the number of allelic combinations. An interim procedure, however, is to treat such interrelated sets as single character states. For the MPI example, this groups five interrelated allelic com- binations as a single character state (Figure 7: character state .3). For other examples, see Emberton (1994). Page 48 77 78 86 87 17.1 —17.2 —> 17.3 17.1 —>17.3 —>17.2 79) oe 17.4 17.4 80 81 89 90 17.1 —>—17.2 —> 17.3 17.1 —>17.3 —> 17.2 a2 A 91 & 17.4 17.4 83 84 92 93 17.1 —> 17.2 —> 17.3 17.1 —> 17.3 —> 17.2 a5 A 94 17.4 17.4 Weights for Transitions 77-94 = 1/6 Figure 8. Alternative character-state trees derived from Figure 7. Numbering of character states and transformations as in Emberton (in review). Another problem that arose in applying step 6 (com- bining all character-state trees using parsimony) was that of alternative character-state trees. For example, in the MPI locus (Figure 7), state .3 has an equal probability (according to the minimum-turnover model: step 5) of being derived from states .1 or .2, but not from state .4. Likewise, state .4 has an equal probability of derivation from any of the three other states. To encode all these possible transformations would unduly weight infor- mation-poor loci over information-rich loci. A solution devised by Emberton (1994) is to use all transformations in all alternative character-state trees, but to weight them by the reciprocal of the number of alternatives. For example, the multiple arrows among MPI character states .1, .2, .3, and .4 (Figure 7) result in six alternative character-state trees that are shown in Figure 8. Each of these trees was encoded for Hennig86 analysis (using the transformation-coding method de- scribed above, producing transformations numbers 77- 94 as used in the complete analysis of polygyrid genera: Emberton, 1994). Before analysis, however, transfor- mations 77-94 were all assigned a relative weight of 1/ 6 (using the ccode command of Hennig86). Transfor- mations in other characters received other weights, de- pending on each character’s total number of alternative character-state trees. Another point needs to be made respecting step 4 (ordering allelic combinations into character-state trees). Although ordering of character states is considered by many cladists to be a central tenet of phylogenetic sys- tematics (Hennig, 1966; Brooks & McLennan, 1991; Har- vey & Pagel, 1991; Wilkinson, 1992), Hauser & Presch (1991; Presch, 1992) and others maintain that, at the very least, ordered data should also be analyzed unor- dered to determine the robustness of the cladogram to hypotheses of character-state order. This procedure ap- plied to the polygyrid-genera allozyme data set (Em- berton, 1994), yielded the comparison shown in Figure 9. The cladogram for ordered allozyme data has the same THE NAUTILUS, Supplement 2 out WN XTVC OAFESYGDRPIMK TT LTE Allozymes unordered out WNXTVC OAFESYGDRPIMK {Ls TLL Figure 9. Cladograms from allozyme data on 20 polygyrid genera (from Emberton, in review). Allozymes ordered topology as that for unordered data, but with greater resolution (an increase from four nodes to eight nodes). DISCUSSION The present author’s obvious preference for discrete- parsimony analysis of allozyme data (= allozyme cla- distics) is mildly supported by the evaluation outlined in Figs. 1-3, which demonstrates a hypothetical case in which cladistics successfully reconstructs the true phy- logeny—even when rare alleles are undetected—while both UPGMA and frequency parsimony (as interpreted and hand-calculated by the present author) fail. Further eval- uations using both hypothetical and real data, and using other distance and parsimony methods, are needed to test these results, but it seems clear at least that the distance method of UPGMA bases phylogenetic infer- ence on both plesiomorphic and apomorphic characters, including autapomorphies. On first principles, therefore, UPGMA (and, in the present author’s incompletely in- formed opinion, other distance methods) should not be used to reconstruct phylogeny (Harvey & Pagel, 1991; Brooks & McLennan, 1991). Although it is true that UPGMA and cladistics results often are congruent, such cases only demonstrate relatively constant rates of allo- zyme evolution within the limitations of the data set, and do nothing to make the phylogenetic inference itself more robust. Frequency parsimony (Swofford & Berlocher, 1987) depends heavily on frequencies, which can vary widely within a taxon (e.g. Emberton, 1993), and furthermore requires an algorithm that is very difficult to program and costly in computer time to run. A preliminary pro- gram is available (FREQPARS: Swofford, 1988; not used for the present paper), but can handle only very small data sets and often does not produce the most parsi- monious solution(s) (Swofford, 1988; D. Lindberg, per- sonal communication, 1992). The practice of allozyme cladistics, as outlined above and demonstrated in Figs. 4-6, seems logical, objective, and empirically validated (Mickevich & Mitter, 1981, 1983; Buth, 1984). For polygyrid land-snail genera, it yielded a phylogenetic hypothesis that was generally both K. C. Emberton, 1994 Page 49 consistent with and complementary to a hypothesis based on an independent anatomical data set (Emberton, 1994, unpublished). Polymorphisms-both in allozyme and in morpholog- ical data— may at first seem a hindrance to cladistics. As Mickevich and Mitter’s (1981, 1983) method points out, however, polymorphisms can offer important clues to character evolution, and hence to taxon evolution. Or- dering of character states within characters may some- times lead to error, however, so it is important to analyze data both ordered and unordered (Hauser & Presch, 1991; Wilkinson, 1992; Hauser, 1992). Ordering does enhance phylogenetic resolution without changing topology among polygyrid genera (Figure 9; Emberton, 1994), but does not among Truncatella snail species (G. Rosenberg, per- sonal communication) and apparently does not among taxa in several non-molluscan data sets (Hauser & Presch, 1991). Clearly, each new data set must be evaluated in its own right. ACKNOWLEDGMENTS Supported in part by N.S.F. grant BSR-87—-00198 to the author and N.I.H. grant TMPI11373 to G. M. Davis. I also thank Gary Rosenberg for helpful discussion, M.G. Harasewych for inviting my participation in this sym- posium, and Harasewych and two anonymous reviewers for useful comments on a previous draft. LITERATURE CITED Berger, E.M. 1983. Population genetics of marine gastropods and bivalves. In: Russell-Hunter, W. D. (ed.). The Mol- lusca, Volume 6, Ecology. Academic Press, New York, p. 563-596. Brooks, D. R. and D. A. McLennan. 1991. Phylogeny, ecology and behavior: A research program in comparative biology. University of Chicago Press, Chicago, IL, 434 p. Buth, D. G. 1984. The application of electrophoretic data in systematic studies. Annual Review of Ecology and System- atics 15:501-522. Cain, A. J. 1983. Ecology and ecogenetics of terrestrial mol- luscan populations. In: Russell-Hunter, W. D. (ed.). The Mollusca, Volume 6, Ecology. Academic Press, New York, p. 597-647. Davis, G. M., W. H. Heard, S. L. H. Fuller, and C. Hesterman. 1981. Molecular genetics and speciation in Elliptio and its relationships to other taxa of North American Unionidae (Bivalvia). Biological Journal of the Linnean Society (Lon- don) 15:131-150. Emberton, K. C. 1988. The genitalic, allozymic, and con- chological evolution of the eastern North American Trio- dopsinae (Gastropoda: Pulmonata: Polygyridae). Malaco- logia 28:159-273. Emberton, K. C. 1991. The genitalic, allozymic, and con- chological evolution of the tribe Mesodontini (Gastropoda: Pulmonata: Polygyridae). Malacologia 33:71-178. Emberton, K. C. 1993. Over-representation of rare alleles in juveniles and lack of pattern in geographic distributions of alleles in a land snail. Malacologia 35:361-369. Emberton, K. C. 1994. Polygyrid land-snail phylogeny: ex- ternal sperm trading, early North American biogeography, iterative shell evolution. Biological Journal of the Linnean Society, 52: 1-81. Farris, J.S. 1972. Estimating phylogenetic trees from distance matrices. American Naturalist 106:645-668. Farris, J. S. 1988. Hennig86, Version 1.5. James S. Farris, Port Jefferson Station, New York, NY 11776. Harvey, P. H. and M. D. Pagel. 1991. The comparative meth- od in evolutionary biology. Oxford University Press, New York, NY, p. 239. Hauser, D. L. 1992. Similarity, falsification and character state order—-a reply to Wilkinson. Cladistics 8:339-344. Hauser, D. L. and W. Presch. 1991. The effect of ordered characters on phylogenetic reconstruction. Cladistics 7:243- 265. Hillis, D. M. 1989. Genetic consequences of partial self-fer- tilization on populations of Liguus fasciatus (Mollusca: Pulmonata: Bulimulidae). American Malacological Bul- letin 7:7-12. Hennig, W. 1966. Phylogenetic systematics. University of Illinois Press, Urbana, IL, p. 263. Hillis, D. M. and C. Moritz (eds.). 1990. Molecular system- atics. Sinauer Associates, Inc., Sunderland, MA, p. 588. Hoeh, W. R. 1990. Phylogenetic relationships among eastern North American Anodonta (Bivalvia: Unionidae). Mala- cological Review 23:63-82. Johnson, M. S., B. Clarke, and J. Murray. 1988. Discrepancies in the estimation of gene flow in Partula. Genetics 120: 233-238. Johnson, M. S., J. Murray, and B. Clarke. 1986. An electro- phoretic analysis of phylogeny and evolutionary rates in the genus Partula from the Society Islands. Proceedings of the Royal Society of London B 227:161-177. Mickevich, M. F. and C. Mitter. 1981. Treating polymorphic characters in systematics: a phylogenetic treatment of elec- trophoretic data. In: Funk, V. A. and D. R. Brooks (eds.). Advances in cladistics. New York Botanical Garden, Bronx, NY, p. 45-60. Mickevich, M. F. and C. Mitter. 1983. Evolutionary patterns in allozyme data: a systematic approach. In: Platnick, N. I. and V. A. Funk (eds.). Advances in cladistics 2. Columbia University Press, New York, NY, p. 169-176. Mickevich, M. F. and S. J. Weller. 1990. Evolutionary char- acter analysis: tracing character change on a cladogram. Cladistics 6:137-170. Platnick, N. I. 1989. An empirical comparison of microcom- puter parsimony programs, II. Cladistics 5:145-161. Richardson, B. J., P. R. Baverstock, and M. Adams. 1986. Allozyme electrophoresis. A handbook for animal system- atics and population studies. Academic Press, Sydney, p. 410. Sarich, V. M. 1977. Rates, sample sizes, and the neutrality hypothesis for electrophoresis in evolutionary studies. Na- ture 265:24-28. Sneath, P. H. A. and R. R. Sokal. 1973. Numerical taxonomy. Freeman, San Francisco, p. 573. Swofford, D. L. 1988. FREQPARS, Release 1.0. David L. Swofford, Illinois State Natural History Survey, 172 Nat- ural Resources Building, 607 E. Peabody, Champaign, IL 61820. Swofford, D. L. and S. H. Berlocher. 1987. Inferring evolu- tionary trees from gene frequency data under the principle of maximum parsimony. Systematic Zoology 36:293-325. Wiley, E. O. 1981. Phylogenetics. The theory and practice of phylogenetic systematics. John Wiley and Sons, New York, p. 439. Page 50 Wilkinson, M. 1992. Ordered versus unordered characters. Cladistics 8:375-385. Woodruff, D.S. 1989. Genetic anomalies associated with Cer- ion hybrid zones: the origin and continuance of new elec- tromorphic variants called hybrizymes. Biological Journal of the Linnean Society 36:281-294. Woodruff, D. S. and A. Solem. 1990. Allozyme variation in THE NAUTILUS, Supplement 2 the Australian camaenid land snail Cristilabrum primum: A prolegomenon for a molecular phylogeny of an extraor- dinary radiation in an isolated habitat. Veliger 33:129- 139. Wright, S. 1978. Evolution and the genetics of populations, Volume 4, Variability within and among natural popu- lations. University of Chicago Press, Chicago, IL, 580 p. THE NAUTILUS, Supplement 2:51-60, 1994 Page 51 Use of Random Amplified Polymorphic DNA (RAPD) Markers to Assess Relationships Among Beach Clams of the Genus Donax S. Laura Adamkewicz Department of Biology George Mason University Fairfax, VA 22030, USA M. G. Harasewych Department of Invertebrate Zoology National Museum of Natural History Smithsonian Institution Washington, DC 20560, USA ABSTRACT The polymerase chain reaction was used to amplify genomic DNA from nine populations of donacid clams representing six taxa occurring in three sympatric pairs. The randomly ampli- fied polymorphic DNA (RAPD) markers produced by this tech- nique successfully distinguished among all taxa. Each taxon possessed a unique subset of markers and one member of each sympatric pair differed from the other by several markers. The taxa also separated clearly into two groups, one North American and the other Caribbean. Use of RAPD markers as characters in a cladistic analysis produced well resolved phylogenetic trees of high consistency. Key words: RAPD, PCR, phylogeny, biogeography, Donax. INTRODUCTION Comparisons of closely related species are often infor- mative with regard to the functional biology of their shared characters, while comparisons of sympatric con- geners can provide insights to the selective forces and adaptive complexes that are important in speciation (Larson, 1989). Western Atlantic species of beach clams in the genus Donax are particularly appropriate candi- dates for such comparative studies. Six species or sub- species have been described from the coastal waters of the eastern United States, and nearly as many have been reported from the Caribbean (Morrison, 1971). All of these species are highly polymorphic for shell colors and patterns, and one (Donax variabilis Say, 1822) serves as a classic example of a hyper-variable species (Moment, 1962). Both in the western Atlantic and world-wide, Don- ax often occur as sympatric species pairs or triplets (Ab- bott, 1974; Ansell, 1983; Morrison, 1971) that partition their shared habitat. The selective forces that promote the rise and maintenance of hyper-variable polymor- phisms are poorly understood (Allen, 1988; Owen & Whitely, 1988) and may be clarified by comparative studies. Furthermore, the interaction of habitat parti- tioning and hyper-variability has been studied only in the snail genus Cepaea (Clarke, 1960). Our long-term goal is to investigate these questions in Donax. However, the sympatric co-occurrences of two or more similar and often highly polymorphic species have placed the sys- tematic status of some of these taxa in dispute. Morrison (1971) recognized six taxa along the Atlantic and Gulf coasts of the United States: 1) Donax fossor Say, 1822, which ranges from New York to North Car- olina; 2) Donax variabilis variabilis Say, 1822 (as Donax roemeri protracta Conrad, 1849!), which occurs from Virginia southward along both coasts of Florida and west- ward along the Gulf coast to Mississippi; 3) Donax par- vulus Philippi, 1849, with a range that extends from North Carolina to the eastern coast of Florida; 4) Donax dorotheae Morrison, 1971, which occurs along the Gulf coast from Florida to Louisiana; 5) Donax variabilis roe- meri Philippi, 1849 (as Donax roemeri roemeri Philippi, 1849), which ranges from the Mississippi delta westward along the coasts of Texas and Mexico; and 6) Donax texasianus Philippi, 1847 with the same range as D. variabilis roemeri. Only in the northern-most part of its American range is Donax represented by a single species, D. fossor. Elsewhere, species of Donax generally occur as sympatric pairs. In the Caribbean and along the coast of northern South America, the genus is represented by Donax denticulatus denticulatus Linné, 1758, D. den- ticulatus stephaniae Petuch, 1992, D. striatus Linne, 1767, and D. vellicatus Reeve, 1855, with two or three species occurring together. Donax denticulatus is the type species of the subgenus Chion Scopoli, 1777 (Gray, 1847). Where species of Donax co-occur, they subdivide the habitat in much the same way. As described by Morrison (1971), Donax parvulus, D. dorotheae, and D. texasianus all occupy the same habitat and are allopatric, replacing one another along the coast, each co-occurring with Don- ax variabilis. In every case, D. variabilis is larger, occurs 1 See Boss (1970) and Melville (1976) for details on the no- menclature of this taxon. Page 52 higher in the inter-tidal zone, and migrates more actively with the tides. The other three species are much smaller, occur at the bottom of the inter-tidal zone and spend much of the year sub-tidally. During at least some parts of the year, these species occur with D. variabilis and both can be collected in the same handful of sand. This co-occurrence, combined with morphological similarity, has placed the status and rank of several taxa in dispute. Abbott (1974) considered D. parvulus to be an offshore ecological form of D. variabilis. Loesch (1957) reviewed the names of donacid taxa reported from the Texas coast and stated that morphological intergrades had been re- ported between D. texasianus and what may have been D. dorotheae near Louisiana and between D. texasianus and D. variabilis roemeri along the Texas coast. Chanley (1969) suggested that D. fossor represents a [temporary, seasonal] summer extension of the range of D. variabilis. In the Caribbean, D. vellicatus, like several of the American taxa, remains primarily sub-tidal. Donax den- ticulatus and D. striatus partition their habitat somewhat differently, but the division is still based on the tendency to migrate with the tide, as well as a preferred position in the inter-tidal zone. Wade (1967, 1968) has observed that, although D. denticulatus and D. striatus did some- times occur together, D. denticulatus migrated actively throughout the tidal cycle while D. striatus maintained a constant position higher in the inter-tidal zone. Historically, separations of donacid species and sub- species have been based exclusively on shell morphology (primarily on size, inflation [obesity], and degree of stri- ation), which is subject to environmentally induced vari- ation. Molecular data are often suitable for resolving taxonomic questions of this nature where the cause of morphological differences cannot be ascribed to either genetic or environmental differences. The restriction of a molecular character, either an allele in an allozyme system or a DNA marker, to one member of a sympatric species pair is taken as evidence of the absence of inter- breeding between the two taxa. While molecular evi- dence cannot confirm genetic isolation when two pop- ulations are allopatric, it can at a minimum demonstrate genetic divergence. A previous attempt to differentiate between sympatric populations of D. parvulus and D. variabilis using allozyme data yielded ambiguous results (Nelson et al., 1993). While allele frequencies differed between the two groups, no alleles unique to either taxon were found. Because the allozyme data could not distin- guish the two taxa, additional molecular markers were sought. The use of randomly amplified polymorphic DNA (RAPD) markers to differentiate between closely related individuals, populations and species was introduced in 1990 by Williams et al. and by Welsh and McClelland. In essence, segments of genomic DNA are amplified by the polymerase chain reaction (PCR) using a single very short primer (9 to 11 nucleotides) whose sequence might occur multiple times within the genome. The RAPD amplification of genomic DNA produces a set of frag- ments of various molecular weights, their number and THE NAUTILUS, Supplement 2 size depending upon the number of times the primer sequence occurs in the genome, as well as the distances between pairs of primer sites. The RAPD technique has already been shown to produce genetic markers for Men- delian segregational analysis (Klein-Lankhorst et all, 1991), genetic markers to distinguish among individuals within one population (Smith et al., 1992), genetic mark- ers that identify cultivars within a species (Hu & Quiros, 1991), and genetic markers that discriminate among spe- cies within a genus (Kambhampati et al., 1992). As a necessary prelude to the long-term goal of in- vestigating morphological polymorphisms and ecological niche-partitioning in Donax, the present study seeks to resolve the systematic status of several western Atlantic donacid taxa and to discern their phylogenetic relation- ships. Failure of the allozyme data to resolve these taxa definitively has led us to investigate the utility of ran- domly amplified polymorphic DNA (RAPD) markers to distinguish between populations, subspecies and species in the genus Donax, as well as to determine the rela- tionships of these populations and taxa using cladistic methodology. MATERIALS AND METHODS 1. COLLECTION OF SPECIMENS Donax were collected at the six locations shown on the map in Figure 7. These collections included samples from nine populations, representing six taxa, which are listed in Table 1 and are further identified as follows. 1) Donax variabilis variabilis (DVF) and 2) Donax parvula (DPF) were collected in the spring of 1990 at Indiatlantic Beach, on the Atlantic coast of Florida. 3) Donax variabilis roe- meri (DVR) and 4) Donax texasianus (DTT) were col- lected at Corpus Christi, on the Gulf coast of Texas, in spring of 1990. 5) Donax variabilis variabilis (DVG) were collected in 1992 at Alligator Point on the Gulf coast of Florida. Collections were made in Jamaica in both May and November of 1991 for 6) Donax denticulatus (DDM) from Port Maria, a town on the northern coast of the island, along the town’s seawall 7) Donax denticulatus (DDN) from Negril, a town on the western side of the island, at the Cosmos Beach Club 8) Donax denticulatus (DDB) and 9) Donax striatus (DSB) from Black River, a town on the southern coast of the island, at the Bridge House Inn. Donax were collected by sieving sand from the inter- tidal zone of a beach. The animals were placed in plastic bags and kept cool until they could be identified to spe- cies, then frozen at —20° C and shipped to George Mason University. Thereafter, clams were maintained at —60° C until DNA was extracted. Attempts to collect Donax fossor and D. dorotheae at their respective type localities were unsuccessful, and these taxa are not included in the present study. Shells of the samples used in this study are deposited in the collections of the National Museum of Natural History, Smithsonian Institution. Catalogue numbers for voucher lots are listed in table 1. S. L. Adamkewicz and M. G. Harasewych, 1994 Page 53 Table 1. Sources of the nine populations of Donax used in this study. Collecting sites are those shown on the map in figure 7. Sample designation Taxon DVF Donax variabilis variabilis DVG Donax variabilis variabilis DVR Donax variabilis roemeri DPF Donax parvulus DTT Donax texasianus DDM Donax denticulatus DDN Donax denticulatus DDB Donax denticulatus DSB ' Donax striatus 2. EXTRACTION OF DNA To avoid contamination from food organisms, only mus- cle dissected from the foot of each clam was used to extract DNA. Approximately 30 mg of tissue was treated according to a protocol derived from that of Reeb and Avise (1990). Tissue was macerated in a 1.7 ml micro- centrifuge tube containing 400 ul TE buffer (10 mM Tris, 1 mM EDTA, pH 7.6) for about 30 seconds with a pestle driven by an electric drill. After adding 25 ul of 10% sodium dodecyl sulfate (SDS), the extract was in- cubated at 65° C for 30 to 60 minutes. Next, 70 ul of 8M potassium acetate was added, the mixture shaken, and chilled on ice for 60 minutes. After the extract was cen- trifuged at 14,000 x g for 10 minutes, the supernatant was transferred to a clean tube and the pellet discarded. The supernatant was chilled at —20° C for 2 minutes, spun again for 10 minutes, and again transferred to a clean tube. Next, 400 ul of chloroform and 400 ul of tris- saturated phenol were added, the tube shaken, and cen- trifuged for 5 minutes at 14,000 x g. The upper, aqueous layer was transferred to a clean tube, 400 ul of chloroform added, the tube shaken, and centrifuged for 5 minutes. The upper, aqueous layer was again decanted to a clean tube, treated with chloroform, and centrifuged. After the upper layer was transferred to yet another clean tube, Iml of cold 95% ethanol was added and mixed by gently inverting the tube. The sample was kept at —20°C for two minutes and centrifuged again for 10 minutes. The supernatant was discarded and the pellet was washed with 1 ml of 80% ethanol. After another centrifugation for 5 minutes at 14,000 x g, the ethanol was discarded and the pellet was dried in an incubator at 38° C for about 30 minutes. The cleaned DNA pellet was dissolved in 300 ul of TE buffer and kept at —20° C until needed. This procedure yielded DNA at concentrations rang- ing from 5 to 35 wg/ml. If initial PCR amplification failed, the DNA was further purified with “GeneClean II” (BIO 101 Inc., P.O. Box 2284, La Jolla, CA 92038) after which it amplified satisfactorily. 3. PCR AMPLIFICATION OF DNA The amplification protocol of Bowditch et al. (1993) was used in this study. “Amplitaq’” DNA polymerase, sup- USNM catalogue Collecting site number Atlantic coast of Florida 869538 Gulf coast of Florida 869539 Gulf coast of Texas 869540 Atlantic coast of Florida 869541 Gulf coast of Texas 869542 Port Maria, Jamaica 869543 Negril, Jamaica 869544 Black River, Jamaica 869545 Black River, Jamaica 869546 plied by Perkin-Elmer/Cetus at an activity of 8 units per ul, was used at a concentration of 0.5 units per sample (0.06 ul). The four nucleotide triphosphates were sup- plied by Pharmacia as 100mM stocks and mixed to make a single stock 0.25 mM for each dNTP. A special RAPD buffer was prepared according to the recipe: 100mM Tris, 5|00mM KCI, 19mM MgCl,, and 10 mg/ml bovine serum albumin (not acetylated). Primers came from two sources: 20 (designated OP-E) were from Operon Tech- nologies Kit E and 40 (designated LMS-P) were provided by the Laboratory for Molecular Systematics, National Museum of Natural History, Smithsonian Institution, where they had been synthesized. Between 5 and 15ng of DNA from an individual clam and 50ng of a single, short primer (10mer in all cases) were combined in 25 ul of a reaction mixture comprised of: 18 yl sterile distilled water, 2.5 ul] RAPD buffer, 2.5 ul deoxynucleotide mix, 1 ul primer, and 1 ul target DNA. The reaction mix was topped with mineral oil and placed in a Perkin-Elmer 4800 Thermocycler for 45 cy- cles of a RAPD amplification profile as follows: dissoci- ation of DNA for 1 minute at 94°C, annealing of primer for 1 minute at 36°C, polymerization of DNA for 2 min- utes at 72°C. The amplification products were loaded onto a 1.4% agarose gel, electrophoresed in TBE buffer (89mM Tris, 89mM Boric Acid, 2mM EDTA) and vi- sualized with ethidium bromide. Each primer produced a characteristic set of amplification products with sizes ranging from 0.3 to 3.0 kilobases, which appeared as bright bands on the agarose gels (Figure 1). Rather than measuring distances from the origin, approximate sizes were determined by comparison to fragments of known size in a mixture of lambda DNA cut with HindIII and $X174 cut with HaelII. To confirm that two bands were identical, samples were run in adjacent lanes of a gel. Throughout this paper, these products are referred to interchangeably as “amplification products,’ “DNA fragments,” or “RAPD markers.” 4. SELECTION OF PRIMERS AND MARKERS Sixty primers were screened on a panel of 24 individuals comprised of three clams from each of the sample pop- ulations except D. striatus (DBS). Each primer was used Page 54 DVF DVR DVG DTT DPF DDM DDNDSB S-1 - B= bee ere eho. Figure 1. RAPD amplifications generated by primer OP-E18. The horizontal bars over the gel join the three individuals from each population that were run on each gel. Sample designations above the bars refer to taxa and populations listed in table 1. The letter M identifies lanes containing molecular weight stan- dards (lambda DNA cut with HindIII + #X174 cut with HaelII). Standard bands and sizes in kilobases are depicted to the right of the gel. at least twice to amplify each screening DNA. A primer was judged to be suitable for use in this study if it met the criteria of: 1) amplification, that is, the production of clearly resolved DNA fragments, 2) reproducibility, with at least one DNA fragment appearing consistently and reproducibly in repeated assays of the same indi- viduals, and 3) commonality, or the presence of at least one DNA fragment in two or more populations (but not necessarily two or more taxa). Primers that met these requirements were not common. Approximately one fourth of the primers tested failed to meet criterion 1, with most of the remainder failing criterion 3. Criterion 2, reproducibility, was not a serious problem. For all of the markers chosen, amplification of DNA from the same individual produced the same results whether the am- plification was repeated in separate PCR experiments or replicated within the same PCR experiment. Identical results were produced when amplifications were repeat- ed using a Coy thermocycler, in which temperature changes much more slowly than in a Perkin-Elmer ma- chine. THE NAUTILUS, Supplement 2 The initial screening procedure identified five primers that produced a total of 17 RAPD markers that were informative for the purposes of this study. These primers and markers are described in Table 2 and representative results are shown in Figure 1. Of the two primers from the Laboratory for Molecular Systematics, primer LMS- PO1 is the same as primer AP8g of Williams et al. (1990) while primer LMS-P56 was designed and synthesized at LMS. A total of nine individuals from each population were assayed at least twice with each of the five primers. Samples with similar markers were run side-by-side in the replicate assay in order to facilitate direct compar- isons. 5. PHYLOGENETIC ANALYSIS: Data were analyzed and trees produced using Hennig86 version 1.5 software (Farris, 1988). The implicit enu- meration (ie;) algorithm was used in each series of anal- yses to insure that all shortest, equally parsimonious trees were found. Each RAPD marker was treated as a separate character regardless of which primer was used to generate it or which other bands from the same or other primers co- occurred with it. In an initial analysis (Analysis 1), the 17 markers listed in Table 4 were scored as either absent (0) or present (1) for each population, and a hypothetical outgroup, scored as lacking all 17 RAPD markers (all characters = 0), was used (for data matrix, see Appendix 1). In subsequent analyses, the markers were scored as absent (0), polymorphic (1), i.e. present in some but not all members of the population, or fixed (2), i.e. present in all members of the population. In Analysis 2, this data set was run unordered, using the same hypothetical out- group as in Analysis 1 (data matrix in Appendix 2). A third series of analyses used the same data matrix but, instead of the hypothetical outgroup, each of the basal taxa from Analysis 2 (DSB, DDN) was used in turn as the outgroup. The arrangement of character states on the resulting trees were examined using the Dos Equis (xx) and xsteps tree diagnostic commands. Table 2. DNA primers used in this study. For each primer the table shows: the sequence; the average number of DNA amplification products detected in an individual, with the range of averages among the nine groups following in parentheses; and the sizes of those DNA amplification products used as markers. Primers designated OP-E are from Operon Technologies Kit E and primers designated LMS-P were provided by the Laboratory for Molecular Systematics. Those fragments followed by an asterisk (*) are characteristic of the Caribbean taxa, while those followed by an ampersand (&) are characteristic of the Carolinian taxa. Note that, for any given primer, the smaller fragments are always characteristic of the Carolinian taxa. Average number of RAPD Primer Sequence OP-E07 5’-AGATGCAGCC OP-E16 5'-GGTGACTGTG OP-E18 5'-GGACTGCAGA LMS-PO1 5’-TGGTCAGTGA LMS-P56 5'/-AGATCTGCAG DNA fragments per clam (range) Size (kb) of useful RAPD markers 3.8 (3.14.3) 0.6 1.5 3.4 (2.6-4.3) 0.3& 0.5& 0.6% 0.9 1.1 3.0 (1.6-4.2) 0.5& 0.6* 0.9 2.8 (2.0-3.2) 0.5& 1.0* 1.2* 3.0 (2.2-3.6) 0.3& 0.6& 1.1* 1.2* S. L. Adamkewicz and M. G. Harasewych, 1994 Page 55 Table 3. Distribution of RAPD DNA markers in May and November samples of Donax denticulatus from Port Maria, Jamaica. As in tables 2 and 4, the RAPD marker identification designates the primer that produced the marker, the approximate size of each marker in kilobases, and our identifying marker number. Each entry in the matrix shows the number of individuals in which the RAPD marker was detected over the number of individuals tested (e.g., 3/9). Because Primer OP-E07 was not used with the May sample, data from this sample were not included in table 4, which serves as the data matrix for phylogenetic analysis. OP OP OP LMS Primer: E16 E16 E18 PO] Marker size (kb): 0.9 Li 0.9 12 Marker number: i 9 10 12 May 9/9 . 9/9 9/9 0/9 November 9/9 9/9 9/9 1/9 RESULTS 1. STABILITY OF RAPD MARKERS OVER TIME To assess the stability of the RAPD markers used in this study, two different samples of Donax denticulatus from Port Maria, Jamaica, were examined, one taken in May and the other in November of 1991. These two samples produced identical RAPD markers with only slight dif- ferences in frequencies between the two collections (Ta- ble 3). The only marker not present in both samples, marker 12 (LMS-P01/1.2kb), was the rarest, appearing in only 3 of a total of 21 individuals assayed. 2. DISTRIBUTION OF RAPD MARKERS AMONG POPULATIONS Table 4 summarizes the distribution of the 17 RAPD markers among the nine assayed populations, but does not contain data from the May sample of Donax den- ticulatus from Port Maria. Because the presence of two species in the sample from Black River was not discoy- ered until after the laboratory work was completed, these two populations have reduced sample sizes (3 D. den- ticulatus, 6 D. striatus). Of the 17 RAPD markers assayed, two (markers 8 and 9), each produced by a different primer, were present at varying frequencies in all nine populations. All three samples of D. variabilis (DVF, DVG, DVR), including the subspecies D. variabilis roemeri, had the same 11 markers appearing in at least one member of each pop- ulation. Of these 11 markers, two appeared in no other taxon. Similarly, all three populations of D. denticulatus (DDM, DDN, DDB) shared a set of 11 markers, three of which appeared in no other taxon. Twelve of the 17 RAPD markers were not shared between D. variabilis (DVF, DVG,DVR) and D. denticulatus (DDM, DDN, DDB). The three remaining taxa, each represented by a single population, showed clear affinities with either D. variabilis or D. denticulatus. The absence of unique markers in these taxa was an artifact of the criteria for primer selection (i.e.- that bands occur in at least two sample populations). Nevertheless, Donax parvulus (DPF) was distinguished from its sympatric congener D. var- iabilis variabilis (DVF) by the absence of RAPD markers LMS LMS OP OP LMS P56 P56 E16 E18 PO1 il. 1.2 0.6 0.6 1.0 13 14 15 16 17 4/8 8/8 9/9 9/9 9/9 6/6 5/8 8/9 9/9 8/9 1 and 2, while D. texasianus (DTT) differed from its sympatric congener D. variabilis roemeri (DVR) in lack- ing markers 1, 2, 3, and 7. Marker 2, which was fixed in all populations of Donax variabilis and present in no other taxon, appears to be a diagnostic marker for this species. The Caribbean species D. striatus (DSB) differed from D. denticulatus (DDB) in lacking 5 RAPD markers (10, 11, 15, 16, 17), of which two (10, 16) were fixed in D. denticulatus. Marker 16, which occurred in all in- dividuals of D. denticulatus tested, was unique to this species and may be used as a diagnostic marker for this species. Although diagnostic markers were not identified for some taxa, the absence of multiple RAPD markers in one member of each sympatric pair (including markers fixed in the other member of the sympatric pair) is taken as evidence that these pairs do not exchange genetic material. An empirical observation is that when primers produced RAPD markers characteristic of both Carolin- ian and Caribbean taxa, markers that distinguished Car- olinian taxa were invariably shorter than markers that were diagnostic of Caribbean taxa (Table 2). The sig- nificance of this observation is not yet clear. 3. PHYLOGENETIC ANALYSES An initial cladistic analysis (Analysis 1), scoring each of the 17 RAPD markers as absent or present (regardless of frequency) in each sample population and using a hy- pothetical outgroup in which all characters were scored as absent, produced a single most parsimonious tree (length = 20, ci = 85, ri = 91) that resolved all species level taxa but left populations and/or subspecies unre- solved (Figure 2). Fourteen of the 17 character trans- formations plotted unambiguously onto this tree (Figure 2). Each of the remaining three characters (markers 7, 10, 11) could either have been present in the common ancestor of all the taxa in this study and subsequently lost in a single taxon (Figure 2), or have arisen twice independently (Figure 3). An analysis of character po- larity using the out-group comparison method (Watrous & Wheeler, 1981) indicated that the presence of markers 7, 10, and 11 is plesiomorphic, as they occur in both the in-group and the out-group, while their loss, in each case Page 56 THE NAUTILUS, Supplement 2 restricted to a functional in-group, is apomorphic (Fig Saor DD od ”), = 2 Ll tl Ss A second analysis, scoring the RAPD markers as absent (0), polymorphic (1) or fixed (2), and employing the same hypothetical outgroup, produced four equally parsimo- nious trees (length = 32, ci = 90, ri = 92) when the data were run unordered. One tree, supported by all markers except 8, 10, and 11, matched the topology of the tree in Figure 8, except that all sample populations were resolved. In the other three trees, which differed only in the resolution of D. denticulatus populations and which were supported by all markers except 7, 8, and 13, Donax striatus emerged as the sister group to all remaining taxa. The nelsen consensus tree (length = 34, ci = 85, ri = 87) of these four trees is shown in Figure 4. Twelve of the 17 character transformations plotted uniquely onto the consensus tree, while markers (7, 8, 10, 11, 18) could be interpreted as evolving in several equally parsimonious scenarios. Analyses of character polarity using the out- group comparison method (Watrous & Wheeler, 1981) ea nom liees ou suggest that marker 7 was fixed in the Donax ancestor, SSS became polymorphic in the North American clade, and eventually lost in Donax texasianus, while marker 11, which was polymorphic in the Donax ancestor, was lost in Donax striatus, but became fixed in Florida popula- tions of Donax variabilis. The remaining markers could not be mapped onto the consensus tree (Figure 4) without reversals (markers 8, 13) or convergences (marker 10). Only marker 8 was incompatible with all of the initial trees. When the data were reanalyzed using Donax striatus as the outgroup, one most parsimonious tree (length = 28, ci =96, ri = 97), resulted (Figure 5). Likewise, when the Black River population of Donax denticulatus, served ; = = = = = = = = as the outgroup, a single, equally parsimonious tree (length = 28, ci =96, ri = 97), was produced (Figure 6). All analyses that employed an intermediate character S8e2,|2Seeeo lid state produced identical tree topologies for the Carolin- atari eprom 2S ian samples but differed in the resolution and/or rela- tionships of the Caribbean taxa and populations. LMS P56 1 13 6/6 5/9 3/3 4/6 LMS POL 1.2 12 1/9 1/9 1/3 6/6 0.6 Il OP E18 0.9 10 9/9 9/9 9/9 9/9 2/9 9/9 9/9 3/3 OP E07 15 8 9/9 9/9 9/9 6/9 9/9 2/9 7/9 2/3 5/6 OP E16 0.9 0.5 DISCUSSION RAPD markers observed in the nine samples, represent- ing six species or subspecies of Donax, showed a high degree of polymorphism both within and among taxa. Nevertheless, the polymorphisms did not obscure rela- tionships among the samples and the presence of these markers was stable over time. The distribution of these RAPD markers supports previously disputed distinctions between members of the following three sympatric pairs of species: Donax parvulus and D. variabilis variabilis, D. texasianus and D. variabilis roemeri, and D. denti- culatus and D. striatus, as well as between the similar but allopatric pair D. parvulus and D. texasianus. Among the earlier applications of RAPD methodology, the technique was used to distinguish between individ- uals, strains, cultivars, populations and species (e.g. Hu & Quiros, 1991, Kambhampati et al. 1992). The char- OP E16 0.5 4 9/9 9/9 9/9 7/9 9/9 0.3 3 8/9 9/9 6/9 5/9 OP E18 2 9/9 9/9 9/9 the marker in kilobases and a sequential number to identify the marker. Each entry in the matrix shows the number of individuals in which the RAPD marker was detected OP E16 ] 2/9 1/9 4/8 over the number of individuals tested (e.g., 3/9). In a few cases, the number of individuals tested was less than nine, either because fewer DNA samples were available (DDB, DSB) or because we were unable to score an individual for a particular marker (markers 1, 13, 14). When no individuals produced a marker, the entry is marked “—” rather than 0/n. Data are grouped to emphasize affinities among sample populations, rather than by primer. Size (kb): Number Table 4. Distribution of the RAPD DNA markers among sample populations. For each RAPD marker, the band designation identifies the primer, the approximate size of DVF DVG DVR DPF DTT DDM DDN DDB! DSB? LMS OP OP L] P56 = C«éEIG~—=é«CECLL' 1.2 0.6 0.6 14 15 16 5/8 8/9 9/9 6/9 8/9 9/9 1/3 2/3 3/3 6/6 1 Only 3 individuals available for testing. 2 Only 6 individuals available for testing DDM DDN DDB DVR DVG DVF 15,14 7,8,9,10,11 Page 57 = Za xe ow Qa a a > > > Q a a Q a a Cl=85 Rl= 91 LENGTH= 20 Figures 2-3. Phylogenetic tree resulting from analysis 1, in which all RAPD markers listed in table 4 were scored as absent or present in each sample population, regardless of frequency (data matrix in appendix 1). 2. Character transformations plotted onto tree. Markers 7, 10 and 11 are plotted as having arisen once and been subsequently lost in a single taxon. 3. Markers 7, 10 and 11 are plotted as having arisen twice independently. acterization of diagnostic markers that would identify Donax species was beyond the scope of this study, and our criteria for primer selection precluded the recogni- tion of markers unique to taxa represented by single samples. Even with this screening bias, fixed, species specific RAPD markers were discovered for Donax var- iabilis s.l. and for D. denticulatus, the two species in this study that were represented by multiple samples. Kambhampatiet al. (1992) successfully applied RAPD methodology to identify mosquito species and were able to cluster individuals of the same species correctly by applying phenetic algorithms (UPGMA) to markers gen- erated by two primers. Although the resulting pheno- gram did not reflect the ancestral relationships of the mosquito species, these authors did not rule out the utility of RAPD data for phylogeny reconstruction but sug- gested that a greater number of primers (>20) should be tested to find lineage-specific markers. Our results confirm their conjecture. A survey of 60 primers was necessary to select the 5 primers and 17 markers that we =) Figure 4. Nelson Consensus Tree of four equally parsimonious trees produced when all RAPD markers listed in table 4 were scored as absent, polymorphic, or fixed (data matrix in appendix 2) and all characters were run unordered. Character transfor- mations are plotted onto the tree. Where alternative, equally parsimonious character transformations were possible, prefer- Oo LL > > Q Q s faa) Zz ag 3 2 aq 2 © Oo. (an) 4 2 17 1] 13 1 Ie A : a ! SS; 1 2 2 7 0 10 1 3 1 13 2 Sh ? jaa) 15 1 ip) 16 2 Q 17 2 4 2 5 1 ; 0 i "1 of ° 12 th ; OF 3 0 14 = =) Cl=85 O 7 2 Ri = 87 aL 2 LENGTH = 34 10 2 ence was given to the ordered, but not polarized, transformation series (0 <-> 1 <-> 2) and to reversals (loss of marker) over convergent origins of a marker. Page 58 O Le > > a) a) THE NAUTILUS, Supplement 2 DVG DVF Cl = 96 Rl = 97 LENGTH = 28 Figure 5. Most parsimonious tree produced when all RAPD markers listed in table 4 were scored as absent, polymorphic, or fixed, and all characters were run unordered. Based upon results of the previous analysis (Figure 4), Donax striatus was selected as the outgroup. used to construct phylogenies. As RAPD markers have previously been shown to segregate in a Mendelian man- ner, behaving as dominant alleles (Williams et al., 1990), we treated the individual markers as homologous char- acters suitable for cladistic analyses. Although the anal- yses were run unordered, when character transforma- tions could be plotted onto the resulting tree in several equally parsimonious ways, preference was given to the ordered, but not polarized, transformation series (0 <- > 1 <-> 2) because this series reflects the manner in which alleles enter, are distributed within, and leave populations. Preference was also given to reversal (loss of a marker) over convergent evolution of a marker. Despite differences in outgroup selection and data scor- ing, all cladistic analyses produced the same, single, high- ly consistent tree for the five Carolinian samples. The inability to stably resolve the Caribbean samples is at- tributed in large part to the low number of samples. A plot of the consensus tree (Figure 4) on a map of geographic distributions of the taxa (Figure 7) illustrates that the Carolinian Donax species form a monophyletic clade, while the Caribbean species may represent either a clade (Figure 2) or grade (Figs. 5,6) depending on how the limited data are analyzed. Although the genus Donax Cl= 96 RI = 97 LENGTH = 28 Figure 6. Most parsimonious tree produced when all RAPD markers listed in table 4 were scored as absent, polymorphic, or fixed, and all characters were run unordered. Based upon results of a previous analysis (Figure 4), the Black River pop- ulation Donax denticulatus, was selected as the outgroup. Figure 7. A plot of the consensus tree (Figure 4) on a map of the locations of the collection sites for the nine sample popu- lations of Donax used in this study. Open circles indicate species that migrate with the tide. Solid circles indicate subtidal species. Sample designations as in Table 1. S. L. Adamkewicz and M. G. Harasewych, 1994 has been represented in the fossil record of the western Atlantic since the Oligocene (Gardner, 1943:105), pro- vincial boundaries have existed between molluscan fau- nas of the Gulf of Mexico and the Caribbean Sea since the late Oligocene or early Miocene (Petuch, 1988:48). The Recent Carolinian and Caribbean Provinces com- prise, respectively, the Caloosahatchian Province and a portion of the larger Gatunian Province, both ranging from the late Oligocene to the early Pleistocene (Petuch, 1988:fig.1). Thus, the considerable divergence between Carolinian and Caribbean Donax faunas, which share at most five of 17 markers (29% similarity), may have ac- cumulated over a period of approximately 25 million years. The topology of the phylogenetic tree indicates that, within the Carolinian province, the non-migratory, lower intertidal to subtidal habitat is the more primitive among Donax. Species occupying this habitat (D. texasianus, D. parvulus and D. dorotheae) likely diverged as a result of barriers to gene flow posed by the Mississippi River and the emergence of peninsular Florida, respectively. As D. parvulus appears to be the sister species of D. variabilis, it is likely that D. variabilis originated in the eastern Carolinian Province, although comparable RAPD data on D. dorotheae may make possible a more precise localization of the area of origin of D. variabilis. ACKNOWLEDGMENTS The authors are indebted to the staff of the Laboratory for Molecular Systematics (LMS), National Museum of Natural History, Smithsonian Institution, and particu- larly to Darrilyn Albright, for technical advice. The au- thors learned the RAPD technique at LMS and, with the help of John Slapcinsky, now at the Field Museum of Natural History, generated all of the RAPD data while working there. George Mason University provided major support for this project in the form of a sabbatical leave for S.L. Adamkewicz while a grant from Jeffress Trust provided her financial support. The authors thank Walter Nelson, Florida Institute of Technology, Paul Mikkelsen, Palm Beach County Department of Environmental Re- source management, and Robert Vega, Texas Parks and Wildlife Department, who collected some of the samples. We are grateful to Prof. Diana Lipscomb and Mr. John B. Wise of George Washington University for helpful discussions on phylogeny reconstruction. $.L. Adamke- wicz also wishes to thank her husband for his untiring help in collecting specimens for this project and for all earlier ones. LITERATURE CITED Abbott, R.T. 1974. American Seashells, Second Edition. Van Nostrand, Reinhold Company, New York. 663pp. Allen, J.A. 1988. Reflexive selection is apostatic selection. _ Oikos 51:251-252. Ansell, A.D. 1983. The biology of the genus Donax. IN De- velopments in Hydrobiology 19, sandy beaches as ecosys- tems, pp. 607-634. A. McLachlan and T. Erasmus editors. Page 59 H.J. Dumont, series editor. Dr. W. Junk Publishers. The Hague. Boss, K. J. 1970. Donax variabilis Say, 1822 (Mollusca: Bi- valvia): proposed validation under the plenary powers. Z. N. (S.) 1923. Bulletin of Zoological Nomenclature 27:205- 206. Bowditch, B.M., D.G. Albright, J.G.K. Williams and M.J. Braun. 1993. The use of RAPD markers in comparative genome studies. Methods in Enzymology 224:294-309. Chanley, P. 1969. Donax fossor: a summer range extension of Donax variabilis. The Nautilus 83(1):1-14. Clarke, B. 1960. Divergent effects of natural selection on two closely-related polymorphic snails. Heredity 14:423-443. Farris, J. S. 1988. Hennig86 Reference. Version 1.5 (18 pp. manual distributed with program diskette). Gardner, J. 1943. Mollusca from the Miocene and Lower Pliocene of Virginia and North Carolina. Part I. Pelecy- poda. United States Geological Survey Professional Paper 199-A. 178 p. 23 pls. Gray, J. E. 1847. A list of the Genera of the Recent Mollusca, their synonyma and types. Proceedings of the Zoological Society of London 1847:129-219. Hu, J. and C. F. Quiros. 1991. Identification of broccoli and cauliflower cultivars with RAPD markers. Plant Cell Re- ports 10:505-511. Kambhampati, S., W.C. Black IV and K. S. Rai. 1992. Ran- dom Amplified Polymorphic DNA of Mosquito Species and Populations (Diptera: Culicidae): Techniques, Statis- tical Analysis, and Applications. Journal of Medical En- tomology 29(6): 939-945. Klein-Lankhorst, R.M., A. Vermunt, R. Weide, T. Liharska, and P. Zabel. 1991. Isolation of molecular markers for tomato (L. esculentum) using random amplified poly- morphic DNA (RAPD). Theoretical and Applied Genetics 83:108-114. Larson, A. 1989. The relationship between speciation and morphological evolution. In: Otte, D. and J. A. Endler (eds. ). Speciation and Its Consequences. Sinauer Associates, Inc. Sunderland, MA. pp. 579-598. Loesch, H.C. 1957. Studies on the ecology of two species of Donax on Mustang Island, Texas. Publications of the In- stitute of Marine Science, University of Texas 4:201-227. Mayr, E. 1963. Animal Species and Evolution. Harvard Uni- versity Press, Cambridge, MA. 797 pp. Melville, R. V. 1976. Opinion 1057. Donax variabilis Schu- macher, 1817) Mollusca: Bivalvia) suppressed under the plenary powers; type species designated for Latona Schu- macher, 1817. Bulletin of Zoolological Nomenclature 33(1): 19-21. Moment, G.B. 1962. Reflexive selection: a possible answer to an old puzzle. Science 136:262-263. Morrison, J. P. E. 1971. Western Atlantic Donax. Proceedings of the Biological Society of Washington 83(48): 545-568, 2 pls. Nelson, W.G., E. Bonsdorff, and L. Adamkewicz. 1993. Eco- logical, morphological, and genetic differences between the sympatric bivalves Donax variabilis Say, 1822, and Donax parvula Phillipi, 1849. The Veliger 36(4):317-322. Owen, D. F. and D. Whiteley. 1988. The beach clams of Thessalonika: reflexive or apostatic selection? Oikos 51: 253-255. Petuch, E. J. 1988. Neogene history of tropical American mollusks, biogeography and evolutionary patterns of trop- ical western Atlantic Mollusca. The Coastal Education and Research Foundation, Charlottesville, 217pp. Page 60 Reeb, C.A. and J.C. Avise. 1990. A genetic discontinuity in a continuously distributed species: Mitochondrial DNA in the American Oyster, Crassostrea virginica. Genetics 124: 397-406. Smith, M.L., J.N. Bruhn, and J.B. Anderson. 1992. The fungus Armillaria bulbosa is among the largest and oldest living organisms. Nature 356:428-431. Wade, B. A. 1967. Studies on the West Indian beach clam, Donax denticulatus Linne. 1. Ecology. Bulletin of Marine Science 17:149-174. Wade, B. A. 1968. Studies on the West Indian beach clam, Donax denticulatus Linne. 2. Life History. Bulletin of Marine Science 18:877-901. THE NAUTILUS, Supplement 2 Watrous, L. E. and Q. D. Wheeler. 1981. The out-group comparison method for character analysis. Systematic Zo- ology 30(1):1-11. Welsh, J. and M. McClelland. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acid Research 18:7213-7218. Williams, J. G. K., Kubelik, A. E., Levak, K. J., Rafalski, J. A. and Tingey, S.C. 1990. DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acid Research 18:6531-6535. Appendix 1. Data matrix for initial cladistic analysis of Donax phylogeny. RAPD markers scored as absent (0) or present (1), regardless of frequency. For sample designations see Table 1. Samples Outgroup DVF o 4 la) OOOO O Or ©] > OOOQqoe SOPH ©] OOQOQooeOH HH i ©]| Ooooc9orH PHC] ss OoOQoc’P Mei © | XK OoOoochP mH © | @& ee i ie OS | RAPD markers eel Oo} —_ j=) — — — bo _ (ce) — > e On _ [o> _ ~] eS EE OO] LO on a) oo a) SiS tI Sa Oooo Se © FSH ooeoeee © Fi FHS Eee eee © OPH oeogegoee © Sm ooeooee © SPH HS oe9g ee oS Appendix 2. Data matrix for subsequent cladistic analyses of Donax phylogeny. RAPD markers scored as absent (0), polymorphic (1), or fixed (2). For sample designations see Table 1. Samples Outgroup DVF SSSCOCOCONNNOC!]W SeoooorrHNe-FoS]w SSSCONHENNNO!]SA Secoornnnnrolu SOSCONNNNNO!D DNONNORPRHEHOI]A S) La La] SeoooorHFHO]+e RAPD markers Serer PNY NNW OO] © co) — — — bo io (ee) — _ — 1S) i (o>) — ~] 10 Se No No) NFrFrFoOoCoOoOOSO FPNFNOOCOCOCSO NFrFrFOOOCOOCSO SHS 26o0Q0000°9 OSOnnwnnoocoocond SOnnNrFoocoooceo NNNNNYHYHEFE OO ONNNrFYNNNN OS THE NAUTILUS, Supplement 2:61-78, 1994 Page 61 Mitochondrial Genomes and the Phylogeny of Mollusks Jeffrey L. Boore! Wesley M. Brown Department of Biology University of Michigan 830 N. University Ave. Ann Arbor, MI 48109-1048, USA am ABSTRACT We are seeking a character set that reliably reflects the evo- lutionary origin of the phylum Mollusca and the relationships among molluscan classes. Such a character set must be: (1) present in all taxa; (2) unambiguously homologous; (3) changing at a rate appropriate for the taxonomic range; and (4) complex enough to make convergence highly unlikely. The arrangement of genes in mitochondrial DNA (mtDNA) appears to meet these criteria. With only a few exceptions, the mtDNA of all met- azoans contains the same 37 genes: 2 for ribosomal RNAs, 22 for transfer RNAs, and 13 for proteins. Comparing the ar- rangements of these genes among the 17 taxa for which they are known suggests that the rate of change is appropriate for resolving higher level relationships. These genes could poten- tially be arranged in >2 X 10° different ways; thus, the prob- ability of the same order by convergence is very small. We have determined the complete mtDNA sequence of the bivalve Mytilus edulis (Hoffmann, Boore & Brown, 1992). ThismtDNA differs from those of other metazoans in its unique gene ar- rangement, in encoding an additional tRNA (tRNAmet(AUA)), and in lacking one protein coding gene (ATPase 8). In order to test whether these features are typical for mollusks, and to investigate mitochondrial gene arrangements as a phylogenetic character for molluscan relationships, we have determined the mtDNA sequence for the polyplacophoran Katharina tunicata. Katharina mtDNA contains the gene for ATPase 8 and has a gene arrangement substantially different from that of Mytilus and much more similar to that of Drosophila. The different gene arrangements of Mytilus and Katharina provide numer- ous character states for investigating molluscan class relation- ships. By screening for gene junctions unique to one of these two arrangements, it may be possible to find patterns of rear- rangements which unite the remaining classes to reflect their evolutionary history. The DNA sequence already obtained al- lows rapid screening of other animals by two methods. First, the polymerase chain reaction (PCR) can be used to selectively amplify gene boundaries unique to one of these two arrange- ments. Second, a large number of animals can be rapidly tested for the general arrangement of several of the large, well-con- served genes by Southern blot analysis. We compare the mi- 1 Present address: Department of Cell Biology and Neuro- anatomy, University of Minnesota, 4-135 Jackson Hall, 321 Church St. SE, Minneapolis, MN 55455, USA tochondrial genome arrangements of Mytilus and Katharina and describe gene arrangement differences which are partic- ularly useful for each of these approaches, based on: (1) the phylogenetic information inherent in shared gene arrange- ments; (2) the availability of well-conserved probe sequences for Southern hybridization; and (3) the likelihood of sequences suitable for PCR primers in adjacent genes. — Key words: Mitochondrial DNA, Mollusca, Mytilus, Kathar- ina, phylogeny, gene order. INTRODUCTION TRADITIONAL APPROACHES TO PHYLOGENY Multicellular animals are grouped with fair confidence into various phyla based on shared general body plans. Establishing evolutionary relationships among the vari- ous phyla, and among the various classes and orders within each phylum, is much more speculative. Animal phylogeny is most ambiguous at these high levels, where the fossil record is least complete, homology of morpho- logical structures is least discernible, and long periods of time have erased traces of relatedness. Paleontological studies are limited in resolving these higher level relationships because of the scarcity of fossils from the very early history of life. By the time of the earliest fossil-rich period, the Cambrian, animal diversity was considerable, with nearly all currently-recognized higher taxa represented. Many early animal forms ex- hibit unique morphologies not recognizable as inter- mediate between other groups of animals. Analysis of morphological data is confounded by the great length of time that has elapsed since these taxa diverged. Determining homologous structures among an- imals with radically different body plans is contentious. Convergence of morphological structures among various animals is common and difficult to recognize. Hypotheses based on considerations of functional morphology or re- capitulation of embryological structures lack the desired rigor and falsifiability. Many alternative, often contra- dictory models of metazoan evolution have been pro- posed, all based on interpretations of the same embry- ological, paleontological, and morphological data sets. Page 62 Nowhere are these difficulties more acute than in de- termining the relationships among the various classes and orders of the phylum Mollusca, or in determining the sister taxon to this phylum. Although most classification schemes agree on uniting the classes Scaphopoda, Bi- valvia, Cephalopoda, Gastropoda, and Monoplacophora into a monophyletic Conchifera, the relationships among these classes are equivocal. Monoplacophora has been suggested as the sister group to Gastropoda based on studies of comparative morphology (Gotting, 1980) and paleontological material (Knight & Yochelson, 1960). Sal- vini-Plawen (1985) views Monoplacophora as basal to a clade of the remaining four conchiferan classes. It has also been suggested that Monoplacophora unites with Gastropoda and Cephalopoda in the trichotomous Cyr- tosoma (Runnegar & Pojeta, 1974). Cephalopoda is sug- gested as the basal conchiferan (Gotting, 1980) or as the sister group to Gastropoda (Wingstrand, 1985; Runnegar & Pojeta, 1974). Finally, Scaphopods may be primitive mollusks (Lindberg, 1985) or they may be included in a clade with the Bivalvia and the extinct rostroconchs, the Diasoma (Runnegar & Pojeta, 1974). Relationships among the non-conchiferans are even more ambiguous. Early classification schemes united Aplacophora with Polyplacophora in the Amphineura, first considering neither to be mollusks (von Ihering, 1876), and later recognizing their molluscan affinity (Pelseneer, 1899). Some modern day classifications view a mono- phyletic Amphineura as the basal group of mollusks (Po- jeta, 1980; Haas, 1981) although some continue to ques- tion their inclusion in Mollusca (Fretter & Graham, 1962). Another view places Aplacophora as the primitive, basal class of mollusk, creating a clade of the Polyplacophora and the Conchifera (Gotting, 1980; Wingstrand, 1985; Scheltema, 1988). Salvini-Plawen (1985) not only views the Aplacophora in this basal position, but splits Apla- cophora into two paraphyletic groups, the Caudofoveata and the Solenogastres, a scheme supported by several later studies (Meglitsch & Schram, 1991; Pearse et al., 1987; Brusca & Brusca, 1990; Barnes, 1987; Nielsen, 1985, 1987). Still other analyses view the Conchifera as the primitive group of mollusks, with the Polyplacophora, Caudofoveata, and Solenogastres secondarily derived (Marcus, 1958; Hadzi, 1953, 1963). The closest sister taxon to the phylum Mollusca has been suggested variously to be Annelida (Gotting, 1980; Vagvolgyi, 1967), Arthropoda (Lemche, 1959a,b; Fretter & Graham, 1962), Sipunculida (Inglis, 1985), Turbellaria (Graham, 1955; Salvini-Plawen, 1972, 1980; Haas, 1981), Echiurida, or Nemertina (see discussion in Salvini-Plaw- en (1985) and Vagvolgyi (1967)). A clade of consisting of Annelida and Mollusca is suggested to have been de- rived from flatworms (Hammarsten & Runnstrom, 1925; Boettger, 1959) or coelenterates (Beklemishev, 1963). The debates on the relationships among molluscan classes and on the evolutionary origin of Mollusca center on alternative models of morphological change. These models differ in interpreting which types of change are feasible or common, to what extent convergence and THE NAUTILUS, Supplement 2 parallelism occur, and which structures are homologous among the various metazoan bauplane. Resolution of molluscan relationships would permit many conclusions regarding the evolution of the coelom, serially repeated structures, and body segmentation. MOLECULAR PHYLOGENIES Molecular phylogenies based on comparing the sequenc- es of nucleotides or amino acids have offered many new insights into the evolutionary relationships among or- ganisms. During the development of the technology to make this feasible, some hoped that all questions of phy- logeny would eventually be answered with a high con- fidence using these techniques. Although it would be hard to overstate the contribution of molecular ap- proaches to evolutionary biology in recent decades, many questions of organismal relationships remain recalcitrant. Numerous molecular studies of metazoan relationships have compared the nucleotide sequence of 18S ribosomal RNAs (rRNAs). This molecule was chosen for sequence comparisons in part because its high copy number in the cells of many organisms allows sequence determination directly (without the need for cloning the gene itself) using the enzyme reverse transcriptase. Although they effectively outline the broad pattern of relationships among kingdoms of organisms (Pace et al., 1986; Woese & Fox, 1977; Woese, 1987; Lake, 1988; Sogin, 1991; Wainright et al., 1993), when applied to relationships among metazoans these comparisons have yielded phy- logenies that are mutually contradictory and difficult to reconcile with patterns of morphological divergence (Field et al., 1988; Lake, 1990; Patterson, 1989). While only three molluscan classes are represented (Bivalvia, Polyplacophora, and Gastropoda), various methods of phylogenetic analysis of the 18S rRNA data yield con- tradictory results of the relationships among these classes and their relationships to other animals (see figure 1). There are numerous limitations of DNA sequence comparisons for phylogeny, especially when considering ancient divergences: (1) With only four possible char- acter states at each aligned position, homoplasy is com- mon and difficult to recognize. (2) These characters states are difficult to reliably polarize as primitive versus de- rived. (3) Unequal rates of nucleotide substitution in various lineages can lead to erroneous linkages when using distance-based phylogenetic methods, or to the “long-branch attraction” problem when using parsimo- ny-based analyses (Felsenstein, 1978). By counting the number of lineage-specific nucleotide substitutions for 20 taxa for a sample of 1000 rRNA nucleotides, Ghiselin (1988) concluded that substitution rate varies over an 18- fold range among the metazoa. (4) The effects of gene conversion in multigene families may produce patterns of change among groups of organisms that are difficult to deduce (Arnheim, 1983; Sasaki et al., 1987; Walsh, 1985; Coen, Strachan & Dover, 1982). (5) Small subunit rRNAs vary in length by almost a factor of two, so many gaps must be introduced for nucleotide alignment. This J. L. Boore and W. M. Brown, 1994 Page 63 Polychaete A Pclychaete B Clams C Brachiopod Brachiopod Sipunculid Chiton Chiton Chiton Pogonophora Pogonophora Snail Sipunculid Oligochaete Brachiopod Oligochaete Clams Pogonophora Snail Sipunculid Oligochaete Clams Snail Polychaete Brine shrimp Human Brine shrimp Insect Sea star Horseshoe crab Millipede Brine shrimp Horseshoe crab Planarian Brine shrimp Polychaete Insect Polychaete D Oligochaete F Millipede Oligochaete Clams Horseshoe crab Pogonophora Brachiopod Clams Clams Chiton Polychaete Chiton Pogonophora Brachiopod Brachiopod Sipunculid Chiton Sipunculid Snail Pogonophora Snail Brittle star Snail Brine shrimp Human Oligochaete Insect Brine shrimp Sipunculid Horseshoe crab Planaria Human Millipede Figure 1. Six evolutionary hypotheses resulting from 18S rRNA comparisons. In some cases, taxa that do not group within the depicted clade have been omitted in order to emphasize the placement of the mollusks. (A) Tree from Field et al. (1988), derived using a distance method of analysis for 839 aligned positions. Mollusca is paraphyletic with the inclusion of two annelids, a brachiopod, a pogonophoran, and a sipunculid. Gastropoda is primitive; however, no scaphopods, cephalopods, aplacophorans, or monoplaco- phorans are included in this analysis. (B) A portion of the rooted evolutionary tree from Lake (1990), using the method of evolutionary parsimony with exactly the same data and alignments as in (A). Mollusca is paraphyletic with the inclusion of a sipunculid and a brachiopod. (C)-(F) Various results of parsimony analyses from Patterson (1989). (C) A portion of the tree produced by a strict consensus of the four shortest trees in a parsimony analysis of 544 aligned nucleotides. Data and alignment are the same as in A and B, but with the addition of two prokaryotic 16S rRNA sequences as outgroups. (D) The tree produced using the same data as for (C), but limiting the taxa analyzed to those depicted, and rooting the tree with the planarian sequence. (E) A portion of the tree produced by limiting the analysis of (C) to only those 273 aligned positions which represent unpaired nucleotides in the rRNA secondary structure. (F) A portion of the tree produced by strict consensus of three equally parsimonious trees, generated by limiting the analysis of (C) to only those 271 aligned positions which represent paired nucleotides in the rRNA secondary structure. Taxa used in these analyses: Clams-M ya arenaria and Spisula solidissima, Polychaete-Chaetopteris sp.; Brachiopod-Lingula reevi; Chiton- Cryptochiton stelleri; Pogonophora-Riftia pachyptila; Oligochaete-Lumbricus sp.; Sipunculid-Goldfingia gouldii; Snail-Anisodoris nobilis; Human-Homo sapiens; Sea star-Asterias forbesi; Brine shrimp-Artemia salina; Planarian-Dugesia tigrina; Insect-Drosophila melanogaster, Millipede-Spirobolus marginatus; Horseshoe crab-Limulus polyphemus; Brittle star-Ophiocoma wendtii. alignment is often so ambiguous that different portions of the molecule are compared between the various or- ganisms to eliminate regions which could not be satis- factorily aligned (by subjective criteria). A sequence alignment is fundamentally an hypothesis of homology at each of the aligned positions; changing the alignment of nucleotides can generate very different evolutionary trees. (6) There is disagreement about which parts of the rRNA sequence provide reliable data. Some contend that the most reliable phylogenetic trees are generated by using only the subset of nucleotides that are paired in rRNA secondary structure (Smith, 1989), some argue that the most reliable information is in unpaired nucleotides (Wheeler & Honeycutt, 1988; in a study using 5S and 5.8S rRNA sequences), and some are unable to find a significant difference between the two (Vawter & Brown, 1993). Various weighting schemes for these two subsets of nucleotides lead to different hypotheses of relation- ships and are based on intractable aspects of the evolu- tionary history of the molecule (Dixon & Hillis, 1993; Patterson, 1989: see Kraus et al., 1992 for discussion about mitochondrial rRNA sequences). (7) Applying dif- ferent tree-making methods to the same data frequently yields very different results (e.g., see figure 1). These methods differ in their assumptions of the evolutionary process, and we have no reliable method for discerning Page 64 which is most realistic. Most insidious is the tendency to judge results most reliable when congruent to previously accepted hypotheses of relationships. Such circularity questions the potential contribution of molecular phy- logenies. MITOCHONDRIAL GENOMES We are interested in discovering a better set of characters for determining molluscan (and other metazoan) rela- tionships. We seek a character set that reliably reflects the evolutionary origin of the phylum Mollusca and the relationships among molluscan classes. To be useful for phylogenetic inference, there are several properties that such a character set must possess: (1) Character states should be determinable for all taxa so no states are shared between organisms as “missing.” (2) Characters should be demonstrably homologous among the organisms. (3) Character states should be complex enough so that it would be highly unlikely that the same character state could have arisen independently in two or more lineages. Therefore, identical character states would likely be shared by two or more taxa only because of common ancestry. (4) Character states should change at a rate appropriate for the time span being investigated. Too little change limits resolution; too much change obscures relationships. We are investigating a set of molecular characters that appear to possess the above properties and show promise to be useful for determining ancient divergences: the arrangement of genes in mitochondrial DNA (mtDNA). MtDNA exists as a discrete genome within the cells of all metazoans, generally as a closed circular DNA of about 14-17 kilobases (kb) (Wolstenholme, 1992a,b; Brown, 1985; Wallace, 1982; Attardi, 1988)!. Although much larger mtDNAs are occasionally found in a variety of metazoan taxa, including Mollusca (Moritz, Dowling & Brown, 1987; Snyder et al., 1987; LaRoche et al., 1990), in none of these cases is there any evidence for variation in gene content. MtDNA typically contains one or more large non-coding sequences, which can vary significantly in length among organisms. This region in vertebrates (Clayton, 1991, 1992; Montoya et al., 1982; Bogenhagen, Cairns & Yoza, 1985; King & Low, 1987) and insects (Clary & Wolstenholme, 1985a) has been shown to include elements for the control of replication and transcription. Large variation in lengths of mtDNAs has been due to variation in the length of this major non- coding region (Brown, 1985; Harrison, 1989; Carr, Broth- ers & Wilson, 1987; Harrison, Rand & Wheeler, 1985; Fauron & Wolstenholme, 1976; Solignac, Monnerot & Mounolou, 1986; Wilkinson & Chapman, 1991; Buroker 1 An exception is found among some (but not all) cnidarians, where mtDNA is present as one or two linear molecules totaling about 16 kb (Warrior & Gall, 1985; Bridge et al., 1992). THE NAUTILUS, Supplement 2 et al., 1990; Monforte, Barrio & Latorre, 1993) or to duplication of some portion of the mitochondrial genome (Moritz & Brown, 1986, 1987; Zevering et al., 1991). The arrangement of the genes in mtDNA appears to meet the above four criteria for a character set to be used for phylogenetic inference. The gene content of metazoan mtDNA is well conserved. With few excep- tions (Wolstenholme et al., 1987; Okimoto et al., 1991, 1992; Hoffmann, Boore & Brown, 1992), the mtDNA of all animals examined contains the same 87 genes: 2 for the small and large rRNAs of the mitochondrial ribosome (s-rRNA and |-rRNA), 22 for transfer RNAs (tRNAs), and 13 for protein subunits of the enzyme complexes of the inner mitochondrial membrane [cytochrome oxidase subunits I-III (CO1-3), NADH dehydrogenase subunits 1-6 and 4L (ND1-6, 4L), cytochrome b (cytb), and ATP synthase subunits 6 and 8 (ATPase 6, 8)]. Mitochondrial genes are certainly homologous among metazoa, not only because of their functional identity and obvious sequence similarity among metazoans, but also because of the high degree of their similarity in these respects to the mito- chondrial genes of non-metazoans. Table 1 shows that each of the 15 protein- or rRNA-encoding genes has a homologue in the mtDNA of one or more non-metazoan organisms. It seems clear that these 15 genes were en- coded in mtDNA prior to the origin of Metazoa, and that the gene content of the mtDNA of extant organisms was established prior to this radiation. The arrangement of mitochondrial genes is a very complex character set for phylogenetic inference, be- cause there are a very large number of possible gene arrangements. Assuming complete positional indepen- dence of all genes and two transcriptional orientations, these 37 genes could potentially be joined in greater than 2 X 10°? different arrangements; thus, the probability of the same order arising independently in more than one taxon is vanishingly small. Identical gene arrangements would likely be shared only as a result of common an- cestry, making homoplasy rare. Preliminary studies suggest that the rate of change in mitochondrial gene arrangement is appropriate for re- solving ancient divergences. Although mtDNA evolves rapidly in sequence (Brown, George & Wilson, 1979), rearrangements in gene order appear rare. All 37 genes are identically arranged in the mitochondrial genomes of several vertebrates, including Homo (Anderson et al., 1981), Bos (Anderson et al., 1982), Mus (Bibb et al., 1981), Rattus (Gadaleta et al., 1989), Balaenoptera (Ar- nason, Gullberg & Widegren, 1991), Phoca (Arnason & Johnsson, 1992), Xenopus (Roe et al., 1985), Crossostome (Tzeng et al., 1992), Cyprinus (Chang, Huang & Lo, 1994), and Gadus (Johansen, Guddal & Johansen, 1990), although minor rearrangements have taken place in mar- supial mammals (Paabo et al., 1991) and in birds (Des- jardins & Morais, 1990, Desjardins, Ramirez & Morais, 1990). The gene arrangement of the mtDNA of the ceph- alochordate Branchiostoma floridae (W. Brown & L. Daehler, unpublished data) is very similar to that of vertebrates. Similarly, mitochondrial genomes have un- J. L. Boore and W. M. Brown, 1994 Page 65 Table 1. The 15 protein- or rRNA-encoding genes typical of metazoan mtDNA, and a listing of non-metazoan taxa in whose mtDNAs homologous genes have been positively identified. Genes are abbreviated as in the text. Additional genes are also present in these non-metazoan mtDNAs. The non-metazoan species are Paramecium aurelia, Leishmania tarentolae, Trypanosoma brucei, Chlamydamonas reinhardtii, Marchantia polymorpha, Neurospora crassa, Saccharomyces cerevisiae, Schizosaccharomyces pombe, Aspergillus nidulans, and Podospora anserina. Data are from Boer and Gray, 1991; Brown et al., 1985; Clark-Walker, 1989; Cummings and Delmenico, 1988; Cummings et al., 1990a,b; Dewey et al., 1985; de Zamaroczy and Bernardi, 1986; Dyson et al., 1989; Feagin et al., 1988; Gray and Boer, 1988; Ise et al., 1985; Lang et al., 1983; Oda et al., 1992; Pratje et al., 1989; Pritchard et al., 1990; Simpson et al., 1987; Stuart and Feagin, 1992; Wolf and Del Guidice, 1988; Vahrenholz et al., 1985. Cytochrome oxidase ATPase Cyt NADH dehydrogenase rRNA J 2 3 6 8 b ] ® 3 4L 4 D 6 Sml Lrg Metazoa xX x xX xX xe xX x Xx xX xX xX xX xX xX x Paramecium x xX xX xX xX xX xX xX xX x Leishmania xX x x xX xX x x xX xX xX Trypanosoma xX xX xX xX x xX xX Xx xX Xx Chlamydamonas x x x x x x Xx x x Marchantia x x x x xX x xX x x x xX x xX x Neurospora x xX xX xX xX x xX xX x x x xX x xX Saccharomyces xX x xX xX x xX xX xX S. pombe x Xx x x x x x xX Aspergillus x x x x x x x x x x x x xX x xX Podospora x x x x xX xX xX xX xX xX xX xX xX xX xX * ATPase 8 appears to be missing from the mitochondrial genomes dergone little gene rearrangement among echinoderm classes, with only a single large inversion separating the arrangements of sea stars and sea urchins (Jacobs et al., 1988; Cantatore et al., 1989; De Giorgi et al., 1991; Himeno et al., 1987; Smith et al., 1989, 1990, 1998). In each of these deuterostome phyla, organisms which have been separated for over 500 million years share very similar mitochondrial gene arrangements. The traditionally accepted superphylum, Protostomia, includes three major phyla: Arthropoda, Mollusca, and Annelida. Drosophila and Apis represent the only genera within arthropods for which the complete mtDNA se- quences are published (Clary & Wolstenholme, 1985a; De Bruijn, 1983; Garesse, 1988; Crozier & Crozier, 1993). These two mitochondrial genomes are nearly identically arranged, differing only by a few tRNA gene translo- cations. Partial gene arrangements for numerous other arthropods (including Artemia, Locusta, Aedes, Daph- nia, Homarus, and Limulus) indicate that gene order is highly conserved in this phylum (Batucus et al., 1988; Hsuchen, Kotin & Dubin, 1984; Dubin, Hsuchen & Til- lotson, 1986; McCracken, Uhlenbusch & Gellissen, 1987; Uhlenbusch, McCracken & Gellissen, 1987; D. Stanton, L. Daehler & W. Brown, unpublished data). Representing pseudocoelomate animals, the mtDNA of two nematodes, Caenorhabditis elegans and Ascaris suum, have nearly identical gene arrangements (Wol- stenholme et al., 1987; Okimoto et al., 1992), although that of a third, Meloidogyne javanica, has a radically different arrangement (Okimoto et al., 1991). Partial genome organization has been determined for the par- asitic flatworm, Fasciola hepatica (Garey & Wolsten- holme, 1989) and is unique among animals examined to of Mytilus and nematodes (see text). date. Although gene rearrangements appear to be gen- erally rare within a phylum, there are substantial dif- ferences among the gene orders of each of these phyla. Furthermore, the radically different arrangements of mi- tochondrial genes that exist among several of the taxa suggest that selection for any particular gene order is minimal, and argues against this as a factor that might promote convergence of gene order in separate lineages. Comparisons of mtDNA gene arrangements may also be useful for phylogeny at lower taxonomic levels. There is evidence that the frequency of rearrangements of the mitochondrial tRNA genes is greater than that of the rRNA and protein genes. For example, the gene orders of two diptera, Aedes (Dubin, Hsuchen & Tillotson, 1986, Hsuchen, Kotin & Dubin, 1984) and Drosophila (Clary & Wolstenholme, 1985a,b), differ in the relative positions of two tRNA genes and by the relative inversion of a third, but are otherwise identical insofar as can be de- termined from the partial sequence of Aedes mtDNA. The partial gene order determined for another insect, Locusta, differs from Drosophila by one tRNA rear- rangement (MacCracken, Uhlenbusch & Gellissen, 1987; Uhlenbusch, MacCracken & Gellessen, 1987). Eight tRNAs must be repositioned to interconvert the mito- chondrial gene arrangements of Apis (Crozier & Crozier, 1993) and Drosophila. The mitochondrial genome ar- rangement of several marsupials differs from that of pla- cental mammals by translocations within a cluster of five tRNAs (Paabo et al., 1991). As more gene orders become available, useful information about phylogenetic rela- tionships among even more closely related animal groups (ordinal and subordinal levels) may occasionally be ob- tained. Page 66 COMPARISON OF THE MITOCHONDRIAL GENOME ARRANGEMENTS OF MYTILUS EDULIS (BIVALVIA) AND KATHARINA TUNICATA (POLYPLACOPHORA) We have determined the DNA sequence of 13.9 kb of the 17.1 kb mitochondrial genome of the bivalve Mytilus edulis, which is sufficient to identify all 37 mitochondrial genes (Hoffmann, Boore & Brown, 1992). The arrange- ment of genes in Mytilus mtDNA is radically different from those that have been found in other metazoans. With few exceptions, the arrangement of mitochondrial genes appears to be very similar or identical in within- phylum comparisons, so it was initially unclear whether the unusual gene arrangements in Mytilus mtDNA was typical of mollusks in general or characteristic of a more restricted group of molluscan taxa. To investigate this, and to evaluate further the potential of mitochondrial genome structure as a useful phylogenetic indicator, we determined the complete mtDNA sequence for the po- lyplacophoran Katharina tunicata. The Katharina mi- tochondrial gene arrangement differs substantially from that of Mytilus and is much more similar to the mt DNAs of other coelomate animals, including an annelid (Boore & Brown, manuscript in preparation), and representa- tives of other classes in the phylum Mollusca (W. Brown, T. Collins & L. Daehler, unpublished data). The mitochondrial genome of Mytilus edulis, in par- ticular, contains several unusual features in comparison with others previously characterized. Mytilus mtDNA lacks a gene for ATPase 8. This gene is also absent from the mitochondrial genomes of the three nematodes men- tioned above, although presumably these absences rep- resent convergent losses in the Mytilus and nematode lineages. Mytilus mtDNA encodes 23 tRNAs, one more than the typical metazoan mitochondrial complement. The anticodon of the additional tRNA is complementary to the codons for methionine, giving Mytilus mtDNA two methionine tRNA genes. One of these genes specifies a tRNA with the anticodon CAT, which is typical of other metazoan mitochondrial tRNA™* genes; the other is nearly unique among all genomes in having the an- ticodon TAT. The arrangement of these 37 genes is high- ly unusual, with few gene boundaries shared with any other metazoan so far investigated. The reading frames of the ND1, CO1, and CO8 genes vary significantly in length from those of other metazoans, more so than found in any previous comparisons. However, in other respects this genome is typical of metazoan mtDNA. Aside from the supernumerary tRNAmet genes, its gene content is typically metazoan, its tRNA and rRNA genes are small relative to those found in prokaryotic and eukaryotic nuclei, its gene organization is highly compact, and its genetic code appears to be identical to that employed in several other metazoan mitochondrial systems. All genes are encoded by the same DNA strand, as is the case in other (but not all) metazoans. Radical variation in the arrangement of mitochondrial genes has been demonstrated previously in comparisons THE NAUTILUS, Supplement 2 among metazoan phyla, most notably between nema- todes and the other phyla examined. However, only mi- nor variation in mitochondrial genome arrangement is usually observed in within-phylum comparisons. It was, therefore, surprising to find the mitochondrial gene ar- rangement of another mollusk, the polyplacophoran Ka- tharina tunicata, to be much more similar those of ar- thropods, chordates, or echinoderms than to that of Mytilus. Katharina mtDNA encodes the 37 genes typical of metazoan mtDNA, including the gene for ATPase 8 which is absent from Mytilus mtDNA. There are at least three possible explanations for the loss of the ATPase 8 gene in the lineage leading to Mytilus after its separation from the Katharina lineage: (1) The normal function of sub- unit 8 of the ATP synthase complex is subsumed by another protein subunit; (2) The function of ATPase 8 has become dispensable in the metabolism of Mytilus mitochondria; or (3) The ATPase 8 gene has been trans- ferred to the nucleus, and its gene product is now im- ported into the mitochondria. If it could be determined whether other mollusks share the absence of the ATPase 8 gene from mtDNA, such molecular or metabolic changes could be a very complex derived character, robust for phylogenetic analysis. In addition to the 22 tRNAs typical of metazoan mi- tochondrial genomes, Katharina mtDNA contains two additional sequences that can be folded into structures resembling tRNAs. If actual tRNAs, their anticodons (AAA and AGA) would presumably recognize the codons UUU and UCU as phenylalanine and serine, respectively. They are, therefore, provisionally identified as tRNAPb(UU0) and tRNAs*UCU) in figure 2. However, there are several reasons to doubt that they actually function as tRNA genes. The anticodons AAA and AGA are unprecedented in metazoan mtDNA. None of the tRNAs encoded in the mtDNAs of Katharina, Mytilus, or Drosophila have an A in the 8rd (‘wobble’) position. Both of these putative tRNAs have several mismatches within their stems and neither has a T preceding the anticodon, as is found in all other Katharina tRNAs. They do, however account for nearly all of the nucleotides in what would otherwise be unassigned sequence, and their predicted secondary structures are no more aberrant than those of many other mitochondrial tRNAs. Figure 2 shows the mitochondrial gene arrangements of Katharina tunicata, Mytilus edulis, and Drosophila yakuba (the latter determined by Clary & Wolstenholme, 1985a). If we ignore the positions of tRNA genes, only two rearrangements are necessary to interconvert the gene arrangements of Katharina and Drosophila: a trans- position of the CO8-ND8 segment, and an inversion of the ND6-Cytb segment. The genes encoding tRNAs ap- pear to rearrange at a much higher frequency, as has been noted previously, with numerous tRNAs differing in position between Drosophila and Katharina for both nearest-neighbor genes, namely those for leu(UUR), lys, asp, gly, ser(AGN), glu, ile, gln, met, and trp. In contrast, there is little in common when comparing J. L. Boore and W. M. Brown, 1994 Page 67 a F GEID \/ KL(CUN) V M(AUA)RAH S(UCN) —S 4 = Ze) ~ ee M(AUe)> S(AGN) F(UUU)? S(UCU)? £ H aT P LC(UUR) Vv MYOE KRI S(AGN) | Vv — ~m carne [co Bf as [olf mf fama m D S(UCN) L(CUN) CWG AN L(UUR) K G ANE H U S(UCN) L(CUN) V 1M wy : sa) = wo Kk Drosophila C01 C02} NDS | ND4 | S}//9 | Cytb NDI |l-rRNAlsrava | + | ND2 D RIE p Q ¢ S(AGN) Figure 2. Comparison of mitochondrial gene arrangements among Katharina tunicata (Boore & Brown, submitted), Mytilus edulis (Hoffmann, Boore & Brown, 1992), and Drosophila yakuba (Clary & Wolstenholme, 1985), with each genome aligned starting at the gene for CO1. All genes of Mytilus are transcribed from left to right, as are all genes in the Katharina and Drosophila genomes other than those designated by underlining to signify reverse orientation. Ignoring tRNA position differences, which are numerous, rearrangements are shown by lines connecting gene pairs or blocks of contiguous genes (marked by a bar). Inversions are indicated by a circular arrow. Gene designations are as follows: cytochrome oxidase subunits I-III, CO1-3; NADH dehydrogenase subunits 1-6 and 4L, ND1-6, ND4L; cytochrome b apoenzyme, Cytb; ATP synthase subunits 6 and 8, A6, A8; small and large ribosomal subunit RNAs, s-rRNA, l-rRNA. Transfer RNAs are designated by the one letter code for the corresponding amino acid; the two tRNAs each for serine and leucine are further differentiated by the codon recognized (UCN and AGN for serine; UUR and CUN for leucine). M(AUA) of Mytilus mtDNA denotes an additional methionine tRNA and F(UUU)? and S(UCU)? designate additional tRNA-like structures of Katharina mtDNA. UNK (unknown) designates the largest unassigned region of the Mytilus and Katharina mtDNAs. A+T in Drosophila designates the A+T rich non-coding region. the mitochondrial gene arrangement of Mytilus with three gene block, tRNA'\CUN)-tRNA!CUR)_NDI. The either Katharina or Drosophila. Ignoring tRNA genes, only gene boundaries shared by the Drosophila and My- only the two rRNA genes are in the same order and tilus mitochondrial genomes are those of CO2-tRNA*, transcriptional polarity in the three animals. However, tRNA"-ND4L (although here the relative polarity of in Drosophila and Katharina mtDNA (and many other tRNA" is reversed), and tRNA™'-ND2 (although here metazoans) the two rRNA genes are separated by the tRNA™* of Mytilus is has the anticodon TAT whereas tRNA”, whereas in Mytilus mtDNA they are separated the tRNA™* of Drosophila has the anticodon CAT). by seven tRNAs, none of which is tRNA”. The only gene The genes of metazoan mtDNA are typically arranged boundaries shared by the Katharina and Mytilus mito- very compactly. Introns are absent, intergenic nucleo- chondrial genomes are those of tRNA''-ND4L and the tides are few, and genes frequently overlap or end on Page 68 Taxa: a ly € | Taxa: a ly) © GG State: 0 ONTO State: 0 Pe os) SS A B Taxa: a b oc d State: 0O @ jl 1 € Figure 3. An explanation of the method for analyzing the evolutionary significance of patterns of mitochondrial genome rearrangement. Each taxon is represented by a letter, with “a” designated as the outgroup. Sharing a number for the character state represents sharing a gene boundary; differing numbers indicate that the taxa differ in the gene boundary. Three types of patterns may occur: (A) taxa “b” and “c” share a primitive gene arrangement (symplesiomorphy), since the “0” state ex- isted prior to the origin of “b’, “c’, or “d”. Hence “b” and “c” cannot be united to the exclusion of “d” by this shared gene boundary. (B) taxa “b’, “c’, and “d” each have unique gene arrangements (autapomorphies). (C) the only pattern of gene arrangements that can be used to unite taxa. Because taxon ‘“b” shares the “0” state with the outgroup, the state is polarized to indicate “c’’ and “d” share the derived “1” state as a synapomorphy. To place taxon “b” within the clade con- taining ‘c’ and “d” would be less parsimonious, since it would require either the reversion to the “0” state in taxon “b” or the convergent gain of the “1” state in taxa “c” and “d”. Consid- ering the large number of potential character states at each gene boundary, such reversion or convergence to identical states is improbable. abbreviated stop codons (Wolstenholme, 1992a,b; Brown, 1985; Moritz, Dowling & Brown, 1987; Attardi, 1988). Because genes often abut directly or overlap, any genome rearrangement would require very precise breakage and recombination for the resultant genome to produce func- tional products. This barrier to recombination has been offered as one possible explanation for the conservation of arrangement of mitochondrial genes over long periods of time (Brown, 1985). In Mytilus mtDNA there are five lengthy intergenic sequences, four of which range in size from 79 to 119 nucleotides and the fifth of which is 1.2 kb. It may be that the relatively lengthy regions of DNA without apparent function between genes in Mytilus mtDNA enable recombination, accounting for what ap- pears to be an unusually rapid rate of rearrangement, although this remains undemonstrated. For most metazoans, mtDNA is inherited maternally (Lansman, Avise & Huettel, 1983; Dawid & Blackler, 1972; Gyllensten, Wharton & Wilson, 1985). This is part- ly due to the vastly greater number of mitochondria in the egg cytoplasm than in the sperm. Sperm cells typi- cally contain only a few mitochondria, and these are localized in the midpiece, which is often excluded from the egg during fertilization. Mytilus is a notable excep- THE NAUTILUS, Supplement 2 tion in this regard. Its mtDNA is frequently inherited biparentally and two or more variant forms of mtDNA often occur within an individual, a condition known as heteroplasmy (Hoeh, Blakely & Brown, 1991; Zouros et al., 1992). Another bivalve, the scallop Placopecten, also exhibits frequent heteroplasmy as well as large variations in mitochondrial genome size (Snyder et al., 1987; Gjet- vaj, Cook & Zouros, 1992; LaRoche et al., 1990). It is possible that one or more of these unusual features is responsible for the highly derived state of mtDNA in Mytilus and, possibly, in other bivalves. This, however, is very speculative, and a much broader survey of bivalve mtDNAs is needed to determine when the radical vari- ation in gene arrangement occurred. ANALYSIS OF GENE ARRANGEMENTS It would be overly simplistic and perhaps wrong to sug- gest that a shared arrangement of mitochondrial genes by itself indicates a close evolutionary relationship. Taxa may share a gene arrangement because they inherited it in an unchanged form which existed ancestral to the divergence of the several taxa being considered (sym- plesiomorphy; figure 3A); likewise, the gene arrange- ments of closely related taxa may differ due to a rear- rangement that is unique to one of the lineages (autapomorphy; figure 3B). Neither of these patterns of gene arrangement is indicative of phylogeny. The gene arrangements that are useful for phylogenetic inference are those which can be demonstrated, by comparison with appropriate outgroups, to be shared in a derived form (synapomorphy; figure 3C). Admittedly, most phy- logenetic branching events will not coincide with mi- tochondrial gene rearrangements. However, when a de- rived arrangement is shared by two or more taxa, it is extremely likely to indicate common ancestry. We cannot infer that Katharina andMyftilus are dis- tantly related simply because of the great number of differences in the arrangement of their mitochondrial genes. To infer such a distant relationship would be to assume that mtDNA gene rearrangements occur in a clock-like manner, an assumption that existing data re- fute. The differences in gene order between these two mtDNAs can only be interpreted as an autapomorphy, given the data at hand, and therefore as phylogenetically uninformative. In the same manner, we cannot infer that Katharina and Drosophila share a more recent common ancestor than Katharina and Mytilus based on the great- er similarity of the Katharina and Drosophila gene ar- rangements; this similarity is a symplesiomorphy, and may represent the state of arrangement in the common ancestor of all three taxa. As in figure 3A, all arrange- ments of the ingroup taxa are equally parsimonious, therefore these characters are not phylogenetically in- formative. It is critical that gene arrangements be determined not only for several representatives of each of the classes of Mollusca, but also for all potential outgroup taxa that might be useful for determining whether gene rear- rangements are primitive or derived. A larger survey of J. L. Boore and W. M. Brown, 1994 animal mitochondrial genomes might reveal interme- diate genome arrangements and, perhaps, identify non- molluscan taxa that have these arrangements. However, with increasing numbers of gene arrangements to com- pare, determining precisely the most parsimonious pat- tern of rearrangement becomes exponentially more dif- ficult. Techniques are currently being developed to provide computer analysis of genome rearrangements (Sankoff et al., 1990,1992). COMPARISON OF OTHER ASPECTS OF THE MITOCHONDRIAL GENOMES OF MYTILUS EDULIS AND KATHARINA TUNICATA Molecular phylogenies have been limited largely to com- parisons of linear sequences of nucleotides or amino ac- ids. An entire field of scientific inquiry has developed from the need to deduce phylogeny most accurately from these sequence comparisons. However, genomes contain many other complex features that can be compared, such as the arrangement of genes, the relative positions of deletions and insertions, the number and position(s) of regulatory sequences, numerous sequence-based aspects of transcription, translation, and DNA replication, vari- ations in the genetic code, and secondary structures of transfer and ribosomal RNAs. The analysis and comparison of such characteristics among the large and complex nuclear genomes of met- azoans will be most informative, but is presently im- practical. However, many of these features can be easily accessed for comparison among the much smaller and simpler mitochondrial genomes. Metazoan mtDNA is 25,000 times smaller than the smallest nuclear genome, contains few non-coding nucleotides, has a consistent gene complement and, at least in some organisms, does not appear to undergo genetic recombination (see Wol- stenholme, 1992a,b; Brown, 1985; Moritz, Dowling & Brown, 1987; Wallace, 1982). The tRNA genes are usually interspersed among the protein- and rRNA-coding genes of metazoan mtDNAs. Their product tRNAs fold into complex secondary struc- tures due to internal base-pairing, and it is likely that these structures are present in the polycistronic RNA transcripts and are used as recognition sites by RNA processing enzymes (Ojala et al., 1980; Ojala, Montoya & Attardi, 1981). In Katharina mtDNA there are four gene junctions which lack an intervening tRNA, and at each there is a potential secondary structure that positions the start codon of the downstream gene at an identical relative location (Boore & Brown, submitted). If these structures actually form in vivo, they may substitute for tRNAs as signals for transcript cleavage. In Mytilus mtDNA there are also sequences capable of forming potential secondary structures in the several lengthy in- tergenic regions, and these may also play a role in the processing of the polycistronic transcript. By investigat- ing mitochondrial RNA processing in mollusks, it may be possible to determine whether these secondary struc- tures actually form, whether they serve as signals for processing enzymes, and whether any of the RNA pro- Page 69 cessing mechanisms are evolutionarily derived for (or within) Mollusca. Such information is, thus, potentially relevant for molluscan phylogeny. Animal mtDNA uses several variations of the genetic code (see Jukes & Osawa, 1990, 1993, and Wolstenholme, 1992a,b). TGA specifies tryptophan. ATA specifies me- thionine in all but echinoderm and cnidarian mtDNA. AGA and AGG specify serine in echinoderms, arthro- pods, nematodes, and platyhelminths, arginine in cni- darians, glycine in ascidians, and are probably stop co- dons in mammalian mtDNA. AAA usually specifies lysine, but in echinoderm and platyhelminth mtDNA it specifies asparagine. Comparisons of codon usage patterns and protein alignments suggest that both Mytilus edulis and Katharina tunicata mtDNAs have a genetic code that is identical to that of arthropods. Two major protostome phyla, Mollusca and Arthropoda, are therefore united in this feature. While nuclear genes initiate translation exclusively with the methionine codon ATG, metazoan mitochondrial genes employ several additional initiation codons, in- cluding ATT, ATA, ATC, GTG, TTG, GTT, and ATAA (see Wolstenholme, 1992b). In all but one case, it is un- clear whether the initial amino acid of mitochondrial proteins varies with the start codon used, or whether the alternate start codons are somehow recognized by a methionyl-tRNA when they occur as the initial codon of a mRNA. For one human mtDNA gene and transcript, Fearnley and Walker (1987) have determined by se- quencing the corresponding protein that ATT in the initiator position specifies methionine, but that it specifies isoleucine when it is in an internal position. Katharina mitochondrial genes appear to initiate translation with ATG, ATA, and GTG. Mytilus mitochondrial genes ap- pear to initiate translation only with ATG or ATA, al- though the possible use of other start codons cannot be ruled out due to significant ambiguity in determining the start point of several genes. Both ATA and ATG code for methionine within the reading frames of mitochondrial proteins. It is not ob- vious how the single methionyl-tRNA encoded in most mitochondrial genomes discriminates initiation codons, which are translated with N-formyl-methionine, from internal methionine codons, since both may be either ATA or ATG. For this reason, it is intriguing thatM ytilus mtDNA contains two tRNAs for methionine. However, since the codons expected to pair most efficiently with each of these tRNAs (ATA and ATG; the anticodons of these two tRNAs are UAU and CAU) are present as both initiation codons and in internal positions, the differential use of these tRNAs in initiation and protein extension is not likely. Termination of translation is also unusual in metazoan mtDNA. In the mtDNA of protostomes all codons are typically used within open reading frames except the stop codons TAA or TAG. However, many genes end with “abbreviated” stop codons of T or TA. In human mitochondria, where it has been investigated, the gene- specific message is precisely cleaved after a T or TA, the first or first and second nucleotides of the terminal codon, Page 70 after which the stop codon TAA is completed by poly- adenylation of the gene specific message (Ojala, Montoya & Attardi, 1981). Both of the mitochondrial genomes characterized in this work appear to employ this mech- anism commonly. Nuclear encoded tRNAs are invariant for a number of primary and secondary structure features, such as the typical “three-leaf clover’ structure, the nucleotides T, pseudo-U, C in one arm (designated the TC arm) and dihydrouracyl in another (the DHU arm; see Lewin, 1987). The tRNAs of metazoan mtDNA are much more variable, both in primary and secondary structure. Nem- atode mitochondrial tRNAs are especially unusual; each tRNA is unpaired for the entire TYC arm. (Wolsten- holme et al., 1987; Okimoto et al., 1991, 1992). This feature is of great potential use for assessing the hypoth- esis of monophyly of the Aschelminthes, the group into which nematodes are often placed. All sequenced metazoan mtDNAs contain two differ- ent tRNA*™ genes, recognizing codons AGN and UCN, respectively. It is common for one of these, the tRNA*"4CN) to lack the potential for base-pairing in the DHU arm, and this characteristic is found in both Ka- tharina and Mytilus. However, in Katharina mtDNA the DHU arm is unpaired in the second serine tRNA, tRNAseUCN) as well. Since both serine tRNAs must be charged with the same amino acid, perhaps in Katharina the DHU portion of the tRNA structure is recognized by the same charging enzyme. If true, and if other mol- luscan taxa share this shift in tRNA structure and mech- anism for serine tRNA charging, this would also be a useful character for phylogenetic analysis. Sequence de- termination of the serine tRNA genes from other mol- lusks will allow us to assess this. Based on analysis of DNA sequence alone, each of the two mollusk classes investigated may encode one or more tRNAs in addition to the normal metazoan complement of 22. Mytilus mtDNA may contain an additional tRNA for methionine; Katharina mtDNA may contain addi- tional tRNAs for serine and phenylalanine. Further in- vestigation of molluscan mtDNAs will reveal whether the presence of supernumerary tRNA genes is common in this phylum. If it is, then this may also indicate that there are molecular mechanisms in the mitochondrial system of some mollusks that are specific to the evolu- tionary history of this phylum and that can be used for phylogenetic inference. Ribosomal RNAs also form elaborate secondary struc- tures through internal base pairing. In comparisons of rRNA gene sequences from various organisms, it is ap- parent that deletion or addition of large structures in the rRNA has been a common mode of evolution (Clary & Wolstenholme, 1985b; Zwieb, Glotz & Brimacombe, 1981). Perhaps these large scale changes accompany a shift in ribosome functioning. We are currently devel- oping models of secondary structure for the small and large rRNAs of these two mollusk mitochondrial genomes in hopes of identifying structural variations that might be used to infer relationships among molluscan lineages. THE NAUTILUS, Supplement 2 FUTURE DIRECTIONS Initially, it was surprising to find that the arrangement of genes in the mtDNA of Katharina was so different from that of Mytilus. Although radical variation in mi- tochondrial gene arrangement has been noted in com- parisons among phyla (as in nematode versus coelomate mtDNAs), mitochondrial genome rearrangements ap- pear to be infrequent within phyla. The very different mitochondrial gene arrangements of Mytilus and Ka- tharina provide numerous character states for investi- gating molluscan relationships. By screening additional molluscan mitochondrial genomes for gene junctions unique to one of these two arrangements, and by com- paring these arrangements with those of non-mollusks, it may be possible to deduce the broad pattern of the evolutionary history of Mollusca. Specifically, the first goal is to investigate mitochondrial gene arrangements in representatives of each of the remaining classes of mollusks. Non-molluscan protostomes must also be in- vestigated to help characterize gene arrangements as primitive or derived. Investigating gene arrangements by determining com- plete mtDNA sequences is very costly and laborious. With the knowledge gained from the complete mtDNA sequences of Katharina and Mytilus, it may be possible to employ less costly and easier methods to screen other molluscan taxa for particular gene arrangements that are likely to be phylogenetically informative. For example, the polymerase chain reaction (PCR) can be used to selectively amplify gene boundaries that are unique to one of these two arrangements, and a large number of animals can be rapidly tested for the arrangement of several of the large, well-conserved genes by Southern hybridization analysis. The development of DNA amplification via PCR (see Innis et al., 1989) has enabled DNA sequence determi- nation without the difficult and time-consuming proce- dures of restriction mapping and cloning. As outlined in figure 4, a segment of DNA is amplified to sufficient quantity for gel analysis and DNA sequence determi- nation by employing two oligonucleotides complimen- tary to flanking sequences. These oligonucleotides serve as primers for the synthesis of new DNA strands by a thermostable DNA polymerase. The success of a PCR amplification is critically de- pendent on the complementarity of the oligonucleotide primers to the sequences flanking the DNA to be am- plified. The complete mtDNA sequences of Katharina and Mytilus aid in primer design for screening additional molluscan mtDNAs in two ways. First, since closely re- lated organisms are more likely to share sequence iden- tities, the sequences of the primers can be chosen to match well conserved portions of these two genomes, increasing the likelihood of a successful amplification in the target genome. Second, the gene arrangements ofMytilus and Katharina mtDNAs give hypotheses of gene arrangement to test on additional animals, since the PCR can only be successful if the primers “face” one J. L. Boore and W. M. Brown, 1994 another (see figure 4). Since there is a practical limit to the size of a DNA sequence that can be successfully amplified, primers must be selected in genes that are likely to be closely spaced in molluscan mtDNA. Gene arrangements to be investigated by PCR can be selected based on both the likelihood of finding sequences suitable for PCR primers in adjacent genes and on the amount of phylogenetic information in the sharing of particular gene arrangements. One strength of this approach is that DNA sequence information is concurrently gained, which can be used in sequence-based phylogenetic analyses as a separate test of relationships. In addition, this technique precisely maps contiguous gene arrangements, and may identify tRNA rearrangements which would be invisible to the technique of Southern hybridization. Crude and highly impure DNA preparations can be used for amplification, including those made from ancient tissues and museum specimens (Thomas, et al., 1990; Kocher et al., 1989). PCR amplification also enables the analysis of DNA from very small organisms and from tiny portions of tissue from rare ones. The main disadvantage of PCR is that it can not iden- tify novel gene arrangements, but only test for hypoth- esized arrangements (other than small gene insertions). Primers must be designed to amplify a specific segment, opposing one another over a short segment of DNA. If the flanking sequences to which the primers are designed have rearranged significantly, no amplification will oc- cur. This negative result provides no information, be- cause amplification may fail for numerous reasons in addition to gene rearrangement (e.g., because there have been mutations in a few nucleotides in the region com- plementary to the primer sequences). Figure 5 illustrates a gene arrangement that may be amenable to investigation in other molluscan mtDNAs through PCR amplification. The region to be amplified is flanked by the genes for the I]-rRNA and cytb. Each of these two genes individually has been successfully amplified by PCR from a variety of organisms (Kocher et al., 1989; D. Stanton and W. Brown, unpublished data). The I-rRNA and cytb genes are well-conserved, based on comparisons among widely divergent taxa; this maximizes the likelihood of finding primer sequences that are useful over a broad taxonomic range. Determining the arrangement of these particular genes in other mollusks may yield phylogenetically useful in- formation. In Katharina and Drosophila mtDNAs, ND1 is between |-rRNA and cytb. This represents a symple- siomorphy, since vertebrate mtDNAs share this charac- teristic as well. The block of genes tRNA!MCUN)- tRNAMCUR)_NDI is shared between the mtDNAs of Ka- tharina and Mytilus, but in the latter this entire block is translocated to another region of the genome. Any mollusks that share this translocation with Mytilus mtDNA are likely to have a common evolutionary history with Bivalvia. In comparing the mtDNAs of Katharina and Drosophila, the block tRNAS*(YC)-cytb-ND6- tRNA? has been inverted. Any mollusks that share this Page 71 A S E Digestion S Electrophoresis B Hybridization E Eee B S eo B E Gene is located here Oligonucleotides matching flanking sequence B ==> ves | uccuny | nor seucny |__| <= Double-stranded DNA TUTTE Heat to denature, cool to anneal primers TTTTTITITITIIITIITITIIIITITITIITIITTTtittitittiittity ——_— LITITITI TTI Extend new strands TTTTTTTTTTTTT TTT TTT TTT TTT ttt rrr tririttiiritT ——pe LLL TOT LUT TPIT TTT Repeat for 20-40 cycles for exponential DNA amplification Figure 4. Two alternative techniques for determining the ar- rangement of mitochondrial genes. (A) The technique of Southern hybridization (Southern, 1975). A cleavage map of the relative locations of restriction enzyme sites in the mtDNA is constructed. Each enzyme recognizes a particular short se- quence of DNA (4-8 bp). E, S, and B in this figure represent the locations of three independent restriction enzyme (EcoRI, SalI, and BamHI) cleavage sites on the circular map of the mitochondrial genome. The mtDNA is cleaved with each en- zyme, the fragments produced are separated according to size by electrophoresis through an agarose gel, the fragments in the gel are transferred to a membrane and probed using a radio- labeled DNA fragment that contains all or part of the gene of interest. The fragment patterns generated by each restriction enzyme are labeled E, S, and B on the depiction of the gel. The radiolabeled probe will hybridize only to the fragments of the mtDNA that contain the corresponding gene, shown in bold. This information, when correlated with the mtDNA cleavage map, provides the gene’s position in the mitochondrial genome. (B) The technique of PCR amplification (Innis et al., 1987). Oligonucleotide primers that are complementary to the DNA sequence flanking the region of interest are determined and synthesized. The double-stranded template DNA is heat denatured, mixed with a vast excess of the oligonucleotides, then cooled to allow annealing of the oligonucleotides to the template. The oligonucleotides serve as primers for the synthesis of new strands of DNA in a reaction using thermostable DNA polymerase. This cycle of heat (denaturation) and cool (anneal and synthesize) is repeated many times (typically 20-40) to exponentially amplify the DNA region between the primers to provide amounts that allow manipulation and determination of the DNA sequence. Page 72 THE NAUTILUS, Supplement 2 CUN Drosophila T=cRNA 2] Apis eer re ea a UCN CUNUUR Katharina (eae ia ae a a P| UCN CUNUUR syed line coz UUR VeAtCREARSS [Tea vine JSS amore B Echinoderms fairies AND He te tdogvas Metridium Figure 5. The arrangements of several mitochondrial genes particularly amenable to investigation by PCR (see text) and potentially informative for molluscan phylogeny. Gene abbreviations are as in figure 2. Genes are transcribed from left to right except those depicted below the main line to designate opposite orientation. The broken line shown for vertebrate mtDNA indicates a large, undepicted portion of the genome. The positional relationships of genes in the insect and mollusk mtDNAs are depicted as in figure 2; the arrangement of these genes in mtDNAs of other organisms are shown for comparison (references in text except for Metridium, D. R. Wolstenholme, personal communication). Genes are not drawn to scale. inversion with Katharina are likely to also share a com- mon ancestor with Polyplacophora. If the arrangements of these genes is determined for representatives of all molluscan classes, it should be possible to formulate a phylogenetic hypothesis for the Mollusca, and to test the hypothesis with additional gene arrangement data. The second technique for rapidly screening mito- chondrial gene arrangements is Southern hybridization (Southern, 1975), which localizes genes relative to a phys- ical map of the mtDNA. This technique is outlined and described in figure 4. Cleavage sites for restriction en- zymes which recognize specific short sequences of DNA determine the physical map. Restriction endonuclease cleavage generates DNA fragments, which are separated by size using gel electrophoresis, visualized, and related back to their position in the mtDNA. Gene-specific probes that are labeled with a radioisotope are exposed to the gel-separated DNA bands under hybridizing conditions. The probes hybridize specifically to the DNA bands which include the probe gene. The position of this gene can then be correlated to the physical map of the mtDNA and localized relative to other probed genes. The sub- cloning necessary for determining the complete mtDNA sequence of Katharina and Mytilus provides the gene- specific probes necessary to this technique. One limitation of the Southern hybridization tech- nique is the requirement for a detailed cleavage map of the mtDNA. This is most effectively accomplished if the mtDNA can be recovered in pure form and in large quantity, which is often difficult and may be impossible for some small or rare organisms. The resolution of the gene map will be limited by the spacing of the restriction enzyme cleavage sites. The main advantages of Southern hybridization are that 1) the relative location of widely spaced genes can be determined, whereas PCR can only be applied to contiguous blocks of closely adjacent genes J. L. Boore and W. M. Brown, 1994 Drosophila CO3 Katharina A Mytilus Page 73 Col AGN SI oR | 2 0) S| 2 [co UA AGN UCN Lae EE Bs eS coe Vertebrates ANCE Echinoderms Figure 6. The arrangements of several mitochondrial genes promising for Southern hybridization analysis (see text). Gene abbre- viations are as in figure 2; genes are transcribed from left to right except those depicted below the main line to designate opposite orientation (references in text). The broken lines shown for Drosophila and vertebrate mtDNAs indicates a large, undepicted portion of the genome. and 2) no prior hypothesis of gene arrangement is nec- essary for Southern hybridization, in contrast to the case for PCR. Figure 6 shows the relative arrangement of four genes in the Mytilus, Katharina, and Drosophila mitochon- drial genomes that may be especially useful for inves- tigating molluscan relationships. In general, the gene arrangements of Katharina and Mytilus are difficult to relate. However, discounting tRNA genes, the arrange- ment of CO8, ND2, ND8, and CO1 differ only in that ND2 and ND3 have exchanged positions. The arrange- ment of these genes in Drosophila mtDNA is similar in that CO8 is near ND3 and ND2 is near CO1, but these two pairs of genes are separated by approximately 10 kb of DNA sequence. Each of these four genes is well con- served enough to expect that they could be detected in the mtDNA of other mollusks by hybridization to probes of Mytilus or Katharina mtDNA. Determining the ar- rangement of these genes in the mtDNAs of other mol- lusks might suggest whether the rearrangement that brought ND3 near to ND2 occurred near the base of the molluscan radiation. If so, it would provide a synapo- morphy suggesting the monophyly of Mollusca. It would also inform us about whether other classes of mollusks share with Mytilus the derived condition of inverting the positions of ND2 and ND§8, thus signaling a common ancestry with Bivalvia. Admittedly, most evolutionary relationships will not be-resolved by comparisons of mitochondrial gene ar- rangements. Genome rearrangements may not have oc- curred during the period of shared history, or subsequent rearrangements may have erased similarity. The main advantage of this data set is that relationships are very reliably inferred when accompanied by a genome re- arrangement. The complete sequences of the mtDNAs of Katharina and Mytilus facilitate even more rapid investigation of the patterns of mitochondrial genome rearrangments among mollusks. As more mtDNAs are investigated for gene arrangement and other complex molecular characteristics, higher level relationships among mollusks and among other taxa may be resolved. LITERATURE CITED Anderson, S., A. T. Bankier, B. G. Barrell, M. H. L. De Bruijn, A. R. Coulson, J. Drouin, I. C. Eperon, D. P. Nierlich, B. A. Roe, F. Sanger, P. H. Schreier, A. J. H. Smith, R. Staden, and I. G. Young. 1981. Sequence and organization of the human mitochondrial genome. Nature 290:457-465. Anderson, S., M. H. L. De Bruijn, A. R. Coulson, I. C. Eperon, F. Sanger and I. G. Young. 1982. Complete sequence of bovine mitochondrial dna: conserved features of the mam- malian mitochondrial genome. Journal of Molecular Bi- ology 156:683-717. Arnason, U., A. Gullberg and B. Widegren. 1991. The com- plete nucleotide sequence of the mitochondrial DNA of the fin whale, Balaenoptera physalus. Journal of Molecular Evolution 33(6):556-568. Arnason, U. and E. Johnsson. 1992. The complete mitochon- drial DNA sequence of the harbor seal, Phoca vitulina. Journal of Molecular Evolution 34:493-505. Arnheim, N. 1983. Concerted evolution of multigene families. In: M. Nei and R. Keohn (eds.). Evolution of Genes and Page 74 Proteins. Sinauer Associates, Inc., Sunderland, Massachu- setts, pp.38-61. Attardi, G. 1988. Biogenesis of mitochondria. Annual Review of Cell Biology 4:289-333. Barnes, R. D. 1987. Invertebrate Zoology, 5th edition. W.B. Saunders, Philadelphia, 893 pp. Batucas, B., R. Garresse, M. Calleja, J. R. Valverde and R. Marco. 1988. Genome organization of Artemia mito- chondrial DNA. Nucleic Acids Research 16:6515-6529. Bibb, M. J., R. A. Van Etten, C. T. Wright, M. W. Walberg and D. A. Clayton. 1981. Sequence and gene organi- zation of mouse mitochondrial DNA. Cell 26:167-180. Beklemishev, V. N. 1963. On the relationship of the Turbel- laria to other groups of the animal kingdom. In: E. C. Dougherty (ed.). The Lower Metazoa. University of Cal- ifornia Press, Berkeley, pp. 234-244. Boer, P. and M. Gray. 1991. Short dispersed repeats localized in spacer regions of Chlamydamonas reinhardtii mito- chondrial DNA. Current Genetics 19:309-312. Boettger, C.R. 1959. Comments to H. Lemche: Protostomian interrelationships in the light of Neopilina. Proceedings of the XVth International Congress of Zoology, London, pp.386-389. Bogenhagen, D. F., S. S. Cairns and B. K. Yoza. 1985. Nu- cleotide sequences involved in the control of transcription and displacement loop DNA synthesis for Xenopus laevis mtDNA. In: E. Quagliariello et al. (eds.). Achievements and Perspectives of Mitochondrial Research, Vol. II: Bi- ogenesis. Elsevier Science Publishers, pp. 175-182 Boore, J. L. and W. M. Brown. Complete DNA sequence of the mitochondrial genome of the black chiton, Katharina tunicata, submitted. Boore, J. L. and W. M. Brown. Complete DNA sequence of the mitochondrial genome of the annelid, Lumbricus ter- restris, manuscript in preparation. Bridge, D., C. W. Cunningham, B. Schierwater, R. Desalle and L. W. Buss. 1992. Class-level relationships in the phylum Cnidaria: evidence from mitochondrial genome structure. Proceedings of the National Academy of Sciences, USA 89:3750-8753. Brown, T., R. Waring, C. Scazzocchio and R. Davies. 1985. The Aspergillus nidulans mitochondrial genome. Current Genetics 9:113-117. Brown, W. M. 1985. The mitochondrial genome of animals. In: R. J. MacIntyre (ed.). Molecular Evolutionary Genetics. Plenum Press, New York, pp. 95-130. Brown, W. M., M. George and A. C. Wilson. 1979. Rapid evolution of animal mitochondrial DNA. Proceedings of the National Academy of Sciences, USA 76:1967. Brusca, R. C. and G. J. Brusca. 1990. Invertebrates. Sinauer, Sunderland, Massachusetts, 922 pp. Buroker, N., J. Brown, T. Gilbert, P. O'Hara, A. Beckenbach, W. Thomas and M. Smith. 1990. Length heteroplasmy of sturgeon mitochondrial DNA: An illegitimate elonga- tion model. Genetics 124:157-163. Cantatore, P., M. Roberti, G. Rainaldi, M. N. Gadaleta and C. Saconne. 1989. The complete nucleotide sequence, gene order and genetic code of the mitochondrial genome of Paracentrotus lividus. Journal of Biological Chemistry 264: 10965-10975. Carr, S. M., A. J. Brothers and A. C. Wilson. 1987. Evolu- tionary inferences from restriction maps of mitochondrial DNA from nine taxa of Xenopus frogs. Evolution 41:176. Chang, Y.-s , F.-l. Huang and T.-b. Lo. 1994. The complete nucleotide sequence and gene organization of carp (Cy- THE NAUTILUS, Supplement 2 prinus carpio) mitochondrial genome. Journal of Molec- ular Evolution 38:138-155. Clark-Walker, G. 1989. In vivo rearrangement of mitochon- drial DNA in Saccharomyces cerevisiae. Proceedings of the National Academy of Sciences, USA 86:8847-8851. Clary, D. O. and D. R. Wolstenholme. 1985a. The mito- chondrial DNA molecule of Drosophila yakuba: Nucleo- tide sequence, gene organization, and genetic code. Jour- nal of Molecular Evolution 22:252-271. Clary, D. O. and D. R. Wolstenholme. 1985b. The ribosomal RNA genes of Drosophila mitochondrial DNA. Nucleic Acids Research 13(11):4029-4044. Clayton, D. A. 1991. Replication and transcription of ver- tebrate mitochondrial DNA. Annual Review of Cell Bi- ology 7:453-478. Clayton, D. A. 1992. Transcription and replication of animal mitochondrial DNA. In: K. W. Jeon and D. R. Wolsten- holme (eds.). Mitochondrial Genomes. International Re- view of Cytology 141:217-232. Coen, E., T. Strachan and G. Dover. 1982. Dynamics of concerted evolution of ribosomal DNA and histone gene families in the melanogaster species subgroup of Dro- sophila. Journal of Molecular Biology 158:17-35. Crozier, R. H and Y. C. Crozier. 1993. The mitochondrial genome of the honeybee Apis mellifera: complete se- quence and genome organization. Genetics 133:97-117. Cummings, D. and J. Domenico. 1988. Sequence analysis of mitochondrial DNA from Podospora anserina. Journal of Molecular Biology 204:815-839. Cummings D., F. Michel, J. Domenico and K. McNally. 1990a. DNA sequence analysis of the mitochondrial ND4L-ND5 gene complex from Podospora anserina: Duplication of the ND4L gene within its intron. Journal of Molecular Biology 212(2):269-86. Cummings, D., K. McNally, J. Domenico and E. Matsuura. 1990b. The complete DNA sequence of the mitochondrial genome of Podospora anserina. Current Genetics 17:375- 402. Dawid, I. B. and A. W. Blackler. 1972. Maternal and cyto- plasmic inheritance of mitochondrial DNA in Xenopus. Developmental Biology 29:152-161. De Bruijn, M. H. L. 1983. Drosophila melanogaster mito- chondrial DNA, a novel organization and genetic code. Nature 304:234-241. De Giorgi, C., C. Lanave, M. D. Musci and C. Saccone. 1991. Mitochondrial DNA in the sea urchin Arbacia lixula: evo- lutionary inferences from nucleotide sequence analysis. Molecular Biology and Evolution 8:515-529. de Zamaroczy, M and G. Bernardi. 1986. The primary struc- ture of the mitochondrial genome of Saccharomyces cer- evisiae—a review. Gene 47:155-177. Desjardins, P., and R. Morais. 1990. Sequence and gene or- ganization of the chicken mitochondrial genome: a novel gene order in higher vertebrates. Journal of Molecular Biology 212:599-634. Desjardins, P., V. Ramirez and R. Morais. 1990. Gene orga- nization of the Peking duck mitochondrial genome. Cur- rent Genetetics 17:515-518. Dewey, R., C. Leving and D. Timothy. 1985. Nucleotide sequence of ATPase subunit 6 gene of maize mitochondria. Plant Physiology 79:914-919. Dixon, M. T. and D. M. Hillis. 1998. Ribosomal RNA sec- ondary structure: compensatory mutations and implica- tions for phylogenetic analysis. Molecular Biology and Evolution 10(1):256-267. ee eee J. L. Boore and W. M. Brown, 1994 Dubin, D. T., C. -C. Hsuchen and L. E. Tillotson. 1986. Mos- quito mitochondrial tRNAs for valine, glycine and glu- tamate: RNA and gene sequences and vacinal genome organization. Current Genetics 10:701—707. Dyson, N., T. Brown, J. Ray, R. Waring, C. Scazzocchio and R. Davies. 1989. Processing of mitochondrial RNA in Aspergillus nidulans. Journal of Molecular Biology 208: 587-599. Fauron, C. M. -R. and D. R. Wolstenholme. 1976. Structural heterogeneity of mitochondrial DNA molecules within the genus Drosophila. Proceedings of the National Academy of Sciences, USA 73:3623. Feagin, J., J. Abraham and.K. Stuart. 1988. Extensive editing of the cytochrome c oxidase III transcript in Trypanosoma brucei. Cell 53:413-422. Fearnley I. and J. Walker. 1987. Initiation codons in mam- malian mitochondria: differences in genetic code in the organelle. Biochemistry 26(25):8247-8251. Felsenstein, J. 1978. Cases in which parsimony and compat- ability methods may be positively misleading. Systematic Zoology 27:401-410. Field, K. G., G. J. Olsen, D. J. Lane, S. J. Giovannoni, M. T. Ghiselen, E. C. Raff, N. R. Pace and R. A. Raff. 1988. Molecular phylogeny of the animal kingdom. Science 239: 748-753. Fretter, V. and A. Graham. 1962. British prosobranch mol- luses. The Ray Society, London, 755 pp. Gadaleta, G., G. Pepe, G. De Candia, C. Quagliariello, E. Sbis and C. Saccone. 1989. The complete nucleotide sequence of the Rattus norvegicus mitochondrial genome: cryptic signals revealed by comparative analysis between verte- brates. Journal of Molecular Evolution 28:497-516. Garesse, R. 1988. Drosophila melanogaster mitochondrial DNA: Gene organization and evolutionary considerations. Genetics 118:649-663. Garey, J. R. and D. R. Wolstenholme. 1989. Platyhelminth mitochondrial DNA: Evidence for early evolutionary or- igin of a tRNA‘46%) that contains a dihydrouridine arm replacement loop and of serine-specifying AGA and AGG codons. Journal of Molecular Evolution 28:374—-387. Ghiselin, M. T. 1988. The origin of molluscs in the light of molecular evidence. Oxford Surveys in Evolutionary Bi- ology 5:66-95. Gjetvaj, B., D. I. Cook and E. Zouros. 1992. Repeated se- quences and large scale size variation of mitochondrial DNA: A common feature among scallops (Bivalvia: Pec- tinidae). Molecular Biology and Evolution 9:106-124. Giitting, K. -J. 1980. Origin and relationships of the Mollusca. Z. zool. Syst. Evolut. -forsch 18:24-27. Graham, A. 1955. Molluscan diets. Proceedings of the Mal- acological Society of London 31:144-159. Gray, M. and P. Boer. 1988. Organization and expression of algal (Chlmydamonas reinhardtii) mitochondrial DNA. Philosophical Transactions of the Royal Society of London 319:135-147. Gyllensten, U., D. Wharton and A. C. Wilson. 1985. Maternal inheritance of mitochondrial DNA during backcrossing of two species of mice. Journal of Heredity 76:321. Haas, W. 1981. Evolution of calcareous hardparts in primitive molluscs. Malacologia 21:403—418. Hadzi, J. 1953. Anattempt to reconstruct the system of animal classification. Systematic Zoology 2:145-154. Hadzi, J. 1963. The evolution of the Metazoa. Macmillan, New York, 499 pp. Page 75 Hammarsten, O. and J. Runnstriim. 1925. Zur Embryologie von Acanthochiton discrepans Brown. Zool. Jahrb., Abt. Anat. Ontog. Tiere 47:261-318. Harrison, R. G., D. M. Rand and W. C. Wheeler. 1985. Mi- tochondrial DNA size variation within individual crickets. Science 228:1446-1448. Harrison, R.G. 1989. Animal mitochondrial DNA asa genetic marker in population and evolutionary biology. Trends in Ecology and Evolution 4:6-11. Himeno, H., H. Masiki, T. Kawai, T. Ohta, I. Kumagai, K. Miura and K. Waranabe. 1987. Unusual genetic codes and a novel gene structure for tRNA*%"°”) in starfish mi- tochondrial DNA. Gene 56:219-230. Hoeh, W. R., K. H. Blakley and W. M. Brown. 1991. Het- eroplasmy suggests limited biparental inheritance of My- tilus mitochondrial DNA. Science 251:1488-1490. Hoffman, R. J., J. L. Boore and W. M. Brown. 1992. A novel mitochondrial genome organization for the blue mussel, Mytilus edulis. Genetics 131:397—412. Hsuchen, C. -C., R. M. Kotin and D. T. Dubin. 1984. Se- quences of the coding and flanking regions of the large ribosomal subunit RNA gene of mosquito mitochondria. Nucleic Acids Research 12:7771-7785. Ihering, H. von. 1876. Versuch eines nattirlichen Systemes der Mollusken. Jahrbucher der Deutschen malacozoolo- gischen Gesellschaft 3:97-148. Inglis, W.G. 1985. Evolutionary waves; patterns in the origins of animal phyla. Australian Journal of Zoology 33:153- 178. Innis, M., D. Gelfand, J. Sninsky and T. White (eds.). 1989. PCR Protocols: A Guide to Methods and Applications. Academic Press, Inc., San Diego, CA, 482 pp. Ise, W., H. Haiker and H. Weiss. 1985. Mitochondrial trans- lation of subunits of the rotenone-sensitive NADH-ubi- quinone reductase in Neurospora crassa. European Mo- lecular Biology Organization Journal 4:2075-2080. Jacobs, H. T., D. J. Elliott, V. B. Math and A. Farquarson. 1988. Nucleotide sequence and gene organization of sea urchin mitochondrial DNA. Journal of Molecular Biology 202:185-217. Johansen, H., P. H. Guddal and T. Johansen. 1990. Organi- zation of the mitochondrial genome of Atlantic cod, Gadus morhua. Nucleic Acids Research 18:411-419. Jukes, T. H. and S. Osawa. 1990. The genetic code in mito- chondria and chloroplasts. Experientia 46(11-12):1117- 1126. Jukes, T. H. and S. Osawa. 1993. Evolutionary changes in the genetic code. Comparative Biochemistry and Physiology 106B(8):489-494. King, T. C. and R. L. Low. 1987. Mapping of control elements in the displacement loop region of bovine mitochondrial DNA. Journal of Biological Chemistry 13:6204-6213. Knight, J. B. and E. L. Yochelson. 1960. Monoplacophora. In: R. C. Moore (ed.). Treatise on Invertebrate Paleontol- ogy, I, Mollusca. University of Kansas, Lawrence, pp. 177- 184. Kocher T., W. K. Thomas, A. Meyer, S. Edwards, S. Paabo, F. Villablanca and A. Wilson. 1989. Dynamics of mito- chondrial DNA evolution in animals: amplification and sequencing with conserved primers. Proceedings of the National Academy of Sciences, USA 86(16):6196-6200. Kraus, F., L. Jarecki, M. M. Miyamoto, S. M. Tanhauser and P. J. Laipis. 1992. Mispairing and compensational changes during the evolution of mitochondrial ribosomal RNA. Molecular Biology and Evolution 9(4):770-774. Page 76 Lake, J. 1988. Origin of the eukaryotic nucleus determined by rate-invariant analysis of rRNA sequences. Nature 331: 184-186. Lake, J. 1990. Origin of the Metazoa. Proceedings of the National Academy of Sciences, USA 87:763-766. Lang, B., F. Ahne, S. Distler, H. Trinkl, F. Kaudewitz and K. Wolf. 1983. Sequence of the mitochondrial DNA, ar- rangements of genes and processing of their transcripts in Schizosaccharomyces pombe. In: R.J. Schweyen, K. Wolf and F. Kaudewitz (eds.). Mitochondria 1983: Nucleo-Mi- tochondrial Interactions. de Gruyter, Berlin, pp. 3138-329. Lansman, R. A., J. C. Avise and M. D. Huettel. 1983. Critical experimental test of the possibility of “paternal leakage” of mitochondrial DNA. Proceedings of the National Acad- emy of Sciences, USA 80:1969-1971. LaRoche, J., M. Snyder, D. I. Cook, K. Fuller and E. Zouros. 1990. Molecular characterization of a repeat element causing large-scale size variation in the mitochondrial DNA of the sea scallop Placopecten magellanicus. Molecular Biology and Evolution 7:45-64. Lemche, H. 1959a. Molluscan phylogeny in the light of Neo- pilina. Proceedings of the XVth International Congress of Zoology, London, pp. 380-381. Lemche, H. 1959b. Protostomian interrelationships in the light of Neopilina. Proceedings of the XVth International Congress of Zoology, London, pp. 381-389. Lewin, B. 1987. Genes III. Wiley & Sons, Inc., New York, pp. 122-143. Lindberg, D. R. 1985. Aplacophoran, monoplacophorans, po- lyplacophorans, and scaphopods: the lesser classes. In: D. J. Bottjer, C. S. Hickman and P. D. Ward (eds.). Mollusks, Notes for a Short Course. University of Tennessee De- partment of Geological Sciences Studies in Geology no. 13, pp. 230-247. Maccracken, A., I. Uhlenbusch and G. Gellissen. 1987. Struc- ture of the cloned Locusta migratoria mitochondrial ge- nome: restriction mapping and sequence of its ND-1 (URF- 1) gene. Current Genetics 11:625-630. Marcus, E. 1958. On the evolution of the animal phyla. Quar- terly Review of Biology 33:24-58. Meglitsch, P. A. and F. R. Schram. 1991. Invertebrate Zo- ology, 3rd edition. Oxford University Press, New York. Monforte, A., E. Barrio and A. Latorre. 1993. Characteriza- tion of the length polymorphism in the A+T region of the Drosophila obscura group species. Journal of Molecular Evolution 36(8):214-223. Montoya, J., T. Christianson, D. Levens, M. Rabinowitz and G. Attardi. 1982. Identification of initiation sites for heavy- strand and light-strand transcription in human mitochon- drial DNA. Proceedings of the National Academy of Sci- ences, USA 79:7195-7199. Moritz, C., and W. M. Brown. 1986. Tandem duplications of D-loop and ribosomal RNA sequences in lizard mitochon- drial DNA. Science 233:1425-1427, Moritz, C. and W. M. Brown. 1987. Tandem duplications in animal mitochondrial DNAs: variation in incidence and gene content among lizards. Proceedings of the National Academy of Sciences, USA 84:7183-7187. Moritz, C., T. E. Dowling and W. M. Brown. 1987. Evolution of animal mitochondrial DNA: relevance for population biology and systematics. Annual Review of Ecology and Systematics 18:269-292. Nielsen, C. 1985. Animal phylogeny in the light of the tro- chaea theory. Biological Journal of the Linnaeus Society 25:243-299. THE NAUTILUS, Supplement 2 Nielsen, C. 1987. Structure and function of metazoan ciliary bands and their phylogenetic significance. Acta Zoologica 68:205-262. Oda, K., K. Yamato, E. Ohta, Y. Nakamura, M. Takemura, N. Nozato, K. Akashi, T. Kanegae, Y. Ogura, T. Kohchi and K. Ohyama. 1992. Gene organization deduced from the complete sequence of liverwort Marchantia polymorpha mitochondrial DNA. Journal of Molecular Biology 223: 1-7. Ojala, D., C. Merkel, R. Gelfand and G. Attardi. 1980. The tRNA genes punctuate the reading of genetic information in human mitochondrial DNA. Cell 22:393-408. Ojala, D., J. Montoya andG. Attardi. 1981. tRNA punctuation model of RNA processing in human mitochondria. Nature 290:470-474. Okimoto, R., H. M. Chamberlin, J. L. Macfarlane and D. R. Wolstenholme. 1991. Repeated sequence sets in mito- chondrial DNA molecules of root knot nematodes (Me- loidogyne): nucleotide sequences, genome location and potential for host race identification. Nucleic Acids Re- search 19:1619-1626. Okimoto, R., J. L. Macfarlane, D. O. Clary and D. R. Wol- stenholme. 1992. The mitochondrial genomes of two nematodes, Caenorhabditis elegans and Ascaris suum. Genetics 180(3):471-498. Paabo, S., W. K. Thomas, K. M. Whitfield, Y. Kumazawa and A. C. Wilson. 1991. Rearrangements of mitochondrial transfer RNA genes in marsupials. Journal of Molecular Evolution 33:426-430. Pace, N. R., G. J. Olsen and C. R. Woese. 1986. Ribosomal RNA phylogeny and the primary lines of evolutionary descent. Cell 45:325-326. Patterson, C. 1989. Phylogenetic relationships of major groups: conclusions and prospects. In: B. Fernholm, K.Bremer and H. Jrnvall (eds.). The Hierarchy of Life. Elsevier Science Publishers, B. V.(Biomedical Division), Amsterdam, pp. 471-488. Pearse, V., J. Pearse, M. Buchsbaum and R. Buchsbaum. 1987. Living invertebrates. Boxwood Press, Pacific Grove, Cal- ifornia. Pelseneer, P. 1899. Recherches morphologiques et phylo- genétiques sur les mollusques archaiques. Mémoires Cu- ronnées et Mémoires Savants trangers Academie Belgique 57:1-113. Pojeta, J. 1980. Molluscan Phylogeny. Tulane Studies in Ge- ology and Paleontology 16(2):55-80. Pratje, E., C. Vahrenholz, S. Buhler, and G. Michaelis. 1989. Mitochondrial DNA of Chlamydomonas reinhardtii: the ND4 gene encoding a subunit of NADH dehydrogenase. Current Genetics 16(1):61-64. Pritchard, A., J. Seilhamer, R. Mahalingam, C. Sable S. Venuti and D. Cummings. 1990. Nucleotide sequence of the mitochondrial genome of Paramecium. Nucleic Acids Re- search 18(1):173-180. Roe, B. A., D. -P. Ma, R. K. Wilson and J. F. -H. Wong. 1985. The complete nucleotide sequence of the Xenopus laevis mitochondrial genome. Journal of Biological Chemistry 260:9759-9774. Runnegar, B. and J. Pojeta. 1974. Molluscan phylogeny: the palaeontological viewpoint. Science 186:311-317. Salvini-Plawen, L. von. 1972. Zur Morphologie und Phylo- genie der Mollusken; die Beziehungen der Caudofoveata und der Solenogastres als Aculifera, als Mollusca, und als Spiralia. Zeitschrift fr wissenschaftliche Zoologie 184:205- 394. J. L. Boore and W. M. Brown, 1994 Salvini-Plawen, L. von. 1980. A reconsideration of systematics in the Mollusca (phylogeny and higher classification). Malacologia 19:249-278. Salvini-Plawen, L. von. 1985. Early evolution and the prim- itive groups. In E. R. Trueman and M. R. Clarke (eds.). The Mollusca, volume 10, Evolution. Academic Press, Inc., Orlando, pp. 59-150. Sankoff, D., R. Cedergren and Y. Abel. 1990. Genomic di- vergence through gene rearrangement. Methods in En- zymology 183:428—438. Sankoff, D., G. Leduc, N. Antoine, B. Paquin, B. F. Land and R. Cedergen. 1992. Gene order comparisons for phylo- genetic inference: evolution of the mitochondrial genome. Proceedings of the National Academy of Sciences, USA 89:6575-6579. Sasaki, T., T. Okazaki, M. Muramatsu and R. Kominami. 1987. Variation among mouse ribosomal RNA genes within and between chromosomes. Molecular Biology and Evolution 4(6):594-601. Scheltema, A. H. 1988. Ancestors and descendents: relation- ships of the Aplacophora and Polyplacophora. American Malacological Bulletin. 6(1):57-68. Simpson, L., N. Neckelmann, V. de la Cruz, A. Simpson, J. Feagin, D. Jasmer and K. Stuart. 1987. Comparison of the maxicircle (mitochondrial) genomes of Leishmania tarentolae and Trypanosoma brucei at the level of nucle- otide sequence. Journal of Biological Chemistry. 262(13): 6182-6196. Smith, A. B. 1989. RNA sequence data in phylogenetic re- construction: testing the limits of its resolution. Cladistics 5:321-344. Smith, M. J., D. K. Banfield, K. Doteval, S. Forski and D. J. Kowbel. 1989. Gene arrangement in sea star mitochon- drial DNA demonstrates a major inversion event during echinoderm evolution. Gene 76:181-185. Smith, M. J., D. K. Banfield, K. Doteval, S. Forski and D. J. Kowbel. 1990. Nucleotide sequence of nine protein-cod- ing genes and 22 tRNAs in the mitochondrial DNA of the sea star Pisaster ochraceus. Journal of Molecular Evolution 31:195-204. Smith, M. J., A. Arndt, S. Gorski and E. Fajber. 1993. The phylogeny of echinoderm classes based on mitochondrial gene arrangements. Journal of Molecular Evolution 36: 545-554. Snyder, M., A. R. Fraser, J. Laroche, K. E. Gartner-Kepkay and E. Zouros. 1987. Atypical mitochondrial DNA from the deep-sea scallop Placopecten magellanicus. Proceed- ings of the National Academy of Sciences, USA 84:7595- 7599. Sogin, M. L. 1991. Early evolution and the origin of eukary- otes. Current Opinion in Genetics and Development 1:457- 463. Solignac, M., M. Monnerot and J. -C. Mounolou. 1986. Con- certed evolution of sequence repeats in Drosophila mi- tochondrial DNA. Journal of Molecular Evolution 24:53- 60. Southern, E. M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. Journal of Molecular Biology 98:503-517. Stuart, K. and J. Feagin. 1992. Mitochondrial DNA of Ki- netoplastids. In: K. W. Jeon and D. R. Wolstenholme (eds. ). Mitochondrial Genomes. International Review of Cytology *141:65-88. Thomas, W. K., S. Paabo, F. Villablanca and A. Wilson. 1990. Spatial and temporal continuity of kangaroo rat popula- Page 77 tions shown by sequencing mitochondrial DNA from mu- seum specimens. Journal of Molecular Evolution $1(2): 101-112. Tzeng, C. -S., C. -F. Hui, S. -C. Shen and P. C. Huang. 1992. The complete nucleotide sequence of the Crossostome lacustre mitochondrial genome: conservation and varia- tions among vertebrates. Nucleic Acids Research 20(18): 4853-4858. Uhlenbusch, I., A. Mccracken and G. Gellissen. 1987. The gene for the large (16S) ribosomal RNA from the Locusta migratoria mitochondrial genome. Current Genetics 11: 621-638. Vagvolgyi, J. 1967. On the origin of molluscs, the coelom, and coelomic segmentation. Systematic Zoology 16:153- 168. Vahrenholz, C., E. Pratje, G. Michaelis and B. Dujon. 1985. Mitochondrial DNA of Chlamydomonas reinhardtii: se- quence and arrangement of URF5 and the gene for cy- tochrome oxidase subunit I. Molecular Genetics 201:213- 224. Vawter, L. and W. M. Brown. 1993. Rates and patterns of base change in the small subunit ribosomal RNA gene. Genetics 134:597-608. Wainright, P.O. 1993. Monophyletic origins of the metazoa: an evolutionary link with fungi. Science 260(5106):340- 342. Wallace, D. C. 1982. Structure and evolution of organelle genomes. Microbiological Reviews 46(2):208-240. Walsh, J. 1985. Interaction of selection and biased gene con- version in a multigene family. Proceedings of the National Academy of Sciences, USA 82:153-157. Warrior, R., and J. Gall. 1985. The mitochondrial DNA of Hydra attenuata and Hydra littoralis consists of two linear molecules. Archives Science Genéve 38:439-445. Wheeler, W. C. and R. L. Honeycutt. 1988. Paired sequence difference in ribosomal RNAs: evolutionary and phylo- genetic implications. Molecular Biology and Evolution 5:90-96. Wilkinson, G. and A. Chapman. 1991. Length and sequence variation in evening bat D-loop mtDNA. Genetics 128: 607-617. Wingstrand, K. G. 1985. On the anatomy and relationships of recent Monoplacophora. Galathea Report 16:7-94. Woese, C. R. and G. E. Fox. 1977. Phylogenetic structure of the prokarotic domain: the primary kingdoms. Proceed- ings of the National Academy of Sciences, USA 74(11): 5088-5090. Woese, C. R. 1987. Bacterial evolution. Microbiology Reviews 51(2):221-271. Wolf, K. and L. Del Guidice. 1988. The variable mitochon- drial genome of ascomycetes: organization, mutational al- terations, and expression. Advances in Genetics 25:185- 308. Wolstenholme, D. R., J. L. Macfarlane, R. Okimoto, D. O. Clary and J. A. Wahleithner. 1987. Bizarre tRNAs inferred from DNA sequences of mitochondrial genomes of nem- atode worms. Proceedings of the National Academy of Sciences, USA 84:1324-1328. Wolstenholme, D. R. 1992a. Genetic novelties in mitochon- drial genomes of multicellular animals. Current Opinion in Genetics and Development 2:918-925. Wolstenholme, D. R. 1992b. Animal mitochondrial DNA: structure and evolution. In: K. W. Jeon and D. R. Wol- stenholme (eds.). Mitochondrial Genomes. International Review of Cytology 141:173-216. Page 78 Zevering, C. E., C. Moritz, A. Heideman and R. A. Sturm. 1991. Parallel origins of duplications and the formation of pseudogenes in mitochondrial DNA from parthenoge- netic lizards (Heteronotia binoei: Gekkonidae). Journal of Molecular Evolution 33:431-441. THE NAUTILUS, Supplement 2 Zwieb, C., C. Glotz and R. Brimacombe. 1981. Secondary structure comparisons between small subunit ribosomal RNA molecules from six different species. Nucleic Acids Research 9(15):3621-3640. THE NAUTILUS, Supplement 2:79-84, 1994 Page 79 The Mitochondrial Genome of Cepaea nemoralis (Gastropoda: Stylommatophora): Gene Order, Base Composition, and Heteroplasmy Jonathan Terrett Sue Miles Department of Genetics Queens Medical Centre Clifton Boulevard Nottingham, NG7 2UH, U.K. Cromwell Road Richard H. Thomas Department of Zoology The Natural History Museum London, SW7 5BD, U.K. ABSTRACT The 14.1 kb mitochondrial genome of the terrestrial gastropod Cepaea nemoralis has been cloned and completely sequenced. Sequences coding for 13 proteins and the large and small sub- units of ribosomal RNA have been identified. All metazoan mtDNAs examined to date, except for the nematodes Caenor- habditis elegans and Ascaris suum, and the bivalve mollusc Mytilus edulis, contain an ATPase subunit 8 gene. The presence of ATPase 8 in Cepaea mitochondrial DNA suggests that this gene has been lost independently from the nematode and bi- valve mitochondrial genomes. Commonly genes are encoded on both strands of mtDNA molecules, and this is true of the Cepaea mitochondrial genome, but not of the nematodes and Mytilus. Base composition is the least biased of any reported metazoan mitochondrial genome. Comparisons are made with other metazoan mitochondrial genomes. The cloned genome has been used to infer the presence of polymorphisms in the size of mitochondrial genomes in Cepaea by analysis of the lengths of restriction fragments. These polymorphisms have aided the identification of heteroplasmic snails. A protocol is described that can be used to extract intact gastropod mito- chondrial DNA. This DNA is sufficiently free from inhibitors for use in restriction enzyme digestions and for amplification using the polymerase chain reaction. Key words: Mitochondrial DNA, Mollusca, molecular evolu- tion, heteroplasmy. INTRODUCTION The number of completely sequenced mitochondrial ge- nomes is increasing rapidly and the evenness of taxo- nomic sampling is improving. Sequences of seventeen metazoan mitochondrial genomes are now available from ten vertebrates [six placental mammals (Anderson et al., 1981; Bibb et al., 1981; Anderson et al., 1982; Gadaleta et al., 1989; Arnason et al , 1991; Arnason et al., 1992), two fish (Chang & Huang, 1991; Tzeng et al., 1992), a bird (Desjardins & Morais, 1990), and a frog (Roe et al., 1985)], two insects (Clary & Wolstenholme, 1985; Crozier & Crozier, 1993), two sea urchins (Cantatore et al., 1989; Jacobs et al., 1989), and three nematodes (Okimoto et al., 1992; Okimoto et al ., 1991). Many other taxa have been sequenced in part and of these the complete gene order is known for three, the mussel, Mytilus edulis (Hoffmann et al., 1992), a sea star (Smith et al., 1989) and the codfish, Gadus morhua (Johansen et al., 1990). Analyses of these genomes are revealing a host of inter- esting phenomena (reviewed in Wolstenholme, 1992), as well as providing tools for population and phylogenetic studies (Avise, 1989). Mitochondrial sequences from molluscs promise to greatly enhance our understanding both of the molecular evolution of metazoan mitochondria and the evolution- ary relationships of the molluscs themselves (Boore & Brown, 1994). In this paper we present the order of protein coding genes and ribosomal RNA (rRNA) genes in the mitochondrial genome of Cepaea nemoralis and compare it with those of other metazoans. Comparisons of DNA sequence across the phyla can be used to infer mechanisms of mtDNA replication and the mode of base substitution. We make preliminary observations on these points. We also present here a reliable protocol for the extraction of mtDNA from gastropods which is suitable for restriction fragment length polymorphism (RFLP) studies and for polymerase chain reaction (PCR) ampli- fication. Extraction of mitochondrial DNA from mol- luscs, and in particular terrestrial gastropods, has been problematic (Stine, 1989; J.S. Jones, personal commu- nication); the availability of a simple and reliable method opens up many possibilities for further work on these animals. MATERIALS AND METHODS mtDNA extractions: Several protocols for the extraction of mollusc mtDNA have already been published (Ski- binski & Edwards, 1987, Stine, 1989). The following Page 80 method combines procedures from both of these and can be performed in less than two hours. It has been used successfully and repeatedly on Cepaea and Helix (Ter- rett, 1992) and Littorina (E. Rumbak, personal com- munication). Snails are killed by placing at —80°C for 1 hour. They are then defrosted at 50°C for 5 minutes. After removing the snail from its shell, the hepatopan- creas is minced thoroughly with a scalpel blade and then homogenised in 3 ml of 0.25M sucrose in TEK (50mM Tris/HCl pH 7.5, 1OmM EDTA, 1.5% KCl (w/v)) in a Dounce homogenizer (7ml) with five to ten strokes of the “A’ pestle. The homogenate is transferred to two 1.5ml microfuge tubes. Nuclei and other cell debris are removed by centrifuging at 6,000 rpm (low speed) in a microfuge for two minutes. The supernatants are trans- ferred to two fresh microfuge tubes, and mitochondria collected by centrifugation at 13,000 rpm (high speed) in a microfuge for five minutes. The soft mitochondrial pellets are resuspended in 600 ul of 0.25M sucrose in TEK, and then 600 ul of 1.1M sucrose in TEK is layered beneath. Mitochondria are collected by centrifuging at high speed in a microfuge for 10 minutes. The mito- chondrial pellet is then resuspended in 600 pl of STE100 (100mM NaCl, 10mM Tris/HCl pH 8.0, 100mM EDTA) and intact mitochondria are lysed by the addition of 80 ul of 10% Nonidet P-40 followed by gentle shaking. Mitochondrial membranes are removed by centrifuga- tion at high speed in a microfuge for one minute. Mu- copolysaccharides are removed by the addition of 100 ul of 5M NaCl to the supernatant followed by 80 ul of CTAB/NaCl (0.7M NaCl / 10% CTAB). This mixture is shaken vigorously before incubating at 65°C for fifteen to thirty minutes. The precipitate that forms contains the mucopolysaccharides and is removed by filling the tube with chloroform, shaking vigorously to form an emulsion, and centrifuging at high speed for three min- utes in a microfuge. The mtDNA containing supernatant is removed and subjected to two phenol/chloroform ex- tractions and one chloroform extraction before the nu- cleic acids are precipitated. This is generally achieved by adding an equal volume of propan-2-ol (20°C) to the aqueous phase, shaking vigorously, and centrifuging im- mediately at high speed in a microfuge for ten minutes. The yield from the hepatopancreas of a single Cepaea is not sufficient for restriction digest fragments to reliably be seen on an agarose gel stained with ethidium bromide (EtBr). However, up to fifteen restriction digests can be visualised after Southern (1975) blotting and detection using digoxigenin (Boehringer Mannheim). Restriction enzyme digestions and cloning: C. nemor- alis mtDNA was extracted as above, electrophoresed on a 0.8% EtBr/agarose gel, and open circles were extracted using GenecleanII (Bio 101). This mtDNA was then di- gested with BamHI and cloned into AGem11 (Promega) using LE392 host. The entire genome was then subcloned into the SstI and HindIII sites of pGem7zf(+) (Promega) maintained in E. coli strain JM101. All clones were ver- ified as containing Cepaea mtDNA by probing against THE NAUTILUS, Supplement 2 mtDNA extractions. Preparation of plasmid, and pha- gemid DNAs were as in Sambrook et al. (1989). Sequencing and sequence analysis: DNA sequences were obtained using the dideoxy chain termination meth- od (Sanger et al., 1977) from sets of deletion subclones (Henikoff, 1984) and subsequent extraction of single stranded phagemid DNA (Vieira & Messing, 1987). Cepaea mitochondrial protein coding genes were identified by the similarity of their inferred amino acid sequences with those of Drosophila yakuba (Clary & Wolstenholme, 1985) and humans (Anderson et al., 1981), and in some cases verified by similarities in hydropathy profiles (Kyte & Doolittle, 1982). rRNA genes were iden- tified by the similarities of the DNA sequences of D. yakuba and Cepaea. Southern blotting and DNA hybrid detection: Standard Southern (1975) blotting techniques were used to transfer DNA onto nitrocellulose. The entire mitochondrial ge- nome of C. nemoralis within its \ vector was used as a probe after labelling with digoxigenin. Labelling and detection procedures were done according to the man- ufacturer’s instructions (Boehringer Mannheim). Blots were washed in 0.1 x SSC at 65°C (2 10 mm shell length) only the digestive gland was removed for DNA extraction, whereas for the smaller species the whole animal was processed. For DNA extraction a mod- ification of the guanidinium thiocynate method of Pitch- er et al. (1989) was used. The coarsely chopped animal was placed in equal volumes of Tris-EDTA (0.1M, pH 8.0) and GE reagent (5M guanidinium thiocynate, 1M EDTA), a half volume of glass beads (Glasperlen, Braun) added, and the sample shaken (2 min, 4°C, 2,000 rpm) in a bead beater (Mikro-Dismembrator, Braun). The beads and debris were removed by centrifugation (13,000 x g, 15 min, 4°C). The supernatant was collected and half a volume of ammonium acetate (7.5 M) added, before treatment with phenol and chloroform to remove pro- teins and pigments. The DNA in the sample was pre- cipitated with ethanol, washed with 70% ethanol, dried, and resuspended in sterile distilled water. The samples were then stored at 4°C, or for long-term storage, — 70°C. Polymerase Chain Reaction (PCR) and DNA sequenc- ing: Amplification of part of the mitochondrial small ribosomal RNA gene (12S rRNA) was carried out using the universal 12Sa and 12Sb primers (Kocher et al., 1989). THE NAUTILUS, Supplement 2 Each PCR was performed in a 100 ul volume consisting of 67 mM Tris-HCL (pH8.8), 2mM MegCl,, 0.05% Tween- 20, 100 ng/ml bovine serine albumin, 40 pmoles of each primer, 100 uM of each dNTP, 2.5 units Taq Polymerase (Perkin-Elmer/Cetus), and 10-1000 ng template DNA. The cycling parameters for amplification were an initial 5 min denaturation at 94°C, and then 30 cycles of 45 sec at 94°C, 1.5 min at 55°C, and 2 min at 72°C. Precipitated PCR products were then directly sequenced by the rapid thermal cycling technique described in Embley (1991). DNA sequence was determined for both strands. The DNA sequences have GenBank accession numbers u@5862-u@5874-v. Sequence analysis and gene phylogeny reconstruction: Sequences were aligned with the aid of the program CLUSTALYV (Higgins et al., 1992), with minor adjust- ments made by eye. Results from a range of analytical methods were compared (review by Swofford and Olsen 1990). Analyses employing maximum parsimony criteria were carried out with PAUP, version 3.0s (Swofford, 1990). Bootstrapped distance-based analyses were done with CLUSTALYV (Higgins et al., 1992), which employs Kimura's (1980) 2-parameter model for the calculation of corrected pairwise distances and the neighbour-joining algorithm of Saitou and Nei (1987) for the construction of trees. The maximum likelihood method (Felsenstein, 1981) was carried out with PHYLIP, version 3.4 (Fel- senstein, 1991). For further details of methodology con- sult legends to figures 2 and 3. RESULTS Sequences of the 12S rRNA gene fragment (not including the primer sequence) from eleven Littorina species and two Nodilittorina species are shown aligned in Figure 1. No within-species variation was found for any of the species used (for sample size see Table 1) except for L. saxatilis. Both the South African and Welsh L. saxatilis populations had two non-identical single base pair dif- ferences, as well as a sequence in common with Isle of Wight and Venice populations. Only the sequence in common is presented here (Figure 1). Alignment of the sequences was not difficult because this gene was found to be fairly well conserved. There are 73 variable posi- tions (including deletions and insertions) among the 374 sites aligned. Phenetic analysis of the sequence data for each pair of taxa is presented in Table 2. The total number of transversions and transitions, as well as the number of transversions relative to transitions, increase from left to right in the table (i.e. L. saxatilis to N. trochoides). Like- wise, the genetic distance (calculated from Kimura’s (1980) 2-parameter estimator, which allows transitions and transversions to occur at different rates) increases from top to bottom in the species matrix. Since the order of the species in Table 2 is based on the morphological cladogram of Reid (1990a), these trends reflect the in- creasingly distant relationships between species. These data show the expected transition bias (Brown et al., E. Rumbak et al., 1994 Page 93 N. radiata TCTTAGGC-A TAAATAAATT TAAATATTTA CTAGAGTACT ACGAATAAAA 50 L. saxatilis N. trochoides MGos= a ACenosoonoe oSo0e TA.A. 50 L. striata {So des C= Sadne esasanusee sodas TT.T. 50 de arcane L. keenae CG. s==ed€os ec@eocesoen Seance Wer 50 15%) 1 L. nigrolineata L. scutulata 6 oooscesess ds==cosde sscescoese so00- Cossm SH) 0.01 Units Tevarine L. plena > oassogooco dGs=Sso0ne aacsccscaa soegosuooS 50 : ‘i L. littorea ol soooned4nouo slls=seba6e stoodeoado sooocgdods 50 L. obtusata L. subrotundata o gocdarso0o one==s0G60 Bsadcn9tsn Hsaobaooade 50 WO = oan cogs=o «oso obuoeey soe ==ceec0 soeencdog0 tooasco0dE 50 Epubrotundats L. mariae mh ocsodssenn coe==Hoodn Saocob0ddd soomgaDOdn 50 L. littorea I GOGH ocoacasdGo oceodeoees soe==ao0es Baooadoa0G0 asodcodeDD 50 L. arcana 6 O0nodo0000 cocsSpo0uc0 vodscroosy ogaboozcae 50 ei L. saxatilis 0 aen00n0009 ooeFHS0006 ancccocods congbsoncos 50 L. scutulata N. radiata CTATTTAAAA CTCAAAGAGC TTGGCGGTGC TITAGACTTC TCAGGGGAAC 100 IN, GuTGIEES = ©€Escancc0s ooeooeenes sogogdeace Woscoacad © cosooseace 100 lsc = = = =BEoocenodo ¢o000600005 aa00d0090 Wococgcas ® cooocoon0d 100 WNC ~=ss0000m0 doocodgnae! vosn0Gb009 Wosocsesne enanxtoogae 100 IL GGHGNEH! = =a 0c000070 annoacdned poasuobass Eoscoavcds os INa 5100.00 100 Innis e=oev0GQ0 voddoDDb0d oopOAOnoOoS BooOtRoDGK ssao0nGbocG 100 L. littorea S=co0g0bUO GOoGdOD00O FoHDdD0Dge oCosoDDODOO HasoDoOnDD 100 IL, GAROUCAG cP sa0g0000 coaneoccdo vovoeeogod seopecegds sau00009000 100 I QU1G3%) = p= aocagenny da0000gg0e BoevecooODD BoCCDBHDOD SHObsogDOD 100 ITE = = op onan0g00 gc0nGbeags FEDGoESedS SododeeDeD Soe pDodDOND 100 Wn OI s= 55000000 G00cG0R000 GocHabdo0S booaDd5GDG so0gG000DD 100 IL CHER = Bc onoand 9000060500 caba2DG0059 s00esGe005 S000000550 100 ik, SEIS = =p ogcG0Go aooceosonO BU0dGE5009 Ssdogbeu0c0 sO00000000 100 N. radiata CTGTCTCGTA ATCGACAGTC CACGAATCAA CCTTACCTTC TTTCGCG-AT 150 N. trochoides ....... Gag coocosd Gee ovoccoge Ws sooSo SWS Cs MGS; , iS0) L. striata L. keenae L. scutulata L. plena L. littorea L. subrotundata . L obtusata L. mariae L. nigrolineata L arcana L. saxatilis N. radiata CAGTATGTAT ACCGTCGTCG TCAGGTAACT TTTAAAAATA TAGAAGTTAG 200 N. trochoides ely L, striata L. keenae L. scutulata L. plena L. littorea L. subrotundata L obtusata L. mariae L, nigrolineata L, arcana L. saxatilis N. radiata N. trochoides L. striata L. keenae L. scutulata L. plena L. littorea L. subrotundata . L obtusata L. mariae L. nigrolineata L arcana L. saxatilis N. radiata N. trochoides L. striata L. keenae L. scutulata L, plena L. littorea L. subrotundata L. obtusata L, mariae L. nigrolineata L. arcana L. saxatilis N. radiata A-ATGACTTA TATGAAGGCG GACTTGAAAG TATGAATTAG TATAGAAATT 350 N, trochoides TGGTTT: -A. sso 9 AEG og ott SIEVO) L. striata €.GTG....A 350 L. keenae SEs woo SIsX0) L. scutulata -ACT....A 350 L. plena ogddCoanao 350 L, littorea sSShoo00ds «aotasodans “souceeouds otoamdoGoo) lOO ODL OEIC 350 IL GALE! o> yaase0g G0FGe0K00d “SoosoonoDS OnoUEOHODe gQoOdoDDODOK 350 L. obtusata Croaltenooa 350 L. mariae (C5 adtse 350 AGOGIEE o=-cocnbo vo=bo00G00 —vadobuaood cdaecdoGns0 DoonoduUdS 350 L arcana soceapaepos ca) L. saxatilis poodG5owo.0 350 N. radiata AGCTCTGAAG ACTG 374 N. trochoides CG... 374 L. striata 5605 374 L. keenae 374 L. scutulata 374 L. plena 374 L. littorea 374 L. subrotundata ... 374 L obtusata 374 L. mariae 374 L. nigrolineata 374 L. arcana 374 L. saxatilis 374 L. keenae L. striata N. radiata N. trochoides Figure 2. Bootstrapped neighbour-joining tree constructed with pairwise distances calculated using Kimura's (1980) 2-parameter correction for multiple substitutions. Branches are drawn pro- portional to their length. Produced with CLUSTALV (Higgins et al., 1992). Sites at which any species has a gap have been included in the analysis. Numbers on nodes are the percentage of replicates in which the taxa to the right of the node occur together. This is a bootstrap 50% majority-rule consensus tree based on 1000 replicates. The tree is shown with the Nodilit- torina species as an outgroup to root the tree. Arbitrarily groups with less than 50% support have been collapsed to polytomies. 1982; Moritz et al., 1987) seen in other metazoan mi- tochondrial sequences, and judging from the relatively low ratio of transversions to transitions these sequences are not yet saturated, even for the deepest divergences. In all pairwise comparisons the number of transitions exceed the number of transversions (Table 2). In general, once transitions have become saturated, the transversions continue to accumulate approximately linearly with time well past the point at which transitions have reached their asymptotic value (Miyamoto & Boyle, 1989). Phylogenetic trees constructed using neighbour-join- ing and maximum parsimony are shown in Figures 2 and 3 respectively. These trees are consistent with each other, though the neighbour-joining tree shows slightly more resolution of L. scutulata and of the L. saxatilis complex. The maximum likelihood method tree (not shown) had the same topology as the most parsimonious tree. In the parsimony analysis a single most parsimo- nious tree was obtained of length 185 steps. By permitting the tree length to increase by 1 step, and taking a strict consensus of the resulting trees, it was found that only four nodes remained in the ingroup (supporting L. stria- ta, L. keenae, the remaining species with the L. saxatilis complex). At 187 steps the L. saxatilis complex collapsed; the next branch to collapse was that separating L. keenae and L. striata at 190 steps. (a Figure 1. Aligned DNA sequences of a fragment of the mi- tochondrial 12S rRNA gene from 13 species of littorinid gas- tropods. Dots indicate that the sequence is identical to N. ra- diata, dashes indicate deletions in one or more sequences relative to other sequences shown. Page 94 THE NAUTILUS, Supplement 2 Table 1. Localities and sample sizes of Nodilittorina and Littorina used in this study. Species N. trochoides (Gray, 1839) N. radiata (Eydoux & Souleyet, 1852) L. striata King & Broderip, 1832 L. keenae Rosewater, 1978 L. scutulata Gould, 1849 L. plena Gould, 1849 L. littorea (Linnaeus, 1758) . subrotundata (Carpenter, 1864) . obtusata (Linnaeus, 1758) . mariae Sacchi & Rastelli, 1966 . nigrolineata Gray, 1839 . arcana Hannaford Ellis, 1978 . saxatilis (Olivi, 1792) Se iS) Si ie DISCUSSION The branches nearer the base of both trees (Figures 2 and 3) are well resolved and are consistent with Reid’s (1989, 1990a) concept of the monophyletic genus Lit- torina with L. striata as its basal member. Strictly, our data provide only a partial test of the inclusion of L. striata in Littorina; clearly it does not cluster between the two Nodilittorina species, but the hypothesis that it could be a basal member of Nodilittorina cannot be falsified without recourse to more distant outgroups. However, it may be noted that the calculated genetic distances (Table 2) show that in 6 of the 10 possible comparisons with other Littorina species, L. striata is closer to the Littorina than to either of the Nodilittorina species. This is reflected in the branch lengths of Figure 2. In addition, midpoint rooting of the maximum par- simony tree placed the tree root between the Nodilit- torina species and L. striata. Previously there has been some debate about the classification of L. striata, with Cape d’Aguilar, Hong Kong Cape d’Aguilar, Hong Kong El Golfo, Lanzarote, Canary Is. Pacific Grove, California, USA Pacific Grove, California, USA Candlestick Park, San Francisco, California, USA St. Lawrence, Isle of Wight, UK West Angle Bay, nr Pembroke, Wales, UK Charleston, Oregon, USA Pembroke Dock, Wales, UK West Angle Bay, nr Pembroke, Wales, UK West Angle Bay, nr Pembroke, Wales, UK St. Govan’s Head, nr Pembroke, Wales, UK St. Govan’s Head, nr Pembroke, Wales, UK Alberoni, Venice, Italy Langebaan Lagoon, South Africa St. Lawrence, Isle of Wight, UK Locality Sample size NPNNWHOWHNWHNWNHWHNWADWAWA KNW most recent authors placing it in Nodilittorina on mor- phological grounds (Rosewater, 1981; Bandel & Kadol- sky, 1982). However, in their analysis of allozyme data Backeljau & Warmoes (1992) also favoured inclusion in Littorina. The terminal groupings of L. saxatilis, L. ar- cana and L. nigrolineata (the L. saxatilis complex) and L. obtusata and L. mariae (the L. obtusata complex) are also supported, agreeing with the results of both mor- phological (Reid, 1989; 1990a) and allozyme studies (Warmoes, 1986; Ward, 1990; Knight & Ward, 1991; Zaslavskaya et al., 1992). At the more intermediate levels, however, neither tree is well resolved. The maximum parsimony tree (Figure 3) is entirely consistent with the morphological clado- gram of Reid (1990a; Figure 4), but does not resolve relationships between the members of the subgenera Lit- torina (L. scutulata, L. plena, L. littorea) and Neritrema (L. saxatilis complex, L. obtusata complex, L. subro- tundata). The neighbour-joining tree shows slightly more Table 2. The number of transversion differences followed by the number of transition differences (above the diagonal) for the 374 sites of the 12S rRNA gene fragment, and the pairwise genetic distance calculated from Kimura's (1980) 2-parameter estimator (below the diagonal). The first three letters of species names are used as abbreviations above columns. sax arc nig mar obt L. saxatilis — 1/0 2/1 1/5 1/5 L. arcana 0.003 — 1/1 0/6 0/6 L. nigrolineata 0.008 0.006 — 1/7 1/7 L. mariae 0.017 0.017 0.022 — 0/2 L. obtusata 0.017 0.017 0.022 0.006 — L. subrotundata 0.022 0.020 0.025 0.025 0.025 L. littorea 0.031 0.028 0.028 0.0384 0.034 L. plena 0.037 0.034 0.040 0.028 0.034 L. scutulata 0.052 0.052 0.058 0.052 0.052 L. keenae 0.108 0.108 0.102 0.098 0.105 L. striata 0.177 0.180 0.180 0.184 0.176 N. radiata 0.185 0.189 0188 0.181 0.189 N. trochoides 0.219 0.219 0.219 0.207 0.207 sub 2/6 1/6 PPT 1/8 1/8 0.031 0.037 0.055 0.111 0.177 0.174 0.212 lit ple scu kee str rad tro 3/8 8/10 2/16 11/25 14/41 25/84 27/41 2/8 2/10 1/17 10/26 15/41 26/34 26/42 3/7- 8/11 2/18 9/25 16/40 27/383 27/41 2/10 2/8 1/17 10/23 15/24 26/32 26/39 2/10 2/10 1/17 10/25 15/40 26/34 26/89 3/8 8/10 2/17 11/26 14/41 25/81 25/41 — 4/12 3/21 11/26 16/40 28/32 28/40 0.046 — 1/19 12/28 17/41 28/32 28/89 0.070 0.058 — 11/29 16/42 27/36 27/438 0.112 0121 0121 — 20/41 28/30 24/85 0.180 0.187 0.187 0.195 — 25/33 22/48 0.188 0.188 0.200 0180 0.183 10/23 0.219 0.215 0.227 0.184 0.096 = E. Rumbak et al., 1994 L. saxatilis L. arcana L. nigrolineata L. obtusata L. mariae L. subrotundata L. littorea L. plena L. scutulata L. keenae L. striata N. radiata N. trochoides Figure 3. Bootstrapped maximum parsimony tree based on analysis of 12S rRNA gene sequences produced with PAUP (Swofford, 1990) using a branch-and-bound search. Options used were furthest’ addition sequence; ACCTRAN; all minimal length trees saved (MULPARS); zero-length branches collapsed; gaps treated as missing (treatment of gaps as a fifth character has no effect on the topology and has only a slight effect on bootstrap values). Numbers on branches are the number of inferred synapomorphies. The numbers on the nodes are the percentage of replicates in which the taxa originating from there occur together. Bootstrap parameters are as described for Figure 2. The tree is shown with the Nodilittorina species as an outgroup to root the tree. There are 158 character states distributed over 57 informative characters. The tree is 185 steps long. resolution, as a result of the contribution of autapomor- phies to the pairwise distances among taxa for which there are few or no sites that are informative for parsi- mony analysis. This tree (Figure 2) separates L. scutulata in the same position as in the morphological cladogram, but does not place L. plena as its expected sister-species (Reid, 1990a). However, in view of the relatively low bootstrap values at the node separating L. scutulata, little confidence can be placed in this result. Disagreements as to the position of L. scutulata and its probable sister- species (Boulding, 1990; Reid, 1990a) are therefore not yet resolved. Other disagreements between the morpho- logical cladogram and allozyme-based trees (Zaslavskaya et al., 1992) involve species from the northwestern Pacific that have yet to be included in our study. The morphological cladogram of Reid (1990a) has been used as evidence of the origin of the Atlantic species of Neritrema from a Pacific ancestor, an example of the migration of marine fauna from the northern Pacific into the northern Atlantic following the opening of the Bering Strait during the Upper Pliocene (Reid, 1990b). Although the present results do not resolve the relationships of Neritrema, there is some support for this biogeographic hypothesis, in that the Atlantic Neritrema species show the lowest genetic distances among themselves (Table 2, Figure 2). The distance measures given in Table 2 show a close relationship among species classified in the subgenera Page 95 S = Sy Sas = iS aS a Se QD S = = Sos S 5s 5 5 SOs ete se eee See: r S$ 8 MSS Sos FS BS Ss 8 SS ss & = SSS SSeS SSUES QE SESS SS SSS BS ESSESESE SBE SESE SEHE SB & RF oS & & ke VD Oy SS Sy SS ey 8 7) ts = : SS SoS SoS Ss 2 sO Ss oS th OS oN & as o 8 8 Fo FG 8 SFTEB RAK GEE De RR Sn Ee ee, Oe CRMC pO Oe ER? EGS Meters (3) Sanita se SASS Asa es & ee Figure 4. Morphological cladogram of 16 Littorina species, redrawn from Reid (1990a). The following modifications were made: exclusion of Mainwaringia rhizophila (now not consid- ered a member of this clade, unpublished data); exclusion of L. kurila (a synonym of L. subrotundata) and L. neglecta (a probable synonym of L. saxatilis). Littorina and Neritrema, which is surprising in view of their rather marked morphological differences (Reid, 1989; 1990a). Both L. striata and L. keenae are relatively distant from these and from each other, supporting their classification in separate subgenera (Reid, 1989). The two species of Nodilittorina used as the outgroup in this analysis, N. radiata and N. trochoides, are also relatively distant from each other; their shells are dissimilar, but anatomically they show few differences and are presently placed in the same subgenus, Nodilittorina. There has not yet been a species-level analysis of phylogenetic re- lationships in Nodilittorina. The inference models used in this study to analyse the sequence data do not assume a molecular clock’ (Swof- ford & Olsen, 1990), and it is uncertain if the same topology would result if a clock were assumed. If, how- ever, the ‘molecular clock’ hypothesis, that genetic dis- tance measures are proportional to the time of phylo- genetic divergence, is tentatively assumed, one could make the following crude estimates. For Littorina, two nodes on the cladogram can be dated: the separation of L. littorea from its Pacific sister-species L. squalida, and the separation of the Atlantic Neritrema species from the sister-taxon of this clade in the Pacific. The earliest possible date for both of these separations is the opening of the Bering Strait, dated at 3.5-4.0 million years (My) Page 96 ago (Hopkins, 1967; Reid, 1990b; Vermeij, 1991). The latest date is taken as the time of the onset of widespread glaciation about 2.4 My (Shackleton et al., 1984; Loub- ere, 1988) when the trans-Arctic migration route may have been cut by climatic cooling. Littorina squalida was not included in this study, but the sister-taxon of the Atlantic Neritrema species is believed to be L. subro- tundata (Reid, 1990a). The average distance between L. subrotundata and the five Atlantic Neritrema species is 0.023 (Table 2), a divergence of 0.00575 per My (ac- cepting the older estimate of the age of the Bering Strait). This approximation permits tenuous estimates of the ages of other nodes on the tree: 8 My for the separation of L. plena plus L. scutulata (assuming these are sister-taxa) from more recent species; 19 My for L. keenae; 32 My for L. striata and therefore a minimum age of the clade as a whole. These ages are probably too young and would have become even more so had the latest separation date been used in the calibration. The fossil record of Litto- rina is very poor, and the older fossils, dating to the Upper Palaeocene, cannot be assigned to the genus with any confidence (Reid, 1989; 1990b). The oldest certain member of the genus is L. sookensis from the lower Miocene of Vancouver Island (approximately 22 My), which is probably close to the modern L. keenae. The Recent species L. squalida (sister-species of L. littorea) has been recorded from the Middle Miocene of Kam- chatka (approximately 15 My). It should be stressed that this means of estimation of ages from molecular data gives only very approximate results. This study has shown that the 12S rRNA gene is in- sufficiently variable to resolve the relationships among recently diverged species such as members of the sub- genera Littorina and Neritrema. However, it has per- mitted well-supported resolution of the deeper branches of the phylogeny. This is part of an ongoing study and we now intend to sequence more variable portions of the mitochondrial genome to resolve relationships more ful- ly, including those at the intraspecific level. We propose to examine as many as possible of the 21 species presently classified as Littorina, in the attempt to derive a well- resolved phylogeny for the genus. This will provide an evolutionary framework for the large amount of ecolog- ical and physiological data already available for the ge- nus, and thus permit tests of adaptational hypotheses (Reid, 1989; 1990a). It will also test hypotheses about speciation and biogeography of marine invertebrates in temperate latitudes for which this well-studied genus has been used as an example (Golikov & Tzvetkova, 1972; Reid, 1990b; Vermeij, 1991). ACKNOWLEDGEMENTS This work was supported by NERC grant GR3/7854. We would like to thank Stephen A. Ridgway and James T. Carlton for providing some of the samples used. As- sistance in the preparation of figures was provided by Gail Altschuler. THE NAUTILUS, Supplement 2 LITERATURE CITED Avise, J. C. 1989. Gene trees and organismal histories: a phy- logenetic approach to population biology. Evolution 43: 1192-1208. Avise, J. C., J. Arnold, R. Martin Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. A. Reeb and N. C. Saunders. 1987. In- traspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annual Re- view of Ecology and Systematics 18:489-522. Backeljau, T. and T. Warmoes. 1992. The phylogenetic re- lationships of ten Atlantic littorinids assessed by allozyme electrophoresis. In: Grahame, J., P. J. Mill and D. G. Reid (eds.). Proceedings of the third International Symposium on Littorinid Biology pp.9-24. London: Malacological So- ciety of London. Bandel, K. and D. Kadolsky. 1982. Western Atlantic species of Nodilittorina (Gastropoda: Prosobranchia): compara- tive morphology and its functional, ecological, phyloge- netic and taxanomic implications. Veliger 25:1—42. Boulding, E. G. 1990. Systematics, ecology and ecological genetics of some northern Pacific Littorina. PhD thesis, University of Washington, Seattle. Brown, W. M., E. M. Prager, A. Wang and A. C. Wilson. 1982. Mitochondrial DNA sequences of primates: tempo and mode of evolution. Journal of Molecular Evolution 18:225- 239. Emberton, K. C., G. S. Kuncio, G. M. Davies, S. M. Phillips, K. M. Monderewicz, and H. G. Yuan. 1990. Comparison of recent classifications of stylommatophoran land-snail families, and evaluation of large-ribosomal-RNA sequenc- ing for their phylogenetics. Malacologia 31:327-352. Embley, T. M. 1991. The linear PCR reaction: a simple and robust method for sequencing amplified rRNA genes. Let- ters in Applied Microbiology 13:171-174. Felsenstein, J. 1981. Evolutionary trees from DNA Sequences: A Maximum Likelihood Approach. Journal of Molecular Evolution 17:368-376. Felsenstein, J. 1991. PHYLIP (phylogeny inference package) version 3.4. Distributed by the author. Department of Ge- netics, University of Washington, Seattle. Golikov, A. N., and N. L. Tzvetkova. 1972. The ecological principle of evolutionary reconstruction as illustrated by marine animals. Marine Biology 14:1-9. Higgins, D. G., A. J. Bleasby and R. Fuchs. 1992. CLUSTAL- V-Improved software for multiple sequence alignment. Computer Applications in the Biosciences 8:189-191. Hillis. D. M. and M. T. Dixon. 1991. Ribosomal DNA: mo- lecular evolution and phylogenetic inference. The Quar- terly Review of Biology 66:411-453. Hopkins, D. M. 1967. The Cenozoic history of Beringia- a synthesis. In: Hopkins, D. M. (ed.) The Bering Land Bridge : pp. 451-484. Stanford, California; Stanford University Press. Kimura, M. 1980. A simple method for estimating evolution- ary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16:111-120. Knight, A. J.and R. D. Ward. 1991. The genetic relationships of three taxa in the Littorina saxatilis species complex (Prosobranchia: Littorinidae). Journal of Molluscan Studies 07:81-91. Kocher, T. D., W. K. Thomas, A. Meyer, S. V. Edwards, S. Paabo, F. X. Villablanca and A. C. Wilson. 1989. Dy- namics of mitochondrial DNA evolution in animals: am- E. Rumbak et al., 1994 plification and sequencing with conserved primers. Pro- ceedings of the National Academy of Sciences, USA 86: 6196-6200. Loubere, P. 1988. Gradual late Pliocene onset of glaciation; a deep-sea record from the northeast Atlantic. Palaeo- geography, Palaeoclimatology and Palaeoecology 63:327- 334. Miyamoto, M. M., and S. M. Boyle. 1989. The potential im- portance of mitochondrial DNA sequence data to euther- ian mammal phylogeny. In: B. Fernholm, K. Bremer and H. Jornvall (eds). The hierarchy of life. Elsevier Science, Amsterdam. Moritz, C., T. E. Dowling and W. M. Brown. 1987. Evolution of animal mitochondrial DNA: relevance for population biology and systematics. Annual Review of Ecology and Systematics 18:269-292. Morris, S. R. 1979. Genetic variation in the genus Littorina. Unpublished thesis, University of Wales. Pitcher, D., N. A. Saunders and R. J. Owen. 1989. Rapid extraction of bacterial genomic DNA with guanidium thio- cyanate. Letters in Applied Microbiology 8:151-156. Reid, D.G. 1986. The Littorinid molluscs of Mangrove forests in the Indo-Pacific region. The genus Littoraria. British Museum (Natural History), London, 228p. Reid, D. G. 1989. The comparative morphology, phylogeny and evolution of the gastropod family Littorinidae. Philo- sophical Transactions of the Royal Society of London, Se- ries B 324:1-110. Reid, D. G. 1990a. A cladistic phylogeny of the genus Lit- torina (Gastropoda): implications for evolution of repro- ductive strategies and for classification. Hydrobiologia 193: 1-19. Reid, D. G. 1990b. Trans-Arctic migration and speciation induced by climate change: the biogeography of Littorina (Mollusca: Gastropoda). Bulletin of Marine Science 47:35- 49. Rosewater, J. 1970. The family Littorinidae in the Indo-Pa- cific. I. The subfamily Littorininae. Indo-Pacific Mollusca 2:417-506. Page 97 Rosewater, J. 1981. The family Littorinidae in tropical West Africa. Atlantide Report, Copenhagen. 13:7-48. Saitou, N. and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mo- lecular Biology and Evolution 4:406-425. Shackleton, N. J., J. Backman, H. Zimmerman, D. V. Kent, M. A. Hall, D. G. Roberts, D. Schnitker, J. G. Baldauf, A. Desprairies, R. Homrighausen, P. Huddlestun, J. B. Keene, A. J. Kaltenback, K. A. O. Krumsiek, A. C. Morton, J. W. Murray and J. Westberg-Smith. 1984. Oxygen isotope calibration of the onset of ice-rafting and history of gla- ciation in the North Atlantic region. Nature 307:620-623. Simon, C. 1991. Molecular systematics at the species bound- ary: exploiting conserved and variable regions of the mi- tochondrial genome of animals via direct sequencing from amplified DNA. In: Hewitt, G. M., A. W. B. Johnston and J. P. W. Young (eds). Molecular Techniques in Taxonomy. Springer, Berlin. p.33-71. Swofford, D. L. 1990. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.0, Illinois Natural History Survey, Champaign, Illinois. Swofford, D. L. and G. J. Olsen. 1990. Phylogeny reconstruc- tion. In: Hillis, D. M. and C. Moritz (eds). Molecular Sys- tematics. Sinauer, Sunderland, Mass. p.411-501. Vermeij, G. J. 1991. Anatomy of an invasion: the trans-Arctic interchange. Paleobiology 17:281-307. Ward, R. D. 1990. Biochemical genetic variation in the genus Littorina. Hydrobiologia 193: 53-69. Warmoes, T. 1986. Een inleidende systematische en taxon- omische studie van het genus Littorina (Gastropoda: Pros- obranchia). Licentiaatsthesis, Universitaire Instelling Ant- werpen, p.128 Zaslavskaya, N. I., S. O. Sergievsky and A. N. Tatarenkov. 1992. Allozyme similarity of Atlantic and Pacific species of Littorina (Gastropoda: Littorinidae). Journal of Mol- luscan Studies 58:377-84. THE NAUTILUS, Supplement 2:98-110, 1994 Page 98 Small Ribosomal Subunit RNA and the Phylogeny of Mollusca Birgitta Winnepenninckx Departement Biochemie Universiteit Antwerpen (U.I.A.) Universiteitsplein 1, B-2610 Antwerpen, Belgium Vautierstraat 29 Thierry Backeljau Afdeling Malacologie Koninklijk Belgisch Instituut voor Natuurwetenschappen Rupert De Wachter! Departement Biochemie Universiteit Antwerpen (U.1.A.) Universiteitsplein 1, B-2610 Antwerpen, Belgium B-1040 Brussel, Belgium ABSTRACT We determined the complete sequence of the small ribosomal subunit RNA of the pulmonate snail Onchidella celtica. This sequence and the one recently determined for the chiton Acan- thopleura japonica were added to an alignment of 25 18SrRNA sequences of Metazoa, including three other Mollusca. The data set was used to assess certain aspects of molluscan phylogeny by distance matrix and character state methods. The trees ob- tained were tested for effects of random and systematic errors. The results of our analyses support: (a) molluscan monophyly; (b) gastropod monophyly; (c) bivalve monophyly; (d) a sister group relationship of Gastropoda and Polyplacophora. The po- sition of the phylum among other Metazoa remains uncertain due to a lack of representatives of many invertebrate phyla in our data set. Most of our results are congruent with existing hypotheses. Key Words: 18S rRNA, phylogeny, Metazoa, Gastropoda, Bi- valvia, Polyplacophora, Onchidella celtica. INTRODUCTION Historical Background Many aspects of molluscan phylogeny are still uncertain. The huge phenotypic diversity within the phylum ob- scures the evolutionary relationships between the larger molluscan taxa (e.g. von Ihering, 1876; Milburn, 1960; von Salvini-Plawen, 1969, 1972, 1985, 1990a,b; Stasek, 1972; Gotting, 1980; Wingstrand, 1985; Scheltema, 1988; Brusca & Brusca, 1990). Nevertheless, it is generally ac- cepted that the “shell- bearing molluscs” (Conchifera, i.e. Cephalopoda, Scaphopoda, Bivalvia, Gastropoda and Monoplacophora) are monophyletic with Polyplacoph- ora as sister group (e.g. von Salvini-Plawen, 1969, 1985, 1990a; Stasek, 1972; Gotting, 1980; Wingstrand, 1985; Scheltema, 1988; Brusca & Brusca, 1990). Indeed, the loss of spicules and the presence of three mantle margin folds, an univalve shell consisting basically of three lay- ers, jaws, a head with cerebrally innervated appendages, 1 Author for correspondence. a nervous system differentiated in axons and ganglia, a crystalline style and statocysts are considered to be syn- apomorphies uniting the five conchiferan classes (e.g. Gotting, 1980; von Salvini-Plawen, 1985; Wingstrand, 1985; Brusca & Brusca, 1990). However, different inter- pretations exist about the phylogenetic relationships within this subphylum. Milburn (1960) suggested three conchiferan clades: Monoplacophora, Cephalopoda and a Bivalvia-Gastropoda-Scaphopoda clade. The branching pattern of these three groups and the topology of the Bivalvia-Gastropoda-Scaphopoda clade, remains unre- solved. Gotting (1980) proposed a Bivalvia-Scaphopoda sister relationship and relied on the shell structure and form of the larval shell to conclude that Gastropoda and Monoplacophora are sister groups. Cephalopoda is then a sister group to the other four conchiferan classes. How- ever, Wingstrand (1985), Brusca and Brusca (1990) and von Salvini-Plawen (1985, 1990a) considered “Monopla- cophora’’ (i.e. class Tergomya sensu Peel, 1991 or Try- blidiida sensu Wingstrand, 1985; see Peel, 1991 for a discussion) as a sister group to the four other conchiferan classes, which in turn consist of a Bivalvia-Scaphopoda clade and a Gastropoda-Cephalopoda clade. The former is characterized by the presence of a mantle surrounding the entire body, reduction of the head and a laterally compressed form. The latter is determined by the pres- ence of a well developed head, dorsoventral elongation, dorsal concentration of the viscera and shell coiling (Brus- ca & Brusca, 1990). Except for the position of the “Mon- oplacophora’”, this view agrees well with the division of the Conchifera into the clades Diasoma (classes Bivalvia, Scaphopoda and the fossil Rostroconchia) and Cyrtosoma (classes ““Monoplacophora”, Gastropoda and Cephalo- poda), which is widely accepted among paleontologists (Runnegar & Pojeta, 1974, 1985; Pojeta, 1980; Steiner, 1992). Yet, Peel (1991) recently suggested that Cyrto- soma and Diasoma are both polyphyletic. The oldest fossil molluscs date from 570 MYA (e.g. Runnegar & Pojeta, 1974, 1985; Valentine, 1980), near the Precambrian-Cambrian boundary. This period was marked by an explosive radiation of animals resulting in the appearance of most extant invertebrate phyla (e.g. B. Winnepenninckx et al., 1994 Table 1. List of 17 oligonucleotides complementary to con- served regions in eukaryotic 18S rRNA genes. These were used to determine the sequence of both strands of the 18S rRNA gene of Onchidella celtica. Corresponding position in the 18S rRNA gene Sequence! Strand? of Onchildella celtica CTGGTTGATYCTGCCAGT R 4-21 GAAACTGCGAATGGCTCATT R 82-101 AATGAGCCATTCGCAGTTTC C 101-82 AGGGYTCGAYYCCGGAGA R 393-410 TCTCCGGRRTCGARCCCT C 410-393 TCTCAGGCTCCYTCTCCGG C 422-404 ATTACCGCGGCTGCTGGC C 605-588 CGCGGTAATTCCAGCTCCA R 097-615 TTGGYRAATGCTTTCGC C 990-974 TTRATCAAGAACGAAAGT R 1002-1019 CCGTCAATTYYTTTRAGTTT C 1188-1169 AATTTGACTCAACACGGG R 1221-1238 GGGCATCACAGACCTGTTAT C 1479-1460 ATAACAGGTCTGTGATGCCC R 1460-1479 TTTGYACACACCGCCCGTCG R 1666-1685 GACGGGCGGTGTGTRC C 1684-1669 CYGCAGGTTCACCTACRG C 1833-1816 1 Sequence positions where both purines (A and G) are present are indicated by “R”, those where both pyrimidines (C and T) are present by “Y”’. 2 Oligonucleotides with a sequence corresponding to that of the RNA-like strand are indicated by a “R’, those whose sequence is complementary to it, by a “C’. Bergstrom, 1991; Erwin, 1991; Valentine, 1991). Several aspects of the metazoan branching pattern still remain confused due to the doubtful homology of the relatively few morphological, anatomical and embryological char- acters shared by different phyla (e.g. Nielsen, 1977; An- derson, 1981; Inglis, 1985; Bergstrom, 1986; Ax, 1989; Schram, 1991; Backeljau et al., 1993). Nevertheless, Mol- lusca appear to be a monophyletic group belonging to the Spiralia (i.e. Platyhelminthes, Nemertini, Mollusca, Sipuncula, Echiura and Annelida, and probably Gna- thostomulida and Entoprocta) (e.g. Wingstrand, 1985; Brusca & Brusca, 1990; Willmer, 1990), but no syna- pomorphies are known linking the Mollusca unambig- uously to any other spiralian phylum (e.g. Wingstrand, 1985; Erwin, 1991). Some authors (e.g. von Salvini-Plaw- en, 1990a) suggest a sister group relationship to Turbel- laria (Platyhelminthes) considering the flat, often cili- ated, ventral creeping foot as a synapomorphy relating both phyla. Many others however, include the Mollusca in the protostome clade (e.g. Wingstrand, 1985; Brusca & Brusca, 1990; Willmer, 1990; Schram, 1991). Biochemical and molecular characters have been in- troduced as an independent source of phylogenetic in- formation. A serological study of molluscs, echinoderms, annelids and arthropods suggested that Mollusca are most closely related to Annelida (Wilhelmi, 1944). Lyddiatt et al. (1978) used cytochrome c amino acid sequence data to deduce a sister group relationship between mol- Page 99 luscs and echinoderms. In studies using 5S ribosomal RNA (rRNA) sequences (Ohama et al., 1984; Hendriks et al., 1986; Hori & Osawa, 1987), the Mollusca (rep- resented by Bivalvia, Gastropoda and Cephalopoda) ap- peared as a polyphyletic group. From the analysis of Lenaers and Bhaud (1992) on the basis of partial se- quences of 28S rRNA, Mollusca (represented by Mytilus edulis) appeared to be a sister group to Annelida. Holland et al. (1991) used partial small subunit (SSU) rRNA (18S tRNA) sequences and suggested that Mollusca (repre- sented by Mytilus edulis) and Arthropoda are sister taxa. On the basis of mitochondrial SSU rRNA sequences the Mollusca, represented by a prosobranch and a chiton, appeared as sister group to the Annelida or as a para- phyletic group including the latter phylum (Ballard et al., 1992) . In all these studies, however, the data sets were too limited to allow reliable conclusions. Field et al. (1988) determined partial sequences of SSU rRNA from representatives of ten different metazoan phyla including four Mollusca, viz. an opisthobranch gastropod, two bivalves and a chiton. Yet, different phylogeny in- ference methods yielded contradictory results. Field et al. (1988; see also Raff et al., 1989) used a distance method to conclude that Mollusca form a clade with Annelida, Sipuncula, Brachiopoda and Pogonophora. However, the relationships between the five groups were not resolved. Ghiselin (1988, 1989) reanalyzed this data set with a ‘signature’ approach and concluded that mol- luscs are a sister group to the Annelida sensu lato (i.e. Annelida sensu strictu, Brachiopoda, Pogonophora and Sipuncula). A maximum parsimony analysis of the same data produced a similar clade containing Sipuncula, Po- gonophora, Brachiopoda, Annelida and Mollusca (Pat- terson, 1989) but with the latter two phyla not being monophyletic. Lake (1989), who applied evolutionary parsimony, also concluded that Mollusca are paraphy- letic. In a preliminary attempt, we use complete SSU rRNA sequences to assess molluscan phylogeny. We consider sequences to be complete if (1) the sequence of the entire 18S rRNA molecule is known or (2) if only a total number on the order of 50 nucleotides at the 5’ and 3’ terminal parts are missing because they are used as PCR primer annealing sites (e.g. Rice, 1990; Littlewood, 1991). Hith- erto, complete 18S rRNA sequences of only three mol- luscan species viz., the bivalves Placopecten magellan- icus and Crassostrea virginica and the gastropod Limicolaria kambeul, have been published respectively by Rice (1990), Littlewood (1991) and Winnepenninckx et al. (1992) . In this paper we present the complete 185 rRNA sequence of the gymnomorphan snail Onchidella celtica (Cuvier, 1817). A fifth molluscan sequence (Win- nepenninckx et al. , 1993), that of the chiton Acantho- pleura japonica (Lischke, 1873) is also included. Small ribosomal subunit RNA sequences SSU rRNA sequences combine several features that make them appropriate for phylogenetic studies (Raff et al., LL Page 100 THE NAUTILUS, Supplement 2 Onchidella celtica 39 43 oA a, ROG A Aw cl Sy ts CCU UA AEA GUUCEC Cady c GaGa GGaGauu UCAAGcaacca Oo A c ueacu” a c au UrA A A G-U @ taemeezs UlA "Gy ESN A (Cota {e) ic} c-a Aru uU c-a eS Q-u G Yo 6:6 44 U-AgUA AsU A) Qc-a © v8 Ye.c oor wu Ura uU “ac 426 oh oA ye cy . uac?, u A uP Ss A aA u shee A c A Boys Oe? “pe oy Pacih aa Ue Ae poly ge 27 29 YA 36 a:¢ ace 46 ACN A gare u u e A G Q-u eo A c ca A Oo c aca 45 ccuaacc @ Cie u AsU 2 SoG) °. aaaucea c A A c:G AA A G A G-c Cc AaeuA c 2 e ca AA c-a r 24 Au Q-u au'G Gesu AeU a Gu a-c c-a E23-8 E23-9 Aor Uc OP a nes a-c CA .UA rach A G 28 .c ec e Gia A A "A ag cc,A aug GAACUGAGGUAAUGAU: A GAA Uc™, gl E23-7 ‘ G fe} eiovoletelenotere UCAA ACGA G. "6S, U 6 GUUGUUUUAUCUUGGCUCC sieiesel sieieke) *Gau iris @GuUU_ uGCcU Oe} A c UAAUGGAAUA GA Cc u u A A A A c co Yaccus Yoon OT: - 23 9, Baca CGGAC ACUU, U A Cyc Q,A au: é AAV ATATU c *a, Gane u 22 3 6 Ae uA. vu c uv Ae ue. tyY cc) E23-6 ,*%.:¢ m8 AU AAA ‘ G 1 Aaea* “ucaccaaccaca ee -u UU, A ~GaccauccU acuGuUUccCGU Gc hn Ao S6a5e nn BN cu A Gc u-a CUGGU gq A c @ GaAGcGucCcGGG eB Ula UgIOU, u Co 6 * SOS SS aca 6 M90 Sh, B ¢ cucGueGcGccc U c c Cc c u uc GcS& u-a c uu Gu 20 “s¢¢ “1gace 3 cy "eega™m G UA u G-u Cc G 49 E23-5 C-G Ge GA GA GPa u u Gc vACUAUS vusAscosuUCAgU” “cascauc — yuggUC cagucyoauaL’c acu, 4 AUGAAAG ,GGUUCGCUAGCUCA CUCGCAG AGCCGG GCCAGGCCACG G A A AG A,@ uc Cc c Gat Ue A i A a So, A c u G, 0S 9 Ane . A A ACCUMETACHG A 50 “oo A ac & So uv Aca G-c, uuuccauace a Hott coesecceee CG, USU¢ GAAGGCGUCC , A a-¢ u 6 G, aaa, & A Us A u 15 oc) -*A8 uu cacue © Uy hy oo BOoo Ih AW? CAA GUGG c c A c,c A A x Sacuace 7 J aw) eaa c 14° G-Cq 6 Cau @:¢ Sua e° ayucucce®" °, 1 Bc UAAGGGGC u AA u A A A Ne 2Uq Cc CUUGGAU uaa a GUAGAUA AUG yy c Nn 6 ti Yeg Figure 1. Secondary structure model for the 18S rRNA of Onchidella celtica. Helix numbering from helix 19 onward has changed with respect to the numbering used by De Rijk et al. (1992), due to the discovery of a tertiary structure interaction in helix 19 (Woese & Gutell, 1989). B. Winnepenninckx et al., 1994 Table 2. The 18S rRNA sequence of the gastropod Onchidella celtica. UAUCUGGUUGAUCCUGCCAGUAGUCAUAUGCUUGUCUCAAAGAUUAAGCC AUGCAUGUCUAAGUUCACACUGUCUCACGGUGAAACCGCGAAUGGCUCAU UAAAUCAGUCGAGGUUCCUUAGAUGACACGAUCCUACUUGGAUAACUGUG GCAAUUCUAGAGCUAAUACAUGCUAUUCAAGCUCCGACCCUCUGGGGAAG AGCGCUUUUAUUAGUUCAAAACCAAUCGCCGUGUGCUCUUCCCGGGGCCG GGCGUCCCCCUUGGUGACUCUGGAUAACUUUGUGCUGAUCGCAUGGCCUU UUGCGCCGGCGACGCAUCUUUCAAAUGUCUGCCCUAUUAAAUGCGAUGGU ACGUGAUAUGCCUACCAUGUUUGUAACGGGUAACGGGGAAUCAGGGUUCG AUUCCGGAGAGGGAGCAUGAGAAACGGCUACCACAUCCAAGGAAGGCAGC AGGCGCGCAACUUACCCACUCCCGGCACGGGGAGGUAGUGACGAAAAAUA ACAAUACGGGACUCUUUCGAGGCCCAGUAAUUGGAAUGAGUACACUUUAA ACCCUUUAACGAGGAUCUAUUGGAGGGCAAGUCUGGUGCCAGCAGCCGCG GUAAUUCCAGCUCCAAUAGCGUAUAUUAAAGUUGUUGCAGUUAAAAAGCU CGUAGUUGGAUCUCAGGCGCAGGCGGGCGGUCCGGCUCGCGCCGCUCACU GCCCGUUGUCUCCUGCCCUACCUGUUGCCGGCUCUCUCCCGUGGGUGCUC UUCGCUGAGCGUCCGGGUGGC CGGCGCGUUUACUUUGAAAAAAUUAGAGU GUUCAAAGCAGGCCUCGCCUGCCUGAAUAAUUGCGCAUGGAAUAAUGGAA UAGGACCUCGGUUCUAUUUUGUUGGUUUUCGGAACUGGAGGUAAUGAUUA ACAGGGACAAACGGGGGGAUUCGUAUUGCGGCGUUAGAGGUGAAAUUCUU GGAUCGCCGCAAGACGAGCUACUGCGAAAGCAUUUGUCAAGAAUGUUUUC AUUAAUCAAGAACGAAAGUCAGAGGCGAGAAGACGAUCAGAUACCGUCGU AGUUCUGACCAUAAACGAUGCCGACCAGCGAUCCGCAGGAGUUGCUUCGA UGACUCUGCGGGCAGCUUCCGGGAAACCAAAGUGUUUGGGUUCCGGGGGA AGUAUGGUUGCAAAGCUGAAACUUAAAGGAAUUGACGGAAGGGCACCACC AGGAGUGGAGCCUGCUGCUUAAUUUGACUCAACACGGGAAAACUCACCCG GUCCGGACACUGUAAGGAUUGACAGAUUGAUAGCUCUUUCUUGAUUCGGU GGGUGGUGGUGCAUGGCCGUUCUUAGUUGGUGGAGCGAUUUGUCUGGUUA AUUCCGAUAACGAACGAGACUCUAGCCUAUUAAAUAGUUCGCCGGUCCCU CGAUGCGCCGGCGCAACUUCUUAGAGGGACGAGUGGCGUUUAGCCAACGA GAUUGAGCAAUAACAGGUCUGUGAUGCCCUUAGAUGUCCGGGGCCGCACG CGCGCUACACUGAAGGAAUCAGCGUGGAUGCCUCCCUGGCCCGAAAGGCU GGGAAACCCGUUGAAUCUCCUUCGUGCUAGGGAUUGGGGCUUGUAAUUCU UCCCCAUGAACGAGGAAUUCCCAGUAAGCGCGAGUCAUAAGCUCGCGUUG AUUACGUCCCUGCCCUUUGUACACACCGCCCGUCGCUACUAUCGAUUGAG CGGUUCAGUGAGGGCAUCGGAUUGGUCUCGGUCUGGUGUUCGCGCACCGG CACCGCUGGCCGAGAAGACGCUCGAACUCGAUCGCUUGGAGAAAGUAAAA GUCGUAACAAGGUUUCCGUAGGUGAACCUGCGGAAGGAUCAUUA Page 101 1989; Hillis & Dixon, 1991; Solignac et al., 1991; Woese, 1991): (1) universality; (2) constancy of function; (3) al- ternation of conserved regions with variable ones, allow- ing phylogenetic studies at a broad range of taxonomical levels; (4) presence of conservative regions that allow the design of “universal” primers; (5) a conservative second- ary structure facilitating the identification of homologous positions in regions with little sequence similarity; (6) apparent absence of lateral gene transfer; (7) a large information content (1800-1900 bp) (8) intraspecific se- quence homogeneity among different gene copies (Ger- bi, 1985; Dover, 1986). Gene cloning Much sequence information on rRNAs has been obtained by direct RNA sequencing using reverse transcriptase (Lane et al., 1985; Solignac et al., 1991) or by direct sequencing of the rRNA genes after PCR amplification (Saiki et al., 1988). Both techniques are very rapid. Yet we prefer to clone and sequence the 18S rRNA genes, for direct RNA sequencing has some disadvantages: (1) RNA is less stable than DNA; (2) subsequent checking of sequences is not possible; (3) sequencing of regions with strong secondary structure is difficult; (4) reverse transcriptase has a rather high error frequency; (5) only one strand is available and thus two-strand verification is not possible. All this results in an overall error rate of about 1% (Lane et al., 1985). Although PCR amplifica- tion eliminates a great deal of these problems, it also has some drawbacks (Hillis & Dixon, 1991): (1) Taq poly- merase has a high error rate, viz. 210-4 to <1X10° according to Eckert and Kunkel (1991) and 2.75 10° according to Bej et al. (1991); (2) the 3’ and 5’ parts of the gene itself have to be used as primer annealing sites, if the sequence of the adjacent regions is unknown; (3) direct sequencing of PCR amplified fragments is difficult (e.g. Gyllenstein, 1989); (4) the product is afterwards not available to others for verification. By cloning the PCR product prior to sequencing, the latter two problems can be overcome, but the sequencing of numerous clones is necessary to avoid an enhancement of the error rate Page 102 Distance 0.1 ee 92, Rattus norvegicus 1 Mus musculus Homo sapiens 1 O16 Echinorhinus cookei 76 Tenebrio molitor Artemia salina Eurypelma californica Crassostrea virginica 100 Acanthopleura japonica | 0 Onchidella celtica Limicolaria kambeul 100 100 Anemonia sulcata Oryctolagus cuniculus Turdus migratorius Heterodon platyrhinos Alligator mississippiensis Xenopus laevis Ps Latimeria chalumnae ee ‘ Squalus acanthias 81| Sebastolobus altivelis Fundulus heteroclitus Herdmania momus Oedignathus inermis Placopecten magellanicus Schistosoma mansoni 1 Opisthorchis viverrini Paramecium tetraurelia THE NAUTILUS, Supplement 2 Vertebrata Chordata Tunicata Crustacea Insecta Crustacea Arthropoda Acyrthosiphon pisum Insecta Chelicerata Pivatvia Polyplacophora |Mollusca _|sestronoaa Trematoda Platy = helminthes Cnidaria Ciliata Figure 2. Neighbor-joining tree based on the 18S rRNA sequences from 27 Metazoa. All sequences were complete except for the following (number of sequenced nucleotides between brackets): Turdus migratorius (1753), Alligator mississippiensis (1691), Heterodon platyrhinos (1717) and Latimeria chalumnae (1777). Paramecium tetraurelia was chosen as an outgroup. Bootstrap values are indicated at the root of each clade, but only if they exceed 50%. (Bevan et al., 1992). This of course reduces the time advantage of PCR amplification. MATERIALS AND METHODS Animals: Specimens of Onchidella celtica collected at Vila Franca do Campo (Sao Miguel, Azores) were frozen alive and preserved at -80°C. Voucher material was de- posited in the collections of the “Koninklijk Belgisch Instituut voor Natuurwetenschappen , Brussels (general inventory number, I.G. No. 28053). DNA extraction: Digestive glands of ten specimens were pooled and homogenized under liquid nitrogen in a pre- chilled mortar and transferred to 15 ml of preheated (60°C) 2% CTAB buffer (2% (w/v) CTAB; 0.2% (v/v) 2-mercaptoethanol, 1.4 M NaCl; 20 mM EDTA; 100 mM Tris-HCl pH=8; 100 ug/ml proteinase K). After incu- bation at 60°C for 30 min., further extraction was done Table 3. Organisms that were used as outgroup in our analyses. Species Position Zea mays angiosperms Neurospora crassa ascomycetes Saccharomyces cerevisiae ascomycetes Rhodosporidium toruloides basidiomycetes Gracilaria lemaneiformis red algae Porphyra umbilicalis red algae green algae green algae dinoflagellates Chlorella ellipsoidea Volvox carteri Prorocentrum micans Giardia duodenalis diplomonads Trypanosoma brucei kinetoplastids Paramecium tetraurelia ciliates Oxytricha nova ciliates apicomplexa (Sporozoa) slime molds Plasmodium berghei Dictyostelium discoideum B. Winnepenninckx et al., 1994 Distance 0.1 90° Rattus norvegicus 1 98! Mus musculus Homo sapiens 1 100 Tenebrio molitor Artemia salina Eurypelma californica Anemonia sulcata Zea mays Echinorhinus cookei Placopecten magellanicus Crassostrea virginica 400 Acanthopleura japonica a 04 Onchidella celtica Limicolaria kambeul Oryctolagus cuniculus Turdus migratorius 91 | 59) Heterodon platyrhinos Alligator mississippiensis Xenopus laevis os Latimeria chalumnae 100] Squalus acanthias ; Fundulus heteroclitus Sebastolobus altivelis Page 103 Vertebrata Chordata Herdmania momus ~ Tunicata Schistosoma mansoni 1 RASMACSSIA ie ee Opisthorchis viverrini minthes Oedignathus inermis Crustacea Insecta Crustacea Arthropoda Acyrthosiphon pisum Insecta Chelicerata [pivaivia _Polyplacophora Mollusca pastropoaa Cnidaria Magnolic- phyta Figure 3. Neighbor-joining tree based on the same set of metazoan 18S rRNA sequences as in Fig. 2, but with Zea mays as an outgroup. Bootstrap values are indicated as in Fig. 2. as described by Winnepenninckx et al. (1993a). The DNA yield amounted to 60 ug. Gene cloning and sequencing: Restriction enzymes suit- able for isolation of a DNA fragment containing the 18S rRNA gene were identified as described by Winnepen- ninckx et al. (1992). After digestion of 1.2 ug DNA with BamHI and separation on a 0.8% (w/v) agarose gel, restriction fragments of 4 kb containing the 18S rRNA gene were eluted (Heery et al., 1990). Competent DH5a E. coli cells (Gibco BRL Life Technologies; Gaithersburg, USA) were transformed with these DNA restriction frag- ments ligated into pBluescriptSK* (Stratagene; La Jolla, California, USA). Colony screening was performed using a PCR fragment of the gastropod Limicolaria kambeul (Winnepennincksx et al., 1992), labeled with **P via nick translation (Rigby et al., 1977). Plasmids were isolated (Birnboim & Doly, 1979) from a single clone and se- quencing was performed by the dideoxynucleotide method (Sanger et al., 1977) using Sequenase 2.0 (USB; Cleveland, Ohio, USA). The 18S rRNA primers used are given in Table 1. Sequence alignment and construction of phylogenetic trees: The Onchidella celtica 18S rRNA sequence was aligned with other SSU rRNA sequences present in our database (De Rijk et al., 1992). Alignment was done manually taking into account the secondary structure features of the molecule, as described by De Rijk et al. (1992). For tree construction, pairwise distances were calculated using the formula of Jukes and Cantor (1969) modified to take into account gaps (Van de Peer et al., 1990). They served to derive neighbor-joining trees (Sai- tou & Nei, 1987), whose reliability was tested by boot- strapping (Felsenstein, 1985) over 100 replicates. Ac- cording to the guidelines of Hillis and Bull (1993), only branching points with bootstrap values higher than 70% were considered to be reliable. Estimated internal branches with bootstrap values above 70% should rep- resent true clades over 95% of the time (Hillis & Bull, Page 104 Distance 0.1 —— THE NAUTILUS, Supplement 2 98, Rattus norvegicus 1 99} Mus musculus 100|L Homo sapiens 1 Turdus migratorius 70 Oryctolagus cuniculus 9 Tenebrio molitor Artemia salina Eurypelma californica Placopecten magellanicus Crassostrea virginica canthopleura japonica Anemonia sulcata Acyrthosiphon pisum Giardia lamblia Heterodon platyrhinos Xenopus laevis Latimeria chalumnae eg Squalus acanthias Echinorhinus cookei " Sebastolobus altivelis Fundulus heteroclitus Alligator mississippiensis Vertebrata Ghoraaes Herdmania momus Tunicata j j hel- EMSS 1 Giensen Ee es eee: Opisthorchis viverrini Naeia Oedignathus inermisCrustacea Insecta Arthropoda Crustacea P Chelicerata ul ‘jpivaivia Polyplacophora| Mollusca Gastropoda Cnidaria Arthropoda Poly - mastigotes Insecta Figure 4. Neighbor-joining tree based on the same set of metazoan 18S rRNA sequences as in Fig. 2, but with Giardia duodenalis (often called Giardia lamblia or Giardia intestinalis) as an outgroup. Bootstrap values are indicated as in Fig. 2. 1993). All calculations were carried out with the TREE- CON package of Van de Peer and De Wachter (1993). Character state analyses using maximum parsimony were performed using the package HENNIG86 (version 1.5; Farris, 1989) with the heuristic algorithms MHENNIG* and BB* combined. The results were summarized in a strict consensus tree, i.e. a tree that contains only those clusters that are common to all competing trees (“nelsen” command of HENNIGS86). Nucleotides were treated as non-additive characters and no differential weighting was done. RESULTS Sequence Alignment The 18S rRNA of Onchidella celtica (EMBL accession number X70211), of which the nucleotide sequence is shown in Table 2, is 1844 nucleotides long. The 3’ and 5’ termini of the gene were located on the basis of sim- ilarity with those of other 18S rRNA sequences. Figure 1 shows a secondary structure model of the molecule in accordance with the one published for Limicolaria kam- beul (Winnepenninckx et al., 1992). Both models show high similarity to each other and are in accordance with the general model proposed for eukaryotic SSU rRNA (De Rijk et al. 1992). Based on our latest insights into the secondary structure of 18S RNA, modifications were made in helices 19, 20, 21 and 38. The new gastropod sequence as well as the one of Acanthopleura japonica (Winnepenninckx et al., 1993b) were added to an align- ment of other SSU rRNA sequences (De Rijk et al., 1992). This alignment can be obtained on request. Trees were constructed on the basis of a set of 27 metazoan sequences which are either complete or nearly complete. Distance Matrix Analyses Figure 2 shows the neighbor joining (NJ) tree obtained with the ciliate Paramecium tetraurelia as outgroup. It B. Winnepenninckx et al., 1994 Page 105 Distance 0.1 ——_—— 95, Rattus norvegicus 1 99! Mus musculus Homo sapiens 1 Turdus migratorius Xenopus laevis ‘a Latimeria chalumnae 9 ‘oof Squalus acanthias Echinorhinus cookei 73 Sebastolobus altivelis Fundulus heteroclitus 80 Herdmania momus Artemia salina Tenebrio molitor Eurypelma californica Placopecten magellanicus Crassostrea virginica 100 Acanthopleura japonica on 109 Onchidella celtica Limicolaria kambeul 100 100 Anemonia sulcata Oryctolagus cuniculus Heterodon platyrhinos Alligator mississippiensis Schistosoma mansoni 1 Opisthorchis viverrini Paramecium tetraurelia Crustacea Insecta Arthropoda Chelicerata Acyrthosiphon pisum Insecta Bivalvia Polyplacophora| Mollusca ta : Vertebra ete See Tunicata Gastropoda 7 Trematoda Platyhel- minthes Cnidaria Ciliata Figure 5. Neighbor-joining tree obtained on the basis of a set containing all the 185 rRNA metazoan sequences of Fig. 2 except Oedignatus inermis. Paramecium tetraurelia was used as an outgroup. Bootstrap values are indicated as in Fig. 2. suggests that (bootstrap values in parentheses): (1) Mol- lusca are a monophyletic group (100/100) within a rel- atively poorly supported protostome clade (62/100); (2) Gastropoda (100/100) and Bivalvia (99/100) are mono- phyletic as well; (3) Polyplacophora appears as a sister group to the Gastropoda (100/100). The tree also indi- cates that : (1) Cnidaria are a sister group to Eubilateria (100/100); (2) Acoelomata, represented by two Trema- toda, are a sister group to the Eucoelomata (76/100); (3) Arthropoda are a monophyletic group (88/100); (4) nei- ther Insecta nor Crustacea are monophyletic; (5) Chor- data (99/100) and Vertebrata (100/100) are both mono- phyletic. We attempted to assess the stability of our tree by testing its sensitivity to the presence of specific taxa. First we studied the influence of the outgroup by successively replacing Paramecium tetraurelia by each of the 14 other organisms listed in Table 3. We observed only two topological changes. In nine out of the 15 cases, the topology shown in Figure 3 was obtained, i.e. the Pla- tyhelminthes appeared as a sister group to the Arthro- poda-Mollusca clade. In one case, the topology shown in Figure 4 was obtained, viz. when the diplomonad Giar- dia duodenalis, was chosen as outgroup. This organism forms a very long branch in previously published trees comprising organisms from different eukaryotic king- doms (e.g. Van de Peer et al., 1993). In this case, the aphid Acyrthosiphon pisum, which is also marked by an exceptionally long branch, became a sister group to all other Metazoa. The latter observation is probably due to the fact that errors in distance estimation increase with the amount of divergence. Long branches will provoke an underestimation of the evolutionary distance and will systematically attract each other, causing biased topol- ogies (Felsenstein, 1978; Olsen, 1987; Swofford & Olsen, 1990; Lake, 1991). The changes in the position of the Platyhelminthes, which do not have exceptionally long branches, is probably not due to such a systematic error. The low bootstrapping values on their branching point, suggests uncertainty as to their position. Inclusion of rep- resentatives of more invertebrate phyla might be helpful in this case. Page 106 _[pescnoste sulcata Paramecium tetraurelia | j-Heramania momus |_| Oats emorehis W/ILW/AS'S IS aL Oat Schistosoma mansoni Ee erases celtica ee ee californica Artemia salina peige Tenebrio molitor f—Latimeria chalumnae ees omo sapiens Lise. musculus f—Acanthopleura japonica LY ptimicotaria kambeul Gignathus inermis p 2 HeoOseerem magellanicus =| es virginica | | L| [racyrthosiphon pisum | iS Squalus acanthias | j-Fundulus heteroclitus psoas colons altivelis LE Pehenorhinus cookei Xenopus laevis po Seeeoson platyrhinos Alligator mississippiensis res migratorius ryctolagus cuniculus THE NAUTILUS, Supplement 2 Cnidaria Ciliata Tunicata Chordata Trematoda Platyhel | minthes Polyplacophora Mollusca Gastropoda Chelicerata Crustacea Arthropoda | a8ecce Vertebrata Chordata Rattus norvegicus | Figure 6. Strict consensus tree constructed from three maximum parsimony trees (length=3114 steps; c.i= 0.51) obtained by applying the MHENNIG* + BB* option of Hennig86 on the 706 informative positions of the same alignment as in Fig. 2 and with Paramecium tetraurelia chosen as outgroup. Subsequently, we constructed 27 trees with Parame- cium tetraurelia as outgroup, but each time omitting one species. Only one topological change was observed: when excluding Oedignathus inermis, Acyrtosiphon pisum branched off first within the arthropod clade (Figure 5). The fact that this change involves the species with the longest branch, again points to the above mentioned “long branch effect” (Felsenstein, 1978; Swofford & Olsen, 1990). Since the placement of the two Platyhelminthes was ambiguous (cfr. Figures 2 and 3) and since we sus- pected Acyrthosiphon pisum to be a source of systematic errors, we removed all three species from our data set to assess their impact. However, the topology of the tree we obtained did not differ from the one in Figure 2. Character State Analyses The 28 species analysed, with Paramecium tetraurelia as outgroup, yielded 706 informative sites. A position is informative if it contains at least two different nucleo- tides, each of them present in at least two species (Nei, 1987). Ambiguous nucleotides were not used to ascertain the informative character of a position. Three maximum parsimony (MP) trees of 3114 steps and with a consis- tency index (c.i.) (Kluge & Farris, 1969) of 0.51 were found. The strict consensus tree shown in Figure 6 sug- gests that (1) Mollusca, Bivalvia and Gastropoda are monophyletic groups; (2) Polyplacophora appear as a sister group to Gastropoda; (3) Arthropoda are a mono- phyletic clade in which Chelicerata branch off first; (4) Insecta are monophyletic but Crustacea are paraphyletic; (5) Vertebrata are monophyletic. Ten different data in- put orders did not change this topology. Again we tested the stability of our results. If the placement of a taxon is biased, its removal should cause an increase of the consistency index (Swofford & Olsen, 1990). We checked this by successively removing those species, the position of which appeared unstable in our distance matrix anal- yses, viz. Acyrthosiphon pisum, Schistosoma mansoni, Opisthorchis viverrini, and a fourth species, Mus mus- culus, which occupied a stable position. Each time we identified the informative positions anew and applied HENNIG86 with Paramecium tetraurelia as outgroup. Omitting Acyrthosiphon pisum increased the c.i. to 0.53, while removing any of the other species did not change the c.i. This again suggests that the placement of Acyr- thosiphon pisum is liable to a systematic error. As for the ambiguous position of the Platyhelminthes, this may B. Winnepenninckx et al., 1994 Page 107 _|Fetetonse sulecata Paramecium tetraurelia __ peesssyosom mansoni Opisthorchis viverrini ee celtica [et eolakia kambeul [jeeetzzcstcee japonica Re Crassostrea virginica Placopecten magellanicus Oedignathus inermis fe p BeMebELo molitor Artemia sali = alina | prreranania Calli ronnaca Herdmania momus p scnamorinaus cookei Ps ebas coLolsus altive | Erunduius heteroclit qualus acanthias Latimeria chalumnae | j-Xenopus laevis Alligator missi ae migrator Homo sapiens Heterodon platyrhinos Cnidaria Ciliata Platyhel — minthes Trematoda i Polyplacophora} Mollusca Gastropoda Bivalvia Se gis eal Crustacea al Insecta Arthropoda Crustacea Chelicerata Tunicata lis us Chordata Vertebrata . ssippiensis ius Crore cee cuniculus , ts musculus Rattus norvegicus Figure 7. Strict consensus tree constructed from two maximum parsimony trees (length=2741 steps; c.i=0.53) obtained with the MHENNIG*+BB* option on the 658 informative sites of the same alignment of Fig. 2 from which the insect Acyrthosiphon pisum was removed. be due to the lack of other invertebrate phyla and classes. Figure 7 shows the strict consensus tree of the two MP trees (length=2741; c.i=0.53) obtained when Acyrtho- siphon pisum was excluded. All our conclusions based on the tree in Figure 6 remain valid, but in addition the bilaterian pentachotomy of Figure 6 is now resolved. Mollusca appear as a sister group to a clade containing Arthropoda and Chordata. It is also suggested that (1) Bilateria are monophyletic; (2) Acoelomata are a sister group to Eucoelomata; (3) Chordata are monophyletic. DISCUSSION The monophyletic character of the Mollusca, the Bivalvia and the Gastropoda, which is supported by all our trees, is generally accepted (e.g. Brusca & Brusca, 1990; Will- mer, 1990; von Salvini-Plawen, 1985, 1990a; Gotting, 1980). Using globin amino acid sequences, Goodman et al. (1988) agreed with these views. The 5S rRNA based analyses of Ohama et al. (1984), Hendriks et al. (1986) and Hori and Osawa (1987) also confirmed gastropod monophyly. Ghiselin (1988, 1989) supported molluscan monophyly. But Patterson (1989) and Lake (1989) did not corroborate these well established views, while the question was not resolved by Field et al. (1988; see also Raff et al., 1989). In both the distance and MP trees, we find the chiton included within the conchiferan clade as a sister group to the Gastropoda. This result is in contrast with the results of anatomical (e.g. Milburn, 1960; Stasek, 1972; Gotting, 1980; Scheltema, 1988; Brusca & Brusca, 1990; von Salvini-Plawen, 1990a) and paleontological (e.g. Runnegar & Pojeta, 1974; Pojeta, 1980; Peel, 1991) stud- ies. Neither Field et al. (1988; see also Raff et al., 1989), nor Ghiselin (1988, 1989) or Lake (1989) were able to resolve the position of the Polyplacophora, while Pat- terson (1989) suggested that Polyplacophora and Brach- iopoda are sister taxa. Using mitochondrial SSU rRNA sequences (Ballard et al., 1992), the class either appeared as a sister group to the Gastropoda-Annelida clade or formed together with the Gastropoda a sister group to the Annelida. Addition of more molluscan representa- tives to our data set is necessary to investigate the con- flicting position of the Polyplacophora. Our current data set is also not sufficiently represen- tative to draw conclusions on the position of the Mollusca among other Metazoa. From our NJ analyses, the phylum appears as a sister group to the Arthropoda (see also eee ee ee rere ———— Page 108 Holland et al., 1991) but this topology is insufficiently supported by bootstrap values. According to the char- acter state analysis it branches off before the Chordata- Arthropoda clade. Data from additional invertebrate phyla should be included. All our current analyses strongly support the view that: (1) Arthropoda is a monophyletic group and (2) Verte- brata, Chordata and Bilateria are monophyletic. These observations are in agreement with the results of some classical (e.g. Ax, 1989; Brusca & Brusca, 1990; Schram, 1991) and molecular studies (Ghiselin, 1988, 1989; Pat- terson, 1989; Raff et al., 1989; Winnepenninckx et al., 1992). Contradictory views on these aspects of metazoan phylogeny were given by e.g. Lake (1989), Willmer (1990) and Fryer (1992). However our analyses do not allow conclusions on the status of the Acoelomata and the mono- or paraphyletic character of the Insecta and Crustacea. It is beyond the scope of this paper to expand on meta- zoan evolution, however the congruence of most of our results with independently derived hypotheses suggests that complete 18S rRNA sequences are a reliable tool to assess the phylogeny of the Mollusca and other metazoan groups. ACKNOWLEDGMENTS We are indebted to Dr. A.M. Frias Martins (University of the Azores) for helping to collect Onchidella celtica. The investigations were performed in the framework of the Institute for the Study of Biological Evolution of the University of Antwerp. This work was supported by FKFO Grants 2.0023.94 and 2.0003.93. B. Winnepen- ninckx holds an I.W.O.N.L. scholarship. LITERATURE CITED Anderson, D. T. 1981. Origins and relationships among an- imal phyla. Proceedings of the Linnean Society of New South Wales 106:151-166. Ax, P. 1989. Basic phylogenetic systematization of the Me- tazoa. in: Fernholm, B., K. Bremer and H. Jornvall (eds). The Hierarchy of life. Elsevier, Amsterdam, p. 229-245. Backeljau, T., B. Winnepenninckx and L. De Bruyn. 1993. Cladistic analysis of metazoan relationships: areappraisal. Cladistics 9:167-181. Ballard, J. W. O., G. J. Olsen, D. P. Faith, W. A. Odgers, D. M. Rowell and P. W. Atkinson. 1992. Evidence from 12S ribosomal RNA sequences that onychophorans are modified arthropods. Science 258:1345-1348. Bej, A. K., M. H. Mahbubani and R. M. Atlas. 1991. Ampli- fication of nucleic acids by polymerase chain reaction (PCR) and other methods and their applications. Critical reviews in Biochemistry and Molecular Biology 26:301-334. Bergstrom, J. 1986. Metazoan evolution - a new model. Zool- ogica Scripta 15:189-200. Bergstrom, J. 1991. Metazoan evolution around the Precam- brian-Cambrian transition. In: Simonetta, A.M. and S.C. Morris (eds). The early evolution of Metazoa and the sig- nificance of problematic taxa. Cambridge University Press, p. 25-34. Bevan, I. S., R. Rapley and M. R. Walker. 1992. Sequencing THE NAUTILUS, Supplement 2 of PCR-amplified DNA. PCR Methods and Applications 1:222-228. Birnboim, H. C. and J. Doly. 1979. A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nu- cleic Acids Research 7:1518-1522. Brusca, R. C. and G. J. Brusca. 1990. Invertebrates. Sinauer Associates, Sunderland, MA 922 pp. De Rijk, P., J.-M. Neefs, Van de Peer, Y. and R. De Wachter. 1992. Compilation of small ribosomal subunit RNA se- quences. Nucleic Acids Research 20:2075-2089. Dover, G. A. 1986. Molecular drive in multigene families: how biological novelties arise, spread and are assimilated. Trends in Genetics 2:159-165. Eckert, K. A. and T. A. Kunkel. 1991. DNA polymerase fidelity and the polymerase chain reaction. PCR Methods and Applications 1:17-24. Erwin, D. H. 1991. Metazoan phylogeny and the Cambrian radiation. Trends in Ecology and Evolution 6: 131-134. Farris, J. S. 1989. Hennig86: a PC-DOS program for phylo- genetic analysis. Cladistics 5:163. Felsenstein, J. 1978. Cases in which parsimony or compati- bility methods will be positively misleading. Systematic Zoology 27:401-410. Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-791. Field, K. G., G. J. Olsen, D. J. Lane, S. J. Giovannoni, M. T. Ghiselin, E. C. Raff, N. R. Pace and R. A. Raff. 1988. Molecular phylogeny of the animal kingdom. Science 239: 748-753. Fryer, G. 1992. The origin of the Crustacea. Acta Zoologica 73:273-286. Gerbi, S. A. 1985. Evolution of ribosomal DNA. In: Mac- Intyre, R.J. (ed). Molecular Evolutionary Genetics. Plenum Press, New York and London, p. 419-517. Ghiselin, M. T. 1988. The origin of molluscs in the light of molecular evidence. In: Harvey, P. H. and L. Partridge (eds). Oxford Surveys in Evolutionary Biology. Volume 5. p. 66-95. Ghiselin, M. T. 1989. Summary of our present knowledge of metazoan phylogeny. In: Fernholm, B., K. Bremer and H. Jornvall (eds). The Hierarchy of life. Elsevier, Am- sterdam, p. 261-272. Goodman, M., J. Pedwaydon, J. Czelusniak, T. Suzuki, T. Go- toh, L. Moens, F. Shishikura, D. Walz and S. Vinogradov. 1988. An evolutionary tree for invertebrate globin se- quences. Journal of Molecular Evolution 27:236-249. Gotting, K.-J. 1980. Origin and relationships of the Mollusca. Zeitschrift fiir zoologische Systematik und Evolutionsfor- schung 18:24-27. Gyllenstein, U. B. 1989. PCR and DNA sequencing. Bio- techniques 7:700-708. Heery, D. M., F. Gannon and R. Powell. 1990. A simple method for subcloning DNA fragments from gel slices. Trends in Genetics 6:173. Hendriks, L., E. Huysmans, A. Vandenberghe and R. De Wach- ter. 1986. Primary structures of the 5S ribosomal RNAs of 11 arthropods and applicability of 5S RNA to the study of metazoan evolution. Journal of Molecular Evolution. 24:103-109. Hillis, D. M. and M. T. Dixon. 1991. Ribosomal DNA: mo- lecular evolution and phylogenetic inference. The Quar- terly Review of Biology. 66:411-453. Hillis, D. M. and J. J. Bull. 1993. An empirical test of boot- strapping as a method for assessing confidence in phylo- genetic analysis. Systematic Biology. 42:182-192. B. Winnepenninckx et al., 1994 Holland, P. W. H., A. M. Hacker and N. A. Williams. 1991. A molecular analysis of the phylogenetic affinities of Sac- coglossus cambrensis Brambell & Cole (Hemichordata). Philosophical Transactions of the Royal Society of London (Series B) 332:185-189. Hori, H. and S. Osawa. 1987. Origin and evolution of organ- isms as deduced from 5S ribosomal RNA sequences. Mo- lecular Biology and Evolution 4:445-472. Inglis, W.G. 1985. Evolutionary waves: patterns in the origins of animal phyla. Australian Journal of Zoology 33:153- 178. Jukes T. H. and C. R. Cantor. 1969. Evolution of protein molecules. In: Munro, H. N. (ed). Mammalian Protein Metabolism. Academic Press, New York, p. 21-132. Kluge, A. G. and J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18:1- 32. Lake, J. A. 1989. Origin of the multicellular animals. In: Fernholm, B., K. Bremer and H. Jornvall (eds). The Hi- erarchy of life. Elsevier, Amsterdam, p. 273-278. Lake, J. A. 1991. Tracing origins with molecular sequences: metazoan and eukaryotic beginnings. Trends in Biochem- ical Sciences 16:46-50. Lane, D. J., B. Pace, G. J. Olsen, D. A. Stahl, M. L. Sogin and N. R. Pace. 1985. Rapid determination of 16S-ribosomal RNA sequences for phylogenetic analyses. Proceedings of the National Academy of Science of the USA 82:6955- 6959. Lenaers, G. and M. Bhaud. 1992. Molecular phylogeny of some polychaete annelids: an initial approach to the At- lantic-Mediterranean speciation problem. Journal of Mo- lecular Evolution 35:429-435. Littlewood, D. T. J., S. E. Ford and D. Fong. 1991. Small subunit rRNA gene sequence of Crassostrea virginica (Gmelin) and a comparison with similar sequences from other bivalve molluscs. Nucleic Acids Research 19:6048. Lyddiatt, A., D. Peacock and D. Boulter. 1978. Evolutionary change in invertebrate cytochrome C. Journal of Molec- ular Evolution 11:35-45. Milburn, P. W. 1960. Further remarks on the interpretation of the Mollusca. The Veliger 3:43-48. Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York, p. 315-319. Nielsen, C. 1977. The relationships of Entoprocta, Ectoprocta and Phoronida. American Zoologist 17:149-150. Ohama, T., T. Kumazaki, H. Hori and S. Osawa. 1984. Evo- lution of multicellular animals as deduced from 5S rRNA sequences: a possible early emergence of the Mesozoa. Nucleic Acids Research 12:5101-5108. Olsen, G. J. 1987. Earliest phylogenetic branchings: compar- ing rRNA-based evolutionary trees inferred with various techniques. Cold Spring Harbor Symposia on Quantitative Biology LII:825-837. Patterson, C. 1989. Phylogenetic relations of major groups: conclusions and prospects. In: Fernholm, B., K. Bremer, and H. Joérnvall (eds). The Hierarchy of life. Elsevier, Amsterdam, p. 471-487. Peel, J. S. 1991. Functional morphology of the class Helci- onelloida nov., and the early evolution of the Mollusca. In: Simonetta, A.M. and S.C. Morris (eds). The early evo- lution of Metazoa and the significance of problematic taxa. Cambridge University Press, p. 157-177. Pojeta, J. Jr. 1980. Molluscan phylogeny. Tulane Studies in Geology and Paleontology 16:55-80. Raff, R. A., K. G. Field, G. J. Olsen, S. J. Giovannoni, D. J. Page 109 Lane, M. T. Ghiselin, N. R. Pace and E. C. Raff. 1989. Metazoan phylogeny based on analysis of 18S ribosomal RNA. In: Fernholm, B., K. Bremer and H. Jornvall (eds). The Hierarchy of life. Elsevier, Amsterdam, p. 247-260. Rice, E. L. 1990. Nucleotide sequence of the 18S ribosomal RNA gene from the Atlantic sea scallop Placopecten ma- gellanicus. Nucleic Acids Research 18:5551. Rigby, P. W. J., M. Dieckmann, C. Rhodes and P. Berg. 1977. Labeling deoxyribonucleic acids to high specific activity in vitro by nick translation with DNA polymerase I. Jour- nal of Molecular Biology 113: 237-251. Runnegar, B. and J. Pojeta, Jr. 1974. Molluscan phylogeny: the paleontolog- ical viewpoint. Science 186:311-317. Runnegar, B. and J. Pojeta, Jr. 1985. Origin and diversification of the Mollusca. In: Wilbur, K. M., E. R. Trueman and M. R. Clarke (eds). The Mollusca. Volume 10. Evolution. Academic Press, Inc. p. 1-57. Saiki, R. K., D. H. Gelfand, S. Stoffel, S. J. Scharf, R. Higuchi, G. T. Horn, K. B. Mullis and H. A. Erlich. 1988. Primer- directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239:487— 491. Saitou, N. and M. Nei. 1987. The Neighbor-joining method: a new method for reconstructing phylogenetic trees. Mo- lecular Biology and Evolution 4:406—425. Sanger, F., S. Nicklen and R. Coulson. 1977. DNA sequencing with chain-terminating inhibitors. Proceedings of the Na- tional Academy of Science of the USA 74:5463-5467. Scheltema, A. H. 1988. Ancestors and descendents: relation- ships of the Aplacophora and Polyplacophora. American Malacological Bulletin 6:57-68. Schram, F.R. 1991. Cladistic analysis of metazoan phyla and the placement of fossil problematica. In: Simonetta, A. M. and S. C. Morris (eds). The early evolution of Metazoa and the significance of problematic taxa. Cambridge Univer- sity Press, p. 35-46. Solignac, M., M. Pélandakis, F. Rousset and A. Chenuil. 1991. Ribosomal RNA phylogenies. In: Hewitt G.M., A. W. B. Johnston and J. P. W. Young (eds). NATO ASI Series. Volume 57. Molecular Techniques in Taxonomy. Volume H57. Springer-Verlag, Berlin, Heidelberg, p. 73-85. Stasek, C. R. 1972. The molluscan framework. In: Florkin, M. and B. T. Scheer (eds). Chemical Zoology. Volume VII. Mollusca. Academic Press, New York and London, p. 1- 44. Steiner,G. 1992. Phylogeny and classification of Scaphopoda. Journal of Molluscan Studies 58:385-—400. Swofford, D. L. and G. J. Olsen. 1990. Phylogeny reconstruc- tion. In: Hillis, D. M. and C. Moritz (eds). Molecular Sys- tematics. Sinauer Associates, Inc., Sunderland, MA, p. 411- 501. Valentine, J. W. 1980. L’origine des grands groupes d’ani- maux. La Recherche 112:666-674. Valentine, J. W. 1991. Major factors in the rapidity and extant of the metazoan radiation during the Proterozoic-Phan- erozoic transition. In: Simonetta, A.M. and S.C. Morris (eds). The early evolution of Metazoa and the significance of problematic taxa. Cambridge University Press, p. 11- 13. Van de Peer, Y., J.-M. Neefs and R. De Wachter. 1990. Small ribosomal subunit RNA sequences, evolutionary relation- ships among different life forms, and mitochondrial ori- gins. Journal of Molecular Evolution 30:463-476. Van de Peer, Y., J.-M. Neefs, P. De Rijk and R. De Wachter. 1993. Evolution of eukaryotes as deduced from small Page 110 ribosomal subunit RNA sequences. Biochemical System- atics and Ecology 21:43-56. Van de Peer, Y. and R. De Wachter. 1998. TREECON: A software package for the construction and drawing of evo- lutionary trees. Computer Applications in the Biosciences 9: 177-182. von Ihering, H. 1876. Versuch eines natiirlichen Systemes der Mollusken. Jahrbuch der Deutschen Malakozoologischen Gesellschaft 3: 97-148. von Salvini-Plawen, L. 1969. Solenogastres und Caudofoveata (Mollusca, Aculifera) Organisation und phylogenetische Bedeutung. Malacologia 9:191-216. von Salvini-Plawen, L. 1972. Zur Morphologie und Phylo- genie der Mollusken; die Beziehungen der Caudofoveata und der Solenogastres als Aculifera, als Mollusca und als Spiralia. Zeitschrift zur Wissenschaftlichen Zoologie 184: 205-394. von Salvini-Plawen, L. 1985. Early evolution and the prim- itive groups. In: Wilbur, K. M., E. R. Trueman and M. R. Clarke (eds). The Mollusca. Volume 10. Evolution. Aca- demic Press, Inc. p. 59-150. von Salvini-Plawen, L. 1990a. Origin, phylogeny and clas- sification of the phylum Mollusca. Iberus 9:1-33. von Salvini-Plawen, L. 1990b. The status of the Caudofoveata and the Solenogastres in the mediterranean sea. Lavori della Societa Italiana di Malacologia 23: 5-30. Wilhelmi, R. W. 1944. Serological relationships between the THE NAUTILUS, Supplement 2 Mollusca and other invertebrates. Biological Bulletin 87: 96-105. Willmer, P. 1990. Invertebrate relationships—Patterns in an- imal evolution. Cambridge University Press, Cambridge. Wingstrand, K. G. 1985. On the anatomy and relationships of recent Monoplacophora. Galathea Reports 16:7-94. Winnepenninckx, B., T. Backeljau, Y. Van de Peer and R. De Wachter. 1992. Structure of the small ribosomal subunit RNA of the pulmonate snail Limicolaria kambeul, and phylogenetic analysis of the Metazoa. FEBS Letters 309: 123-126. Winnepenninckx, B., T. Backeljau and R. De Wachter. 19938a. Extraction of high molecular weight DNA from molluscs. Trends in Genetics 9:407. Winnepenninckx, B., T. Backeljau and R. De Wachter. 1993b. Complete small ribosomal subunit RNA sequence of the chiton Acanthopleura japonica (Lischke, 1873) (Mollusca, Polyplacophora). Nucleic Acids Research 21:1670. Woese, C. R. 1991. The use of ribosomal RNA in recon- structing evolutionary relationships among Bacteria. In: Selander, R. K., A. G. Clark and T. S. Whittam (eds). Evolution at the molecular level. Sinauer Associates, Inc., Sunderland, MA p. 1-25. Woese, C. R. and R. R. Gutell. 1989. Evidence for several higher order structural elements in ribosomal RNA. Pro- ceedings of the National Academy of Science of the USA 86:3119-3122. THE NAUTILUS, Supplement 2:111-121, 1994 Page 111 Preliminary Ribosomal RNA Phylogeny of Gastropod and Unionoidean Bivalve Mollusks Gary Rosenberg Academy of Natural Sciences 1900 Benjamin Franklin Parkway Philadelphia, PA 19103 USA George M. Davis Academy of Natural Sciences 1900 Benjamin Franklin Parkway Philadelphia, PA 19103 USA Gerald S. Kuncio Department of Medicine University of Pennsylvania Philadelphia, PA 19104 USA M. G. Harasewych Department of Invertebrates National Museum of Natural History Smithsonian Institution Washington, D.C. 20560 USA ABSTRACT Sequences of about 150 nucleotides in the D6 region of the large (28S) ribosomal molecule were obtained from 20 union- oidean bivalves and 13 gastropods, including 9 truncatellids, 1 muricid, 1 cancellariid, 1 melongenid, and 1 pleurotomariid. These were analyzed along with sequences from Emberton et al. (1990) for 8 pulmonates, a helicinid and a pomatiopsid. Rates of divergence varied by a factor of two, with unionoidean and the helicinid sequences differing by 15-20% relative to mouse, pulmonates by 22-24%, neogastropods by 24-27%, and rissooideans by 27-32%. Length variation among sequences occurred mainly in the D6 loop, and complementary mutations were seen in the D6 stem. Cladistic analysis found 24 equally parsimonious trees; the strict consensus supports monophyly of Unionoidea, Rissooidea, Pulmonata and Stylommatophora. Monophyly of Neogastropoda is not contradicted. Some groupings are anomalous when compared to mor- phology-based phylogenies. Helicinidae groups with Uniono- idea, Pleurotomariidae with Neogastropoda, and Geomelania (Truncatellidae) with Pomatiopsidae. In each case, addition of taxa that intersect long branches (e.g. Chitonidae, Patellogas- tropoda) might show that characters interpreted as synapo- morphic are plesiomorphic or convergent. The observed group- ing of Muricidae and Cancellariidae is well-supported, indicating that cancellariids are a highly derived group within the Sten- oglossa. Pleurotomariidae are more closely related to the other gastropods in the analysis than are Helicinidae, supporting Ha- szprunar s (1988) anatomy-based conclusion. To date, sequence studies of mollusks have not overturned phylogenies based on morphology, but rather have helped in choosing among competing morphology-based hypotheses. Like morphological data, sequence data are subject to problems of convergence, unequal rates of evolution, and choice of taxa. Classifications must be based on all available data to maximize the potential for detecting convergences and correctly resolving phylogenetic relationships. Key words: Gastropoda, Bivalvia, 28S ribosomal RNA, phylog- eny, cladistics, rates of divergence. INTRODUCTION Ribosomal RNA (rRNA) and rDNA sequences have prov- en to be valuable and versatile sources of data for phy- logenetic inferences. The great variation in rates of evo- lution of different parts of rDNA allows evolutionary investigations from the level of population and species through kingdom, by study of appropriately variable regions. This variation has caused debate as to the reli- ability of some types of rDNA sequence data in phylo- genetic analysis, but it has become clear that, when an- alyzed with care, all parts of the sequence are potentially informative (Wheeler & Honeycutt, 1988; Swofford & Olsen, 1990; Hillis & Dixon, 1991; Dixon & Hillis, 1993). Ribosomal sequences have most often been used in de- termining relationships among bacteria (e.g., Woese & Olsen, 1986) and vertebrates (e.g., Hedges et al., 1990), but can be used with any organism (e.g., Sogin et al., 1986; Field et al., 1988). Of the more than 150 phylogenetic studies of rDNA sequences published to date (Hillis & Dixon, 1991), only a few have been devoted to mollusks. Ghiselin (1988) looked at molluscan origins using 18S rRNA, Emberton et al. (1990) at pulmonate relationships using D6 28S rRNA and Tillier et al. (1992) at gastropod phylogeny using D1 28S rRNA. These studies have demonstrated the potential for rDNA sequences to sharpen and resolve ideas of molluscan phylogeny, particularly at the ordinal level and above. We have supplemented the 10 gastropod sequences obtained by Emberton et al. (1990) with data from 33 more molluscan species. All 43 sequences were used in this study with the aims of 1) analyzing aspects of caen- ogastropod, archaeogastropod, and unionoidean relation- ships, 2) surveying variability in 28S rRNA sequences in mollusks, and 3) examining how choice of taxa affects Page 112 THE NAUTILUS, Supplement 2 Table 1. Localities and catalogue numbers of voucher specimens for this study. Depository is the Academy of Natural Sciences of Philadelphia (ANSP) unless otherwise noted; USNM = United States National Museum. Condition: f = frozen; | = lyophilized; w = whole live animal. Voucher lot of Anodonta imbecillis was collected in 1974; tissue sample was from same population in 1975. See Emberton et al. (1990) for vouchers of Helicina orbiculata, Oncomelania hupensis, Biomphalaria glabrata, Mesodon inflectus, Mesodon normalis, Neohelix albolabris, Triodopsis hopetonensis, Haplotrema concavum, Mesomphix latior, Ventridens cerinoi- NE of Swedesboro, Gloucester Co., New Jersey Ramah Borrow Canal, Iberville Parish, Louisiana Lake Lacawac, Lake Ariel, Wayne Co., Pennsylvania Deep Creek, Nanticoke River, Sussex Co., Delaware Pit River, SW of Canby, Modoc Co., California Kyles Ford, Clinch River, Hancock Co., Tennessee Ramah Borrow Canal, Iberville Parish, Louisiana Ouachita River, Arkadelphia, Clark Co., Arkansas Kyles Ford, Clinch River, Hancock Co., Tennessee North of Quickstep, Trelawny Parish, Jamaica deus. Species Condition Locality Anodonta cataracta ] ] Swartswood, Sussex Co., New Jersey A. grandis ] A. imbecillis ] Magnolia Springs, Jenkins Co., Georgia Amblema plicata f Bogue Chitto Creek, Dallas Co., Alabama Elliptio complanata ] Swartswood, Sussex Co., New Jersey ] ] Fusconaia cerina f Bogue Chitto Creek, Dallas Co., Alabama Gonidea angulata ] Lampsilis teres f Bogue Chitto Creek, Dallas Co., Alabama L. claibornensis f Bogue Chitto Creek, Dallas Co., Alabama Megalonaias boykiniana ] Ochlockonee River, Leon Co., Florida Obliquaria reflexa f Bogue Chitto Creek, Dallas Co., Alabama Quadrula cylindrica l Q. quadrula f Bogue Chitto Creek, Dallas Co., Alabama Plectomerus dombeyanus ] Pleurobema cordatum ] Unio pictorum f Shropshire Canal, north of Chester, near Mollington Grange, England Uniomerus tetralasmus ] Magnolia Springs, Jenkins Co., Georgia Cumberlandia monodonta l Margaritifera falcata ] Siletz River, Lincoln Co., Oregon M. margaritifera ] Locust Creek, Schuylkill River, Pennsylvania Perotrochus maureri f 90 miles east of Charleston, South Carolina Truncatella sp. WwW Harrison Point Lighthouse, Barbados T. caribaeensis Ww Bay side, Mile 57, Grassy Key, Florida Keys T. clathrus Ww Bay side, Mile 57, Grassy Key, Florida Keys T. pulchella Ww Bay side, Mile 57, Grassy Key, Florida Keys w Falmouth, Trelawny Parish, Jamaica T. reclusa Ww Cumaca, Northern Range, Trinidad T. scalaris Ww Falmouth, Trelawny Parish, Jamaica T. subcylindrica w The Fleet, Dorset, England Geomelania sp. Ww G. typica Ww Wallingford, St. Elizabeth Parish, Jamaica Busycon carica f Cape Henlopen, Sussex Co., Delaware Mancinella deltoidea f South Beach, Miami, Dade Co., Florida Progabbia cooperi f Off La Jolla, San Diego Co., California Catalogue no. 333526, 341937 334429, 341946 341888 333563 373820, A12742 334428 339430 339340 397248 339965 373821, A12744 397249 346111 397247, A12722 335041 397246, A12728 vouchers not kept 340629 350622 353138 341956 339339 334867 USNM 875218 397286 397275 397278 397274 397264 397285 397263 397280 397283 397284 USNM 847010 USNM 870850 USNM 846054 phylogenetic inference. The results presented here must be considered preliminary until data for longer sequenc- es and additional ordinal level taxa are available. MATERIALS AND METHODS We obtained sequences from 20 unionoidean bivalves and 13 gastropods, including 9 truncatellids, 1 muricid, 1 cancellariid, 1 melongenid, and 1 pleurotomariid. Spe- cies names, localities, voucher information, and higher classification are given in Tables 1 and 2. Truncatellid RNA was obtained by homogenizing live animals, other gastropod RNA from frozen tissue, and unionoidean RNA from lyophilized or frozen tissue. Methods for sequenc- ing followed Emberton et al. (1990) and were done in the same laboratory, using the same primer for the D6 region, complementary to nucleotides 2099 through 2118 for mouse as published by Hassouna et al. (1984). Each species was sequenced at least twice, or more often as necessary to resolve ambiguities in nucleotide identity. Sequence alignment: Gross alignment of the sequences was easily achieved because of large conservative stretch- es in the D6 flanks. The MALIGN program of Wheeler and Gladstein (version 1.73, 1993), was used to refine the manual alignment. The following weights (costs), with options alignaddswap and treeaddswap, yielded align- ments that matched overall the manual alignments of the D6 flanks, while providing improvement in details: G. Rosenberg et al., 1994 Page 113 transitions 1, transversions 3; internal gaps 10; leading gaps 5; trailing gaps 5. Phylogenetic analysis: Informative and variable nucle- otide positions, indicated by “i” or “v” in Figure 1, were analyzed using Hennig86. The sequence of commands “mhennig; bb; ie*;” was used, which guarantees finding all of the most parsimonious trees. All characters were equally weighted and unordered (command “‘cc-.”). Those gaps marked with a hyphen (-) in Figure 1 were scored as characters, except for the deletion from posi- tions 74 to 79 in Truncatella clathrus. Species that showed no differences in sequence were combined for the pur- pose of the phylogenetic analysis. RESULTS Aligned sequences are shown in Figure 1. The 5’ flanking region (positions 1-46), and the 3’ flank (positions 99- 161) are conservative, and only a few gaps were inserted to align the sequences. The D6 loop shows considerable variation in length, and were too variable in the non- pulmonate gastropods to be reliably aligned. A number of complementary changes can be seen in the stem re- gion, positions 47-55 and 90-98( Figure 2). Among the 48 species, 26 different sequences were found. All species with identical sequences were confam- ilial. As reported by Emberton et al. (1990), sequences were invariant among the four polygyrids. Among 20 unionoideans, only 6 different sequences were found. Cumberlandia, Gonidea and the two Margaritifera all had distinct sequences. The other 16 unionids differed from each other by at most one nucleotide, falling into two groups, referred to here as the Anodonta and Am- blema groups. In contrast, sequences differed strongly among truncatellids: each of the nine species had a unique sequence. Within the genus Truncatella, all species dif- fered from each other by at least five nucleotides. Dif- ferences were concentrated in the D6 loop, and involved significant variation in length, in addition to nucleotide substitution. Only a partial sequence was obtained for Perotrochus. Molluscan sequences differed from those of mouse at 15 to 32 percent of the sites (Table 3). Sequences from the unionoideans and Helicina were the most conser- vative, differing from mouse at 15 to 20 percent of sites. Gastropods, excluding Helicina, differed from mouse at 22 to 32 percent of sites, with pulmonates showing less sequence divergence (22 to 24 percent) than neogastro- pods (24 to 27 percent) and rissooideans (27 to 32 per- cent). é Use of published sequences (Gutell & Fox, 1988) from Mus, Rattus, Xenopus or Homo as outgroup did not affect polarization of characters. Sequences for Caenor- habditis, Physarum and Saccharomyces were more di- vergent from molluscan sequences than were vertebrate sequences, and in some regions could not be aligned satisfactorily with them. They were therefore judged less appropriate as outgroups. The sequence from Drosophila (Tautz et al., 1988), shown in Figure 1, could be aligned, Table 2. Higher classification of genera for which sequence data were analyzed. Classification follows Davis and Fuller (1981) for Unionoidea, Haszprunar (1988b) for Gastropoda, Rosenberg (1989) for Rissooidea, Kantor and Harasewych (1992) for Neogastropoda and Emberton et al. (1990) for Pulmonata. Bivalvia Neogastropoda Paleoheterodonta Stenoglossa Unionoidea Muricoidea Unionidae Muricidae Unioninae Mancinella Unio Melongenidae Ambleminae Busycon Amblema Cancellarioidea Megalonaias Progabbia Plectomerus Pulmonata Quadrula Basommatophora Pleurobemini Planorboidea Elliptio Planorbidae Fusconaia Biomphalaria Pleurobema Stylommatophora Uniomerus Holopoda Gonideini Polygyroidea Gonidea Polygyridae Lampsilini Mesodon Lampsilis Neohelix Obliquaria Triodopsis Anodontinae Holopodopes Anodonta Rhytidoidea Margaritiferinae Haplotrematidae Margaritifera Haplotrema Cumberlandia Aulacopoda Gastropoda Zonitoidea Neritopsina Zonitidae Neritoidea Mesomphix Helicinidae Ventridens Helicina Vetigastropoda Pleurotomarioidea Pleurotomariidae Perotrochus Caenogastropoda Neotaenioglossa Rissooidea Pomatiopsidae Oncomelania Truncatellidae Truncatellinae Truncatella Geomelaniinae Geomelania but differs from the molluscan sequences at twice as many sites as does the mouse sequence. Of 148 nucleotide positions scored in bivalves, 35 (24%) are variable relative to mouse, whereas 72 (49%) are variable relative to Dro- sophila. This degree of divergence made Drosophila un- reliable as an outgroup. Out of 153 alignable sites, 73 (48%) are variable in mollusks relative to mouse and 55 are potentially infor- mative for cladistic analysis. With mouse as the outgroup, cladistic analysis yielded 24 equally parsimonious trees, the strict consensus tree of which is shown in Figure 3. THE NAUTILUS, Supplement 2 Page 114 ‘(% GInBIy eas) FJasyt YM sired aseq pure yorq SPO} e[Noajour [eWMOsoq!1 sy} SITY “WI9}S ,G 9Y} WUN[ep sour] [LONIEA sayy, ° 1, peyxIeUI oe sisAyeuR ONSIpK[O 10J aAtyeuLIOJUT AT[eryUA}0d asoyy SA. peyxreur oie sysnjjour ut a[qetea suontsod ‘souenbas asnour ay] eAoqy ‘sauo UkOsN][OUI UY} 1eBU0] o1e saousnbes asoy} a194M sIsdifja ue are (W10}}0q ye) DjVYdoso1g puR ssnouL Jo doo] 9q 24) ul (+) sasniq °,.N,, PexeUI a1e pai0os aq jou p[nos yeyy sapyoaponyNy ‘(-) ysep eB Aq sdes ‘(*) jop e& Aq paquasesdes ore asnow wi0Iy pesueyoun sapnospony (Pg6l) ]Y 72 eunosseyy JO 9E8T UoNIsod ssnour 0} spuodsai109 ssouanbeas ay) Jo pua ,G sy], ‘UoNIsod apryoejonu ay} 0} Jezor syo[diay [eons9A ur do} ay) sso10¥ sraquinyy ‘ssouenbes pousiyy *[ enay NNVIISNISWISNVISNIIOVWNSISSNANINIVVSVNIISNYNISNVVS - VNDSVVISWNSY | ONOVNNNIV | VWONNN-NVVWNWNVONSNN++++++++V9-WOV | NOVWWWNIID | 9-YANVIWN-wNIDWNYNONNSW-wANINDNONVNONvVVVWNNIII 8) 1Yudosoig ¥I9999999V99N9I9999I999IIIIINWVNIISVWINNIISVINIIIONIIIIIISVISWNSY | 99V999I999 | IWNID9-V999999999999++++++++99-NIV | 999999N9I | 9-99999WN-VIII999NINIIOW-99NINIVINVIINVWVV9NIID ASNOW seers yeenyet eee e steepest esse eggs sc test eee ees | sceqyeres lyen Dy YY-*")" WYO yO ONINR oT RO OWNN octet ccs ccsscsseeeeecee suaplJqua, SODGOOOO DO) O00) O00) OOGOOOO. G0 OGbOGON fh) DobOtotGomO onl Inkl. SOtnl iy “WoWW-" nN? -Wony trey legge | eg et OMNNG eye ctes reese x Lyduosay A pag etegye sly “W°WW-" "Nn? -Wonv Pe OTATTNINE OT eC UNNG ety cesses ey eulasj30)dey wren eenyeniyssge etsy gg sss ssss | se egyens |g “D"WW-" "NN" -VOND seleyttes seca le leer ee eee yene cece a yeee ence eee ce cece eens sisdopotd Noccoatcawenty' "9° "g 7p 9°9N" soya 19 "9" WW-" "NN" -WOND seleye[ee PONT cero nts t sess see cen eee XL] 94OaN 9 non 9 “9° 5 . sees] el legecee ls ress leyeees ND AUNT pone yeas: WV-""/n*-von9 seleyedee "WIN ** st]euluou Uoposay --WW-"9N"-n-"¥ seleye |e weeds leer es legqeeyent eae icyrc st sc tects ec esees elueyeyduorg BXSIN Nicmmeecbetesele marcy c!atsig sic kyjeichensisininnyeygyisic/gyisieie/=)sisixyolsiNwjnieiene.e/eie sels) “"OWW""" --Nv-° “OnvnvoNNony 99-9" -"9)°nnND"" De Ae OGO/ BOGUT TOD OO CO GOO OOOO ORO COR On0DO0 etqqeBbold yt ct ences seeeyergee sess sgggessss se segeenye esses sl soya: “<"NN-"DD9WM"ONOINNN DD-DV°|"CD°ANOT TT -NCOT MC CWNND ac e] )auLoueW DOG p)SOeeo eee “"OVW'N'” -"N°""-"WO-"N9" 99-"n* "-"9°nNO"” “AW KH WNNE To -Wron oe. uosAsng regepesrstesaeee: "= *yyone | -cwoo*-cwW-neat-n "p7-tye | "99nn °° -W"°9° "MT -“ANTWOONN ON’ oe e Lue ]auioou9 seers sees pee este tees cece meng | wes = 9nATAS -Y *-9y°]"99nna"* | °-W°9° "At -"ANT WOON os eoidA3 *9 seer tet eege escent testers lem yan | yet = 9-AAS WY s*-9y'|-"99nno°* | t-WroccAt-cAN WOO ee os “ds e1uejauoay rregegrrctssegetsscss sect esse | ee ygns =| sys s*-"¥9-99NWNWII"DVDDN WW-2°"}""99MNI "| *-WOT A -TAN WOOT eee seests* sisuaseqiues “1 Dg) COPO Soong poco onooosnogar "-9yVON'” -"W°W"-"W"-N"99 "WWW99999"V "9-99 "*99NND"* OSPR) POMP Shine AR eer es o00 DDS oG000R0000000000 e891 4pul }Aoqns a "979" ““OWWON"* | -"W"o""-"WO-Ww"SWWW""9NNID = ND - “*99NND"* |} *-W° 9°" N"- "NN WI9" =" ** esn}Ie4 “1 2°9 “"OWWON" =| -"W"**-"WW-"9V999°9NINNIININ"N- 2) ]949)nd “1 We gge sess sssssceceee le eounge ss | tyes? -"W9-99NNNDD°999-99 -°9- | "*9DAMOT | -NCD “AT -ANT WOON ON ONO-NS Se siue]eos "1 yee cesses ssc s cece ce ees | eoniygers | yess yy 8 ne 99°99999N *D-"W | 9ONMOT TT-NTDO AM -CAN WDD ne snayje)2 “1 ee) 999 teen |e llegeree see leyecee } 9° o°n een AY ATW "9° Soon aaa Ea) -oyeAT: WW-" "a" -von9 reeeye [reggie | seecgees: "WN C0 sn}92)JUL “WH o7n o7n SEDO DOOOOO 077.) 0000. 20) OO OOO0) 0) OO COoOo) Getto Oooo cocoon ISton ir on he) “wd “wey wy “wey Wey “WV “Wwe “yo" 2 . Li a i } ° 22 E) > = = = i) aqaqaqaaaacatacc <