A Systematic Study of the Clam Shrimp Genus

Home , Clam shrimp, Diplostraca, Eulimnadia, Imnadia, Leptestheriidae, Limnadiidae

A SYSTEMATIC AND BIOGEOGRAPHIC STUDY OF THE CLAM SHRIMP

GENUS EULIMNADIA PACKARD, 1874 (BRANCHIOPODA: SPINICAUDATA:

LIMNADIIDAE) AND AN INVESTIGATION INTO THE EVOLUTION AND

MAINTENANCE OF ANDRODIOECY IN EULIMNADIA DAHLI DAKIN, 1914

A dissertation submitted to Kent State University

in cooperation with The University of Akron

in partial fulfillment of the requirements

for the degree of Doctor of Philosophy

Sadie K. Reed Stimmell

August 2013

Dissertation written by

Sadie K. Reed Stimmell B.S., Northern Arizona University, 2000 M.S., Northern Michigan University, 2003 Ph.D., Kent State University, 2013

Approved by

______, Chair, Doctoral Dissertation Committee Dr. Stephen C. Weeks

______, Co-advisor, Doctoral Dissertation Committee Dr. Walter R. Hoeh

______, Members, Doctoral Dissertation Committee Dr. Andrea Case

______, Dr. Lisa E. Park

______, Dr. Alison Smith

Accepted by

______, Chair, Department of Biological Sciences Dr. Laura G. Leff

______, Dean, College of Arts and Sciences Dr. Raymond Craig

TABLE OF CONTENTS LIST OF FIGURES………………………………………………………………..v

LIST OF TABLES…………………………………………………………………vi

ACKNOWLEDGMENTS…………………………………………………………vii

CHAPTERS

I. GENERAL INTRODUCTION……………………………………………..1

II. A SYSTEMATIC STUDY OF EULIMNADIA

Abstract…………………………………………………………………….3

Introduction………………………………………………………………..4

Materials and Methods…………………………………………………….11

Results………………………………………………………………………16

Discussion…………………………………………………………………...29

Acknowledgements………………………………………………………….38

III. A BIOGEOGRAPHIC ANALYSIS OF THE EULIMNADIA WITH AN

ASSESSMENT OF THE EVOLUTION OF ANDRODIOECY

Abstract……………………………………………………………………..39

Introduction…………………………………………………………………40

Materials and Methods………………………………………………………42

Results……………………………………………………………………….44

Discussion……………………………………………………………………50

Acknowledgements…………………………………………………………..55

IV. AN INVESTIGATION OF THE MAINTENANCE OF ANDRODIOECY VIA

METAPOPULATION DYNAMICS IN EULIMNADIA DAHLI

Abstract………………………………………………………………….56

iii

Introduction……………………………………………………………...57

Materials and Methods…………………………………………………..63

Results…………………………………………………………………….65

Discussion…………………………………………………………………71

Acknowledgements………………………………………………………..78

V. GENERAL DISCUSSION AND CONCLUSIONS………………………..80

REFERENCES…………………………………………………………………..86

LIST OF FIGURES

Figure 2.1 Phylogenetic tree resulting from analysis of both cytb and EF1

Figure 2.2 Phylogenetic tree resulting from analysis of EF1

Figure 2.3 Phylogenetic tree resulting from maximum likelihood analysis of cytb

Figure 2.3a Clade from phylogenetic tree resulting from maximum likelihood analysis of cytb

Figure 2.4 Phylogenetic tree resulting from Bayesian inference analysis of cytb

Figure 2.4a First clade from phylogenetic tree resulting from Bayesian inference analysis of cytb

Figure 2.4b Second clade from phyloenetic tree resulting from Bayesian inference analysis of cytb

Figure 3.1 Continental plates used for distributional areas

Figure 3.2 Ancestral area reconstruction for Eulimnadia based on both EF1 and cytb Bayesian analysis

Figure 3.3 Present day species distributions and biogeographic dispersal events

Figure 4.1 Percent males versus genetic diversity

Figure 4.2 Gene flow versus distance of outcrops

LIST OF TABLES

Table 2.1 List of valid Eulimnadia species names in Brtek (1997)

Table 2.2 Additional Eulimnadia species

Table 2.3 List of specimens included in phylogenetic analysis

Table 2.4 Primer Pairs used for DNA amplification

Table 2.5 Current taxonomic status of the North American Eulimnadia species

Table 4.1 Population Descriptive Statistics

Table 4.2 Pairwise FST and Nm Estimates

Table 4.3 Heirarchical model of population genetic structure

ACKNOWLEDGEMENTS

Many times throughout my dissertation research, I heard the quest for a PhD described as a lonely pursuit. This is very much not the case as I look back and find I am overwhelmed with the number of people who helped me along the way. I am filled with the deepest gratitude to all of the people who have supported me in this degree.

First, I would like to extend my thanks to my doctoral advisor, Dr. Steve Weeks.

He was a true mentor in every aspect of my academic career, from research and writing to teaching and field work. He has been a friend and was always quick to offer encouragement and praise when I needed it. He never gave up and always believed I could accomplish this dissertation, and that is truly why I have. I would like to thank my co-advisor, Dr. Walter R. Hoeh for much time learning techniques in the lab and much of his encouragement along the way. I would also like to thank my committee members,

Dr. Lisa Park Bousch, Dr. Andrea Case, and Dr. Alison Smith for many helpful comments and suggestions that have strengthened this document.

I have been so fortunate to receive a great amount of help and support from a number of professors at the University of Akron. Dr. Joel Duff never hesitated to offer so much of his time in the lab and was always willing to take time to explain things over and over until I understood. This research would certainly not have been completed without him. Dr. Francisco Moore was a huge help in all of my statistical questions and always had an open ear for ideas and frustrations. Thank you also to Dr. Peter Niewiarowski, Dr.

Brian Bagatto, Dr. Richard Londraville and Dr. Randy Mitchell. I would also very much like to thank Dr. Mark Kershner for his help and support at Kent State University.

vii

My field experience in Western Australia was filled with adventure and once-in- a-lifetime experiences. It would not have been possible without the limitless generosity of Dr. Brenton Knott, who welcomed us into his lab and home. I will cherish the memories of Friday lab meetings complete with “drinkable” reds and cheese and classical music. All of the wonderful people at the School of Animal Biology at the University of

Western Australia, in Perth contributed to our successful research and fantastic experience. Thank you to Kerry, Debra, Danny, Magdalena, Wally, Bonnie, and Rob.

I am so grateful to have enjoyed three field seasons in Las Cruces, NM at the

Jornada LTER. We were so fortunate to have the help and friendship of Dr. Naida

Zucker and Dr. Richard Spellenberg, along with Dr. Michele Nishiguci’s and her lab team. So many friendships were formed with the students in the Department of biology at

NMSU, including Vinod, Bryan, TuShun, and Will. Thank you for the lab space and so much fun.

I can truly say that one person was with me through all the ups and downs and experiences of graduate school, Dr. Chiara Benvenuto. She was with me in every field season all over the globe and shared every adventure. Thank you for your continuous support and friendship through thick and thin.

The “clam shrimp crew” was a conglomeration of wonderful people who were always helpful and friendly and supportive in every way. Thank you to Allisa Calabrese,

Tom Sanderson, Beth Wallace, and Amanda Crow.

I formed a number of friendships with so many fellow students at the University of Akron and I am particularly thankful to Jenn Purrenhage for all of her talks through my anxieties.

viii

So much happens in the many years it takes to complete a doctorate. I was fortunate to marry a loving and supportive person who has always encouraged me and done everything in his power to help be along the way. Thank you so very much, Sean

Stimmell, for always believing in me and adding a healthy dose of humor to every situation. We were also blessed with two children, Finnbar and Josette, during the course of this degree and they have added to my life in so many ways and given me the perspective needed to finish.

Thank you to the number of generous people who have helped with childcare during my final writing phase: Kate Stimmell, Amy Carlson, and Jenn DePiano.

Thank you to my parents who have encouraged me in every step, from my very first. Tish Lacy Reed and Raymond Reed, this would not have been accomplished without you.

This dissertation was by no means a lonely pursuit with the number of people who stood by me and helped in every way they could. Thank you to any one I may have left out as I was blessed to be surrounded by so many wonderful folks. I am indebted to all of you and I am sincerely grateful.

I. GENERAL INTRODUTION

The clam shrimp genus Eulimnadia is a cosmopolitan taxon comprising approximately 54 species (Brtek 1997; Martin & Belk 1989; Roessler 1990; Pereira &

Garcia 2001; Durga-Prasad & Simahachalam 2004; Simahachalam 2004, 2005; Timms &

McLay 2005; Babu & Nandan 2010; Rogers et al. 2010). It is the most speciose spinicaudatan genus and is of particular interest due to its exhibition of an extremely rare mating system - androdioecy (Weeks et al. 2006; Weeks 2012). Androdioecy is the co- occurrence of males and hermaphrodites in populations in the absence of true females.

The evolutionary benefits of such a makeup are unclear. Males can play an important role in hermaphroditic populations to avoid inbreeding depression and to supplement male allocation in sperm-limited hermaphrodites (Wolf & Takebayashi, 2004). Yet males have difficulty invading hermaphroditic populations because of reduced mating opportunities due to the ability of hermaphrodites to self-fertilize. This requires males to more than double their mating success (i.e., successful fertilizations) relative to hermaphrodites to effectively invade the population. Theoretical models of mating system evolution that have incorporated these constraints show that androdioecy is unlikely to evolve, and that those few cases found in nature are likely to be short-term, transitional populations that are evolving from hermaphroditism to dioecy, or vice versa (Lloyd, 1975; Charlesworth,

1984).

The rarity of androdioecious mating systems was exemplified when Charlesworth

(1984) showed most of the species described as androdioecious to that date were actually functionally dioecious. Pannell (2002) reviewed six plant species that are truly androdioecious, and Weeks et al. (2006) reviewed androdioecy in the animal kingdom

1 2 and documented this mating system in one chordate, seven nematodes, and 28 crustacean species. Weeks et al. (2006) present four additional androdioecious crustaceans. Within the Crustacea, the clam shrimp genus Eulimnadia represents the most widespread occurrence of androdioecy within any one generic taxon with 16 species confirmed as androdioecious (Weeks et al. 2005, 2006, 2008, 2012). For this reason, the genus

Eulimnadia presents a unique opportunity to study this rare mating system.

My PhD dissertation research comprises three projects with the objective of furthering our understanding of this interesting genus and unique mating system. The first project (Chapter II) is the construction of a molecular phylogeny of the genus Eulimnadia utilizing nuclear and mitochondrial markers to investigate infra-generic relationships and to explore the validity of current taxonomic delimitations. The second project (Chapter

III) aims to evaluate Eulimnadia species distributional patterns in a biogeographic context to gain insight into the evolution of androdioecy throughout the genus. The third project (Chapter IV) is an empirical field test of Pannell’s (1997, 2000) metapopulation model of the maintenance of androdioecy using E. dahli populations occurring on granite outcrops in Western Australia. These studies will contribute to our knowledge of this interesting crustacean genus and further our understanding of the evolution and maintenance of androdioecy.

II. A SYSTEMATIC STUDY OF EULIMNADIA

Abstract

The clam shrimp genus Eulimnadia Packard, 1874 is the most speciose and widely distributed genus in the Spinicaudata. It is also an important genus in the study of mating system evolution due to its exhibition of the rare mating system, androdioecy, where males and hermaphrodites occur in the absence of true females. Yet, in order to study the prevalence of this rare mating system, a solid systematic foundation is necessary for understanding species delimitations and the evolutionary relationships among these species. Taxonomic determinations based on morphology have been controversial because of the intraspecific variability shown in many of the characters used to date, although recently, much emphasis has been placed on egg shell morphology for reliable species-specific characters. In this study, I explore the phylogenetic relationships of 22 species of Eulimnadia and test previously proposed synonymies of the

North American taxa based solely on egg shell morphology. Phylogenetic studies were based on sequence data of the mitochondrial DNA-encoded cytochrome b (cyt b) and the nuclear-encoded elongation factor 1 α (EF1α), analyzed using Maximum Likelihood

(ML) and Bayesian Inference (BI) approaches. Phylogenetic analyses support the monophyly of Eulimnadia, yet a large amount of polyphyly for specimens identified as the same species highlights the troubles in morphological taxonomic determinations.

Belk (1989) synonymized seven North American species into Eulimnadia diversa and the data here show that specimens of Eulimnadia diversa and these synonyms are highly unresolved and polyphyletic. Species-level phylogenetic resolution is low, but specimens

from the same population tend to show monophyly. These data emphasize the great need

for a total taxomomic revision of the Eulimnadia.

Introduction

The genus Eulimnadia Packard, 1874 is classified within the Limnadiidae along

with Limnadia Brongniart, 1820, Limnadopsis Spencer & Hall, 1896, Imnadia Hertzog,

1935, Metalimnadia Mattox, 1952, Paralimnadia Sars 1896, Afrolimnadia Rogers et al.,

2012, Calalimnadia Rogers et al., 2012, and Austrolimnadia Timms & Schwentner,

2012. Of these genera, Eulimnadia is the most speciose and widely distributed (Brtek

1997). This is compared to the next most speciose genus, Limnadopsis, which is made up

of 12 species (Schwentner et al. 2012). Brtek (1997) compiled a checklist of the valid

and invalid species names of the large Branchiopods, within which he listed 52 described

species of Eulimnadia, but only recognized 43 valid species (Table 2.1). Additionally,

Brtek (1997) did not mention three species that had been described prior to his checklist

(Table 2.2: Martin & Belk 1989; Roessler 1990), and since his list, an additional eight

species have been described (Table 2.2: Pereira & Garcia 2001; Durga-Prasad &

Simahachalam 2004; Simahachalam 2004, 2005: Timms & McLay 2005; Babu &

Nandan 2010; Rogers et al. 2010) yielding a total of 54 species. There is also at least one

species that has yet to be described (Table 2.2).

TABLE 2.1 - LIST OF VALID EULIMNADIA SPECIES NAMES IN BRTEK (1997) Species Locality Eulimnadia acutirostris Daday, 1913 West Africa Eulimnadia aethiopica Daday, 1913 Sudan Eulimnadia africana (Brauer, 1877) Eastern & Southern Africa Eulimnadia agassizii Packard, 1874 Masachussetts, US Eulimnadia stoningtonensis Berry, 1926 Eulimnadia alluaudi Daday, 1913 Madagascar Eulimnadia antillarum (Baird, 1852) Mexico, West Indies, Brazil Eulimnadia antlei Mackin, 1940 US

Eulimnadia astraova BeIk, 1989 US Eulimnadia texana - in Moore [1965] Eulimnadia azerbaidshanica Smirnov, 1936 Azedrbaidjan Eulimnadia behningi Smirnov, 1949 Eulimnadia belki Martin, 1989 Mexico, Venezuela Eulimnadia brasiliensis Sars, 1902 Brazil Eulimnadia braueriana Ishikawa, 1895 Japan Eulimnadia chacoensis Gurney, 1931 Paraguay Eulimnadia colombiensis Roessler, 1990 Colombia Eulimnadia compressa (Baird, 1860) India Eulimnadia chaperi (Simon, 1886) Eulimnadia curvirostris Roan, 1952 Eulimnadia cygnorum Dakin, 1914 Eulimnadia cylindrova Belk, 1989 US, Baja California, Galapagos Eulimnadia dahli Sars, 1896 Australia Eulimnadia diversa Mattox, 1937 US Eulimnadia alineata Mattox, 1953 Eulimnadia francesae Mattox, 1953 Eulimnadia inflecta Mattox, 1939 Eulimnadia oryzae Mattox, 1954 Eulimnadia thompsoni Mattox, 1939 Eulimnadia ventricosa Mattox, 1953 Eulimnadia dubia Daday, 1913 New Guinea Eulimnadia feriensis Dakin, 1914 Australia Eulimnadia garretti (Richters, 1882) Tahiti Eulimnadia geayi Daday 1913 Venezuela, Colombia, Mexico Eulimnadia gibba Sars, 1900 India Eulimnadia gunturensis Radhakrishna et Durga-Prasad. 1976 India Eulimnadia kobai Ueno, 1940 China Eulimnadia margaretae Bond, 1934 South Arabica Eulimnadia mauritiana (Guerin, 1837) Ilse de France, Maritiana Eulimnadia michaeli Nayar et Nair, 1968 India Eulimnadia minuta Daday, 1913 West Africa Eulimnadia ovata Nayar, 1965 India Eulimnadia ovata ssp. inversa Battish, 1981 India Eulimnadia packardiana Ishikawa, 1895 Japan Eulimnadia pulchra Mohammad, 1986 Eulimnadia similis Sars, 1900 India Eulimnadia sordida (King 1855) Eulimnadia rivolensis Brady, 1886 Eulimnadia subtropica Daday, 1913 Madagascar Eulimnadia taoluoensis Hu, 1986 China Eulimnadia texana Packard, 1871 US, Mexico Eulimnadia tropica Rammer, 1933 Eulimnadia victoriensis Sayce, 1903 List of valid Eulimnadia species names included in Brtek (1997) along with any known locality information obtained from the literature or from Weeks (pers. comm.). Synonymies are indented underneath the valid taxon.

TABLE 2.2 - ADDITIONAL EULIMNADIA SPECIES Species Locality Eulimnadia azisi Babu & Nandan, 2010 India Eulimnadia caraibensis n. sp. (not yet described) Carribean Eulimnadia follisimilis Pereira & Garcia, 2001 Venezuela

Eulimnadia graniticola Rogers et al. 2010 Georgia, US Eulimnadia indocylindrova Durga Prasad & Simhachalam, 2004 India Eulimnadia krishnaensis Simhachalam, 2005 India Eulimnadia magdalensis Roessler, 1990 Colombia Eulimnadia marplesi Timms & McLay 2005 New Zealand Eulimnadia ovilunata Martin & Belk, 1989 Argentina Eulimnadia ovisimilis Martin & Belk, 1989 Paraguay Eulimnadia pseudocylindrova unknown Eulimnadia radhakrishanai Simahachalam, 2004 India List of Eulimnadia species not included in Brtek (1997) along with any known locality information obtained from literature or Weeks (pers. comm.).

Interestingly, Brtek (1997) chose to synonymize Eulimnadia with Limnadia,

generally citing that within the Spinicaudata, genus-diagnostic characters can be highly

variable in different stages of ontogeny, in different sexes, and in different species within

the same genus. This argument is not a new one, as Webb & Bell (1979) also suggested

the synonymization of Eulimnadia and Limnadia due to variation in the only

distinguishing morphological character between the two genera: the presence

(Eulimnadia) or absence (Limnadia) of a ventral spine on the telson at the base of the

cercopods. Belk (1989) disputed this reasoning, suggesting it was based on a survey of

poor taxonomic drawings in the literature, and found Eulimnadia and Limnadia to be

reliably morphologically distinct.

The authors that do not support the synonymy of Eulimnadia and Limnadia have

still recognized the morphological similarities of the genera and assumed them to be

sister taxa (Straskraba 1964; Sassaman 1995). In 1964, Straskraba presented a discussion

of limnadiid generic relationships based on morphological characters. She chose to

recognize the similarities of Eulimnadia and Limnadia by including them in the

subfamily Limnadiinae Burmeister, 1843. She also recognized the distinctiveness of

Imnadia and Metalimnadia by assigning each to a monogeneric subfamily (Imnadiinae

Botnariuc & Orghidan, 1941 and Metalimnadiinae Straskraba, 1964, respectively). She

7 also grouped Limnadiopsis and Limnadiopsium Novojilov 1958 (this latter genus is no longer recognized (Timms, 2009, Schwentner et al., 2009)) together into the

Limnadiopsinae Novojilov, 1958. This taxonomic classification was based on synapomorphies established in the assessment of 10 morphological characters (Straskraba

1964).

Given the dearth of morphological characters, it is ideal to draw upon molecular analyses to test these hypotheses. Recently, Rogers et al. (2012) revised the extant genera of the Limnadiidae based on the molecular analyses of Weeks et al. (2009) and provided morphological characters that are specific to each genus. Yet species distinctions are still in need of in-depth study.

Most molecular phylogenies, to date, have focused on higher-level analyses of the families in the Branchiopoda (Braband et al. 2002; Spears & Abele 2000; Hanner &

Fugate 1997). Braband et al. (2002) is the only study from which conclusions may be drawn because they used representatives from four of the six Limnadiid genera. This analysis of 12S rDNA and elongation factor 1 (EF1) actually found Imnadia and

Limnadia formed a monophyletic group with 83% bootstrap support in the combined analysis, while Eulimnadia and Limnadiopsis formed a monophyletic group with 75% bootstrap support in the combined analysis. Although this study seemingly contradicts the conclusions based on morphological characters, the purpose of the study was to delineate higher-level relationships within the Branchiopoda and only single species were used to represent each genus.

Hoeh et al. (2006) conducted a molecular phylogenetic analysis specifically aimed at evaluating the monophyly of Eulimnadia using 28S rDNA, 12S rDNA, and

8 cytochrome b (cytb) sequences. This study found support for the monophyly of

Eulimnadia in all of the analyses. Metalimnadia was found to be the sister taxon to the

Eulimnadia clade in the 28S analysis, while Limnadiopsis was sister in the 12S, cytb, and combined analyses. However, data for Metalimnadia was missing in the 12S and cytb analysis and thus the three trees might not be contradictory. Most importantly, in all analyses run, Limnadia was never placed as a sister genus to Eulimnadia. Hoeh et al.

(2006) concluded that the placement of Limnadia and Eulimnadia into Limnadiinae, as proposed by Straskraba (1965), is based on symplesiomorphic morphological characters and not synapomorphies.

The Hoeh et al. (2006) study used nine species of Eulimnadia (represented by 36 specimens) for the 28S analysis and six species for the 12S and cytb analyses. Even with these few species, Hoeh et al. recognized three distinct sublineages: 1) E. texana + E. cylindrova, 2) E. diversa + E. magdalensis (+ E. colombiensis + E agassizzi in the 28S tree), and 3) E. braueriana. Given the global distribution of the Eulimnadia, it is likely that this speciose genus might necessitate subgeneric taxonomic groupings that could be revealed through molecular analysis.

The most recent phylogenetic analyses conducted on the clam shrimp are those of

Schwentner et al. (2009) and Weeks et al. (2009). The Schwentner et al. (2009) study analyzed 41 species with data from three genes: 28S, 16S, and cytochrome c oxidase I

(COI). Fifteen species of Eulimnadia were represented. This study follows the same conclusions reached by Hoeh et al. (2006) regarding the monophyly of Eulimnadia. They also confirm that Limnadia is not the sister taxon to Eulimnadia, and further find the

9 former to be a paraphyletic genus. Rogers et al. (2012) formally moved the Australian

Limnadia into the newly resurrected genus, Paralimnadia.

Weeks et al. (2009) included a phylogenetic analysis to address the evolutionary transitions of mating systems within the limnadiid clam shrimp. This study included 15 species of Eulimnadia and utilized 28S, EF1, and COI as well as 27 morphological characters. The trees resulting from these analyses provided strong support for the monophyly of the Limnadiidae including the genera Eulimnadia, Metalimnadia, Imnadia, and Limnadopsis. Two undescribed eulimnadoid species are included in the analysis that fall in distinct clades, the first from the island Republic of Mauritius and the second from

South Africa. In a companion paper, Rogers et al. (2012) propose these to be given the new generic names Calalimnadia and Afrolimnadia, respectively. Further, another undescribed limnadopsoid lineage from Australia was clearly found which may reveal another new genus in this family. Weeks et al. (2009) also confirm Limnadia is paraphyletic, appearing in two well-supported distinct clades, similar to those exhibited in Schwentner et al. (2009). Rogers et al. (2012) have resurrected the genus

Paralimnadia Sars, 1896 to encompass the Australian limnadioid clade. They further note the monophyly of (Eulimnadia + Metalimnadia) and the (Australian “Limnadia” +

Limnadopsis). Weeks et al. (2009) analyzed a total of 71 specimens representing the 15 species of Eulimnadia they included in the study. The analysis of multiple specimens per species highlighted problematic species determinations for specimens of E. diversa, E. follisimilis, and E. cylindrova which were not monophyletic in the phylogenetic analyses

(Weeks et al. 2009).

While Brtek (1997) did recognize 43 valid species of Eulimnadia (disregarding its inclusion in Limnadia), he also suggested that many of these would be synonymized in the future after careful examination of additional material. This underscores the taxonomic difficulties that have faced this group. Many of the early species descriptions are based on very few specimens, often from one locality. Frequently details of characters are vague or missing and drawings have been misleading (Belk 1989). An excellent example of these issues is that of E. diversa. Mattox (1954) recognized 12 North

American species of Eulimnadia. He found these to be distinct through the use of several morphological characters including: growth lines on the carapace, size and shape of the carapace, telson spines, shape of the head, male gnathopods, and antennae length (Belk

1989). However, Belk (1989) cites these characters as unreliable due to high levels of intraspecific variability. He also found that in some cases, such as that of E. alineata, the type specimens were immature. He found egg morphology to be the most important in species diagnoses and subsequently synonymized six (E. inflecta, E. thompsoni, E. ventricosa, E. francesae, E. oryzae, and E. alineata) of the North American species into

E. diversa due to similarities in egg structure, specifically, the overall shape of the egg and the pattern and shape of ridges and valleys on the egg. However, while Belk (1989) did find egg morphology to be the most useful character found to date, he recognized that the full extent of its value, as well as its limitations, needs to be further explored. In fact, as a caution, he cites Mura (1986) as showing that four distinct species of Chirocephalus

(Anostraca) showed no differences in egg morphology.

Analysis of molecular data can adequately test these hypotheses of synonymies based strictly on egg morphology and whether this is a reliable means of diagnosing

11 species. The current study presents the most comprehensive molecular analysis of the genus Eulimnadia to date with 22 species across 5 continents analyzed using a portion of the nuclear-encoded EF1 and a portion of the mitochondrion-encoded cytb. The resulting phylogenetic trees are used to evaluate sublineages within the genus

Eulimnadia. These data are further used to test the hypotheses of synonymization put forward by Belk (1989) for Eulimnadia diversa.

Materials and Methods

Study Organisms

Accessions included in this study are listed in Table 2.3. The phylogenetic analysis included all available Eulimnadia species as the ingroup. The new genus

Calalimnadia (Rogers et al. 2012) and Limnadia lenticularis (Linnaeus, 1761) served as outgroup taxa in all analyses. Limnadopsis parvispinus Henry, 1924 was included as an outgroup in the cytb analysis. Limnadopsis birchii (Baird, 1860), Imnadia yeyetta

Hertzog, 1935, Leptestheria dahalacensis (Rüppel, 1837), and Caenestheria lutraria

(Brady, 1886), were included as outgroup taxa in the EF1 analyses.

TABLE 2.3 – LIST OF SPECIMENS INCLUDED IN PHYLOGENETIC ANALYSES Species Specimen cytb ef1- Collection Locality No. Caenestheria lutraria (Brady, 1886) 592 -- EF189592 Australia Calalimnadia mahei (Rogers et al., 2012) 124 New New Republic of Mauritius 125 -- New Republic of Mauritius 126 New New Republic of Mauritius 127 New New Republic of Mauritius N47 AY779719 -- Republic of Mauritius Eulimnadia africana (Brauer, 1877) 42 New --- South Africa 59 New New South Africa 60 New --- South Africa

Eulimnadia agassizii Packard, 1874 111 New --- United States: MA 113 --- New United States: MA 74 New --- United States: MA 75 New --- United States: MA 75 New New United States: MA N59 AY779709 --- United States: MA Eulimnadia brasiliensis Sars, 1896 132 New New Brazil 133 New --- Brazil 134 New --- Brazil 135 New --- Brazil Eulimnadia braueriana Ishikawa, 1895 141 --- New Japan 142 New New Japan 143 New New Japan 146 New New Taiwan 593 --- New Japan N40 AY779724 --- Japan N41 AY779726 --- Eulimnadia colombiensis Roessler, 1990 136 New New Venezuela 137 New --- Venezuela N106 AY779715 --- Venezuela Eulimnadia cylindrova Belk, 1989 102 New New Mexico 103 New New Mexico 44 New New Mexico N103 AY779715 --- Venezuela N104 AY779716 --- Venezuela N11 AY779697 --- Mexico N16 AY779699 --- Mexico N17 AY779698 --- Mexico N64 AY779701 --- Ecuador: Galapagos Islands N65 AY779700 --- Ecuador: Galapagos Islands Eulimnadia dahli Sars, 1896 87 New --- Australia 88 New New Australia 89 New --- Australia Eulimnadia diversa Mattox, 1937 21 New New United States: IN 22 --- New United States: IN 23 --- New United States: NE 24 --- New United States: NE 37 New New United States: IN 38 New --- United States: IN 46 New New United States: NE 47 New New United States: NE 48 New New United States: NE N4 AY779720 --- United States: AZ N8 AY779721 --- United States: AZ Eulimnadia feriensis Dakin, 1914 90 New --- Australia 91 New New Australia 92 New New Australia Eulimnadia follisimilis Pereira & Garcia, 35 New New United States: NM 2001 36 New --- United States: NM 76 New New United States: NM Eulimnadia graniticola Rogers et al. 100 New New United States: GA 2010 101 New New United States: GA

31 New New United States: GA 32 --- New United States: GA 54 --- New United States: GA 72 --- New United States: GA Eulimnadia gunterensis Radhakrishna & 4 New --- India Durga-Prasad, 1976 Eulimnadia inflecta Mattox, 1939 62 New New United States: LA 63 New New United States: LA 64 New New United States: LA Eulimnadia magdalensis Roessler, 1990 N100 AY779728 --- Venezuela N107 AY779729 --- Venezuela N108 AY779730 --- Venezuela N99 AY779731 --- Venezuela Eulimnadia michaeli Nayar & Nair, 1968 68 New New Thailand 69 New New Thailand 85 New New Thailand Eulimnadia oryzae Mattox, 1954 108 --- New United States: MS 109 New New United States: MS 110 New New United States: MS 77 --- New United States: MS 78 New New United States: MS 79 --- New United States: MS Eulimnadia packardiana Ishikawa, 1894 28 New New Japan 29 --- New Japan 55 New New Japan 93 New New Japan N85 AY779726 --- Eulimnadia radhakrishanai A6 New --- India Simahachalam, 2004 Eulimnadia sp. 128 New New India 129 --- New India Eulimnadia azisi Babu & Nandan, 2010 I1 New New India Eulimnadia texana Packard, 1871 11 --- New United States: NM 12 New New United States: NM 21 New New United States: NM 5 New --- United States: NM N35 AY779712 --- United States: NM N50 AY779702 --- Mexico: Baja California N51 AY779703 --- Mexico: Baja California N5 AY779714 --- United States: NM N70 AY770707 --- United States: NM N71 AY779706 --- United States: NM N72 AY779717 --- United States: NM N73 AY779704 --- United States: NM N9 AY778708 --- United States: NM W65 AY779704 --- United States: NM Eulimnadia thompsoni Mattox, 1939 49 New New United States: IL 50 New --- United States: IL 51 --- New United States: IL 80 New --- United States: IL 81 New --- United States: IL 81 New --- United States: IL 97 New New United States: IL

Imnadia yeyetta Hertzog, 1935 289 -- AF526290 Austria Leptestheria dahalacensis (Rüppel, 291 -- AF526291 Austria 1837) Limnadia lenticularis (Linnaeus, 1761) W66 AY779732 -- United States: FL NS25 AY779733 -- United States: FL Limnadopsis birchii Baird, 1860 290 -- AF526290 Australia Limnadopsis parvispinus Henry, 1924 NS45 AY779735 -- Australia NS44 AY779734 -- Australia List of accessions included in this study. Newly obtained sequences are designated “New” and previously published sequences obtained from GenBank are noted by accession numbers. General locality information is listed.

Specimen collection, identification, and rearing

Adult clam shrimp that were collected in wet field conditions were preserved in

95% ethanol for DNA extraction. In dry field conditions, sediment was collected from dry pools and placed in plastic bags for subsequent lab hydrations. Collections were obtained by the Weeks lab team or obtained through colleagues. Those samples reared from sediment collections were kept in a temperature-controlled animal facility at the

University of Akron at 26-28C and under 24-hour light conditions (Durotest Sunlight

Simulating fluorescent bulbs). A small sample of sediment (~500 ml) was placed in the bottom of a 38 L glass aquarium and hydrated with deionized water. Each aquarium received ~10 ml of food each day. Food was composed of 2.5 g of ground Tetramin® flake food for algae-eating fish and 2.5 g of baker’s yeast suspended in 500 ml of deionized water. This protocol has been shown to be the most successful in raising a variety of species of clam shrimp (Weeks et al. 1997, 2006, 2009). When clam shrimp reached sexual maturity, they were frozen at -70C for DNA extraction. All samples were identified to species by Dr. Clay Sassaman or Mr. Christopher Rogers.

DNA Sequencing

Total DNA was extracted from individual clam shrimp using the Qiagen DNeasy

Animal Tissue Kit (Qiagen, Germantown, MD, USA). Polymerase chain reactions (PCR)

15 were used to amplify fragments of EF1 and cytb using the primer pairs presented in

Table 2.4. Each reaction consisted of 1.5 l of template DNA, 0.75 l of each primer

(125mg/ml), 18.0 l Q-master mix, and 10.5 l of H2O for a total volume of 30 l. The reactions were run in an Eppendorf personal thermocycler (Eppendorf, Hamburg,

Germany) for 40 cycles. Each cycle consisted of a 1-minute denature step at 94C, a 1- minute annealing step at 45C for cytb and 53C for EF1, and a 1.5-minute extension step at 72C.

TABLE 2.4 – PRIMER PAIRS USED FOR DNA AMPLIFICATION Gene Primer Sequence (5’-3’) Amplicon size/Citation cytb UcytB151F TGTGGAGCNACYGTWATYACTAA ~400 bp UcytB270R AANAGGAARTAYCAYTCNGGYTG Merrit et al., 1998 EF1- M44-1 GCTGACCGYGARCGTGGTATCAC ~1300 bp 3’EF1 GGAAGTCAGAGAAGGACTC Braband et al., 2002 Primer pairs used for DNA amplification in this study: mitochondrion-encoded cytochrome b: cytb and nuclear-encoded elongation factor 1-: EF1.

PCR products were cleaned with the Qiagen MinElute PCR Purification kit

(Qiagen, Germantown, MD, USA). Sequencing was carried out with an ABI 3130XL

Genetic Analyzer (Applied Biosystems, Inc., Foster City, CA, USA). Sequences were evaluated and initially aligned using Sequencher 4.7 (Genecodes Corp, Ann Arbor, MI,

USA). Subsequent editing and alignment was conducted with SeqApp (Gilbert, 1992).

These were combined with sequences acquired from the GenBank® genetic sequence database (http://www.ncbi.nlm.nih.gov/) (Benson et al. 1997). Further sequence evaluation was performed in MEGA (Molecular Evolutionary Genetics Analyis) v. 5.1

(Tamura et al. 2011).

Phylogenetic Analyses

Analyses were conducted using concatenated data sets with all genes partitioned by codon position. Maximum Likelihood analyses were performed using Randomized

Axcelerated Maximum Likelihood version 7.0.3 (RAxML) (Stamatakis, 2006). Separate analyses were performed for cytb and EF1, as well as both genes together. The General

Time Reversible (GTR) model (Taver, 1986) of nucleotide substitution with I’ model

(Yang, 1996) of rate heterogeneity was implemented for all analyses. The rapid bootstrapping algorithm was utilized with 1000 runs followed by a thorough ML search on the original alignment.

Bayesian Inference (BI) analyses were computed with MrBayes v. 3.1.2

(Huelsenbeck and Ronquist, 2003). The GTR+I+G substitution model (Ronquist et al.

1990) was implemented with two simultaneous runs in each analysis and terminated when the average standard deviation of the split frequencies reached less than 0.01. The two-gene analysis consisted of 27 million generations and resulted in 54,240 saved trees.

The cytb analysis consisted of 8 million generations and resulted in 17,014 saved trees.

The EF1 analysis consisted of 5 million generations and resulted in 100,002 saved trees.

In all analyses, the first 50% of trees were considered as the burn-in and were discarded.

Majority rule consensus trees were created incorporating posterior probabilities (PP) which are percentage values representing the frequency at which each clade was recovered in the sampled trees.

Results

Data

DNA from a total of 118 specimens representing 28 species (6 outgroup and 22 ingroup taxa) were sequenced or obtained from Genbank in this study (Table 2.3).

Ninety-six of these specimens yielded cytb sequences and 66 yielded EF1 sequences.

Both genes were successfully sequenced in 44 specimens. The cytb alignment resulted in

400 characters and the EF1 alignment contained 936 characters. Of the 400 characters sequenced from cytb, 189 were variable with 172 phylogenetically informative sites. Of

936 characters sequenced from EF1 region, 200 sites were variable with 103 being phylogenetically informative.

Combined Cytb and EF1 Analyses

Both BI and ML analyses yielded similar tree topologies for the combined analyses of 44 taxa for which both genes were obtained thus only the BI tree is presented in Figure 2.1 with both posterior probabilities and bootstrap support values indicated on each branch. Eulimnadia is confirmed to be a monophyletic group (PP=84, BS=84). The samples included in Eulimnadia form a trichotomy with E. brasiliensis and two clades of limited support. Lineage 1 (PP=66, BS=84) includes E. packardiana, E. cylindrova, E. follisimilis, E. diversa, E. texana, E. colombiensis, E. oryzae, E. thompsoni, E. inflecta, E. agasizzi, and E. graniticola. Within this lineage, E. packardiana and E. cylindrova are a monophyletic group (PP=55, BS=76). The remaining taxa also show two discrete subgroupings: 1a, PP=83 and BS=71 and 1b, PP=74 and BS=48. However individuals identified as E. diversa fall within both subgroups. In fact, within these subgroupings, individuals of E. diversa, E. texana, E. follisimilis, E. thompsoni, E. inflecta, and E. oryzae do not clade by species designation. Conversely, individual samples of E.

18 graniticola form a well-supported monophyletic group (PP=100, BS=98). Lineage 2 contains E. dahli, E. feriensis, E. braueriana, E. asizi, E. africana, E. brasiliensis, E. michaeli, and one unidentified Eulimnadia from India. Those species for which multiple samples formed monophyletic groupings show high support for the identifications (E. michaeli, PP=100, BS=100; E. braueriana, PP=100, BS=100; E. feriensis, PP=100,

BS=87).

FIGURE 2.1 – PHYLOGENETIC TREE RESULTING FROM ANALYSIS OF BOTH CYTB AND EF1

Phylogenetic tree resulting from an analysis of 44 taxa with both cytb and EF1 sequences. Calalimnadia mahei and Limnadia lenticularis comprise the outgroups. Values along branches represent Bayesian posterior probabilities (PP) first and maximum likelihood bootstrap support values (BS) second, both expressed as percentages. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

EF1 Analyses

Trees inferred from BI and ML for the 66 taxa using only EF1 sequences differed only in the placement of one taxon (E. brasiliensis), thus only the Bayesian tree with posterior probabilities (PP) and ML bootstrap support values (BS) as percentages is shown (Figure 2.2) and the one difference is discussed below. In this analysis,

Limnadopsis birchii, Imnadia yeyetta, Leptestheria dahalacensis, and Caenestheria lutraria were included as outgroup taxa, along with Calalimnadia mahei and Limnadia lenticularis. In this case, the Eulimnadia does not form a monophyletic group, but instead forms a paraphyletic clade (PP=100, BS=17) with Calalimnadia mahei and L. lenticularis. Additionally, there is discrepancy as to where E. brasiliensis is placed in the

BI and ML trees. In the MLtree, E. brasiliensis falls sister to the entire Eulimnadia +

Calalimnadia + Limnadia clade, albeit with rather weak support (BS=17), while in the BI tree E. brasiliensis is included in a clade with E. michaeli, E. asizi, E. africana, E. braueriana, the unidentified species from India, E. dahli and E. feriensis (PP=62). As expected for a single gene tree, resolution is limited and there are a large number of polytomies. In the cases where multiple individuals from a species were analyzed, only four species were found monophyletic: E. braueriana (PP=56, BS=86), E. sp. from India

(PP=100, BS=92), E. graniticola (PP=99, BS=86), and E. cylindrova (PP=100, BS=100).

The rest of the species had individuals falling in different clades, or throughout polytomies with other species.

FIGURE 2.2 – PHYLOGENETIC TREE RESULTING FROM ANALYSIS OF EF1

Phylogenetic tree resulting from an analysis of 66 taxa with EF1 sequences. Values along branches represent Bayesian posterior probabilities (PP) first and maximum likelihood bootstrap support values (BS), second, expressed as percentages. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

Cytb Analyses

The analyses of 96 taxa using only cytb sequences showed somewhat different topologies when inferred with BI (Figure 2.3) and ML (Figure 2.4). These analyses included Calalimnadia mahei, Limnadia lenticularis and Limnadopsis parvispinus as outgroup taxa. In both the BI and ML analyses, Eulimnadia is a monophyletic group

(PP=100, BS=77). These trees show much better resolution than the EF1 trees. In both trees, two specimens of E. diversa from Arizona, US (N8 and N4) are sisters to all remaining Eulimnadia lineages (PP=73, BS=39). However, there are six other specimens of E. diversa that are in much more apical positions of the tree. Both trees include a clade of 4 specimens of E. magdalensis from Venezuela and one specimen of E. colombiensis from Venezuela (PP=100, BS=94). As with E. diversa, there are other specimens of E. colombiensis that are found in different lineages of the trees. Both trees also contain a strongly supported clade of four specimens of E. brasiliensis from Brazil (PP=100,

BS=99) which forms a monophyletic group. Both the BI tree and ML tree include a clade with all E. michaeli specimens grouped together (PP=100, BS=100), sister to those is E. radhakrishnai (PP=100, BS=94), sister to (E. michaeli+ E. radhakrishnai) is a clade comprised of E. braueriana from Taiwan + an unidentified Eulimnadia from India

(PP=100, BS=99) and sister to this inclusive clade is E. gunteriensis. The BI tree shows a strongly supported clade (PP=99) including E. fereiensis, E. dahli, E. africana, E. asizi,

E. braueriana, and E. packardiana. In the ML tree, the clade including E. fereiensis, E. dahli, E. africana, E. asizi, E. braueriana, and E. packardiana is only weakly supported

(BS=28). In both trees, E. feriensis and E. dahli from Australia are monophyletic

(PP=100, BS=95), but in the BI tree, the clade of E. braueriana + E. packardiana is sister

23 to E. feriensis and E. dahli (PP=88). In the ML tree, the specimens of E. africana are sister to E. feriensis and E. dahli (BS=51). In the BI tree, E. asizi from India and E. africana from Botswana are basal members in the clade, while in the ML tree, the clade of E. braueriana and E. packardiana falls basal to the rest of the taxa.

In the ML tree, a large clade, denoted by an arrow in Figure 2.3 is shown in more detail in Figure 2.3a. This clade (BS=93) includes E. cylindrova from Mexico and

Venezuela, E. packardiana from Japan, E. colombiensis from Venezuela, E. follisimilis and E. texana from Mexico and the US, E. graniticola from the US, E. agassizi from the

US, and E. diversa and its synonyms (E. thompsoni, E. oryzae, and E. inflecta) from the

US. This clade corresponds to one found in the BI tree (PP=100), also denoted by an arrow in Figure 2.4. Figures 2.4a and 2.4b are sections of the clade found in Figure 2.4 to show detail not seen in the larger figure. The topology between the ML and BI trees for the arrow-denoted clade in Figures 2.3 and 2.4 is very different and often poorly supported. However, it is clear that in the species for which multiple samples were included, very few are monophyletically grouped according to species identification.

Only, E. graniticola (PP=64, BS=74) and E. agassizi (PP=94, BS=61) are monophyletic, while all other specimens fall in polyphyletic groups.

FIGURE 2.3 – PHYLOGENETIC TREE RESULTING FROM MAXIMUM LIKELIHOOD ANALYSIS OF CYTB

Phylogenetic tree resulting from maximum likelihood analysis of 96 taxa with cytb sequences. Values above branches maximum likelihood bootstrap support values (BS) in percentages. The clade denoted for Figure 2.3a does not include BS values for space constraints. See Fig 2.3a for BS values on that portion of the tree. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

FIGURE 2.3A – CLADE FROM PHYLOGENETIC TREE RESULTING FROM MAXIMUM LIKELIHOOD ANALYSIS OF CYTB

Clade extracted from Figure 2.3 to show detail. Maximum likelihood analysis of cytb sequences. Values above branches or directly beside branches are maximum likelihood bootstrap support values (BS) in percentages. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

FIGURE 2.4 – PHYLOGENETIC TREE RESULTING FROM BAYESIAN INFERENCE ANALYSIS OF CYTB

Phylogenetic tree resulting from Bayesian inference analysis of 96 taxa with cytb sequences. Values above branches are posterior support values in percentages (PP). Two clades denoted for Figure 2.4a and 2.4b do not include PP values for space constraints. See Fig 2.4a and 2.4b for details on those portions of the tree. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

FIGURE 2.4a – FIRST CLADE FROM PHYLOGENETIC TREE RESULTING FROM BAYESIAN INFERENCE ANALYSIS OF CYTB

Clade extracted from Figure 2.4 to show detail. Bayesian Inference analysis of cytb sequences. Values above branches are posterior probabilities values (PP) expressed as percentages. Specimens in color are those that show polyphyly (they do not group along species delimitations). Eulimnadia diversa and its synonyms are all labeled in red.

FIGURE 2.4b – SECOND CLADE FROM PHYLOGENETIC TREE RESULTING FROM BAYESIAN INFERENCE ANALYSIS OF CYTB

Discussion

The taxonomic status of the genus Eulimnadia has been thoroughly debated in the literature (Sars, 1895; Sayce, 1903; Daday, 1925; Ueno, 1927; Barnard, 1929; Brehm,

1933; Mattox, 1954; Straskraba, 1965; Webb an Bell, 1979; Belk, 1989; Martin, 1989;

Martin and Belk, 1989; Brtek, 1997; Pereira and Gardica, 2001). Recently, several molecular phylogenetic studies have validated its status (Hoeh et al. 2006; Schwentner et al.2009; Weeks et al. 2009) by showing strong monophyly of the Eulimnadia species included in the analyses. The current study adds four new species of Eulimnadia from

India (E. azisi, E. gunterensis, E. radhakrishnai, and an undescribed Eulimnadia sp.) and

122 new sequences to those already found in GenBank from previous analyses. The two gene analyses, as well as the single gene cytb analyses presented here confirm

Eulimnadia as a monophyletic group. The single gene EF1 analysis does not support

Eulimnadia monophyly as Calalimnadia mahei and Limnadia lenticularis render

Eulimnadia paraphyletic. Much of this single gene tree is ambiguous, with many polytomies, and low nodal support for many of the branches.

Two methods of analysis were used here: (a) the two gene concatenation approach, in which the gene sequences were concatenated head-to-tail to form a two gene alignment, and (b) the single gene consensus approach, where each gene was analyzed individually to look for consensus in the phylogenies. It has been shown that the concatenated approach yields more accurate trees (Sudhinra et al 2005). Both the concatenated analysis and the single gene cytb analysis support the monophyly of the

Eulimnadia, and these results are congruent with those of Hoeh et. al. (2006) and

Schwentner et al. (2009).

The alpha taxonomy of the species of Eulimnadia has been the subject of much

debate and confusion. Straskraba (1966) found most of the characters used to diagnose

species were highly variable and cast much doubt on the validity of the named species

throughout the world, with particular note of the 12 North American species recognized

by Mattox (1937, 1939, 1953a, 1953b, 1954). Belk (1989) expressed similar frustration in

attempts to identify the species of Eulimnadia in North America. Mattox (1937, 1939,

1953a, 1953b, 1954) diagnosed the North American Eulimnadia species based on growth

lines on the carapace, size and shape of the carapace, telson spination, form of the head,

the male gnathopods, and relative antennae length. Yet, several studies have found most

of these characters highly variable and not suitable for species delineation (Martin and

Belk, 1988; Sissom, 1971; Straskraba, 1966; Vidrine et al., 1987; Zaffagnini, 1971). Belk

(1989) argues that morphology of the shell of resting eggs, or cysts, is species-specific

and provides a solution to the problems of identifying the North American Eulimnadia.

This led to the synonymization of seven of the species that Mattox (1954) recognized into

Eulimnadia diversa (see Table 2.5) (Belk, 1989).

TABLE 2.5 – CURRENT TAXONOMIC STATUS OF THE NORTH AMERICAN EULIMNADIA SPECIES North American Species Included in Study Eulimnadia agassizii Packard, 1874 Yes Eulimnadia stoningtonensis Berry, 1926 Eulimnadia antlei Mackin, 1940 Eulimnadia astraova BeIk, 1989 Eulimnadia texana - in Moore [1965] Eulimnadia belki Martin, 1989 Eulimnadia cylindrova Belk, 1989 Yes Eulimnadia diversa Mattox, 1937 Yes Eulimnadia alineata Mattox, 1953 Eulimnadia francesae Mattox, 1953 Eulimnadia inflecta Mattox, 1939 Yes

Eulimnadia oryzae Mattox, 1954 Yes Eulimnadia thompsoni Mattox, 1939 Yes Eulimnadia ventricosa Mattox, 1953 Eulimnadia graniticola Rogers et al. 2010 Yes Eulimnadia texana Packard, 1871 Yes The currently recognized taxonomic status of the North American species of Eulimnadia. Note the synonyms of E. diversa proposed by Belk (1989). Synonymies are indented underneath the valid taxon. Those included in the present study are noted.

Results from a molecular phylogenetic analysis allow for independent testing of

some of the taxonomic conclusions based on morphological observations. Currently,

eight species of Eulimnadia are recognized from North America (Table 2.5). This study

includes five of these, as well as three of the species that Belk (1989) synonymized into

E. diversa (Table 2.5). The specimens representing these species were identified by

Christopher Rogers and Dr. Clay Sassaman, both leading clam shrimp taxonomists, based

on the original type characters used to diagnose the species prior to Belk’s (1989)

synonymization based on egg characters. Multiple specimens were included in each

analysis to test for monophyly at the species level. Only one North American species

demonstrates monophyly of all specimens used: E. graniticola. Eulimnadia cylindrova

from Mexico is monophyletic in the combined gene analysis, but when more specimens

are added in the cytb and EF1 single gene analyses, the species becomes polyphyletic.

Yet, it is interesting to note that specimens are monophyletic with respect to populations

(Mexico, Ecuador, and Venezuela). Eulimnadia agassizi is monophyletic in the combined

gene analyses and the cytb analyses, but is polyphyletic in the EF1 analyses. In all

analyses, E. texana specimens are polyphyletic, dispersed in the tree with specimens of E.

diversa, E. follisimilis, and E. colombiensis. Eulimnadia diversa specimens, and those

identified as its synonyms, are also not monophyletic.

These results highlight the troubles in identification of the Eulimnadia species. In this study, we could assume that all specimens were correctly identified based on the diagnostic characters of the species descriptions. If this is the case, then the molecular analyses would suggest that most of the current species descriptions are not diagnostic.

Alternatively, we could assume that some of the specimens were incorrectly identified and thus led to such polyphyly as seen in the molecular analyses. This is certainly possible given the variability in morphological characters. Under either scenario, it is suggested that the characters used to identify species, including those of the egg, or cyst, as proposed by Belk (1989) are not reliable.

Despite the uncertainty of the species designations, given the polyphyly of the three Mattox (1954) species that Belk (1989) synonymized within E. diversa, we can conclude that Belk was correct in his assessment that these are not valid species.

However, in all analyses E. diversa is also polyphyletic. In the two-gene analysis, three

E. diversa specimens from a population in Nebraska do form a monophyletic group

(PP=78, BS=57), but two specimens from a population in Indiana are found in very different parts of the tree. In the EF1 analysis, there is very little resolution for E. diversa and its synonyms. This is also the case in the cytb analysis, with the exception of two specimens identified as E. diversa from a population in Arizona, which actually fall basal to all of the other Eulimnadia taxa.

This study includes the largest number of E. diversa specimens (and its synonyms) sampled from a wide range throughout North America. There is not a lot of resolution at this level with the genes included and it might be necessary to seek out genes better suited to species and population level phylogenetics to further test the

33 validity of E. diversa as a species including its synonyms, or whether populations of E. diversa might be distinct species.

Eulimnadia texana has been extensively studied as a model for the androdioecious mating system. Belk (1989) found the eggs of E. texana to be morphologically conservative between two wide-ranging populations, one in Texas and one in California. In this study, we were only able to obtain both genes for two specimens and in the two-gene tree, these are not monophyletic, falling in a clade with E. follisimilis, E. colombiensis, and the E. diversa specimens from Nebraska. In the EF1 tree, three specimens are included, but the tree is poorly resolved at this level. The cytb trees offer the most information on this species as we were able to include 13 specimens,

11 of which are from populations in New Mexico and two from a population in Mexico

(Baja of California). In both the BI and ML analyses, these specimens form a polyphyletic group with specimens of E. follisimilis (also from New Mexico), E. colombiensis from Venezuela, and the E. diversa specimens from Nebraska. Also, in both analyses, the specimens from the population in Mexico are significantly grouped into a clade together. It is also interesting to note that the two specimens of E. colombiensis and the two specimens of E. follisimilis do represent monophyletic groups within the E. texana polytomy. Yet, these specimens are from single populations, thus it seems specimens from populations form monophyletic groups, while specimens from different populations identified as the same species do not. This highlights that it is the species diagnosis that is in question.

Eulimnadia agassizii is a North American species with a very small range in

Massachusetts, US (Packard, 1874; Zinn and Dexter 1962). In the two-gene analysis and

34 in the single cytb analysis, E. agassizii specimens are monophyletic (PP=65, BS=46 and

PP=94, BS=61, respectively), while in the EF1 analysis, there is very little resolution and the specimens fall in a polytomy with several other species. Still, given the monophyly shown in the two-gene and cytb analyses, this data supports this as a valid species.

Eulimnadia cylindrova is a widely distributed species, occurring throughout North and South America. Here, we have included specimens from Mexico, Venezuela, and

Ecuador (the Galapagos Islands). Although all of these specimens do not form a monophyletic group, some do occur in monophyletic groups at the population level.

Brendonck et al. (1990) conducted a large survey of the Branchiopod fauna of the

Galapagos Islands, including E. cylindrova. In this study, Brendonck et al. (1990) noted that there were slight differences in the morphology of the eggs from those described by

Belk (1989), but not enough to warrant a new species designation. However, these differences led the authors to observe egg morphology across populations of E. cylindrova from one population in Arizona, US, two populations in Mexico, as well as those populations in the Galapagos Islands. Brendonck et al. (1990) noted variability in morphology and size characteristics of the eggs from these different populations. This is in contrast to the suggestions by Belk (1989) that egg morphology is a conservative character. They concluded that further study is needed to clarify the taxonomic level at which these populations should be considered and resort to classifying all populations into an extended species group: E. cylindrova sensu lato. The molecular data in this study, particularly the cytb analysis (which included three different populations of E.

35 cylindrova) would actually support a conclusion that these populations are quite different and might warrant species-level distinction.

Given the strong monophyly of E. graniticola in all the analyses, we can be very confident in this species. Interestingly, this is the most recently described Eulimnadia species from North America (Rogers et al. 2010) and the designation of it as a new species is based primarily on egg morphology. It is noted that E. graniticola eggs are most similar to E. follisimilis eggs and these two species might form a species group as described in Brendonck et al. (1990). However, Rogers et al. used molecular analysis as support for E. graniticola as a unique species: they found three apomorphies in the sequences of 28S of four specimens. The molecular data presented here further support the hypothesis that E. graniticola is a distinct species.

Overall, from these data, it is evident that Belk (1989) is correct in his assessment that the North American species designated by Mattox (1954) are most likely not valid.

However, it is unclear as to whether they should all be synonymized into E. diversa.

Eulimnadia diversa, E. texana and E. cylindrova are all wide-ranging species and seem to form species groups with other North and South American species. Eulimnadia graniticola and E. agassizii are narrowly distributed and they are well supported as valid species. In fact, in this study, the specimens for these species were from single populations, while the other American species had multiple populations represented. In general, specimens tended to be monophyletic relative to populations rather than species diagnosis. While Belk (1989) found egg morphology to be consistent across populations,

Brendonck (1990) showed it to be variable and it is unclear if these observations represent intraspecific or interspecific comparisons. This molecular study highlights the

36 need to further investigate inter-population variability within species to better understand what constitutes a species in the Eulimnadia.

Only two other molecular studies included multiple specimens of Eulimnadia species (Hoeh et al. 2006, Weeks et al. 2009). Hoeh et al. (2006) included multiple specimens of E. texana, E. cylindrova, E. texana, E. magdalensis, E. diversa, E. agassizii,

E. braueriana and an unidentified Eulimnadia from Japan. In this study, Hoeh et. al.

(2006) actually found well supported monophyly at the species level in the concatenated mitochondrial (12S and cytb) analysis, as well as the concatenated mitochondrial + nuclear (12S + cytb + 28S) analyses. The single 28S analysis showed little resolution between E. cylindrova and E. texana specimens, but still, the rest of the specimens of

Eulimnadia were monophyletic along species lines. Alternatively, Weeks et al. (2009) included a greater number of specimens representing E. africana, E. agassizi, E. brasiliensis, E. cylindrova, E. dahli, E. diversa, E. follisimilis, E. magdalensis, E. michaeli, E. texana, and 3 unidentified Eulimnadia species. In this study, a concatenated analysis of 28S + COI + EF1 showed very similar polyphyly of specimens as those found in the phylogenies presented herein, particularly in those specimens identified as E. diversa, E. agassizi, E. cylindrova, and E. follisimilis. While not the focus of their paper,

Weeks et al. (2009) did note that these results were most likely due to the problems of species determination, which could have also been a problem in the current study.

Rabet (2010) most recently stressed that egg morphology is the most important character in Eulimnadia systematics. In a review of the samples stored in The National

Museum of Natural History (MNHN) of Paris, Rabet (2010) confirmed egg morphology as a reliable and consistent systematic character, but he noted several points in using this

37 diagnostic. First, scanning electron microscopy (SEM) technology is a must in identification because egg shape can be misinterpreted at different angles in light microscopy. Second, Rabet (2010) found egg contamination to be a frequent problem.

Mixture of eggs was common in the MNHN collections, which could make species determination impossible. Further, Rabet (2010) noted that cohabitation of two

Eulimnadia species is frequent, particularly in the Neotropical region, and can lead to egg contaminated collections and confusion in identification. Lastly, Rabet (2010) found that some species do have extremely similar egg morphology (E. texana with E. ovissimilis and E. colombiensis with E. belki); thus diagnosis must be associated with other morphological characters. Overall, Rabet (2010) suggested an entire revision of the genus with SEM observation and standard terminology for the eggs.

For the current study, identifications were not confirmed using SEM of the eggs of the specimens. We also made the assumption that single populations represented single species. Based on this assumption, collections were split with some specimens reserved for DNA analysis and others sent for identification. This could certainly lead to identification problems if more than one species were present in a population. Obviously, those specimens that were identified as species that were synonyms of E. diversa had to be diagnosed based on adult morphological characters that have been deemed variable and unreliable. Molecular phylogenetic studies of this kind can be useful in understanding species relationships within such a diverse, speciose, and cosmopolitan genus like Eulimnadia. However, it might be best to postpone such studies until a comprehensive taxonomic revision is conducted so that the species descriptions and the specimens used can be reliably identified. Still, the combination of molecular and

38 morphological data can lend even further support to species hypotheses (as in the case of

E. graniticola) and it can highlight interpopulation diversity within a species (as in the case of E. cylindrova).

Acknowledgements

Many thanks are extended to Clay Sassaman and Christopher Rogers for sample identification. I am very grateful to Joel R. Duff for laboratory space, much help with

DNA sequencing, and extensive help in early phylogenetic analyses and data interpretation. I thank Walter R. Hoeh for running the final phylogenetic analyses.

Francisco Moore was kind enough to review an early draft of this chapter. This project was supported by the National Science Foundation [grant number DEB-0235301] and the

National Science Foundation Doctoral Dissertation Improvement Grant [DEB-0709618].

III. A BIOGEOGRAPHIC ANALYSIS OF EULIMNADIA WITH AN

ASSESSMENT OF THE EVOLUTION OF ANDRODIOECY

Abstract

The clam shrimp genus Eulimnadia is the most speciose and widely distributed genus in the Spinicaudata, with species occurring on every continent except Antarctica.

The Eulimnadia life history consists of an encysted egg stage where dispersal is likely to occur via wind or animal vectors across relatively short distances. Dispersal across long distances and ocean barriers has been shown to be unlikely and it has been argued that the present day distributions of extant taxa is the result of ancient dispersals when the continental positions held landmasses contiguous, up to 180 million years ago. This study uses Statistical Dispersal-Vicariance Analysis (S-DIVA) to reconstruct the biogeographic history of the Eulimnadia to test this hypothesis. The Eulimnadia have a New World origin and I inferred nine dispersal events in the history of the Eulimnadia species studied, of which five could support the ancient dispersal hypothesis and four have more recent dispersal explanations.

Introduction

The present day distribution of a species can be attributed to its origin and evolution as well as its ancestor’s distribution, its ability to disperse and establish in a given environment, and the evolution of the environment over time (Crisci 2001).

Historical biogeographic reconstructions based on phylogenetics are valuable means to elucidate the evolutionary history of particular taxa (Crisci 2001). Specifically, the method of dispersal vicariance analysis (DIVA; Ronquist 1997, 2000) is a popular and

40 simple tool for inferring ancestral areas. Herein, the phylogenetic data produced in

Chapter II are used to construct a biogeographic history for the clam shrimp genus

Eulimnadia.

The Eulimnadia is the largest and most widely distributed clam shrimp genus and is particularly interesting because most, if not all, of the species in this genus are androdioecious. Weeks et al. (2008) showed that 14 species of Eulimnadia are functionally androdioecious and the most parsimonious conclusion is that the ancestor to the Eulimnadia was also androdioecious. This contradicts the proposed transitory nature of androdioecy (Lloyd 1975; Charlesworth 1984). For this reason it is important to understand when and where Eulimnadia evolved and how they became distributed across the globe to understand the origins and persistence of this rare mating system.

The fossil record of clam shrimp has been studied extensively by Tasch (1987), and these fossils date back to the Devonian (408-360 mya). However, it is unclear as to how the fossil taxa relate to Eulimnadia. Tasch and Shaffer (1964) suggest Eulimnadia arose in the Mesozoic era (~65 mya). Chen and Shen (1981) propose the extinct

Yunmenglimnadia from the Paleogene period (66-24 mya) as a possible ancestor to the

Eulimnadia. Weeks et al. (2006) infer that the genus Eulimnadia is up to 180 million years old based on the global distribution of the present day species. They argued that the dispersal capabilities of the genus are restricted to overland and therefore that the dispersal of Eulimnadia had to have occurred during a time when the continents were contiguous.

Tasch (1987) argued wind dispersal as the most likely mechanism for clam shrimp dispersal. Tasch (1987) proposed that clam shrimp cysts are windblown along

41 with the dried substrate of the pond in which it resides. He envisioned this occurring across short distances in a “stepping stone” method from one shallow pond to another such that large distances could be covered overland, but it would not allow for dispersal across oceans. There is evidence from several studies that newly formed pools can be quickly colonized by dispersing cysts (Brendonck 1996; Vanderkerhov et al. 2005), validating Tasch’s assumptions of short distance wind dispersal.

Another mechanism proposed for dispersal of clam shrimp cysts is transport by birds. Darwin (1859) noted the ability of migratory waterfowl to disperse aquatic plants and invertebrates across long distances. This could occur externally (epizoochory) by adhering to the bird, or internally (endozoochory) by passage through the digestive tract.

Multiple studies have shown this to be a viable dispersal mechanism for many aquatic invertebrates, including Branchiopods (reviewed in Charalabidou & Santamaria 2002;

Figuerola and Green 2002; Green and Figuerola 2002); however it has not been investigated specifically for any clam shrimp. Furthermore, Tasch (1987) argued that this is unlikely for clam shrimp because no clam shrimp have ever been found in Antarctic lakes where both cladocerans and anostracans are thought to have been transported via birds. Tasch (1987) additionally points out that there is no evidence for circumpolar birds prior to the Cretaceous (~65 mya) and thus if Eulimnadia was globally distributed prior to this time, it could not have occurred via bird dispersal. The generated molecular phylogeny from Chapter II and the area cladograms and biogeographic analysis of this chapter can help to elucidate how Eulimnadia species dispersed throughout the globe.

This analysis provides a test of the hypothesis put forth by Weeks et al. (2006) that the present day distributions of the Eulimnadia species is a result of dispersal prior to the

42 break-up of the continental landmasses, and has direct bearing on our understanding of the persistence of androdioecy in the genus.

Materials and Methods

Dispersal-vicariance analysis (DIVA) (Ronquist, 1996, 1997, 2001) is a widely used method for inferring the distributional history of individual groups (taxon biogeography). This method is desirable because it provides rapid results and requires little a priori knowledge or assumptions (Ronquist, 1997). Ronquist (1997) puts forth specific rules for this analysis where there are two modes of allopatric speciation: (1) the ancestor is distributed in a single unit area and after a speciation event both descendant lineages are found in the same area or (2) the ancestor occurs in more than one unit area and speciation results in the division of the ancestral distribution into two exclusive unit areas. Allopatric speciation associated with geographical vicariance is accepted as the null model in historical biogeography. Thus Ronquist (1997) uses character optimization techniques which minimize dispersal and extinction events in the reconstruction of ancestral distributions (Ronquist, 1997). Statistical-DIVA (S-DIVA; Yu et al. 2010) is a program that uses non-parametric, empirical Bayesian analysis to handle phylogenetic uncertainty and uncertainty in DIVA optimization, as discussed in Nylander et al. (2008) and Harris and Xiang (2009), where ancestral area reconstructions are averaged across all trees and are weighted by the probability that each tree is correct. The sample of trees from a Bayesian analysis is sampled in direct proportion to their posterior probabilities, and the frequencies derived are treated as probabilities (Nylander et al. 2008). The probability (p) of ancestral range (x) at node (n) on the final tree is calculated as:

m p(xn) = Σ t=1 F(xn)t * pn

Here, t is the selected tree and m is the total number of trees sampled. F(xn)t is the occurrence of an ancestral range x at node n for tree t, and pn is the support for that node.

Thus the probability of the ancestral reconstructions given in the final tree is dependent on the frequency of occurrence of the node and reveals the underlying phylogenetic uncertainty. The optimal reconstruction of the final tree is that with the highest S-DIVA value (SV). The program requires a set of trees from a phylogenetic analysis, a final representative tree, and range information for the terminal taxa. The Reconstruct

Ancestral State in Phylogenies (RASP) is the 2.0 version of S-DIVA (Yu et al. 2011).

The phylogenetic tree files from the analyses of both genes generated in Chapter

II with MrBayes v. 3.1.2 (Huelsenbeck and Ronquist, 2003) were used for the biogeographic analyses in RASP. RASP requires a fully resolved tree, thus final trees were generated with the “Condense” option in the RASP program and polytomies were arbitrarily resolved. RASP requires a single defined outgroup. The most basal taxon in the phylogenetic trees in Chapter II was used here (124CalspRM). Areas of distribution were broadly defined using the continental plates on which each species is located: A –

North American Plate, B – South American Plate, C – African Plate, D – Eurasian Plate,

E – Indian Plate, and F – Australian Plate (Figure 3.1). Root distribution was set to

“Wide,” where the ancestors of the tree were distributed in all the ranges defined.

Optimization of ancestral reconstruction was unconstrained as to the number of unit areas for each ancestral node. Bayesian Markov Chain Monte Carlo analysis was run under default conditions in RASP for the 54,240 trees obtained from the phylogenetic analysis in Chapter II.

FIGURE 3.1 – CONTINENTAL PLATES USED FOR DISTRIBUTIONAL AREAS

Map outlining the continental plates that were used to define distributional areas in the RASP analyses. Courtesy of the U.S. Geological Survey.

Results

The biogeographical reconstruction by S-DIVA analysis suggests both vicariance and dispersal were important factors in shaping the present-day distributions of the species examined in this study. Figure 3.2 is the final phylogenetic tree with ancestral areas reconstructed in RASP represented in pie charts at each hypothetical ancestral node.

The XY designations represent inferred dispersal events from area X to area Y. The nodes are those points in the phylogenetic tree representing ancestors, and each node is assigned a pie chart illustrating the probabilities of the ancestral distributional areas at each node. The areas are assigned letters so that the distribution of those ancestors can be discussed. The legend in each figure shows which area corresponds to which letter. An asterisk refers to undefined ancestral areas. If one node (ancestor) is inferred to be in a particular area, and the following nodes branching from that ancestor are the descendants

45 that are placed in a different area, then a dispersal event is inferred between those two nodes (area 1  area 2). I also distinguish nodes as being terminal in the tree or internal in the tree. Here, I define terminal nodes as those representing extant species on the tree and in some cases a node representing a clade of multiple specimens of a single extant species in those cases where all multiple specimens of a single species are represented.

Thus, the terminal nodes in the Eulimnadia clade are nodes 1, 2, 4, 6, 11, 12, 16, 18, 20,

26, 27, 34, and 35.

Nine dispersal events can be inferred from this analysis (Figure 3.2), with five of these occurring between internal nodes (node 38 Anode 10 E; node 10 Enode 9 C; node 9 Cnode 8 E; node 8 E node 7 F; and node 39 A 132 E. brasiliensis B) and four of these ending in terminal nodes/taxa (node 3 E node 2 D; node 7 F node 6 D; node 36 A node 34 D; and node 30 A 136 E. colombiensis B). Node 39 represents the ancestor to the Eulimnadia clade, thus while 132 E. brasiliensis is the basal taxa branching from this node and represents a terminal taxa, I define this as an internal dispersal event.

Six of the internal ancestral node reconstructions in Eulimnadia are ambiguous. S-

DIVA indicates the ancestor to the Eulimnadia (node 39) had a New World (North

America 59.5%, South America 36.2%, * 4.4%) distribution, with a high probability of occurrence in North America. Node 39 occurred in 81% of the trees sampled. The most basal taxon in this analysis, branching from node 39, is E. brasiliensis, known from

Brazil; thus this is the first inferred dispersal from North America to South America

(AB). Node 10 indicates a dispersal event from North America to India (AE), where the India is favored at 54.8% while other probabilities include D: Eurasia 18.7%, C:

Africa 15.6%, A: North America 8.7% and undefined areas (*) at 2.2%. This node is only found at a frequency of 51% in the trees sampled. Within this clade, it is difficult to discern the sequence of dispersal events given the ambiguity of the ancestral area reconstructions and the low support for the internal nodes 9, 8, and 7. Node 9 indicates a dispersal from India to Africa (EC) (Africa 64.3%, India 28.0%, * 7.7%) followed by another dispersal at node 8 from Africa back to India (CE) (India 67.7%, Eurasia

14.6%, Australia 10.0%, Africa 6.5%, * 1.2%). Node 8 is found at a frequency of 53%.

The favored ancestral range for node 7 would indicate another dispersal from India to

Australia (EF) (Australia 48.3%, Eurasia 43.8%, India 6.0%, * 2.0%) and this node is found in 100% of the trees sampled. One more dispersal event in this clade is indicated at node 6 where the ancestral reconstruction favors dispersal from Australia to Eurasia

(FD) at 98.7% in 100% of the trees sampled at this terminal. Node 2 is also a terminal and shows a dispersal event from India to Eurasia (CD) and is favored 100% of the time.

Nodes 11 through 33 all favor ancestral reconstructions in North America indicating vicariant or duplication events, although it is important to note that not all of these nodes show speciation because in some cases, multiple specimens of a given species were studied. In fact, this clade is fraught with problems where some specimens of particular species are polytypic (see Chapter II for a full discussion). All the species in this clade and the ancestral reconstructions are distributed in North America, excepting one specimen of E. colombiensis found in Venezuela, which would indicate a dispersal event from North America to South America after node 30 (AB).

The ancestral reconstruction of node 36 favors an ancestral distribution in North

America and is found at a frequency of 56%. The terminal node 34 is found at a frequency of 90% and is distributed in Eurasia, indicating the dispersal from North

America to Eurasia (A D).

In this analysis, seven of the nine dispersal events noted would have been across oceans given present continental positions (node 2 ED, node 6 FD, node 7 EF, node 8 CE, node 9 EC, node 10 AE, and node 34 AD), and four of these were at internal ancestral nodes (node 2 ED, node 6 FD, node 7 EF, node 8 CE, node 9

EC, node 10 AE).

Inferred dispersal events were superimposed on the maps presented in Weeks et al. (2006) to indicate whether these dispersals were across oceans in the present day configuration (Figure 3.3a), 60 mya continental configuration (Figure 3.3b) and 180 mya continental configuration (Figure 3.3c).

FIGURE 3.2 – ANCESTRAL AREA RECONSTRUCTION FOR EULIMNADIA BASED ON BOTH EF1 AND CYTB BAYESIAN ANALYSIS

Ancestral area scenarios for Eulimnadia reconstructed with Reconstruct Ancestral State in Phylogenies (RASP). The phylogeny is a Bayesian tree with ambiguities arbitrarily resolved. Node frequencies are shown next to nodes. Nodes are numbered in white. Pie charts at internal nodes represent the marginal probabilities for each alternative ancestral area derived in RASP. Arrows indicate inferred dispersal events (XY). Color key legend shows possible ancestral ranges at different nodes with black asterisk representing undefined ancestral areas.

FIGURE 3.3 – PRESENT DAY SPECIES DISTRIBUTIONS AND BIOGEOGRAPHIC DISPERSAL EVENTS

Maps are adapted from Weeks et. al. (2006). Known Eulimnadia distributions are marked with circles and squares on the maps showing continental positions at present (A), approximately 60 mya (B), and approximately 180 mya when the major continents were contiguous (C). Circles refer to those species studied in Weeks et al. (2006) and squares are from other known localities from the literature included in Weeks et al. (2006)(see Chapter II, Table 2.1 and Table 2.2). The maps show the outlines of continental boundaries, shelf margins, major tectonic boundaries, active plate boundaries and seafloor isochrones (Scotese 2001). Colors and letters correspond to the key in Figure 3.1 which are the assigned species areas for the RASP analysis. Arrows indicate inferred dispersal events based on RASP analysis. Solid lines are those dispersals referred to as internal, while dotted lines are those dispersals referred to as terminal.

Discussion

The clam shrimp genus Eulimnadia is the most speciose and widely distributed clam shrimp genus (see chapter II), with species found across the globe on every continent except Antarctica. Yet, its means of dispersal are thought limited to land-based mechanisms. This genus is of much interest because it represents the most widespread example of androdioecy with at least 16 species shown to be functionally androdioecious

(Weeks et al. 2008, Weeks 2012). Weeks et al. (2008) concluded that the ancestor to the

Eulimnadia was most likely androdioecious despite most mathematical and ecological models predicting that androdioecy cannot be a successful long-term evolutionary strategy (Charlesworth 1984, Lloyd 1975, Pannell 2002). Weeks et al. (2006) inferred that androdioecy is indeed exceptionally long lived in the Eulimnadia because it is the ancestral state in this group and thus rejected the prediction from these theoretical models of mating system evolution that androdioecy must be short-lived.

The argument put forth by Weeks et al. (2006) hinges on the dispersal capabilities of clam shrimp. The Eulimnadia life history consists of the aquatic adult stage in a temporary pool and an encysted egg stage in the dry environment. Thus dispersal might occur in the adult stage via water pathways connecting temporary pools or in the encysted stage via wind or animal vectors. Short-distance dispersal via wind has been shown for various branchiopod cysts (Brendonck 1996; Vanderkerhov et al. 2005). Short-distance dispersal has also been documented for branchiopod cysts via birds (reviewed in

Charalabidou & Santamaira 2002), mud-wallowing mammals such as elephants, warthogs, rhinoceroses, and buffalo (Vanshoenwinkel et al. 2011), and even humans

(Waterkeyn 2010). However, these short-distance dispersal mechanisms do not explain the wide distribution seen in the Eulimnadia. Birds migrating over long-distances might be able to explain these wide distributions, or even human transport across oceans.

However, if the Eulimnadia are as old as the fossil record indicates (up to 65 mya, see intro: Chen & Shen 1984; Tasch and Shaffer 1964), these might not be viable explanations for their historical distribution. Tasch (1987) reasons that circumpolar birds were not known prior to the Cretaceous (~65 mya) when Eulimnadia was likely already distributed globally. Also, human global transport would be a relatively recent event.

Tasch (1987) proposed short distance dispersal via wind would lead to broader distributions in a “stepping stone” fashion.

Weeks et al. (2006) suggest that due to the limited dispersal capabilities of these animals, the biogeographical history of the Eulimnadia ought to exhibit dispersal during a time when the continents were contiguous. The continental positions at 65 million years ago, when some suggest the fossil record indicates that Eulimnadia arose (Chen & Shen

1984; Tasch and Shaffer 1964) still requires dispersal across oceans (Figure 3.3b).

However the continental positions of ~180 million years ago (Figure 3.3c) are contiguous and might help date Eulimnadia to this time period.

If the ancestor to the Eulimnadia dates back to 180 million years ago when the continents were contiguous and the wide distribution pattern of the present day extant taxa is to be explained by short-distance wind dispersal, I would expect to find a biogeographical history where speciation occurs via vicariance and short-distance dispersal. I would expect major dispersal events to occur in internal nodes representing past hypothetical ancestors in the phylogenetic and biogeographic reconstructions and I

52 would also expect that dispersal should occur in areas adjacent to each other in the contiguous land mass (Figure 3.3c). Alternatively, dispersal inferred at terminal nodes

(nodes representing extant species) and/or into areas that are disjunct could indicate more recent expansion and require different dispersal mechanisms, such as bird migration and/or human transport. The results of this analysis suggest a combination of both of these scenarios.

The ancestral area reconstructions in the deeper nodes within the Eulimnadia clade are all ambiguous and none of these nodes are found 100% of the time. However,

Ronquist (1996) does point out that ancestral reconstructions do become less reliable closer to the root node. Yet, if the most probable ancestral area reconstructions are considered, several interesting internal dispersal events are seen. Node 39 represents the ancestor to the eulimnadian clade and the ancestral reconstructions support a New World origin, specifically in North America. Node 39 branches into the most basal taxon, E. brasiliensis from South America, and node 38. Node 38 branches into two clades, one that remains in North America and another at Node 10 whose members are found in

India, Africa, Australia, and Eurasia (Thailand and Japan). The internal nodes 10, 9, and

8 are found at frequencies just around 50%; thus it is difficult to confidently say which dispersal occurred first, but from node 38 to this clade, the Eulimnadia likely dispersed into India, followed by dispersals into Africa and Australia. If these dispersals occurred given present day (Figure 3.3a) and 60 mya (Figure 3.3b) continental configurations, they would have to have occurred across ocean barriers. However, at 180 mya (Figure 3.3c) these areas were contiguous and land-based dispersal mechanisms would be enough to

53 explain the present day Eulimnadia species distribution throughout India, Africa, and

Australia. This supports the analysis presented by Weeks et al. (2006).

Within the node 10 clade, there are two more dispersal events closer to the terminals. The first is from Australia to Eurasia (at node 6). Even at 180 mya (Figure

3.3c) this dispersal would likely have occurred across an ocean. We do not have a molecular clock for dating these phylogenetic events, but given this dispersal is at the terminal, it is certainly more recent than those at internal nodes. Thus, the occurrence of

Eulimnadia braueriana in Japan requires an explanation of dispersal either of this species or its ancestor across a long, oceanic barrier. Eulimnadia sp. in Japan are found in rice paddies along with other large Branchiopods, the Notostracans and Anostracans (Grygier et. al. 2002). The Notostracans, or tadpole shrimp (Triops sp.) have actually been reported as an important biological control agent of weed in rice paddies (Takahashi,

1977) and human introduction of the North American tadpole shrimp, Triops longicaudatus, has been found (Hamasaki, 1999). Given the similar life histories of the notostracans and clam shrimp, it seems likely that human introduction of clam shrimp could have easily occurred concomitantly with tadpole shrimp.

The second terminal dispersal event in this clade is that from India to Eurasia at node 2, which represents E. michaeli from Thailand. Being at the terminal, this is likely a more recent dispersal event, and if this dispersal occurred after India collided with Asia, land-based dispersal methods would also be sufficient in achieving this distribution.

Another clade showing important dispersal events branches off of node 37, which has an ancestral area reconstruction in North America, to node 36 where the clade members are found in Eurasia (Eulimnadia packardiana from Japan) and North America

(Eulimnadia cylindrova from Mexico). Thus, another dispersal into Japan across an oceanic barrier is found at a terminal node, which, again, could be explained as due to human movement of branchiopods that are commonly used in rice paddies, as noted above.

The last terminal inferred dispersal event is that of Eulimnadia colombiensis found at node 30. This event is represented by one specimen whose locality is in

Venezuela. This specimen falls in a large clade of species from the United States at node

33. Recent dispersal from North America into South America could have been achieved via land-based dispersal methods.

Despite the ambiguity of many of the ancestral reconstructions of this analysis, the results presented do support the thesis presented by Weeks et al. (2006) that the majority of current-day distributions of clam shrimp in the genus Eulimnadia can be explained by dispersal at a time when the continents were contiguous. Thus, the inference that androdioecy is an exceptionally long-lived reproductive mode in these crustaceans is supported by the current biogeographic analyses. Not all current-day species can be explained in this way; there are dispersal events that had to have occurred across oceans that are likely more recent, when human and/or bird dispersal could have dispersed

Eulimnadia cysts over oceans (much more likely the former), still the majority of dispersal events are in line with ancient dispersal events when the continents were contiguous.

This study is most certainly the first step in investigating how the species of

Eulimnadia came to such a cosmopolitan distribution given their limited dispersal capabilities. The development of a molecular clock is much needed to narrow down the

55 timing of the dispersals inferred here. Further, more robust taxon sampling throughout the areas not represented in this analysis will be essential to discerning how these animals dispersed across the continents via land-based mechanisms. With more species represented, and a broader understanding of the extent of the distributions of each species, we could expect a better supported phylogeny to access the ancestral area reconstructions.

Acknowledgements

I am grateful to Raymond Reed for extensive help in how to edit figures and illustrations, as well as providing a computer and software for running these analyses and developing illustrations. Thanks also goes to Francisco Moore for reading an earlier draft of this chapter.

IV. AN INVESTIGATION OF THE MAINTENANCE OF ANDRODIOECY VIA

METAPOPULATION DYNAMICS IN EULIMNADIA DAHLI

Abstract

Androdioecy is a most rare mating system where males and self-compatible hermaphrodites co-occur in populations with no true females. Clam shrimp in the genus

Eulimnadia represent the most widespread occurrence of androdioecy in the animal kingdom. Herein, we test Pannell’s (2000; 2002) ecological model whereby metapopulation dynamics of extinction and recolonization allow for long-term persistence of androdioecy when traditional models suggest it is not a stable system. This model requires that the populations most recently colonized should contain only hermaphrodites (monoecious) and males invade secondarily as population numbers increase. Thus it is expected that monoecious populations will be of a lower density, and show lower within-population diversity and higher genetic differentiation than androdioecious populations. Eulimnadia dahli was found occurring in isolated metapopulations (many distinct temporary pools) atop nine granite outcrops sampled

throughout Western Australia. Gene flow estimates (based on FST) were found to correlate with geographic distance between outcrops, but not with the geographic distance of pools on the outcrops (no isolation by distance at the pool within outcrop level) suggesting differentiation can be attributed to the processes of extinction and recolonization, the essential components in Pannell’s (2000; 2002) model. I found only two outcrops to have both monoecious and androdioecious populations; thus within-population diversity and genetic differentiation comparisons between the two types of populations were not significant, although the data trend towards supporting the criteria of this metapopulation

57 model. This is the first test of Pannell’s (2000; 2002) model in an animal system and the data suggest it might be a reasonable model regarding the persistence of androdioecy in

Eulimnadia. More investigation is required to determine the prevalence of monoecious populations in the metapopulation framework and to allow further comparisons of these populations with androdioecious populations for a significant test of this promising model.

Introduction

Androdioecy is an exceptional and rare mating system in which males and hermaphrodites co-occur in populations in the absence of true females. Males can play an important role in hermaphroditic populations to reduce inbreeding depression and to supplement male allocation in sperm-limited hermaphrodites (Wolf & Takebayashi,

2004). Yet males have difficulty invading hermaphroditic populations because of reduced mating opportunities due to the ability of hermaphrodites to self fertilize. This requires males to more than double their mating success (i.e., successful fertilizations) relative to hermaphrodites to effectively invade a hermaphroditic population (Lloyd, 1975;

Charlesworth, 1984). Theoretical models of mating system evolution that have incorporated these constraints predict that androdioecy is unlikely to evolve, and that those species exhibiting androdioecy are likely to be short-term, transitory states that are evolving from hermaphroditism to dioecy, or vice versa (Lloyd, 1975; Charlesworth,

1984).

While many species have been purported to be androdioecious, Charlesworth

(1984) showed most of the species described as androdioecious to that date were actually functionally dioecious. This exemplifies the rarity of this mating system. In 2002, Pannell

58 reviewed only six plant species that are truly androdioecious. Weeks (2012) conducted a review of the literature and found 115 animal species to be androdioecious, and the clam shrimp genus Eulimnadia represents one of the most widespread occurrences of androdioecy within any one generic taxon, providing us a unique opportunity to investigate the evolutionary implications of this mating system.

The clam shrimp genus Eulimnadia comprises approximately 44 species (Brtek,

1997) and of those, 16 species have been shown to be androdioecious (Weeks et al.,

2006b, Weeks 2012). These species are globally widespread and phylogenetic evidence suggests the ancestor to the genus had an androdioecious mating system (Hoeh et al.,

2006; Weeks et al., 2006b, 2009). The ancestor to this genus has been proposed to have existed 24 to 66 million years ago, based on fossil evidence (Tasch & Schaffer, 1964;

Chen & Shen, 1981). Weeks et al. (2006b) also used biogeographic evidence to evoke the vicariance hypothesis given the widespread distribution of Eulimnadia, and they argued that the ancestor to Eulimnadia occurred upwards of 180 million years ago. This evidence contradicts the prediction of the above noted models (Lloyd, 1975; Charlesworth, 1984) that androdioecy is a short-term, tranisional state in the evolution of hermaphroditism from dioecy or vice versa.

The models constructed by Lloyd (1975) and Charlesworth (1984) were based on characteristics of plant species that exhibit androdioecy. Eulimnadia spp. differ in several ways. The males are morphologically distinct, in that the first and second phyllopods in males have been modified into claspers used to grasp the carapace of the hermaphrodite for outcrossing. Hermaphrodites lack these claspers and thus they are unable to exchange sperm with one another; this is important because the above models assume mating

59 between hermaphrotites. The hermaphrodites also represent two distinct genotypes, termed “amphigenics” and “monogenics” (Sassaman & Weeks, 1993). The amphigenics are heterozygous for a sex determining gene or chromosome and produce hermaphrodites and males in a 3:1 ratio when selfed and a 1:1 ratio when outcrossed. The monogenics are homozygous and always produce hermaphroditic offspring (Sassaman & Weeks, 1993).

These unique characteristics of androdioecy in Eulimnadia are important to consider in modeling as they directly affect the type and amount of outcrossing that can occur in the population and the subsequent offspring and population sex ratios.

Otto et al. (1993) developed a model specific to Eulimnadia that incorporated the unique variables noted above. This model accounts for the presence of males when the proportion of eggs fertilized by the male () times the viability of the males (1-) is greater than two times the proportion of eggs that can be successfully selfed () times one minus the inbreeding depression of selfed offspring (). There have been many experimental studies to assess these parameters in one species, Eulimnadia texana

(Weeks et al., 1999, 2000, 2001; Zucker et al., 2001; Hollenbeck et al., 2002). The parameter estimates obtained in all these studies tend to show high variation that leads to a large range of possible values, particularly for , and no robust acceptance or rejection of the Otto et al. (1993) model has been published. However, among all these estimates of the parameters of the Otto et al. (1993) model, no combination has successfully accounted for the observed sex ratios in natural populations (Weeks & Bernhardt, 2004).

Thus, this model does not appear to adequately explain the maintenance of androdioecy in Eulimnadia.

More recently, Pannell (2000; 2002) presented an ecological model that predicts that extinction and colonization within a metapopulation can lead to the evolution and maintenance of androdioecy. This model relies on the interaction of local mate competition and reproductive assurance (increased fecundity through selfing when mates are in low abundance: Baker’s Law; Baker, 1955), and assumes that dioecy is the evolutionarily stable strategy in large, demographically stable populations. However, in areas where local extinction followed by colonization is common, self-compatible hermaphroditism is selectively advantageous due to reproductive assurance. Selfing hermaphroditism is maintained in the early portions of newly established populations due to low encounter frequencies and thus limited outcrossing opportunities. Once populations grow in size, and encounter rates are more frequent, males are able to secondarily invade and spread due to the benefits of outcrossing. With these expectations, we should find an assortment of populations with differing male frequencies (which correlate to differing times since population establishment) that are maintained over ecological timescales. It appears that this newer model of the evolution and maintenance of androdioecy is more relevant to studies of androdioecy in Eulimnadia than the Otto et al. (1993) model. The operation of metapopulation dynamics, with local extinction and colonization, might be our best explanation for the long-term persistence of androdioecy documented in Eulimnadia (Weeks et al., 2006a).

Pannell’s (2000; 2002) model makes specific, testable assumptions and predictions regarding the evolution of androdioecy. First, across the metapopulation, males should be the minority sex. This is true of all Eulimnadia species so far examined, where male frequencies typically range from 0% to ~30% (Sassaman, 1989; Weeks and

Zucker, 1999; Weeks et al., 2006b, 2008). Second, populations within a species should exhibit broad variation in male frequencies, with many populations having no or extremely few males. This is also shown in E. texana (Strenth, 1977; Sassaman, 1987) and E. dahli (Weeks et al., 2006). Third, hermaphrodites should show female-biased sex allocation. This has been documented in E. texana (Zucker et al., 1997) and E. agassizii

(Weeks et al., 1999) and recently in four more species of Eulimnadia (E. diversa, E. radhakrishnai, E. gunturensis, and E. michaeli; Brantner, 2011). Fourth, the species should have a colonizing habit. Eulimnadia inhabit temporary freshwater pools and their desiccation-resistant eggs rest in the soil during dry periods (Weeks et al., 1997), which is common to all large Branchiopods (Brendonck, 1996). Erosion is continuously changing the landscape and potentially opening up new depressions which are then available for colonization via wind or bird dispersal. Brendonck and Riddoch (1999) showed that anostracan eggs are wind-dispersed over short distances and Vandekerhov and colleagues (2005) found cladocerans could colonize newly formed ponds within one year. This suggests Branchiopods with this life history are well adapted to a colonizing habit. Fifth, the androdioecious species should have been derived from a dioecious ancestor. Hoeh et al. (2006) generated a phylogeny of the family of clam shrimp to which Eulimnadia belongs (the Limnadiidae) and found that the dioecious Metalimnadia

(Fig.2) is the most likely sister taxon to Eulimnadia. Additionally, the genus, Imnadia, sister to the Eulimnadia + Metalimnadia clade is also dioecious. These results suggest

Pannell’s model might well explain the maintenance of androdioecy in these shrimp.

The most important untested assumptions critical to Pannell’s (2000; 2002) hypothesis for the evolution of androdioecy in clam shrimp are: (1) populations should

62 have low inter-population gene flow; and (2) there should be high population turnover.

For Pannell’s model to hold, it is essential that the species studied be sub-structured into demes (constituting a metapopulation) with broadly varying male frequencies, and have a history of localized extinctions and recolonizations. If these assumptions do not hold, then Pannell’s metapopulation hypothesis cannot explain the evolution of androdioecy in the Eulimnadia.

Pannell and Charlesworth (2000) reviewed the expectations associated with measures of genetic diversity and metapopulation dynamics: when population turnover exceeds the rate of migration between populations, within-population genetic diversity is expected to be reduced. At the same time, genetic differentiation should be increased among populations. Pannell’s (2000) model assumes that the populations most recently colonized should contain only hermaphrodites (monoecious) and that males invade and persist secondarily as population numbers increase. Thus it is expected that monoecious populations will be of a lower density, on average, and should show lower within- population diversity and higher genetic differentiation than androdioecious populations.

Herein I report the results of a broad survey of sub-populations of the clam shrimp Eulimnadia dahli in an effort to test the metapopulation hypothesis in an animal system. Eulimnadia dahli is an androdioecious clam shrimp known to occur in rock-pools atop granite outcrops throughout Western Australia (Weeks et al. 2006). Each rock outcrop has a mosaic of rock-pools with populations of E. dahli that represent subpopulations, while each outcrop may represent a metapopulation. Thus analyses were conducted on a hierarchical basis. Specifically, I surveyed sex-ratio variation, population size, gene flow rates within and between rock pools (sub-populations) and within and

63 between outcrops (metapopulations) to test the assumptions of Pannell’s model. I further use these data to address the question of population turnover and whether a history of metapopulation dynamics, including extinction and recolonization, can be established in

E. dahli.

Materials and Methods

Field Sampling

A field trip to Western Australia was conducted from February to April, 2007 in the Wheatbelt Region in the southwest of Western Australia. This is one of nine regions in WA and encompasses 154,862 square miles. It partially surrounds the city of Perth and is bordered in the north by the Mid-West Region, in the east by the Goldfields-Esperance

Region, and in the south by the South West and Great Southern Regions. Eleven granite outcrops were surveyed within this region. All were found to be dry, and thus approximately 1 liter of sediment from the bottom of several (5-9) rock-pools was collected from each of nine outcrops sampled. Latitude/longitude coordinates were taken at each rock pool using GPS. Geographic distances between rock pools and between outcrops were calculated using the simple equation: square root ((69.1 x ((latitude 1 - latitude 2)2)) + (53 x ((longitude 1 – longitude 2)2))) = distance in miles. The sediment was transported back to the University of Akron for laboratory hydrations.

Laboratory Rearing of Field-Collected Soil

For each rock-pool, a small sample of sediment (~200 ml) was placed in the bottom of a 38 L glass aquarium. The aquarium was filled with either 38 L (Full) or 19 L

(Half) deionized water and aerated with one small air stone. These aquaria were kept in a

64 temperature controlled animal facility at the University of Akron at 26-28C and under

24-hour light conditions (Durotest Sunlight Simulating fluorescent bulbs). Each aquarium received ~10 ml of food each day. Food was composed of 2.5 g of ground Tetramin® flake food for algae-eating fish and 2.5 g of baker’s yeast suspended in 500 ml of deionized water. This protocol has been shown to be the most successful in raising a variety of species of clam shrimp (Weeks et al. 1997, 2006, 2009).

The aquaria were monitored daily for hatching and shrimp were allowed to grow to maturity. Once fully developed, the shrimp were identified and all Eulimnadia dahli individuals were counted. The sex of each individual was identified, and the sex ratio for each population was calculated. Up to 100 individuals were randomly sampled from this population and frozen at -80C.

Allozyme Extraction and Analysis

Electrophoretic assays were run using cellulose acetate gels (Richardson et al.,

1986) to visualize six enzyme loci: Fum (fumarate hydratase, EC 4.2.1.2), Hk

(hexokinase, EC 2.7.1.1), Idh (isocitrate dehydrogenase, EC 1.1.1.42), Mpi (mannose- phosphate isomerase, EC 5.3.1.8), 6Pgd (6-phophogluconate dehydrogenase, EC

1.1.1.44) and Pgm, (phosphoglucomutase, EC 5.4.2.2). Alleles were designated by increasing anodal mobility.

Estimates of genetic diversity were calculated: (1) across all outcrops; (2) across all rock-pool populations on each outcrop; and (3) within each rock pool population.

Allelic richness, Nei’s gene diversity He (expected heterozygosity, Nei, 1987), and

Wright’s (1951) F-statistics were calculated and significant differences between each comparison above were assessed through randomization tests in FSTAT 2.9.3 (Goudet,

1995). Significant differences in the above variables between monoecious and androdioecious populations were assessed in FSTAT 2.9.3 (Goudet, 1995). The hierarchical model of population structure was examined with hierarchical F-statistics in

FSTAT 2.9.3 (Goudet, 1995).

Multiple regression analyses were performed using JMP (SAS Institute 2003) to test the predictions that genetic diversity, population size, and population densities should be positively correlated with male frequencies in androdioecious populations (Pannell,

2000). Input variables included rock pool size, rock pool population size, and within- population genetic diversity. Regression analyses were also used to determine whether the geographic distances between rock-pools and outcrops were correlated with gene flow (isolation by distance). These were done in FSTAT 2.9.3 (Goudet, 1995) using the

Partial Mantel/Regression test. All comparisons were analyzed in a hierarchical fashion: among the rock-pool populations on each outcrop and then combined to compare across all population samples across all outcrops.

Results

Eulimnadia dahli was hatched from 32 rock-pools from nine of the eleven outcrops sampled (Table 4.1). Population numbers ranged from only one individual

(Dingo Rock, pool 4 and Elachbutting, Pool 7) to 971 (The Humps, pool 3). Sex ratios ranged from 0% male (monoecious pools) to 47.4% male (with the exception of Dingo

Rock, pool 4, which only had one individual, which was male). Only two rock outcrops had a combination of both monoecious and androdioecious pools. Tammin Rock had three monoecious pools (4, 5, and 7) and two androdioecious pools (2 and 6), while

Yorkrakine Rock had just one monoecious pool (9) and three androdioecious pools (1, 2, and 5). It should be noted that Yorkrakine pool 9 corresponds to Yorkrakine pool 2 in

Weeks et al. (2006).

TABLE 4.1 – POPULATION DESCRIPTIVE STATISTICS

Outcrop Pool Total Population %Male N Ho He Fis Shrimp Density (per L) Bunjil 1 268 7.09 31.2 62 0.12 0.23 0.47 2 456 12.06 11.0 62 0.11 0.15 0.26 3 131 3.47 10.7 53 0.11 0.14 0.09 4 6 0.16 33.3 5 32 0.85 21.9 30 0.13 0.10 -0.30 6 15 0.40 26.7 10 0.18 0.34 0.40 7 396 10.48 21.2 53 0.12 0.28 0.56 8 15 0.40 26.7 10 0.20 0.23 0.12 Cairn 1 64 3.39 25.0 54 0.12 0.13 0.11 2 5 0.13 20.0 5 55 1.46 12.7 40 0.04 0.04 -0.06 Dingo 4 1 0.05 100.0 Elachbutting 5 37 1.96 27.0 34 0.13 0.17 0.28 7 1 0.05 0.0 Kookaburra 3 6 0.32 16.7 5 7 0.37 14.3 6 5 0.26 20.0 Tammin 2 108 2.86 0.9 51 0.09 0.07 -0.23 4 46 2.43 0.0 40 0.07 0.08 0.17 5 923 48.84 0.0 55 0.09 0.12 0.30 6 910 48.15 2.8 58 0.10 0.12 0.18 7 45 2.38 0.0 40 0.04 0.04 -0.16 The Humps 3 971 25.69 21.3 61 0.06 0.09 0.37 5 255 6.75 12.9 60 0.02 0.02 -0.07 6 23 0.61 13.0 20 0.06 0.05 -0.23 Wave 3 121 3.20 25.6 61 0.07 0.12 0.44 6 19 0.50 47.4 16 0.08 0.12 0.35 8 8 0.21 25.0 Yorkrakine 1 15 0.40 7.1 15 0.01 0.01 0.00 2 25 0.66 8.7 20 0.11 0.13 0.19 5 145 3.84 16.6 50 0.08 0.18 0.57 9 159 4.21 0.0 45 0.01 0.01 -0.04 Outcrop name, pool number, total shrimp hatched, population density per liter, percent males in population, number of shrimp genotyped (N), mean observed (Ho) and expected (He) heterozygosities for the six enzyme loci and Fis for each of the sampled Eulimnadia dahli populations.

Pairwise FST estimates ranged from -0.02 to 0.87 and were significant in 170 of

276 cases after Bonferroni adjustment for multiple comparisons (Table 4.2). Nm estimates

were calculated from pairwise FST estimates and ranged from -1250.25 to 45.20 (Table

4.2).

TABLE 4.2 – PAIRWISE FST AND Nm ESTIMATES

B1 B2 B3 B5 B6 B7 B8 C1 C5 E5 H3 H5 H6 T2 T4 T5 T6 T7 W3 W6 Y1 Y2 Y5 Y9 B1 -- 1.73 1.06 0.73 2.58 3.93 1.07 0.15 0.16 0.22 0.15 0.12 0.14 0.08 0.10 0.12 0.12 0.07 0.18 0.22 0.17 0.20 0.28 0.12 B2 0.13 -- 1.89 1.05 2.08 1.34 3.62 0.14 0.07 0.17 0.14 0.09 0.13 0.08 0.08 0.10 0.09 0.05 0.16 0.18 0.08 0.18 0.20 0.06 B3 0.19 0.12 -- 19.28 1.96 0.66 1.01 0.12 0.05 0.15 0.11 0.07 0.10 0.07 0.07 0.09 0.08 0.05 0.13 0.14 0.06 0.14 0.16 0.05 B5 0.26 0.19 0.01 -- 1.24 0.51 0.67 0.10 0.04 0.12 0.08 0.04 0.07 0.08 0.05 0.07 0.07 0.03 0.11 0.10 0.04 0.11 0.13 0.03 B6 0.09 0.11 0.11 0.17 -- 1.59 3.65 0.20 0.10 0.31 0.11 0.07 0.16 0.11 0.06 0.07 0.07 0.06 0.23 0.32 0.15 0.30 0.23 0.07 B7 0.06 0.16 0.28 0.33 0.14 -- 1.57 0.17 0.19 0.22 0.16 0.13 0.16 0.07 0.14 0.16 0.15 0.10 0.18 0.23 0.22 0.23 0.29 0.14 B8 0.19 0.06 0.20 0.27 0.06 0.14 -- 0.15 0.05 0.20 0.11 0.05 0.11 0.33 0.06 0.09 0.08 0.04 0.16 0.21 0.06 0.21 0.20 0.03 C1 0.63 0.64 0.67 0.71 0.55 0.60 0.62 -- 0.15 0.39 2.51 1.20 2.74 0.05 0.17 0.29 0.24 0.11 1.26 1.28 0.17 0.74 0.47 0.12 C5 0.62 0.79 0.82 0.87 0.71 0.56 0.85 0.62 -- 0.10 0.21 0.05 0.07 0.22 0.16 0.20 0.21 0.07 0.14 0.09 45.20 0.09 0.41 14.04 E5 0.54 0.59 0.62 0.67 0.45 0.53 0.55 0.39 0.72 -- 0.27 0.15 0.26 0.15 0.15 0.25 0.21 0.09 0.51 0.66 0.12 0.40 0.44 0.08 H3 0.63 0.65 0.69 0.75 0.69 0.61 0.69 0.09 0.55 0.48 -- 6.89 13.87 0.07 0.11 0.18 0.16 0.07 1.99 1.57 0.20 0.71 0.66 0.16 H5 0.68 0.74 0.79 0.87 0.79 0.66 0.84 0.17 0.85 0.63 0.04 -- 3.19 0.11 0.07 0.13 0.11 0.02 1.22 0.50 0.03 0.29 0.46 0.03 H6 0.64 0.66 0.71 0.79 0.61 0.60 0.70 0.08 0.78 0.49 0.02 0.07 -- 0.13 0.07 0.14 0.12 0.04 2.12 1.38 0.05 0.66 0.52 0.04 T2 0.75 0.76 0.78 0.83 0.75 0.70 0.78 0.43 0.83 0.53 0.62 0.78 0.69 -- 0.20 0.61 -0.24 0.10 0.15 0.13 0.05 0.13 0.13 0.04 T4 0.71 0.76 0.79 0.84 0.81 0.65 0.81 0.60 0.61 0.62 0.69 0.79 0.78 0.55 -- 2.75 4.87 2.20 0.11 0.10 0.13 0.09 0.16 0.11 T5 0.67 0.71 0.73 0.78 0.79 0.61 0.74 0.47 0.56 0.50 0.58 0.65 0.64 0.29 0.08 -- -1250.25 0.61 0.17 0.16 0.17 0.15 0.21 0.14 T6 0.68 0.73 0.75 0.79 0.79 0.62 0.75 0.51 0.55 0.54 0.61 0.69 0.67 0.36 0.05 0.00 -- 0.84 0.15 0.15 0.18 0.14 0.19 0.14 T7 0.78 0.83 0.85 0.90 0.79 0.72 0.87 0.70 0.77 0.73 0.78 0.92 0.88 0.72 0.10 0.29 0.23 -- 0.07 0.05 0.05 0.05 0.10 0.04 W3 0.58 0.61 0.65 0.70 0.52 0.58 0.61 0.17 0.65 0.33 0.11 0.17 0.11 0.63 0.69 0.59 0.63 0.78 -- -16.92 0.16 1.17 0.88 0.12 W6 0.53 0.59 0.64 0.71 0.44 0.52 0.55 0.16 0.73 0.28 0.14 0.33 0.15 0.66 0.72 0.60 0.63 0.83 -0.02 -- 0.10 1.36 1.00 0.06 Y1 0.59 0.76 0.80 0.87 0.63 0.53 0.81 0.60 0.01 0.68 0.55 0.89 0.83 0.84 0.67 0.59 0.58 0.83 0.62 0.71 -- 0.11 0.45 -12.45 Y2 0.55 0.58 0.63 0.69 0.46 0.53 0.54 0.25 0.73 0.39 0.26 0.46 0.28 0.67 0.73 0.62 0.65 0.83 0.18 0.16 0.70 -- 1.93 0.07 Y5 0.47 0.56 0.61 0.66 0.52 0.46 0.55 0.35 0.38 0.36 0.27 0.35 0.33 0.66 0.62 0.55 0.56 0.72 0.22 0.20 0.35 0.11 -- 0.30 Y9 0.68 0.81 0.85 0.91 0.78 0.64 0.89 0.67 0.02 0.77 0.61 0.90 0.87 0.87 0.70 0.64 0.64 0.86 0.68 0.80 -0.02 0.79 0.46 --

Pairwise FST (below diagonal) and Nm (above diagonal) estimates for Eulimnadia dahli populations. FST values were averaged across all loci. Values in bold are significant after Bonferroni adjustment. Letters refer to the outcrop and numbers refer to the population (pool) as listed in Table 1: B, Bunjil Rock; C, Cairn Rock; E, Elachbutting Rock; H, The Humps; T, Tammin Rock; W, Wave Rock; Y, Yorkrakine Rock.

A hierarchical model of population genetic structure (Table 4.3) shows significant non-random mating at both the outcrop (Theta-P) and at the pool (Theta-S) levels, confirming that each outcrop represents a metapopulation with a suite of distinct subpopulations at the rock pool level.

TABLE 4.3 – HEIRARCHICAL MODEL OF POPULATION GENETIC STRUCTURE

FIS 0.270 -0.087  0.569

FIT 0.749 F - Statistics 0.352  0.922 Theta - S 0.656 0.362  0.833 Theta - P 0.538 0.212  0.814 Summary F-statistics and variance obtained through 1000 bootstrap replicates at 95% confidence level (random seed = 1). FIS and FIT are total estimates. Theta – S is at the subpopulation (pool) level. Theta – P is at the population (outcrop) level.

No significant differences in expected heterozygosity were found between monoecious (He= 0.071) and androdioecious (He= 0.127) rock pool populations (P =

0.1310) across all outcrops. Again, within just Tammin rock, no differences were found

between the monoecious (HS = 0.080) and androdioecious (Hs = 0.095) rock-pools (P =

0.8110).

Percentage of males was found to be significantly correlated with genetic diversity (P = 0.0045) (Figure 4.1), but not with population density (P = 0.1245) or pool size (P = 0.6930).

FIGURE 4.1 – PERCENT MALES VS. GENETIC DIVERSITY

0.3

0.25

) e 0.2

0.15

0.1 Genetic DiversityGenetic

Genetic Diversity (H Genetic 0.05

0 0 10 20 30 40 50 Percent Males

The relationship of percentage of males in rock-pools and genetic diversity is significant (P = 0.0045).

Estimates of FST were found to correlate with geographic distance between outcrops (P = 0.0165), but not with the geographic distance of pools on outcrops (P =

0.6875). Thus geographic distance is related to genetic differentiation (isolation by distance) at the outcrop level, but not on the pool level.

FIGURE 4.2 – GENE FLOW VS. DISTANCE OF OUTCROPS

0.8

0.7

0.6

0.5

0.4

Fst Fst 0.3

0.2

0.1

0 0 10 20 30 40 50 Distance between Outcrops (Km)

The relationship of gene flow (Fst) between outcrops and the distance of those outcrops (P = 0.0165).

Discussion

The rare androdioecious mating system is widespread in the clam shrimp genus

Eulimnadia (Weeks et al. 2006a, 2006b) and evidence suggests that it has persisted long- term throughout the evolution of the genus (Weeks et al 2006b). Pannell (2000, 2002) provided an ecological model that argued androdioecy could persist where metapopulation dynamics acted upon the populations in the form of extinction and recolonization, although long-term persistence is still unlikely. Pannell’s (2000, 2002) model outlined the specific situations for which androdioecy would be maintained within a metapopulation. Males are expected to be the minority sex and populations should exhibit broad variation in male frequencies, including many monoecious populations (no males). Further, populations should have low inter-population gene flow and high

72 population turnover. This is best tested through a comparison between monoecious and androdioecious populations (Obbard et al. 2006).

Eulimnadia dahli populations observed in this study satisfied the criteria that males are the minority sex and vary in frequency between the populations. This is consistent with all Eulimnadia species examined to date (Sassaman, 1989; Weeks and

Zucker, 1999; Weeks et al. 2006b, 2008). Weeks et al. (2006b) observed populations of

E. dahli on five rock outcrops and reported sex ratios ranging from 0% - 31% males.

They only sampled from multiple pools on one outcrop, but found two out of six pools to be monoecious. In this study, we did not find nearly the number of monoecious populations we had expected under the Pannell model across the outcrops. Only two outcrops of the nine investigated yielded monoecious pools, and only one contained multiple monoecious pools.

Monoecious populations should correspond to small, newly formed and highly selfing populations as described in Pannell’s model. It is only when monoecious populations grow in size that males are able to secondarily invade due to increased outcrossing opportunity. The absence of monoecious pools on so many of the outcrops could point to poor sampling strategy. This is possible in that we were only able to sample populations in their dry state. We collected sediment from what we found to be obvious depressions in the outcrop that would hold water during the wet season, but this sampling strategy could have led to a bias towards larger and longer established pool environments. Alternatively, if our sampling strategy was sufficient (as we did find some monoecious populations on two outcrops), it would imply that there were just not many newly established small populations. Assuming Pannell’s model is correct, this would

73 indicate all populations on the outcrops were older, well established populations where males had invaded and proliferated (Pannell, 1997). This could indicate that population turnover is not as high as required by Pannell’s model to maintain an assortment of monoecious and androdioecious populations in the metapopulation. Under the scenario described by Pannell (1997), this will eventually lead to dioecious populations as the evolutionary stable strategy, which does not seem to be the case for E. dahli, or any

Eulimnadia species studied to date (Weeks et al., 2008).

Pairwise FST and migration estimates showed significant, non-random mating among many pools due to the spatial structure between these pools. Further, a hierarchical model of population genetic structure showed significant non-random mating at both the among-outcrop and among-pools-on-outcrops levels. A metapopulation is often broadly defined as any spatially structured population (Hanski 1998, Pannell &

Obbard 2003). Thus our results confirm that E. dahli populations are structured in metapopulations on each outcrop with a suite of sup-populations at the rock pool level.

Pannell’s model is based on classic metapopulation theory (Levins 1969, 1970) in which geographically distinct subpopulations are connected via dispersing migrants and where extinction and recolonization are fundamental in maintaining the population (Hanski

1997, 1998). It is the simultaneous benefits of reproductive assurance and outcrossing in the metapopulation which sustain androdioecious populations in the long-term (Pannell,

1997, 2002).

Obbard et al. (2006) conducted the first empirical test of the metapopulation hypothesis (Pannell 2000, 2002) through a comparison of genetic variation between sexual systems in the plant Mercurialis annua. It is essential to use a comparative

74 approach between monoecious and androdioecious populations to test the prediction that within population diversity is reduced and that genetic differentiation is greater among monoecious relative to androdioecious populations (Pannell 2000; 2002). Obbard et al.

(2006) compared 29 monoecious populations to 16 androdioecious populations, using genetic diversity estimates obtained from six allozyme loci. They found within- population genetic diversity was significantly higher (P<0.001) in the androdioecious populations relative to the monoecious populations, and they found genetic differentiation (Fst) was significantly higher (P<0.001) in monoecious relative to androdioecious populations.

Pannell (2002) finds the predictions of his model seem to hold well for all taxa that had been confirmed as androdioecious to that date, whether plant or animal. Our study is the first empirical test of this hypothesis in an animal model and closely follows the same comparative approach as conducted by Obbard (2006). The androdioecious system in Eulimnadia does differ from this mating system in plants because hermaphroditic clam shrimp are unable to outcross with one another, lacking the claspers necessary to couple with other hermaphrodites and transfer sperm. Thus, the avoidance of inbreeding depression and the assumption of delayed selfing become key conditions for the successful invasion of males (Otto et al. 1993; Pannell 2002). Pannell (2002) finds this to be consistent with the metapopulation model where selfing rates are variable due to fluctuation population structure. Further, one criticism of the metapopulation model in plants is the difficulty in definining discrete habitat patches (Obbard et al. 2006). For this reason, Eulimnadia dahli is an ideal system for testing this model due to its occurrence in discrete pools on the granite rock outcrop. However, in this study, we found no

differences in either population diversity (He) or genetic differentiation between monoecious and androdioecious populations (Fst) at either the outcrop or pool levels.

Thus, Pannell’s (1997) predictions were not upheld in these two respects.

Pannell’s (1997) model posits a scenario in which males secondarily invade monoecious populations once they are established and grow in size. From this, we predicted that we should find higher male frequencies in larger populations and these populations should be more genetically diverse. Percentage of males was found to be significantly correlated with genetic diversity, but not with population density or pool size. We do expect that a higher percentage of males should correlate with higher genetic diversity, as males are the only mode of outcrossing in Eulimnadia (Weeks 1997). The lack of correlation with population density and pool size lends no support for Pannell’s model.

Gene flow estimates (FST) were found to correlate with geographic distance between outcrops, but not with the geographic distance of pools on the outcrops. Thus, geographic distance is related to genetic differentiation (isolation by distance) at the outcrop level, but not at the pool level. Classic metapopulation theory is dependent upon four criteria: (1) habitats are discrete patches with local breeding populations, (2) all populations are at risk of extinction, (3) population isolation does not inhibit recolonization, and (4) local dynamics ensure that simultaneous extinction of all populations is unlikely (Hanski 1997, 1999). Due to the isolation by distance found between outcrops, but not pools, our hypothesis that each outcrop represents a metapopulation is verified. Gene flow estimates were significant between pools, but we found no isolation by distance at the pool within outcrop level. Therefore, at the pool

76 level, we deduce that differentiation is consistent with the processes of extinction and recolonization. This supports the assumption of population turnover in Pannell’s (1997) metapopulation model.

Our study shows the androdioecious clam shrimp, Eulimnadia dahli, most certainly exists in a structured metapopulation with some evidence that the subpopulations experience turnover in the form of extinction and recolonization. Our study did not provide as many monoecious populations as anticipated, thus limiting our ability to fully accept or reject the Pannell’s (2000, 2002) metapopulation hypothesis. Pannell’s

(1997) model for the evolution of androdioecy in a metapopulation seemed the most relevant to explain the long-term persistence of androdioecy in Eulimnadia, but because of the small number of monoecious populations, it was difficult to adequately test the criteria of low inter-population gene flow and high population turnover, which is best assessed via a comparison between monoecious and androdioecious populations.

Certainly, there was a trend within Tammin Rock that could support the predictions of

Pannell’s model, but we would need more monoecious pools to further explore this trend.

Pannell (1997, 2002) indicates that the processes of extinction and re-colonization should be occurring continuously to maintain androdioecy over ecological timescales.,

Direct measures of population turnover in natural populations would be best to fully test this model. Our sampling only occurred during one field season. A more rigorous study over multiple seasons might show changes in the subpopulations observed here, and sampling in different environments for other androdioecious Eulimnadia would strengthen the test of Pannell's model. Additionally, we were unable to sample E. dahli when the pools were filled onsite. This could have led to us missing smaller pools as we

77 searched for depressions when they were dry, which might have resulted in our biasing toward larger, better-established populations. Such a potential bias would likely yield a higher number of androdioecious populations.

Eulimnadia survive the dry periods as encysted eggs that hatch as the rock pools fill with rain. These eggs banks are analogous to seed banks of plants where prolonged seed dormancy allows for persistence in ephemeral habitats, and only a portion of the eggs hatch during any given hydration event (De Stasio, 1989). It is interesting to consider how the presence of these egg banks could be a factor in population turnover and the processes of extinction and recolonization as implied in a metapopulation model.

During this same field season, we witnessed differential hatching in the dioecious shrimp

Limnadia badia (Benvenuto et al. 2009). The disproportion of hatching of these diapausing eggs could be thought of in terms of temporal dispersal and recolonization

(Hairston and Kearns 2002) and thus, while the pool containing an egg bank might never actually experience true extinction, it may not be populated during any particular hydration event. Further, the population composition and whether the population is monoecious or androdioecious may not necessarily be a factor of population size and/or age as outlined in Pannell’s (1997, 2002) model, but may be influenced by fluctuation in environmental hatching cues. In this way, true extinction and recolonization dynamics may not hold for the life history of egg banks, but differential hatching could still allow for Pannell’s (1997, 2002) model of metapopulation dynamics to adequately explain the persistence of androdioecy in these organisms. For this reason, it would be very important to monitor these populations across many field seasons and in the natural environment. The random sampling of sediment in the field and the ideal hatching

78 conditions we strive for in the lab might not be good indicators of how populations are composed in the natural environment and how well they conform to the requirements of population turnover in the metapopulation model.

The current study does confirm that E. dahli exists in a metapopulation framework, and shows evidence that androdioecy in E. dahli could persist due to the dynamics outlined by Pannell (1997, 2002). Therefore, further investigations that include a greater number of monoecious populations are certainly warranted.

Acknowledgements

I am sincerely grateful to Chiara Benvenuto, Alissa Calabrese, Rob Davis, Wally Gibb,

Debra Judge, Kerry Knott, Danny Tang, Magdalena Zofkova, and the School of Animal

Biology at The University of Western Australia for much help during the field study.

Above all, Brenton Knott’s help and advice were essential throughout the field study.

Those in the shire offices of Western Australia provided valuable information in locating field sites and Danny Stefoni at the Department of Conservation and Land Management of Western Australia helped to obtain the required collecting permits. Tom Sanderson helped with laboratory animal rearing and Amanda Crow helped with initial electrophoretic analysis. I would like to thank Francisco Moore for his help with statistical analysis and interpretation. This work was supported by the National Science

Foundation [grant number IBN-0213358] and the National Science Foundation Doctoral

Dissertation Improvement Grant [DEB-0709618]; the Kent State University Graduate

Senate InTraGrant and the Kent State University David B. Smith Fellowship; Sigma Xi’s

Grants in Aid of Research, and the Crustacean Society Denton Belk Memorial

Scholarship in Graduate Studies focusing on Large Branchiopod research.

V. GENERAL DISCUSSION AND CONCLUSIONS

The clam shrimp genus Eulimnadia is the most speciose and widely distributed

Spinicaudatan genus. It is comprised of approximately 54 species found on every continent except Antarctica. These species exhibit an ephemeral life cycle where adults occur in temporary freshwater pools and eggs encyst to persist through dry periods. These species are of particular interest because they exhibit androdioecy, an exceptionally rare mating system where males and self-compatible hermaphrodites occur in populations with no true females. Fourteen species of clam shrimp have been confirmed to be androdioecious and phylogenetic data show this to be the ancestral condition in this group (Weeks et al. 2005, 2006, 2008). This means the Eulimnadia represent the most widespread occurrence of androdioecy in the animal kingdom.

In fact, the prevalence of androdioecy in the Eulimnadia is a puzzle because the theoretical models of Lloyd (1975) and Charlesworth (1984) show that androdioecy is unlikely to evolve and in those cases where it is found, it is likely a short-term occurrence in populations evolving from hermaphroditism to dioecy, or vice versa. Weeks et al.

(2006, 2008) make the case that not only is androdioecy the likely mating system for all the species of this large genus, and thus the ancestral condition, but that it is also exceptionally long-lived, persisting for 24 to 180 million years. For this reason, the

Eulimnadia present a unique opportunity for studying mating system evolution and the persistence of androdioecy. Yet, given the importance of the Eulimnadia, much is not known about the phylogenetic relationships of the species within the genus or its biogeographic history.

In Chapter II, I sought to investigate current species distinctions and the intrageneric relationships of the species within the Eulimnadia. I used 22 species to conduct a molecular phylogenetic analysis. I used sequence data from the mitochondrion- encoded cytochrome b (cyt b) and the nuclear-encoded elongation factor 1 α (EF1α) and I contributed 122 new sequence fragments to this analysis. I analyzed these new sequences, along with those obtained from GenBank, using Maximum Likelihood (ML) and

Bayesian Inference (BI) approaches. My data support previous studies upholding the monophyly of the Eulimnadia (Hoeh et al., 2006; Schwenter et al., 2009; Weeks et al.,

2009). This study also found a high degree of polyphyly for specimens identified as a single species, highlighting the difficulties in species identification.

Morphological taxonomic determinations have been fraught with difficulties due to the intraspecific variability in a number of the characters used to date and the limited numbers of characters used to distinguish one species from another (Belk, 1989; Martin and Belk, 1988; Sissom, 1971; Straskraba, 1966; Vidrine et al., 1987; Zaffagnini, 1971).

Recently, much emphasis has been placed on the morphology of the shell of the encysted eggs as an important species-specific diagnosis. Belk (1989) synonymized seven North

American species described by Mattox (1954) into just one, Eulimnadia diversa based on egg morphology. The results of Chapter II suggested that these seven species are not, in fact, distinct species. However Eulimnadia diversa is also not an easily recognized and identified species because I found that the specimens identified as E. diversa were polyphyletic. Overall, specimens tended to be monophyletic among populations, but not at the species level, such that specimens identified as the same species from multiple

82 populations were polyphylectic, but multiple specimens from the same population readily grouped within the same clade.

The results of Chapter II emphasize the need to further investigate inter- populational variability within species to better understand what constitutes a species in the Eulimnadia. Further, a full taxonomic revision is called for to better identify species before their phylogeny can be investigated in more depth.

In Chapter III, I aimed to investigate the biogeographic history of the Eulimnadia.

I was most interested in trying to test the hypothesis put forth by Weeks et al. (2006) which reasoned the Eulimnadia likely dispersed during a time when the continental positions held landmasses contiguous, up to 180 million years ago, to account for the cosmopolitan distributions of the extant species. This hypothesis was based on the limited dispersal capabilities of the large Branchiopods. Wind dispersal during the encysted stage is the likely mode of dispersal (Brendonck, 1996; Vanderkerhov et al. 2005) where long distance dispersal would be achieved in a “stepping stone” method (Tasch, 1987).

Animals such as birds might also serve as dispersal vectors, but no animals could account for long distance dispersal if the Eulimnadia were globally distributed prior to the

Cretaceous (~65 mya).

I was able to use the 54,240 phylogenetic trees computed from the two gene

Bayesian Inference (BI) analysis conducted in Chapter II to reconstruct the biogeographic history of the Eulimnadia using Statistical Dispersal-Vicariance Analysis (S-DIVA). The results suggest a combination of vicariance and dispersal shaped the present day distribution patterns of the Eulimnadia species. I found the Eulimnadia have a New

World origin and I inferred nine dispersal events in the history of the Eulimnadia species

83 studied, of which five could support the ancient dispersal hypothesis of Weeks et al.

(2006) and four have more recent dispersal explanations. Future studies of this nature should include a more extensive taxon sampling throughout the areas not represented in this analysis to more fully understand the distributional areas of each species as well as yield better supported phylogenetic trees for less ambiguous ancestor area reconstructions. Further, the development of a molecular clock would help to reveal when these dispersal events occurred.

In Chapter IV, I conducted a population genetic study on the androdioecious clam shrimp Eulimnadia dahli to investigate Pannell’s (2000; 2002) ecological model using the metapopulation dynamics of extinction and recolonization to account for the long- term persistence of androdioecy despite traditional mathematical models (Charlesworth

1984; Lloyd 1975) which suggest it to be a transitory stage in the evolution of dioecy or hermaphroditism. In Pannell’s (2000, 2002) metapopulation model, recently colonized populations should be monoecious (hermaphrodites only) while larger and longer sustained populations allow for secondary male invasion transforming these populations into androdioecious populations. Such a scenario predicts that a metapopulation should exhibit a mosaic of hermaphroditic and androdioecious populations. Under these conditions, it is expected that monoecious populations will be of a lower density and show lower within-population diversity and higher genetic differentiation among populations relative to androdioecious populations.

To test these predictions, I sampled Eulimnadia dahli occurring in isolated metapopulations atop granite outcrops throughout Western Australia. A total of 32 populations were sampled across nine granite outcrops. Gene flow estimates (FST) were

84 found to correlate with geographic distance between outcrops (metapopulations), but not between populations on the outcrops. With no isolation by distance at the population level within an outcrop, differentiation can be attributed to the processes of extinction and recolonization, which are the foundation of Pannell’s (2000; 2002) model. However, I found only two of the nine outcrops to have both monoecious and androdioecious populations: Tammin Rock had three monoecious populations and two androdioecious populations and Yorkrakine Rock had just one monoecious population and three androdioecious populations. Consequently, within-population diversity and genetic differentiation comparisons were not significant. While there was a trend in Tammin

Rock that could support the predictions of this model, the absence of monoecious populations in the majority of outcrops sampled limited my ability to fully accept or reject Pannell’s (2000; 2002) model for the evolution of androdioecy and the overall lack of monoecious populations suggests that population turnover might not be adequate for this model. This study represents the first empirical test of this model in an animal system, but it was only conducted over one field season and the clam shrimp had to be reared in the laboratory. Future studies should include sampling across multiple field seasons with particular emphasis on collections during the wet period when pools on the outcrops are filled.

The three studies conducted in this dissertation contributed to several different aspects of knowledge concerning the phylogeny, biogeography, and population genetic structure of the androdioecious clam shrimp genus Eulimnadia. The long-term stability of androdioecy in this large and widely distributed group contradicts the models of the evolution of this breeding system (Lloyd 1975; Charlesworth1984) put forth to date, and,

85 therefore has major implications in our understanding of the benefits of dioecy versus hermaphroditism (Weeks et al. 2006). It is imperative to have a good taxonomic foundation for this genus to investigate the phylogeny and biogeographic history of the androdioecious species, and it is vital to further investigate the metapopulation model

(Pannell 2000; 2002) to understand the persistence of this rare mating system. This system presents an exceptional opportunity to explore major questions in the evolution of mating systems.

REFERENCES

Babu, K. K. S. and S. B. Nandan. 2010. Two new clam shrimp species (Crustacea: Branchiopoda: Spinicaudata) from Kerala, India. Zootaxa. 2713: 55-64.

Baird, W. 1852. Monograph of the family Branchiopodidae, a family of crustaceans belongin to the division Entomostraca, with a description of a new genus and species of the family, and two new species belonging to the family Limnadiidae. Proceedings of the Zoological Society of London. 20: 18-37.

Baird, W. 1860. Description of two new species of Entomostracous Crustaceans from India. Proceedings of the Zoological Society of London.

Baker, H. G. 1955. Self-compatibility and establishment after "long-distance" dispersal. Evolution. 9: 347-349.

Barnard, K. H. 1929. Contributions to the crustacean fauna of South Africa. A revision of the South African Branchipoda (Phyllopoda). Annals of the South African Museum. 29: 181-270.

Battish, S. K. 1981. On some conchostracans from Punjab with the description of three new species and a new subspecies. Crustaceana. 40 (2): 178-196.

Bayly, I. 1999. Rock of Ages: Human use and natural history of Australian granites. University of Western Australia Press, Nedlands, Western Australia.

Belk, D. 1989. Identification of species in the conchostracan genus Eulimnadia by egg shell morphology. Journal of Crustacean Biology 9: 115-125.

Benson, D. A., Boguski, M. S., Lipman, D. J. and Ostell, J. 1997. GenBank. Nucleic Acids Research 25 (1): 1-6.

Benvenuto, C. Calabrese, A., Reed, S. K., and Knott, B. 2009. Multiple hatching events in clam shrimp: Implications for mate guarding behaviour and community ecology. Current Science. 96 (1): 130-136.

Berry, E. W. 1926. Description and notes on the life history of a new species of Eulimnadia. American Journal of Science Ser 5, 11: 429-433.

Bond, R. M. 1934. Report on Phyllopod Crustacea (Anostraca, Notostraca, an Conchostraca), in cluding a revision of the Anostraca of the Indian Empire. Mem. Connecticut Academy of Arts and Sciences, 10: 29-62.

Botnariuc, N. & Orghidan, T. 1941. Sur une nouvelle espece due genre Imndaid trouve en Toumanie et sure les Imnadiidea, n. fam. Bull. Sect. Sci. Acad. Roum. 24 (10): 239-246.

Braband, A., S. Richter, R. Hiesel and G. Scholtz. 2002. Phylogenetic relationships within the Phyllopoda (Crustacea, Branchiopoda) based on mitochondrial and nuclear markers. Molecular Phylogenetics and Evolution 25: 229-244.

Brady, G. S. 1886. Notes on freshwater entomostraca from South Australia.

Brauer 1877. Beitrage zur kenntniss der Phyllopoden. S. B. Akad. Wiss. 75: 583-614.

Brehm, V. 1933. Phyllopoden. Mitteilungen von der Wallacea – Expedition Wolterek, 5. Zoologischer Anzeiger. 104: 31-40.

Brendonck, L. and B. J. Riddoch. 1999. Wind-borne short-range egg dispersal in anostracans (Crustacea: Branchiopoda). Biological Journal of the Linnean Society. 67 (1): 87-95.

Brendonck, L. 1996. Diapause, quiescence, hatching requirements: What we can learn from large freshwater branchiopods (Crustacea: Branchiopoda: Anostraca, Notostraca, Conchostraca). Hydrobiologia. 320: 85-97.

Brongniart, A. 1820. Memoire sur le Limnadia, nouveau genre des Crustaces. Mem. Mus. Hist. Nat., Paris.

Brtek, D. 1987. Checklist of the valid and invalid names of the “large branchiopods” (Anostraca, Notostraca, Spinicaudata, and Laevicaudata), with a survey of the taxonomy of all Branchiopoda. Zbornik Slovenskeho Narodneho Muzea Prirodne Vedy. 43: 2-65.Burmeister 1843

Charalabidou, I. & Santamaria, L. 2002. Waterbirds as endozoochorous dispersers of aquatic organisms: a review of experimental evidence. Acta Oecologica. 23: 165- 176.

Charlesworth, D. 1984. Androdioecy and the evolution of dioecy. Biological Journal of the Linnean Society. 23: 333-348.

Chen, P. J. & Shen, Y. B. 1981 Paleogene conchostracan faunas of China. Geological Society of America. Special Paper 187: 193-197.

Crisci, J.V. 2001. The Voice of Historical Biogeography. Journal of Biogeography 28: 157-168.

Daday, E. 1913. Quelques Phyllopodes Anostraces nuveaux. Appendice a la monographie Systematique des Phyllopodes Anostraces, Annales des Sciences Naaturelles, Zoologies S9, 18: 217-218.

Daday, E. 1925. Monographie systematique des Phyllopodes Conchostraces. Troisieme partie. Annales des Sciences Naaturelles, Zoologies 10e serie 8: 143-184.

Dakin, W. J. 1914. Fauna of Western Australia. – II. The Phyllopoda of Western Australia. Proceedings of the Zoological Society of London, 1914: 293-305.

Darwin, C. 1859. On the origin of species by means of natural selection. John Murray, London.

De Stasio, B. T. J. 1989. The seed bank of a freshwater crustacean – Copepodology for the plant ecologist. Ecology. 70: 1377-1389.

Drummond AJ & Rambaut A (2007) "BEAST: Bayesian evolutionary analysis by sampling trees." BMC Evolutionary Biology 7, 214.

Durga Prasad, M.K. and Simhachalam G. 2004. Eulimnadia indocylindrova n. sp. (Branchiopoda: Spinicaudata) from South India with a review of the genus Eulimnadia in Indomalayan region. Proceedings of the International Conference. The Great Himalayas: Climate, Health, Ecology, Management, and Conservation.

Figuerola, J. and Green, A. J. 2002. Dispersal of aquatic organisms by waterbirds: a review of past research and priorities for future studies. Freshwater Biology. 47: 483-494.

Goudet, J. 1995. Fstat v-1.2: A computer program to calculate F-statistics. Journal of Heredity. 86: 485-486.

Green, A. J. and Figuerola, J. 2005. Recent advances in the study of long-distance dispersal of aquatic invertebrates via birds. Biodiversity Research. 11: 149-156.

Gurney, R. 1931. A new species of Branchiopod from southern India. Records of the Indian Museum. 32: 63-64.

Hairston, N. G. and Kearns, C. M. 2002. Temporal Dispersal: Ecological and Evolutionary Aspects of Zooplankton Egg Banks and the Role of Sediment Mixing. Integrative and Comparative Biology. 42: 481-491.

Hamasaki, K. 1999. Occurrence of American Tadpole Shrimp, Triops longicaudatus(LeConte), in Paddy Fields Cultivated under Conventional and Organic Farming Methods. Japanese Journal of Applied Entomology and Zoology. 43 (1): 35-40.

Hanner, R. & Fugate, M. 1997. Branchiopod phylogenetic reconstruction from 12S rDNA sequence data. Journal of Crustacean Biology. 17: 174-183.

Hanski, I. 1998. Metapopulation dynamics. Nature. 396: 41-49.

Hanski, I. 1999. Habitat connectivity, habitat continuity, and metapopulations in dynamic landscapes. Oikos. 87, 209-219.

Harris, A. J. and Xiang, Q. Y. 2009. Estimating ancestral distributions of lineages with uncertain sister groups: a statistical approach to Dispersal-Vicariance Analysis and a case using Aesculus L (Sapindaceae) including fossils. Journal of Systematics and Evolution. 47: 349-368.

Henry, M. 1924. A monograph of the freshwater Entomostraca of New South Wales. Part IV. Phyllopoda. Proceedings of the Linnaen Society, N.S.W. 49: 120-137.

Hertzog, L. 1935. Crustaces. Notes faunistiques de la Camargue, 1. Bulletin of the Zoological Society of France. 60: 265-281.

Hoeh, W. R., N. D. Smallwood, D. M. Senyo, E. G. Chapman, and S. C. Weeks. 2006. Evaluating the monophyly of Eulimnadia and the Limnadiinae (Branchiopoda: Spinicaudata) using DNA sequences. Journal of Crustacean Biology.

Hollenbeck, V. G., S. C. Weeks, W. R. Gould, and N. Zucker. 2002. Maintenance of androdioecy in the freshwater shrimp Eulimnadia texana: sexual encounter rates and outcrossing success. Behavioral Ecology. 13 (4): 561-570.

Huelsnebeck, J. P. and Ronquist, F. 2003. MrBayes: Bayesian inference of phylogeny. Bioinformatics. 17: 754-755.

Ishikawa, G. 1895. Phyllopod Crustacea of Japan. Zool. Mag. 7 (76): 13-21.

King, R. L. 1855 On Australian Entomostracans. Proc. Roy. Soc. Van Diemen's Land, iii (1): 56-75, pls. 9, 10.

Levins, R. 1969. Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America. 15:237–240.

Levins, R. 1970. Extinction. pp. 77–107. In M. Gesternhaber (ed.), Some Mathematical Problems in Biology. American Mathematical Society, Providence, Rhode Island.

Linnaeus, C. 1761. Fauna Seucica. Ed. II. Stockholmiae.

Lloyd, D. G. 1975. The maintenance of gynodioecy and androdioecy in angiosperms. Genetica. 45: 325-339.

Mackin, J. G. 1940. A new species of conchostracan phyllopod, Eulimnadia antlei, from Oklahoma. American Midland Naturalist. 23: 219-221.

Martin, J. W. 1989. Eulimnadia belki, a new clam shrimp from Cosumel, Mexico (Conchostraca: Limnadiidae), with a review of Central and South American species of the genus Eulimnadia. Journal of Crustacean Biology. 9: 104-114.

Martin, J. W., and D. Belk. 1989. Eulimnadia ovilunata and Eulimnadia ovisimilis, new species of clam shrimps (Crustacea, Branchiopoda, Spinicaudata) from South America. Proceedings of the Biological Society of Washington 102: 894-900.

Mattox, N. T. 1937. Studies on the life history of a new species of fairy shrimp Eulimnadia diversa. Transactions of the American Microscopical Society of Menaska. 56: 249-255.

Mattox, N. T. 1939. Description of two new species of the genus Eulimnadia and notes on the other Phyllopoda of Illinois. American Midland Naturalist. 22: 642-653.

Mattox, N. T. 1952. A new genus and species of Limnadiidae from Venezuela (Crustacea: Conchostraca). Journal of the Washington Academy of Sciences 42: 23-26.

Mattox, N. T. 1953. Two new species of Eulimnadia from Maryland and Virginia (Crustacea: Conchostraca). Journal of the Washington Academy of Sciences. 42: 23-26.

Mattox, N. 1954. A new Eulimnadia from the rice fields of Arkansas with a key to the American species of the genus (Conchosraca, Limnadiidae). Turlane Studies in Zoology 2: 3-10.

Mohammad, M. B. M. 1986. Two new species of Conchostraca (Crustacea: Branchiopoda) from Iraq. Hydrobiologia. 140: 165-168.

Moore, W. G. 1965. New distribution records for Conchostraca (Crustacea, Branchiopoda) with an account of their occurrence in Louisiana. Proceedings of the Louisiana Academy of Sciences. 28: 41-44.

Mura, G. 1986. SEM morphological survey of the egg shell in the Italian anostracans (Crustacea, Branchiopoda). Hydrobiologia. 134: 273-286.

Nayar, C. K. G. 1965. Three new species of Conchostraca (Crustacea: Branchiopoda) from Rajasthan. Bulletin of Syst. Zool. (Calcutta). 1: 19-24.

Nayar, C. K. and Nair, K. K. 1968. On a collection of Conchostraca (Crustacea: Branchipoda) from South India, with the description of two new species. Hydrobiologia. 32: 219-224.

Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York.

Novojilov, N. 1958. Recueil d’articles sur les Phyllopodes Conchostraces. Ann. Serv. Inform. Geol. B.R.G.G.M., 26: 1-135.

Nylander, J. A. A., U. Olsson, P. Alstrom, and I. Sanmartin. 2008. Accounting for the phylogenetic uncertainty in biogeography: a Bayesian approach to dispersal- vicariance analysis of the thrushes (Aves: Turdus). Systematic Biology. 57: 257- 268.

Obbard, D. J., S. A. Harris, and J. R. Pannell. 2006. Sexual systems and population genetic structure in an annual plant: testing the metapopulation model. American Naturalist. 167 (3): 354-366.

Otto, S. P., C. Sassaman, and M. W. Feldman. 1993. Evolution of sex determination in the conchostracan shrimp Eulimnadia texana. The American Naturalist. 141: 329- 377.

Packard, A. S. Jr. 1871. Preliminary notice of new North American Phyllopoda. American Journal of Science and Arts, 3rd series. 2 (102): 108-113.

Packard, A. S. 1874. Synopsis of the freshwater phyllopod crustacean of North America. Annual Report of the United States Geological and Geographical Survey of the Territories for 1873. pp. 613-622.

Pannell. J. R. and D. J. Obbard. 2003. Probing the primacy of the patch: what makes a metapopulation? Journal of Ecology. 91 (3): 485-488.

Pannell, J. R. 1997. The maintenance of gynodioecy and androdioecy in a metapopulation. Evolution 51:10-20.

Pannell, J. R. 2000. A hypothesis for the evolution of androdioecy: the joint influence of reproductive assurance and local mate comptetition in a metapopulation. Evolutionary Ecology. 14: 195-211.

Pannell, J. R. 2002. The evolution and maintenance of androdioecy. Annual Review of Ecology and Systematics. 33: 397-425.

Pannell, J. R. and B. Charlesworth. 2000. Effects of metapopulation processes on measures of genetic diversity. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences. 355: 1851-1864.

Pereira, G., and J. V. Garcia. 2001. A review of the clam shrimp family Limnadiidae (Branchiopoda, Conchostraca) from Venezuela, with the description of a new species. Journal of Crustacean Biology. 21: 640-652.

Rabet, N. 2010. Revision of the egg morphology of Eulimnadia (Crustacea, Branchiopoda, Spinicauata). Zoosystema. 32 (3): 373-391.

Radhakrishna, Y. and M. K. Durga-Prasad. 1976. Eulimnadia gunturensis sp. nov. (Branchiopoda, Conchostraca) from South India. Crustaceana. 31 (2); 131-136.

Rammer, W. 1933. Süss-und and Brackwasser-Phylloen von Bonaire. Zool. Jahrb (Syst.). 64: 357-368.

Richardson, B. J., P. R. Baverstock, and M. Adams. 1986. Allozyme Electrophoresis: A handbook for Animal Systematics and Population Studies. Academic Press, New York.

Roen, U. 1952. On some Euphyllopoda from North China. Vidensk. Medd. Fra Sansk nature. foren. 114: 203-215.

Roessler, E. W. 1990. Estudios sobre los entomostraceos de Colombia. III. Estudio taxonomico de una nueva especie Colombiana, Eulimnadia colombiensis (Arthropoda, Crustacea, Conchostraca). Caldasia. 16: 58-65.

Rogers, D. C., Weeks, S. C., and Hoeh, W. R.. 2010. A new species of Eulimnadia (Crustacea: Branchiopoda: Diplostraca: Spinicaudata) from North America. Zootaxa, 2413: 61-68.

Rogers, D. C., Rabet, N. and Weeks, S. C. 2012. Revision of the extant genera of Limnadiidae (Branchiopoda: Spinicaudata). Journal of Crustacean Biology. 32 (5): 827-842.

Ronquist, F. 1996. DIVA, version 1.1. Computer program and manual available by anonymous FTP from Uppsala University (http://www.ebc.uu.se/systzoo/research/diva/diva.html)

Ronquist, F. 1997. Dispersal-vicariance analysis: a new approach to the quantification of historical biogeography. Systematic Biology. 46: 195-203.

Ronquist, F. 2001. DIVA, version 1.2 Computer program for Mac OS and Win32. Evolutionary Biology Centre, Uppsala University (http://www.ebc.uu.se/systzoo/research/diva/diva.html)

Sars, G. O. 1895. Descriptions of some Australian Phyllopoda. Archiv for Mathematik og Naturvidenskab. 17: 1-27.

Sars 1896 1896. Phyllocarida og Phyllopoda. Fauna Norvegiae, 1: i-vii, 1-140.

Sassaman, C. 1995. Sex determination and evolution of unisexuality in the Conchostraca. Hydrobiologia 298:45-65.

Sassaman, C., and S. C. Weeks. 1993. The genetic mechanism of sex determination in the conchostracan shrimp Eulimnadia texana. The American Naturalist. 141: 314- 328.

Sayce, O. A. 1903. The Phyllopoda of Australia, including descriptions of some new genera and species. Proceedings of the Royal Society of Victoria. 15: 224-261.

Scotese, C. R. 2002. Point Tracker. In PALEOMAP project. Arlington, TX.

Schwentener, M. B. V. Timms, R. Bastrop, and S. Richter. 2009. Phylogeny of Spinicaudata (Branchiopoda, Crustacea) based on three molecular markers – An Australian origin for Limnadopsis. Molecular Phylogenetics and Evolution. 53: 716-725.

Simahachalam, G. 2004. Studies on the taxonomy and a few biochemical aspects of clam shrimp species (Branchiopoda: Crustacea) of coastal Andhra Pradesh, India. Ph.D. dissertation. Acharya Nagarjuna University.

Sissom, S. L. 1971. Morphological variation in Eulimnadia texana, Texas’ most common eulimnadian Conchostraca. Texas Journal of Science. 23: 295-297.

Smirnov, S. 1936. Zweiter Beitrag zur Phyllopodenfauna Transkaukasiens. Zool. Anz. 113: 311-320

Smirnov, S. 1949. Novyi vid roda Eulimnadia pack. (Crustacea, Conchostraca) iz Uzebekistana. Dokl. Akad. Nauk SSSR, N. S. 67: 1159-1162

Spears, T., and L. G. Abele. 2000. Branchiopod Monophyly and Interordinal Phylogeny Inferred from18s Ribosomal DNA. Journal of Crustacean Biology 20: 1-24.

Spencer, W. B. and T. S. Hall. 1896. Crustacea report on the work of the Horn scientific expedition to central Australia. 11. Zoology. London (Dulau) and Melbourne. 8: 227-248.

Stamatakis, A. 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics. 22 (21): 2688- 2690.

Straskraba, M. 1965. Taxonomical studies on Czechoslovac Conchostraca. I. Family Limnadiidae. Crustaceana 9: 263-273.

Straskraba, M. 1966. Taxonomical studies on Czechoslovak Conchostraca. III. Family Leptestheriidae, with some remarks on the variability and distribution of Conchostraca an a key to the Middle European species. Hydrobiologia. 27: 571- 589.

Sudhinra, R. G., M. S. Rosenberg and S. Kumar. 2005. Inferring species phylogenies from multiple genes: concatenated sequence tree versus consensus gene tree. Journal of Experimental Zoology (Mol Dev Evol). 304B: 64-74.

Takahashi, F. 1977. Triops spp. (Notostraca: Triopsidae) for the biological control agents of weeds in rice paddies in Japan. Entomophaga 22: 351-357.

Tamura K, Peterson D, Peterson N, Stecher G, Nei M, and Kumar S 2011. MEGA5: Molecular Evolutionary Genetics Analysis using Maximum Likelihood, Evolutionary Distance, and Maximum Parsimony Methods. Molecular Biology and Evolution 28: 2731-2739.

Tasch, P. 1987. Fossil Conchostraca of the Southern Hemisphere and Continental Drift: Paleontology, Biostratigraphy, and Dispersal. The Geological Society of America, Inc. U.S.A.

Tasch, P. & Shaffer, B. L. 1964 Conchostracans: Living and fossil from Chihuahua and Sonora, Mexico. Science. 143: 806-807.

Taver, S. 1986. "Some Probabilistic and Statistical Problems in the Analysis of DNA Sequences". Lectures on Mathematics in the Life Sciences (American Mathematical Society) 17: 57–86.

Timms, B. 2009. A revision o fthe Australian endemic clam shrimp genus Limnadopsis Spencer and Hall (Crustacea: Branchiopoda: Spinicaudata: Limnadiidae). Records of the Australian Museum. 61: 49-72.

Timms, B. and McLay, C. 2005. A new species of Eulimnadia(Crustacea: Spinicaudata: Limnadiidae) from New Zealand. Journal of the Royal Society of New Zealand. 35 (4): 409-415.

Timms, B. and Schwentner, M. 2012. A new genus and species of large Limnadiid clam shrimp from Australia (Spinicaudata: Limnadiidae). Journal of Crustacean Biology. 32 (6): 981-990.

Ueno, M. 1927. Freshwater Branchiopoda of Japan. 1. Memoirs College of Science. Kyoto Imperial University B2: 259-311.

Ueno, M. 1940. Phyllopod Crustacea of Manchoukuo. Bulletin of the Biogeographical Society of Japan. 10 (5): 87-102.

Vandekerkhov, J., G. Louette, L. Brendonck, and L. De Meester. 2005. Development of cladoceran egg banks in new and isolated pools Archives of HydrobiologiaI. 162 (3): 339-347.

Vanschoenwinkel, B., Waterkeyn, A., Nhiwatiwa, T., Pinceel, T., Spooren, E., Geerts,

A., Clegg, B., and Brendonck, L. 2011. Passive external transport of freshwater invertebrates by elephant and other mud-wallowing mammals in an African savannah habitat. Freshwater Biology. 56: 1606-1619.

Vidrine, M. F., S. L. Sissom, and R. E. McLaughn. 1987. Eulimnadia texana Packard (Conchostraca: Limnadiidae) in rice fields in southwestern Louisiana. Southwestern Naturalist. 32: 1-4.

Waterkeyn, A., Vanshoenwinkel, B., Elsen, S. Anton-Pardo, M., Grillas, P., and Brendonck, L. 2010. Unintentional dispersal of aquatic invertebrates via footwear and motor vehicles in a Mediterranean wetland area. Aquatic Conservation: Marine and Freshwater Ecosystems. 20: 580-587.

Webb, J. A., and G. D. Bell. 1979. A new species of Limnadia (Crustacea: Conchostraca) from the Granite Belt in Southern Queensland and Northern New South Wales. Proceedings of the Linnaen Society of New South Wales. 103: 237-245.

Weeks, S. C. 2012. The role of androdioecy and gynodioecy in mediating evolutionary transitions between dioecy and hermaphroditism in the animalia. Evolution. 66 (12): 3670-3686.

Weeks, S. C. and R. L. Bernhardt. 2004. Maintenance of androdioecy in the freshwater shrimp, Eulimnadia texana: field estimates of inbreeding depression and relative male survival. Evolutionary and Ecological Research. 6: 227-242.

Weeks, S. C. and E. G. Chapman, D. C. Rogers, D. M. Senyo and W. R. Hoeh. 2009. Evolutionary transitions among dioecy, androdioecy, and hermaphroditism in limnadiid clam shrimp (Branchiopoda: Spinicaudata). Journal of Evolutionary Biology. 22: 1781-1799.

Weeks, S. C., B. R. Crosser, R. Bennett, M. Gray, and N. Zucker. 2000. Maintenance of androdioecy in the freshwater shrimp Eulimnadia texana: Estimates of inbreeding depression in two populations. Evolution. 54 (3): 878-887.

Weeks, S. C., B. R. Crosser, and M. M. Gray. 2001. Relative fitness of two hermaphroditic mating types in the androdioecious clam shrimp, Eulimnadia texana. Journal of Evolutionary Biology. 14: 83-94.

Weeks, S. C., V. Marcus, and S. Alvarez. 1997. Notes on the life history of the clam shrimp Eulimnadia texana. Hydrobiologia. 359: 191-197.

Weeks, S. C., V. Marcus, and B. R. Crosser. 1999. Inbreeding depression in a self- compatible, androdioecious crustacean, Eulimnadia texana. Evolution. 53 (2): 472-483.

Weeks, S. C., R. T. Posgai, M. Cesari, and F. Scanabissi. 2005. Androdioecy inferred in the clam shrimp Eulimnadia agassizii (Spinicaudata: Limnadiidae). Journal of Crustacean Biology. 25: 323-328.

Weeks, S. C., T. F. Sanderson, S. K. Reed, M. Zofkova, B. Knott, U. Balaraman, G. Pereira, D. M. Senyo, and W. R. Hoeh. 2006. Ancient androdioecy in the freshwater crustacean Eulimnadia. Proceedings of the Royal Society of London Series B-Biological Sciences. 273 (1587): 725-734.

Weeks, S. C., M. Zofkova, and B. Knott. 2006. Limnadiid clam shrimp biogeography in Australia (Crustacea: Branchiopoda: Spinicaudata). Journal of the Royal Society of Western Australia. 89: 155-161.

Weeks, S. C., and N. Zucker. 1999. Rates of inbreeding in the androdioecious clam shrimp Eulimnadia texana. Canadian Journal of Zoology. 77: 1402-1408.

Wright, S. 1951. The genetical structure of populations. Annals of Eugenics. 15: 323- 354.

Wolf, D. E. and Takebayashi, N. 2004. Pollen limitation and the evolution of androdioecy from dioecy. The American Naturalist. 163(1): 122-137.

Yu, Y. A. J. Harris, and X. He. 2010. S-DIVA (Statistical Dispersal-Vicariance Analysis): A tool for inferring biogeographic histories. Molecular Phylogenetics and Evolution. 56: 848-850.

Zaffagnini, F. 1971. Alcune precisazioni sullo sviluppo post-embrionale del concostraco Limnaida lenticularis (L.). Memorie dell’ Istituto Italiano di Idrobiologia. 27: 45- 60.

Zucker, N., M. Cunningham and H. P. Adams. 1997. Anatomical evidence for androdioecy in the clam shrimp Eulimnadia texana. Hydrobiologia. 359: 171-175.

Zucker, N., B. Stafki, and S. C. Weeks. 2001. Maintenance of androdioecy in the freshwater clam shrimp Eulimnadia texana: longevity of males relative to hermaphrodites. Canada Journal of Zoology. 79: 393-401.