Assessing diversification patterns in an ancient tropical lake: Gomphocythere (Ostracoda) in Lake Tanganyika.

Item Type text; Dissertation-Reproduction (electronic)

Authors Park, Lisa Ellyn.

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 04/10/2021 07:10:10

Link to Item http://hdl.handle.net/10150/187316 INFORMATION TO USERS

This manuscript pas been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

'!be quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely. event that the author did not send UMI a complete mannscript and there are missing pages, these will be noted. A1so, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

UMI A Bell & Howellinformalion Company 300 North Zeeb Road. Ann Arbor. MI48106-1346 USA 313/761-4700 800:521-0600

ASSESSING DIVERSIFICATION PATTERNS IN AN ANCIENT TROPICAL LAKE: GOMPHOCYTHERE (OSTRACODA) IN LAKE TANGANYIKA

by

Lisa Ellyn Park

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF GEOSCIENCES

In Partial Fulfillment of the Requirements For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

1995 OMI Number: 9620378

UMI Microfonn 9620378 Copyright 1996, by UMI Company. All rights reserved.

This microfonn edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 2

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have read the dissertation prepared bY ____~T~,iws~a~E~1~1~y~n~P~a1r~k~ ______entitled Assessing Diversification Patterns in an Ancient Tropical

Take' Gomphocythere (Ostracoda) in T.a ke Tanganyika

and recommend that it be accepted as fulfilling the dissertation

Doctor of philosophy re:Zflccze of ? !/lj /1'5 And::.,ew Co~ ~ L-U ~.c.. Karl Flessa Date ~ Owen Davis Date o~/qqo- Date S S"fll 11f5 ~~Wayn addison Date I •

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I have read this dissertation prepared under my that it be accepted as fulfilling the dissertation

,fiR 1'-1 (/,'is Andrew Cohen Da 3

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author. 4

ACKNOWLEDGMENTS

I wish to thank and acknowledge the following people, without whom, this project would not have been possible. Dr. Andy Cohen (University of Arizona) served as my graduate advisor and provided guidance, input and his expertise in this project. Dr. Karel Wouters and Dr. Koen Martens of the Koninklijk Belgisch Instituut voor N atuurwetenschappen provided access to materials, technical support and lent their taxonomic expertise while I was working at the museum. Dr. Boudwin Goddeeries (also KBIN) allowed me generous use of his microscopes and taught me much about art during my visits to Belgium. Drs. Karl Flessa, Owen Davis and Wayne and David Maddison (University of Arizona), served as members of my dissertation committee, and were very helpful and understanding in this process. On a personal note, I wish to thank my parents, who have always supported me in whatever I do. Thanks especially to my father, Mr. Richard Park for printing my SEM photographs. I thank Dr. Allan Saaf, Heartland Community College for being such a kind and understanding boss, as well as Mrs. Patricia Bozarth, Mrs. Stephanie Johnson, Mrs. Janet Beach Davis and Dr. Edith Wallace, also of Heartland Community College, who provided clerical support. I would like to thank Ms. Kathleen Nicoll and Mr. and Mrs. John Downing for providing an intermittent place for me to stay and work while completing this project. Ms. Lois Roe and Mr. Kevin Moodie provided critical logistical support in Tucson in the final days of preparing this manuscript. And finally, I wish to thank Dr. Kevin Downing (DePaul University) for his unending support, insight, love and friendship. This project was supported by student research grants from the following: The Sulzer Fund, Geological Society of America, University of Arizona Student Research Fund, Chevron, and the University of Arizona Analysis of Biological Diversification Research and Training Grant. 5

This dissertation is dedicated with love to my sister Cynthia, who left us too early, but gave us so much. Your spirit remains with me always. 6

TABLE OF CONTENTS

I. LIST OF FIGURES ...... 9 II. LIST OF TABLES ...... 11

III. ABSTRACT ...... 12

IV. CHAPTER 1: PHYLOGENETIC RELATIONSHIPS OF GOMPHOCYTHERE(CRUSTACEA:OSTRACODA) IN LAKE TANGANYIKA, EAST AFRICA ...... 14 Introduction ...... 14 Biological Diversification in Lake Tanganyika ...... 14 Major Groups of Ostracods...... 15 History and Definition of Gomphocythere ...... 15 Taxonomic Issues with Gomphocythere ...... 16 Global Distribution and Evolutionary Significance of Gomphocythere ...... 19 Purpose of Study...... 20 General Method ...... 21 Philosophy of Cladistic Analysis ...... 21 Materials and Methods ...... 22 Taxa and Characters ...... 22 Collection of Specimens ...... 22 Dissections ...... 25 Outgroup Selection ...... 25 Species Delimitation...... 26 Characters and Character States ...... 27 Character Definitions...... 30 Hard Parts ...... 30 Hingement ...... 30 Muscle Scar Patterns ...... 31 Brood Pouch and Alar Prolongations...... 36 Sieve Pore Canals ...... 36 Surface Ornamentation ...... 37 Nodes, Alae and Longitudinal Ridges ...... 38 Soft Parts ...... 38 A1-First Antennae ...... 39 A2-Second Antennae or Antennule ...... 39 Md-Mandibles ...... 39 Mx-Maxilla ...... 42 P1-First Thoracic Legs ...... 42 P2-Second Thoracic Legs ...... 43 P3-Third Thoracic Legs ...... 43 7

TABLE OF CONTENTS -- Continued

Hemipenes ...... 44 Cladistic Analysis ...... 44 Signal and Noise in the Data ...... 48 Comparison of Hard versus Soft Part Phylogenies as Inferences to Ostracod Biodiversity in the Fossil Record...... 51 Results ...... 52 Results of Phylogenetic Analysis ...... 52 Definition of Ordering and Multistate Characters ...... 52 Definition and Analysis of Phylogenetic Structure...... 59 Favored Phylogeny...... 64 New Gomphocythere Species Recognized ...... 67 Subclade Synapomorphies ...... 70 Species Clustering...... 70 Hard Part Character Distributions ...... 75 Soft Part Character Distributions...... 76 Results for Hard versus Soft Part Phylogeny Reconstructions...... 77 Hard Part Reconstructions...... 77 Soft Part Reconstructions...... 81 Discussion ...... 84 New Species ...... 84 Phylogeny of Gomphocythere ...... 84 Homoplasy and Speciation Patterns within the Phylogenetic Reconstruction...... 86 Homoplasy Model for Multiple Invasions-Radiations ...... 87 Hard Parts versus Soft Parts ...... 96 Conclusions...... 97

V. CHAPTER 2: ECOLOGY AND SPECIATION OF THE OSTRACOD CLADE, GOMPHOCYTHERE, IN LAKE TANGANYIKA ...... 99 Introduction...... 99 Lake Tanganyika Speciation Patterns ...... 99 Lake Tanganyikan Environments...... 99 Problem to be Addressed ...... 100 Rationale, Significance and Need for Study ...... 100 Theoretical Framework for the Proposed Study ...... 103 Previous Work on Gomphocythere Ecology and Speciation ...... 104 Dispersal Strategies ...... 107 Geographic and Habitat Distribution Hypotheses ...... 108 Ecological Character Correlation Hypotheses ...... 109 Methods ...... 110 General Phylogenetic Method ...... 110 8

TABLE OF CONTENTS -- Continued

Geographic Location Data Analysis and Methods ...... 110 Interlake Distributional Patterns ...... 110 Intralake Distributional Patterns ...... 113 Ecological Distribution Data ...... 113 Results ...... 116 Biogeographic Distributions of Gomphocythere ...... 116 Interlake Distribution...... 116 Endemic Tanganyikan Gomphocythere ...... 119 Intralake Distribution...... 119 Results of Ecological Character Mapping ...... 120 Substrate Distributions...... 120 Substrates Mapped onto Phylogenetic Tree ...... 145 Depth Distributions ...... 148 Body Size Distribution...... 154 Ecologically Correlated Morphologies...... 154 Discussion...... 162 Gomphocythere Distributions ...... 162 Speciation Barriers...... 163 Substrates ...... 164 Depth ...... 165 Ecologically Correlated Morphologies ...... 165 Speciation Mechanisms of Gomphocythere in Lake Tanganyika...... 166 Single Invasion Scenario ...... 167 Multiple Invasion Scenarios ...... 167 Immigration ...... 167 Fluctuation ...... 168 Implications for the Fossil Record...... 169 Tempo and Mode of Gomphocythere Speciation ...... 169 Conclusions ...... 171 VI. APPENDIX A: PHYLOGENETIC CHARACTERS AND CODING ...... 173 VII. APPENDIX B: CHARACTERS, CHARACTER STATES AND CHARACTER DISTRIBUTIONS OVER THE FAVORED PHYLOGENY ...... 180

VIII. APPENDIX C: TAXONOMIC SPECIES DESCRIPTIONS AND NOTES ON OTHER SPECIES ...... 251

IX. REFERENCES ...... 309 9

LIST OF FIGURES

Figure Page

1.1 Distribution map of Gomphocythere in Africa ...... 24 1.2 Schematic illustrations of hard part morphologies ...... 33

1.3 SEM photographs of hard part morphologies ...... 35

1.4 Schematic illustrations of soft part morphologies ...... 41

1.5 Schematic illustrations of soft part morphologies ...... 46

1.6 Phylogenetic trees from unordered analysis...... 54

1. 7 Favored phylogenetic reconstruction ...... 58

1.8 Phylogenetic trees from ordered analysis ...... 61

1.9 Treelength distribution for unordered analysis ...... 66

1.10 Phylogenetic tree from analysis without sexual characters ...... 69

1.11 Phylogenetic tree from hard part analysis ...... 80

1.12 Phylogenetic tree from soft part analysis ...... 83

1.13 Speciation scenario for single invasion ...... 89

1.14 Speciation scenario for multiple invasion ...... 91

1.15 Speciation scenario for multiple invasion part b...... 94

2.1 Map of Lake Tanganyika with substrate distributions ...... 102

2.2 The favored phylogeny of Gomphocythere ...... 112

2.3 Distribution map of Gomphocythere in Africa ...... 118

2.4 Distribution of endemic species mapped onto phylogeny ...... 122

2.5a Map of distribution of Gomphocythere species in Lake Tanganyika and locality information ...... 124 10

LIST OF FIGURES - Continued

2.5b Relative frequency data for Gomphocythere ...... 126

2.5c Presence/absence data for Gomphocythere species ...... 130 2.6 R-mode cluster analysis of substrate distributions...... 139

2.7 Principal components analysis for substrates ...... 142

2.8 Phylogenetic reconstruction and substrate distribution ...... 147

2.9 Phylogenetic reconstruction and depth distribution ...... 153

2.10 Distribution of Gomphocythere size on phylogeny ...... 156

2.11a Character state distribution for node number ...... 159 2.11b Character state distribution for Al terminal claw ...... 161 11

LIST OF TABLES

Table Page

1.1 Taxonomic list of species in analysis ...... 18 1.2 Number of specimens used for each species ...... 29

1.3 Phylogenetic structure skewness statistics ...... 50

1.4a List of apomorphies supporting each terminal taxon for DELTRAN optimization...... 72 l.4b List of apomorphies supporting each terminal taxon for ACCTRAN optimization...... 74

2.1 Results from One-Way General Linear Model for pooled substrates ...... 134

2.2 Substrate affinities of Gomphocythere species in Lake Tanganyika ...... 137

2.3 Eigenvalues from Principal Component Analysis of 11 variables ...... 144

2.4 Depth affinities of Gomphocythere species in Lake Tanganyika ...... 151 12

ABSTRACT I examined the distribution of 70 morphological characters for 17 extant species of an ostracod genus, Gomphocythere, in Africa, to test hypotheses concerning character development and speciation patterns. Using heuristic searches conducted with the phylogenetic reconstruction program PAUP, I found 9 trees of 269 steps (CI = 0.45). The skewness of tree length distribution reveals significant phylogenetic structure in the data. Nodes are supported by 2 to 13 character state changes, and these character changes are sometimes reversed or paralleled elsewhere, accounting for much of the homoplasy in the reconstructions. The results suggest the presence of four new species (Gomphocythere coheni, G. downingi, G. wilsoni, G. woutersi). These species are relatively derived within the favored tree. Analyses were done to elucidate the effect of the exclusion of hard and soft part characters and the effects of differential fossil preservation on phylogenetic reconstruction. Eliminating the hard part characters caused the collapse of many branches as polytomies and decreased the agreement of the hard part trees. Excluding the soft part characters increased the number of most parsimonious trees, and decreased the resolution of the trees by creating many unresolved polytomies, but produced similar islands of stability as the complete analysis. I integrated phylogenetic and ecological data sets to examine ecological variables of substrate and depth range and their role in speciation of Gomphocythel'e in Lake Tanganyika. Using one-way MANOVAs on substrate distribution data, I found significant variability 13 in the substrate distribution of (p=.OOl) seven species. In addition, principal components analysis and R-mode cluster analysis revealed close associations between certain species with respect to substrate ranges. Correlations exist between species and their ecological tolerances, indicating that environmental fluctuations could have had a profound effect on speciation. In addition, evidence for multiple invasions of Gomphocythere species exist from mapping endemism onto the phylogenetic tree. From this information, a model of speciation emerges in which there have been multiple invasions and subsequent radiations. 14

CHAPTER 1

PHYLOGENETIC RELATIONSHIPS OF GOMPHOCYTHERE (CRUSTACEA: OSTRACODA) IN LAKE TANGANYIKA, EAST AFRICA

INTRODUCTION Biological Diversification in Lake Tanganyika The East African rift lake system has long been known as an area of megadiversity (sensu Mittermeier, 1988; Mittermeier and Werner, 1990), particularly with respect to the large, endemic species flocks that originated within various lakes in this geologic and geographic setting (Coulter, 1991; Sturmbauer and Meyer, 1992; Coulter, 1994; Snoeks et aI., 1994). Over 185 cichlid fish species (180 of which are endemic), 68 gastropod species (45 of which are endemic), and 200 ostracod species (almost all of which are endemic), have been found in Lake Tanganyika alone, providing an excellent means for studying diversification processes and diversity changes of endemic organisms over time. Recent analyses of cichlid fish (Fryer and nes, 1972; Sturmbauer and Meyer, 1992) and thiarid molluscs (Cohen and Johnston, 1987; Michel et aI., 1992) show that patterns of morphological diversity are often linked to environmental differences as well as incidences of multiple invasions and subsequent radiations in the lake. Whether or not similar patterns of evolution have produced the diversity observed in more poorly studied groups, such as the ostracods that occur in these lakes, can be tested through careful analyses of their phylogeny. 15

Major Groups of Ostracods Ostracods are small, bivalved crustaceans, that are found in both marine and lacustrine environments. There are four major suborders of ostracods: Myodocopa, Cladocopa, Platycopa, and Podocopa (Hartmann and Puri, 1974). Gomphocythere, the Tanganyikan ostracod genus that I have examined, is a member of Podocopa. Like all podocopans, Gomphocythere share the following features: a straight ventral margin, a five jointed antennula (A1), an antenna that does not function as a natatory organ (as in Myodocopa), a respiratory plate (epipodite) that is always present, 3 masticatory endites, a palp on the maxilla, the second thoracic leg as a walking leg with a cleaning organ directed inwards, highly specialized, paired genital organs, no paired side eyes, no heart and blood vessels, and lack ofa frontal organ (Hartmann and Puri, 1974).

History and Definition of Gomphocythere Sars (1910) described Limnicythere (sic) obtusata from Lake Victoria. He later redescribed this taxon and designated it as the type species (Gomphocythere obtusata) of the genus Gomphocythere (Sars, 1924). The genus shows a wide range of carapace morphologies, but typically has pronounced brood pouches in females, robust carapace reticulation, the presence or absence of alae and nodes, as well as the distinct inverse lophodont hinge structure (Grekoff, 1953). Intraspecific variation of carapaces is limited, but can sometimes lead to misidentification of species. Table 1.1 is a composite listing of the 16

taxonomy of Gomphocythere to date including new Gomphocythere species as well as the outgroups included in this analysis.

Taxonomic Issues with Gomphocythere Gomphocythere species can be recognized by the inverse lophodont hinge structure, hemipenis morphology and sieve pore canal pattern which are synapomorphies for the genus. Ventral ridges and lateral crests were also thought to be typical of Gomphocythere, until Klie (1939) described Gomphocythere angusta, which has neither feature. However, DeDeckker (1981) cited the presence of these ventral ridges and lateral crests as the main difference between Gomphocythere and his newly erected Australian genus, Gomphodella. Currently, Gomphocythere Sars, Cytheridella Daday and Gomphodella DeDeckker, belong to a special tribe, Cytheridellini Danielopol and Martens, within the subfamily Limnocytherinae, created by Danielopol et al. (1990). This designation does not reflect the close resemblance that Gomphocythere and Cytheridella have to members of the subfamily Timiriaseviinae. In their present taxonomic position within the subfamily Limnocytherinae, Gomphocythere and Cytheridella must be considered cases of parallel evolution occurring within the two subfamilies: Limnocytherinae and Timiriaseviinae because of their parallel acquisition of similar character states. A broader systematic analysis must be done to test the validity of Gomphocythere and Cytheridella in Limnocytherinae and the tribe Cytheridellini. It is most likely that Gomphocythere, Gomphodella and Cytheridella will be moved 17

Table 1.1: Taxonomy of species used in this analysis, including the 4 newly designated species that are not formally described, but are designated in this dissertation as separate species for the purpose of this analysis. 18

TABLE 1.1

TAXONOMY OF SPECIES USED IN THIS ANALYSIS

INGROUP TAXA: Subclass Ostracoda LATREILLE, 1806 Order Podocopida MOLLER, 1894 Suborder Podocopa SARS, 1866 Superfamily Cytheracea BAIRD, 1850 Family Limnocytheridae KLIE, 1938 Subfamily Limnocytherinae KLIE, 1938 Tribe Cytheridellini DANIELOPOL and MARTENS, 1990 Genus Gomphocythere SARS, 1924 Gomphocythere aethiopis ROME, 1970 Gomphocythere alata ROME, 1962 Gomphocythere angulata LOWNDES, 1932 Gomphocythere angusta KLIE, 1939 Gomphocythere capensis MULLER, 1914 Gomphocythere cristata ROME, 1962 Gomphocythere curta ROME, 1962 Gomphocythere expansa MARTENS, 1993 Gomphocythere lenis ROME,1962 Gomphocythere obtusata SARS, 1910 Gomphocythere parcedilatata ROME, 1977 Gomphocythere simplex ROME, 1962 Gomphocythere n. sp. "coheni" Gomphocythere n. sp. "downingi" Gomphocythere n. sp. "wilsoni" Gomphocythere n. sp. "woutersi"

OUTGROUP TAXA: Order Podocopida MOLLER, 1894 Suborder Podocopa SARS, 1866 Superfamily Cytheracea BAIRD, 1850 Family Limnocytheridae KLIE, 1938 Subfamily Limnocytherinae KLIE, 1938 Tribe Cytheridellini DANIELOPOL and MARTENS, 1990 Genus Cytheridella DADAY, 1905 Genus Gomphodella DEDECKKER, 1981 Tribe Limnocytherini KLIE, 1938 Genus Limnocythere BRADY, 1867 Limnocythere africana KLIE, 1939 Limnocythere dadayi MARTENS, 1990 Limnocythere thomasi MARTENS, 1990 19

out of the subfamily Limnocytherinae and into the subfamily Timiriaseviinae (Martens, personal comm.), based on features such as the hinge structure. One goal of this analysis is to elucidate some of these problematic areas concerning the validity of Gomphocythere and its relationship between it and its sister group Cytheridella.

Global Distribution and Evolutionary Significance of Gomphocythere Gomphocythere is an ideal genus for documenting and understanding diversification processes, because it is a diverse genus whose individual species show specificity to various environmental conditions, such as substrate and depth. Fourteen described living species of Gomphocythere can be found today in Africa and parts of the Middle East (e.g. Israel). The sister or subgroup taxon, Cytheridella, is found in South America, North America and Australia, whereas Gomphodella is restricted entirely to Australia. The distribution of Gomphocythere in East African lakes is poorly understood. Several Gomphocythere species appear to be endemic to individual, large, inland lakes; however, there are also cosmopolitan species that can be found in many of the lakes throughout the African continent (Sars, 1924; Rome, 1962; Martens, 1990). Gomphocythere is represented in Lake Tanganyika by five described endemic species: G. alata, G. cristata, G. curta, G. lenis and G. simplex (Rome, 1962) and 4 additional, yet formally undescribed species, that are discerned here for the first time. Despite their diversity, Rome 20

(1962) is the only previous author to specifically address the taxonomy of Gomphocythere species in Lake Tanganyika. His specimens, and hence descriptions, were almost all females, except for G. alata and G. cristata. By contrast, I found both males and females of most Gomphocythere species and used sexually dimorphic characters and additional morphologic information in my analysis.

PURPOSE OF STUDY The purpose of this study is to provide the first detailed phylogenetic analysis of the evolutionary relationships of the ostracod genus Gomphocythere. The phylogeny produced provides the means for documenting character acquisitions and their evolution in a historical context. This analysis also provides the phylogenetic tramework required for testing hypotheses of Gomphocythere speciation mechanisms in the ecologically complex Lake Tanganyikan system (see Chapter 2). In addition, systematic removal of all hard part characters and then all soft part characters in separate phylogenetic analyses provides a framework for testing whether or not the selection of characters influences phylogeny reconstruction and if there are phylogenetic ally important characters that are not represented in the fossil record. Whether or not the same tree topologies result from these two different data sets is important for paleobiologists who must rely on only hard parts for their phylogenetic reconstructions, and are significant in the measurement of bias in the assessment of evolutionary relationships. 21

GENERAL METHOD Philosophy of Cladistic Analysis This paper evaluates the phylogenetic relationships of the Gomphocythere clade occurring in Lake Tanganyika and elsewhere in Africa, using parsimony methods. A phylogenetic tree represents the history of evolutionary connections through time, and serves as a hypothesis of historical relationships of character acquisition (Hennig, 1966). It is derived assuming that evolution has caused the branching patterns. A phylogenetic perspective adds a critical historical dimension to the study of evolution, offering the ability to test further hypotheses of underlying causal mechanisms within the framework of phylogenetic relationships. Such testable hypotheses include questions of whether or not speciation could be attributed to stochastic process (e.g. random drift) or more deterministic ones. In the context of Gomphocythere evolution, the latter might include questions about the role that .lake level fluctuations or geochemical changes in the lake water, may have played in the clade's radiation. In addition, questions can be asked concerning the importance of adaptation as a mechanism for explaining observed morphologies. It is critical to understand evolutionary relationships before erecting taxonomic divisions or testing speciation hypotheses. Otherwise, there is a substantial risk of creating artificial categories and patterns that do not reflect evolutionary history. 22

MATERIALS AND METHODS Taxa and Characters Collection of specimens My analysis includes 17 Gomphocythere speCIes from Lake Tanganyika and elsewhere in Africa, including the West Transvaal region (Republic of South Africa), the Ethiopian rift lakes, Lake Albert, Lake Kivu and Lake Turkana. I used all known Gomphocythere available to me from my own collections, collections of the Royal Belgian Museum of Natural History or documented in the literature. The species from Lake Tanganyika are: Gomphocythere alata, G. curta, G. cristata, G. lenis, G. simplex, and the new species Gomphocythere n. sp. IIcoheni,1I G. n. sp. IIwilsoni,1I G. n. sp. IIdowningi,1I and G. n. sp. IIwoutersi. 1I Species from outside of Lake Tanganyika include: Gomphocythere aethiopis, G. angulata, G. angusta, G. capensis., G. expansa, G. obtusata, G.ortali, and

G. parcedilatata. Specimens used in my analysis from Lake Tanganyika were taken from over 350 samples collected on many separate expeditions, including the Belgian Hydrobiological expedition of 1946-47, and the expeditions of Cohen, 1986, 1989, Martens, 1990, Cohen, 1992, and Martens, 1992 (Figure 1.1). Specimens from the 1990; 1991, and the two 1992 expeditions were sorted and stored in alcohol until dissections were made. Outgroup material from other parts of Africa was from the Royal Belgian Natural History Museum ostracod collections and descriptions in the literature. 23

Figure 1.1: Distribution map of Gomphocythere species in Africa. The numbers on the diagram represent the following species: l=Gomphocythere aethiopis; 2=G. alata; 3=G. angulata; 4=G. angusta; 5=G. capensis; 6=G. cristata; 7=G. curta; 8=G. expansa; 9=G. lenis; 10=G. obtusata; 11=G. ortali; 12= G. parcedilatata, 13=G. simplex; 14=G. n.sp. "coheni"; 15=G. n.sp. "downingi"; 16=G. n.sp. "wilsoni"; 17=G. n.sp. "woutersi." Scale is indicated. 24

Distribution of Gomphocythere

1000 Ian

Figure 1.1 25

Dissections Dissections of all specimens were made using a Leica Wild M-10 binocular dissecting microscope and a Leica Wild polarizing light microscope. Soft parts were dissected in glycerin and sealed on a glass slide. Valves were stored dry. All specimens were curated and given Royal Belgian Museum numbers, except for material from the Cohen 1992, NURC expedition, which were curated and given University of Arizona Laboratory of Paleontology numbers. Camera lucid a drawings were made and photographs were taken of all soft part material as well as carapaces. Additional qualitative character analysis was accomplished by comparing Scanning Electron Microscopy photographs at a variety of magnifications that were taken of all Gomphocythere and outgroup species in this analysis.

Outgroup Selection An outgroup is defined as any group used in a phylogenetic analysis that is not included in the taxon under study. It is used to provide a broader phylogenetic context and thus aid in determining the root of the ingroup or ancestral states (Farris, 1972, 1982; Watrous and Wheeler, 1981; Maddison et al., 1984; Maddison and Maddison, 1992). Outgroup comparison requires a priori identification of the closest relatives to the ingroup (Wiley et al., 1991). By including many sister groups that are considered close taxonomically, it is possible to do outgroup analysis to determine character polarity (Maddison et al., 1984). Cytheridella species are similar to Gomphocythere, and currently placed in the same subfamily 26

by Martens (In Prep.). They occur in North America, South America and Australia (Purper, 1974). Gomphodella, another group with similar characteristics, occurs in Australia (DeDeckker, 1981). Species of Cytheridella were included as outgroups in my analysis, but Gomphodella was excluded because of lack of available material or adequate descriptions in the literature. In addition to Cytheridella, 3 species of Limnocythere were chosen because they are in the same family (Limnocytheridae) as Gomphocythere.

Species Delimitation Species have always been regarded as a basic unit of nature, systematics, taxonomy and evolution, but whether or not species are objective, real entities or just human-made constructs has been debated for centuries (Mayr, 1969 and 1984). Our concept of a species must be separate from the process of speciation and from the methodology by which a species is recognized. Confusion of concept, process, and methodology has engendered much of the debate about species. In this analysis, I designated species by separating specific morphologies and creating a IIspecies hypothesisll which I tested by character coding individual hypothetical species separately, using more than one individual in making character state assignments. In this process, individuals grouped into like kinds (with similar homologous characters) were collapsed into larger groups with similar morphologies of homologous characters. Any differences were examined in the context of the entire sample of ostracod specimens. Once specific groups of specimens were 27 delineated, they were compared with the type material, if available. Problems arose with Gomphocythere lenis and G. simplex, first described in Rome (1962). These species' type materials curated in the collections of the Royal Belgian Museum of Natural History, lack hard parts because they decalcified in the storage medium used by Rome. However, line drawings of these specimens are figured in Rome's (1962) treatise on Lake Tanganyika ostracods and were used for comparative purposes here. Although it would have been more robust to code each individual specimen as a terminal taxon and delineate species by doing a phylogenetic analysis, time did not permit me to do this. The number of specimens available for coding hard and soft parts in this analysis are summarized in Table 1.2.

Characters and Character States Characters are defined based upon homologous structures and are coded as a numeric or alphabetic symbol that represents a particular character state (Wagner, 1989; Pogue and Mickevich, 1990). I defined and coded 70 characters as unordered for all 21 taxa in my analysis. Similar numbers of hard part characters (34) as well as soft part characters (36) were identified for the composite analysis, using male and female data sets separately. The complete matrix showing the presence and absence of characters and multistate values is provided in Appendix A. Multistate characters (25) were then coded as ordered if they had successively additive states. 28

Table 1.2: Number of specimens used in the acquisition of character state information for all phylogenetic analyses. Hard and soft parts are designated as well as specific numbers of specimens for each sex. 29

Table 1.2

Number of Specimens Species Hard Parts SoftParts

G. aethiopis 2 1 2 1 G. alata 121 91 1 5 G. angulata 2 1 2 2 G. angusta 1 1 1 0 G. capensis 8 4 7 4 G. cristata 14 17 5 7 G. curta 28 23 1 6 G. expansa 1 1 1 1 G.lenis 5 4 1 1 G.obmsata 2 1 1 1 G.ortali 4 4 4 4 G. parcedilatata 2 1 1 1 G. simplex 3 2 0 1 G. n.sp. "coheni" 16 23 1 3 G. n.sp. "downingi" 87 52 3 5 G. n.sp. "wilsoni" 42 69 2 1 G. n.sp. "woutersi" 2 2 1 1 30

Character Definitions Certain morphological characters of Gomphocythere are easily observed or measured and are highly distinct between species, and therefore have been thought critical for diagnosis. These include the hinge structure, muscle scar, brood pouch in the female, pore canals, and surface ornamentation, including longitudinal ridges and alae. In addition, soft part anatomical features such as number of segments in the

limbs~ position of setae on the first thoracic appendage (PI), and hemipenes hinge structure, have been used by taxonomists for grouping Gomphocythere into a single genus (Colin and Danielopol, 1978). Some of these characters are more significant and stable than others, as the analysis below will demonstrate. A complete set of figures depicting all characters and character states defined below as well as their distribution on the favored phylogeny is given in Appendix B.

Hard Parts Hingement Valves of an ostracod are joined at their dorsal regions by a hinge mechanism. The hinge plays an important role in ostracod taxonomy because it is the fundamental defining morphological element that controls all other valve features (Moore, 1961; Van Morkhoven, 1962; Hartmann and Puri, 1974). There are 3 basic hinge types: 1) the adont type, which is made of a single element in each valve, 2) the merodont type, which is a compound hinge, where the median element is not subdivided and the teeth are present only in the smaller valve and 3) the 31

amphidont type which is also a compound hinge, where the median element is subdivided into an anteromedian and a posteromedian element and the teeth are present in both valves (Figure 1.2-1.3). (Van Morkhoven, 1962). Gomphocythere has an inverse lophodont hinge as described by Martens (1990), which is considered a subtype of the merodont hinge style. In general, Gomphocythere species have simple anterior and posterior cardinal teeth in one valve that fit into corresponding sockets of the opposite valve (Figure 1.2). In the outgroups, Limnocythere, and Cytheridella, the hinge mechanisms range from lophodont, adont, and amphidont (Colin and Danielopol, 1978). Previous researchers have placed taxonomic weight on the inverse lophodont hinge structure. However, the reversal of hinge structure is not uncommon in many ostracod genera, particularly in those groups closely related to Gomphocythere. For example, Cytherissa, Cypridopsis and Cypridea all have documented cases of hinge reversal. According to Martin (1940) and Van Morkhoven (1962), the reversal of the valve size, and hence the hinge structure, is not enough by itself for establishing separate genera. Both recommended using this character for discriminating species only.

Muscle Scar Patterns The central (or adductor) muscle scars are the attachment locations of the closing muscles which run from valve to valve (Figures 1.2 and 1.3f). The number and arrangement of the individual scars are considered to be 32

Figure 1.2: Schematic drawings of hard part morphologies of Gomphocythere ostracods. Anterior, posterior, dorsal and ventral margins are labeled. Diagram A represents the outer carapace morphologies, while diagram B represents the inner carapace morphologies. Scale is indicated. 33

( Dorsal Margin)

Sieve pores Spines

( Ventral Margin) Posterior ventral flare Alae

A 400 Ilm Exterior Left Valve

Teeth

400 Ilm B Interior Left Valve

FIGURE 1.2 34

Figure 1.3: SEM photo micro graphs of selected hard part morphological features. A-anterior tooth of inverse lophodont hinge structure, G. alata; scale bar is 10 J.lm; B-posterior tooth of inverse lophodont hinge structure, G. alata, scale bar is 10 J.lm. Both teeth are located on the left valve. C­ sockets of inverse lophodont hinge structure, G. alata; scale bar is 100 J.lm; D-muscle scar pattern on interior part of valve, G. n.sp. "coheni", scale bar is 10 J.lm; E-alar prolongation, G. alata; scale bar is 100 J.lm. F-sieve pore on external surface of valve, G. n.sp. "coheni", scale bar is 10 J.lm. G­ tubercles on anterior margin of valve, G. n.sp. "coheni", scale bar is 100 J.lm; H-nodes present on carapace surface, G. n.sp. "coheni'" scale bar is 100 J.lm; G. n.sp. "coheni" is described in Appendix C of this dissertation. 35 36 the most important elements to this character (Moore, 1961). Adductor muscle scar patterns are critical for diagnosing ostracod families. Within a family, they can be used in delineating genera and even species, because they are morphologically stable and readily identifiable.

Brood Pouch and Alar Prolongations Gomphocythere often have ventral swellings, alar prolongations, and hollow tubercles. Unlike brood pouches found in Paleozoic ostracods, brood chambers in post-Paleozoic ostracods are not as conspicuous from the surrounding valve surface. The pouches cause the carapace to exhibit an inverted, heart-shaped form in transverse vertical section. Presence or absence of a brood pouch was coded as a character in this analysis because it is morphologically prominent. A brood pouch is present in all Gomphocythere, whereas in the outgroup, it is absent. Alar prolongations appear as wing-shaped lateral extensions, causing a distinct arrow-shape in dorsal view and a triangular outline in transverse vertical section. Alar prolongations always point backwards and can be modified into different forms (Figures 1.2-1.3c).

Sieve Pore Canals Sieve pores are sensory organs similar to the mechanoreceptors of insects. They serve to connect the inside of the shell with the outside environment. Their presence or absence are features commonly used by taxonomists. Sieve pores are larger than open pores. Larger sized pores are usually sieve pores rather than open pores, and are either rounded or 37

sub-rounded (Figure 1.2 and 1.3h). At present, the absence of open pores and the presence of sieve pores is a characteristic feature of the tribe Cytheridellini, into which Gomphocythere has been placed by Danielopol et al. (1990). The morphology of pore systems and their position on the carapace is thought to have a close relationship with the habitat and the behavior of ostracods (Tsukagoshi, 1990). The sieve pores were coded as: presenUabsent as well as penetrative/non-penetrative. All Gomphocythere have penetrative sieve pores in contrast to the outgroups where they are uniformly non-penetrative. Unfortunately, little previous work has been undertaken to determine the utility of sieve pores as species-defining morphological features in ostracods (Rosenfeld and Vesper, 1977).

Surface Ornamentation The surfaces of the lateral outer margins of the ostracod valves are either smooth or ornamented with pits, punctuations, spines, striae, ridges or costae and reticulation (Figure 1.3e). These features are important for strengthening and armoring the carapace. Ornamentation on Gomphocythere valves typically does not differ between male and female of the same species, but is not necessarily a morphologically stable feature and may show clinal variation in dispersed populations (Carbonnel and Jacobzone, 1977; Okada, 1982). While using ornamentation as a character, a distinction should be made between primary and secondary features. Primary ornamentation in this analysis includes the major morphological features which often encompass 38 structural elements. Secondary ornamentation characteristics include the more superficial details such as small nodes and ridges (Figure 1.2-1.3e).

Nodes. Alae and Longitudinal Ridges A node is a bulbous protrusion found on the anterior or posterior regions of the valve surface. Nodes are considered modified flanges, keel­ like thickenings of the outer calcareous lamella (Moore, 1961; Van Morkhoven, 1962; Hartmann and Puri, 1974). Flanges play an important role in closure of the valves and provide support for the selvage. Alae or alar prolongations are conspicuous, wing-shaped lateral extensions found on the valve surface resulting from the duplicature of the outer lamella. These structures do not affect the inner lamella and serve to support the animal in soft substrates. Alae always point posteriorly and can help in orienting isolated valves. Longitudinal ridges or extensions are similar to alae but traverse the entire lateral valve width. Nodes, alae and longitudinal ridges are similar to surface ornamentation in their variability, but their presence or absence on ostracod carapaces is quite stable intraspecifically (Figure 1.2-1.3e). In Gomphocythere, these characters are important as synapomorphies for the various subclades within the group.

Soft Parts Several serial characters, such as segments, joints on the dorsal spine, terminal claws, setae on the second joint, setae on inner joints were 39 all coded with respect to their numbers and where applicable, their position. A schematic diagram of soft parts is given in Figure 1.4.

AI-First Antennae The cephalon is separated from the thorax by a slight constriction of the body and has 4 appendages attached: the anntennule (AI), antenna (A2), mandible (Md) and maxilla (Mx). The first cephalic appendages are important organs for swimming and feeling. Numerous hairs or setae occur on these structures that vary according to species. In this analysis, characters were coded for the number of claws present, the number of segments and the presence/absence and number of spines on specific segments (Figure 1.4).

A2-Second Antennae or Antennule The second pair of cephalic appendages are the most important locomotory appendages. They are joined to the forehead near its junction with the upper lip, and they are entirely encased within the shell when it is fully closed. Like the first antennae, the presence of long natatory setae on the second antennae are diagnostic at the species level (Figure 1.4). In Podocopa, the second antenna consists of a protopodite (2 fused podomeres) and an endopodite (3 podomeres) (Hartmann and Puri, 1974).

Md-Mandibles The mandibles are the third pair of cephalic appendages, lying at the sides of the mouth opening. Their singular function is mastication. 40

Figure 1.4: Schematic drawings of dissected soft part morphologies of Gomphocythere ostracods. A. AI-First antenna; B. A2-Second antenna; C. Mx-Maxilla; D. Md-Mandible; E. PI-First thoracic leg; F. P2-Second thoracic leg; G. P3-Third thoracic leg. Scale = 100 !lm. 41

100~

Figure 1.4 42

They retain the full suite of crustacean appendages, having both endopodite and exopodite parts, in addition to the protopodite (Figure 1.4). There are at least two different types of mandibular palps. The first type is a 'straight' type with normal setae and the last three segments subequal. The second type is an 'knee-like' type, with bifurcated setae and two penultimate segments fused together. A last, minute segment, is inserted perpendicularly on the basal segment (Van Morkhoven, 1962).

Mx-Maxilla The fourth pair of cephalic appendages lie behind the mandibles and on the sides of the hypostome (Figure 1.4). The maxillae function as supporting organs of the mandibles, carrying food to the mouth and creating water currents used in respiration (Moore, 1961). All Cytheridellini have reduced maxillular palps with fused segments and reduced number of setae. Although Sars (1924) illustrated a normal palp for Gomphocythere obtusata, Martens (1993) investigated several Mrican populations of this species and always found reduced palps, from which, he deduced that Sars' illustration was inaccurate.

PI-First Thoracic Legs The structure and shape of the first thoracic legs (PI) differ radically in various groups of ostracods, depending on the purpose that they serve. In Gomphocythere, the PI appendages serve as locomotory organs, primarily walking legs. In all Cytheracea, the endopodite on the PI is well-developed and segmented into 3 distinct podomeres and is thus 43 invariable in this particular analysis (Moore, 1961; Van Morkhoven, 1962). The thoracic appendages were coded to include the number of setae, the position of the inner spines and the presence and absence of setae (Figure 1.4).

P2-Second Thoracic Legs In the suborder Podocopa, the second thoracic legs are conspicuous, uniramous appendages that are projected out by individuals as soon as their carapace opens. In all groups within Podocopa, their structure is similar, and their function is for walking, crawling or clinging to inclined surfaces (Van Morkhoven, 1962). The powerful end claw is movable and is used to support the rear part of the body during walking. There are no muscles connecting the second thoracic legs in Cytheracea with the dorsal region of the carapace, such as in the 1st and 2nd antennae and mandibles. The P2 characters were coded in much the same manner as the AI, A2, and PI (Figure 1.4).

P3-Third Thoracic Legs In the Podocopan families: Cytheridae, Bairdiidae and Darwinulidae, the third thoracic legs serve as locomotory organs (Moore, 1961). They are structurally similar to the second pair of thoracic legs and directed downward. The place of attachment of the setae and the presence, absence and number of bristles on the setae is of taxonomic and phylogenetic importance (Figure 1.4). The characters coded on the P3 44 include the presence or absence of setae on the first joint, number of setae on the second segment and the first segment.

Hemipenes The hemipenes serve the function of copulation in ostracods and are an important diagnostic feature in all male ostracods. The size of the copulatory apparatus is usually so large that it takes up most of the posterior part of the valve, making the valves longer in the male than in the female. This is the general reason for the sexual dimorphism in this group. The external copulatory organ and the testes are connected via the vas deferens. In all extant Limnocytherinae, the distal lobe (dl) forms an integral part of the peniferum and does not hinge (Figure 1.5). In both Timiriaseviinae and Cytheridellini there is a movable distal lobe, which is connected to the remainder of the peniferum with a hinge (Van Morkhoven, 1962).

Cladistic Analysis The computer program PAUP (Phylogenetic Analysis Using Parsimony, version 3.1, Swofford, 1993) was used to find the most parsimonious tree from a character set of the 70 characters listed in Appendix A. Two types of runs were undertaken using the PAUP parsimony algorithm. In the first runs, I counted any change from one state to another state as one step. In subsequent analyses, 25 characters were redefined as ordered based upon their successive, additive states. 45

Figure 1.5: Schematic drawings of dissected soft part sexual morphologies of Gomphocythere ostracods. A. Hemipenis; B. Furca; C. Sternum; Scale = 100 Jlm. 46

100 flDl

c

Figure 1.5 47

Separate analyses were then done using the dataset with the redefined characters. Exact methods, which guarantee optimal reconstructions cannot be used for 21 taxa because the data matrix is too large for the algorithm and computer capabilities. Therefore, the data matrix (Appendix A) was analyzed using the heuristic searching option. The heuristic method builds an initial tree or set of trees by stepwise addition of taxa and continuously searches for shorter and shorter trees by rearranging that tree (Swofford, 1993). I used the tree bisection and reconstruction (TBR) search option on trees reconstructed with the random addition sequence. To increase the likelihood of finding all islands of equally parsimonious trees (sensu Maddison, 1991), I included 100 random replications in each analysis. An island of equally parsimonious trees is a set of trees in which each tree is connected to every other tree in an island through a series of trees, each member of the series differing from the next by a single, minor rearrangement of branches. All characters in my initial analysis were both unpolarized and undirected following Hauser and Presch (1991) and Swofford and Maddison (1987), making the initial trees produced by PAUP unrooted. Since PAUP includes the outgroup taxa in the parsimony analysis, the assumption of monophyly of the ingroup is tested in the analysis. The trees were then rooted using outgroup analysis (Maddison, 1991). ACCTRAN (transformation that accelerates changes towards the root) and DELTRAN (transformation that delays changes away from the root) options were used to analyze the most-parsimonious reconstructions 48 by optimizing for reversals and parallelisms, respectively (sensu Swofford and Maddison, 1987). The computer program MacClade (version 3.0, Maddison and Maddison, 1992) was used to aid in the exploration of equally parsimonious character distributions within the minimal-length topology discovered by PAUP. Trees were then compared with respect to their phylogenetic structure and compared with trees produced with characters that were randomized over the taxa (sensu Archie, 1989).

Signal and Noise in the Data I measured the phylogenetic structure in the character distribution data by measuring the skewness (g1) of the distribution of tree lengths of all possible trees (Hillis, 1991; Huelsenbeck, 1991; Hillis and Huelsenbeck, 1992). I estimated gl using PAUP's random trees option which generated

105 trees according to the equiprobable tree distribution for my characters. The critical value of gl for 20 taxa and 70 binary characters (for the probability p=0.01, gl =-0.20) provided by Hillis and Huelsenbeck

(1992) offers a conservative test (Table 1.3) for the significance of phylogenetic structure. This test evaluates the structure of the data as a whole as compared to random noise. If all the variation among taxa is essentially random with respect to phylogenetic history, then there would be no basis to expect the most-parsimonious tree to be a good estimate of phylogeny. Therefore, phylogenetic signal can be determined by comparing the number of most parsimonious trees generated by the data to a randomized dataset. Measurements of phylogenetic signal were 49

Table 1.3: Values for skewness statistics for testing phylogenetic structure based on comparisons with randomly generated trees. Values for this analysis are in the lower box. 50

Table 1.3

Hillis and Huelsenbeck Estimations of Number of

Multistate Characters 51

calculated and compared with those calculated by Hillis and Huelsenbeck (1992) who also subsequently tested those calculations using phylogenetic simulations, demonstrating the close relationship between these measures and the effectiveness of parsimony to correctly estimate phylogeny. No tests were done to evaluate the support for individual nodes.

COMPARISON OF HARD VERSUS SOFT PART PHYLOGENIES AS INFERENCES TO OSTRACOD BIODIVERSITY IN THE FOSSIL RECORD Another aspect of my analysis was to determine the effect of the absence of specific anatomical features (e.g. soft parts) in a phylogenetic analysis. This was done because preservational bias against soft tissues has long been recognized as a barrier to clarifying evolutionary relationships and estimating biodiversity in the fossil record. Unfortunately, not all phylogenies can be reconstructed from extant taxa. There are some noted discrepancies between paleontologists and neontologists about the legitimacy of ostracod species that were designated solely on the basis of hard, skeletal carapace features versus a combination of hard and soft anatomical features such as hemipenes and antennules (Martens, 1990). But because ostracod soft tissues are rarely preserved in the fossil record, paleobiologists have no other choice but to use carapace morphology in their taxonomic and systematic analyses. The discrepancies between hard and soft parts suggest that there will be corresponding discrepancies between phylogenies produced by separate hard and soft part character sets, as well. 52

In this study, separate phylogenies were constructed using two unique subsets of the same phylogenetic database, in order to test whether or not similar topologies from both subsets result, and how they individually compare with the phylogeny produced with the combined datasets. By comparing the effect of the types of information that can be lost when the hard skeletal features are not used in a phylogeny or the soft anatomical details are lost, we gain an appreciation of how the biases in the fossil record impact the assessment of biological diversification patterns and their mechanisms.

RESULTS Results of Phylogenetic Analysis The computer program PAUP found 9, minimal-length cladograms (Figure 1.6) with a consistency index of 0.45 and a length of 269 steps. The same 9 trees were found in 100 replications of the search. Figure 1.7 shows the favored parsimonious reconstruction of the phylogenetic relationships using 70 characters across all taxa. The composite distribution of characters and character states in the Gomphocythere clade and outgroups is depicted in detail in Appendix B which maps each character on the favored phylogeny.

Definition of Ordering and Multistate Characters The unordered analysis identified 9 most parsimonious trees (269 steps, CI= 0.45, RI= 0.57) belonging to a single island (Figure 1.6). I employed my preferred ordering for multistate characters, where only 53

Figure 1.6: The 9 most parsimonious reconstructions A-I using 70 unordered characters. All trees have 269 steps, CI=0.45, RI=0.57, RC=0.25, Min. No. of Steps=121, Max. No. of Steps=462. The favored phylogeny and semi-strict consensus tree is shown in Figure 1.7. G. aelhiopis G. aelhiopis G. aelhiopis ~ G. parcedilalata G. parcedilalala G. parcedilalala G.alala G. alata G.alata G. crislala G. crislala G. crislala G. n. sp. 'coheni' G. n. sp. 'coheni' G. n. sp. 'coheni" G. n. sp. 'woulersi' G. n. sp. 'woulersi' G. n. sp. "woulersi" G. angulala G. angulala G. angulala G.obtusala G.oblusala G. oblusala G. n. sp. 'downingi' G. n. sp. 'downingi' G. n. sp. "downingi" G. n. sp. 'wilsoni' G. n. sp. 'wilsoni' G. n. sp. 'wilsoni' G.capensis G.capensis G.capensis G. expansa G. expansa G. expansa G. curta G.curla G. curIa G. ortali G. ortali G. orlali G. simplex G.lenis G.lenis G.·lenis G. simplex G.simplex G. angusta G.angusla G. angusta C. chariessa C. chariessa C. chariessa l. african a l. african a l. alricana "'=j l. dadayi l. dadayi l. dadayi ~. l.lhomasi l.lhomasi l.lhomasi ..... ;"

01 ~ t:; G. aethiopis G. aethiopis G. aethiopis ~ G. parcedilatata G. parcedilatata G. parcedilatata G.alata G.alata G.alata G. cristata G. cristata G. cristata G. n. sp. "coheni" G. n. sp. "coheni" G. n. sp. "coheni" G. n. sp. "woutersi" G. n. sp. "woutersi" G. n. sp. "woutersi" G. angulata G.angulata G. angulata G.obtusata G.capensis G. capensis G. n. sp. "downingi" G.expansa G. expansa G. n. sp. "wilsoni" G.obtusata G.obtusata G. capensis G. n. sp. "downingi" G. n. sp. "downingi" G.expansa G. n. sp. "wilsoni" G. n. sp. "wilsoni" G. curta G.curta G. curta G.ortali G.ortali G. ortali G.lenis G.lenis G.lenis G.simplex G. simplex G.simplex G.angusta G.angusta G.angusta I'%j C. chariessa C.chariessa C. chariessa L. africana L. africana L. africana ~. L. dadayi L. dadayi L. dadayi L. thomasi L. thomasi L. thomasi ..... ~

01 01 I-t G. aelhiopis G. aelhiopis G. aelhiopis G. parcedilalala G. parcedilalala G. parcedilalala G.alala G.alala G.alala G. crislala G.crislala G.crislala G. n. sp. "coheni" G. n. sp. "coheni" G. n. sp. "coheni" G. n. sp. "woulersi" G. n. sp. "woulersi" G. n. sp. "woulersi" G.expansa G.angulata G. angulala G.obtusala G.obtusata G.obtusala G. n. sp. "downingi" G. n. sp. "downingi" G. n. sp. "downingi" G. n. sp. "wilsoni" G. n. sp. "wilsoni" G. n. sp. "wilsoni" G. angulala G.capensis G. capensis G. capensis G.expansa G.expansa G. curta G. curta G. curta G.ortali G.ortali G.ortali G.lenis G.lenis G.simplex G.simplex G.simplex G.angusla G.angusla G.angusla G.lenis c. chariessa c. chariessa c. chariessa ~ L. africana L. africana L. africana L. dadayi L. dadayi L.dadayi l' L.lhomasi L.lhomasi L.lhomasi .... 0)

01 0) 57

Figure 1.7: A. The favored phylogeny of the 9 most parsimonious reconstructions using unordered characters. This tree has 269 steps, 01=0.45, R1=0.57, RO=0.25, Min. No. ofSteps=121, Max. No. ofSteps=462. Sub clades are indicated as A, B, 0, D, E, and F. Nodes are numbered as indicated. For list of apomorphies for each node, see Table 1.4 a,b. B. Semi-strict consensus tree of the 9 most parsimonious reconstructions. Strict consensus tree was identical to semi-strict tree. * G. angulata fossils from near Lake Turkana, indicate the minimum age of the species is 7 rna. + G. obtusata fossils from near Lake Turkana indicate the minimum age of the species is 2.5 rna. Outgroups are located on the right of the cladogram (C. chariessa, L. africana, L. dadayi, and L. thomasi) l:d ;> rn G. aethiopis ] > (I) G. aethiopis ] > G. parcedilatata ~ !3 < G. parcedilatata G.aJata ...0 ....en (I) G.aJata ::!. G. cristata j;).. (') G. cristata .... G.cohcoi "C Q ::r G. coheni 0 } ~ G. woutcl5i 0 G. woutcl5i J~ t:l ....~ '"(I) G. angulata ~ iil G. expansa ~ G~~I~] ~ s:: "I:I G. obtusata+ I"I:I "1 en G.capensis CD G. downiogi (") ~ G.obtusata ..... til 0 G. wilsoni 10 (I) ~ ]n G. capensis G.wilsoni G. expansa G. curta G. curta G.ortali G.ortali G.lcnis G.lcois G.simplex Jm G.simplex Jm G. angusta G. angusta C. chariessa C. chariessa L. africana L. africana L. dadayi L. dadayi L. thomasi L. thomasi

01 00 59

those characters with incremental addition of states (i.e. state 1-1 seta; state 2-2 setae; state 3-3 setae) were used. The trees generated from this analysis are similar to those from the unordered analysis, with the principal difference being the greater ambiguity in relationships within sub clade D discussed below. The trees found in the different analyses done with ordering of multi state characters are summarized in Figure 1.8. The number of most parsimonious trees, and fit of characters to these trees are shown in Table 1.3. The level of homoplasy revealed by these analyses (CI= 0.45) is close to that expected for 21 taxa with unordered, binary characters. The formula given by Sanderson and Donoghue (1989) predicts CI= 0.60 for unordered, binary characters, and using the parameters from their multiple regression yields CI= 0.40.

Definition and Analysis of Phylogenetic Structure The distribution (e.g. skewness) of tree lengths of all tree topologies (or a random sample of them) provides a sensitive measure of phylogenetic signal. Data matrices that have a strong phylogenetic signal produce treelength distributions that are strongly skewed. Simulated phylogenies have shown that data sets that are significantly structured typically have one or two shortest trees and a skewed distribution. A data set that is more or less random will have a normal distribution of treelengths (Huelsenbeck, 1991; Hillis and Huelsenbeck, 1992). The strongly skewed treelength distribution for the unordered analysis of Gomphocythere is 60

Figure 1.8: The 10 most parsimonious reconstructions A-J for the analysis using 25 ordered characters and 45 unordered characters. All trees have 270 steps, CI=0.45, RI=0.56, RC=0.25, Min. No. of Steps=121, Max. No. of Steps=462. G. aelhiopis G. parcedilalala G. aethiopis G.alala G. parcedilatata G. crislala G.alata G. n. sp. ·coheni· G. crislala G.angulala G. n. sp. ·coheni· G. n. sp. ·woutersi· G. angulala G.capensis G. n. sp. ·woulersi· G.obtusala G. capensis G. n. sp. ·downingi· G.oblusala G. n. sp. ·wilsoni" G. n. sp. ·downingi· G.expansa G. n. sp. "wilsoni" G.lenis G.expansa G.simplex G.lenis G. curta G. simplex G.ortali G. curia G.angusta G.ortali c. chariessa G.angusla L. african a C. chariessa L. africana ~ L. dadayi L. thomasi L. dadayi 1· L.lhomasi G. aethiopis G. aethiopis ..... G. parcedilatata G. parcedilalata (XI G.alata G.alala G. cristala G. cristata G. n. sp. ·coheni· G. n. sp. "coheni" G. angulala G. angulata G. n. sp. ·woutersi" G. n. sp. "woutersi" G.capensis G.oblusata G.oblusala G. n. sp. "downingi" G. n. sp. ·downinpi" G. n. sp. "wilsoni" G. n. sp. ·wilsoni G. capensis G.expansa G.expansa G. anQusta G.lenis G.lems G. simplex G.simplex G. curia G. curia G.ortali G.ortali G. angusta c. chariessa C. chariessa L. african a L. africana L. dadayi L. dadayi L.lhomasi L. thomasi .....~ G. aelhiopis G. aelhiopis / - G. parcedilatata G. parcedilatata 0 /G.alata G.alata G. cristata G. cristata G. n. sp. "coheni" G. angulala G. angulata G. n. sp. "woulersi" G. n. sp. ·woutersi" G.capensis G. capensis G.expansa G.obtusata G.oblusata G. n. sp. "downingi" G. n. sp. "downingi" G. n. sp. "wilsoni" G. n. sp. "wilsoni' G.expansa G. n. sp. "coheni" G.angusta G.lenis G.lenis G. simplex G.simplex G. curta G. curta G.ortali G.ortali G.angusla C. chariessa C. chariessa L. african a . L. africana L. dadayi L. dadayi L.lhomasi L. thomasi I-::!j.... G. aelhiopis G. aelhiopis G. pa""dHalala G. parcedilatata ~ A G.alata G.alala (I) ~ G. cristata G. cristata = ~ G. angulat!l G. n. sp. "coheni" .... G.capensls G.angulala 00 G.expansa G. n. sp. "woulersi" G. n. sp. "woutersi" G.obtusata G.obtusata G. n. sp. "downingi" G. n. sp. "downingi" G. n. sp. "wilsoni" G. n. sp. "wilsoni' G. capensis G. n. sp. "coheni' G.expansa G.lenis G.angusta G.simplex G.lenis G. curta G. simplex G.ortali G. curta G. angusta G.ortali C.chariessa C. chariessa L. africana L. africana L. dadayi L. dadayi L.lhomasi L.lhomasi (j) ~ JoooI G. aethiopis G. parcedilatata G.alata G. cristata G. angulata G. n. sp. ·woutersi­ G.capensis G. expansa G.obtusata G. n. sp. "downingi­ G. n. sp. ·wilsoni­ G. n. sp. ·coheni· G.lenis G. simplex G. curta G.ortali G.angusta I'%j C. chariessa L. africana ~. L. dadayi L. thomasi I-' 00 G. aethiopis G. parcedilatata G.alata G. cristata G. angulata G. n. sp. -woutersi­ G.capensis G.obtusata G. expansa G. n. sp. "downingi­ G. n. sp. "wilsoni" G. n. sp. "coheni" G.lenis G.simplex G. curta G.ortali G. angusta C. chariessa L. africana L. dadayi L. thomasi ~ CI:) 64

shown in Figure 1.9. The skewness and permutation tests each find significant phylogenetic structure in both the unordered and ordered Gomphocythere data matrices. For ordered Gomphocythere character runs, gl= -0.65, where 0 indicates a normal distribution (i.e. no skewness) and both -1 and 1 indicates the highest skewness. For unordered character runs, gl= -0.71.

Favored Phylogeny One of the 9 equally parsimonious trees from the unweighted analysis of the unordered characters is presented in Figure 1. 7 as a working hypothesis of the Gomphocythere phylogeny. This tree was chosen for several reasons. First, it retains all of the subclades that are supported in each of the 9 trees. These were islands of stability that were identified in each tree. For purposes of discussion below, the subclades have been defined by letters A-F (see Figure 1.7 for groupings). In addition, the favored tree was also chosen because it is similar to the tree generated without any sexually dimorphic characters (Figure 1.10). This tree was generated in order to see if some of the strongly sexually dimorphic characters controlled any specific topologies in the analysis. Whether or not their presence in the analysis affects the outcome because they are strongly sexual or whether they effect the outcome because they provide additional characters, is unclear. Finally, I favor this reconstruction because the characters reversed in other reconstructions are known to have high ecophenotypic plasticity (e.g. tubercles and valve reticulation) and therefore are less phylogenetically stable (Carpenter, 65

Figure 1.9: Treelength distribution of most parsimonious reconstruction using unordered characters. No. Trees

~ ~ ~ ~ tr.:l tr.:l ~ CJ1 o CJ1 o CJ1 o CJ1 o o o o o o o o o 0 o o o o o o o o 324 330 336 342 348 354 360 ::? 366 ~ ~ 372 i· -~ 378 ~ ~ 384 Q'l "'"" 390 = 396 402 408 414 420 426

~ ~ 67

1988). This is a biological assumption that I made concerning the validity of these characters. The complex nature of ostracod reproductive morphology and resultant sexual dimorphism was factored out of the analysis by completing a separate analysis using only non-dimorphic characters. In all, 13 characters were eliminated (out of the initial 70). The analysis from the dataset without sexually dimorphic characters yielded a single tree of 274 steps with a CI=0.44 and a RI=0.55 (Figure 1.10). The resulting phylogeny is similar to the favored phylogeny, indicating that sexual dimorphic characters have less effect on the structure of the tree than other morphological elements. Skewness statistics show significant phylogenetic structure with gl=-0.46 (Table 1.3). This single tree produced is congruent with the chosen cladogram for the complete dataset using all characters.

New Gomphocythere Species Recognized Four new Gomphocythere species were recognized in this analysis based upon specific hard and soft part anatomical features. They will be formally named and described in a subsequent publication. Their proposed names and the names used for this analysis are: Gomphocythere n.sp. II co h enI, . II G . n.sp. lid"owningI, II G . n.sp. 11'1'WI sonI, II and G . n.sp. IIwoutersL II A detailed taxonomic description of these proposed new species is given in Appendix C. These descriptions are given and species designations presented for the sake of clarity in this dissertation and I do 68

Figure 1.10: Single most parsimonious tree of 274 steps, generated using only non-dimorphic characters (57 characters) in the analysis. CI= 0.44, RI=0.56, RC=0.25, Min. No. of Steps=121, Max. No. of Steps=462. Subclades indicated as, A, B, C, D. G. aelhiopis J:.> G. parcedilatata G. alata G. cristata G. coheni J~ G. woutersi Itj

~ OQ-. ~ In >-t ~ .I-" I-" 0 G. curta G.ortali L. africana L. dadayi L. thomasi G.lenis G. simplex G. angusta C. chariessa

~ CO 70 not intend them to be available names until subsequent publication of this analysis.

Subclade Synapomorphies The monophyly of the ingroup in the favored phylogeny is supported by 24 synapomorphies. The specific synapomorphies and autapomorphies discerned for each node are listed in Table 1.4. In specific cases, certain subclades were supported by both hard and soft part character synapomorphies of morphologies that are not considered functionally related. For example, subclade B (0. alata, G. cristata, and G. n.sp. IIcohenill)(Figure 1.7) is defined by the following synapomorphies: bend, dorsal view, node number, presence of alae, hinge angle, radial pore canals, and A2 number of setae. These characters are not part of any correlated functional character complex (sensu Chippindale and Wiens, 1994; Huelsenbeck et aI., 1994) in ostracods.

Species Clustering There are several subclades that are supported by all 9 equally parsimonious phylogenetic trees. Those subclades, referred to as A-F, are as follows: A= G. aethiopis and G. parcedilatata; B= G. alata, 0. cristata, G. n. sp. IIcoheni li and G. sp. IIwoutersi ll ; C= G. angulata, G. obtusata, G. n.sp. IIdowningili G. n.sp. IIwilsoni,1I G. capen sis , and G. expansa; D= subclades B, C, E and G. curta and G. ortali; E= 0. lenis and 0. simplex; F= all ingroup taxa. Most sub clades are supported by between 2-13 synapomorphic characters (Figure 1.7). 71

Table 1.4a: List of apomorphous characters for each terminal taxa determined by DELTRAN optimization. 72

TABLE 1.4a

~~1.~1~~.~~~~~ 22 G. aethiopis 25,26,27,51,68 G. parcedilatata 34 23 G. alata 2,8,10,17,20,21,26,27,40,42,44,55,64 G. cristata 13,17,18,20,26,40,42,48,60,62,68,70 24 G. n.sp.'coheni' 6,10,13,15,16,21,22,23,25,34,46 25 G. n.sp. 'woutersi' 40,42,43,70 26 G. n.sp. 'downingi' 13,27,34,46,53,62,65 G. n.sp. 'wilsoni' 1,6,13,27,40,42,43,51,56,66 27 G.obtusata 7,43,44,49,54,60,65,71 28 G. angulata 13,26,41,42,62 29 G. capensis 8,16,18,54,62 G. expansa 5,8,13,20,21,31,61 32 G. curta 15,42,43,49,55,56,62,66,70 G.ortali 3,25,26,27,40,41,50,53,63,71 34 G.lenis 4,5,8,16,25,26,27,34,42,50,53,54,55 G. simplex 6,18,19,30,42,43,44,49,60,64 36 G. angusta 3,6,7,8,10 38 L. dadayi 1,2,3,6,15,16,20,43,44,48,50,51,65 L. thomasi 1,8,19,20,50,63,65,71 39 L. africana 2,48,71 40 C. chariessa 5,6,19,31,32,33,46,51,53,56 73

Table l.4b: List of apomorphous characters for each terminal taxa determined by ACCTRAN optimization. 74

TABLE 1.4b

M~lEfl~'~\"'1l!~7f}!.m:!'{ffiqtF~' _~f~~0i~~~~""~~~;~' '!Z><-".... !i".. • "'~~~~ "'''~ A~'I~"~~'W,.;.~~~" 's ~l".,,,.~~,..;:t~~~ O;IM\~~~ ~ ~.LJkilltj:~t.\.:i.~~ ,Cf.",.~ ... ~ "J 22 G. aethiopis 25,26,27,51,68 G. parcedilatata 34 23 G. alata 2,8,17,21,27,44,55,64 G. cristata 13,10,18,20,26,40,42,48,60,62,68,70 24 G. n.sp.'coheni' 6,13,15,16,21,22,23,34,46 25 G. n.sp. 'woutersi' 40,42,43,70 26 G. n.sp.'downingi' 13,27,34,46,53,62 G. n.sp. 'wilsoni' 1,6,40,42,43,51,56,65,66 27 G.obtusata 7,13,26,43,44,49,54,60 28 G. angulata 41,42,62,65 29 G. capensis 8,13,16,18,62 G. expansa 5,20,21,31,61 32 G. curta 15,42,43,49,62 G.ortali 3,25,26,27,40,41,50,53,63,71 34 G. Lenis 4,5,8,25,26,27,34,50,53,54,55,64 G. simplex 6,16,18,19,30,42,43,44,49,60 36 G. angusta 3,6,7,10 38 L. dadayi 1,3,6,15,16,43,44,48,51 L. thomasi 2,8,19,20,50,63,65 39 L. africana 48 40 C. chariessa 5,6,19,31,32,33,46,51,53,56 75

Hard Part Character Distributions Results of hard part distributions on the phylogeny indicate that the hingement is an important synapomorphy for the Gomphocythere subclade. Gomphocythere has a characteristic inverse lophodont structure, whereas the outgroup members have either a lophodont or adont hinge structure. There was no intraspecific variation in this character and it is extremely stable within each individual terminal taxa grouping (Appendix B). In Gomphocythere, three muscle scar states were coded based on the three distinct patterns (see Appendix B). Little to no variability was observed between patterns from individuals of the same species. When traced on the tree, this character is seen to define the subclade F, with the exception of G. alata and G. cristata. The distribution of the type and size of muscle scars support earlier suggestions that these characteristics are important diagnostic features for delineating species (Moore, 1961). Surface ornamentation was coded as smooth or reticulately robust and as seen distributed over the tree, shows a high amount of homoplasy. Reticulation shows little correlation with other shell morphologies such as presence or absence of nodes, alae, valve shape or size. Ornament morphology is sometimes correlated with water turbulence and possibly with sieve pore development and is considered to be strongly ecophenotypic (Liebau, 1977). In fact, the carapace surface structure is highly sensitive to adaptive evolutionary change (Benson, 1981) and is an important morphology to examine with respect to "long term" environmental change. 76

All members of subclades have prominent nodes and three species (G. alata, G. cristata, G. n.sp. "wilsoni" and G. n.sp. "downingi") also have significant alae. The node features on these two species are stable between sexes and along ecological gradients (e.g. substrate and depth).

Soft Part Character Distributions Results of the soft part character distributions indicate that soft parts are very important in this phylogenetic reconstruction. The number of segments and the number of claws present in the Al appendage in Gomphocythere is highly homoplastic. The number of setae present on the 'second joint, is synapomorphic for subclades Band C. Al in Gomphocythere is quite robust, particularly in the G. n.sp. "coheni" and G. alata , unlike the Al of more distantly related Limnocytherid groups. The mandibles (Md) were coded either as normal or bent knee and the corresponding mandibular palp setae, were coded as either straight or bifurcated. All Gomphocythere are united by their possession of the bent­ knee mandible and by a bifurcation in the mandibular palp setae, in contrast to the outgroups. A reduced maxillular palp (Mx) also supports the monophyly of the Gomphocythere relative to the outgroups. The thoracic appendages show similar patterns to the cephalic appendages and were highly homoplastic. On the first thoracic appendage (PI), the presence of setae on the first outer segment is synapomorphic for subclades Band C. The number of setae on the first outer segment of P2 was found synapomorphic with respect to the subclades Band C, except for the new 77 species G. n. sp. "downingi" and G. n. sp. "wilsoni". The presence of two setae on the first joint is apomorphic with respect to the subclade which includes: G. n. sp. "woutersi", G. angulata, G. obtusata, G. n. sp. "downingi", G. n. sp. "wilsoni", G. capensis, and G. expansa. There is little variation in hemipenes characteristics among Gomphocythere. In all species, the furca are on the lower lobe, away from the copulatory organ, whereas in the sister groups Limnocythere and Cytheridella, the furca are located near the copulatory organ. A movable distal lobe on the hemipenes unites all Gomphocythere with Cytheridella and is distinct from the fixed lobe of the Limnocythere.

Results for Hard Versus Soft Part Phylogeny Reconstructions Hard Part Reconstructions Eliminating the 34 soft part characters unavailable for fossil material diminishes the resolution among most parsimonious trees for both unordered characters and ordered characters, but also supports a large portion of the favored phylogeny. The majority-rule consensus tree is constructed from 47 trees. It differs from the unordered trees in denying the existence of subclade F and collapsing many taxa, with the exception of the subclade A, into unresolved polytomies or by swapping the taxon's placement on the tree. Although some of the major subclades or islands still exist with the exclusion of soft part data, there is more ambiguity in the specific relationships within these subclades (Figure 1.11). Still, many of the components of the subclades are supported in the hard parts only analysis. 78

Despite these areas that are well-supported, the hard part tree is less resolved than the composite analysis because there are areas of instability and dissimilarity between the two analyses. For example, Gomphocythere capensis and G. expansa, are supported as sister species in the composite (hard and soft parts) analysis, but are collapsed in the hard part analysis. In this case, the missing soft part anatomies: A2 setae on the outer third segment, Pl setae on first and second segments impact the resolution of the position of the two taxa in the subclade C. The subclade

that includes G. obtusata, G. n.sp. II downingi II , and G. n.sp. "wilsoni" is still supported in the hard part analysis, but not the sister group relationship between G. lenis and G. simplex. The base of the subclade G resolves now as a polytomy, which joins the sub clade C and part of B in a polytomy. Members of subclade E are now at the base of subclade Band G. simplex and G. angusta are grouped together as sister taxa at the base of sub clade F. The strict consensus tree obtained with all characters from the composite analysis differs from the hard-parts-only consensus tree primarily in the arrangement of the taxa that are fairly labile in the composite analysis. The stable "islands" from the composite analysis are partially supported, but with more limited resolution. The characters driving these differences from the combined character phylogeny and supporting the resolution found in the hard part analysis are the marginal ridge, posterior dorsal tubercle, valve shape, alae presence, hinge angle and shell thickness, which are all 79

Figure 1.11: A. Majority rule (50%) consensus tree for the analysis of hard part characters only (36 characters). The numbers on the branches indicate the percentage of the replications that support the group at that branch. B. Strict consensus tree for hard part character phylogeny. G. aethiopis G. aethiopis ~ G. parcedilatata ~. G. parcedilatata 0 > ::1..... u.aIata G. aIata '< G. cristata :;0 G. cristata c G. coheni G. angulata -t'll G.lenis G. capensis G.ortaIi G. curta G. curta G. expansa G. angulata ~ ..... G.lenis G. capensis ~ G.obtusata G. expansa I-i (t) G.ortaIi G.obtusata .~ G. simplex G. downingi ~ G. wilsoni ~ G. angusta G. coheni G. woutersi G. downingi G. simplex G. wilsoni G. angusta G. woutersi C. chariessa C. chariessa L. africana L. africana L. dadayi L. dadayi L. thomasi L. thomasi

(Xl 0 81 synapomorphic for either the subclade B or the sister group relationship of G. n.sp. "downingi" and G. n.sp. IwilsonL" Hence, many of the major structural features of the valves seem to support the various subclade designations. Results for Cl and Rl from the composite (hard and soft) analyses when compared to the hard part analyses show nearly identical values (Cl= 0.47 and 0.47 respectively; Rl= 0.57 and 0.62 respectively). Thus, the hard part characters that would be present in the fossil record exhibit levels of homoplasy similar to those shown by soft part characters. This conclusion is supported by the lack of significant differences in the means of Cl, and of Rl, for the two character sets over the shortest trees from the composite analyses.

Soft Part Reconstructions Eliminating the 36 characters that represent the hard skeletal features and using only the soft part characters increases the disagreement among most parsimonious trees for both unordered characters and ordered characters (Figure 1.12). But, unlike the hard parts only tree, the majority-rule consensus tree of the 156 trees from the soft parts unordered trees is nearly identical to the majority-rule consensus tree for the trees found using the composite set of all characters (Figures 1.7, 1.11 and 1.12). All subclades are supported in this soft parts only analysis. The values for Cl and Rl from the soft parts analysis and the composite analysis are nearly identical. However, the treelength is 82

Figure 1.12: A. Majority rule consensus tree for the analysis of soft part characters only (34 characters). The numbers on the branches indicate the percentage of the replications that support the group at that branch. B. Strict consensus tree for soft part character phylogeny. rn G. aethiopis ~ G. aethiopis M- h ::t e; ~ G. parcedilatata n ~ ~.alata. rEo M- 0 > ::t ~ ~ G. alata ~ s::~ '' G.obtusata .....~ G. coheni G. capensis ~ G.downin,"' G.expansa ~ CD G. wilsoni .I-l G. woutersi I-l G. curta t\:) G.lenis G.ortali G. parcedilatata G. simplex G. angusta C. chariessa L.africana L. dadayi L. thomasi L. thomasi

00 C.&J 84

substantially shorter in the soft part analysis as is expected when fewer characters are used.

DISCUSSION New Species There are at least four new species of Gomphocythere m Lake Tanganyika, all of which fall within my designated subclade D. Autapomorphies distinguishing these species occur in both hard and soft part characters. For Gomphocythere n. sp. "coheni", autapomorphies include: rounded valve shape, muscle scar attachments, four nodes present and presence of weak alar development. Gomphocythere n. sp. "woutersi" is delimited by the development of small surficial nodes, the number of Al segments and terminal claws and P3 setae on the first segment. Gomphocythere n. sp. "downingi" has a prominent alae developed along with a rectangular shape, A2 setae on the second segment, PI setae on second segment and P2 setae on first segment. Gomphocythere n. sp. "wilsoni" is defined by the marginal ridge, A2 terminal claws, A2 setae on first segment and P2 setae on first segment.

Phylogeny of Gomphocythere Comparisons of the results of the different analyses (Figures 1.6- 1.12) reveal consistent support for relationships among the major subclades or "islands" of the extant Gomphocythere. For example, the position of subclade B that includes Gomphocythere alata, G. cristata and G. n.sp. "coheni" is extremely well-supported in every analysis and 85

appears to be supported by such features as presence of alae or nodes, high reticulation of the carapaces as well as soft part characters such as the number of segments in the first antenna and the number of setae inserted apically on some segments. Similarly, the subclade containing G. aethiopis and G. parcedilatata is also stable and occurs in the same position in all analyses. It has been suggested by Martens (1993) that these two species might be synonymous. However, there are some clear differences between the two species in the presence of an alae on Gomphocythere aethiopis, the elongated nature of G. parcedilatata, the number of terminal claws on the second antennae, and the lateral position of the inner apical setae. Another area of stability within the tree is the relationship between the new species, Gomphocythere n.sp. "downingi" and G. n.sp. "wilsoni," which is supported in every analysis. Gomphocythere len is and G. simplex are often supported in the various analyses, but not in the hard parts tree, where the taxa are separated and G. lenis is inserted near the subclade that includes G. n.sp. "downingi" and G. n.sp. "wilsoni." Areas of instability and uncertainty include species that are both endemic to Lake Tanganyika and more cosmopolitan in other lakes. These species are not positioned at the base of the tree, whereas the other, more stable subclades, are more highly derived. This is also where the homoplasy levels are highest. The species Gomphocythere n.sp. "downingi" and G. n.sp. "wilsoni" are positioned within subclade C and represent the only area of consistency and certainty in this problematic area of the tree. 86

Homoplasy and Speciation Patterns within the Phylogenetic Reconstruction Homoplasy (the multiple appearance of the same character state in a phylogeny) is an important concept in phylogeny reconstruction and is something that is looked upon by some with much disdain, since it suggests possible error in the phylogenetic analysis (Sanderson, 1991). However, homoplasy may be a natural phenomena and not attributed to poor character choices, if the tendancy for parallelism in closely related taxa is high, which can often occur at the base of adaptive radiations (McCune, 1990) or close to the origin of major morphological innovations (Donoghue, 1989). If the homoplastic distribution of a character state or states is nonrandom over a phylogenetic tree, the presence of these character changes are still indicative of shared genetic similarity, even if they are not synapomorphic (Simpson, 1961; Mayr, 1969, Sanderson, 1991). It is very possible that the high level of homoplasy within this phylogeny are generated from parallelisms close to an adaptive radiation which may also be a result of the multiple colonizations of the lake. One line of evidence supporting this is that the more widely distributed species, defined as those occurring in other lakes besides Lake Tanganyika, are distributed throughout the phylogeny, instead of as a sister group to a single, monophyletic clade of Gomphocythere endemic to Lake Tanganyika. This pattern of speciation is mirrored by the Thiarid gastropods (Michel et aI., 1992) and Haplochromine cichlid fish that occur 87

in Lake Tanganyika (Sturmbauer and Meyer, 1992; Meyer, 1993; Meyer et aI., 1994).

Homoplasy Model for Multiple Invasions-Radiations The level of homoplasy expected in a multiple invasion-radiation scenario versus a single invasion with a subsequent radiation has not been thoroughly documented or modeled. However, there are certain scenarios that could be envisioned for both situations. Below, I develop a qualitative model to suggest a plausible explanation for the observed patterns in rift lakes. First, consider a hypothetical radiation of a "parent" species (p) and five descendant, or "daughter" species (d1-d5) that evolve in a single lake system, taking advantage of the available niche space (ns1-ns6). A pattern that could result could be one of relatively low homoplasy since each species would evolve into its own niche space. Autapomorphies would reflect this divergence into well-spaced niches (Figure 1.13). By contrast, a second scenario involving multiple radiations from two lakes could reveal a distinctive pattern from that of a single parent and subsequent radiation. Consider the case of six species found currently in one lake, but actually derived from two different lakes (Figure 1.14). Lake "T", a larger lake, supports, as before, six individual niche spaces. The daughter species in this lake evolve taking advantage of the most accommodating available niche spaces in the lake (ns2-ns4). Similarly, in the second lake, Lake "K," the daughter species evolve taking advantage 88

Figure 1.13: Speciation scenario of a single radiation with 5 descendant species in a large lake. p=parent species; d=daughter species; ns=niche space. Expectations would be low levels of homoplasy and autapomorphies that reflect the divergence of the daughter species into well-spaced niches. 89

Scenario 1: Single Radiation (with 5 descendant species) p = parent species d = daughter species ns =niche spaces

P d1 d2 d4 d5

dl--->d5 evolve, taking advantage of available niche space (nsl-->ns6)

Assume: total niche space in lake to be filled = 6

"Lake System"

Pattern: 1. Autapomorphies reflect divergence into well-spaced niches

2. Homoplasy relatively low

Figure 1.13 90

Figure 1.14: Speciation scenario of two adjacent lakes, Lake "T" and Lake "K" that will subsequently be joined. Lake "T" has 6 niche spaces, while Lake "K" has 3. d1-d3 evolve to fill ns2-ns4 in Lake "T", while d4 and d5 evolve to fill ns2 and ns3, respectively in Lake "K". 91

Scenario 2:Multiple Radiations (from 2 lakes) (with 5 descendant species) d1 d2 d3

dl--->d3 evolve, ~ taking advantage of available niche space (nsl-->ns6); fill most accomodating niches in lake first, ns2--->ns4 Lake "T" (ns! 'ns2 'ns3' ns4' nsS' ns6 J

p

d4-->dS evolve, taking advantage of niche space ns2 and ns3-obtaining similar features to dl Lake "K" and d2 of Lake "T" (most accomodating (ns! 'ns2' ns3 J niches in the lake)

Figure 1.14 92

of the most accommodating niche space in that lake (ns2 and ns3, where ns2 and ns3 are ecologically equivalent in both lakes). When the species of these two lakes are brought together via serendipitous invasion or lakes joining together from water level fluctuations, several apparent patterns could emerge, complicating or making spurious parts of the phylogeny reconstruction. The first pattern would be one in which the dichotomy represents the actual evolution of the sister groups within each lake and reflecting actual homoplasy (Figure 1.14). A second possibility could occur if the number of homoplastic characters overwhelms the phylogenetic sorting role of the true synapomorphies (i.e. autapomorphies playing a role of synapomorphies themselves). This could yield spurious associations of sister groups, or alternatively, imply a type of "pseudo" reticulation (Figure 1.15). The patterns emerging from this type of scenario, can potentially include high levels of homoplasy, because d4 and d5 would evolve in the same approximate niche space as dl and d2, demonstrating parallelism of adaptively derived characters. The autapomorphies would reflect the divergence into well-spaced niches but "like" autapomorphies would be perceived as synapomorphies in the latter example. Paradoxically, the latter example could actually lower homoplasy values for the tree. Of the two scenarios, single radiation and multiple radiation, the latter, multiple invasion scenario could very well account for the high level of homoplasy seen in the analysis of Gomphocythere in Lake Tanganyika. There is evidence that Lake Tanganyika has been linked to adjacent drainages in the past (Tiercelin and Mondeguer, 1991), which would also 93

Figure 1.15: Speciation scenario of two joined adjacent lakes, Lake "T" and Lake "K" and the patterns produced by the joining of two similar clades. In the first pattern, the true relationships would be seen as the tree produced is an unresolved dichotomy. Whereas, in Pattern II, dl and d4 and d2 and d5 would be falsely reconstructed as sister taxa, having similar autapomorphies that are falsely recognized as synapomorphies. This latter pattern could be misconstrued as reticulation (i.e. pseudo­ reticulation). (note: wavy lines indicate similar autapomorphies; dl-d3 are from Lake "T" and d4 and d5 are from Lake "K", as indicated in Figure 1.14). 94

Scenario 2b: Introduction of "K" species into Lake "T"

High homoplasy overall, but not enough to deny true relationships

d 1 and d4 share similar niches and have similar autapomorphies; high homoplasy in "like" Sister Clusters Dichotomy habitats (niches) Pattern I

(ns2) (ns3) dl d4 d2 d5 d3

not enough synapomorphies to hold d4 and d5 separate from dl-->d3; dl and d4 falsely represented as sister­ groups and d2 and d5 falsely represented as sistergroups; dl and d4 share same niche Pattern II space, as does d2 and d5 Pattern: Autapomorphies can either be reversed (gained or lost), as in Pattern I; or could become false synapomorphies, forcing the construction of false sister groups.

Homoplasy relatively high because d4 and d5 evolve in the same approximate niche space as dl and d2-----> parallelism results Figure 1.15 95

be consistent with the model, if the species within these drainage basins were similar to, or the same as the ones found in Lake Tanganyika. Of course, the dispersal of Gomphocythere may be much more complex than the simple radiation from adjacent lakes. Comparison of the favored phylogeny (Figure 1.7) to the currently known distribution of Gomphocythere species (Figure 1.1) suggests that a simple radiation from Lake Kivu is not solely responsible for the diversity of Gomphocythere in Lake Tanganyika. Although the multiple radiation model discussed here has used 2 lakes, it could be equally applied to divisions (i.e. basins) within one lake. Another possibility that must be explored is whether or not the base of subclade D was derived from Lake Tanganyika and then spread outwards via widely distributed species. Fossil evidence can be used to test this scenario by indicating the minimum age of any species within this subclade. Gomphocythere angulata has been found in the Late Miocene Lothagam III Formation from Lake Turkana and the Middle-Late Pliocene Koobi Fora Formation from east Lake Turkana (Cohen, 1982), showing that its associated branching event must pre-date 7 rna (see Figure 1.7a). Gomphocythere obtusata is also known from the Koobi Fora Formation of Middle to Late Pliocene age (approximately 2.5 rna). Neither of these species is present in Lake Tanganyika, but both are present in other Mrican lakes besides Lake Turkana. Lake Tanganyika is estimated to be 9-12 million years old. This age estimate makes the Koobi Fora Formation and Lothagam III Gomphocythere angulata and G. obtusata occurrences equivocal for testing the Lake Tanganyikan origin hypothesis, 96

since the Koobi Fora Formation (3.3-2.6 rna) and Lothagam III (4.5-3.5 rna) have been dated as younger than Lake Tanganyika. This age relationship suggests that G. angulata and G. obtusata are younger than the formation of Lake Tanganyika. However, few deposits besides the Lothagam Hill sites are known to be older than Early Pliocene, making further testing of the out-of-Lake Tanganyika hypothesis untenable. Furthermore, even if Lake Tanganyika was the crucible of Gomphocythere origination, not all species would be endemic to Lake Tanganyika but found in other African lakes. The minimum age estimate for G. angulata

and G. obtusata also suggest that at least parts of the clade could be as old or older than Lake Tanganyika itself.

Hard Parts Versus Soft Parts The effect of eliminating all hard part characters in the analysis of Gomphocythere was to decrease the resolution of the phylogeny, while retaining some of the important islands or subclades in the analysis. The majority rule consensus tree reconstructed using soft part characters was similar to the one reconstructed from the composite data set, except that it was shorter in length. The close affinity that the soft part tree had to the composite tree suggests that the soft parts are more stable and are more phylogenetically conservative than the hard parts. The hard part tree is less well-resolved because those features might be more ecophenotypic and therefore more plastic as indicated in studies of the effects of environmental conditions on ostracod carapace features (see Palacios-Fest et aI., 1994 for review). However, no similar studies have been done on 97 soft part features. This is an important issue to resolve with independent data sets. The implications and cautionary note for paleobiologists is that our ability to estimate phylogenetic relationships based on hard part features alone is diminished. This result is consistent with other analyses, such as Suter's (1994) work on the echinoderm class cassiduloids, where he determined that eliminating characters not available for fossil material reduced the agreement among most parsimonious trees for both unordered and ordered analyses. If species cannot be clearly defined or distinguished based on their skeletal features, then the fossil record, rich as it may be, could not provide an accurate phylogeny, but rather a coarser estimate of it. As far as Gomphocythere diversity in Tanganyika and other lakes is concerned, the absence of soft parts in lake core data has the potential of blurring trends of biodiversity changes and speciation patterns that are based solely on hard parts.

CONCLUSIONS This study represents the first phylogenetic analysis of any ostracod group in a rift lake system. The analyses undertaken here supports the monophyly of Gomphocythere, supported by 10 synapomorphies. The preferred phylogeny has a consistency index of 0.45, reflecting the large amount of homoplasy in the reconstruction. However, the phylogenetic structure of this analysis indicates that there is a reliable signal in the data, despite the high amount of homoplasy. Furthermore, after removing the homoplastic characters in the analysis, the tree topology remains, 98 indicating that despite many parallelisms, the basic topology of the tree is supported as species clusters or subclades. Four new species were found, bringing the total number of described endemic Tanganyikan Gomphocythere species to nine. The low number of autapomorphies defining species and the high amount of parallel character acquisition is consistent with the models of species flock originations in lakes via allopatric speciation or multiple invasions. A qualitative model of how such patterns might arise is given in Figures 1.13-1.15. In a comparison of phylogenies produced by using hard versus soft part characters, I found that removing soft part characters decreased the resolution of the analysis. This is also true of data sets consisting only of hard parts. This is not surprising, since loss of data generally reduced resolution in any type of analysis. It is significant, however, that more resolution was lost with the omission of the soft parts and not the hard parts, suggesting that the preservational bias present in the fossil record will have an effect on the analysis of biological diversification. In particular, the absence will significantly reduce the resolution of any analysis and will impede determining the processes behind the diversification patterns. The phylogeny also provides a basis for evaluating speciation mechanisms and in the future, the role of ecological factors in the diversification of ostracods in this lake system. 99

CHAPTER 2

ECOLOGY AND SPECIATION OF THE OSTRACOD CLADE, GOMPHOCYTHERE, IN LAKE TANGANYIKA

INTRODUCTION Lake Tanganyika Speciation Patterns Speciation in the East African rift lakes: Tanganyika, Malawi, Turkana, and Victoria is remarkable, with many endemic species flocks, each consisting of 10's to 100's of species of cichlid fish, gastropods, ostracods, and other aquatic organisms. Lake Tanganyika, in particular, supports one of the most diverse faunas of any lake system on Earth. The origin of the Tanganyikan fish and gastropod flocks has been attributed to divergence of populations caused by the creation of physical barriers to dispersal such as the formation of separate lakes or habitat fragmentation within a single lake during major lake level falls (Fryer, 1972, Brown and Mandahl-Barth, 1987; Ribbink, 1991; Michel et aI, 1992; Sturmbauer and Meyer, 1992, and Coulter, 1994). Similar mechanisms may have been important in the history of the diverse ostracod fauna of Tanganyika (>200 species) although very little prior work has been done on this problem (Cohen and Johnston, 1987).

Lake Tanganyikan Environments Lake Tanganyika is situated in the western branch of the East African Rift Valley between 3° and 9° S latitudes. It is the deepest of the African rift lakes (1470 m). It is long (-650 km) and narrow (-80 km maximum width) and divided into eight tectonically-formed sub-basins (3 100

bathymetric basins) that are expressions of linked, asymmetrical half­ grabens (Tiercelin and Mondeguer, 1991). The tectonic activity producing these geometries also produces patchy distributions of rock and sand substrates that dramatically alternate along the shore, providing a wide variety of habitats and potential for niche subdivision. The substrates vary in character from silt, sand, shell lags, stromatolites, and rocks (Figure 2.1). This patchy distribution of substrates contrasts with lakes like Lake Victoria, where the littoral zone is more uniform in substrate and slope, not having the contrasting shallow platforms and steep fault block grabens that are characteristic of Lake Tanganyika (Hecky and Degens, 1973; Rosendahl et al., 1986; Tiercelin and Mondeguer, 1991). The geologic history and resultant sedimentation patterns may be important to ostracod distribution. Distribution patterns of species according to substrate and depth often have corresponding morphological innovations.

PROBLEM TO BE ADDRESSED Rationale, Significance and Need for Study The purpose of my research was to examine the relationships between phylogeny and ecology within the genus Gomphocythere. Specifically, I investigated how the distributions of individual species and corresponding morphologies are correlated with ecological variables, such as substrate and depth tolerances. In turn, I examined these correlates in a phylogenetic context to elucidate and develop hypotheses of speciation 101

Figure 2.1: Map of Lake Tanganyika showing the various substrate distributions for mixed rock and sand (m), rocks (r), macrophyte meadows (p) and sand (s). Major faults are also represented on this map, showing the major sub-basins of the lake. 102

29 30 31

Burundi

4

Zaire Tanzania 5

6

( fault

~ structural high

Substrate Designations s=sand 7 10= mixed rock and sand r= rock p= macrophyte meadows Lake Tanganyika o 50 100km I I I 8

Zambia 103

mechanisms within the Gomphocythere clade, which could be tested in future field observations.

Theoretical Framework for the Proposed Study Most literature on speciation in ancient lakes focuses on the ways that the sub-division of a lake, or the limited distribution of a species within a lake, isolates populations. For example, lake level fluctuations cause either breaks or reunification of habitats that might result in evolutionary change in separate lake systems or by intra-lacustrine allopatric radiation. The main hypotheses of these studies involve either disjunct shoreline habitats (Fryer, 1959, and 1977), substrate barriers to dispersal (Michel et aI., 1992) or division of a lake into segments (Greenwood, 1981; Sturmbauer and Meyer, 1992). Basin morphometry, depth and substrate variability, including barriers such as deltas, have important bearing on the application of these theories that invoke physical isolation of some sort. It is within this theoretical framework of speciation by intralacustrine fragmentation that various hypotheses of relationships between ecological and biogeographical distributions and the evolution of Gomphocythere species and species clusters can be developed. For example, I developed hypotheses to test whether a substrate exists that has the most or fewest species occurring in it, in order to see if character acquisition and hence, evolution, is correlated to habitat occurrence. 104

The specific hypotheses I will test are summarized below: 1. There is a biogeographic congruence between Gomphocythere phylogeny and species distribution in African lakes. 2. There was a single invasion and subsequent radiation of Gomphocythere species in Lake Tanganyika, demonstrated by a single clade of endemic Gomphocythere species. 3. Gomphocythere subclades consist of species with the same geographic distributions within Lake Tanganyika. 4. Gomphocythere subclades consist of species with the same substrate distributions. 5. Gomphocythere subclades consist of species with the same depth distributions. 6. Character acquisitions traced on the phylogeny show a relationship to the ecological variables available for this study.

Previous Work on Gomphocythere Ecology and Speciation Most of the - 200 species and 25 genera of ostracods in Lake Tanganyika belong to three taxonomic groups: the Family Candonidae, the Family Cytherideidae and the Family Limnocytheridae. Gomphocythere is a member of the Family Limnocytheridae. In Lake Tanganyika, there are 9 known species of Gomphocythere that are endemic to the lake. The genus provides a useful study group for investigating ecological controls on speciation because it is known to be monophyletic (Martens, 1993). 105

Habitat requirements of Gomphocythere species in African lakes are not well known. Gomphocythere species have been recorded from a variety of littoral and profundal habitats (Colin and Danielopol, 1979; Martens, 1994). Water chemistry tolerance ranges have been documented for Gomphocythere angulata and G. obtusata, the most widespread species of the genus in Africa, neither of which occur in Lake Tanganyika. These studies indicate fairly broad tolerance ranges for Gomphocythere dominated assemblages (conductivity: 1500-4000 /lmho/cm; alkalinity: 15- 30 meq/l)(Cohen et aI., 1983). Because there is a lack of direct observations of Gomphocythere ecology, the data gathered for this study consists of the general ecological conditions existing at the collecting localities of live and dead specimens and provides the only information about Gomphocythere species' environmental tolerances and niche specifications. Although these data are not ideal, they are similar to the data available to paleontologists when they reconstruct paleoecological tolerances. Carbonel (1983) demonstrated that Gomphocythere and Limnocythere are more abundant in conditions that are at the beginning of the alkaline pathway (-HC03» Ca2+ + Mg2+) in lacustrine hydrochemical regimes and disappear as the brine becomes more concentrated (towards a Na-C03-CI brine).

There have been some studies done on the link between various morphologies and environmental conditions such as salinity, temperature, sediment texture, organic matter content, oxygen content, and alkalinity. Species with robust carapace ornamentation are often found in coarse grained sediments or hard substrates (i.e. rocks and stromatolites) 106

(Henderson, 1990), whereas species with smooth carapaces are typically found in a wider variety of sediment-substrate types and are often more abundant in finer grained sediments (Holmes, 1992). And whereas substrate zonation in ostracod species has been demonstrated in several lakes (Carbonnel et al., 1988; Holmes, 1992), depth as a spatial factor is less likely to have a direct impact on ostracod distribution. Instead, depth-related factors such as temperature, oxygen and salinity are more likely to regulate ostracod distribution patterns (Kotzian, 1974, Keen, 1975, Danielopol et al., 1990; Cohen, 1986; Palacios-Fest et al., 1994). Specific morphologies, such as the development of sieve pores in Podocopida (the order that includes Gomphocythere) correlate to the salt concentration in the environment. In low salinity concentrations, round sieve pores occur, whereas in high concentrations, irregular sieve pores predominate (Rosenfeld and Vesper, 1977). Lake Tanganyika has low salinities (TaIling and TaIling, 1965; Coulter, 1991) and the ostracods in this analysis all have rounded sieve pores. In addition to morphological features, most endemic species of ostracods in ancient lakes are fairly small « 1 mm). Size in ostracods has been inversely correlated with intensity of predation in other ostracod groups. Size, as indicated here, is compared across species within the genus. No comparison of size of Gomphocythere species to other species was used since the comparison is not relevant to this analysis. Whether or not size is also correlated to an infaunal lifestyle has not been previously documented. 107

Dispersal Strategies Dispersal capability may have been important to producing the speciation patterns of ostracods in Lake Tanganyika. Whereas dispersal can be directly related to various physical and biological factors, the means by which ostracods are dispersed within a lake system is controversial. Dispersal is easiest through the production and subsequent release of eggs into the water. Brooding in organisms such as Gomphocythere limits their dispersal abilities. Cohen and Johnston (1987) suggested that brooding species in lakes might have more localized gene pools with limited gene flow between intra-lacustrine populations, than non-brooders. Therefore, it might be expected that brooding species would show different patterns of speciation from non-brooders. However, intra-lacustrine dispersal of brooding species could be enhanced if the species in question had broad ecological tolerances, or strong swimming abilities. In ostracods, strong swimming abilities are indicated by the presence of well-developed setae on their cephalic and thoracic appendages (Van Morkhoven, 1962). These features are present in Gomphocythere. Intra-lacustrine dispersal is also facilitated in non-brooders by the presence of desiccation-resistant eggs. Some Limnocytherids (the family that includes Gomphocythere) have these types of eggs. Gomphocythere have fairly durable eggs, as observed in dissections of females, but whether or not they are desiccation resistant is not known. However, since Gomphocythere are all brooders, durable eggs are probably not critical for survival. More important for Gomphocythere dispersal is 108

evidence that Gomphocythere species are effective swimmers. Their dispersal ability is probably more dependent upon their ability to swim than on their egg durability or brooding behavior.

Geographic and Habitat Distribution Hypotheses Existing data on Gomphocythere species distributions in Mrica shows that certain species are endemic to specific lakes whereas other species are found in many lakes throughout the continent. From this information, it would be expected that the biogeographic distribution of these species would reflect aspects of their overall evolutionary and dispersal history. Comparison of Gomphocythere species distributions in African lakes with the phylogenetic history of the clade, allows evolutionary and dispersal hypotheses to be tested. Likewise, the distribution of endemic species within the phylogenetic tree may provide important insight on the long-term historic interlake associations between Gomphocythere species. The species endemic to Lake Tanganyika would be expected to cluster into one subclade, if there was one invasion and no subsequent outward dispersals. If there were multiple invasions and subsequent radiations, there may be a greater propensity for endemics and non-endemics to be interspersed throughout the phylogenetic tree. In cores and grab samples, I have observed consistent relationships between certain Gomphocythere species and lithologies. Correlations between species and their habitat affinities (i.e. substrate and depth) provide a better understanding of the relationship between niche 109

specifications and evolutionary processes. Whether higher order factors such as grain size, compaction, water saturation, sediment disturbance or organic content, control distributions is not known, and could not be specifically tested here because of limited sampling information, but would be fruitful areas of future inquiry.

Ecological Character Correlation Hypotheses Specific morphological characters, particularly those considered ecologically driven features (e.g. surface ornamentation, nodes and alae development, shell thickness, spines, sieve pores, and terminal claw development) may be associated with specific substrates or water depths in Lake Tanganyika and could help explain some of the significant homoplasy revealed in the phylogenetic analysis of Gomphocythere (see Chapter 1). If the development of adaptive morphologies is tightly linked to habitat variables in Gomphocythere, this could help to explain the speciation patterns. This is so because there is a great repetition of habitats within and between East African lakes which can potentially support isolated populations of ostracods. Gomphocythere species and subclades may have specific autapomorphies that are adaptive to their specific lake environment. Integrating ecological and phylogenetic datasets has been the focus of many recent studies that attempt to show the historical connections of ecology and genealogy (Brooks and McClennan, 1991). The value of this approach is that it allows correlations between character state acquisition and ecological variables to be identified. This information provides a 110 better understanding of the possible evolutionary mechanisms of the clade. It also provides structure for additional testing of speciation mechanisms in subsequent field investigations.

METHODS General Phylogenetic Method I constructed a phylogeny for 17 ingroup species of Gomphocythere from lakes in eastern Mrica, including the Ethiopian rift lakes, the East African rift lakes and lakes in South Africa. I chose four outgroup species from sister group taxa, also from Africa. I defined and coded seventy characters as either unordered or ordered for all 21 taxa in my analysis. I chose similar numbers of hard part characters (34) as well as soft part characters (36) for the complete analysis, using male and female data sets separately. The composite phylogeny is given in Figure 2.2. The full details of the composite phylogenetic analysis of Gomphocythere are described in Chapter 1.

Geographic Location Data Analysis and Methods Interlake Distributional Patterns I examined geographical distribution on two scales in this analysis. The first scale I examined was the interlake distribution of all of the Gomphocythere known in Africa. Distributional data used for this analysis were compiled from the literature (see Chapter 1), especially from distributions given in Martens (1993) and from collections at the Royal Belgian Institute for Natural Sciences. It should be noted here that there 111

Figure 2.2: The favored phylogeny of the 9 most parsimonious reconstructions using unordered and ordered characters. This tree has 269 steps, CI=0.45, RI=O.57, RC=O.25, Min. No. ofSteps=121, Max. No. of Steps=462. Subclades are identified as A-F (see Chapter 1 for details). G. aethiopis J;.. G. parcedilatata G. alata G. cristata G. coheni G. woutersi J~ G. angulata G.obtusata I~ ~ ~. In I~ (Jq s:: 1-1 CD .tv tv G. curta G.ortali G.lenis G. simplex }~ G. angusta C. chariessa L. africana L. dadayi L. thomasi

~ ~ t-:) 113

is a paucity of data for Gomphocythere in Mrican lakes other than Lake Tanganyika and Lake Turkana. Future studies on other African lakes are necessary for better documentation of Gomphocythere species in these lakes. Endemic species of Lake Tanganyika are defined as those species known only from this lake.

Intralake Distributional Patterns The distributions of taxa within sub-basins of Lake Tanganyika were plotted on a base map for comparison. Data obtained for this purpose were extracted from various expeditions, including the Belgian Hydrobiological expedition of 1946-47, and the expeditions of Cohen, 1986, 1989, Martens, 1990, Cohen, 1992, and Martens, 1992. Within this intralake comparison, I designated eurytypic species as those broadly occurring at many locations, within many substrates and depths. Stenotypic species were defined as those species with narrow distributions, particularly with respect to substrate and depth tole:rances.

Ecological Distribution Data Ecological distribution data of Gomphocythere species from Lake Tanganyika were extracted from a database of ostracod occurrences from Lake Tanganyika (compiled by Cohen et aI., 1985-1995) as well as the Belgian Hydrobiological expedition of 1946-47, and the expeditions of Cohen, 1986, 1989, Martens, 1990, Cohen, 1992, and Martens, 1992. Substrate and depth data were used as well as Gomphocythere abundance data from each sample site (n=128 sites). Cohen's database was 114

constructed from counts of 350 randomly chosen valves of adult ostracods from grab samples. For each site I calculated the proportion of the total fauna made up by each Gomphocythere species. Substrates for each sample were determined in the field by the collecting team (see Figure 2.1). Each substrate designation represents the dominant substrate feature (Le. rocks, stromatolites, sand, shells, silt, and oolites or mixtures of these). Water depth (m) was recorded in the field using a diver's gauge or a fathometer during sample collection. Specimens for the study were recovered between 0-48 meters. Only a short fossil record of less than 1,000 years obtained from gravity cores is currently available for the ostracod group Gomphocythere in Lake Tanganyika. Therefore, fossil data were not available for my study of Lake Tanganyikan Gomphocythere. However, fossil data is available for elsewhere in Africa, most notably the fossil deposits surrounding Lake Turkana (Cohen, 1982). One-way MANOVAs (multiple analysis of variation) were performed on the substrate distribution data using the statistical package SAS/STAT (SAS Institute, 1989) in order to test whether Gomphocythere species' occurrences within 12 defined substrates are random. I chose a GLM (with MANOVA) over ANOVA because it is more efficient for unbalanced data (SAS Institute, 1989). MAN OVA was chosen because there was more than one dependent variable. An R-mode cluster analysis was performed to place taxa into groups according to the taxa's substrate distribution. All of these analyses were run on SAS/STAT (1989). Cluster associations were generated using the average linkage method (UPGMA 115

unweighted pair-group method using arithmetic averages), where the similarity is expressed as squared Euclidean distance. Principal component analyses (PCA) were performed on 7 out of 9 taxa (rare taxa were excluded) and 12 substrates to test whether there are close associations between certain species with respect to substrate affinities. The function of a PCA is to project points from a multidimensional space into fewer dimensions so that the arrangement of points suffers the least distortion. PC axes were chosen that captured as much variance as possible along the ordination axis (i.e. PC1, PC2, and PC3). Taxa were then plotted in multidimensional space to see if any taxa clustered together. I used PCA as an independent exploratory technique and a way to assess the validity of the cluster analysis. I measured body size on the original carapace specimens, using a microscopic micrometer and also from Scanning Electron Microscopy images. Size was used as a sexually dimorphic character in my phylogenetic analysis (see Chapter 1). The distribution of size on the Gomphocythere phylogeny was examined with respect to the distribution of ecological features mapped onto the tree, in order to determine if species size correlates with substrate, particularly since the substrate grain sizes vary from boulder size (RKS and ROS) to silt size (SIS and SIL). Ecological variables and autapomorphous morphological characters were examined to determine if a correlation exists between various character states (that mayor may not have an adaptive origin) and substrate or depth specificity. An analysis of adaptation as it relates to 116

historical changes (sensu Lauder et aI., 1993) could not be accomplished in this study since no independent measure of performance or fitness of morphology was available. The performance measurement is critical for testing adaptational hypotheses, since it determines if the trait is really an adaptation, an exaptation (Gould and Vrba, 1982) or a nonadaptation (Baum and Larson, 1991). Therefore, specific traits were mapped onto the preferred phylogeny and the change in "selective regime" (i.e. environment) was indicated in order to determine if there was correlation between them, but no adaptational argument was inferred from such a comparison. In addition, the homoplasious patterns of character state acquisition were compared with ecological variables.

RESULTS Biogeographic Distributions of Gomphocythere Interlake Distribution Plotting occurrence data for Gomphocythere in Mrica shows no congruence between species distribution and phylogeny (Figure 2.3). Many of the lakes that are geographically distant share similar Gomphocythere species (e.g. Molopo Oog, RSA and Lake Victoria, East Africa). It may be that this is an artifact of the inadequate ostracod sampling in many of the East African rift lakes, but the species present do not support the hypothesis that there is a congruence between Gomphocythere phylogeny and species distribution in adjacent African lakes. Lake Kivu, Lake Tanganyika and Lake Victoria, which are adjacent to one another have completely different Gomphocythere species. 117

Figure 2.3: Map of the biogeographic distributions of all Gomphocythere species in Africa. The numbers on the diagram represent the following species: 1=Gomphocythere aethiopis; 2=G. alata; 3=G. angulata; 4=G. angusta; 5=G. capensis; 6=G. cristata; 7=G. curta; 8=G. expansa; 9=G. lenis; 10=G. obtusata; 11=G. ortali; 12= G. parcedilatata, 13=G. simplex; 14=G. n.sp. "coheni;" 15=G. n.sp. "downingi;" 16=G. n.sp. "wilsoni;" 17=G. n.sp. "woutersi." Scale is indicated. 118

Distribution of Gomphocythere

lOOOkm

Figure 2.3 119

Endemic Tanganyikan Gomphocythere Species endemic to Lake Tanganyika include: Gomphocythere alata, G. cristata, G. curta, G. lenis, G. simplex, G. n.sp. "coheni," G. n.sp. "downingi," G. n.sp. "wilsoni" and G. n.sp."woutersi." Endemic species, as traced over the phylogenetic tree, indicate that endemics do not cluster together, in a single clade, but instead, are interspersed with non­ endemics (Figure 2.4). Therefore, either there were multiple invasions and subsequent radiations of Gomphocythere species in Lake Tanganyika as would be demonstrated by a single clade of endemic Gomphocythere species, or Lake Tanganyikan endemics subsequently dispersed from the lake. The latter possibility is unlikely, given the fact that Gomphocythere angulata (a cosmopolitan species that rests within the Lake Tanganyikan radiation) is known from northern Kenya in Late Miocene aged sediments, only slightly younger than the maximum age for Lake Tanganyika itself.

Intralake Distribution The following analyses focus only on Lake Tanganyika Gomphocythel'e species and their distributional patterns in the lake. The distribution map shown in Figure 2.5 indicates the widespread nature of all Gomphocythere species throughout the entire lake. Very little information is available from the west side of the lake as a result of logistical problems with collecting in this area. No species is restricted to one sub-basin or area of the lake. Little is known about the life histories of either the widespread or narrowly distributed species to assess whether or not their ecological 120

ranges remain constant through their ontogenetic development. Instar data were not used in my analyses because many ins tars are morphologically convergent and could not be identified accurately at the species level. In addition, I do not know whether or not these species have maintained their environmental affinities through geologic time. Long cores throughout the lake are necessary for documenting any long term changes over time. Cores presently available do not extend that far back in time and their geographic coverage in the lake is restricted to deeper waters and certain parts of the basin.

Results of Ecological Character Mapping Substrate Distributions One-way General Linear Model results with MANOVA can test whether there is significant variation in the distribution of individual species with respect to different substrates (Null hypotheses rejected when p<.05). Results from these analyses are summarized in Table 2.1.

Rare species, such as Gomphocythere lenis and G. simplex were excluded. Rare substrates (with no observations) were also excluded. To increase the power of the statistical test, substrates were pooled according to composition and relative grain sizes. Results show that 3 species, G. cristata, G. woutersi and G. downingi do not show significant variation in the distribution of individual species in different (pooled) substrates, suggesting they are randomly distributed. Using a less-powerful test (i.e. DF=12) with non-pooled substrates, there were five species (G. cristata, G. 121

Figure 2.4: A) Predicted pattern of what would be expected if there was a single invasion and subsequent radiation of a single species in Lake Tanganyika; B) Distribution of Lake Tanganyika species mapped onto the favored phylogeny. O=species found in other lakes in Mrica; 1=endemic species to Lake Tanganyika. 122

Other African Lake Tanganyika Lakes

A Predicted Pattern

o Species outside L. Tanganyika 1 Endemics B Actual Pattern

Figure 2.4 123

Figure 2.5a: Distribution of Gomphocythere species in the sub-basins of Lake Tanganyika. Numbers represent localities that are summarized in Figure 2.5b. 124

~l

40·4 -4

Zaire Tanzania -5

-6

-7

Lake Tanganyika o 50 100km ! -8

Zambia

Figure 2.5a 125

Figure 2.5b: Relative frequency data for Gomphocythere species in Lake Tanganyika for localities represented in map in Figure 2.5a. In all samples, n=350 valves. Latitude and longitude are indicated in degrees and minutes. Substrate designations are given below. Depth is indicated in meters. Figure2.5b LOCATION MAP SPECIES DlSmmUTION

SITE Long. Lat. Substrate Depth(m) G. cristata G. curta G. alata G.lenis G. cnheni G. woutersi G. wilsoni G. cIowningi G.simplex 1 89IMSI56 29.18 3.21 SAN 4.00 0.00000 0.00000 0.00000 0.00000 0.00286 0.00000 0.00000 0.00000 0.00000 2 891MS1133 29.12 3.22 SAN 7.00 0.00000 0.00000 0.00000 0.00000 0.01143 0.15429 0.00000 0.00000 0.00000 3 851ACl52 29.21 3.22 SIL 2.00 0.00000 0.00290 0.00000 0.00000 0.01710 0.00000 0.00000 0.00860 0.00000 4 851ACl19 29.21 3.30 SIS 0.30 0.00570 0.00000 0.00290 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 5 921LUHJ1 29.09 3.31 RKS 1.00 0.00571 0.05714 0.02000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 6 921LUHJ5 29.09 3.31 RKS 5.00 0.01430 0.05710 0.01430 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 7 921LUHJ10 29.09 3.31 RKS 10.00 0.04000 0.03710 0.11710 0.00000 0.00000 0.00570 0.00000 0.00000 0.00000 8 921LUHl20 29.09 3.31 RKS 20.00 0.02860 0.07430 0.11140 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 9 921LUHJ40 29.09 3.31 RKS 40.00 0.00290 0.07140 0.02000 0.00000 0.00290 0.00000 0.00000 0.00000 0.00000 10 851ACl43 29.21 3.33 SIS 3.50 0.00000 0.00000 0.00000 0.00000 0.02000 0.00000 0.00000 0.00000 0.00000 11 851ACl44 29.21 3.34 SIS 1.50 0.00000 0.00000 0.01430 0.00000 0.03140 0.00000 0.00000 0.09430 0.00000 12 851ACl06 29.21 3.34 SIL 4.80 0.00000 0.00000 0.00900 0.00000 0.00000 0.00000 0.00000 0.00300 0.00000 13 861RJ127 29.21 3.34 SAS 40.00 0.00000 0.00000 0.00570 0.00000 0.00570 0.00000 0.00000 0.00000 0.00000 14 891MS1225 29.09 3.36 ROS 31.00 0.03429 0.12000 0.07143 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 15 851ACl48 29.21 3.36 SSS 2.50 0.00000 0.04000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 16 851ACl26 29.21 3.36 SIL 30.00 0.00000 0.00000 0.00000 0.00000 0.00290 0.00000 0.00000 0.05430 0.00000 17 921KM2515 29.21 3.36 SIL 5.00 0.00000 0.01430 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 18 921KM25120 29.21 3.36 SAN 20.00 0.00000 0.03429 0.00000 0.00000 0.00000 0.00000 0.01714 0.00000 0.00000 19 851ACl26 29.21 3.36 SIL 30.00 0.00000 0.00000 0.00000 0.00000 0.00290 0.00000 0.00000 0.05430 0.00000 20 921KM25140 29.21 3.36 SIS 40.00 0.00000 0.00000 0.00000 0.00000 0.00571 0.00000 0.02286 0.01429 0.00000 21 921KM2817 29.21 3.37 MSR 1.00 0.00000 0.09710 0.03430 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 I 22 921KM28f5 29.21 3.37 MSR 5.00 0.00000 0.14860 0.18570 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 23 921KM28f10 29.21 3.37 MSR 10.00 0.00571 0.12857 0.15143 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 24 921KM28120 29.21 3.37 MSR 20.00 0.03143 0.06286 0.24286 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 25 921KM28f40 29.21 3.37 MSR 40.00 0.12857 0.08857 0.16571 0.00000 0.00000 0.00000 0.00000 0.05143 0.00000 26 851ACl11 29.21 3.38 ROS 10.00 0.00000 0.09430 0.00570 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 27 891MS1172 29.21 3.41 ROS 4.00 0.00000 0.00000 0.00000 0.00000 0.00000 0.02000 0.00000 0.00000 0.00000 28 891MS1178 29.20 3.41 ROS 13.00 0.00000 0.17143 0.03429 0.00000 0.00286 0.00571 0.00571 0.00000 0.00000 29 891MS117 29.20 3.41 ROS 13.00 0.00000 0.22600 0.04300 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 30 891MSll84 29.19 3.42 SHS 28.00 0.00571 0.22571 0.02000 0.00000 0.01143 0.02571 0.00571 0.02000 0.00000 31 851ACl28 29.19 3.42 SIS 35.00 0.00000 0.00571 0.00571 0.00000 0.02286 0.00000 0.00000 0.22571 0.00000 ..... 32 851ACl15 29.19 3.42 ROS 8.00 0.00000 0.00000 0.01714 0.00000 0.00286 0.00000 0.00000 0.00000 0.00000 tv 33 851ACl17 29.18 3.43 SIS 20.00 0.00000 0.00286 0.00571 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 ~ - --- SITE Long. Lat. Substrate Depth(m) G. cristata G. curta G. alata G.lenis G. coheni G. woutersi G. wilsoni G.tiowningi G. simpleJ: I 34 861RJ/56 29.19 3.43 ROS 26.00 0.00571 0.07429 0.20000 0.00000 0.00857 0.00571 0.00000 0.01143 0.00000 I 35 851ACl53 29.22 3.50 SAS 5.00 0.00600 0.00900 0.00300 0.00000 0.00600 0.00000 0.00000 0.00000 0.00000 36 861RJ/39 29.24 3.57 SIL 42.00 0.00000 0.00000 0.01400 0.00000 0.02300 0.00000 0.00000 0.22600 0.00000 37 851ACl41 29.26 3.58 SIL 2.50 0.00000 0.00570 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 38 861RJ/44 29.25 3.59 SIS 30.00 0.00000 0.00000 0.00600 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 39 851RJ/45 29.26 4.02 SAN 47.00 0.00000 0.00000 0.02571 0.00000 0.16571 0.00000 0.00000 0.20857 0.00000 40 921CBR15 29.15 4.03 RKS 1.00 0.00600 0.08300 0.02300 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 41 921CBRl10 29.15 4.03 ROS 10.00 0.04000 0.13140 0.05430 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 42 921CBRI20 29.15 4.03 ROS 20.00 0.01100 0.10600 0.10000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 43 921CBRl40 29.15 4.03 ROS 40.00 0.00860 0.09710 0.18570 0.00000 0.00290 0.00000 0.00000 0.00000 0.00000 44 891MS1108 29.15 4.03 SAN 48.00 0.00570 0.32000 0.04570 0.00000 0.01710 0.00000 0.00860 0.01140 0.00000 45 851ACl22 29.33 4.12 SAS 4.00 0.00000 0.03143 0.22571 0.00000 0.00286 0.00000 0.00000 0.00000 0.00000 46 851ACl21 29.32 4.13 SIS 1.50 0.00290 0.02290 0.01430 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 47 851ACl60 29.33 4.14 SIL 4.50 0.00000 0.00570 0.01710 0.00000 0.00860 0.00000 0.00000 0.00000 0.00000 48 851ACl27 29.34 4.17 SIL 40.00 0.00000 0.00000 0.03429 0.00000 0.00571 0.00000 0.00000 0.14571 0.00000 49 861RJ/48 29.34 4.17 SSS 30.00 0.00000 0.00000 0.00000 0.00000 0.00860 0.00000 0.00000 0.01710 0.00000 50 851RJ/49 29.34 4.19 SHL 18.00 0.00570 0.00000 0.20290 0.00000 0.10860 0.00000 0.04000 0.05710 0.00000 51 861RJ/17 29.36 4.21 SAN 7.50 ·0.00000 0.00000 0.00000 0.00000 0.00286 0.00000 0.00000 0.00000 0.00000 52 86IRJ/01 29.37 4.22 SIS 15.25 0.00000 0.00000 0.09710 0.00000 0.02860 0.00000 0.00000 0.02570 0.00000 53 891MS1102 29.12 4.31 SHS 27.00 0.00570 0.24000 0.28860 0.00000 0.00000 0.00000 0.00000 0.01430 0.00000 54 861RJ157 29.37 4.52 SHS 33.50 0.03710 0.01140 0.09140 0.00000 0.01430 0.00000 0.00000 0.02860 0.00000 55 861RJ181 29.48 5.05 MSS 9.25 0.00000 0.00000 0.00000 0.00000 0.13140 0.00000 0.00570 0.00860 0.00000 56 86IRJ/03 29.50 5.13 RKS 2.00 0.00000 0.00000 0.00000 0.00000 0.03100 0.00000 0.00000 0.00600 0.00000 57 86IRJ/04 29.50 5.13 OOS 4 .. 5 0.00000 0.00000 0.00000 0.00000 0.01143 0.00000 0.00000 0.00000 0.00000 58 86IRJ/05 29.50 5.13 RKS 5.00 0.00000 0.00000 0.00000 0.00000 0.01429 0.00000 0.00000 0.00000 0.00000 59 86IRJ/06 29.50 5.13 OOS 2.00 0.00000 0.00000 0.00000 0.00000 0.00857 0.00000 0.00000 0.00000 0.00000 60 86IRJ/07 29.50 5.13 SIL 2.00 0.00000 0.00000 0.00000 0.00000 0.00857 0.00000 0.00000 0.00000 0.00000 61 86IRJ/08 29.47 5.17 SIL 13.75 0.01714 0.02000 0.06000 0.00000 0.06000 0.00000 0.00000 0.00571 0.00000 62 861RJ/15 29.46 5.21 SIS 5.00 0.00000 0.02000 0.08300 0.00000 0.20300 0.00000 0.00000 0.04600 0.00000 63 861RJ/14 29.46 5.22 SIS 1.00 0.00000 0.00000 0.00571 0.00000 0.14571 0.00000 0.00571 0.01429 0.00000 64 861RJ/59 29.52 5.37 SAN 6.00 0.00000 0.01143 0.01429 0.00000 0.01429 0.00000 0.00857 0.00000 0.00000 65 861RJ/60 29.54 5.47 ROS 9.00 0.00000 0.02000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 66 861RJ161 29.47 6.00 SAN l.75 0.00286 0.06286 0.04571 0.00000 0.05714 0.00000 0.00000 0.00000 0.00000 67 861RJ162 29.44 6.02 SHS 2.75 0.00286 0.09143 0.01714 0.00000 0.00000 0.00000 0.00286 0.00571 0.00000 68 861RJ/63 29.44 6.15 SAN 3.00 0.00857 0.22000 0.06000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 .... 69 861RJ/64 29.57 6.31 ROS 3.00 0.00857 0.27143 0.05143 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 tv 70 861RJ165 29.57 6.31 ROS 7.50 0.00286 0.19429 0.06571 0.00000 0.00000 0.00000 0.00000 0.03429 0.00000 -.J ------_. ------SITE Long. Lat. Substrate Depth(m) G. cristata G. curta G. alata G.lenis G.coheni G. woutersi G. wilsoni G. downingi G.simpiu 71 861RJ/66 30.12 6.33 RKS 24.50 0.00000 0.13429 0.05714 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 72 861RJ/67 30.23 6.46 SAN 3.00 0.00000 0.54857 0.00000 0.00000 0.00571 0.00000 0.01143 0.00000 0.00000 73 861RJ/68 30.25 6.50 SAN 3.00 0.00000 0.11429 0.01143 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 74 861RJ/69 30.25 6.50 SAN 9.25 0.00000 0.00000 0.00571 0.00000 0.00286 0.00000 0.00000 0.00000 0.00000 75 861RJnl 30.31 7.02 SHS 80.00 0.00000 0.00000 0.00000 0.00000 0.15429 0.00000 0.00000 0.40857 0.00000 76 861RJn2 30.31 7.10 MSR 12.25 0.01714 0.09714 0.03429 0.00000 0.00571 0.00000 0.00000 0.00000 0.00000 77 861RJn3 30.34 7.24 SHL 9.00 0.00000 0.02571 0.06571 0.00000 0.04857 0.00000 0.01143 0.00000 0.00000 78 861RJn4 30.35 7.27 MSS 30.50 0.00570 0.01710 0.16570 0.00000 0.71400 0.00000 0.01140 0.03430 0.00000 79 861RJn5 30.47 7.49 MSR 24.50 0.00000 0.04000 0.14286 0.00000 0.08857 0.00000 0.00000 0.01714 0.00000 80 861RJn6 30.47 7.52 MSR 33.50 0.02571 0.10857 0.14571 0.00000 0.01429 0.00000 0.01143 0.02286 0.00000 81 861RJ/80 30.30 8.20 RKS 15.25 0.00000 0.12286 0.06000 0.00286 0.00000 0.00000 0.00000 0.00000 0.00000 82 861RJn8 30.38 8.29 SHL 28.00 0.00290 0.09710 0.26570 0.00000 0.09430 0.00000 0.00000 0.01710 0.00000 83 861RJn9 30.38 8.29 SHS 27.50 0.00286 0.14286 0.29143 0.00000 0.10000 0.00000 0.01714 0.01714 0.00000 -

Substrate Designations

SIL Silt SIS Silt and Sand SAN Sand SAS Sand and Stromatolites SHL Shells SHS Shells and Sand SSS Sand. Shells and Stromatolites OOS Oolitic Sand ROS Rocks and Stromatolites RKS Rocks MSR Mixed Sand and Rocks MSS Mixed Sand and Stromatolites

N=350/sample Value is relative % of species in sample: (number of species present/350) I-' ~ 00 129

Figure 2.5c: Presence/absence data for Gomphocythere species in Lake Tanganyika for localities represented in map in Figure 2.5a. Latitude and longitude are indicated in degrees and minutes. Substrate designations are given below. Depth is indicated in meters. Figure2.5c LOCATION MAP SPECIES DISTRIBUTION

SITE Long. Lat. Substrate Deptb(m) G. cristata G. curta G. alata G.lenis G. ccheni G. woutersi G. wilsoni G. downingi G.simplu J

84 SITE 138 29.20 5.40 NA plankton X X X X X X X X X I

85 LTI9lf17 29.25 3.58 SAN 5 X X X X I 86 LT/91f20 29.25 3.58 SIS 50 X X X X I 87 LTJelf21 29.25 3.58 SAN 50 X X X X X X X X 88 LTI91f27 29.25 3.58 SIS 8 X X 89 LTI9lf12 29.28 4.60 PLA 1 X X X X 90 LT/9lf14 29.28 4.60 SIS 45 X 91 LTI9lf16 29.28 4.60 SIL 50 92 LT/9lf8 29.33 4.12 SAN 7 X X X 93 LT/91f5 29.34 4.17 SAN 12 X X X 94 LT/9lf6 29.34 4.17 MSR 20 X X 95 LT/9117 29.34 4.17 SIS 60 X X X X X X X X 96 LTI92163 29.36 4.42 SIL 16 X X X X X 97 LT/9lf3 29.39 4.26 SAN 6 X X X X 98 LTI92133 29.43 6.02 RKS 4 X X X 99 LTI92134 29.44 6.01 RKS 4 X X X X X 100 LTI92129 29.44 6.14 SHS 25 X X 101 LTI92156 29.45 5.27 SIS 18 X X 102 LTI92157 29.45 5.27 SAN 8 X 103 LTI92158 29.45 5.27 SAN 2 X X X X X 104 LTI92130 29.45 6.01 RKS 4 X X 105 LTI92159 29.45 6.27 SAN 4 X X X X 106 LTI92160 29.45 6.27 SHS 25 X 107 LTI92161 29.45 6.27 SSS 15 X X 108 LTI92162 29.45 6.27 SSS 7 X X 109 LTI92167 29.45 6.27 SIL 10 X X 110 LTI92152 29.53 5.40 SIS 7 X X 111 LT/92136 29.54 5.55 SHL 12 X X I-' 112 LTI92137 29.54 5.55 SHL 25 X X X CI:) 113 LTI92138 29.54 5.55 SHL 30 X o ~T/9213~ ~9.54 5.55 SIL 125 X X J.14 ~- - - - - SITE Long. Lat. Substrate Depth(m) G. cristata G. curta G. alata G.lenis G. roiu!ni G. woutersi G. wilsoni G. downingi G. simplex 115 LTI92143 29.54 5.55 SIL 12 X X X X X 116 LTI92144 29.54 5.55 SHL 12 X X 117 LTI92147 29.54 5.55 SAN 0 X X 118 LTI92150 29.56 5.48 SIL 5 X 119 LTI92125 30.08 6.27 SAN 65 X 120 LTI92126 30.08 6.27 SHL 40 121 LTI92123 30.16 6.35 SIL 90 X X X X 122 LTI92124 30.16 6.35 SAN 20 X X X 123 LTI92115 30.30 7.07 SAN 10 X X X 124 LTI92114 30.31 7.25 SHL 35 X X X X X X X X 125 LTI92120 30.32 6.52 SAN 3 X X X X 126 LTI92121 30.32 6.52 SHS 18 X X X X 127 LT/9213 30.34 7.27 MSR 4 X X X ,].27 SIL 50 X_ X X 12~ '--- LT/9214_ ~O.~ ------

Substrate Designations • Substrate assignments are not necessarily equilavalent to the substrate designations from Table 2.5b

SIT.. Silt SIS Silt and Sand SAN Sand SAS SandandSbromatolites SIlL Shells SUS Shells and Sand SSS Sand, Shells and Stromatolites OOS Oolitic Sand ROS Rocks and Sbromatolites RKS Rocks MSR Mixed Sand and Rocks MSS Mixed Sand and Sbromatolites PLA Macrophyte Meadows

X= present in sample No data available for relative % calculations f-O CI:) f-O 132 curta, G. coheni, G. woutersi, and G. downingi) that do not show significant difference in occurrence between substrates (Table 2.1). Distributions of the relative abundance of each Gomphocythere species with respect to the standardized samples show that species found within specific subclades share similar substrate distributions, but each member species has a different substrate in which it is most abundant (Table 2.1). For example, Gomphocythere alata, is a member of subclade B, and occurs on almost every substrate type, except for macrophyte meadows (PLA) , oolitic sand (008), and sand, shells and stromatolites (888). It is most abundant in the shelly substrate (8HL), mixed sand and rock (M8R), and shelly sand substrate (8H8). Gomphocythere cristata, also a member of subclade B, occurs in the same substrates as G. alata (Table 2.2). G. cristata are most abundant on mixed sand and rock (M8R). Gomphocythere n.sp. "coheni", another member of subclade B, shares a similar distribution with the other two species, but is most abundant on mixed sand and stromatolites (M88) (Table 2.2). Whereas these three species in the same subclade share similar substrate distributions, in terms of presence and absence (previously shown in Table 2.1), they each reach their maximum abundance on different substrates, suggesting niche specificity within subclades. These abundances are confirmed by the results from the peA and cluster analysis. A second interesting substrate-species correlation occurs with Gomphocythere curta, a species occurring within subclade D (see Figure 2.2) that is present in almost every environment, but is most abundant on rocks and stromatolites (R08). On this substrate, the G. curta carapaces 133

Table 2.1: Results from One-Way General Linear Model (GLM) for pooled substrate data. Substrates were pooled according to composition and grain size in order to strengthen the power of the statistical test. Rare taxa were excluded. F-values, p-values, DF=degrees of freedom are indicated. 134

Table 2.1

Results of One-Way General Linear Model (GLM) Pooled Data Substrates Pooled Substrate 1 Substrate 2 Substrate 3 Substrate 4 Substrate 5 SAN ROS MSR SIS SHL SAS RKS MSS SIL SHS OOS SSS

No significant Taxa F-value P-value DF Variation G. alata 7.26 0.0001 4 G. cristata 2.44 0.0555 4 * G. curta 2.61 0.0437 4 G. coheni 3.32 0.0156 4 G. woutersi 1.19 0.325 4 * G. wilsoni 4.81 0.0019 4 G. downingi 1.76 0.1474 4 * 135

are generally tuberculate, retaining all other morphological features of the species, whereas on sand and shell lags, their carapaces are more elongate. It is likely that these morphologies reflect ecophenotypic variations within this species. Gomphocythere lenis and Gomphocythere simplex are members of subclade E and occur in a narrower range of environments, primarily rocks (RKS), and sand (SAN) for G. lenis and sand and silt for G. simplex. They are rare and occur only at a few sites. The sister species, Gomphocythere n.sp. "downingi" and G. n.sp. "wilsoni" co-occur in similar substrates of sand and rocks (MSR), mixed sand and stromatolites (MSS), rocks and stromatolites (ROS), silt and sand (SIS), sand (SAN), shells (SHL) and shells and sand (SHS). G. n.sp. "downingi" is most abundant in shelly sand (SHS), whereas G. n.sp. "wilsoni" is most abundant in shells (SHL) and more generally on coarser substrates (Table 2.2). R-mode cluster analysis groups taxa into three main associations based on the substrate in which they are most abundant (relative to standardized samples of all ostracods at these localities) (Figure 2.6). This R-mode clustering with pooled data was run at a coarser level than the individually defined relative abundances per substrate and the MANOVA to determine the significance of their distributional pattern. This clustering is not a replicate of the phylogeny (i.e. given in Fig. 2.2), but merely shows substrate affinities among the common species. The first cluster (Cluster A) groups the species: Gomphocythere cristata, G. wilsoni, and G. woutersi together. Gomphocythere cristata and G. woutersi both 136

Table 2.2: Substrate affinities for Gomphocythere species in Lake Tanganyika. Rare taxa were excluded. Substrates in which each species is most abundant are indicated. Other substrates in which each species is common are also indicated. Relative abundances are calculated with respect to all ostracods found in each standardized sample (n=350) and confirmed by T tests (at p<0.05) of pairwise comparisons between means of each substrate distribution as calculated in the One-way GLM MANOVA. Table 2.2

Mean Frequency of Species per Substrate

Species Substrate G. cristata G. curta G. alata G. coheni G. woutersi G. wilsoni G. downingi SIL(n=11) .002 (±.005) .004 (±.007) .012 (±.019) .012 (±.O18) 0 0 .045 (±.075) SIS (n=11) .001 (±.002) .005 (±.009) .021 (±.034) .042 (±.068) 0 .003 (±.007) .038 (±.069) SAN (n=12) .001 (±.014) .109 (±.172) .017 (±.022) .023 (±.048) .013 (±.045) .004 (±.006) .018 (±.06) SAS (n=3) .002 (±.003) .013 (±.016) .078 (±.128) .005 (±.002) 0 0 0 SHL(n=3) .003 (±.003) .041 (±.05) .178 (±.102) .084 (±.031) 0 .017 (±.021) .025 (±.029) SHS (n=6) .009 (±.014) .119 (±.103) .118 (±.137) .047 (±.065) .004 (±.Ol) .004 (±.007) .082 (±.16) SSS (n=2) 0 0 0 0 0 0 .009 (±.012) OOS(n=2) 0 0 0 .061 (±.075) 0 0 0

MSS (n=2) .003 (±.004) .009 (±.012) .083 (±.117) .423 (±.412) 0 .009 (±.004) .021 (±.018) I MSR(n=8) .026 (±.043) .096 (±.034) .138 (±.072) .014 (±.031) 0 .001 (±.004) .011(±.019) ROS (n=12) .009 (±.014) .126 (±.08) .069(±.064) .001 (±.003) .007 (±.017) .0005 (±.002 .004(±.01) RKS (n=10) .01 (±.014) .064 (±.045) .042 (±.043) .005 (±.Ol) .001 (±.002) 0 .001 (±.002)

I--' c,., -..:] 138

Figure 2.6: R-mode cluster analysis on species grouped according to substrate distribution. Black circles indicate the relative abundance of each species with respect to the total ostracod fauna for that sample in each substrate. Clusters are indicated by the letters: A, B, C. Phylogenetic sub clades are indicated by: triangle= phylogenetic subclade B; circle= phylogenetic sub clade C; square= phylogenetic subclade D. A =member of phylogenetic subclade B Relative % of species in substrate member of phylogenetic subclade C o = •.110-.500% 8.011-.050% •.001-.005% member of phylogenetic subclade D [] = •.060-.109% •.006-.Q\0%

Average Distance Between Clusters 2.0 1.0 0.0 I I I I I I II I I SIL SIS SAN SAS SHL SHS SSS OOS MSS MSR ROS RKS ...------fGomphocythere alata A. • • • 1_- • • • • I..-----Gomphocythere curta [] • • • • • • • • • •• ...-----4rJ.omphocythere downingi 0 • • • • • • • • • • 1..----4.J.i omphocythere coheni A • • • • • • • • omphocythere woutersi •- . A • • • • omphocythere wilsoni 0 • • • • • • • omphocythere cristata A. • • • • • A • • • • • Figure 2.6 .... CI,j c.o 140

belong to the phylogenetic sub clade B, whereas G. wilsoni belongs to subclade C. Gomphocythere cristata is most abundant in mixed sand and rocks (MSR), G. wilsoni is most abundant in shells (SHL) and G. woutersi is most abundant on sand (SAN). The second cluster (Cluster B) contains

G. downingi and G. coheni which are most abundant in stromatolite, shell and silt dominated substrates and are members of different subclades. Cluster C contains G. alata and G. curta which frequently occur on shells and rocks and sand. They too are members of different subclades. This data supports the pattern previously documented with relative abundances and MAN OVA , because it demonstrates that within individual phylogenetic subclades, each species has an affinity for a different substrate, even though the members of the subclade may occur (present, but in low numbers) in similar substrates to each other. In other words, substrate affinity (defined as the substrate(s) in which a species is most abundant) is not phylogenetic ally constrained within a subclade but follows the same (homoplasious) pattern as many of the morphological character distributions. Principal component analyses of species according to substrate was done to test whether there is significant variance in the data. The coordinate system of the first three principal components explains 43.04% of the variation (Fig. 2.7 and Table 2.3). The species cluster according to similar substrate affinities. The species group consistently with their substrate affinities along PC1 (SAN, SHS, MSR, RKS, ROS strongly positive), and PC2 (MSS, SIS strongly positive) confirming the results from the cluster analysis, that individual species from the same subclade 141

Figure 2.7: Principal component ordination of species distribution according to substrate affinities. Species are labeled. A) PC1 versus PC2; B) PC1 versus PC3. Each point represents a species. Principal Component 1 contains 43.04% of the variation, Principal Component 2 contains 25.15% of the variation and Principal Component 3 contains 20.97% of the variation. 142

A

?coheni

.... G.alata ?downingi

G. wilsoni /. L~ G. cristata G. curta --..a. woutersi

Principal component 1 (43.04%)

B ...... 3.0 ~ t- O) G. alata 0 ./ C\l 2.0 '-'

~..., 1.0 G.wilsoni I=l G. cristata OJ ! I=l ?G. woutersi 0 0.0 Q.. Il.. G. downingi S ". 0 G. coheni C) -1.0 ~ ,.,- -Q...... G. curta C) .....I=l -2.0 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 Pot'"' Principal component 1 (43.04%) Figure 2.7 143

Table 2.3: Eigenvalues from Principal Component Analysis of 11 variables (substrates) and the Gomphocythere species' distributions. Table 2.3

Eigenvalues of the Covariance Matrix Total=0.52686988 Average=0.04789726

1 2 a 4 5 6 7 8 9 10 11 Eigenvalue 0.2268 0.1325 0.1105 0.0558 0.0010 0.0002 0.0000 0.0000 0.0000 0.0000 0.0000 Difference 0.0943 0.0221 0.0546 0.0548 0.0008 0.0002 0.0000 0.0000 0.0000 0.0000 Proportion 0.4304 0.2515 0.2097 0.1059 0.0020 0.0005 0.0000 0.0000 0.0000 0.0000 0.0000 Cumulative 0.4304 0.6820 0.8917 0.9976 0.9995 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000

Principal Comoponents Correlation Matrix Substrate PC1 PC2 pca PC4 PC5 PC6 MSS -0.28 0.8974 -0.2253 -0.02559 -0.0086 0.0026 SAN 0.83623 0.0463 -0.5266 -0.05284 0.13359 0.026 SAS 0.5523 0.1777 0.8135 -0.03944 0.00364 0.0058 Sm.. 0.45049 0.5314 0.70891 -0.07496 0.05501 -0.0589 SHS 0.08649 0.3139 0.14752 0.36211 0.0233 0.0052 sn. -0.0443 0.3683 0.03657 0.92789 -0.0094 0.0041 SIS -0.0794 0.8723 0.05904 0.47832 0.0193 0.0114 SSS 0.7076 0.1105 -0.6556 0.24031 -0.0328 -0.0231 MSR 0.86615 0.2932 0.38413 -0.09534 -0.0704 0.0472 RKS 0.96698 0.0534 -0.2058 -0.13764 -0.0222 -0.0178 ROS 0.97931 -0.0468 -0.1652 -0.10271 -0.0284 0.0102 ..... ~ ~ 145

can be found in similar substrates as their fellow subclade members, but are most abundant in substrates that are shared by species outside of their subclade. The variation contained in PC3 (SAS and SHL strongly positive) further supports the cluster pattern which groups Cluster A and Cluster B together. PCA indicates that the variation of species in terms of their substrate occurrences is consistent with the R-mode cluster analysis.

Substrates Mapped onto Phylogenetic Trees Species clustering according to substrate affinities were mapped onto the phylogenetic reconstruction of endemic Gomphocythere from Lake Tanganyika. With the exception of the sister species Gomphocythere len is and G. simplex, all species are present in most substrates, except for oolitic sand (OOS), sand, shells and stromatolites (SSS)(G. coheni occurs on OOS and G. downingi occurs on SSS). Substrate affinities were established by first determining the substrate in which the species is most abundant via relative percentage of the total ostracod fauna (n=350) at that site. This relationship was then tested using MANOVA to determine if the distribution of the species is random (null hypothesis). I further explored and defined substrate ranges using cluster analysis and PCA which grouped species with similar distributions and affinities together Table 2.2. Gomphocythere n.sp. "downingi" and G. n.sp. "wilsoni" share similar substrate distributions, but G. n.sp. "downingi" is most abundant in SHS and G. n.sp. "wilsoni" is most abundant in SHL, demonstrating possible specializations for these particular habitats. This is also the case for G. lenis and G. simplex, as well as for G. alata and G. cristata. In all of 146

Figure 2.8: Phylogenetic reconstruction for Gomphocythere species in Lake Tanganyika and substrate distribution. White squares designate species absence, lightest gray represent mean frequencies between .001- .005%; light gray represents mean frequencies between .006-.010%; gray represent mean frequencies between .011-.050%; dark gray designates mean frequencies between .060-.100% and black representing mean frequencies between .110-.500%. Where quantitative data were not available (designated by *), + represents present, ++ represent common­ abundant. 147

* SAN * * * * ROS MSS MSR RKS OOS SSS SAS SIS SIL SHL SUS

I< c

_ .110-.500% _ .060-.109% _ .011-.050% I'ml .006-.010% o .001-.005% lSI Absent ++ Common-Abundant + Present * Non-endemic; not comparable data Figure 2.8 148

these cases, substrate distributions overlap in these subclades, but individual members in the subclades have their specific affinities for different substrates, indicating niche specialization or partitioning. The results are summarized in Figure 2.8.

The relative abundances of Gomphocythere speCIes In each substrate are summarized in Tables 2.1 and 2.2. From these distributions, it appears that some species are more abundant in specific substrates than others. Some substrates have not been extensively sampled because they are relatively rare compared to the other samples and grab samples were not uniform for each site. However, the counts of ostracods from these individual samples were uniform (n=350) and randomized.

Depth Distributions Depth range data for Gomphocythere species show that some species have broad ranges and others narrow ranges. In all groups examined, distributions of the relative abundance of individual species with respect to the total ostracod fauna for that depth range often spans a broad range of depths (just as in the substrate distribution), but there are specific depths at which each species is most abundant. More importantly, just as with substrate, depth affinities appear to differ among those members of closely related species (i.e. subclades). In sub clade B, both G. alata and G. cristata are found at all depths up to 50 meters, but the two species reach their maximum abundances at slightly different depths (20-29 meters vs. 40-49 meters, respectively). 149

Gomphocythere n.sp. "coheni" is found at all depths, but is most abundant in deeper waters (30-39 meters). Gomphocythere. n.sp. "woutersi" is concentrated at shallower depths, particularly around 10 meters. Comparing the four species' depth distributions, it is apparent that each has a distinct depth range which is the same pattern that I observe in substrate differentiations (Table 2.4). In subclade C, Gomphocythere n. sp. "downingi" and G. n.sp. "wilsoni" show different depth affinities. Gomphocythere n.sp. "downingi" is most abundant relative to all ostracods found at varied depths. G. lenis and G. simplex occur in such small sample sizes that any pattern may not be distinguishable. Gomphocythere curta, which is distributed in many different substrates, is also distributed at many different depths, but is particularly abundant in waters between 20-29 meters. When looking at individual subclades, each species can be found in greatest abundance at narrow depth ranges that are different from sister species within that subclade. This is consistent with the substrate distribution data and also with the concept of niche segregation or specialization for Gomphocythere species. Whereas the expected pattern might be one that has an entire subclade with a specific substrate or depth affinity, this pattern, instead, shows niche partitioning within the subclade (Figure 2.9). The high homoplasy and patterns of independent character acquisition are consistent with the substrate and depth distributions (see discussion, Chapter 1). 150

Table 2.4: Depth affinities for Gomphocythere species in Lake Tanganyika. Rare taxa were excluded. Depth at which each species is most abundant is indicated. Other depths at which each species is common are also indicated. Relative abundances are calculated with respect to all ostracods found in each standardized sample (n=350) and confirmed by T tests (at p<0.05) of pairwise comparisons between means of each depth distribution. Table 2.4

Mean Frequency of Species per Depth Range

Species

Depth G. cristata G. curta G. alata G. coheni G. wilsoni G. downingi G. woutersi I

0-9 (n=41) .002 (±.003) .069 (±.132) .027 (±.053) .019 (±.043) .001 (±.003) .005 (±.017) .004 (±.024) i

10-19 (n=10) .013 (±.016) .086 (±.072) .083 (±.06) .02 (±.037) .004 (±.013) .009 (±.019) .001 (±.002) I 20-29 (n=12) .008 (±.Ol1) .103 (±.073) .144 (±.111) .025 (±.042) .003 (±.007) .008 (±.009) .003 (±.007) . 30-39 (n=9) . 011 (±.016) .029 (±.049) .054 (±.067) .087 (±.235) .003 (±.005) .049 (±.069) 0 40-49 (n=9) .016 (±.042) 0.061 (±.105 .055 (±.07) .025 (±.053) .003 (±.008) .073 (±.094) 0 80 (n=l) *0 *0 *0 *0.154 *0 *0.409 *0

* Note: only one sample

...... 01 152

Figure 2.9: Phylogenetic reconstruction for Gomphocythere species in Lake Tanganyika and depth distribution. White squares designate species absence, light gray represent mean frequencies between .001- .010%; gray represents mean frequencies between .011-.090%; and black represents mean frequencies between .100-.900%. Where quantitative data were not available (designated by *), + represents present. 153

50 * * * * *

40 ---~ ~Q) 30 S "-' ..s 20 fr Q 10 + 0 .., ,~ ".;j 'c tU 0 «='" tU (;j a'" 1:: :a '" (;J ,r:J '" §. ::s 1:: 'c 0 ~ u u 0 ~ d d d d d d d d d d d d d d

•.100-.900% _ .010-.090% I.)}I .001-.009% D Absent * Non-endemic; not comparable data Figure 2.9 154

Body Size Distribution I mapped body size distributions onto the phylogenetic tree (Figure 2.10). Species in particular subclades have similar sizes. The smallest sized Gomphocythere species in Lake Tanganyika are members of subclades Band D, whereas the larger sized species occur within sub clade C. There appears to be little or no correlation between size and substrate affinities, since species that are dominant in silt and in deeper waters are similar in size to those that are dominant in rocks and sand in shallower waters (Figure 2.10). It should be noted that these comparisons are relative ones only for Gomphocythere species in Lake Tanganyika and no comparisons outside of the lake were made. The issue of size versus other ecological variables, such as fish predation, was not examined here, but could be the focus of future work. It should be noted that the largest Gomphocythere in Lake Tanganyika, G. n.sp. "wilsoni" and G. n.sp. "downingi" are found on shell dominated substrates. A hypothesis that could be tested in the future is whether or not these species use empty shells as refugia from fish predators.

Ecologically Correlated Morphologies Synapomorphies and autapomorphies can represent adaptive morphologies for a clade or species. However, for an adaptation to be documented, the character has to be shown to provide current utility to the organism and to have been generated historically through the action of natural selection for its current biological role (Baum and Larson, 1991). This definition leads to a methodology requiring independent 155

Figure 2.10: Distribution of Gomphocythere species size as mapped onto the favored phylogenetic reconstruction. 0=<.5 Ilm; 1=>.5Ilm. 156

& : '0, '§ c:::: : '2 Ql '2 '2 Ql '5 0 ..c:: 0 .$ 0 0== 0 p ctI = ~ ~ X 1a Ql ctI ci. ci. ci. ci. en a. ctI en en en en en '2 1:: 1a 'L:- E :J (ij 0 c:::: c:::: c:::: c:::: ~ 'iii 0 CJ ci ci ci ci ci ci ci ci D D D liD m D D D

size unordered

c::J 0 I11III11

Figure 2.10 157 measurement of utility. That was not possible in my study. Therefore, I will discuss only the acquisition of various traits and the corresponding changes in selective regimes (i.e. environment). There are at least four characters that appear to be correlated to specific subclades and environmental parameters when distributed over the phylogenetic tree. These include the presence and characteristics of nodes and alae on the carapaces and the presence of specific features on the first antenna and thoracic leg. Nodes are protuberances on the valve surface that have developed into distinct lobate structures (Fig. 2.11a). Earlier work has suggested that theses features are linked to carapace rigidity and stability in coarser grained substrates (Carbonnel, 1983; Carbonnel et aI., 1988). They are present and apparently independently acquired in Gomphocythere n.sp. "coheni", and G. alata (Fig. 2.11a). The acquisition of this trait, as present in these species, corresponds to the dominant substrate affinities of these two species. Both are most abundant in the coarser grained substrates of mixed sand and stromatolites (MSS) and shells (SHL) (Figure 2.11a). Within the soft anatomical features, the number of terminal claws on the first antennae (AI) appear to correlate with substrate. Terminal claws are used for digging or for movement along a substrate surface. The number of terminal claws varies from 2 to 4 on all Gomphocythere species. Gomphocythere n.sp. "coheni" and G. n.sp. "downingi" each have 4 claws and are most abundant on either shells or mixed sand and stromatolites (Figure 2.11b). Gomphocythere alata and G. n.sp. "wilsoni" have 3 terminal claws and both are most abundant in shell lag substrates. 158

Figure 2.11a- Character state distribution for node number mapped onto the phylogeny of Lake Tanganyikan Gomphocythere species. Substrate affinities (the substrate in which the species is most abundant) are indicated in the boxes above the terminal taxa. The character is indicated in the picture above the cladogram. Character states are as follows: 0= nodes absent; 1=1 node present; 2= 4 nodes present. 159

node ~ c = = 'in... c tn Q) °2 °2 °2 0 Q) ::J .c -0 == .E2 0 ;p0 p It! =(.J === .£9 >< It! Q) It! It! - c.. c.. "'5 c.. c.. en a. It! en en en Ol en en °2 t iii 0;::- C E ::J 10 (.J c: c: It! c: c: !!! °en (.J CJ CJ cd CJ 0 • 0 0

unordered

r=:J 0 l:o:o:o:o:oJ 1 1 node _ 3 4nodes

Figure 2011a 160

Figure 2.11b- Character state distribution for Al terminal claws mapped onto the phylogeny of Lake Tanganyikan Gomphocythere species. Substrate affinities (the substrate in which the species is most abundant) are indicated in the boxes above the terminal taxa. The character is indicated in the picture above the cladogram. Character states are as follows: 0= claw absent; 1=1 claw; 2= 2 claws; 3= 3 claws; 4= 4 claws. 161

MSR SHL SHS SHL RKS ROS IIi 1111 1111 .. 'ED .. c: .. '2 '2 '2 Q) 3: ~ .s= 0 0 ;p .~ .m =0 = cu ci ci ci .;::-UJ UJ UJ UJ 0 C C C

unordered

c::::J 0 ':·:·:·:·:·1 1 1 claw 1:-:-:-:-:-:1 2 2 claws Figure 2.11h IIIIlI 3 3 claws _ 44claws 13 equivocal 162

Gomphocythere cristata and G. simplex both have 2 claws, plus small setae and both are most abundant in mixed rock and stromatolite and rocks substrates.

DISCUSSION Gomphocythere distribution patterns have been examined at two levels to better understand their speciation patterns and mechanisms in Lake Tanganyika. On one level, biogeographic distribution of Gomphocythere species throughout Africa demonstrates that closely related species are found in geographically distant lake systems, which defies a simple model of adjacent lake basin immigrations. It further suggests that dispersal for some Gomphocythere species has been rapid relative to speciation; and their dispersal events could have occurred at any time, with extinction in intervening lakes. On a second level, within Lake Tanganyika, the lack of phylogenetic constraint on substrate and depth distribution across the Gomphocythere phylogeny, including within subclades, appears consistent with the homoplastic distribution of many independently character states for individual species.

Gomphocythere Distributions Dispersal strategies for Gomphocythere remain enigmatic. The distribution patterns of Gomphocythere species in lakes throughout Africa suggest that these species can disperse over long distances. Gomphocythere ortali occurs in both South Africa and Lake Victoria, with no apparent corridor for dispersal via rivers or connecting lake systems. 163

Lake Kivu, the lake most likely to be connected to Lake Tanganyika through time, has a Gomphocythere faunal association different from that of Lake Tanganyika. In fact, there is little faunal similarity between the two lakes. This is perhaps because prior to the Pleistocene, Lake Kivu drained to the north. If a simple model of connections of adjacent lake basins over time were applicable, then it would be expected that these two lakes would share Gomphocythere species. This is not the case. An alternative means for dispersal between lakes would be via bird transport. If the migration pattern of birds over this geographic region was thoroughly examined, perhaps a correlation between relevant bird migratory routes and ostracod biogeographic patterns could be found. Within Lake Tanganyika, most Gomphocythere species are widely distributed, however each species is most abundant in particular habitats. The fact that almost all species can be found at some frequency in all substrates and at all depths suggests that species can disperse over barriers but favor specific conditions (e.g. substrate and depth). One way that these species might be able to disperse is via swimming. Strong swimming appendages are present on all Gomphocythere species, which would enable them to move from place to place.

Speciation Barriers Whether or not there was a single invasion or multiple invasions of the lake by one or more Gomphocythere species does not explain the entire diversity pattern of the clade. Once in the lake, an ancestral population of Gomphocythere would be affected by the ecological conditions within the 164

Tanganyikan system. Dispersal abilities, substrate and/or depth affinities of the original colonizing species might play an important role in the derivation of new species. Gomphocythere species have been demonstrated above to have affinities for certain substrates and depths, although species that are only found in one or two types of substrates or depths are rare (G. lenis and G. simplex). Physical barriers have been shown to occur in the lake basin throughout the basin's history (Coulter, 1991) (Figure 2.1), but based on the distributional patterns of extant Gomphocythere species, it is clear that most species are widely distributed throughout the lake. This is so even though they are brooding organisms, which may limit their dispersal abilities at least over short geologic intervals (Cohen and Johnston, 1987).

Substrates Niche specificity or partitioning in substrate range could be the result of competitive exclusion between species after sympatric speciation, or could be 1) constrained by competition with other species outside of the clade, or 2) allopatric speciation over very short niche space distances within the same subclade for dominance in that habitat. In addition, species that occur at the base of some of the subclades appear to be generalists, whereas other species have specific substrates in which they are most abundant. For example, the generalist Gomphocythere curta occurs in many different substrates and appears at the base of subclade D, while G. n.sp. "downingi" and G. n.sp. "wilsoni" are more narrowly distributed to specific substrates. These distributions 165

indicate that, whereas some of the synapomorphies that define sub clade D may be adaptations for certain types of substrates that are shared by all members, individual autapomorphies defining species have likely been derived from associations with specific substrates and corresponding micro-environments. This pattern is corroborated by the observed in other subclades such as Band C.

Depth The pattern of depth distributions, like the pattern of substrate distributions within subclades suggests partitioning of the environment according to individual species' affinities and may also connote possible mechanisms described above for substrates between species that occupy similar niche spaces. In addition, it is significant to note that the species that are most abundant at all depths (i.e. G. curta) are situated at the base of certain subclades (not the entire phylogeny), indicating that these species are most broadly distributed and most abundant, with the most generalized (plesiomorphic) morphologies.

Ecologically Correlated Morphologies The results suggesting a correlation of node and alae development to environmental distributions are interesting because they suggest possible adaptation to environments. Autapomorphous node characters are consistent with substrates in which a species is most abundant, even though the species may not occur within the same subclade. Node and alae development are both recognized as features that strengthen the 166

carapace and give it stability (Van Morkhoven, 1962). They occur on species that are most abundant in coarse grained substrates of mixed rock, sand or shells. For this argument to be further substantiated, more study of the function and improved fitness by acquisition of the character state is required. The distribution of the number of terminal claws on the Al appendage over the phylogeny also appears to be consistent with coarse grained substrates. The number of claws and the robust nature of the animal suggests that this appendage is used primarily for digging and locomotion through coarse-grained substrates. Species that are most abundant in finer grained substrates, or that are planktonic would be expected to show reduced appendages (Moore, 1961). These ideas still need to be confirmed by field observation of Gomphocythere.

Speciation Mechanisms of Gomphocythere in Lake Tanganyika The purpose of 1) integrating biogeographic distributions and phylogeny, 2) mapping endemic species onto the phylogeny, 3) establishing the pattern of substrate and depth distribution for each species and mapping it onto the phylogeny, and 4) examining individual character distributions and their correlations with ecological variables, was to provide a basis for better understanding the speciation mechanisms of Gomphocythere in Lake Tanganyika and to generate new testable hypotheses of Gomphocythere speciation. Whether or not Gomphocythere speciation in Lake Tanganyika is consistent with other 167 ostracod groups' patterns is also an important area of inquiry and could be addressed in future work.

Single Invasion Scenario If a single radiation produced the diversity of Gomphocythere species observed in Lake Tanganyika, it would be expected that there would be a single endemic clade within the phylogenetic tree. The distribution of non-endemic species within endemic subclades throughout the phylogenetic tree indicates that it is likely that there was more than one invasion of the lake and that the Gomphocythere species in Lake Tanganyika are not the result of a single radiation, but of more than one radiation. Some species in these bursts probably went extinct and might be uncovered in a longer fossil record, if and when it becomes available.

Multiple Invasion Scenarios Immigration An immigration scenario explaining the origin of species flocks or clusters requires that dispersal of a sister population from a lake with subsequent reproductive isolation and reintroduction to the lake through immigration occur. If this were the case, then many species within the lake should have a sister group outside of the lake (Smith and Todd, 1984). The presence of some Gomphocythere species throughout Africa and the restriction of other Gomphocythere species to specific lake systems could be explained by such multiple invasions. Gomphocythere species in shallow lakes are likely to be dispersed via wading birds. However, wide 168 distribution of Gomphocythere species might also be explained by their age alone. If they originated long ago in a single lake or in a series of connected drainages, their presence in these various lakes could be the result of geologically distant isolation and not immigration. The problem with this explanation is that some of these Gomphocythere species are found in lakes that were never connected in the geologic past.

Fluctuation in Lake Levels Another hypothesis for species flock formation involves lake level fluctuations. Low lake levels would isolate populations into separate water bodies, with different evolutionary pressures (or enough time for random mutations) such that the populations would be reproductively isolated. When the lake level rose again, the two (or more) newly evolved species would become sympatric. This theory is also compatible with the concept of allopatric speciation. Through the repeated connection and isolation, there would be opportunities for invasions and also for extinctions of species, which would open up possible new niche space for these invading species to occupy. There is abundant evidence of lake level fluctuation in Lake Tanganyika's late Pleistocene history. For example, there were low stands of at least -200 m below present at 50-45 kybp and 25-15 kybp. High lake level stands occurred between 15-5 kybp, when the Kivu basin was open to the Tanganyikan trough. Presently, lake levels are rising, but are -20 m below the maximum lake level estimates. These estimates are based on dated shorelines and pollen sequences (Jolly et al., 1994). 169

Although a fluctuation hypothesis is an appealing mechanism to invoke, the evidence needed to support this mechanism, including essential core data spanning such events, is not available. One might consider the possibility of vestigial populations, with newly evolved species, separated from each other in sub-basins of Lake Tanganyika (i.e. not having yet become sympatric) as circumstantial evidence of the fluctuation hypothesis. However, the Gomphocythere species distribution on Lake Tanganyika are well mixed between sub-basins.

Implications for the Fossil Record The implications for paleoecological interpretations for Gomphocythere and other ostracod species in the fossil record in Lake Tanganyika highlights the strong correlation between substrate and ostracod lifestyle, including specific morphologies and faunal abundances. In some cases, in substrates of equal grain size, such as mixed rock and sand and mixed rock and stromatolite, the faunal components are similar but there is a switch in the dominant species from one to another, suggesting niche partitioning within these habitats. Such observations concerning Gomphocythere species in current habitats should provide guidance when interpreting environmental and faunal change as represented in lake cores, and may provide future testable hypotheses.

Tempo and Mode of Gomphocythere Speciation There is no unequivocal evidence that supports gradualistic or explosive speciation for Gomphocythere or for ostracods, in general, in 170 ancient lakes. As aforementioned, there is less than 1000 years represented in the record of cores currently available for Lake Tanganyika, so rates of evolution were not evaluated, nor can they be evaluated from the cores presently available. General evolutionary models of ostracod evolution forwarded have been speculative. Martens (1990) proposes that there is some evidence, when contrasting the great number of limnocytherid species now present in African rift lakes versus their potential origins within the Holocene, to support rapid speciation. The scenario is that there have been a number of discrete bursts of species, interspaced by relatively long periods of stasis. However, there is no quantitative documentation or geological data to support this speciation pattern. Lake Tanganyika is -9-12 rna (Cohen et al., 1993). Unlike the haplochromine cichlid fish radiations of Lake Victoria and Lake Malawi (Sturmbauer and Meyer, 1993), the Gomphocythere radiation appears to have occurred early on in the lake's history with little evidence for cladogenesis since the Miocene. It is unrealistic to assume that the diversification of Gomphocythere and other ostracod groups is a strictly late Pleistocene or Holocene phenomena. Based upon inclusion of non-endemics within Lake Tanganyikan subclades as revealed in my study, the tempo of speciation of Gomphocythere within Lake Tanganyika might actually have been slow. The data do not support a recent and rapid radiation from a single ancestor. Instead, the topology of the tree suggests that many of these species are old (pre-Pliocene) and that there have been multiple invasions of species into the lake, some of which led to subsequent speciation. The 171

rate of speciation of the Gomphocythere clade prior to G. angulata is still unknown. Whether or not the ancestors and their descendants spread over a larger and larger geographical area, with descendant species fulfilling the same or very similar ecological roles in different locations is not known precisely. However, the pattern that a situation like that would produce would show a widely distributed ancestor with overlapping ecological distributions with different substrate affinities for younger-aged species. This is the pattern that the Gomphocythere show in Lake Tanganyika.

CONCLUSIONS The analysis of the Gomphocythere clade and its ecology and biogeography suggest that Gomphocythere diversified multiple times, or invasions of new species potentially from disjunct lakes occurred by incidental transport. This is evidenced, in part, by the occurrence of non­ endemics and endemics in the same subclades. Biogeographic distributions of all Gomphocythere species in Africa, indicate that there are no systematic corridors or connected pathways between Gomphocythere faunas from one lake to an adjacent one. In addition, the position of many of the species that occur widely in Africa, is at the base of various sub clades , suggesting that they might be older in age than the subclades above them on the tree. Their presence interspersed within the Gomphocythere phylogenetic tree also suggests multiple invasions. Gomphocythere species differ in their environmental affinities in Lake Tanganyika, as defined by their relative abundances in the various 172

substrates and depths recorded along the lake margin. Members of the various sub clades within the phylogeny are present in similar substrates, but species show some specificity in substrate and depth distributions suggesting niche partitioning. In addition, certain morphologies are correlated with ecological variables. Nodes, alae, the number of terminal claws on the first antennae and other morphologies are highly, and possibly multiply derived, and are correlated to substrate and depth affinities. These ecologically correlated morphologies are linked to these habitat affinities and distributional patterns and represent major morphological innovations in these groups, some of which are recurrent in various species. 173

APPENDIX A

PHYLOGENETIC CHARACTERS AND CODING 174

CHARACTERS

HARD PART CHARACTERS Marginal Structures: 1. Marginal ridge: O-absentj 1-smoothj 2-spinosej 3-crenulated 2. Spines: O-absentj 1-roundedj 2-sharply pointed 3. .H..e.n.d.: O-angle of bend <900 j 1- angle of bend >900 4. Ventral marginal rim: O-absentj I-present 5. Posterior Dorsal Tubercle: O-absentj I-present 6. Posterior Dorsal Flare: 0- absentj 1- present 7. Surface ornamentation: O-reticulate robustj 1- reticulate reduced 8. Frill: O-absentj 1-denticulatej 2-spinosej 3-undulatingj 4-concentric 9. Brood pouch on female: O-absentj I-present 10. Valve shape (dorsal yiew-anterior-posterior): 0- roundj 1- ovalj 2- rectangular 11. Dorsal view of valye shape: 0- conveXj 1- heart-shaped (inflated posterior)j 2- triangular shaped (inflated velate) 12. Siz.g: 0- < .5 mmj 1- .5-.8 mmj 2- .8-1.2 mm 13. Reticulation of carapace: 0- absentj 1- <= 11j 2- 12-17j 3- >= 18 14. Sieve Pore Penetration: O-does not penetratej 1- penetrates Muscle-scar Pattern: 15. Al Muscle Scar: 0- absentj 1- present 16. Md Muscle Scar: 0- absentj 1- present 17. Central muscle scar: O-serialj I-directed anteriorj 2-directed posterior 18. Central muscle scar appearance: O-squashed togetherj I-slightly spaced apartj 2-spaced widely apart External Features: 19. Tubercles: O-absentj I-present 20. Node position: O-absentj 1-anteroventralj 2-posteroventralj 3-anterodorsalj 4- posterodorsal 21. Node number: O-absentj 1-lj 2-3j 3-4 22. Maximum dorsal node size: O-absentj l-0.lmm 23. Maximum ventral node size: O-absentj 1<0.1 mmj 2>0.1 mm 24. ~: O-absentj I-present 25. ~: O-absentj I-one presentj 2- two present 26. Alae position: O-absentj 1- ventromedianj 2-ventral-dorsal 27. Alae lenlrt;h: O-absentj 1-<1/2 of carapacej 2->1/2 of carapace Interior Features: 28. Hin~e an~e: 0- acutej 1- obtuse 29. Shell Thickness: 0- thinj 1- thick 30. Rid~e on inner lamella: O-prominentj 1-reduced 31. Radial pore canals: O-absentj 1<5 per .lmm2j 2->= 5 per .lmm2 32. Sieve pores: O-in center of depressionj I-in between the pits on the valve surfacej 2-center of depression 33. Hingement: O-lophodontj 1-adontj 2-inverse lophodont Valve: 34. Male L:H ratio: 0-1.2-1.4j 1-1.4-1.75j 2-1.75-1.9 175

SOFT PART CHARACTERS Furca: 35. Size of Furca: 0- reduced; 1- as long a thoracic appendages 36. Total number offurcae setae: 0-2 setae each; 1-2 setae and 3 lobes Copulatory Organ: 37. Distal lobe on hemipenis: O-fixed; I-moveable 38. Position offurca: 0- above copulatory appendage; 1- below copulatory appendage 39. Distal lobe apex: 0- ridged; 1- smooth AI: First Antenna 40. AI-Number of joints: 0-3; 1-4; 2-5 41. AI-Number of setae on second joint: 42. AI-Number of claws: 0-2; 1-3; 2-2 plus small setae; 3-3 plus small setae; 4-4 setae 43. AI-Number spines on outer last insert(4th joint): 0-1; 1-2; 2-3; 3-4 44. AI-Number spines on inner last insert(4th joint): 0-1; 1-2; 2-3 45. AI-Setae on first segment: O-apical; I-lateral 46. AI-Number of setae 4th segment (outer): 0- 0; 1-1; 2- 2 47. AI-Setae on second segment spine: 0- half the way down on spine; 1- absent 48. AI-3rd joint setae (outer): 0-0; 1-1; 2-2 A2: Second Antenna 49. A2-Number of segments: 0-3; 1-4; 2-5 50. A2-Number of joints on long spine: 0-0; 1-1; 2-2 51. A2-Number of terminal claws: 0-2; 1-3 52. A2-Position ofterminal claws: O-all from same level; 1-3 from lower level, 1 from higher level; 2-2 from lower level, 1 from middle level, 1 from higher level 53. A2-Number of setae on inner 3rd joint: 0-1; 1-2 54. A2-Number of setae on outer 3rd segment: 0-1; 1-2; 2-3 55. A2-Number of setae on inner 3rd segment: 0-1; 1-2; 2-3 56. A2-0-1; 1-2; 2-3 Md: Mandibles 57. Mandibular palp: O-normal; I-bent knee 58. Mandibular palp setae: O-straight; I-bifurcated Mx: Maxilla 59. Maxillular palp: O-normal; I-reduced PI: First Thoracic Leg 60. PI-Position of inner apical: O-lateral; I-apical 61. PI-Number of setae on 1st joint: 0-0; 1-1; 2-2; 3-3; 4-4 62. PI-Number of setae on 2nd joint: 0-0; 1-1; 2-2 63. PI-Number of setae on 1st (outer) segment: 0-1; 1-2; 2-3 P2: Second Thoracic Leg 64. P2-Position of inner apical: O-lateral; I-apical 65. P2-Number of setae on 1st joint: 0-0; 1-1; 2-2 66. P2-Number of setae on 2nd joint: 0-1; 1-2 67. P2-Number of setae on 1st (outer) segment: 0-1; 1-2 P3: Third Thoracic Leg 68. P3-Position of inner apical: O-lateral; I-apical 69. P3-Number of setae on 1st joint: 0-0; 1-1; 2-2 70. P3-Number of setae on 1st (outer) segment: 0-1; 1-2 Phylogenetic Character Coding

Characters and States I Taxa 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 margiJ spines bend ventra post. d post. v surfaCl frill brood I valve I dorsal size reticul sieve J: Al mu Mdml centra cent.scl

G. aethiopis 0 2 0 I 0 I I 2 I I 0 I a I 0 0 0 2 I G. alata 3 1 I I I I 0 I I 0 2 0 0 I 0 0 I I I G. angulata I 2 0 1 0 1 0 0 1 1 1 1 0 1 0 0 2 1 G. capensis 2 2 0 1 0 I 0 2 1 I 1 I a I 0 1 2 o I G. cristata a 0 1 1 1 1 0 3 I 1 2 0 2 1 0 0 0 2 G. curta 1 2 0 1 0 I 1 0 1 1 0 0 a I I I 2 1 I G. expansa 2 2 0 1 1 1 0 a 1 1 1 1 0 1 0 0 2 I I

G. lenis 0 0 0 0 1 1 I a 1 I 0 0 0 1 0 1 2 1 i G.obtusata 1 0 0 1 0 1 1 0 1 1 I 1 a 1 0 0 2 0 I

G.ortali 1 2 1 1 0 1 1 0 I 1 0 0 a 1 0 1 2 1 I G. parcedilatata 0 2 0 1 0 1 1 2 1 1 0 1 a 1 0 0 0 2 I G. simplex 0 0 0 1 0 0 1 0 1 I ? 0 0 1 0 0 2 o : G. n. sp. "coheni" a 0 0 1 1 0 0 a 1 0 1 0 I 1 1 1 2 1 I G. n. sp. "downingi" I 0 0 1 1 1 0 0 1 2 2 1 2 1 0 0 2 0 G. n. sp. "wilsoni" 0 0 0 1 1 0 0 0 1 2 2 1 I 1 0 0 2 0 G. n. sp. "woutersi" 1 2 0 1 ? 1 0 ? 1 ? ? ? ? 1 ? ? 2 1 C. chariessa 0 2 0 1 I 0 I 0 1 1 0 1 ? 0 ? ? 0 2 L. africana 0 0 0 0 0 1 1 0 0 1 0 2 a 0 0 0 0 2 L. dadayi a 0 1 0 0 0 0 0 0 I 0 2 a 0 1 1 2 2 L. thomasi 1 2 0 0 0 1 0 4 0 1 0 2 a 0 0 0 2 2 G. angusta 0 2 1 I ? 0 0 2 1 2 ? 1 ? 1 0 0 2 ?

~ '1 0) Phylogenetic Character Coding

Characters and States Taxa 19 3) 21 Z! 2J at 25 31 Z1 28 2J 3) 31 32 3J 3i 3) :B uberc: nodep node D dorsal ventra sulcus alae alae p« alae Ie hinge shell t ridge radial sieve 1= hinger male I size of no. of G. aethiopis 1 0 0 ? ? 0 1 1 2 0 0 0 0 2 2 2 1 1 G. alata 0 3 1 ? ? 1 2 2 1 1 1 0 0 2 2 0 1 1 G. angulata 0 0 0 0 0 1 0 1 0 0 0 1 ? 2 2 2 1 1 G. capensis 0 0 0 0 0 1 0 0 0 0 0 0 2 2 2 2 1 1 G. cristata 0 4 0 0 0 1 2 3 2 1 1 0 0 2 2 0 1 1 G. curta 0 0 0 0 0 0 0 0 0 1 0 0 2 2 2 2 1 1 G. expansa 0 2 1 ? ? 1 0 0 0 0 0 0 0 2 2 2 1 1 G. lenis 0 0 0 0 0 0 1 1 1 1 0 0 1 2 2 1 1 1 G.obtusata 0 0 0 0 0 1 0 0 0 0 0 1 2 2 2 2 1 1 G.ortali 0 0 0 0 0 0 1 1 2 1 0 0 2 2 2 2 1 1 G. parcedilatata 1 0 0 0 0 0 0 0 0 0 0 0 0 2 2 1 1 1 G. simplex 1 0 0 0 0 0 0 0 0 1 0 1 1 2 2 2 1 1 G. n. sp. "coheni" 0 0 3 2 1 1 1 0 2 0 1 0 2 2 2 1 1 1 G. n. sp. "downingi" 0 0 0 0 0 1 1 1 2 0 0 1 2 2 2 1 1 1 G. n. sp. "wilsoni" 0 0 0 0 0 1 1 1 1 0 0 1 2 2 2 2 1 1 G. n. sp. "woutersi" 0 0 0 0 0 ? 0 0 0 0 0 ? ? 2 2 ? 1 1 C. chariessa 1 0 0 0 0 1 0 0 0 0 0 0 1 1 1 2 0 1 L. africana 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 2 0 0 L. dadayi 0 1 2 1 1 1 0 0 0 0 0 0 0 0 0 2 0 0 L. thomasi 1 2 2 1 1 1 0 0 0 0 0 0 0 0 0 2 0 0 G. angusta ? 0 0 0 0 0 0 0 0 1 0 0 ? ? 2 2 1 1

I-' " Phylogenetic Character Coding Characters and States

Taxa ~ M

G. aethiopis 11110121?1421011111011111210101111 G. alata 1111012101121111121011111211111210 G. angulata 1111011121021011121011111211121111 G. capensis 1111011101421011121010111211121112 G. cristata 1111011101221011121010111211111210 G.curta 1111012111311011111011101011101111 G. expansa 111101110142101112101011121112111'1 G. lenis 1111012111021011121011111111101210 G.obtusata 1111011101431111131011121211121110 G. ortali 1111011121421011111011111111101011 G. parcedilatata 1111012111421011111011111211101111 G. simplex 1111012111211111121011101011101111 G. n. sp. "coheni" 1 1 o 2 o 4 2 o 1 o o 1 1 2 1 1 o 1 G. n. sp. "downingi"' 1 1 o 1 o 4 2 o 1 1 o 2 1 2 1 2 2 1 G. n. sp. "wilsoni" 1 1 o o o 1 1 o 1 3 o 2 1 2 o 2 1 1 G. n. sp. "woutersi"' 1 1 o o o o o o 1 2 o 1 1 2 1 2 o 1 c. chariessa 1 1 o 2 2 4 2 o 1 o o 1 1 o o ? o 1 L. africana 0 o 1 1 2 4 2 o o 1 1 2 1 o 1 ? 1 1 L. dadayi 0 o 1 1 2 4 1 1 o 1 1 o 1 1 o ? 1 1 L. thomasi 0 o 1 1 2 4 2 o o 1 1 1 1 2 1 ? 1 1 G. angusta 1 1 o ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

I-' ~ 00 Phylogenetic Character Coding Characters and States

Taxa 55 Iii m IB m m 61 m m 61 m III f11 III Ell 'jO A2no. A2no. MdpaJ MdpaJ MxpaJ PI inn PI no. PI no. Plno.P2 pos P2no. P2no. P2no. pa pes P3no. P3no. G. aethiopis I 2 I 1 I 0 I 0 1 0 0 0 0 0 I 1 G. alata I 0 I I 1 0 I 0 2 0 0 I I I I 2 G. anguiata I I I I I 0 0 I 2 I 0 I I 1 2 2 G. capensis I 2 I I I 0 0 I 2 I I I I I 2 2 G. cristata 0 0 I I I I I I 2 1 0 I I 0 1 0 G. curta 0 0 I I I 0 0 1 I I 0 1 I I 0 0 G. expansa ? 2 I I I 0 I 0 2 I I I I I 2 ? G.lenis 0 2 I I I 0 I 0 I 0 0 0 0 I 0 1 G.obtusata 0 I I I I I I 0 2 I I I I I 2 2 G.ortaIi 1 2 1 I 1 0 0 0 0 I 0 0 I I 0 I G. parcedilatata 1 2 I I I 0 I 0 I 0 0 0 0 I I I G. simplex 1 2 I I I I I 0 I I 0 0 0 I 0 I G. n. sp. "coheni" 0 0 I 1 I 0 0 0 2 I 0 I I I I ? G. n. sp. "downingi" 0 I I I I 0 I I 2 0 I I 0 I 2 2 G. n. sp. "wilsoni" 0 0 I I I 0 I 0 2 0 0 0 0 I 2 2 G. n. sp. "woutersi" 0 0 I I I 0 0 0 2 ? ? I I I 2 a c. chariessa 1 I 1 0 1 0 I 0 I 0 0 0 0 I 0 I L. africana I 2 0 0 0 0 I I I I 0 0 0 I 0 I L. dadayi I 2 0 0 0 0 I I I I I 0 0 I 0 I L. tbomasi I 2 0 0 0 0 I I 0 I 2 0 0 I 0 I G. angusta ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

..... -..J c.o 180

APPENDIXB

CHARACTERS, CHARACTER STATES AND CHARACTER DISTRIBUTIONS OVER THE FAVORED PHYLOGENY o I r-" ~ Dl o G. aethiopis CD o G. parcedilatata F.- .... = I • G. alata Dl • G. cristata Bo .G. n. sp. "coheni" o ~~G " . F.' a::a • n. sp. woutersl" ~ o leD I M) •• ~ ~CD Er CDc .. :;~~.~A·-/:~:~:~::: ... o ~~ .~~' ... :: .: ~vfo.:.< I;;)G. n. sp. "downingi" Dl= ~s- ...... '.. OG. n. sp. "wilsoni" ~ CI:I I CD ~. . • II G. capensis • a - ·;~IIG. expansa ~ !.o '" ~I::JG.curta ~ ~ ::s •••. (;3 G. ortali :3. ~ e, oG.lenis ~ ~ oG.simplex ~ dl Dl oG.angusta c: 3 :::l IlJ o C. chariessa ~ o :!. ~ Q.:::l CD ~llmDm m e. o L. african a .c t.:)ro ...... o ...m ... e- c: a. 0: .L. dadayi o <"o cc () m .;'\1;;) L. thomasl ~ e. ~ CD co ...... co 182

2. Spines: an elongate projection from valve surface, with rounded or sharply pointed distal extremity O-absent; I-rounded; 2-sharply pointed

1 2

spines unordered

c:::J 0 IPBI_2 1 ~equivocal 183

3. ~: a sharp angulation of the valve in the ventral part, usually parallel to the free edge of the valve

O-angle of bend <90 0 ; 1- angle of bend >900 J.\ • •

o 1

unordered c:J_1 0 ~equivocal 0 .;.. I

~m aG. aethiopis CD a G. parcedilatata ~~ 1-0 aG.alata I a G. cristata ~m aG. n. sp. "cohen!" CD aG. n. sp. "wouters!" ~ a • G. angulata e • G. obtusata , I \ \ • G. n. "downing!" ~ sp. A I ~ .G. n. sp. "wilsoni" I I P.- I - • G. capensis ~ 0 .G.expansa ! § aG. curta (D"" P.- aG.ortali :3. 0 :t-CD oG.lenis '1 CD< .G.simplex ::s .... ~ .• G. angusta !. ~.c. chariessa S miD II OL africana ~ CD CD - .0 -4 0 CD 0;9. c= Cl. ~" []L dadayi ::s <" 3 0 o0 OL thomasi ..., ~ :t-CD

1-0co .;.. 0 I ?l

~rn o G. aethiopis CD OG. parcedilatata F.- .... aG.alata I 't:S "'1 CD rn CD • G. n. sp. "coheni" t:I .-- c+- G. n. sp. "woutersi" -. O G. angulata o G. obtusata • G. n. sp. "downingi" = 1\ n aG. n. sp. "wilsoni" § oG.capensis i -0 ~ .G.expansa :3. "'1 0 DG. curta ~ § oG.ortall • ~ ~ • CD V• ~ 't:S oG.simplex "'1 """ 0 _ G.angusta If \\ ~. t:I chariessa CD ~.c. t:I miD 11 n CD Q. o L. african a CD CD .., 0 .c -'" 0 CD .., 0 c: Q.fIl OL. dadayi t:I <"0 !!!. 0 C OL. thomasi ~ !!!.. c- CD aCD CD fIl .... 00 01 0 I ~ SIl C" • G. aethlopis UI (t) • G. parcedilatata F.- aG.alata I-l = I • G. cristata ~ (t) OG. n. sp. "coheni" UI (t) aG. n. Sp. "woutersr a • G. angulata • G. obtusata aG. n. sp. "downingl" A •I oG. n. sp. "wilsoni" • 13 • G. capensis ~PI- ~. .G.expansa t'D ::se. • G.curta :3 • (t) 0 p.. aG.ortali '1 (J'Q (t) aG.lenis G- oG. simplex SIl ....c+ _0 G. angusta UI

~DC. chariessa c+~ miD II (t) CD CD < • L. african a ::sp.. .c -" 0 CD g: c: 0._ OL. dadayi """ (t) ~:- p.. 0 !. 0 .L. thomasi !!!. -iil ...CD

I-l ao ~ 187

7. Surface omam~ntation: raised, upward growths of portions of the valve surface that reticulate O-reticulate robust; 1- reticulate reduced

anterior ....>

o 1

surface ornamentation unordered C1_1 0 ~equivocal 188

8. .Erill: a wide velate structure, often extending beyond the free edge of the valve O-absent; I-denticulate; 2-spinose; 3-undulating; 4-concentric anterior ....>

unordered

Cl 0 1·;·;·;·;·;·1 1 F;;.:..:-:-:-' 2 1111111111113_4 ~equivocal 0 I ~ ~ aG. aethiopis m CD a G. parcedilatata F.- m,"i ...... G.alata I 'tS 1[9. a G. cristata ::s a1m aG. n. sp. "coheni" CD::s a G. n. sp. "wouters!" =E .~ ~ • G. angulata a G. obtusata aG. n. sp. "downingi" aG. n. sp. "wilsonl" a G. capensis anterior ._ •• > aG.expansa aG. curta aG.ortali aG.lenis ... BG.simplex aG.angusta C tr ::s -. aC. chariessa miD as. o L. african a CD ~"O e, .a ..... 0 CD 0 C Q.Co oL.dadayl ~ <"o :::T CD o OL. thomasi 'tSo !!!. m ~ ::tCD o '1 ..... co00 0 ~ aI 0 aG. aethiopis § II G. parcedilatata .?- ~ oG.alata I 0 II G. cristata < oG. n. sp. "cohenl" ~ t\:) G. n. sp. "woutersr I ~ II G. angulata n II G. obtusata =0 § .G. n. sp. "downlngl" 1\ • e. • G. n. sp. "wilson I" • ~ II G. capensis ... II 1\ ~ III G. expansa pt. (D IIIG. curta ::3. IIG.ortali ~ IIIG.lenis II G. simplex .G.angusta ~ c: < :J !!!. o < IIC. charlessa ~ ~IID a. CD CD CD en II L. african a .c "' ..... 0 ... :::T ~ c: ~.g ilL. dadayi til oc:" CD =r' () ilL. thomasl .g !!!. JD

co~ o 191

11. Dorsal view ofyalye shape: view of shell shape based on dorsal perspective 0- convex; 1- heart-shaped (inflated posterior); 2- triangular shaped (inflated velate) A • • ~ of: i

o 1 2

dorsal view shapes unordered

c:::J 0 IIIlIII_2 1 ~equlvocal 0 ~ I t..:I 1\ II G. aethiopis Crt II G. parcedilatata j oG. alala r ~ o G. cristata I 13 Crt OG. n. sp. "coheni" I ~. 00 G. n. sp. "woutersi" a..... III G. angulata j (D::s t..:I II G. obtusata I ~ .?" 00 IIIG. n. sp. "downingi" § I ~ aG. n. sp. "wilsoni" i i..:l ~ II G. capen sis :3. ::l. 0 0 "1 "1 III G. expansa ~ e+ • 0 oG.curta • 'tS • 0 V tll OG. ortali e+ (D oG.lenis ::l. 0 OG. simplex "1 .11 G. angusta ~IIDl~" ~III c. chariessa CD CD • L. african a .0 I\) ..... 0 c: ~ :;:" Co .L. dadayl 0 0 .L. thomasl ~

co~ t..:I 0 ~ I co ~rn • G. aethiopis (I) • G. parcedilatata ~.a ~ oG.alata I '~I!IIG. cristat~ ." 1\ EJ G. n. sp. cohem ~" ~ ~. G. n. sp. "woutersl" t..:l I o G. angulata ~ t..:l I • G. obtusata ~ ....:J IIG. n. sp. "downlngl" A ~. co •g I ~. EI G . n. sp. "w'I'"I som • V • G. capen sis ~" oG.expansa ~ 00 C'"~ a(I) (I) .G. curta '1 ::3. 0 0 M:I aG.ortali '1 ::s~ oG.lenls CD OG. simplex q-rn G.angusta ~ C.chariessa ~ ~llrnD iI' -(I) m m­ • L. african a ~(I) .c t.:I 1\)-" 0 a;~ rn E, a. 0' .L. dadayi rn .... a G. simplex ... CI G. angusta £: > :l -4 C. chariessa a.3o miD CD £: a L. africana CD .0 -4 0 CD £: ~ .r c.~ .L. dadayi o (I) fir o o a L. thomasi !!!.. ..,ID

.... ento 0 ~ I ~ []G. aethiopis rn~ (t) [] G. parcedilatata ft G.alata ~ I [] G. cristata ~ (t) .G. n. sp. "coheni" = rn (t) G. n. sp. "woutersr 0 a [] G. angulata I;! [] G. obtusata as- G. n. sp. "downlngl" (") []G. n. sp. "wilsonl" a(t) .G. capensis t:S e"+- [] G. expansa ~ .G. curta $Il ~ .G.ortali "1 ~ "'" ~ nnOD S ~ =rn ~-c. ch~riessa (") miD I; -(t) CD CD c: [] L. africana .c -0. 0 m ~ , c: 0.- .L.dadayl <-. CD 0 0 ~ [] L. thomasl ~ ....ID ~

co~ 0) 197

17. Central muscle scar: mark on shell interior for attachment of muscle, generally distinguishable by localized differences in texture of surface, elevation, depression, or delimiting narrow groove O-serial; I-directed anterior; 2-directed posterior

<---- anterior ~ [ J Co ) o 1 2

& "2

central muscle unordered c::::J 0 1IIIIIIII_2 1 ~equivocal 198

18. Central muscle SCar appearance: scar pattern shows little intraspecific variation O-squashed together; I-slightly spaced apart; 2-spaced widely apart

o 1 2

cent.scar appearr. unordered "1c:J_2 0 ~equivocal 0 I ..... ~ 5-0 0'" • G. aethiopis III ct> .G. parcedilatata .rt"=..... oG.alata I / 't:I n >; nl ct>_~I III OG. n. sp. "coheni" ct> OG. n. sp. "woutersI'" =~ -0 ~ o G. angulata >; 0 OG.obtusata § p.. OG. n. sp. "downingi" ct> p.. OG. n. sp. "wIlsoni" 't:I OG. capensis a oG.expansa ... ~. OG. curta ct> n oG.ortali =ct> 0 oG.lenis ....M) .G.simplex ~ =ct> _ G.angusta chariessa ~ ~.c. ~ miDCD IfCD _ o L. african a .Q -0 (ilCD S- 5. a. en OL. dadayi III < .... 0 toil 0 .L. thomasi ct> ~ 0 =<: e. <: ..... ct> CO CO 200 20. Node position: protuberance on valve surface (larger than tubercle, smaller than knob), clearly distinct lobes

O-absent; l-anteroventral; 2-posteroventral; 3-ante~odorsal; 4- posterodorsal <••.. anterior

node pOSition unordered

c::J 0 1·:·:·:·:·:·1 1 E:.;:-:-:-:-, 2 l1li_4 3 ~equlvocal o t.:l I ~ ~ lOG. aethiopis III o G. parcedilatata ~ ~ ~G. a1ata I .-z ~ ~. OG. cristata t.:l ..... I .G. n. sp. "coheni" Ca:I Ca:I e oG. n. sp. "woutersr i J:.. ~~"1 '0 G. angulata :3. ~~ 10 G. obtusata ~ :;.~ G. n. sp. "downlngl" • §~ • §""1 oG. n. sp. "wilsonl" v o § 0"'(") G.capensls ...... (t) .~ ~ 0 I:;IG.expansa f2.::s (t) -< OG.curta ~e. OG. ortaJi ~~ oG.lenis ~a oG.simplex g.g, (") (") [J G. angusta ~(t) C ::::J 0_ co ~ 5 8. [J C. chariessa --g"~ aC!) ~llmD C!) C!) ::::J !O L. africana rn~ .Q t.) I\l -4 0 ~ C "1 c Q.3 C" III L. dadayl <"o -..C!) o III L. thomasl ~ ~

~o.... 0 t\:) I t\:) ~ Ul G. aethiopis CD .0 G. parcedilatata ~ ..... G.alata I A 1\ OG. cristata • • ~...... G. n. sp. "coheni" • G. n. sp. "woutersr § o Pl- ~ (I) t\:) o G. angulata I :3. V o G. obtusata .....0 eG. n. sp. "downlngl" ~ eG. n. sp. "wilsoni" ~ OG. capensls 't:J G.expansa a ~ .eG.curta g. CD eG.ortali loS

eG.lenis (")~ (1) eG.slmplex 0 .0 G. angusta ::s ~ec. chariessa

t\:) 0 t\:) 0 ~ I CI:I ~ til G. aethiopis ~ ,[] G. parcedilatata F.- ..... G.alata 1\ 0 o G. cristata ...... G. n. sp. "coheni" oG. n. sp. "woutersi" ~ oG. angulata ~v 0 oG.obtusata ..... oG. n. sp. "downingl" ~ oG. n. sp. "wilsoni" i::J. oG. capensis ~ 't:I G.expansa ~ • 0 • ~ OG. curta v• s- oG.ortali ~ .0G.lenis § (") oG.simplex ~ 0 .0 G. angusta ::s ~oc. chariessa e.<: II o L. african a ~ miDCD CD - .c --0 (il:J til c:_0 Q. Q.0 .L. dadayi ~ CD 0 thomasi en .L. (")~ !!!. N° ~ CD

~ 0 CI:I 204

24. Sulcus: more or less prominent groove or trench on valve surface trending dorsoventrally and generally best developed in dorsal half carapace; may be reduced to faint depression O-absent; I-present

o 1

sulcus unordered c::J_1 0 ~equivocal 205

25. ~: wing-shaped protrusion on the ventral portion of the shell, parallel to ventral margin, directed posteriorly O-absent; I-one present; 2- two present <.... anterior

unordered c:J_1_2 0 §I equivocal 206

26. Alae position: wing-shaped protrusion on the ventral portion of the shell, parallel to ventral margin, directed posteriorly O-absent; 1- ventromedian; 2-ventral-dorsal

anterior ....>

alae pOSition unordered

r=I 0 1·:·:·:·:·:·1 1 II!IlI!IIIIIII_3 2 EJ equivocal 207

27. Alae length: wing-shaped protrusion on the ventral portion of the shell, parallel to ventral margin, directed posteriorly O-absent; 1-<112 of carapace; 2->112 of carapace <•... anterior

alae length unordered

c::J 0 IlIII!IIII_2 1 ~equivocal 208

28. Hinge angle: angle between the dorsal hinge and an imaginary line the is perpendicular to the ventral margin 0- acute; 1- obtuse

o 1

hinge angle unordered CJ_1 0 ~equivocal 0 ~ I e: ~ o G. aethiopis F- IJ G. parcedilatata ~ I .G.alata ~e: • G. cristata ~ .G. n. sp. "coheni" IJG. n. sp. "woutersr e IJ G. angulata IJ G. obtusata eG. n. sp. "downingl" IJG. n. sp. "wilson I" III ::r' ~ CD IJ G. capensls ;- ...... - eG.expansa :3. ~e: eG. curta ~ IJG.ortali • • ~III IJG.lenis V• III ~ IJG. simplex III S .IJ G. angusta CD ~ ~IJ c. chariessa .... III I! ~ miD IJ L. african a CD CD ..,CD -;:r j:l.. .0 ..... 0 CD - c: Q.~ IJL. dadayi 0 <" :J t:S 0n enCD IJ L. thomas! !!!. en st CD j:l.. 0 '"S III ~ !. 0 co 0 ~ I 't:I 0 [] G. aethiopis a [] G. parcedilatata ::s~. []G.alata ::sCD ~ []G. cristata ~ I "1 []G. n. sp. "coheni" CD G. n. sp. "woutersr (")~ CD • G. angulata = r;:l.. • G. obtusata ~ • G. n. sp. "downing'" S- .G. n. sp. "wilsoni" 't:I :1. "1 0 [] G. capensis ~ [] G. expansa • ~. • CD []G. curta • V a []G.ortan :::t []G.lenls 13 ~ .G.simplex 0 _[] G. angusta I-l ~ ~[]C. chariessa 0 ::s miD 11 african a ~ m m 0 [] L. ..D ..... 0 CD :J t:r' c::: a. _. []L. dadayi CD :J .... <"0 :J 0 iii" [] L. thomas! g e!. CD ? "1

t.:I ~ 0 0 Cal I ~ ~ Dl o G. aethiopis CD o G. parcedilatata ::s .rt.... OG. alata A tTl 't:I CD aG. n. sp. "coheni" <= "1 G. n. sp. "woutersr .... G.angulata ~I.\:) • G. obtusata ~. I.\:) In i I aG. n. sp. "downlngl" :3. v 't:I II Sl:I .G. n. sp. "wllsonl" Dl ~ tTl Dl In .G. capensis 't:I OQ • CD • "1 CD oG.expansa V .... • G. curta §

~I.\:) Jg ~

IIG.simplex ~ In . G.angusta """ ~ ~IID II ~I!I c. chariessa Dl::r L. africana CD ~ 1\) ...... 0 ~~ c: CD 0 Dl <" 0.; oL. dadayi C'f- g n ::l. III_ III:::s oL. thomasi 't:I IDen

I.\:) .... 212

32, Sieve pores: penetrations through valve surface O-in center of depression; l-inbetween the pits on the valve surface; 2- center of depression

1 2 !!.. CI. I: e:- m "2 -e: 0e: ~ rn E -8- m "~ E2 ., J!l rn m . rn rn m "iii m CD e: "~ =g m rn "e: m "~ m e - "m 0 E == "~-... 1: '2 "I: ~ 0 .!E a; "C .c:-

sieve pores unordered

I:::J 0 1IIIIIIII1_2 ~equivocal 213

33. HinfWment: he area ofa valve which is articulated to the opposite valve during the life of the ostracod; includes all special interlocking structures that strengthened the hingement O-lophodont; 1-adont; 2-inverse lophodont

L R

o 1 2

hingement unordered "1c:::J_2 0 ~equivocal 214

34. Male L:H ratio: greatest length and greatest height

0-1.2-1.4; 1-1.4-1.75; 2-1.75-1.9

<..... anterior

male L:H ratio unordered

CJ 0 DIll_2 1 ~equlvocal 215

35. Size of Furca: appendage-like structure attached at the posterior edge of the body; undivided, fork-like structure 0- reduced; 1- as long a thoracic appendages

1

size of furea unordered c=J_1 0 ~equivoeal 0 to I to.:) ?l til aG. aethlopis CD S- a G. parcedilatata CD CD aG.alata III aG. cristata Jr ~ aG. n. sp. "coheni" I to.:) .G. n. sp. "woutersr til CD • G. angulata S- CD aG.obtusata .... ~ • G. n. sp. "downingi" s::l.o to aG. n. sp. "wilsoni" 0 -0'" aG. capensis CD til .G.expansa til aG. curta [ aG.ortali ::r-- aG.lenls ~. aG.slmplex ,!... ~ _8 G. angusta CD til ~ac. chariessa [ miD l~ african a (') CD - o L. ] ..... 0 iDe !:. a. a oL. dadayi [ < I\) 0 til n oL. thomasi 0 !!!. t:I

~(') ~ l\:) ~ 0) 0 CAl I ~

(I)~ • G. aethiopls .P.- • G. parcedilatata .... I .G.alata 0a .G. cristata C5 .G. n. sp. "coheni" ~ .G. n. sp. "woutersl" -(I) • G. angulata • G. obtusata .G. n. sp. "downlngl" .G. n. sp. "wilsonl" .G. capensis ~ ~(I) • G. expansa "1 't:S .G. curta 0 .G.ortall ~.... 0 .G.lenis ::s 0 .G.slmplex lot) n 0 .• G.angusta 't:S chariessa e. ~.c. ~ rot- miD l~ 0 CD CD - o L. african a .c ...... o ~g. ~ E. a.. CD oL.dadayl < 0 0n OL. thomasl aa e. §

....t.:) ~ 0 ~ I 00 ~ .G. aethiopis 0

~ ~ 00 0 Col:! :::tI .CO IJG. aethiopis ~ IJ G. parcedilatata .?- ~ IJG. alata I rn IJ G. cristata B 0 IJG. n. sp. "coheni" 0 IJG. n. sp. "wouters!" ~ IJ G. angulata C'f- IJ G. obtusata t:r' y!I CD .IJG. n. sp. "downingl" tt. 't:J 0 IJG. n. sp. "wilsonl" 0 '"S IJG. capensis !:I) 't:J IJ G. expansa CD ~ IJG. curta 0 I-+) IJG.ortali ~ IJG.lenis CD e:rn IJ G. simplex C'f- .IJ G. angusta a 0.... ~IJ C. chariessa C'" Ii CD miDCI) CI) II) african a ..0 ..... 0 .L @-o c: <. a.~ .Ldadayi 0 0 thomas! !!!. .L ~ ~ ~ H ...

~ co~ 0 ~ I I:.\) 0 • G. aethiopis ~ I • G. parcedilatata ~~ t.:) .G.a1ata I en II G. cristata n .G. n. sp. "coheni" eG. n. sp. "woutersr <= II G. angulata II G. obtusata a G. n. sp. "downingi" IZI eG. n. sp. "wilsoni" CD

aG. capensis aCD ::s aG.expansa f!'+ fIl • G.curta .. g-f!'+ IIG.ortali f!'+ • G.lenis CD ~ . IZI .G.simplex f!'+ G.angusta s=~ CD ~.c. chariessa s- ~IID II african a IZI CD i:... L. ~ g. .cc: '" -" 0 C.CD o·_. CD -. ::::I ilL. dadayi <0 - a n en ilL. thomasi f!'+.... ~ N IX! f!'+.... ::s0 t'-' t.:) 0 0 ~ I 0 ..... ~. II G. aethiopis ...... I II G. parcedilatata ~. ~ I IIG. alata ~ ~ eG. cristata mG. n. sp. "coheni" = mG. n. sp. "woutersl" II G. angulata mG. obtusata • G. n. sp. "downingl" .G. n. sp. "wilsonl" eG. capensis eG.expansa IIG. curta ... m mG.lenis 13 e. II G. simplex .!""' • G.angusta ::r'e. ~.c. chariessa ,.... ~IIDCD 1:CD ::J -p;' ..c 1\) ..... 0 ala. (I) E. a.:; m < en 0 CD 0 p; l'-' a(') ~ CD ~

~(I) III ~ .....~ 0 ~ I ~ ~ ~. • G. aethiopls ~I • G. parcedilatata S~ ~I mG. alata ~ '~m G. cristata 't:S a-m aG. n. sp. "coheni" m []G. n. sp. "woutersI'" S [] G. angulata e. -m CD • G. obtusata S' aG. n. sp. "downlngl" S~ S' ·'ElG. n. sp. "wilsonl" CAlI CAl ~. II G. capensis 't:S t:I c .G.expansa -m e. 0 aG. curta m S -~ .G.ortall e. m m []G.lenls -CD '!hm G. simplex S' ~~ _ G.angusta ~I ::: :.: 5 ..... chariessa ~ rnll'II~D1,1 ••• c> ~ac. m '.' .:. a :J CD II) II) ? • L. african a ~ ,g ~ (0) I\) ...... 0 in n SI:I c a,- .L. dadayi CD <" 0 ~ n en .L. thomasl ~ ~~

~ ~ ~ 0 ~ .....I c,., ~ ...... III G. aethiopis I t-:) II G. parcedilatata t-:) flG.alata c,.,I ~. c,., • G. cristata I mG. n. sp. "coheni" "- ~ DG. n. sp. "woutersr II G. angulata • G. obtusata IIG. n. sp. "downingi" "EJ G. n. sp. "wilson I" aG. capensls .. G. expansa I3G. curta IIG.ortali flG.lenls "EJ G. simplex G.angusta ~.c. chariessa ~llrnD i: II L. africana ~ til ] (,)1\)-"'0 a;; 't:I c: a,cn 13L. dadayi 5' <' CD 0 iii I n !e. aL. thomasl t=: e.. Pi' CD

~ ~c,., oI ..'"7 t ~ o G. aethiopis I t\:) o G. parcedilatata .G.alata o G. cristata DG. n. sp. "coheni" DG. n. sp. "wouters!" o G. angulata • G. obtusata DG. n. sp. "downingi" = DG. n. sp. "wilsoni" o G. capensis DG.expansa DG. curta DG.ortaii .DG.lenis .G.simplex G.angusta c ::l »..... o en DC. chariessa -''0 Q. -" miD CD ::l L. african a CD o .c ..... 0 ~ CD ~.... c Q.en • L. dadayi t:S <" 5" CD o ::l I C) CD DL. thomasi ~ -. -p;.: CD

t\:) t\:) ~ 0,. ~ .§ ?l .G. aethiopis ....n • G. parcedilatata ~ I-' .G.alata ~ Sll • G. cristata fi aG. n. sp. "cohenl" A f!- .G. n. sp. "woutersi" • G. angulata .G.obtusata .G. n. sp. "downlngl" aG. n. sp. "wilsonl" • G. capensis =r- • G.expansa ~, . aG. curta .... -~ aG.ortali CD ~ Ell .G. simplex na ~ G.angusta ( i Ell .c. chariessa VI' II mID~:CD Q. CD o l. african a .0 ..... 0 ~iii c CD CD -< Q...... Ol. dadayi o en o -en Ol. thomas I !!!.. CD cp

t-.:) t-.:) t11 0 ~ I 0 ~ ~.

~ I ~ ~. I.\:) I I.\:) • G. cristata CI:I I CG. n. sp. "coheni" CI:I illiG. n. sp. "woutersi" II G. angulata • G. obtusata . !;;IG. n. sp. "downingi" .G. n. sp. "wilsoni" aG. capensls .. G. expansa c:JG. curta .... 13 G. ortali IIIIG.lenls [ ';S g ~ CC. chariessa acn -~ ~IIDD CD CD CD !! .c (.)"' ..... 0 I'll c: a;m :c. Q.~ o a- o :J­ n o ~ c: :J- [ ..,CD I'll

I.\:) I.\:) en 0 ~ I -..J o G. aethiopis [ M) o G. parcedilatata ~ CD oG.alata ~ o G. cristata ~ s:l,. o G. n. sp. "coheni" 0 OG. n. sp. "woutersi" ~ OG. angulata 0::s o G. obtusata til 't:S.... o G. n. sp. "downlngi" ::s R ~ OG. n. sp. "wilsoni I-" I OG. capensls ~ oG.expansa til ::sCD OG. curta C"!- oG.ortali ::r' e? oG.lenis '"t I oG.slmplex -~ CD G.angusta til C"!- ~oc. chariessa 2(') miD Ii african a C"!- (I) (I) - .L .Ll ...... 0 ~m ~ 5..o 0...... _ .L.dadayi CD < (/) til 0 thomasl (/) .L ~ (I) Cfl

~ ~ -..J 0 .r;:.. 00 0• ~ . G. aethiopis ....• aG. parcedilatata ~..... l\:) []G.alata A ~ [] G. cristata ." = []G. n. sp. "cohen! ." [] G. n. sp. "woutersl • G. angulata G obtusata [].[]G. n. sp. "downlngl"." []G. n. sp. "wilson! []G.capensis . P'" ... e1. oG.expansa ';'l =-: IIG. curta I:Ir (t) aG.ortali III IIG.lenis a(') III G. simplex ~ G.angusta ~ DC» IA III ~.c. chariessa mil H • L. africana CD 1\) ..... 0 CD CD ...., .0 0._ C III .L. dadayl --< CD 0 0 .L. thomasl !!!. , l\:) l\:) 00 0 ~ C..:I• CO ~. ~• a G. parcedilatata ~~ I.\:) IIG.alata • .- en III G. cristata aG. n. sp. "coheni" IIG. n. sp. "woutersr III G. angulata = • G. obtusata aG. n. sp. "downingl" aG. n. sp. "wilson'" a G. capensls III G. expansa C'"~ .... CD DG.curta "1 S, aG. ortal! c..o. ~ ....0 ~ ~ III G.angusta ~ CD ~III c. chariessa milD 1:CD • aL africana i ~ 1\) ..... 0 Cilen a III c:<" a.~. aLdadayi 0 n ilL thomasl ~

I.\:) I.\:) co 0 01 I 0 0 ~..... • G. aethiopis I • G. parcedilatata r:' t..:l I .G.alata t..:l • G. cristata aG. n. sp. "coheni" aG. n. sp. "woutersi" • G. angulata • G. obtusata • G. n. sp. "downingl" ~ ~ = • G. n. sp. "wilsonl" • G. capen sis .G.expansa DG. curta IIG.ortali IIG.lenis DG. simplex G.angusta § ~ !D C. chariessa ~IID aso CD CD· o L. africana .c 1\) .... 0 c: ...CD -0 Co :::s- ilL. dadayi <"o (ii o en .L. thomasi !!!. "E. :::s CD

t..:l CI:) o 0 CJ1 I J'? .... oG. aethiopis .... I • G. parcediJatata c.o .G. alala • G. cristata aG. n. sp. "coheni" aG. n. sp. "woutersr • G. angulata .G.obtusata <= .G. n. sp. "downingl" o G. n. sp. "wllsoni" • G. capensis .G.expansa .G. curta .G.ortali aG.lenls .G. simplex ~ ~G.angusta !~ C. chariessa miDCD CD !J .L african a -g ..... 0 Cil- _. Q. CD OLdadayi ~ 3 0 :i" thomasl ~ ~ .L en cpCD

t.:) c.o.... 232 52. A2-Position of terminal claws: O-all from same level; 1-3 from lower level, 1 from higher level; 2-2 from lower level, 1 from middle level, 1 from higher level

1

A2 pos. term. claws unordered

c:J 0 l1l\I1_2 EJ equivocal 0 en I to .!":' ..... • G. aethiopls I J'? • G. parcedilatata ~ I aG.aJata to a G. cristata =~ :::: aG. n. sp. "cohenl" aG. n. sp. "woutersr II G. angulata II G. obtusata IIIG. n. sp. "downlngl" eG. n. sp. "wilsonl" • G. capensis ~~::: .G.expansa G. curta .G.ortali .G.lenis ~ .G.slmplex G.angusta § ~ lie. chariessa a::so milD CD ? CD • L. africana .c 1\) ..... 0 ii1 en c::: a. CD dadayi ~::: ~. in .L. o CD thomasl (.0) .L. !!!. a

~ to ~ 0 en I ~ !7.... • G. aethiopis I • G. parcedilatata J'P I.\:) I .G.aJata e ~ II ~ C..:I a G. cristata aG. n. sp. "cohenl" aG. n. sp. "wouters!" • G. angulata a G. obtusata a G. n. sp. "downlngl" a G. n. sp. "wilsonl" ... • G. capensls G.expansa aG. curta .G. orta" aG.lenls gG.simplex G.angusta miD • C. charlessa CD .a ...... 0 • L. afrlcana c c:" .L.dadayi o n o .L. thomasi !!. c -,CD

I.\:) C..:I ~ 0 en .....I en ~ . I!I G. aethlopis ..... I III G. parcedifatata J'9 l\:I I ~ oG.alata to o G. cristata ( IIG. n. sp. "cohenl" aG. n. sp. "woutersr II G. angulata G.obtusata mG. n. sp. "downing'" aG. n. sp. "wilson'" .G. capensls G.expansa [ G.ortali oG.lenis II G. simplex • G.angusta 1:1:1 ""lie. chariessa ~IIDCD I:CD· ilL africana oC "' ..... 0 men c: a. CD ilL dadayi <" ~ o CD 0 ilL thomasi ~ 5" =r !!!.. ::J ..,CD

~ toen 0 en I 0) !":" ~ a G. aethiopis I II G. parcedilatata w~ ~ I ~ .G.a/ata ~ ::s ? .G. cristata 0 M) IZI e G. n. sp. "coheni" CD C'I- I» eG. n. sp. "wouters!" C!) aG. angulata 0::s IZI II G. obtusata CD 0 .G. n. sp. "downing!" ::s0 .. G. n. sp. "wilson!" j:l. C'I- II G. capensis IIG. expansa ~.::s IIG. curta a eG. ortali III .G.lenis ~ ~CD a G. simplex ::s ~ C'I- ~ G.angusta e C. chariessa @IID i: Ill. africana I\) ..... o~· -g_ (I) Sl < Q. - ilL. dadayi o III g (I) II L. thomasi - :JI\) Q. en rP ~ ~ 0) o c:n ....::J • G. aethiopis • G. parcedilatata 1~ aG. aJata I 0'" aG. cristata (I) a G. n. sp. "coheni" a a G. n. sp. "woutersr = ~ (I) (I) • G. angulata • G. obtusata aG. n. sp. "downlngl" aG. n. sp. "wIlsoni" aG. capensis aG.expansa aG. curta aG.ortali aG.lenls .G.simpJex G.angusta ... • C. chariessa miD L. african a CD o .c ...... 0 c: OL. dadayi <:­o o OL. thomasi !!!.

t..:l ~ ....::J o en m 00 e.~ ·G. parcedilatata .G.alata % ~ aG. cristata ....Co .G. n. sp. "coheni" ~ .G. n. sp. "woutersi" (") • G. angulata = ~ ~ .G.obtusata aG. n. sp. "downingl" .G. n. sp. "wilsonl"

aG. curta .G.ortali .G.lenls .G.simplex ... G.angusta i! ~o c. chariessa miD

t.:» ~ 00 0 I en =:s ~ 0 ~ • G. parcedilatata ~ aG. alata .... I • G. cristata ~ \, !:l,. s::n CD aG. n. sp. "woutersi'" = ~~ ~- !:l,. • G. angulata aG.obtusata • G. n. sp. "downingi" a G. n. sp. "wilsoni" • G. capen sis a G. expansa . • G.curta aG.ortali aG.lenis aG.simplex """' G.angusta !! ~.c. chariessa miDCD CD !!!. o l. africana .D ..... O ;"0 s. Co ol.dadayi < 0 n Ol. thomasi !!.

t-.:) ~co 0 0) I 0 o G. aethiopis -~ a G. parcedilatata ~ I-" oG. a1ata I ~ • G. cristata '1:j.... (') aG. n. sp. "cohen!" a oG. n. sp. "woutersr oG. angurata • G. obtusata OG. n. sp. "down!ngi" ;0 G. n. sp. "wilson!" a G. capensis I-l oG.expansa OG. curta oG.ortali OG. renis .G.simprex _ G.angusta miD 1; ~oc. chariessa CD CD ::I o L. african a -D -" 0 CD~ OL. daday! ~o a.~_. n n ~ ~

~ ~o 0 0) • f-l 9 .G. aethiopis f-l f-l• • G. parcedilatata lJ.t .G.alata • G. cristata eG. n. sp. "cohen I" DG. n. sp. "woutersr = e G. angulata • G. obtusata a G. n. sp. "downlngl" aG. n. sp. "wilsonl" [] G. capensis .G.expansa eG. curta eG.ortan .G.lenis .G.slmplex G.angusta ~.c. char/essa ... miD(I) !~(I) • .0 ..... 0 ~cn E. a.~ < 0 0 !!?. --

t.:) ~ f-l 0 0) I t\:) ~9 e G. aethiopis '"""I e G. parcedilatata '""" t\:) I I~ eG.alata t\:) • G. cristata e eG. n. sp. "cohen! ." '= j[ d' = eG. n. sp. "wouters!" • G. angulata e G. obtusata .G. n. sp. "downing!" eG. n. sp. "wilsoni" • G. capensls eG. expansa • G. curta ... eG.orlali eG.lenls e G. simplex G.angusta c: "'0 :l ..... e C. chariessa g :l 0.0 miD(I) (I). ., en • L. african a .0 0 (I) (I) c: o.g; .L. dadayi <'o (I) o I\J .L. thomasi ~ ~ :l 0.

t\:) ~ t\:) 0 m .....I ~ ..... I II G. aethiopis I.\:)

aG. parcedilatata I.\:) I .G.alata e /1 ~ a G. cristata aG. n. sp. "coheni" aG. n. sp. "woutersr • G. angulata • G. obtusata .G. n. sp. "downingl" aG. n. sp. "wilsonl" • G. capensls .G.expansa ~ BG.curta []G. ortali aG.lenls ,II G. simplex _ G. angusta ~a c. chariessa milDCD i~CD' III L. african a .0 1\) ...... 0 (i;cn E. c. ~ ilL. dadayi < III ~ g CD ...... [] L. thomasi ~ ~ ~--- I.\:) ,;:.. C/.:) 0 I ~ -~ ~ DG. parcedilatata !. ~. DG.alata .... I • G. cristata ~.... aG. n. sp. "coheni" (') -(!) G. n. sp. "woutersl" \ • G. angulata = • G. obtusata aG. n. sp. "downlngl" G. n. sp. "wilsonl" .G. capensis .G.expansa .G.curta .G.ortali aG.lenis .G.slmplex _ G.angusta ~ miD Ii ~DC. chariessa ~ -... 0 ~ $I' • L. africana c_. CoCD -::J .L.dadayi < ::J o CD g ., thomasl - .g .L. 0- !!!.

.;:...~ .;:... 0 0) I 0 en ~ ...... oG. aethiopis .....I o G. parcedilatata t..:I I oG.alata t..:I o G. cristata OG. n. sp. "coheni" G. n. sp. "woutersi'" o G. angulata II G. obtusata IIG. n. sp. "downingi" OG. n. sp. "wilsonl" II G. capensis .G.expansa

G.angusta c: "'C :::l N OC. chariessa g :::l ~IID 0.0 CD CD ' OL african a .D N ...... 0 -. rn CD CD c: o.S' ilL dadayi <'o CD o ...... L thomasl ~ !e.

t..:I en~ o 0) ..:.. 0)

o G. aethiopis ~ t\:)• o G. parcedilatata .G. alata • G. cristata .G. n. sp. "cohenl" aG. n. sp. "woutersr • G. angulata • G. obtusata .G. n. sp. "downlngl" OG. n. sp. "wilsonl" • G. capensis .G.expansa aG. curta oG.ortali oG.lenls oG.slmplex G.angusta c: "'0 :J I\) OC. chariessa Q :J 0.0 (I) • miD(I) o L. africana -.. en .0 0 (I) (I) c:.c: 0._ OL. dadayl o ~ o I\) [J L. thomasi e. :Ja.

~ 0) o (j') I ...:J !7 I-l a G. aethlopis I t\:) aG. parcedilatata .G.alata • G. cristata .G. n. sp. "coheni" • G. n. sp. "woutersl" • G. angulata • G. obtusata aG. n. sp. "downlngl" aG. n. sp. "wilsonl" aG. capensis .G.expansa

aG.lenis aG.simplex G.angusta

:lc: I\)" o :l IJ C. chariessa a. o miDCl) Cl)o aL african a .0 0 ... (II :coc: Q.S'Cl) Cl) aLdadayi o Cl) o ~ aL thomasi ~ ~

t\:) ~ ...:J o 0) ,!- (Xl DG. aethiopis S­ • G. parcedilatata ~ ~ .G. aJata I S» DG. cristata 'S. (') • G. n. sp. "cohenl" a .G. n. sp. "woutersr • G. angulata .G.obtusata .G. n. sp. "downlngi" .G. n. sp. "wilsonl" • G. capen sis Q .G.expansa .G. curta .G.ortall .G.lenis .G. simplex . G.angusta ~.c. chariessa miD I:CD 0 • L. african a .LlCD ..... 0'"CD_ !I' £. 0. 5 .L. dadayi

t.:) .;... (Xl 0 0) 0• ~ ...... • mG. parcedilatata .... ~• IIIG. alata ~ III G. cristata mG. n. sp. "cohenl" .G. n. sp. "woutersr • G. angulata • G. obtusata .G. n. sp. "downlngl" .G. n. sp. "wilson I" • G. capensls .G.expansa

G.angusta c:: "tJ ::::I Cal o C. chariessa ~ ::::I 0.0 milD CD· CD '0 L. african a -'l I\) ..... 0 ... en c:: CD CD OL. dadayi o<" o.i o ..... OL. thomasi ~ !e.

~ co~ 0 -..1 I 0 !:".... I t.:l e ~

c: "0 :::l (,) o :::l a o ~llrnD CD" ~ (,)"' .... 0 ~ (II c: a. !a o<" II: o .... ~ !!l

t.:l o01 251

APPENDIXC TAXONOMIC SPECIES DESCRIPTIONS AND NOTES ON OTHER SPECIES 252

TAXONOMIC DESCRIPTIONS OF NEW SPECIES

Subfamily Limnocytherinae KLIE, 1938 Tribe Cytheridellini DANIELOPOL and MARTENS, 1990 Amended diagnosis. (from Martens, 1993) Carapace with important sexual dimorphism; females have widely dilated valves in dorsal view, forming brooding pouches for eggs and first and possibly second instars. Valves with or without additional longitudinal ridges and crests on the external surface. Hinge adont to weakly lophodont or inverse lophodont. Mxl with palp weakly sclerified, mostly undivided, with reduced number of apical and lateral setae. P(3) the largest walking limb (Gomphocythere and Gomphodella) or reduced to a short and curved cleaning limb (Cytheridella).

Genus Gomphocythere SARS, 1924 Type species: Limnicythere obtusata SARS, 1910 Amended diagnosis. Females with a brooding pouch. Hinge inverse lophodont, with posterior cardinal tooth on LV larger than anterior one. Ventral side in some species set with ridges. Al with penultimate segment set with 7 claw-like setae and weakly or not divided. P(3) the largest walking limb. Posterior part of female abdomen with one furca, bearing two furcal setae and three hairy furcal lobes. Hemipenis with large, articulating distal lobe, without a lateral seta (present in Cytheridella). 253

ABBREVIATIONS

LT Lake Tanganyika O.C. Ostracod Collections of the Royal Belgian Museum of Natural History R.B.I.N.Sc. Royal Belgian Museum of Natural History LPOOXX Dissection done by Lisa Park LT/91/21 Collection taken in Lake Tanganyika on May 21, 1991 RV Right valve LV Left valve A1 First antenna A2 Second antenna Mx Maxilla Md Mandible P1 First thoracic appendage P2 Second thoracic appendage P3 Third thoracic appendage dl Distal lobe cop Copulatory processes ofhemipenis ur Upper ramus of clasping organ of hemipenis L Length W Width H Height 254

Gomphocythere downingi sp.n.

Type locality. LT/92/21 (Martens and Gooderis expedition); Lake Tanganyika, East Mrica. Bay south of Karema, Tanzania; Water depth 18 m; Coordinates: 6°52' S, 30°32' E.

Type material. The following collections have been used for the present descriptions:

LT/92/21 (for holotype and allotype): sample collected from sand and shell debris, by Dr. K. Martens and Dr. B. Gooderis on 5/20/1992. All illustrated specimens originate from this sample.

Holotype : a male, with soft parts dissected in glycerine in a sealed slide and with valves stored dry (LP 0019).

Allotype : a female, dissected and stored along with the holotype (LP 0021). Paratypes : 2 male and 4 females stored dry after use for SEM (LP 0026, LP 0050, LP 0038, LP 0039, LP 0048, LP 0051). Paratype localities LT/92/57; LT/91/l.

Repository : all types, both dissected and in spirit, are stored in the Ostracod Collection of the R.B.I.N.Sc. 255

Derivation of name : this species is named after Dr. Kevin Downing, DePaul University, in appreciation of his contribution to this project. I also wish to acknowledge and thank him for the warm friendship that we have shared over the years.

Diagnosis. Large and elongated species, with straight dorsal margin. Posterior part of carapace squarishly rounded in dorsal view. Prominent alae developed that is greater than one half the length of the carapace. Carapace strongly reticulated. Mx1 palp weakly sclerified and undivided. Hemipenis and dl sub-triangular, with rounded base, blunt tip set with comb-like structure, medio-Iateral indentation without a protuberance; ur relatively short, cop short and curved in its distal half. Additional description of male: Valves (Figure 1) strongly sclerified and externally reticulated. Prominent alae developed ventrally. Anterior margin broadly rounded, posterior margin more squarish, but still evenly rounded. Greatest height situated at about one fourth from the anterior. Ventral margin weakly sinuous. Inverse lophodont hinge (Figure 1), with prominent and elongated, but simple cardinal teeth on LV fitting in cardinal sockets on RV; intercardinal bar on RV smooth. Both valves (Figure 1) with prominent selvage; calcified inner lamellae narrow on both anterior and posterior sides. Carapace in dorsal and ventral (Figure 1) views narrow, with nearly parallel sides. A1 (Figure 2) with 5-segmented endopodite, penultimate segment undivided. First segment of endopodite twice as long as wide, with one long, medio-ventral seta. Second segment short and squarish, with one 256

long apical seta. Third segment with three apical (one ventral, two dorsal) setae. Fourth segment with 3 apical (one ventral, two dorsaD setae. Terminal segments elongated, with one sub-apical setae and one apical seta, slightly fused at the base with a seta, this seta longer than the actual sensorial club. A2 (Figure 2) with endopodite three-segmented. First segment short, bearing one long apical seta. Second segment long, with medio­ ventrally two setae and one long seta (the latter reaching to tip of segment); one stout, claw-like seta. Terminal segment with three stout and short claws, one subapically and two apically inserted. Md (Figure 2) without special features. Palp four-segmented. First segment with respiratory plate consisting of three rays only (two long, one short) and with two subequal ventral setae. Second segment short and squarish, bearing three apical setae. Third segment of similar shape, with 5 apical setae, three dorsally and two ventrally inserted. Terminal segment pyramidal, with two claws. Mxl (Figure 2) with three endites of normal shape and chaetotaxy; palp weakly sclerotized and unsegmented, set with three lateral and two apical setae. P(1-3) (Figure 3) all walking limbs; pel) the shortest and P(3) the longest and with the longest claw. First segment in these limbs with 2 long dorsal and one long and hirsute ventral seta (the latter proximally inserted) and with one (on P(2) and P(3) or two (on pel»~ knee-setae as is typical of the subfamily. 257

Hemipenis (Figure 4) of the typical Gomphocythere type, with a large muscular body and a prominent, articulating distal lobe (dl), devoid of setae. Shape of the dl in this species triangular, with a rounded base, a blunt tip set with a comb-like structure and with a sinuous dorsal margin with a protuberance. On main body, ur relatively short and straight, well­ sclerotized; cop short and distally curved.

Additional description of females. Valves (Figure 1) generally as in the male, but larger and with straight dorsal margin only slightly sloping towards the caudal sides. Structure of hinge and valve margin as in the male. Carapace in dorsal and ventral view with posterior part more swollen than in the male, but considerably less so than in most other species of this genus. All appendages (Figures 5-7) generally as in the male; Two pairs of "hypostomeal lobes" present posterior to labium, anterior lobes approximately twice as long as posterior ones. Sternum as illustrated in Figure 4. Posterior part of abdomen with one furca bearing two furcal setae and three hirsute furcal lobes, one seta and one forked organ. Genital orifice small and rounded.

Measurements (in /lm): Male: L = 769 (n = 1); H = 446 (n = 1) Female: L = 650-720 (n = 4); H = 410-435 (n = 4) 258

Differential diagnosis. The new species can easily be distinguished from other species by the large, oval shaped reticulated valves with the prominent alae and by the shape of the hemipenis. 259

Figure 1: Gomphocythere downingi n.sp.: SEM micro graphs of the valves. A. male LV, external view (LP0026). B. male RV, external view (LP0050). C. male RV, internal view (LP0026). D. male LV, internal view (LP0050). E. female LV, external view (LP0021). F. female RV, external view (LP0039). G. female RV, internal view (LP0021). H. female LV, internal view (LP0039). (Scale bars = 100 /lm). 2~O 261

Figure 2: Gomphocythere downingi n.sp. (male LP0019). A. First antenna (AI), B. Maxilla (Mx), C. Second antenna (A2), D. Mandible (Md). (Scale = 100 !lm for A-D). 262

-<

Figure 2 263

Figure 3: Gomphocythere downingi n.sp. (male LP0019). A. Third thoracic leg (P3), B. Second thoracic leg (P2), C. First thoracic leg (PI). (Scale = 100 11m for A-C). 264

Figure 3 265

Figure 4: Gomphocythere downingi n.sp. (male LP0019). A. Sternum, B. Hemipenis. (Scale = 100 Jlm for A-B). 266

100 Jlm

Figure 4 267

Figure 5: Gomphocythere downingi n.sp. (female LP0021). A. First antenna (A1), B. Second antenna (A2). (Scale = 100 J.lm for A-B). 268

Figure 5 269

Figure 6: Gomphocythere downingi n.sp. (female LP0021). A. Maxilla (Mx), B. Mandible (Md and mandibular palp). (Scale = 100 /lm for A-B). 270

Figure 6 271

Figure 7: Gomphocythere downingi n.sp. (female LP0021). A. Third thoracic leg (P3), B. Second thoracic leg (P2), C. First thoracic leg (P1). (Scale = 100 J.lm for A-C). 272

Figure 7 273

Gomphocythere wilsoni sp.n.

Type locality. LT/92/21 (Martens and Gooderis expedition); Lake Tanganyika, East Africa. Bay south of Karema, Tanzania. Water depth 18 m; Coordinates: 6°52' S, 30°32' E.

Type material. The following collections have been used for the present descriptions:

LT/92/21: sample collected amongst sand and shell debris, by Dr. K. Martens and Dr. B. Gooderis on 5/21/1992. All illustrated specimens originate from this sample.

Holotype : a male, with soft parts dissected in glycerine in a sealed slide and with valves stored dry (LP 0027). Paratypes : 1 male and 0 females stored dry after use for SEM (LP 0032). Locality for paratype material: LT/89/903 Caljon

Repository: all types, both dissected and in spirit, are stored in the Ostracod Collection of the R.B.I.N.Sc.

Derivation of name: this species is named after Dr. Mark Wilson, The College of Wooster, in appreciation for his encouragement and support over the years. 274

Diagnosis. Moderately large and elongated species, with straight dorsal margin. Posterior part of carapace in female relatively narrow and rounded in dorsal view. Mxl palp weakly sclerified and undivided. Hemipenis and dl sub-triangular, with squarish base, blunt tip set with comb-like structure, medio-Iateral indention without a protuberance; ur relatively short, cop short and curved in its distal half. Additional description of male: Valves (Figure 8) strongly sclerified and externally reticulated, but without ridges or crests. Anterior margin broadly rounded, posterior margin more narrow, but still evenly rounded. Greatest height situated at about one fourth from the anterior, from there the straight dorsal margin sloping towards the caudal side. Ventral margin weakly sinuous. Hinge (Figure 8) inverse lophodont, with prominent and elongated, but simple cardinal teeth on LV fitting in cardinal sockets on RV; intercardinal bar on RV smooth. Both valves (Figure 8) with prominent selvage; calcified inner lamellae narrow on both anterior and posterior sides. Ventro-caudal flanges irregularly serrated. Carapace in dorsal and ventral views narrow, with nearly parallel sides. Al with 5-segmented endopodite. First segment of endopodite c. twice as long as wide, with one long, medio-ventral seta. Second segment short and squarish, with one long apical seta. Third segment with one ventral and one dorsal apical setae. Fourth segment with one medio ventral and four apical (one ventral and three dorsal) setae. Terminal segments elongated, with two sub-apical setae and one apical seta, slightly fused at the base with a seta, this seta longer than the actual sensorial club. 275

A2 (Figure 9) with three-segmented endopodite. First segment short, bearing one long apical seta and one apical ventral setae. Second segment long, with medio-ventrally two setae and one long setae (the latter reaching to tip of segment); medio-dorsally two unequal setae and apically one stout, claw-like seta. Terminal segment with three stout and short claws, one subapically and two apically inserted. Md without special features. Four-segmented palp. First segment with respiratory plate consisting of three rays only (two long, one short) and with two subequal ventral setae. Second segment short and squarish, bearing three apical setae. Third segment of similar shape, with 5 apical setae, three dorsally and two ventrally inserted. Terminal segment pyramidal, with two claws. Mx1 (Figure 10) with three endites of normal shape and chaetotaxy; palp weakly sclerotized and unsegmented, set with three lateral and two apical setae. P(1-3) (Figure 10) all walking limbs; P(1) the shortest and P(3) the longest and with the longest claw. First segment in these limbs with 2 long dorsal and one long and hirsute ventral seta (the latter proximally inserted) and with one (on P(2) and P(3) or two (on P(1» knee-setae as is typical of the subfamily. Hemipenis (Figure 9) of the typical Gomphocythere type, with a large muscular body and a prominent, articulating distal lobe (dl), devoid of setae. Shape of the dl in this species broadly sub triangular, with a squarish base, a blunt tip set with a comb-like structure and with a sinuous dorsal margin devoid of protuberances. On main body, ur 276 relatively short and straight, well-sclerotized; cop short and distally curved.

Measurements (in /lm): Male: L = 669-680 (n = 2); H = 446-460 (n = 2); W = 200-230 (n = 2)

Differential diagnosis. The new species can easily be distinguished from almost all of its congeners by the elongated valves and by the shape of the hemipenis. 277

Figure 8: Gomphocythere wilsoni n.sp.: SEM micro graphs of the valves. A. male LV, external view (LP0032). B. male RV, external view (LP0027). C. male RV, internal view (LP0032). D. male LV, internal view (LP0027). Gomphocythere coheni n.sp. E. female LV, external view (LP0045). F. female RV, external view (LP0030). G. female RV, internal view (LP0045). H. female LV, internal view (LP0030). (Scale bars = 100 J.lm). 278 279

Figure 9: Gomphocythere wilsoni n.sp. (male LP0027). A. Hemipenis, B. Second antenna (A2). (Scale = 100 J.lm for A-B). 280

100 11m

Figure 9 281

Figure 10: Gomphocythere wilsoni n.sp. (male LP0027). C. Maxilla (Mx), D. First thoracic leg (P1), E. Third thoracic leg (P3). (Scale = 100 Ilm for C­ E). 282

100 Jlm

Figure 10 283

Gomphocythere coheni sp.n.

Type locality. LT/91/21 (Martens expedition). Lake Tanganyika, East Africa. Dama River delta, Rumonge, Burundi. Water depth 50 m; Coordinates: 3°57' S, 29°26" E.

Type material. The following specimens have been used for the present descriptions:

LT/91/21 A and B: sample collected amongst sand and green algae, by Dr. K. Martens on 512111991. Measurements at time of collection: pH= 9-9.5, alkalinity= 6.07 meqn. All illustrated specimens originate from this sample.

Holotype : a male, with soft parts dissected in glycerine in a sealed slide and with valves stored dry (LP 0045). Allotype : a female, dissected and stored along with the holotype (LP 0045). Paratypes : 1 male and 2 females stored dry after use for SEM (official RBINSc O.C. numbers not yet assigned) (LP 0049, LP 0030, LP 0046).

Repository: all types, both dissected and in spirit, will be stored in the Ostracod Collection of the R.B.I.N.Sc. 284

Derivation of name : this species is named after Dr. Andrew Cohen, my graduate advisor at the University of Arizona, in appreciation of his contribution to this project and service as my advisor.

Diagnosis. Short and round species, with straight dorsal margin, especially in the male. Posterior part of carapace in female bulbous and rounded in dorsal view. Mxl palp weakly sclerified and undivided. Hemipenis and dl sub-triangular, with squarish base, blunt tip set with comb-like structure, medio-Iateral indention without a protuberance; ur relatively short, cop short and curved in its distal half.

Additional description of female : Valves (Figure 8) strongly sclerified and externally reticulated. Anterior margin rounded, posterior margin more broad. Greatest height situated at about one fourth from the posterior. Ventral margin straight. Hinge (Figure 8) inverse lophodont, with prominent and elongated, but simple cardinal teeth on LV fitting in cardinal sockets on RV; intercardinal bar on RV smooth. Both valves (Figure 8) with prominent selvage; calcified inner lamellae narrow on both anterior and posterior sides. Four prominent nodes present postero­ anteriorly, dorso-ventrally. Sulcus present dorsally and ventrally. Weak alae developed from ventral nodes. Carapace in dorsal and ventral views narrow, with nearly parallel sides. Al (Figure 11) with 5-segmented endopodite. First segment of endopodite twice as long as wide, with one long, medio-ventral seta. Second segment short and squarish, with one long apical seta. Third 285 segment with one ventral and one dorsal setae. Fourth segment short and squarish with 4 apical (one ventral, three dorsal) setae. Terminal segments elongated, with two sub-apical setae and one apical aesthetasc, slightly fused at the base with a seta, this seta longer than the actual sensorial club. A2 (Figure 11) with endopodite three-segmented. First segment short, bearing one long apical ventral seta. Second segment long, with medio-ventrally one setae and one long aesthetisc (the latter reaching to tip of segment); medio-dorsally one setae and apically one stout, claw-like seta. Terminal segment with two stout and short claws, one subapically and one apically inserted. Md (Figure 11) without special features. Palp four-segmented. First segment with respiratory plate consisting of three rays only (two long, one short) and with two sub equal ventral setae. Second segment short and squarish, bearing three apical setae. Third segment of similar shape, with 5 apical setae, three dorsally and two ventrally inserted. Terminal segment pyramidal, with two claws. Mx1 with three endites of normal shape and chaetotaxy; palp weakly sclerotized and unsegmented, set with three lateral and two apical setae. P(1-3) (Figure 12) all walking limbs; P(l) the shortest and P(3) the longest and with the longest claw. First segment in these limbs with 2 long dorsal and one long and hirsute ventral seta (the latter proximally inserted) and with one (on P(2) and P(3) or two (on P(l)) knee-setae as is typical of the subfamily. 286

Hemipenis of the typical Gomphocythere type, with a large muscular body and a prominent, articulating distal lobe (dl), devoid of setae. Shape of the dl in this species broadly sub triangular, with a squarish base, a blunt tip set with a comb-like structure and with a sinuous dorsal margin devoid of protuberances. On main body, a relatively short and straight, well-sclerotized; cop short and distally curved.

Measurements (in flm): Male: L = 510 (n = 1); H = 319 (n = 1) Female: L = 547 (n = 2); H = 320 (n = 2)

Differential diagnosis. The new species can easily be distinguished by the elongated valves and by rounded valve and the presence of the bulbous nodes and by the shape of the hemipenis. 287

Figure 11: Gomphocythere coheni n.sp. (female LP0030). A. First antenna (AI), B. Second antenna (A2), C. Mandible (Md). (Scale = 100 flm for A-C). 288

Figure 11 289

Figure 12: Gomphocythere coheni n.sp. (female LP0030). A. Third thoracic leg (P3), B. Second thoracic leg (P2). (Scale = 100 Ilm for A-B). 290

Figure 12 291

Gomphocythere woutersi sp.n.

Type locality. LT/91/21 (Martens expedition); Lake Tanganyika, East Africa. Dama River delta, Rumonge, Burundi. Water depth 50 m; Coordinates: 3°57' S, 29°26' E.

Type material. The following collections have been used for the present descriptions:

LT/91/21 A and B: sample collected amongst sand and green algae, by Dr. K. Martens on 5/21/1991. Measurements at time of collection: pH= 9-9.5, alkalinity= 6.07 meqll. All illustrated specimens originate from this sample.

Holotype : a male, with soft parts dissected in glycerin in a sealed slide and with valves stored dry (LP 0047).

Allotype: a female, dissected and stored as the holotype (LP 0023).

Repository: all types, both dissected and in spirit, are stored in the Ostracod Collection of the RB.I.N .Sc.

Derivation of name : this species is named after Dr. Karel Wouters, Royal Belgian Museum of Natural History, in appreciation of his contribution to this project. 292

Diagnosis. Large and elongated species, with curved dorsal margin sloping towards the caudal side, especially in the male. Posterior part of carapace in female relatively narrow and rounded in dorsal view. Small, prominent protuberances evenly spaced over carapace. Mxl palp weakly sclerified and undivided. Hemipenis and dl sub-triangular, with squarish base, blunt tip set with comb-like structure, medio-Iateral indention without a protuberance; ur relatively short, cop short and curved in its distal half.

Additional description of male: Valves (Figure 13) strongly sclerified and externally reticulated. Anterior margin broadly rounded, posterior margin more narrow, but still evenly rounded. Greatest height situated at about one fourth from the anterior, from there the straight dorsal margin sloping towards the caudal side. Ventral margin weakly sinuous. Hinge (Figure 13) inverse lophodont, with prominent and elongated, but simple cardinal teeth on LV fitting in cardinal sockets on RV; intercardinal bar on RV smooth. Both valves (Figure 13) with prominent selvage; calcified inner lamellae narrow on both anterior and posterior sides. Carapace in dorsal and ventral views narrow, with nearly parallel sides. Al (Figure 14) with 3-segmented endopodite, penultimate segment undivided. First segment of endopodite twice as long as wide, with one long, medio-ventral seta. Second segment short and squarish, with one long apical dorsal seta. Third segment with one medio-ventral and two medio-dorsal setae and one apical dorsal setae. Terminal segments 293 elongated, with two sub-apical setae, slightly fused at the base with a seta, this seta longer than the actual sensorial club. A2 (Figure 14) with endopodite three-segmented. First segment short, bearing one long seta. Second segment long, with medio-ventrally one setae and one long dorsal seta (the latter reaching to tip of terminal segment); apically one stout, claw-like ventral seta. Terminal segment with three stout and short claws, one subapically and two apically inserted. Md (Figure 14) without special features. Palp four-segmented. First segment with respiratory plate consisting of three rays only (two long, one short) and with two subequal ventral setae. Second segment long with one medial-ventral setae. Third segment of similar shape, with 5 apical setae, three dorsally and two ventrally inserted. Terminal segment pyramidal, with two claws. Mx1 with three endites of normal shape and chaetotaxy; palp weakly sclerotized and unsegmented, set with three lateral and two apical setae. P(1-3) (Figure 15) all walking limbs; P(1) the shortest and P(3) the longest and with the longest claw. First segment in these limbs with 2 long dorsal and one long and hairy ventral seta (the latter proximally inserted) and with one (on P(2) and P(3) or two (on P(1» knee-setae as is typical of the subfamily. Hemipenis has a large muscular body and a prominent, articulating distal lobe (dl), devoid of setae. Shape of the dl in this species broadly sub triangular, with a squarish base, a blunt tip set with a comb-like structure 294

and with a sinuous dorsal margin devoid of protuberances. On main body, ur relatively short and straight, well-sclerotized; cop short and distally curved.

Additional description of females. Valves (Figure 13) generally as in the male, but larger and with straight dorsal margin only slightly sloping towards the caudal sides; anterior margin still more broadly rounded than posterior one. Structure of hinge and valve margin as in the male. Carapace in dorsal and ventral view with posterior part more swollen than in the male, but considerably less so than in most other species of this genus. Anterior tip in dorsal view bluntly pointed, posterior edge rounded. All appendages (Figures 16-17) generally as in the male; Two pairs of "hypostomeal lobes" present posterior to labium, anterior lobes approximately twice as long as posterior ones. Posterior part of abdomen with one furca bearing two furcal setae and three hirsute furcal lobes, one seta and one forked organ. Genital orifice small and rounded.

Measurements: Male: L = 485-490 (n = 1); H = 215-220 (n = 1); W = NA Female: L = 600-650 (n = 1); H = 240-250 (n = 1); W =NA

Differential diagnosis. The new species can easily be distinguished from almost all of its congeners by the elongated valves and by the shape of the hemipenis. 295

Figure 13: Gomphocythere woutersi n.sp.: SEM micro graphs of the valves. A. female LV, external view (LP0023). B. female RV, internal view (LP0023). C. female RV, external view (LP0047). D. male RV, external view (no number). (Scale bars = 100 ~m). 296 297

Figure 14: Gomphocythere woutersi n.sp. (male LP0033). A. First antenna (AI), B. Second antenna (A2), C. Mandible (Md). (Scale = 100 /lm for A-C). 298

o

Figure 14 299

Figure 15: Gomphocythere woutersi n.sp. (male LP0033). A. Third thoracic leg (P3), B. First thoracic leg (PI). (Scale = 100 J..lm for A-B). 300

100 Jim

Figure 15 301

Figure 16: Gomphocythere woutersi n.sp. (female LP0023). A. First antenna (AI), B. Second antenna (A2). (Scale = 100 11m for A-B). 302

Figure 16 303

Figure 17: Gomphocythere woutersi n.sp. (female LP0023). A. Third thoracic leg (P3), B. First thoracic leg (PI). (Scale = 100 /lm for A-B). 304

E! :::1. o o ~

Figure 17 305

ADDITIONAL NOTES ON SPECIES USED IN THIS STUDY

The following are short notes about each species to help those interested in working on this group in the future.

G. aethiopis- Martens (1993) speculates that this is the same species as G. parcedilatata. Few specimens are know from either species for comparison.

G. alata- the most abundant species In Lake Tanganyika. Occurs throughout the lake at depths between 0-80 m. Males and females are usually equal in abundance.

G. angulata- the most abundant Gomphocythere species in all of Mrica. Many geochemical studies have been done using this species.

G. angusta- no voucher specimens available. Only literature descriptions.

G. capensis- could very well be the same species as G. expansa. The male hemipenes are identical in both G. capensis and G. expansa. G. capensis was originally described as Cytheridella capensis by Muller (1914). He also misidentified females of Cytheridella obtusata (later renamed Gomphocythere obtusata) as Gomphocythere expansa. Specimens for comparison were not available. K. Martens is currently addressing this issue (see Martens, 1993). 306

G. coheni- the easiest Gomphocythere species to identify because of its small size and big nodes. Very abundant. Often found in cores. Soft parts are much harder to find. Nodes can vary somewhat. Referred to in Cohen's database as G. n.sp.

G. cristata- next to G. alata, one of the easiest to identify. Less common

than G. alata, but males and females are present in most samples.

G. curta- easily mistaken for G. Lenis and G. simplex. Appears to have ecophenotypic variation that can be analyzed morphometrically.

G. downingi- easily recognized by its long alae. Many soft parts material available. Also, the easiest to dissect because of its large size. Referred to in Cohen's database as G. sp. 3.

G. expansa- informally synonymized with G. capensis Muller (1914). See G. capensis for discussion. Some minor carapace features appear to separate them and so I treated them as separate species in the analysis to see how they would fall out in the analysis. They were always a stable sister group. This taxonomic issue will be thoroughly examined by myself and Martens in the immediate future and before publication of this phylogeny. 307

G. lenis- male specimen is erroneously designated as a female in Rome's collection (O.C. 669a-c) and publication (Rome, 1962). Published description of G. lenis by Rome (1962) does not contain sexual characters because of the lack of recognition of the semi-erect hemipenis in the dissection. Species easily confused with G. curta and G. simplex. G. lenis will be re-illustrated in publication of this phylogeny.

G. obtusata- one of the most abundant species throughout Africa. Designated as the type species for the genus. There are doubts of the conspecificity of the populations of this species from East Africa and South Africa (see Martens, 1993 for further discussion). Males are only known from the South African populations.

G. ortali- new species described by Martens (1993). Extends the range of Gomphocythere northward. Type material in Royal Belgian Museum of Natural History.

G. parcedilatata--see notes for G. aethiopis

G. simplex- difficult to tell fl'om G. curta. Rome did not adequately illustrate the outer carapace. Will be re-illustrated in a subsequent publication, along with new species referred to here.

G. wilsoni- sometimes difficult to identify. Identified in Cohen's database as G. sp. 13. 308

G. woutersi- has interesting sieve pore development. Easily recognized. Identified in Cohen's database as G. sp. 11. 309

REFERENCES

Archie, J.W., 1989. A randomization test for phylogenetic information in systematic data. Systematic Zoology, v. 38. p.239-252.

Baum and Larson, 1991. Adaptation reviewed: A phylogenetic methodology for studying character macroevolution. Systematic Zoology, v. 40. p. 1-18.

Benson, R.H., 1981. Form, function and architecture of ostracod shells. Annual Review of Earth and Planetary Science, v. 11. p. 59-80.

Brooks, D. R., and McClennan, D.A., 1991. Phylogeny, Ecology and Behavior: A Research Program in Comparative Biology. University of Chicago Press, Chicago. 434p.

Brown, D.S., and Mandahl-Barth, G., 1987. Living molluscs of Lake Tanganyika: a revised and annotated list. Journal of Conchology, v. 32. p. 305-327.

Carbonnel, P., and Jacobzone, M., 1977. L'ornamentation chez Loxoconchidae: un moyen de prediction (paleo)ecologique. In: H. Loffler and D. Danielopol (eds.). Aspects of Ecology and Zoogeography of Recent and Fossil Ostracoda,. Proceedings of the 6th International Symposium on Ostracods. Dr W. Junk Publishers: The Hague. p. 121-133.

Carbonel, P., 1983. Ostracoda as indicators of ionic concentrations and dynamic variations: methodology (Lake Bogoria, Kenya). In: R.F. Maddocks (ed.) Applications of Ostracoda. 8th International Symposium Ostracoda. University of Houston Geosciences, Houston. p. 264-276.

Carbonnel, P., Colin, J.P., Danielopol, D., Loffier, H., and Neustrueva, 1., 1988. Paleoecology of limnic ostracodes: A review of some major topics. Palaegeography, Palaeoclimatology, Palaeoecology, v. 62. p. 413-461.

Carpenter, J.M., 1988. Choosing among multiple equally parsimonious cladograms. Cladistics, v. 4: p. 291-296.

Chippindale, P.T., and Wiens, J.J., 1994. Weighting, partitioning, and combining characters in phylogenetic analysis. Systematic Biology, v.43. p. 278-287. 310

Cohen, A.S., 1982. Ecological and Paleoecological Aspects of the Rift Valley Lakes of East Africa. Ph.D. Dissertation. University of California, Davis. 314 p.

Cohen, A.S., 1986. Distribution and faunal associations of benthic invertebrates at Lake Turkana, Kenya. Hydrobiologia, v. 74. p. 179-197.

Cohen, A.S., and Johnston, M., 1987. Speciation in brooding and poorly dispersing lacustrine organisms.. PALAIOS, v. 2. p. 426-435.

Cohen, A.S., Dussinger, R., and Richardson, J., 1983. Lacustrine paleochemical interpretations based on eastern and southern African ostracods. Palaeogeography, Palaeoclimatology, Palaeoecology, v. 43. p. 129-151.

Cohen, A.S., Soreghan, M.S., and Scholz, C.A., 1993. Estimating the age of formation of lakes: An example from Lake Tanganyika, East African Rift system. Geology, v. 21. p.511-514.

Colin, J.P. and Danielopol, D.L., 1978. New data on the systematics of the Limnocytheridae (Ostracoda, Cytheracea). Geobios, v. 11: p. 563-567.

Colin, J.P. and Danielopol, D.L., 1979. Why most of the Timiriaseviinae (Ostracoda, Crustacea) became extinct. Geobios, v. 12: p. 745-749.

Coulter, G.W., 1991. Lake Tanganyika and its Life. Oxford University Press: London. 354 pp.

Coulter, G.W., 1994. Lake Tanganyika. In: K. Martens, B. Goddeeris, and G. Coulter (eds.). Speciation in Ancient Lakes, Archiv fur Hydrobiologie Ergebnisse der Limnologie. Stuttgart. p. 13-18.

Danielopol, D.L., Martens, K., and Casale, L.M., 1990. Revision of the genus Leucocythere KAUFMANN, 1892 (Crustacea, Ostracoda, Limnocytheridae), with the description of a new species and two new tribes. Bulletin Kon. Belg. Inst. Natuurwetensch., Biol., v. 59. p.63-94.

DeDeckker, P., 1981. Taxonomy and ecological notes of some ostracods from Australian inland waters. Transactions of the Royal Society of Southern Australia, v. 105: p. 91-138. 311

Donoghue, M.J., 1989. Phylogenies and the analysis of evolutionary sequences, with examples from seed plants. Evolution v. 43: p. 1137-1156.

Farris, J., 1972. Estimating phylogenetic trees from distance matrices. American Naturalist, v. 106, p. 645-668.

Farris, J., 1982. Simplicity and informativeness in systematics and phylogeny. Systematic Zoology, v. 31. p.413-444.

Fryer, G., 1959. The trophic interrelationships and ecology of some littoral communities of Lake Nyasa with especial reference to the fishes, and a discussion of the evolution of a group of rock­ frequenting Cichlidae. Proceeding of the Zoolological Society of London., v. 132. p. 153-281.

Fryer, G., 1972. Conservation of the Great Lakes of East Africa: A lesson and a warning. Biological Conservation, v. 4. p. 256-262.

Fryer, G., 1977. Evolution of species flocks of cichlid fishes in African lakes. Zeitschrift fur Zoologische Systematik Evolutionsforschung. v. 15. p. 141-165.

Fryer, G. and lIes, T.D., 1972. The Cichlid Fishes of the Great Lakes of Africa: Their Biology and Evolution. Oliver and Boyd: Edinburgh. 641 pp.

Gould, S.J., and Vrba, E.S., 1982. Exaptation-A missing term in the science of form. Paleobiology, v. 8. p. 4-15. Greenwood, P.H., 1981. Species flocks and explosive evolution, In: P.H. Greenwood and P.L. Forey (eds.). Chance, Change and Challenge, The Evolving Biosphere, Cambridge University Press, London. p. 61-74.

Grekoff, N., 1953. Sousclasse des Ostracodes. In: J. Piveteau (ed.). Traite de Paleontologie, Paris, 3: p. 269-294.

Hartmann, G. and Puri, H. S., 1974. Summary of neontological and paleontological classification of Ostracoda .. Mitteilungen aus dem Hamburgischen Zoologischen Museum und Institut, v. 70. p. 7-73

Hauser, D.L., and Presch, W., 1991. The effect of ordered characters on phylogenetic reconstruction. Cladistics, v. 7. p. 243-265. 312

Hecky, R.E., and Degens, E.T., 1973. Late Pleistocene-Holocene chemical stratigraphy and paleolimnology of the rift valley lakes of Central Africa. Woods Hole Oceanographic Institution Technical Report, v. 73. 93p. Henderson, P.A., 1990. Freshwater ostracodes. Synopses of the British Fauna, 42. The Linnean Society of London and the Estuarine and Coastal Sciences Association. Universal Book Services, The Netherlands. 228p.

Hennig, W., 1966. Phylogenetic systematics. University of Illinois Press. Urbana. Hillis, D.M., 1991. Discriminating between phylogenetic signal and random noise in DNA sequences. In: M.M. Miyamoto and J. Craycraft (eds.). Phylogenetic analysis of DNA sequences. Oxford University Press: New York. p. 278-294.

Hillis, D.M., and Huelsenbeck, J.P., 1992. Signal, noise, and reliability in molecular phylogenetic analyses. Journal of Heredity, v. 83. p. 189- 195. Holmes, J.P., 1992. Nonmarine ostracodes as Quaternary palaeoenvironmental indicators. Progress in Physical Geography v. 16. p. 425-431.

Huelsenbeck, J.P., 1991. Tree length distribution skewness: an indicator of phylogenetic information. Systematic Zoology, v. 40. p. 257-270.

Huelsenbeck, J.P., Swofford, D.L., Cunningham, C.W., Bull, J.J., and Waddell, P.J., 1994. Is character weighting a panacea for the problem of data heterogeneity in phylogenetic analysis. Systematic Biology, v. 43. p.288-291.

Jolly, D., Bonnefille, R., and Roux, M., 1994. Numerical interpretation of a high resolution Holocene pollen record from Burundi. Palaeogeography, Palaeoclimatology, and Palaeoecology, v. 109. p. 357-370.

Keen, M., 1975. The paleobiology of some upper Paleogene freshwater ostracodes. In: F. Swain (ed.). Biology and paleobiology of ostracoda. Bulletin American Paleontology, v. 65. p. 271-283.

Klie, W., 1939. Ostracoden aus dem Kenia-Gebiet, vornhemlich von dessen Hochgebirgen. Internationale Revue der gesamten Hydrobiologie und Hydrogeographie, v.39. p.99-161. 313

Kotzian, S., 1974. New freshwater ostracodes of the genus Chlamydotheca from Brazil. Anales Academia Brasileira do Ciencias, v. 46. p. 423-467.

Lauder, G.V., Leroi, A.M., and Rose, M.R., 1993. Adaptations and history. Tree. v.8. p.294-297.

Liebau, A., 1977. Carapace ornamentation of the Ostracode Cytheracea: principles of evolution and functional significance. In: H. Lelffler and D. Danielopol (eds.). Aspects of Ecology and Zoogeography of Recent and Fossil Ostracoda,. Proceedings of the 6th International Symposium on Ostracods. Dr W. Junk Publishers: The Hague. p. 107-120.

Maddison, D.R., 1991. The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology, v. 40. p.315-328.

Maddison, W.P., and Maddison, D.R., 1992. MacClade: analysis of phylogeny and character evolution. Sinauer: Sunderland, Massachusetts. Maddison, W.P., Donoghue, M.J., and Maddison, D.R., 1984. Outgroup analysis and parsimony. Systematic Zoology, v. 33. p. 83-103.

Martens, K., 1990. Revision of African Limnocythere s.s. BRADY, 1867 (Crustacea, Ostracoda) with special reference to the Eastern Rift Valley Lakes: morphology, taxonomy, evolution and (palaeo)ecology. Archiv fuer Hydrobiologie Ergebnisse der Limnologie. v.83. p. 453-524.

Martens, K., 1993. On the taxonomy and zoogeography of the genus Gomphocythere SARS, 1924 (Crustacea, Ostracoda), with the description of a new species from the Nahal Dan (Israel). Belg.ian Journal of Zoology, v. 123. p. 39-54.

Martens, K., 1994. Ostracod speciation in ancient lakes: a review. In: K. Martens, B. Goddeeris, and G. Coulter (eds.). Speciation in Ancient Lakes, Archiv fur Hydrobiologie Beih. Ergebn. Limnol. Stuttgart. p. 203-222.

Martin, V.G.P.R., 1940. Ostracoden des norddeutschen Purbeck und Wealden. Senckenbergiana, v. 22p. 276-319.

Mayr, E., 1969. Principles of systematic zoology. McGraw-Hill, New York. 314

Mayr, E., 1984. Evolution of fish species flocks: a commentary, In: AA Echelle and I. Kornfield (eds.). Evolution of fish species flocks. University of Maine at Orono Press. 257p.

McCune, A. R., 1990. Evolutionary novelty and atavism in the Semionotus complex: Relaxed selection during colonization of an expanding lake. Evolution, v. 44. p. 71-85.

Meyer, A., 1993. Phylogenetic relationships and evolutionary processes in African cichlids. Trends in Ecology and Evolution, v. 8. p. 279- 284.

Meyer, A., Montero, C., and Spreinat, A., 1994. Evolutionary history of the cichlid fish species flocks of the East African great lakes inferred from molecular phylogenetic data. In: K. Martens, B. Goddeeris, and G. Coulter (eds.). Speciation in Ancient Lakes, Archiv fUr Hydrobiologie Ergebnisse Limnologie Stuttgart. p. 407- 423.

Michel, E., Cohen, AS., West, K., Johnston, M.R., and Kat; P.W., 1992. Large African lakes as natural laboratories for evolution: examples from the endemic gastropod fauna of Lake Tanganyika. Mitteilungen der Internationalen Vereinigung fuer Limnologie, v. 23. p. 85-99.

Mittermeier, R.A, 1988. Primate diversity and the tropical forest: case studies from Brazil and Madagascar and the importance of the megadiversity countries. In: Wilson, E.O. and Peter, F.M. (eds.), Biodiversity. National Academic Press, Washington, D.C. pp. 145- 154.

Mittermeier, R.A, and Werner, T.B., 1990. Wealth of plants and animals unites 'megadiversity' countries. Tropicus, v. 4. p. 4-5.

Moore, R.C., 1961. Ostracoda. Treatise of Invertebrate Paleontology. University of Kansas Press, Lawrence. 442p.

Okada, Y., 1982. Structure and cuticle formation of the reticulated carapace of the ostracode Bicornucythere bisanensis. Lethaia, v. 15. p.85-101.

Palacios-Fest, M.R., Cohen, AS., and Anadon, P., 1994. Use of ostracodes as paleoenvironmental tools in the interpretation of ancient lacustrine records. Revista Espanola de Paleontologia, v. 9. p. 145- 164. 315

Pogue, M.G., and Mickevich, M.F., 1990. Character definitions and character state delineation: the bete noire of phylogenetic inference. Cladistics. v. 6. p. 319-361.

Purper, 1., 1974. Cytheridella boldii Purper, sp. nov. (Ostracoda) from Venezuela and a revision of the Genus Cytheridella Daday, 1905. An. Acad. Brasil. Cienc. v. 46. p.635-662.

Ribbink, A.J., 1991. Distribution and ecology of the cichlids of the Mrican Great Lakes. In: M.H.A. Keenleyside (ed.) Cichlid Fishes, Fish and Fisheries Series 2. Chapman and Hall, London. p. 36-59. Rome, D.R., 1962. Exploration hydrobiologique du Lac Tanganyika (1946-1947). v. 3 Ostracodes. Inst. Royal Sciences Nat. Belg. Brussels. 309p.

Rosendahl, B., Reynolds, D., Lorber, P., Burgess, C., McGill, J., Scott, D., Lambiase, J., and Derksen, S., 1986. Structural expression of rifting: Lessons from Lake Tanganyika, Mrica. In: L. Frostick, et al. (eds.). Sedimentation in the African rifts: Geological Society of London Special Publication, v. 25. p. 29-43.

Rosenfeld, A., and Vesper, B., 1977. The variability of the sieve pores in recent and fossil species of Cyprideis torosa (Jones, 1850), as an indicator for salinity and paleosalinity. In: H. LOffler and D. Danielopol (eds.). Aspects of Ecology and Zoogeography of Recent and Fossil Ostracoda,. Proceedings of the 6th International Symposium on Ostracods. Dr W. Junk Publishers: The Hague. p. 55-69.

Sanderson, M.J., 1991. Confidence limits on phylogenies: the Bootstrap revisited. Cladistics, v. 5. p. 113-129.

Sanderson, M.J., and Donoghue, M.J., 1989. Patterns of variation in levels of homoplasy. Evolution, v. 43. p. 1781-1795. Sars, G.O., 1910. Zoological results of the 3rd Tanganyika expedition (1904-1905). Report on Ostracoda. Proceedings of the Zoological Society of London, v. 48. p. 732-760.

Sars, G.O., 1924. The freshwater Entomostraca of the Cape Province (Union of South Mrica). Ostracoda. Annals of the South African Museum, v. 20. p. 2-20.

SAS Institute, 1989. SAS / STAT® User's Guide: Version 6. 316

Simpson, G.G., 1961. Principles of Animal Taxonomy. Columbia University Press: New York. 247p.

Smith, G.R., and Todd, T.N., 1984. Evolution of species flocks in north temperate lakes. In: AA Echelle and I. Kornfield (eds.). Evolution offish species flocks. University of Maine at Orono Press. 257p.

Snoeks, J., Riiber, L., and Verheyen, E., 1994. The Tanganyika problem: comments on the taxonomy and distribution patterns of its cichlid fauna. In: K. Martens, B. Goddeeris, and G. Coulter (eds.). Speciation in Ancient Lakes, Archiv fUr Hydrobiologie. Stuttgart. p. 355-372.

Sturmbauer, C., and Meyer, A, 1992. Genetic divergence, speciation and morphological stasis in a lineage of Mrican cichlid fishes. Nature, v. 358. p. 578-581.

Suter, S.J., 1994. Cladistic analysis of the living cassiduloids (Echinoidea), and the effects of character ordering and successive approximations weighting. Zoological Journal of the Linnc~an Society, v. 112, p. 363-387.

Swofford, D.L., 1993. PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1. Computer Program. Illinois Natural History Survey, Champaign, IIlinois.

Swofford, D.L., and Maddison, W.P., 1987. Reconstructing ancestral character states under Wagner parsimony. Mathematical Biosciences, v. 87. p. 199-229. TaIling, J.F., and TaIling, I.B., 1965. The chemical composition of African lake waters. Internationale Revue der gesamte Hydrobiologie, v. 50. p.421-463. Tiercelin, J., and Mondeguer, A, 1991. The geology of the Tanganyika Trough. In: G. Coulter (ed.). Lake Tanganyika and its life. New York, Natural History Museum. p.7-48.

Tsukagoshi, A, 1990. Ontogenetic change of distributional patterns of pore systems in Cythere species and its phylogenetic significance. Lethaia, v. 23. p. 225-241.

Van Morkhoven, F.P.C.M., 1962. Post-Paleozoic Ostracoda: Their Morphology, Taxonomy, and Economic Use, v. 1 and 2. Elsevier Publishing Co.: New York. p.204. 317

Wagner, G.P., 1989. The origin of morphological characters and the biological basis of homology. Evolution, v. 43. p. 1157-1171.

Watrous, L.E., and Wheeler, Q.D., 1981. The outgroup comparison method of character analysis. Systematic Zoology, v. 30. p.l-l1.

Wiley, E.O., Siegel-Causey, D., Brooks, D.R., and Funk, V.A, 1991. The Compleat Cladist: A Primer of Phylogenetic Procedures. University of Kansas Museum of Natural History, Special Publication no. 19. Lawrence, Kansas. 158 p.