Louisiana State University LSU Digital Commons

LSU Doctoral Dissertations Graduate School

2014 Model-based Tests of Historical Demography and Species Delimitation in the Caribbean Coral Reef Sponge Callyspongia Melissa Barrett eBD iasse Louisiana State University and Agricultural and Mechanical College, [email protected]

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_dissertations

Recommended Citation DeBiasse, Melissa Barrett, "Model-based Tests of Historical Demography and Species Delimitation in the Caribbean Coral Reef Sponge Callyspongia" (2014). LSU Doctoral Dissertations. 3540. https://digitalcommons.lsu.edu/gradschool_dissertations/3540

This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Doctoral Dissertations by an authorized graduate school editor of LSU Digital Commons. For more information, please [email protected]. MODEL-BASED TESTS OF HISTORICAL DEMOGRAPHY AND SPECIES DELIMITATION IN THE CARIBBEAN CORAL REEF SPONGE CALLYSPONGIA

A Dissertation

Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy

in

The Department of Biological Sciences

by Melissa Barrett DeBiasse B.Sc., Indiana University, 2003 M.Sc., Nova Southeastern University, 2008 August 2014 Acknowledgements This document is years in the making and the effort of many people. First I would like to thank my advisor Dr. Michael Hellberg. From the very first email he sent in response to my inquiry about the availability of positions in his lab he has shown enthusiasm about my research and my development as a scientist. He gave me the freedom to pursue the projects that interested me but also guidance when my ideas needed to be reigned in. He never hovered but was always available. He has made me an infinitely better writer and helped me to think about the big picture questions that motivate my research. The Louisiana Boards of Regents, National Science Foundation, LSU Department of Biological Sciences, American Museum of Natural History (Lerner Gray Grant), Smithsonian Tropical Research Institute, and the LSU local and national chapters of Sigma Xi funded my research.

My committee members Dr. Robb Brumfield, Dr. Bryan Carstens, Dr. Joseph Neigel, and Dr. Joseph Powers have been incredibly helpful and generous with their time. Robb and Bryan exponentially expanded my knowledge about evolutionary biology through the courses they taught and they were always happy to provide recommendation letters for my grant proposals (some of which were successful!). Joe Neigel provided insightful comments that greatly improved the first chapter of my dissertation and helped lead to its publication. I had the pleasure of working with my outside committee member Joe Powers through the EnvironMentors program at LSU. Volunteering with EnvironMentors was a wonderfully rewarding and I am grateful to have had that experience. Dr. Jeremy Brown provided advice and computational resources for the species delimitation chapter of my dissertation and graciously invited me to attend his lab’s weekly paper discussions. I am also grateful to Dr. Vinson Doyle and Dr. Tom Giarla for their professional advice and friendship.

Dr. Carlos Prada was the best “lab brother” I could have hoped for. Over the last six years, Carlos and I have had countless discussions about ecology and evolution in marine systems. His unique perspective has challenged and expanded the way I think about science from basic, applied, and philosophical points of view. My interactions with Carlos have been integral to my development as a scientist and I consider him to be a valuable collaborator and friend. Tara Pelletier and Sarah Hird have been an incredible support system for me as a scientist generally and as a female scientist specifically. Their professional and personal encouragement helps me believe in myself when the going gets tough. Reid Brennan is my calm in the storm and always brings out the best in me.

I have learned more from interacting with my fellow graduate students than I could have ever imagined was possible. They have provided advice on project ideas, analytical methods, weird results, grant proposals, and manuscripts. They have also been incredible friends and helped to create an environment of hard work and fun that makes the LSU SEE division so special: Dr. Lori Patrick, Dr. Noah Reid, Dr. Eric Rittmeyer, and Cesar Sanchez (members of my cohort), Dr. Verity Mathis, Dr. Meche Gavilanez, Caroline Judy (my running and crossfit buddies), Dr. Ron Eytan, Dr. Alice Dennis, Karine Posbic (past and present Hellholers), Mike Harvey, Clare Brown, Jordan Satler, Dr. John McVay, Cathy Newman, Glenn Seeholzer, John Mittermeier, Ryan Terrill, Matt Brady,

ii Vivian Chua,Valerie Durouen, Bill Ludt, Caleb McMahon, and Maria Vozzo (great colleagues and friends).

Finally I would like to thank my parents Tony and Martha DeBiasse and my sister Jaime DeBiasse for their unconditional support. They never questioned my career path and lived my (momentary) defeats and triumphs as if they were their own. I would not have made it this far had it not been for them. I dedicate my dissertation to the memory of my grandparents, Malcolm and Dorothy Watson and Anthony and Joanne DeBiasse. Their love and support stays with me and I think of them always.

iii Table of Contents Acknowledgements……………………………………………………………………...... ii

Abstract………………………………………………………………...……………....….v

Chapter 1. Introduction……………………………………………………………………1

Chapter 2. Evaluating Summary Statistics Used to Test for Incomplete Lineage Sorting: Mito-nuclear Discordance in the Coral Reef Sponge Callyspongia vaginalis………………………………………………………………………………...…4

Chapter 3. Phylogeography of the Coral Reef Sponge Callyspongia vaginalis Across the Caribbean……...………………………………………………….………….22

Chapter 4. Species Delimitation in the Caribbean Coral Reef Sponge Genus Callyspongia……………………………………………………………...... ……………38

Chapter 5. Conclusions………………………………………………………..…………51

References………………………………………………………………………………..53

Appendix A. Supplementary Text, Tables, and Figures for Chapter 2….…...…………..73

Appendix B. Supplementary Tables and Figures for Chapter 3……………………..…..87

Appendix C. Supplementary Tables and Figures for Chapter 4…………..…………....106

Appendix D. Permission to Reproduce Chapter 2……………………………………...111

Vita………………………………………………………………..……………….……118

iv

Abstract Coral reefs are the most productive and species rich ecosystems in the ocean yet we lack knowledge about the distribution of genetic variation the within and among reef species, particularly for the sponges (Porifera).

My dissertation describes how genetic variation at mitochondrial and nuclear genes is partitioned among and within species in the sponge genus Callyspongia. I compared patterns of genetic diversity and population subdivision in the mitochondrial and nuclear genomes of one species, C. vaginalis, in Florida (Chapter 2). Previous work revealed three divergent mitochondrial lineages, but nuclear alleles did not correspond to either mitochondrial clade or geography. Coalescent simulations showed mito-nuclear discordance was not the result of incomplete lineage sorting. Instead, patterns in mitochondrial and nuclear DNA were consistent with changes in population size and sperm-mediated gene flow.

Across the Caribbean, I found subdivision in C. vaginalis for mitochondrial and nuclear DNA was concordant, suggesting geographic features and habitat discontinuity are important for structuring populations at large spatial scales (Chapter 3). Clustering analyses found C. vaginalis populations were divided west-east and model-based tests of species boundaries supported a cryptic lineage in Central America. Phylogeographic patterns for three invertebrate sponge commensals also showed a west-east split between Florida and the Bahamas.

I tested for mismatches between morphological and molecular species boundaries for seven Caribbean Callyspongia species (Chapter 4). Genetic distances calculated within and among species support C. fallax, C. tenerrima, and C. plicifera as distinct species. However, C. armigera, C. longissima, C. ?eschrichtii and C. vaginalis shared alleles across loci and genetic distances among these taxa overlapped distances within them. Model-based species delimitation supported the hypothesis that these latter four taxa represent one evolutionarily significant unit.

This dissertation demonstrates that in the common reef sponge Callyspongia vaginalis, demographic processes and geography influence population structure at small and large spatial scales, respectively, and genetic markers from different genomes can show contrasting patterns. My work also shows the relationship between morphology and evolutionary history is not straightforward in sponges and points to the importance of inter- and intraspecific genetic data for a thorough documentation of biodiversity in marine invertebrates.

v Chapter 1. Introduction

Coral reefs are the most species rich habitats in the ocean (Odum & Odum 1955; Connell 1978) and play an important role in global ecosystem function (Mumby et al. 2008). Coral reefs act as breeding and nursery areas for a multitude of taxa (Nagelkerken et al. 2000). The heterogeneity and complex three-dimensional structure of coral reefs creates habitat and promotes species diversification (Paulay 1997). Coral reefs connect other important marine ecosystems and facilitate movement of species, nutrients, and energy among sea grass beds, mangroves, and the open ocean (Moberg & Folke 1999). Coral reefs also provide extensive services and goods to human society (Hoegh-Guldberg & Bruno 2010). For example, they reduce coastal erosion by buffering waves and abate storm surge during hurricanes (White et al. 2000). Fish harvested from the ocean provide the main source of protein for many humans populations around and the world and generate more than US$95 billion dollars of commercial and recreational revenue annually (Sumaila et al. 2007). Coral reef invertebrates are an important source for chemical compounds of pharmaceutical interest (Leal et al. 2012).

Like many other terrestrial and aquatic ecosystems, coral reefs are threatened by anthropogenic degradation. Every marine habitat on the planet has been affected by human influences and a large fraction (41%) are impacted by multiple factors (Halpern et al. 2008). For example, the removal of apex predators through overfishing has cascading affects across coral reef ecosystems and results in the loss of the coral and coralline algae species that grow the reef (Dulvy et al. 2004). Atmospheric carbon dioxide (CO2) levels have risen at an unprecedented rate in the last 250 years (Doney & Schimel 2007) and the resulting global temperature increase has been implicated in coral reef bleaching events and disease outbreaks (Glynn 1996; Pandolfi et al. 2003). Because the world’s oceans act as a sink for ever increasing atmospheric CO2, in addition to becoming warmer, oceans are becoming more acidic (Doney et al. 2009). Subsequent shifts in seawater chemistry and alterations in biogeochemical nutrient cycles will especially impact benthic organisms, such as corals, echinoderms, and molluscs, that use calcium carbonate to form their skeletons and shells (Orr et al. 2005). Particularly concerning, but perhaps not surprising, is evidence that biodiversity loss impairs the recovery or marine ecosystems from subsequent disturbances (Worm et al. 2006).

Accurate assessment of biodiversity at the species and population levels is invaluable to conservation and management plans for coral reef ecosystems (Turner et al. 2003). However, quantifying this diversity is difficult due to the simple and plastic morphological characteristics of many coral reef organisms. This is particularly true for the major reef building taxa: corals, coralline algae, and sponges. Several studies have found incongruence between morphological taxonomy and genetic lineages, with taxa being overly split, conservatively lumped, or a combination of the two (Klautau et al. 1999; Miller et al. 2001; van der Strate et al. 2002; Duran & Rützler 2006; Erwin & Thacker 2007; Forsman et al. 2010; Xavier et al. 2010; Pinzon & LaJeunesse 2011; Richards et al. 2012; Prada et al. 2014a). Furthermore, rates of nucleotide substitution in the mitochondrial genome of basal metazoans are up to 50- to 100-times slower than in

1 bilateral animals (Shearer et al. 2002; Hellberg 2006), which complicates the identification of suitable genetic markers for molecular species delimitation in sponges and corals. Accurate assessments of coral reef species richness and an understanding of the processes that influence patterns of genetic diversity and structure require a robust identification of species boundaries. This task may be best undertaken by integrating multiple approaches: morphological, molecular, biochemical, and ecological.

Members of the phylum Porifera are among the most diverse taxa on coral reefs, with some data indicating sponge biomass may surpass that of corals and algae in the Caribbean (Rützler 1978). Sponges promote reef species richness by providing refuge for many commensal invertebrates and fish, particularly during critical juvenile or reproductive life history phases (Ribeiro et al. 2003), and harbor a substantial biomass of diverse microbial endosymbionts, many of which produce secondary metabolites of ecological and pharmacological importance (Taylor et al. 2007). As early metazoans, sponges are important models of animal complexity and development (Philippe et al. 2009; Srivastava et al. 2010; Ludeman et al. 2014). Despite their ecological, evolutionary, and pharmaceutical importance, few studies have investigated lower level relationships in sponges. Currently, species boundaries among sponges are poorly defined, obscuring the processes involved in the origin and maintenance of their biodiversity.

This dissertation aims to describe the distribution of genetic and morphological variation among species of the sponge genus Callyspongia and among populations at small and large spatial scales in the species C. vaginalis. Sponges in this genus display a wide range of morphological variation within and among species and most species lack genetic data at the population and species levels. Callyspongia vaginalis is the most abundant Callyspongia species in Florida, the Bahamas, and the Caribbean and is host to three invertebrate taxa, two amphipods, Leucothoe ashleyae and L. kensleyi, and a brittle star, Ophiothrix lineata, that live commensally within the sponge. Callyspongia vaginalis is hermaphroditic and broods its larvae to an advanced developmental stage (Maldonado 2006), such that they are competent to settle immediately after release and dispersal distance is limited (Lindquist et al. 1997). The extensive morphological variation, abundant biomass, ecological importance, and dearth of genetic data for Callyspongia species make this genus an excellent model in which to test hypotheses about population subdivision and species boundaries in coral reef invertebrates.

The first research chapter of my dissertation focuses on genetic structure among populations of C. vaginalis at a small spatial scale along the Florida reef tract. I found contrasting levels of genetic diversity and subdivision in the mitochondrial and nuclear genome - mitochondrial DNA (mtDNA) lineages were divergent and geographically subdivided while nuclear DNA (nucDNA) was admixed with respect to the mtDNA and geography. I used coalescent simulations in a statistical framework to test if this discordance was due to incomplete lineage sorting (ILS) of the nucDNA with respect to the mtDNA and evaluated five common test statistics for their power to distinguish among simulated datasets with various levels of ILS (Chapter 2). Next, I expanded the geographic scale of my sampling to test patterns of mitochondrial and nuclear subdivision

2 across the range on C. vaginalis in the Caribbean. I also compared patterns of phylogeographic structure in C. vaginalis to those in its amphipod and brittle star commensals. Results from a Bayesian clustering analysis suggested the presence of a cryptic species within C. vaginalis in Central America. I used model-based species delimitation methods to test this hypothesis (Chapter 3). Motivated by the phenotypic variation present within and among species in Callyspongia and evidence of cryptic species in C. vaginalis, I used a model-based approach to test molecular species boundaries for seven Caribbean Callyspongia species (Chapter 4). Finally, I synthesize the results from all of my research chapters and discuss the implications of my findings for our understanding of the origin and maintenance of biodiversity in coral reef ecosystems (Chapter 5).

3 Chapter 2. Evaluating Summary Statistics Used to Test for Incomplete Lineage Sorting: Mito-nuclear Discordance in the Coral Reef Sponge Callyspongia vaginalis*

2.1. INTRODUCTION

The number of studies finding conflicting phylogeographic patterns between mitochondrial and nuclear genomes (mito-nuclear discordance) has increased in the last three decades as it has become easier to generate multi-locus datasets (Fig. 3 in Toews & Brelsford 2012). Mito-nuclear discordance may be expected due to the intrinsic characteristics of each genome. The mitochondrial DNA (mtDNA) of most animals is generally free of recombination (Birky 2001), haploid, and maternally inherited, while nuclear DNA (nucDNA) recombines, is diploid, and is transmitted by both parents. Mitochondrial DNA thus has a four-fold smaller effective population size than the nuclear genome in taxa with separate sexes. Mitochondrial genes are generally expected to undergo lineage sorting faster than nuclear genes, as the rate of sorting is inversely proportional to effective size (Funk & Omland 2003; Zink & Barrowclough 2008), and may provide signals of population differentiation when nucDNA does not. Mito-nuclear discordance may also arise as a function of the difference in rates of nucleotide substitution between the mtDNA and nucDNA. Typically, nucDNA has lower rates of nucleotide substitution compared to mtDNA (Brown et al. 1979) and in some cases, variation in the nucDNA may be insufficient to reveal phylogeographic patterns observed in the mtDNA (Hare 2001; Hickerson & Cunningham 2005).

Most studies finding structure in the mtDNA but not nucDNA have corresponding historical or biogeographic evidence that suggests mtDNA haplotypes diverged in allopatry (Toews & Brelsford 2012). In these cases, the absence of complimentary genetic divergence in the nucDNA is often explained by incomplete lineage sorting (ILS) of the nucDNA (Zink & Barrowclough 2008; McKay & Zink 2010). However, mito- nuclear discordance without strong barriers to gene flow can arise due to selection (Cheviron & Brumfield 2009), nonrandom mating (Pardini et al. 2001), or differences in substitution rates (Orozco-Terwengel et al. 2008; Eytan & Hellberg 2010), among other mechanisms (Irwin 2002).

Coalescent simulations have been used to test whether mito-nuclear discordance results from ILS. In this approach, nucDNA are simulated under the demographic history inferred from the mtDNA and the distribution of a summary statistic calculated for the simulated data is constructed. This null distribution represents the range of values that would be expected in the nucDNA given the differences in effective population size (and ______*This chapter previously appeared as: DeBiasse M, Nelson B, Hellberg M. 2014. Evaluating summary statistics used to test for incomplete lineage sorting: mito-nuclear discordance in the reef sponge Callyspongia vaginalis. Molecular Ecology, 23 (1): 225- 238. Reproduced with publisher’s permission.

4 therefore sorting time) of the nuclear and mitochondrial genomes. The hypothesis of ILS is rejected if the test statistic of the empirical nuclear data falls outside of the expected distribution. For example, Peters et al. (2012) rejected ILS in favor of male-biased dispersal in a sea duck based on observed and simulated ΦST values. Wilson and Eigenmann Veraguth (2010) failed to reject ILS in a pipefish when FST and DEST values for the empirical nucDNA sequence data fell within their respective null distributions and concluded that mito-nuclear discordance was due to the stochasticity of the coalescent. Wilson and Eigenmann Veraguth (2010) and Peters et al. (2012) used summary statistics that describe how genetic variation is partitioned among populations. However, it is possible that some summary statistics might not fully capture the signal of ILS in coalescent genealogies. To my knowledge, no one has evaluated the performance of different statistics to compare empirical and simulated data when testing drivers of mito- nuclear discordance.

Sponges offer a novel perspective on the mechanisms underlying mito-nuclear discordance due to their reproductive biology and attributes of their mtDNA that differ from those of most animals. Nucleotide substitution rates among some basal metazoans (including the Porifera and Anthozoa) are up to 100 times slower than in most bilateral animals (Shearer et al. 2002; Hellberg 2006) and studies surveying COI sequence variation in the Porifera have found little to no intraspecific variation over distances up to 20,000 km (Duran et al. 2004c; Wörheide 2006; Whalan et al. 2008; López-Legentil & Pawlik 2009; Duminil et al. 2011). In corals, slow mtDNA rates are not matched by corresponding slow rates for nucDNA (Eytan et al. 2009) and the pattern of fast mtDNA evolution compared to nucDNA seen in bilateral animals (Brown et al. 1979) is reversed in corals (Hellberg 2006) as it is in plants (Wolfe et al. 1987). Rates for mtDNA and nucDNA genes are lacking for sponges, but Duran et al. (2004a; 2004c) found higher sequence variation for multi-copy nuclear ITS than for COI in the Mediterranean sponge Crambe crambe. More equitable rates of nucleotide substitution in the sponge nucDNA should reduce mito-nuclear discordance caused by difference in the resolving power of the mtDNA and nucDNA.

Although factors such as mating system and variation in reproductive success can alter effective population size (Nunney 1993), the hermaphroditic reproductive strategy in most sponges (Bergquist 1978) allows every individual to pass on a mitochondrial genome, and should ideally make the ratio of the effective size of the mtDNA to the nucDNA 1:2 as opposed to 1:4 as in gonochoristic taxa. The reduction in the difference between effective sizes of the two genomes may decrease the likelihood of discordance due to ILS, thus making sponges a useful model to reveal other drivers of mito-nuclear discordance.

Callyspongia vaginalis is a common sponge that occurs on coral reefs throughout Florida and the Caribbean (Humann & DeLoach 2003) and broods larvae that are competent to settle almost immediately after release (Maldonado 2006). Sequence data from the mitochondrial COI gene revealed significant genetic subdivision among populations of C. vaginalis collected from 465 km of the Florida reef tract (ΦST = 0.33, p<0.0001, DeBiasse et al. 2010). Most sampled sponges possessed one of three divergent mitochondrial

5 haplotypes, each separated by an average of 12 mutational steps (2.4% sequence divergence). Preliminary COI data from other locations show the three major Florida haplotypes are more divergent from each other than they are from any other haplotype sampled in C. vaginalis across the Caribbean.

Given the low rates of mtDNA evolution in sponges and problems associated with drawing phylogeographic conclusions based on one locus, we collected sequence data from six novel, nuclear protein-coding genes to test for concordant patterns of genetic diversity and subdivision between the mtDNA and nucDNA in C. vaginalis in Florida. In contrast to the patterns observed in the mtDNA, nucDNA sequences showed no geographic subdivision nor differentiation corresponding to the three mitochondrial clades. Motivated by these findings, I tested for selection on the mtDNA and used coalescent simulations to test the hypotheses that the mito-nuclear discordance in C. vaginalis is due to ILS of the nucDNA and changes in population size that might differentially affect the mtDNA and nucDNA. I also evaluated five commonly used summary statistics for their ability to reveal discordance.

2.2. METHODS

2.2.1. Geographic sampling and development of novel nuclear markers

My aims were to determine how variation in the nucDNA was partitioned within and among populations of C. vaginalis and how well any patterns of structure and/or sequence divergence in nucDNA corresponded to those observed the mtDNA. Therefore, we randomly subsampled 61 individuals from DeBiasse et al. (2010) from the three mtDNA lineages in each of locations where the lineages occurred sympatrically (Figure 2.1) and generated nucDNA sequences for each individual. Genetic resources for sponges are lacking compared to other taxa, particularly at the species level. Therefore, nuclear markers for this study were generated de novo from a C. vaginalis cDNA library following the protocol of Hale et al. (2009) using 454 sequencing technology. Details about the development of nuclear markers, PCR amplification, and sequencing are in Appendix A.

Primers for six nuclear genes showed consistent PCR and sequencing results and had sequence variation among individuals (Appendix A, Table 1). Some genes failed to amplify or sequence for some individuals (8 for cata; 3 for cps; 5 for ef). As appropriate, analyses were performed on datasets including all individuals for three sequenced genes (cir, fil, mep), only the individuals sequenced for all six genes, and/or including individuals with missing data. Sequences generated here are available from the European Nucleotide Archive (accession numbers HG000670 – HG001246).

2.2.2. Phasing of alleles and tests for recombination

I resolved alleles in heterozygous individuals using PHASE v2.1 (Stephens et al. 2001). Individuals heterozygous for one insertion/deletion were resolved using CHAMPURU v1.0 (Flot 2007). Individuals with alleles that could not be phased probabilistically to a

6 probability > 90% were cloned using the Invitrogen TOPO TA kit. At least eight clones per reaction were sequenced. Individuals that could not be resolved after several rounds of cloning and phasing (1 for cata; 3 for cps; 4 for ef) were removed from the dataset. Each gene region was tested for intra-locus recombination using GARD and SBP implemented in Hy-Phy (Pond & Frost 2005; Pond et al. 2006) and with the DSS method implemented in TOPALi v2.0 (Milne et al. 2004). Recombination was not detected in any gene region.

1

2

3

Key Largo

Haplotype 1: n = 7 Long Key Haplotype 2: n = 7 Dry Tortugas Marquesas Keys Haplotype 3: n = 3

Haplotype 1: n = 5 Haplotype 2: n = 7 Haplotype 1: n = 6 Haplotype 1: n = 6 Haplotype 3: n = 2 Haplotype 2: n = 8 Haplotype 2: n = 2 Haplotype 3: n = 2 Haplotype 3: n = 6

Figure 2.1. Distribution of C. vaginalis COI haplotypes along the Florida reef tract in four locations sampled in DeBiasse et al. (2010). The three most common haplotypes are represented in yellow, light orange, and dark orange. Less frequent haplotypes are in white. The area of each sector is proportional to haplotype frequency at each site. Sample sizes listed below each pie chart show the number of individuals of each haplotype for which nucDNA sequences were collected. Inset shows haplotype network with common haplotypes 1, 2, and 3 represented by the colors dark orange, light orange, and yellow, respectively.

2.2.3. Analysis of nuclear sequence data

I used three approaches to determine the distribution of nucDNA genetic variation in C. vaginalis according to the geographic location and mitochondrial lineage from which each individual was sampled. To visualize the relationships among nucDNA sequences, I constructed allelic networks for each nuclear locus using TCS v1.21 (Clement et al. 2000). These analyses were conducted using the default settings and resulted in the most parsimonious connections among alleles at the 95% confidence level. STRUCTURE v2.3.2 (Pritchard et al. 2000) was used to infer genetic subdivision for the nucDNA. First,

7 nucDNA sequences were recoded into frequency data using DnaSP v5.0 (Rozas et al. 2003; Librado & Rozas 2009) and individuals were genotyped. No individuals shared a multi-locus genotype meaning subdivision measures cannot be skewed by clonal reproduction. I used the admixed ancestry model with correlated allele frequencies and set the number of clusters, “K,” to 1 – 8, 8 being twice the number of geographic locations. I ran the program with 1 million MCMC steps and discarded the first 10,000 steps as burnin. I performed 20 iterations for each K. I used the Evanno method (Evanno et al. 2005) implemented in STRUCTURE HARVESTOR (Earl & vonHoldt 2011) to determine the most likely number of clusters in the data. Finally, I used a hierarchical AMOVA (Excoffier et al. 1992) implemented in Arlequin (Excoffier et al. 2005) (sequence data) and Genodive (Meirmans & Van Tienderen 2004) (frequency data) to determine the partitioning of genetic variation when individuals were grouped according to geographic location or mitochondrial lineage. The AMOVAs were performed on datasets consisting of only individuals sequenced for all six genes and on datasets consisting of all individuals for three genes (cir, fil, mep).

To compare genetic diversity among nuclear genes and between mitochondrial and nuclear genes we calculated summary statistics in DnaSP. To test for selection in COI, I performed the McDonald-Kreitman (MK) (McDonald & Kreitman 1991) and Hudson- Kreitman-Aguadé (HKA) (Hudson et al. 1987) tests using COI sequences from Niphates digitalis (GenBank accession numbers EF519655-8), the sister genus to Callyspongia (Erpenbeck et al. 2007), as the outgroup. Tests for selection that do not rely on allelic distributions, such as the MK and HKA tests, are robust to potentially confounding non- equilibrium processes, such as demographic change and recombination (Hudson et al. 1987; Sawyer & Hartl 1992; Nielsen 2001; Eyre-Walker 2002). I also calculated Tajima’s D (Tajima 1989b), although cognizant that changes in population size can affect the outcome of this test (Tajima 1989a).

2.2.4. Hypothesis testing of population processes in a coalescent framework

In contrast to my expectation, nucDNA alleles did not cluster according to mtDNA lineage or sampling site: nucDNA was admixed with respect to both and had a different Tajima’s D than mtDNA (see Results). To test (non-mutually exclusive) hypotheses for explaining mito-nuclear discordance in C. vaginalis, I used two approaches employing coalescent simulations in a statistical framework. First, I tested whether ILS could explain why mtDNA haplotypes cluster in divergent clades while nucDNA remains admixed using a parametric bootstrapping approach. Second, I tested whether changes in population size could simultaneously produce significant and nonsignificant Tajima’s D values in the mtDNA and nucDNA, respectively.

2.2.5. Coalescent simulations of lineage sorting in the nucDNA

I used the following procedure to test for ILS of nucDNA in C. vaginalis (see Appendix A, Figure 3 for an example). I simulated 1000 nuclear coalescent gene trees matching the demographic history of COI in C. vaginalis using ms (Hudson 2002). I then simulated

8 sequence data with the same model of nucleotide evolution (estimated in jModelTest v2.0.2 (Posada 2008)) for each empirical nuclear gene on the 1000 coalescent trees using seq-gen (Rambaut & Grassly 1997). Gene trees were estimated for each simulated nuclear dataset in PAUP* (Swofford 2003). I calculated the summary statistics theta, nucleotide diversity, and ΦST for the simulated nuclear sequence data using the R package pegas (Paradis 2010) and calculated the number of deep coalescent events (DCEs) and Slatkin and Maddison’s s for simulated nuclear gene trees compared to the mtDNA tree in Mesquite v2.74 (Maddison & Maddison 2010). I calculated the same suite of summary statistics for the empirical nuclear data and compared the empirical values to the distribution of each summary statistic calculated for the simulated data (i.e. the null distributions).

Because different statistics summarize data in different ways, I may expect variation in their performance when comparing null and simulated data in tests for ILS. For example, ΦST describes how variation is structured within and among populations. Slatkin and Madidison’s s and the number of DCEs document the discordance between an individual’s position in a gene tree and another feature of the data- in the case of s, the geographic location, and in the case of the DCE, the species/population tree. In contrast, theta and nucleotide diversity are determined by variation in the sequences themselves, not by how variation is partitioned in geographic or genealogical space. I expect ΦST, DCEs, and s will be better able to reveal ILS than nucleotide diversity and theta because their values should better trace the accumulation of sequence differences among populations (ΦST) and lineages (DCEs and s). I test this prediction here.

The estimation of parameters from the mtDNA used in ms (effective population sizes, divergence times) to simulate coalescent trees requires mitochondrial and nuclear substitution rates. Although slow substitution rates are well documented for sponge mtDNA, less is known about specific rates of nucleotide substitution for either mitochondrial or nuclear genes. Therefore, I identified two rates encompassing the upper and lower range of mitochondrial evolution for the best-studied and most closely related animal clade with slow mtDNA, the anthozoans (lower rate: 5.0x10-10 substitutions/site/year (Hellberg 2006); upper rate: 1.0x10-9 substitutions/site/year (Romano & Palumbi 1996)), that likely encompass the similarly slow mtDNA rates in sponges (Schröder et al. 2003; Duran et al. 2004a; Erpenbeck et al. 2006; Wörheide 2006). I used the lower rate as reported by Hellberg (2006) and increased the fast rate to 5.0x10-9 to make the difference between rates one order of magnitude. Although uncertainty in substitution rates leads to variance around parameter estimates, here it works to my advantage because I am less interested in knowing specific population parameters in C vaginalis and more interested in creating a broad distribution of values that will encompass the true parameters. Using a wide range of values makes the test of ILS more conservative and my ability to reject the null hypothesis more robust.

To obtain substitution rates for each nuclear gene, I estimated their rates relative to the C. vaginalis COI gene in BEAST v1.7.1 (Heled & Drummond 2008) and multiplied the relative rate by the fast and slow mitochondrial substitution rates. I estimated parameters for the demographic history of the COI gene (lineage divergence times and effective

9 sizes: Appendix A, Figure 3, Table 2), which are required for the simulations in the parametric bootstrap, from phylogenies constructed in BEAST and *BEAST v1.7.1 (Heled & Drummond 2010). For each run, two independent MCMC analyses were conducted for 10 million (BEAST) or 50 million (*BEAST) steps, sampling every 1000 steps. Convergence was determined by viewing the log files in Tracer v1.5. All parameters had effective sample sizes (ESS) greater than 300. Treefiles were combined in LogCombiner v1.7.1 with a 10% burnin and the maximum clade credibility (MCC) tree for the combined file was calculated in TreeAnnotator v1.7.1. Specific details about the BEAST settings are available in Appendix A.

To determine which summary statistics were useful for detecting ILS, I calculated the statistics mentioned above for datasets simulated with various population divergence times, thus producing trees with varying levels of lineage sorting (Rosenberg 2003). I simulated five sets of 1000 coalescent trees with divergences of 0.1N, 0.5N, 1.0N, 5N and 10N generations in ms and simulated sequence data onto those trees in seq-gen. I calculated nucleotide diversity, theta, and ΦST in R and DCEs and s in Mesquite as described above.

2.2.6. Coalescent simulations of demographic change

Fay and Wu (1999) showed with coalescent simulations that the discordance in Tajima’s D for humans (negative for the mtDNA and positive for the nucDNA) could be explained by a recent population expansion after a bottleneck. Because mtDNA has a smaller effective population size than nucDNA, it responds more quickly to population expansion after a contraction and will achieve a negative Tajima’s D while the value for the nucDNA remains positive. In C. vaginalis, I found Tajima’s D was positive for the mtDNA and not different from zero for the nucDNA (Table 2.1), consistent with the early stages of a bottleneck (Fig. 2 in Fay & Wu 1999).

Table 2.1. Genetic diversity indices for Callyspongia vaginalis for each gene by sampling location. n, sample size (alleles); bp, base pairs; S, segregating sites; h, haplotypes; Hd, haplotype diversity; π, nucleotide diversity. Underlined Tajima’s D values are significant ______

Location n bp S h Hd π Tajima’s D ______mtCOI Key Largo 17 511 18 3 0.669 0.0143 2.1103 Long Key 14 511 18 3 0.641 0.0137 1.6144 Marquesas Keys 14 511 18 3 0.659 0.0150 2.1341 Dry Tortugas 16 511 18 3 0.633 0.0133 1.6176 All Locations 61 511 18 3 0.656 0.0142 3.3527

10 (Table 2.1. continued) ______

Location n bp S h Hd π Tajima’s D ______cata Key Largo 30 346 13 10 0.770 0.0099 -0.0765 Long Key 26 346 13 8 0.776 0.0077 -0.7390 Marquesas Keys 20 346 17 10 0.758 0.0131 -0.1949 Dry Tortugas 30 346 13 9 0.851 0.0095 0.0175 All Locations 106 346 20 22 0.802 0.0103 -0.1982 cir Key Largo 34 159 7 6 0.729 0.0090 -0.0748 Long Key 28 159 7 6 0.635 0.0064 -0.9755 Marquesas Keys 28 159 7 6 0.730 0.0094 -0.5035 Dry Tortugas 32 159 10 9 0.833 0.0121 -0.4323 All Locations 122 159 11 10 0.742 0.0095 -0.4758 cps Key Largo 34 113 7 6 0.754 0.0261 2.0813 Long Key 26 113 7 7 0.763 0.0277 2.1556 Marquesas Keys 28 113 7 6 0.733 0.0231 -0.6803 Dry Tortugas 28 113 7 7 0.741 0.0255 1.8119 All Locations 116 113 8 10 0.758 0.0240 1.9038 ef Key Largo 28 161 3 6 0.726 0.0064 0.8728 Long Key 26 161 6 8 0.843 0.0110 0.3764 Marquesas Keys 26 161 7 6 0.794 0.0145 0.8309 Dry Tortugas 28 161 6 6 0.698 0.0071 -0.7663 All Locations 112 161 10 13 0.808 0.0108 -0.1935 fil Key Largo 34 127 12 14 0.884 0.0240 0.7571 Long Key 28 127 7 9 0.839 0.0256 2.4217 Marquesas Keys 28 127 10 11 0.749 0.0199 -0.0509 Dry Tortugas 32 127 10 8 0.792 0.0234 0.6120 All Locations 122 127 16 22 0.840 0.0251 0.5938 mep Key Largo 34 91 10 8 0.816 0.0351 0.9391 Long Key 28 91 10 8 0.812 0.0297 0.1677 Marquesas Keys 28 91 9 6 0.759 0.0274 0.2448 Dry Tortugas 32 91 9 7 0.786 0.0201 -0.5622

11 (Table 2.1. continued) ______

Location n bp S h Hd π Tajima’s D ______

All Locations 122 91 10 11 0.835 0.0204 1.0192 ______

However, the difference in the effective population size of mtDNA and nucDNA is smaller in C. vaginalis than in humans (1:2 versus 1:4) and the results of Fay and Wu (1999) may not apply to our system. We used coalescent simulations to test if we could recover higher Tajima’s D values in the mtDNA than nucDNA in the early stages of a bottleneck given the ratio of population effective sizes in C. vaginalis. I used ms to simulate population contractions of various reductions and timescales (Table 2.3). Initial population sizes were scaled to reflect the ratio of the effective population size of the mtDNA and nucDNA in C. vaginalis and in a typical gonochoric species such as humans. Each simulation was run for 5000 replicates. Sample_stats (Hudson 2002) was used to calculate Tajima’s D for each replicate.

2.3. RESULTS

2.3.1. Genetic diversity and distribution of genetic variation in nuclear loci

Nuclear loci were more diverse than COI in terms of haplotype number and nucleotide diversity (Table 2.1). Cata and fil had the highest number of alleles (22) and fil had the highest nucleotide diversity (0.025) Results for the MK (G = 0.348, p = 0.555) and HKA (p = 0.944) tests for selection were nonsignificant. Tajima’s D values were significantly positive for COI for Key Largo and the Marquesas Keys and for all locations combined (Table 2.1). Tajima’s D was nonsignificant for each sampling location individually and for all locations combined for all nuclear genes, with the exception of cps in Key Largo and Long Key and fil in Long Key (Table 2.1). Substitution rates for nuclear genes were faster relative to COI (cata, 2.3 times faster; cir, 1.6; cps, 2.7; ef, 3.0; fil, 7.1; mep, 3.6).

All methods used to determine how nucDNA variation was partitioned showed nuclear alleles correspond to neither mitochondrial lineage nor geography. TCS networks for each nuclear gene joined all alleles at the 95% confidence level and most networks had equally parsimonious connections among alleles. The network for each nucDNA locus also showed that alleles did not cluster according to mitochondrial lineage or geography (Figure 2.2).

The Evanno et al. (2005) method concluded that the most likely number of clusters in the nucDNA was 2, but the clusters did not correspond to either mitochondrial lineage or geographic location (Figure 2.3). Because the Evanno et al. (2005) method determines the most likely K based on the second order rate of change of the likelihood function with respect to K, it is unable to find the best K if K is 1. In the plot for the mean likelihood of

12

cata cir cps

ef fil mep

Key Largo Long Key Marquesas Dry Tortugas

cata cir cps

ef fil mep Haplotype 1 Haplotype 2 Haplotype 3 Figure 2.2. Unrooted statistical parsimony networks for nuclear genes, with geographic locations (top) and mitochondrial lineage (bottom) indicated. Circles represent individual alleles with circle size proportional to frequency of occurrence. Lines connecting alleles represent one mutational step and small black circles represent possible, but not sampled, alleles. The area of each sector or circle is proportional to haplotype frequency at that site.

K, K = 1 had the best likelihood score and the likelihood for K = 2 through 8 declined from there (Appendix A, Figure 1). This suggests that the true K for the nucDNA in Florida is 1. All hierarchical AMOVAs for the nucDNA indicated that most genetic variation (>90%) occurred within, not among groups, regardless of whether groups were

13 based on mitochondrial lineage or geography (Table 2.2 and Appendix A, Table 3). The trend in the results was the same regardless of which dataset (only individuals sequenced for all 6 nuclear genes or all individuals sequenced for cir, fil, and mep) was tested or which analysis (Arlequin or Genodive) was conducted.

Haplotype1 Haplotype 2 Haplotype 3

Key Largo Long Key Marquesas Dry Tortugas Figure 2.3. Bayesian clustering indicated by STRUCTURE, with individuals grouped by mitochondrial haplotype (top) and geography (bottom). Each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

2.3.2. Distribution of simulated and empirical summary statistics

The distributions of summary statistics calculated from datasets simulated under the fast and slow substitution rates for all nuclear genes were in general agreement, although values for ΦST were smaller and more variable under the slow rate (Figure 2.4 and Appendix A, Figures 4-8). The empirical value fell within the null distribution for theta and nucleotide diversity for each gene and fell outside the null distribution for ΦST, DCEs, and s. Based on these results, we fail to reject the hypothesis of ILS based on nucleotide diversity and theta and reject ILS based on ΦST, DCEs, and s.

The results of the simulations to evaluate summary statistics are shown in Figure 2.5. The distribution of theta values for all population divergence times overlapped. There was also overlap in nucleotide diversity for datasets simulated with population divergences of 0.1N, 0.5N, and 1N generations (Figure 2.5).

14 Table 2.2. Results of hierarchical analyses of molecular variance (AMOVA) from Arlequin (sequence data). Groups were defined by mitochondrial lineages or geographic locations. Two datasets were analyzed: A) only individuals sequenced for all six nuclear genes and B) all individuals sequenced for cir, mep, and fil. Underlined values are significant ______

Dataset: Arlequin_A Groups: mtDNA lineages

Source of variation % variance Φ statistic P value ______

Among groups -0.77 ΦCT = -0.008 0.736

Among populations 7.11 ΦSC = 0.071 0.002 within groups

Within populations 93.66 ΦST = 0.063 < 0.001 ______

Dataset: Arlequin_A Groups: Geographic locations

Source of variation % variance Φ statistic P value ______

Among groups 5.47 ΦCT = 0.055 < 0.001

Among populations 1.94 ΦSC = 0.021 0.173 within groups

Within populations 92.59 ΦST = 0.074 0.002 ______

Dataset: Arlequin_B Groups: mtDNA lineages

Source of variation % variance Φ statistic P value ______

Among groups 0.23 ΦCT = 0.002 0.468

Among populations 8.29 ΦSC = 0.083 < 0.001 within groups

Within populations 91.48 ΦST = 0.085 < 0.001

15 (Table 2.2 continued) ______

Dataset: Arlequin_B Groups: Geographic locations

Source of variation % variance Φ statistic P value ______

Among groups 3.64 ΦCT = 0.036 0.068

Among populations 5.39 ΦSC = 0.056 0.004 within groups

Within populations 90.97 ΦST = 0.090 < 0.001 ______

Nucleotide diversity for datasets with older divergences (5N and 10N generations) had less overlap and generally increased as divergence times increased. As expected, ΦST was lowest in the 0.1N divergence dataset and highest in datasets with 5N and 10N population divergences. There was a wide range of ΦST values (-0.030 to 0.877) for 0.5N and 1N diverged datasets. DCEs and s decreased as divergence increased. There was overlap in DCEs and s values in 5N and 10N datasets and the 0.1N divergence dataset had a wide variation in DCEs and s values.

2.3.3. Coalescent simulations of demographic change

Tajima’s D values for each replicate within a simulation varied widely but the average Tajima’s D for those replicates matched our expectations when comparing across population effective size categories (Table 2.3). For example, in simulations 1a-c the mtDNA had the fastest response to the population contraction and therefore the highest Tajima’s D (2.02; Table 2.3, simulation 1a) while the simulated nucDNA datasets had slower responses and lower positive Tajima’s D values. The simulated nucDNA with four times the effective size of the mtDNA had the lowest positive Tajima’s D (1.13; Table 2.3, simulation 1c) while the simulated nucDNA with double the mtDNA effective size had an intermediate Tajima’s D (1.63; Table 2.3, simulation 1b).

2.4. DISCUSSION

Our data show that patterns of genetic diversity and population subdivision in the mitochondrial and nuclear genomes in C. vaginalis are discordant, in contrast to previously published sponge studies where genetic data are available from both genomes (Duran et al. 2004a; Duran et al. 2004b; Duran et al. 2004c; Whalan et al. 2008; Dailianis et al. 2011; Voigt et al. 2012). Our analyses show that discordance in C. vaginalis is not due to ILS of nucDNA or selection on mtDNA but may be explained by population size change or sperm-mediated dispersal.

16 fast fast fast 8

0.6 slow slow slow 2.34 0.007 -0.046 100 0.5 6 80 0.4 60 4 Density 0.3 40 0.2 2 20 0.1 0 0 0.0

0 1 2 3 4 5 6 0.00 0.01 0.02 0.03 0.04 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST

0.25 fast fast

slow 0.08 slow 19 31 0.20 0.06 0.15 Density 0.04 0.10 0.02 0.05 0.00 0.00

-5 0 5 10 15 20 0 10 20 30 DCE S Figure 2.4. Density plots showing distributions of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and model of nucleotide substitution for cata. Black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. Red arrows indicate the value calculated for the empirical data. Density plots for the other nuclear genes are in Appendix A, Figures 4-8.

2.4.1. Mito-nuclear discordance in Callyspongia vaginalis cannot be explained by incomplete lineage sorting or selection

We simulated 1000 datasets under each of 5 different population divergence times (0.1N, 0.5N, 1N, 5N, and 10N generations). The range of nucleotide diversity values calculated for each group of 1000 datasets was the same regardless of the divergence time under which the data were simulated. The same pattern was also true for theta. Given this insensitivity, neither of these summary statistics provides resolution for testing the null hypothesis of ILS as the cause of mito-nuclear discordance.

In contrast, distributions of ΦST, DCEs and s across datasets with different divergence times were more distinct, suggesting they are superior in comparing simulated and empirical datasets in a hypothesis-testing framework. Based on these findings, we reject the hypothesis that ILS explains the failure of the nucDNA to correspond to the mtDNA lineages. Given the population size and split times estimated for the mtDNA, nucDNA loci had sufficient time to sort and their failure to cluster according to mtDNA lineage is not due to differences in effective size of the two genomes. In the parametric bootstrapping approach we employed here, the expected distribution of summary statistics for the nucDNA if discordance in the mtDNA and nucDNA was due differences

17 in effective population sizes (and therefore sorting time) is generated. Had we employed only nucleotide diversity and/or theta, two widely used summary statistics, we would have erroneously failed to reject the hypothesis of ILS for our data. Our findings show that summary statistics that describe a genealogical or geographic pattern in the data (ΦST, DCE, s) are superior to those that describe sequence level variation (nucleotide diversity, theta) when testing for ILS by comparing simulated and empirical data. We suggest researchers test a suite of metrics to be sure that they are using the summary statistics most informative for their data when hypothesis testing.

0.1N 0.1N 0.1N 40

0.30 0.5N 0.5N 0.5N 1N 1N 1N 8 5N 5N 5N

0.25 10N 10N 10N 30 6 0.20 20 Density 0.15 4 0.10 10 2 0.05 0 0 0.00

4 6 8 10 12 14 16 0.00 0.02 0.04 0.06 0.08 0.10 0.0 0.5 1.0

Theta Nucleotide Diversity PhiST

1.0 0.1N 0.7 0.1N 0.5N 0.5N 1N 1N 0.6 5N 5N 0.8 10N 10N 0.5 0.6 0.4 Density 0.3 0.4 0.2 0.2 0.1 0.0 0.0

0 10 20 30 40 50 60 70 0 5 10 15 20 25 30 DCE S Figure 2.5. Density plots showing the distribution of summary statistics calculated for datasets simulated under various population divergences: 0.1N generation population divergence, blue; 0.5N, black, 1N, grey; 5N, red; 10N, green.

Selection can create discordant patterns of genetic diversity between the mtDNA and nucDNA. However, the results of the MK and HKA tests indicated that mtDNA in C. vaginalis is evolving according to neutral expectations. Selective sweeps on maternally inherited symbionts (i.e. Wolbachia, Hurst & Jiggins 2005) can also eliminate variation in the mtDNA with respect to the nucDNA (Meiklejohn et al. 2007). Many sponges have complex (Schmitt et al. 2011), often vertically transmitted (Webster et al. 2009) microbiomes, but C. vaginalis harbors small amounts of bacterial cells and may acquire symbionts from the environment (Giles et al. 2012), reducing the likelihood that the low variation in the mtDNA is due to selection on a symbiont.

18 Table 2.3. Parameters used in the coalescent simulations of population contraction and Tajima’s D values for each set of parameters. Population sizes reflect the ratio of the mitochondrial to nuclear genome for C. vaginalis (1:2) and humans (1:4). N1, initial population size; N0, final population size; θ, theta (4*N0*µ*locus length [bp]); T, duration of bottleneck in generations; D, mean of Tajima’s D (and range of D) ______

Simulation 1a-c: a) mtDNA b) 1:2 nucDNA c) 1:4 nucDNA N1 = 10 000 000 N1 = 20 000 000 N1 = 40 000 000 N0 = 1 000 N0 = 2 000 N0 = 4 000 θ = 0.04 θ = 0.8 θ = 1.6 T = 1000 T = 1000 T = 1000 D = 2.02 (-3.02 – 4.91) D = 1.63 (-2.84 – 4.81) D = 1.13 (-2.26 – 4.42)

Simulation 2a-c: a) mtDNA b) 1:2 nucDNA c) 1:4 nucDNA N1 = 1 000 000 N1 = 2 000 000 N1 = 4 000 000 N0 = 10 000 N0 = 20 000 N0 = 40 000 θ = 0.4 θ = 8 θ = 16 T = 1000 T = 1000 T = 1000 D = 0.527 (-2.31 – 3.98) D = 0.280 (-2.46 – 3.48) D = 0.103 (-2.40 – 3.40)

Simulation 3a-c: a) mtDNA b) 1:2 nucDNA c) 1:4 nucDNA N1 = 1 000 000 N1 = 2 000 000 N1 = 4 000 000 N0 = 200 000 N0 = 400 000 N0 = 800 000 θ = 8 θ = 160 θ = 320 T = 500 T = 500 T = 500 D = -0.088 (-2.48 – 3.06) D = -0.111 (-2.40 – 3.11) D = -0.109 (-2.43 – 3.12)

Simulation 4a-c: a) mtDNA b) 1:2 nucDNA c) 1:4 nucDNA N1 = 1 000 000 N1 = 2 000 000 N1 = 4 000 000 N0 = 200 000 N0 = 400 000 N0 = 800 000 θ = 8 θ = 160 θ = 320 T = 1000 T = 1000 T = 1000 D = 0.079 (-2.40 – 3.04) D = -0.041 (-2.38 – 3.49) D = -0.109 (-2.43 – 3.12) ______

2.4.2. Discordant mito-nuclear patterns of geographic subdivision are consistent with sperm-mediated dispersal

Male-biased dispersal is often invoked to explain mito-nuclear discordance when mtDNA is more structured than nucDNA. The pattern of geographically structured mtDNA and unstructured nucDNA for the hermaphroditic C. vaginalis could result if dispersal by

19 sperm (which transmit only nucDNA) is greater than that for eggs and larvae (which also transmit mtDNA).

Many coral reef taxa are broadcasters that release enormous quantities of gametes during synchronous mass spawning events (Harrison et al. 1984; Colin 1996). In contrast, many sponges, including C. vaginalis, employ spermcast mating where sperm released over an extended period disperse to find eggs retained for internal fertilization (Bishop & Pemberton 2006). The combined effects of finite gamete viability and dilution can reduce the fertilization success of broadcast spawners (Oliver & Babcock 1992; Levitan & Petersen 1995). However, gametes that remain viable for up to 72 hours (Bishop 1998) and fertilization at low sperm concentrations have been documented in spermcasting marine invertebrates (Pemberton et al. 2003). Sponges, which continuously pump water at high rates (3.5 liters/sec/kg dry mass for C. vaginalis, Weisz et al. 2008) may avoid the effects of gamete dilution by filtering and concentrating sperm (Bishop 1998).

Sponge larvae are competent to settle almost immediately after release from the parent and dispersal is brief (Maldonado 2006) so sperm (i.e. nucDNA) dispersal potential in spermcasting sponges may be substantial relative to larval (i.e. mtDNA and nucDNA) dispersal. Although, to our knowledge, sperm dispersal has not been measured directly in sponges, and most fertilizations in spermcasters come from individuals close to the broodparent (Grosberg 1991), fertilization success at distances of 207 meters were recorded in the field for the ascidian Botryllus schlosseri (Yund et al. 2007). Two studies examining multiple paternity in spermcasting marine invertebrates found that some of offspring were fathered by individuals outside the study area (Yund 1998, 300 m2; Lasker et al. 2008, 400 m2). In two spermcasters, a gorgonian and a sponge, reproductive success was not correlated with proximity to mates suggesting distance may not limit fertilization (Lasker et al. 2008, 400 m2; Maritz et al. 2010, 50 m2). In light of my results, these examples indicate sex-biased dispersal could contribute to mito-nuclear discordance in C. vaginalis.

2.4.3. Discordant mito-nuclear patterns of genetic diversity are consistent with a population bottleneck

Changes in population size can create conflicting patterns in mtDNA and nucDNA due to differences in the effective sizes of the two genomes; mtDNA responds faster to contractions and expansions than nucDNA (Fay & Wu 1999). My simulations showed Tajima’s D can be higher for mtDNA than nucDNA under a model of population contraction despite the 1:2 ratio of effective population size for mtDNA and nucDNA in C. vaginalis (Table 2.3). When we simulated a bottleneck of similar size and duration as one produced by a cyanobacterial bloom that killed 80% of surveyed sponges (including C. vaginalis) over 700 km2 in the Florida Keys (Butler et al. 1995), the average Tajima’s D was positive for the mtDNA and negative for the nucDNA (Table 2.3, simulations 4a- c). Field studies in Florida and the Caribbean show sponge populations can experience rapid, severe declines (Butler et al. 1995; Wulff 1995; Laboy-Nieves et al. 2001; Cowart et al. 2006), supporting the idea that demography plays a role in determining genetic diversity in the Porifera.

20 2.4.4. Genetic population structure in the Porifera

Of the few population level studies in sponges employing nuclear sequences, most report minimal subdivision among populations, although on much larger spatial scales (along the Great Barrier Reef, Bentlage & Wörheide 2007; within the Mediterranean Sea, Dailianis et al. 2011) than the approximately 260 km considered in this study. However, other studies report significant subdivision at large spatial scales (from the Red Sea to central South Pacific Ocean, Wörheide et al. 2008) and at spatial scales similar to those here for C. vaginalis (tens to hundreds of kilometers, Blanquer & Uriz 2010). Phylogeographic studies in the Porifera have been hampered by the conservation of the sponge mitochondrial genome but the high levels of sequence variation and substitution rates uncovered here suggest nucDNA sequence data hold promise for future population level work.

21 Chapter 3. Phylogeography of the Coral Reef Sponge Callyspongia vaginalis Across the Caribbean

3.1. INTRODUCTION

Comparing phylogeographic patterns for multiple species with overlapping ranges often identifies common phylogeographic breaks, but these breaks may not represent shared demographic history caused by the same processes or at the same period in time. However, phylogeographic inference has progressed beyond qualitative explanations about genetic patterns so we can now test hypotheses of evolutionary history in a statistical framework that considers the coalescent stochasticity (Knowles 2009). Testing shared vicariance for multiple populations that span a phylogeographic break is of central importance to understanding the evolutionary history of co-distributed taxa (Hickerson et al. 2007; Huang et al. 2011; Oaks 2014).

In terrestrial systems, barriers that generate phylogeographic structure, such as mountains or rivers, are often readily apparent. In the ocean, isolating barriers are usually less obvious, particularly in geographically compact regions such as the Caribbean Sea. Furthermore, the potential for extensive dispersal in many marine taxa should theoretically reduce opportunities for population differentiation. However, the Caribbean contains over 12,000 marine species (Miloslavich et al. 2010) and includes several endemic radiations (Morrison et al. 2004; Taylor & Hellberg 2005; Thornhill et al. 2009; Richards et al. 2012) suggesting complex interacting factors such as philopatry (Jones et al. 1999), ocean currents that separate populations or retain dispersers (Schott et al. 1988; Colin 1995), and isolation during low sea level stands (Barber et al. 2000) are responsible for segregating populations and ultimately creating biodiversity in this region.

Previous studies have documented phylogeographic breaks in the Caribbean Sea. Baums et al. (2005) found populations of the elkhorn coral Acropora palmata were genetically subdivided west-east between Hispanola and Puerto Rico at the Mona Passage. Taylor and Hellberg (2003, 2006) documented genetic breaks at the Mona Passage and at the Central Bahamas in the goby genus Elactinus. The Florida Current has been identified as a barrier to gene flow between Florida and the Bahamas for a range of taxa, including fish, mussels, and corals and their zooxanthellate symbionts (Carlin et al. 2003; Lee & Ó Foighil 2004; Brazeau et al. 2005; Baums et al. 2010; Hemond & Vollmer 2010; Andras et al. 2011). In concordance with the genetic evidence for phylogeographic breaks in these regions of the Caribbean, biophysical oceanographic models predict highly restricted larval dispersal across the Mona Passage and the isolation of the Bahamas (Baums et al. 2006; Galindo et al. 2006).

Here we investigate patterns of population structure for four Caribbean species that live in close association - the facultative commensal brittle star Ophiothrix suensonii, the obligate commensal amphipods Leucothoe ashleyae and L. kensleyi, and their host, the branching vase sponge Callyspongia vaginalis. The intimate connection among these species may lead to more tightly correlated spatial and temporal patterns of subdivision.

22 For example, congruent phylogeographic patterns have also been document in beetle/fungus (Roe et al. 2011), shrimp/goby (Thompson et al. 2005), and coral/zooxanthellae (Prada et al. 2014b) symbiotic pairs. However, symbioses do not always guarantee shared phylogeographic structure (McMullin et al. 2003; Parker et al. 2004; Crandall et al. 2008). Furthermore, differences in dispersal strategy among these species (see below) may also influence population structure (Hellberg 2009). Our aim is to test how symbiotic associations and life history interact to influence genetic structure in a four-species commensalism. We also report here for the first time the phylogeography of a sponge across the Caribbean.

3.2. METHODS

3.2.1. Study system

The brittle star O. suensonii is a sponge commensal most commonly found on C. vaginalis (Henkel & Pawlik 2005; Evans 2012). This passive suspension feeder spawns year round and its planktotrophic larval development times extend to 49 days in culture (Mladenov 1983). The obligate commensal amphipods Leucothoe ashleyae and L. kensleyi live and feed within the inner canals of C. vaginalis. Females brood fertilized eggs in abdominal pouches and the young disperse as crawl away juveniles (Thomas & Klebba 2006). Callyspongia vaginalis is abundant on coral reefs and promotes species richness by providing refugia for many commensal vertebrates and invertebrates (Ribeiro et al. 2003). This spermcasting hermaphrodite broods larvae that are competent to settle immediately upon release (Lindquist et al. 1997). Sponge biomass surpasses that of corals and algae in the Caribbean (Rützler 1978) and is expected to increase as oceans become warmer and more acidic (Bell et al. 2013). Nutrient cycling by sponges allows coral reefs, one of the world’s most productive and species rich ecosystems, to persist in oligotrophic seas (de Goeij et al. 2013). Despite their ecological importance on coral reefs and elsewhere, few studies have investigated lower level relationships in sponges. Inaccurate estimates of sponge biodiversity and species boundaries has implications for conservation and bioprospecting (Bickford et al. 2007; Leal et al. 2012).

3.2.2. Sample collection and data generation

Samples were collected from 18 locations across Florida, the Bahamas, and the Caribbean, although not all species were sampled from every location. Table 3.1 contains sample sizes and locations for all four species. Details about sample collection and genetic data generation for the brittle star and amphipods can be found in Richards (2010) and Richards et al. (2012), respectively. Accession numbers for previously published sequences for the brittle star and amphipods and those generated here for the sponge are available in Appendix B.

For Callyspongia vaginalis, the dataset generated here contains 248 individuals from 10 locations (Table 3.1, Figure 3.1). We sequenced the cytochrome oxidase I (COI) gene in each individual using protocols described in DeBiasse et al. (2010) and six nuclear

23 protein-coding genes in a random subset of individuals from each mitochondrial clade in each geographic location (n = 152) using protocols described in DeBiasse et al. (2014).

Table 3.1. Collection locations and sample size information for the host sponge C. vaginalis and its invertebrate commensals L. ashleyae, L. kensleyi, and O. suensonii based on COI data. ______

Location C. vaginalis L. ashleyae L. kensleyi O. suensonii⌘ ______

Bimini 29 30u 14u - Crooked Island 23 - - 33 Vieques 30 31u 20u - St. Croix 28 - - 31 Curaçao 30 27u 26u 32 Bocas del Toro 24 - - - Roatan 32 79u - - Utila 25 - - 32 Glover’s Reef 33 41u - - Cayman Island - - - 31 Veracruz 16 - - - Dry Tortugas 29* - - - Marquesas Keys 29* - - 28 Key West 30* 29n 33n 24 Long Key 34* 23n 31n 27 Key Largo 30* - - 18 ______*DeBiasse et al. (2010); nRichards et al. (2007); ⌘Richards (2010); uRichards et al. (2010);

All sequences were aligned and edited using Geneious v4.5.5 (Drummond et al. 2012). We resolved alleles for nuclear genes in heterozygous individuals using PHASE v2.1 (Stephens et al. 2001) with a 90% probability limit. Individuals heterozygous for a single insertion/deletion were resolved using CHAMPURUv1.0 (Flot 2007). Individuals with nuclear alleles that could not be phased to a probability >90% were cloned using the Invitrogen TOPO TA kit. At least 8 clones per reaction were sequenced. Individuals that could not be resolved after several rounds of cloning and phasing were removed from the dataset (1 for cata; 2 for cps; 2 for cir; 4 for ef; 1 for fil; 3 for mep). Each nuclear gene region was tested for intra-locus recombination using GARD and SBP as implemented in Hy-Phy (Pond & Frost 2005; Pond et al. 2006) and with the DSS method implemented in TOPALi v2.0 (Milne et al. 2004). Recombination was not detected within any gene region. The data set analyzed here also includes COI and nuclear gene sequences from 61 individuals from four Florida locations published previously (DeBiasse et al. 2010; DeBiasse et al. 2014).

24 Key Largo Straits of Florida Long Key Florida Marquesas Keys Crooked Dry Tortugas Bimini Island

Veracruz

Glover’s Reef St. Croix Utila Mona Passage Roatan Vieques

Curacao Bocas del Toro

Figure 3.1. Distribution of COI haplotypes in C. vaginalis. The sector area at each sampling location is proportional to haplotype frequency at that site. The inset shows the haplotype network for COI. Two haplotypes (purple) are connected to each other but not to the main network at the 95% confidence interval.

3.2.3. Genetic diversity indices for Callyspongia vaginalis

We calculated basic summary statistics (number of segregating sites, number of alleles, and haplotype and nucleotide diversity) in DnaSP v5.0 (Rozas et al. 2003; Librado & Rozas 2009). Because previous research suggested that demographic change influences patterns of genetic diversity in C. vaginalis at small spatial scales (DeBiasse et al. 2014), we calculated Tajima’s D (Tajima 1989b) to test for population bottlenecks and expansions for each gene and each sampling location, also in DnaSP.

3.2.4. Distribution of genetic variation for Callyspongia vaginalis

We constructed allele networks to visualize the relationships among sequences using TCS v1.21 (Clement et al. 2000) using the default settings. We used STRUCTURAMA v2.0 (Huelsenbeck et al. 2011) and STRUCTURE v2.3.2 (Pritchard et al. 2000) to determine how genetic variation was distributed geographically. DNA sequences were recoded into frequency data using DnaSP. Individuals were considered homozygous for their mitochondrial haplotype. Because previous evidence suggests patterns of mitochondrial and nuclear DNA are discordant in the sponge at small spatial scales (DeBiasse et al. 2014), we ran all clustering analyses with the nuclear DNA (nucDNA) alone and with the mitochondrial DNA (mtDNA) and nucDNA combined. In STRUCTURAMA, we allowed the number of populations to be a random variable and

25 set the DPP concentration parameter prior to gamma with the shape and scale set to 0.1 and 1.0, respectively. We ran the program for 10 million MCMC steps, retaining samples every 100 steps, and used a 10% burnin when summarizing the results of the population assignment. We conducted at least two runs with the same infile and program settings to ensure consistency of the population assignments. In STRUCTURE, we applied the admixture model with correlated allele frequencies and performed 20 iterations, each of which consisted of 1 million steps and a 100,000 step burnin, for every possible number of clusters, “K”. For each run, K was set to 1 through the number of geographic locations included in that particular run. The Evanno et al. (2005) method, implemented in STRUCTURE HARVESTOR (Earl & vonHoldt 2011), was used to determine the most likely number of genetic clusters in the data. Results for the STRUCTURE runs were merged with CLUMPP v1.1.2 (Jakobsson & Rosenberg 2007) and bar plots for the STRUCTURE and STRUCTURAMA runs were created with DISTRUCT v1.1 (Rosenberg 2004).

The results suggested that a subset of individuals from Bocas del Toro (n = 3), Utila (n = 4), and Veracruz (n = 7) were genetically divergent from all the other individuals, so we also performed the STRUCTURAMA analyses as described above excluding these 14 individuals. Finally, given the frequency at which we recovered a west-east division of the data (see Results), we performed additional clustering analyses following the procedures described above on individuals from the western and eastern clusters alone to test for fine scale subdivision within the two regions.

3.2.5. Testing the hypothesis of cryptic species in Callyspongia vaginalis

We tested whether divergent individuals from Bocas del Toro and Utila represented an independently evolving lineage using model testing approaches implemented in spedeSTEM v2.0 (Ence & Carstens 2011). This method compares the probability of trees with different numbers of OTUs to determine the optimal number of partitions in the data under the multispecies coalescent model. SpedeSTEM estimates the probability of gene trees input by the user given the species tree for all possible combinations of OTU groupings. Although phylogenetic uncertainty in the species tree does not influence species delimitation, the accuracy of spedeSTEM is dependent on the quality of the gene tree estimates (Ence & Carstens 2011).

For the spedeSTEM analyses, we assigned individuals to one of four groups based on the results of STRUCTURAMA (Appendix B, Figure 1). We excluded individuals from Veracruz from these analyses due to uncertainty in their population assignment (affiliation with Bocas del Toro and Utila based on nucDNA but with Florida based on mtDNA, Figures 3.1, 3.2, and Appendix B, Figure 1) In spedeSTEM, we subsampled six alleles per group per locus (Hird et al. 2010) and estimated maximum likelihood gene trees based on this subsample in PAUP* v3.1 (Swofford 2003) using a model of nucleotide substitution determined in jModelTest v2.1.4 (Darriba et al. 2012). This process was repeated with ten different subsamples of six alleles per group per locus. In one set of spedeSTEM runs, we used the sponge Amphimedon queenslandica as an outgroup. In a second set of spedeSTEM runs, we removed A. queenslandica due to our

26 concerns that it may be too divergent from C. vaginalis to be an appropriate outgroup (Graham et al. 2002). In its place we used one of the four groups identified by STRUCTURAMA as the outgroup and designated the remaining three groups as lineages within one species, assuming that if the individuals from Bocas del Toro and Utila represent an independently evolving lineage, then the tree topology where they are assigned as the outgroup will have the highest model probability. SpedeSTEM analyses following the second strategy were conducted with 10 replicate subsamples of six alleles per group per locus for each of the four outgroup options (40 runs total). For both strategies, we used the validation option in spedeSTEM to calculate the likelihood of all possible hierarchical permutations of the species tree in which lineages were kept separate or collapsed, given the gene trees. We used information theory (Anderson 2008; Carstens & Dewey 2010) to rank all possible species trees, which represent alternative models, according to the following procedure: i) calculate the AIC for each model where the number of parameters equals the number of tips of the tree not including the outgroup, ii) calculate ΔAIC, the difference between the best (lowest) AIC score and the score of each remaining model, iii) calculate the model likelihood using the ΔAIC, and iv) divide each model likelihood by the sum of likelihoods for all models to get the model probability, wi, for each model.

3.2.6. Testing for concordant demographic history among coral reef invertebrate taxa

Populations of both species of Leucothoe amphipods and the brittle star show a west-east pattern of structure across the Caribbean similar to what we found for C vaginalis with the sharpest break occurring between Florida and the Bahamas over a distance of approximately 100 km (see Results). Shared phylogeographic structure among co- distributed taxa do not necessarily imply that the same biogeographic barriers or processes were responsible for creating these patterns or that they arose at the same point in time. To explicitly test for simultaneous divergence of the sponge and its invertebrate commensals, we used hierarchical approximate Bayesian computation (ABC) as implemented in MTML-msBayes v20140305 (Huang et al. 2011). This method incorporates mutation rate heterogeneity among loci and variation in demography among taxa (Hickerson et al. 2007; Huang et al. 2011) and estimates population specific sub- parameters for each taxon and three hyper-parameters that describe the mean, variance, and number of divergence events across all of the population pairs. First, we estimated summary statistics from the empirical COI datasets of the four co-distributed species. Next we simulated 3,000,000 datasets drawing population sub-parameters and hyper- parameters from a prior distribution. The upper bound for prior distribution of the number of possible of divergence events (ψ) was set to equal the number of taxa in the analysis (n = 4). We approximated the posterior distribution for the hyper-parameters by retaining 900 simulated models whose summary statistic vectors had the shorted Euclidian distances from the summary statistic vector in the empirical data. We used ψ and Ω, which measures the incongruence among population divergence times, to evaluate the relative support of each model (Stone et al. 2012). Ω values of 0 (e.g. no variation in divergence times among taxa)and ψ values of 1 (e.g. a single divergence event for all

27 taxa) are indicative of simultaneous divergence of all taxa across a common phylogeographic barrier.

3.3. RESULTS

3.3.1. Genetically divergent individuals in Callyspongia vaginalis represent an independently evolving lineage

Multiple analyses showed that some individuals from Bocas del Toro and Utila were genetically divergent from all other samples. For pairwise K2P distances, these differences were 4% at COI and 3% at cata. In TCS networks generated for the COI and cata genes, these individuals had gene sequences that joined to each other but not to the main network at the 95% confidence level (Figure 3.1 (inset), Appendix B, Figure 2). STRUCTURAMA assigned these individuals to their own cluster, whether based on nucDNA alone or on mtDNA and nucDNA combined. (Appendix B, Figure 1). As K increased in the STRUCTURE analyses, these individuals were assigned to their own cluster before any other individuals were partitioned into groups corresponding to finer scale geography (Appendix B, Figures 3 and 4).

Motivated by these results, we tested the hypothesis that these divergent individuals represent an independently evolving lineage by ranking alternative phylogenetic models using information theory. The best model with A. queenslandica as the outgroup was one in which the four groups suggested by STRUCTURAMA were collapsed into a single lineage (Table 3.2a). The second ranked model was one in which Florida and the western and eastern Caribbean were collapsed into one lineage and the divergent individuals from Bocas and Utila were sister to that group (Table 3.2a). When we removed A. queenslandica from the analysis, the best model was one in which the divergent individuals from Bocas and Utila were designated as the outgroup and the remaining three groups were collapsed into one lineage (Table 3.2b). Based on these results, we conclude that the divergent genotypes from Bocas del Toro and Utila are from a cryptic species.

Table 3.2 Information-theoretic metrics for analyses testing evolutionary independence of divergent individuals from Bocas del Toro and Utila. Putative lineages within Callyspongia vaginalis are based on the groups identified by STRUCTURAMA (Appendix B, Figure 1): dark green, fl; light green, west; yellow, east; blue, div. A) Out group is A. queenslandica. B) Outgroup (capitalized) is one of the four groups identified by STRUCTURAMA. ______

A. Model mean -lnL k AIC ΔAIC Model likelihood wi ______aq, fl_west_east_div 626.120 1 1245.25 0 1 1 aq, div, fl_west_east 4158.40 2 8320.80 7066.55 0 0 aq, east, fl_west_div 4613.66 2 9231.32 7977.08 0 0

28 (Table 3.2 continued) ______

A. Model mean -lnL k AIC ΔAIC Model likelihood wi ______aq, fl_west, east_div 4938.95 2 9881.90 8627.65 0 0 aq, div, fl_west, east 4941.44 3 9888.88 8634.64 0 0 aq, fl, west_east_div 4995.60 2 9995.19 8740.95 0 0 aq, west, fl_east_div 5011.66 2 10027.32 8773.07 0 0 aq, fl_div, west_east 5027.20 2 10058.41 8804.16 0 0 aq, div, fl, west_east 5030.73 3 10067.46 8813.21 0 0 aq, div, fl_west, east 5066.88 2 10013.76 8883.51 0 0 aq, west_div, fl, east 5082.27 3 10170.54 8916.30 0 0 aq, div, fl_east, west 5095.22 3 10196.44 8942.19 0 0 aq, west, fl_div, east 5109.95 3 10225.89 8971.65 0 0 aq, west, fl, east_div 5111.13 3 10228.25 8974.01 0 0 aq, div, west, fl, east 5113.65 4 10235.30 8981.05 0 0 ______

B. Model mean -lnL k AIC ΔAIC Model likelihood wi ______

DIV, west_east_fl 3555.64 1 7113.28 0 1 1 EAST, west_fl_div 4379.11 1 8760.21 1646.94 0 0 DIV, west_fl, east 4462.51 2 8929.02 1815.74 0 0 EAST, west_fl, dv 4536.67 2 9077.34 1964.07 0 0 WEST, east_fl_di 4677.45 1 9356.90 2243.62 0 0 FL, west_east_div 4752.26 1 9506.52 2393.24 0 0 DIV, east_fl, west 4760.34 2 9524.68 2411.40 0 0 FL, west_east, div 4763.27 2 9530.54 2417.26 0 0 DIV, fl, west_east 4763.32 2 9530.63 2417.36 0 0 WEST, east_fl, div 4834.68 2 9673.35 2560.07 0 0 DIV, west, fl, east 4840.72 3 9687.43 2574.16 0 0 EAST, west_div, fl 4910.35 2 9824.71 2711.43 0 0 FL, west_div, east 4910.35 2 9824.71 2711.43 0 0 FL, west, east_div 4911.01 2 9826.02 2712.74 0 0 WEST, east_div, fl 4911.01 2 9826.02 2712.74 0 0 EAST, fl_div, west 4912.96 2 9829.92 2716.64 0 0 WEST, fl_div, east 4912.96 2 9829.92 2716.64 0 0 FL, div, west, east 4915.10 3 9836.21 2722.93 0 0 WEST, div, fl, east 4915.10 3 9836.21 2722.93 0 0 EAST, div, west, fl 4915.10 3 9836.21 2722.93 0 0 ______

29 3.3.2. Genetic diversity and distribution of genetic variation in Callyspongia vaginalis

Nuclear loci were more diverse than COI in terms of haplotype number and nucleotide diversity (Appendix B, Table 2). Cata and fil had the highest number of alleles (36 and 35, respectively) and fil had the highest nucleotide diversity (0.0239). For COI, Tajima’s D values were significantly positive for Key Largo, the Marquesas Keys, and Curaçao, significantly negative for Glover’s Reef, and nonsignificant for all locations combined. For the nuclear loci, Tajima’s D was nonsignificant for each sampling location individually and for all locations combined, with the exception of cps in Key Largo and Long Key and fil in Long Key.

The five major COI haplotypes present in C. vaginalis were generally geographically restricted (Figure 3.1). The brown haplotype was the most frequent overall and had the widest distribution, ranging from Bimini to Bocas del Toro. It was the only haplotype found in Vieques and St. Croix. The orange haplotype was restricted to Central America (Glover’s Reef, Utila, and Roatan). The red haplotype was restricted to Veracruz, Florida, and Bimini. The blue haplotype was common in Florida and Bimini and occurred in a few individuals from Central America. The green haplotype was found in half the individuals sampled in Curaçao and was also found in Florida and in a few individuals in Central America.

When the cryptic species from Bocas del Toro and Utila and individuals from Veracruz were excluded from the STRUCTURAMA analyses, there was high support for two clusters that corresponded to the western and eastern regions of the Caribbean (Figure 3.2). The Evanno et al. (2005) method also found the best K was 2 with western and eastern clusters the same as those identified by STRUCTURAMA (Appendix B, Figures 5 and 6). The western region included Florida, Glover’s Reef, Utila, and Roatan. The eastern region included Bimini and Crooked Island in the Bahamas, Vieques, St. Croix, and Curacao.

The cluster affiliation of Bocas del Toro and Veracruz changed depending on which data set was analyzed. Non-divergent individuals from Bocas del Toro clustered with the west when the nucDNA was analyzed alone (Figure 3.2, Appendix B, Figure 5) but with the east when mtDNA was included in the clustering analyses (Figures 3.2b, Appendix B, Figure 6). All but one non-divergent individual from Bocas del Toro had a mtDNA haplotype common in the eastern Caribbean (Figure 3.1). All individuals from Veracruz grouped with the cryptic species from Utila and Bocas del Toro based on the nucDNA alone (Appendix B, Figure 1) but with Florida according to mtDNA (Appendix B, Figure 1). All individuals from Veracruz had a mtDNA haplotype common in the southern Florida Keys (Figure 3.1). When all individuals were included in the STRUCTURE analyses, the Evanno et al. method found the best K was 2 with large secondary peaks at 7 for the nucDNA alone (Appendix B, Figure 3) and for the nucDNA and mtDNA combined (Figure 3.3), indicative of substructure within the larger western and eastern regions.

30 Probability of assignment 0.98 Probability of assignment 0.99

Key Largo Long Key Bimini Bimini Marquesas Keys Florida Florida Dry Tortugas Crooked Island Crooked Island

Vieques Vieques Glover’s Reef Glover’s Reef Utila Utila Roatan St. Croix Roatan St. Croix

Curacao Curacao Bocas del Toro Bocas del Toro

A B

Figure 3.2. Maps showing the distributionK = 8 of genetic clusters inferred from K = 6 STRUCTURAMA based on (A) the nucDNA alone excluding the divergent individuals from Veracruz,Florida Utila, andBimini Bocas del Toro and (B) the nucDNAFlorida andBimini mtDNA combined excluding the divergent individuals.Crooked Island Results of the Evanno et al. methodCrooked match Island those obtained in STRUCTURAMA (excluding divergent individuals K = 2 for nucDNA alone Vieques Vieques Glover’s Reef Glover’s Reef and for nucDNAUtila and mtDNA combined). The sector area inUtila the pie chart at each sampling location isRoatan proportional to numberSt. Croix of individuals in aRoatan particular clusterSt. at Croix that site. Curacao Curacao Bocas del Toro Bocas del Toro WeC performed additional STRUCTURAMAD and STRUCTURE analyses to test for fine- scale subdivision within these western and eastern regions. Datasets including the mtDNA and nucDNA combined revealed higher levels of substructure than the nucDNA only dataset. For the nucDNA alone, no subdivision was supported within either region based on STRUCTURAMA (Appendix B, Figure 11). The Evanno et al. (2005) method found K = 3 within both the western and eastern regions, but assignments did not correspond well to geography (Appendix B, Figures 7 and 9). When mtDNA was included, K = 2 was best for both west and east according to STRUTURAMA and K = 2 and 4 for the west and east, respectively, according to the Evanno et al. method (2005), with clusters corresponding to each sampling location (Appendix B, Figures 8 and 10). Therefore, we conclude that hierarchical subdivision exists among C. vaginalis populations within regions and that the major signal of differentiation is between the western and eastern regions of the Caribbean.

3.3.3. Co-phylogeography and tests for simultaneous divergence for Callyspongia vaginalis and its commensals

Qualitatively, C. vaginalis had similar patterns of phylogeographic structure at COI compared to its invertebrate commensals, with populations divided west-east across the Caribbean (Figure 3.4; Richards 2010; Richards et al. 2012). In locations where the sponge and amphipods were sampled together (Florida, the Bahamas, Vieques, St Croix, Curaçao, Roatan, and Glover’s Reef), the west-east pattern matched exactly.

31 L(K) (mean +- SD) DeltaK = mean(|L”(K)|) / sd(L(K))

-3800 40 a t a d

f

o 30

b

o -4200 r K

p

a t n 20 l L

e . t D s e

f -4600

o 10

n a e

M 00 -5000 2 4 6 8 10 12 14 2 4 6 8 10 12 14 K K 1.0

0.5

0.0 1.0

0.5

0.0

Utila Roatan Bimini Veracruz CuracaoSt. CroixVieques Key Largo Long Key Dry Tortugas Glover’s Reef Bocas del Toro Crooked Island Marquesas Keys Figure 3.3. Population assignment information for all individuals based on the nucDNA and mtDNA combined. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with peaks at K = 2, 7, and 10. Bottom: bar plots for K = 2 and K = 7 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

The sponge and the two amphipods shared a break between Florida and the Bahamas, a distance of approximately 100km. In the brittle star, this break was not as complete, with some haplotypes shared between Florida and the Bahamas. Ophiothrix suensonii had high haplotype diversity clustered in two clades. Clade I dominated the eastern regions but was found with lower frequency in the west. Clade II was restricted to western locations in Florida and Central America (Richards 2010). Sampling for all species was insufficient to precisely determine the location of the phylogeographic break separating western and eastern populations at the southern edge of the Caribbean basin. None of the taxa examined here show signs of population subdivision across the central Bahamas or at the Mona Passage.

32 V = Callyspongia vaginalis A = Leucothoe ashleyae K = Leucothoe kensleyi S = Ophiothrix suensonii V A V S K A K V S

V S V S V A A S K

V A S

V

Figure 3.4. Distribution of major genetic clades based on COI data for C. vaginalis, L. ashleyae, L. kensleyi, and O. suensonii. Warm colors represent the western clade and cooler colors represent the eastern clades.

The results for the test of simultaneous divergence among taxa were ambiguous. The mode and mean for Ω, which measures the incongruence among divergence times across the same phylogeographic barrier, were 0.0 and 0.190, respectively, however 95% of the values in the approximated posterior distribution (95% quantile) were between 0.0 and 0.658. While a value of zero indicates simultaneous divergence, values above zero in the 95% quantile here suggest variation around the divergence of each of the four population pairs. For ψ, the mode and mean were 1.0 and 1.7, respectively, and the 95% quantile contained all possible numbers of divergence events (1 through 4). The posterior probabilities for models with 1, 2, 3, and 4 divergence events were 0.518, 0.308, 0.126, and 0.048, respectively. Although the range of Ω values contained zero, values greater than zero, which indicate variation around the divergence times of co-distributed taxa, were contained within the distribution. For ψ, 80% of the posterior support was divided between models with one and two divergence events, preventing us from distinguishing the best model to describe divergence history for the host sponge and its commensals.

3.4. DISCUSSION

3.4.1. Phylogeographic structure in Callyspongia vaginalis

Here we present the first study of phylogeography for a sponge throughout the Caribbean. Early sponge phylogeography studies were hampered by slow rates of nucleotide substitution in the mitochondrial genome (Duran et al. 2004c; Wörheide 2006) but we found COI variation in C. vaginalis to be an order of magnitude higher compared to other sponges (π = 0.0118, this study; π = 0.0006, Duran et al. 2004c; π = 0.00049, Wörheide

33 2006; π = 0.00058, López-Legentil & Pawlik 2009). By combining COI sequences with data from six single-copy protein coding nuclear markers, which evolve between 1.6 and 7.1 times faster than COI in C. vaginalis (DeBiasse et al. 2014), we showed C. vaginalis populations were divided between western and eastern regions in the Caribbean, with a secondary signal of subdivision within regions. Our findings of significant population structure were consistent with previous sponge studies employing a variety of markers, which often attribute genetic differentiation to limited dispersal of sponge larvae among populations (Duran et al. 2004a; Duran et al. 2004b; Blanquer et al. 2009; López- Legentil & Pawlik 2009; Blanquer & Uriz 2010; Dailianis et al. 2011).

Our previous work on C. vaginalis showed genomically discordant patterns of structure: COI haplotypes were geographically subdivided along the Florida Keys while nuclear alleles were panmictic among these same populations (DeBiasse et al. 2014). Coalescent simulations and neutrality tests suggested that mito-nuclear discordance in the Florida populations was not the result of incomplete lineage sorting of the nuclear allele or selection on COI, but was likely caused by changes in population size or sperm biased dispersal. At a larger spatial scale across the Caribbean, the mitochondrial and nuclear loci are largely concordant and we see no evidence of change in population size (i.e. nonsignificant Tajima’s D values for Caribbean locations, Appendix B, Table 2). Taken together, our results for C. vaginalis suggest geographic distance influences population structure on larger spatial scales while demographic events are important at smaller spatial scales.

3.4.2. Co-phylogeography of Callyspongia vaginalis and its invertebrate commensals

The sponge and its symbiotic amphipods and brittle star had similar patterns of phylogeographic structure, with populations for all four taxa divided west-east across the Caribbean. The northern break was between Florida and the Bahamas (the Straits of Florida) and the southern break occurred between Bocas del Toro and Curaçao. Acropora palmata is also divided west-east across the Caribbean, sharing a southern break between Bocas del Toro and Curaçao with the taxa examined here, although the northern break for this coral is located at the Mona Passage (Baums et al. 2005). Taylor and Hellberg (2003, 2006) also showed the Mona Passage to separate differentiated lineages in Elacatinus gobies although this region did not appear to influence genetic subdivision in the sponge or its commensals.

The sharpest break for C. vaginalis and the Leucothoe amphipods occurred at the Straits of Florida. The sponge showed some genetic exchange across this break but both amphipod species were reciprocally monophyletic between Florida and the Bahamas (Richards et al. 2012). Florida populations of the brittle star shared a portion of haplotypes with individuals in the Bahamas (Richards 2010). Although we tested whether this phylogeographic barrier resulted from simultaneous population divergence within all four taxa, our results left us unable to distinguish between models with one and two divergence events. We were constrained by the data available for the amphipods and brittle star (COI sequences only) and relying on a single genetic marker likely limited our ability to resolve shared divergences statistically.

34

The Straits of Florida are a major phylogeographic break for several other marine species. Andras et al. (2013) found allele frequency differences in the sea fan Gorgonia ventalina that suggest restricted gene flow between Florida and the Bahamas, while its algal symbionts shared no alleles across this gap (Andras et al. 2011). Two other species of coral and a mussel also have phylogeographic breaks across the Straits of Florida (Brachidontes exustus, Lee & Ó Foighil 2004; Agaricia agaricites, Brazeau et al. 2005; Acropora cervicornis, Baums et al. 2010). The geographic proximity of Florida and Bahamian coral reefs (~100 km) makes it unlikely that distance alone is responsible for restricting gene flow between populations. A more plausible cause is the depth (800 m) of the Florida Straits, lack of suitable intervening habitat, and rapid transport (3.0 x 107 m3s-1) of the Florida Current, which flows northward between Florida and the Bahamas (Baringer & Larsen 2001).

Within the eastern region of the Caribbean, from the Bahamas to Curaçao, the sponge and both amphipod species were further subdivided in coincident fashion. Genetic distances for L. ashleyae and L. kensleyi among sites suggested island endemism on a scale of 10’s to 1000’s of km in these amphipods (Richards et al. 2012). In contrast, nuclear data for the sponge and COI data for both amphipod species showed no subdivision along the Florida reef tract (Richards et al. 2007; DeBiasse et al. 2014). Limited dispersal among islands separated by deep water and little suitable intervening habitat could drive population isolation in the Caribbean for these taxa while Florida’s shallow coastline and continuous habitat may allow extensive stepping stone dispersal among populations. For example, Richards et al. (2007) proposed that gene flow in the amphipod was facilitated via rafting in sponge fragments torn off the reef during storms while DeBiasse et al. (2014) hypothesized nuclear allele frequencies in the sponge were homogenized by greater dispersal of sperm relative to eggs and larvae. Populations of A. cervicornis were also genetically connected within Florida but isolated across Caribbean locations (Vollmer & Palumbi 2007; Hemond & Vollmer 2010). In contrast to the sponge, amphipods, and coral, O. suensonii populations were highly connected within the eastern region over a distance of ~1700 km (Richards 2010), possibly a result of higher fecundity and extended larval duration of the brittle star compared with the sponge and amphipods.

The phylogeographic break at the Straits of Florida was well defined for the taxa examined here, but the location of the break between western and eastern regions at the southern end of the Caribbean was less clear. As we found for C. vaginalis, studies have shown population differentiation between Bocas del Toro and Curaçao for coral reef taxa (Baums et al. 2005; Vollmer & Palumbi 2007; Hemond & Vollmer 2010; Andras et al. 2011; Andras et al. 2013). The Santa Marta Massif, a triangular-shaped mountainous geomorphic feature is isolated from the main mountain ranges of Colombia and bounded on the north by the Caribbean Sea, may contribute to breaks across this region. The Santa Marta Massif was formed due to tectonic activity after the Miocene, covers 13,700 km2 and rises from sea level to 5700 m (Tschanz et al. 1974). A narrow coastal shelf, cold water upwelling, strong offshore currents (Cowen et al. 2006), and freshwater outflow from the Magdalena River may also help reduce population connectivity across the Santa Marta-Guajira Peninsula region. For example, Betancur-R et al. (2010) found the marine

35 catfish Cathorops sp. and C. mapalé were reciprocally monophyletic over 150 km across the Santa Marta Massif (Campbell 1965).

3.4.3. Endemism and isolation in Central America

Genetic data have shown that cryptic species are prevalent in the Porifera (Klautau et al. 1999; Blanquer & Uriz 2007; Xavier et al. 2010; de Paula et al. 2012). The morphological characters traditionally used to define sponges taxonomically (spicules, spongin fibers) tend to be simple and can be influenced by environmental conditions, predation, sedimentation, and hydrology (Palumbi 1984, 1986; Barnes & Bell 2002; Loh & Pawlik 2009). A meta analysis by Hart and Sunday (2007) found that DNA sequences that fail to connect in a statistical parsimony network, as we see here (Figs. 1 and S2), are likely different species. Moreover, when enforcing the default 95% confidence limit, as we did, the TCS method gives a low rate of false positive errors across a broad range of taxa and genetic markers (Hart & Sunday 2007). Although the use of a single marker, particularly COI, to delimit species is controversial (DeSalle et al. 2005), our observation of 4% COI sequence divergence for a species in a phylum characterized by slow mitochondrial evolution (Shearer et al. 2002; Hellberg 2006; Huang et al. 2008) and evidence of COI being invariant across sponge species (Pöppe et al. 2010) lends support to the idea that the divergence observed here in C. vaginalis is meaningful. The results of clustering methods have also been used to delimit species or to motivate hypotheses of lineage divergence for further testing (Pinzon & LaJeunesse 2011; Carstens & Satler 2013; Prada & Hellberg 2013; Satler et al. 2013). Simulations by Rittmeyer and Austin (2012) showed that given a total species tree depth of 6N generations, STRUCTURAMA accurately delimits species using just 5 alleles from 5 loci per species, far below the number of alleles we sampled here. The results of the clustering and TCS analyses, combined with high probability for phylogenetic models where the divergent individuals are the outgroup, support the hypotheses of cryptic species in C. vaginalis.

Populations of C. vaginalis and O. suensonii from Central America were genetically isolated from the rest of the Caribbean. Almost all sponges sampled from Belize and Honduras had a COI haplotype restricted to Central America. For the brittle star, all pairwise tests of population differentiation between Honduras and other Caribbean locations were significant and migration rates into and out of Honduras were an order of magnitude lower than those among other locations (Richards 2010). Northwestern Central America has been shown to be an area of endemism for terrestrial and marine species (Briggs 1984; Rauchenberger 1988; Losos et al. 1998; Roberts et al. 2002). For example, Andras et al. (2011; 2013) reported populations of the sea fan G. ventalina in Belize and Panama were strongly differentiated from the wider Caribbean, as were populations of the sea fan’s symbiotic zooxanthellae. The goby Elacatinus oceanops was monophyletic for mitochondrial and nuclear markers between Florida and Belize (Taylor & Hellberg 2006) and Colin (2002) described E. lori as endemic to Honduras and Belize. The coral reef fish genera Acantheblemaria and Sanopus also contain Gulf of Honduras endemics (Collette 1983; Johnson & Brothers 1989). Currents in Central America likely play a role in structuring populations in this region. Gyres circulating in the Gulf of Honduras (Heyman & Kjerfve 2000) and off the coast of Panama and Colombia

36 (Andrade & Barton 2000; Richardson 2005) may retain disperses from leaving and prevent migrants from other locations from arriving. Indeed, a biophysical model predicting larval reef fish dispersal across the Caribbean found self-recruitment was higher in Belize, Honduras, and Panama than other areas of the Caribbean (Cowen et al. 2006).

3.4.4. Conclusions

We found populations of C. vaginalis to be subdivided west-east across the Caribbean with subtle structure at smaller scales within regions. The sponge and its invertebrate commensals were largely phylogeographically concordant although statistical tests of simultaneous divergence across the Straits of Florida were inconclusive. Patterns of structure observed for the invertebrates examined here agreed with some previously documented breaks (Straits of Florida) but not others (across Central Bahamas). Divergent lineages in Central America for the sponge and brittle star reinforce previous evidence of isolation of this region.

37 Chapter 4. Species Delimitation in the Caribbean Coral Reef Sponge Genus Callyspongia

4.1. INTRODUCTION

Coral reefs are the most species-rich habitats in the ocean but accurately quantifying their diversity is difficult due to the simple and plastic morphological characteristics of many of their inhabitants. This is particularly true for the major reef building taxa: corals, algae, and sponges. Several studies have found incongruence between morphological taxonomy and genetic lineages with taxa being overly split, conservatively lumped, or a combination of the two (Klautau et al. 1999; Miller et al. 2001; van der Strate et al. 2002; Duran & Rützler 2006; Erwin & Thacker 2007; Forsman et al. 2010; Pinzon & LaJeunesse 2011; Prada et al. 2014a). Additionally, rates of nucleotide substitution in the mitochondrial genome of sponges and anthozoans are 50- to 100-times slower than for bilateral animals (Shearer et al. 2002; Hellberg 2006), which can make cytochrome oxidase one (COI) uninformative for identifying species (Neigel et al. 2007; Huang et al. 2008; Pöppe et al. 2010). Sponges would particularly benefit from a model-based approach to species delimitation. The paucity of informative taxonomic characters in this group makes morphological species delimitation difficult and their long generation times and large effective populations sizes can lead to incomplete lineage sorting and gene tree/species tree discordance (Degnan & Rosenberg 2009). Recently, multi-locus species delimitation methods that account for the ancestral coalescent process have been developed (Yang & Rannala 2010; Edwards & Knowles 2014) and validated empirically (Leaché & Fujita 2010; Harrington & Near 2011; Satler et al. 2013; Prada et al. 2014a) but have not yet been applied to the Porifera.

Members of the phylum Porifera are among the most diverse taxa on coral reefs. Sponge biomass surpasses that of corals and algae in the Caribbean (Rützler 1978) and is expected to increase as oceans become warmer and more acidic (Bell et al. 2013). Nutrient cycling by sponges allows coral reefs, one of the world’s most productive and species rich ecosystems, to persist in oligotrophic seas (de Goeij et al. 2013). Sponges promote reef species richness by providing refugia for many commensal invertebrates, particularly during critical juvenile or reproductive life history phases (Ribeiro et al. 2003; Henkel & Pawlik 2005; Richards et al. 2007), and harbor a substantial biomass of diverse microbial endosymbionts, many of which produce secondary metabolites of ecological and pharmacological importance (Taylor et al. 2007). As basal metazoans, sponges are important models of animal complexity and development (Srivastava et al. 2010; Ludeman et al. 2014; Riesgo et al. 2014). Despite their ecological and evolutionary importance, few studies have investigated lower level relationships in sponges.

Sponge species in the genus Callyspongia have a wide range of morphologies and are phenotypically variable within species, which makes them an excellent model for answering questions about the correspondence of morphological and molecular species boundaries in the Porifera. The most common Caribbean species, C. vaginalis, has at least two recognized growth forms (fan and tube shaped, Humann & DeLoach 2003; Zea

38 et al. 2009) and varies in color and texture (López-Legentil et al. 2010). Callyspongia longissima and C. ?eschrichtii were described in 1864 (Duchassaing & Michelotti 1864; Van Soest et al. 2014) but doubt exists as to whether these are distinct species or should be considered growth forms of C. vaginalis (Zea et al. 2009).

Previous phylogenetic results have shown the genus Callyspongia and the species C. vaginalis to be paraphyletic based mitochondrial and nuclear ribosomal sequence data (Erpenbeck et al. 2007; Raleigh et al. 2007; López-Legentil et al. 2010; Redmond et al. 2011; Redmond et al. 2013). López-Legentil et al. (2010) found sequences for COI, 18S, and 28S were identical for C. vaginalis and C. fallax specimens sampled in Key Largo, Florida. In an ordinal level phylogeny estimated by Redmond et al. (2011) using sequences from the 5’ end of COI, the C. vaginalis and C. fallax individuals from López- Legentil et al. (2010) clustered together, a C. vaginalis specimen from Itskovich et al. (2007) clustered in a clade with Xestospongia muta, and a C. armigera specimen from Erpenbeck et al. (2007) clustered together with a group of C. vaginalis samples from DeBiasse et al. (2010). These phylogenetic patterns may be explained by the slow evolution nature of COI in sponges, misidentification of specimens, and/or DNA contamination and motivate our multi-locus, model-based comparison of molecular and morphological species boundaries in Callyspongia from photo-vouchered specimens. Once species boundaries are accurately identified, it will be possible determine how factors such as life history, molecular evolution, environment, and geography influence the distribution of genetic and phenotypic diversity within and among species in Callyspongia.

4.2. METHODS

4.2.1. Sample collection and generation of genetic data

Sponge samples from seven of the nine Caribbean Callyspongia species were collected from Islamorada and Key Largo in the Florida Keys, South Acklins Island in the Bahamas, and Bocas del Toro, Panama, between March and July 2013. Each sponge was photographed in situ before a tissue sample was collected and stored in 95% ethanol. Genomic DNA was extracted using the Qiagen DNeasy kit.

We amplified the 5’ end of the COI gene (Folmer et al. 1994) and the 3’ end of COI based on evidence that this downstream region of the gene may be more informative for sponge taxa (Erpenbeck et al. 2006). We also amplified the D1 region of the 28S gene and two single copy nuclear protein coding gene regions in the macrophage expressed protein, (mep) and filamin (fil) genes, which in C. vaginalis evolve 3.6 and 7.1 times faster relative to COI, respectively (DeBiasse et al. 2014). PCR primers were designed from mitochondrial alignments of sequences downloaded from GenBank (Table 4.1). Primers for the nuclear protein coding genes and the 28S gene were obtained from DeBiasse et al. (2014) and Redmond et al. (2011), respectively.

39 Table 4.1. Primer sequences and expected amplicon size for gene regions used in this study. ______

Gene name Primer sequence (5’ – 3’) Amplicon size ______

5’ COI1 F: ATATAATGTTATAGTGACAGCTCATGC 350 bp R: ATTGGAACTATCGGCTCCGGG

3’ COI1 F: TGGTTATTTGGGGATGGTATATGC 350 bp R: TAACAATACCCMGAWATTTTCCC

28S2 5F: TAGGTCGACCCGCTGCCYTTAAGC 400 bp 300R: CAACTTTCCCTCACGGTACTT

Filamin3 F: CAGTACAATATAAGCATGGTCCCTCG 230 bp (fil) R: GTATAGGCTCACACAGTCCTTCCC

Macrophage F: CTTTCTCTATAAGCACTAATATTGG 210 bp expressed protein3 R: AGTAGTCATAGCAAATGAGAGAGG (mep) ______1Designed for this study, 2Redmond et al. (2011), 3DeBiasse et al. (2014)

PCR amplifications were conducted in 25-µl reactions consisting of 2.5 µl of 10x buffer, 10 µM of dNTPs and each primer, and 0.25 units of One TaqTM DNA polymerase (New England Biolabs Inc.). Amplifications were performed in a Bio-Rad T100 thermocyler under the following conditions: an initial denaturation cycle of 3 minutes at 95°C, 2 minutes annealing at 47 to 50°C and 2 minutes extension at 72°C, followed by 38 cycles of 30 seconds at 95°C, 45 seconds at 47 to 50°C and 45 seconds at 72°C with a final extension cycle at 72°C for 10 minutes. Samples were sequenced in the forward and reverse direction using BigDye chemistry v3.1 on an ABI 3130XL at the Louisiana State University Genomics Facility. Forward and reverse sequences were aligned and edited using Geneious v4.5.5 (Drummond et al. 2012).

We resolved nuclear alleles in heterozygous individuals using PHASE v2.1 (Stephens et al. 2001). Individuals with alleles that could not be phased probabilistically to a probability > 85% (mep: 1 C. vaginalis; fil: 2 C. armigera, 2 C. eschrichtii, 1 C. longissima, 2 C. vaginalis) were removed from the dataset. Nuclear gene regions were tested for intra-locus recombination using GARD and SBP implemented in Hy-Phy (Pond & Frost 2005; Pond et al. 2006). Recombination was not detected in any gene region. We found no evidence of intragenomic variation in the 28S gene for any individuals. Sequences from Amphimedon compressa (COI, NC_010201; 28S, JN178945) and A. queenslandica (mep, GCF_000090795) were used as outgroups. Sequences for all individuals and genes are available from the European Nucleotide

40 Archive. Kimura 2 Parameter (K2P) genetic distances within and between species for all gene regions were calculated in MEGA v5.2.2 (Tamura et al. 2011).

4.2.2. Gene network and single locus gene tree estimation

Parsimony gene networks and maximum likelihood gene trees for each locus were constructed to determine the relationship among alleles within and among the Callyspongia species. Networks were estimated in TCS v1.21 (Clement et al. 2000) using the default settings and resulted in the most parsimonious connections between alleles at the 95% confidence level. Models of nucleotide substitution were determined in jModelTest v2.0.2 (Darriba et al. 2012) based on the AICc criterion and maximum likelihood trees were estimated in PAUP* (Swofford 2003) with 100 bootstrap replicates. We downloaded from GenBank 5’ COI sequences available for the Caribbean Callyspongia species examined here (Table 4.2) and estimated the maximum likelihood tree to compare the placement of previously published sequences with those generated in this study. We also included species from the genus Haliclona because Redmond et al. (2011) found this genus and Callyspongia to be paraphyletic. We determined the model of nucleotide substitution and estimated the gene tree topology as described above.

4.2.3. Species tree estimation and species delimitation

Callyspongia plicifera and C. tenerrima each formed clades in all gene trees (see Results) be unambiguous species, suggesting they are unambiguous species. However, C. armigera, C. ?eschrichtii, C. fallax, C. longissima, and C. vaginalis shared alleles for all loci and shared alleles with two individuals of C. fallax for the nuclear protein coding genes (see Results). Motivated by these findings, we used a model-based approach that estimates the species phylogeny in the face of gene tree discordance due to incomplete lineage sorting to test the hypotheses that C. armigera, C. ?eschrichtii, C. fallax, C. longissima, and C. vaginalis represent distinct evolutionary lineages. We used *BEAST v1.7.3 (Heled & Drummond 2010) to infer the species tree for these five taxa using different combinations of loci: 1) mep, fil, 28S, COI (all loci); 2) mep, fil, COI, excluding the multi-copy 28S gene, which evolves under concerted evolution, to determine whether these violations of the coalescent model influenced species tree estimation; 3) 28S, COI, excluding the nuclear protein coding loci due to their low resolving power. For each run, two independent MCMC analyses were conducted for 50 million steps, sampling every 5000 steps. Convergence was determined by viewing the log files in Tracer v1.5. All parameters had effective sample sizes (ESS) >300. Treefiles were combined in LogCombiner v1.7.1 with a 10% burn-in, and the maximum clade credibility (MCC) tree for the combined file was calculated in TreeAnnotator v1.7.1.

Using the species tree estimated in *BEAST as a guide tree, we evaluated nodal support for competing tree topologies where lineages represented by the Callyspongia species were maintained or collapsed in BPP v2.2 (Yang & Rannala 2010). Two independent BPP analyses with identical priors were run for 500,000 steps, sampling every 5 steps after a 50,000 step burnin. We specified priors for theta (2,2000) and tau (1,10) that represented small ancestral population size and older divergence times, a conservative

41 combination of priors that should favor maintaining current species designations (Leaché & Fujita 2010).

Table 4.2. COI sequences downloaded from GenBank and used in Figure 4.3 ______

Species Accession No. Reference ______

Callyspongia armigera EF519578 Erpenbeck et al. (2007)

Callyspongia fallax JN242192 Redmond et al. (2011) Callyspongia fallax JN242193 Redmond et al. (2011) Callyspongia fallax GQ415417 López-Legentil et al. (2010) Callyspongia fallax GQ415416 López-Legentil et al. (2010)

Callyspongia plicifera NC010206 Lavrov et al. (2008) Callyspongia plicifera EU23477 Kayal and Lavrov (2008)

Callyspongia vaginalis GQ415415 López-Legentil et al. (2010) Callyspongia vaginalis GQ415414 López-Legentil et al. (2010) Callyspongia vaginalis GQ415413 López-Legentil et al. (2010) Callyspongia vaginalis GQ415412 López-Legentil et al. (2010) Callyspongia vaginalis EF095182 Itskovich et al. (2007) Callyspongia vaginalis EF519577 Erpenbeck et al. (2007) Callyspongia vaginalis EF519579 Erpenbeck et al. (2007) Callyspongia vaginalis EF519581 Erpenbeck et al. (2007) Callyspongia vaginalis EF519580 Erpenbeck et al. (2007)

Haliclona cinerea JN242198 Redmond et al. (2011) Haliclona coerulea EF519619 Erpenbeck et al. (2007) Haliclona implexiformia EF519624 Erpenbeck et al. (2007) Haliclona simulans JN242201 Redmond et al. (2011) Haliclona toxius JN242206 Redmond et al. (2011) Haliclona tubifera EF519624 Erpenbeck et al. (2007) ______

4.3. RESULTS

Callyspongia fallax, C. plicifera, and C. tenerrima did not share TCS network connections with other species for COI or 28S (Figures 4.1 and Appendix C, Figures 1- 3). A meta analysis by Hart and Sunday (2007) found that DNA sequences failing to join a statistical parsimony network are likely different species and when enforcing the default 95% confidence limit, as we did here, the TCS method gives a low rate of false positive errors (Hart & Sunday 2007). The mean K2P genetic distance between C. fallax, C. plicifera, and C. tenerrima and all other species was between 10% and 28% (Table 4.3).

42 One C. fallax individual shared fil alleles with C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis. A second C. fallax individual shared mep alleles with C. armigera, C. ?eschrichtii, and C. vaginalis (Figure 4.1, Appendix C, Figures 4 and 5). No other C. fallax, C. plicifera, or C. tenerrima individuals amplified for fil or mep markers, consistent with sequence divergence at these loci for these species.

C. armigera C. eschrichti C. vaginalis C. longissima C. plicifera C. fallax C. tenerrima

28S 5’ COI 3’ COI

FIL MEP Figure 4.1. Unrooted statistical parsimony networks for each gene region with species represented by colors. Circles represent alleles with size proportional to frequency of the sequence. Small black circles represent possible, but not sampled alleles and lines between circles represent one mutational change between sequences.

Callyspongia armigera, C. ?eschrichtii, C. longissima, and C. vaginalis all shared alleles in each of the parsimony networks and maximum likelihood gene trees (Figure 4.1, Appendix C, Figures 1-5). For all loci the average mean K2P genetic distance among species for these four taxa (0-2.8%) was within the range of the mean K2P genetic distance within them (0-2.9%) (Table 4.3). In the 28S sequence alignment, C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis shared a one 6 base pair indel and two 1 base pair indels that distinguished them from C. fallax, C. plicifera, and C. tenerrima. In an alignment with no missing individuals, the 3’ region of COI had slightly higher haplotype (0.871) and nucleotide diversity (0.125) than the 5’ region (0.840, 0.120). However, the 5’ and 3’ COI regions had the same gene tree topology (Appendix C, Figures 1 and 2) and for the *BEAST and species delimitation analyses, we concatenated the 5’ and 3’ COI gene regions into one marker.

43 Table 4.3. Mean Kimura 2 Parameter genetic distances within and among species for each gene region. Italicized values are those estimated within species. Dashes indicate species for which only one individual was sequenced.

A: 5’COI ARM ESC FAL LON PLI VAG ARM 0.018 ESC 0.017 0.000 FAL 0.268 0.273 0.004 LON 0.011 0.012 0.273 - PLI 0.260 0.264 0.098 0.264 - VAG 0.020 0.017 0.265 0.020 0.260 0.015

B: 3’COI ARM ESC FAL LON PLI TEN VAG ARM 0.016 ESC 0.013 0.006 FAL 0.273 0.275 0.002 LON 0.010 0.005 0.275 - PLI 0.271 0.270 0.117 0.270 0.000 TEN 0.261 0.261 0.096 0.264 0.101 - VAG 0.018 0.018 0.274 0.019 0.276 0.259 0.015

C: 28S ARM ESC FAL LON PLI TEN VAG ARM 0.000 ESC 0.000 0.000 FAL 0.209 0.209 0.001 LON 0.000 0.000 0.209 - PLI 0.171 0.171 0.092 0.171 0.000 TEN 0.171 0.171 0.090 0.171 0.066 0.000 VAG 0.000 0.000 0.171 0.000 0.171 0.171 0.000

D: FIL ARM ESC FAL LON VAG ARM 0.026 ESC 0.026 0.026 FAL 0.027 0.019 0.024 LON 0.021 0.023 0.020 0.024 VAG 0.024 0.025 0.023 0.022 0.022

E: MEP ARM ESC FAL LON VAG ARM 0.029 ESC 0.026 0.000 FAL 0.026 0.000 0.000 LON 0.024 0.024 0.024 0.000 VAG 0.028 0.026 0.026 0.026 0.027

44 The species tree estimated using 28S and COI had the highest nodal support and its topology was consistent with the relationships recovered in the maximum likelihood trees and parsimony networks (Figure 4.2a). The species tree estimated using all loci had low posterior probability for all nodes and placed C. fallax and C. ?eschrichtii as sister to a clade containing C. armigera, C. longissima, and C. vaginalis (Figure 4.2b), in contrast to the relationships supported by the parsimony networks and the maximum likelihood gene trees. When we removed the 28S gene and repeated the analysis with COI, fil, and mep, the resulting species tree topology remained the same (C. fallax and C. eschrichtii sister) and the posterior support remained low (Figure 4.2b). Although the 28S gene violates coalescent theory because it is multi-copy and evolves under concerted evolution, which homogenizes copies within an individual, the effective population size of this marker is reduced, thereby decreasing sorting time and increasing resolution. The rRNA genes have been useful in estimating phylogeny in sponges (Redmond et al. 2013; Thacker et al. 2013) and many other taxa (Mindell & Honeycutt 1990; Hillis & Dixon 1991; Hamby & Zimmer 1992). Here we find that it improves support in the species tree analyses.

A B C. fallax C. fallax 0.67/0.78

C. eschrichtii C. eschrichtii

0.99 C. vaginalis C. armigera * 0.60 0.54/0.56 * C. armigera C. vaginalis 0.54 0.81/0.66 *

C. longissima C. longissima 3.0 E -4 3.0 E -4 Figure 4.2. Species trees estimated in *BEAST. Dashed boxes indicate the change in position of C. eschrichtii between trees. A) Species tree topology estimated using 28S and COI gene regions. Posterior probabilities from the *BEAST analysis are listed at the nodes. Collapsed nodes in the highest probability BPP species delimitation model are indicated with asterisks. B) Species tree topology recovered in the analysis of dataset 1 (all loci) and dataset 2 (nuclear protein coding loci and COI). Posterior probabilities for the topology based on dataset 1 are listed first and posterior probabilities for dataset 2 are listed second.

The results of the BPP species delimitation analyses indicated that C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis belong to a single evolutionary lineage. The model with highest posterior support (0.99) was one that maintained only the node splitting C. fallax and the other four Callyspongia species (Figure 4.2a). The posterior support for this node was 1.0 while the support for the node splitting C. longissima from C. ?eschrichtii, C. armigera, and C. vaginalis was 0.0012. There was zero support for the node splitting C. armigera and C. vaginalis and the node splitting C. armigera and C. vaginalis from C. ?eschrichtii.

45 H. tubifera 1 [1] TUB 2, 3 C. plicifera [3] NC_010206 EU237477 4 IMP H. implexiformiaPLI66 [1] H. cinerea 5 [1] CIN 6 TOXH. toxius [1] FAL12 FAL13 FAL14 FAL15 FAL1C.6 fallax [9] FAL17 Clade A FAL18 FAL19 FAL20 7 JN24219C. fallax2 [1] FAL269 FAL27C.0 fallax [2] JN242193 GQ415417 8, 9 [3] GQ41541C. fallax6 GQ415415 GQ415414 10 GQ41541C. vaginalis3 [4]

11 GQ415412 EF09518C. vaginalis2 [1]

H. simulans 12 [1] 13, 14 SIM C. vaginalis [2] EF519577 15 BDC.T022 vaginalis.18 [1] UTL012.16 VAG4C.1 vaginalis [1] ESC09 ESC1C.0 eschrichtii [3] ESC11 EF519581 16 EF51958C. 0vaginalis [2] ARM08 ARM5C.3 armigera [2] VAG4C.7 vaginalis [2] VAG83 Clade B EF51957C. armigera8 17 [1] ARM54 ARM5C.5 armigera [2] LON0C.1 longissima [1] EF519579 18 ARM5C. 7vaginalis [1] VAG2C.4 armigera [1] VAG26 A.COM0 compressa1 VAG3C.9 vaginalis [5] VAG40 VAG42 COEH. coerulea 19 [1]

0.04

Sequences from the 5’ region of COI sampled in this study clustered with Callyspongia sequences deposited on GenBank by other authors (Figure 4.3).

51 H. tubifera 1 TUB 94 C. plicifera2, 3 NC_010206 EU237477 4 IMP H. implexiformiaPLI66 83 H. cinerea 5 CIN 6 TOXH. toxius FAL12 93 FAL13 FAL14 59 FAL15 FAL1C.6 fallax Clade A 73 FAL17 FAL18 FAL19 FAL20 7 JN24219C. fallax2 95 FAL269 FAL27C.0 fallax JN242193 GQ415417 8, 9 GQ41541C. fallax6 GQ415415 GQ415414 10 GQ41541C. vaginalis3

11 GQ415412 EF09518C. vaginalis2

H. simulans 12 13, 14 SIM 99 C. vaginalis EF519577 15 BDC.T022 vaginalis.18 UTL012.16 VAG4C.1 vaginalis ESC09 57 ESC1C.0 eschrichtii ESC11 EF519581 16 EF51958C. 0vaginalis ARM08 80 ARM5C.3 armigera VAG4C.7 vaginalis VAG83 Clade B EF51957C.8 armigera17 99 ARM54 ARM5C.5 armigera 51 LON0C.1 longissima EF519579 18 ARM5C. 7vaginalis VAG2C.4 armigera 89 VAG26 A.COM0 compressa1 VAG3C.9 vaginalis VAG40 VAG42 COEH. coerulea 19

0.04 Figure 4.3. Maximum likelihood tree estimated in PAUP* using COI sequences generated in this study (bold) and those downloaded from GenBank (with superscript numerals). Colored boxes indicate each species according to Figure 4.1 and height of the bar is proportional to the number of individuals with that sequence (for example, one H. tubifera individual and three C. plicifera individuals). Support values from 100 bootstrap replicates appear at the nodes when greater than 50. Black bars show clades A and B recovered here and by Redmond et al. (2011). Dashed brackets represent evolutionary significant units in Callyspongia. The tree is rooted with Amphimedon compressa (accession number JN178945). GenBank sequences: 1Erpenbeck et al. (2007), EF519624; 2Kayal and Lavrov (2008), EU23477; 3Lavrov et al. 2008, NC010206; 4Erpenbeck et al. (2007), EF519624; 5Redmond et al. (2011) JN242198; 6Redmond et al. (2011), JN242206; 7Redmond et al. (2011), JN242192; 8Redmond et al. (2011), JN242193; 9López-Legentil (2010), GQ415416-17; 10López-Legentil (2010), GQ415412-15; 11Itskovich et al. (2007), EF095182; 12Redmond et al. (2011), JN242201; 13Erpenbeck et al. (2007), EF519577; 14DeBiasse (Chapter 3); 15DeBiasse (Chapter 3); 16Erpenbeck et al. (2007), EF519580-81; 17Erpenbeck et al. (2007), EF519578; 18Erpenbeck et al. (2007), EF519579; 19Erpenbeck et al. (2007), EF519619; see also Table 4.2.

We recovered the same two divergent clades reported by Redmond et al. (2011) and also found Haliclona and Callyspongia to be paraphyletic. The sequence for C. plicifera

46 collected here was identical to two other C. plicifera sequences obtained from complete mitochondrial genomes of this species (Kayal & Lavrov 2008; Lavrov et al. 2008). There was high bootstrap support (100%) for a clade containing the C. fallax sequences generated here and those from López-Legentil et al. (2010) and Redmond et al. (2011). This clade also contained C. vaginalis sequences generated by López-Legentil et al. (2010). Callyspongia armigera and C. vaginalis sequences from Erpenbeck et al. (2007) and C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis sequences generated here formed a clade with high bootstrap support (98%). A C. vaginalis sequence from Itskovich et al. (2007) was sister to the clade containing C. plicifera and C. fallax. In the phylogeny estimated by Redmond et al. (2011), this sequence clustered with Xestospongia, Petrosia, and Neopetrosia species with high support (94%). Given the divergent placement of this sequence outside clade A and B it may be the result of misidentification or contamination. Shared COI haplotypes between C. vaginalis individuals collected by López-Legentil et al. (2010) and C. fallax individuals examined here could be the result of hybridization between these two taxa with backcrossing of offspring with the parental species. Hybridization occurs between the Caribbean corals Acropora cervicornis and palmata (Vollmer & Palumbi 2002; Fogarty et al. 2012) and is an important process in the evolution of coral species worldwide (Willis et al. 2006). Data from more individuals and larger genomic regions, ideally combined with reproductive crosses, are needed to test the hypothesis of hybridization in Callyspongia.

4.4. DISCUSSION

The analyses performed here support the species status of Callyspongia fallax, C. plicifera, and C. tenerrima but not C. armigera, C. ?eschrichtii, C. longissima, or C. vaginalis. We found extensive allele sharing among individuals of the four latter species and the multi-locus, model-based species delimitation analysis strongly supported collapsing these taxa into a single genetic lineage. Our genetic results suggest that these taxa have been liberally split based on uninformative morphology and should all be considered growth forms of C. vaginalis (Lamarck1814).

4.4.1. Molecular evolution in the Porifera and its implications for species delimitation

Molecular evolution in the sponge mitochondrial genome has complicated its utility for phylogeographic and phylogenetic inference. In many sponge species, the mitochondrial genome is highly conserved. For example, Wörheide (2006) found only three COI haplotypes in the sponge Astrosclera willeyana between the Red Sea and Fiji, a distance of over 20,000 km, and Pöppe et al. (2010) found COI was invariant across species within the sponge genera Psammocinia and Ircinia. However, for some taxa, particularly the Haplosclerids, a diverse order containing many marine and all freshwater species (Hooper et al. 2002), the mitochondrial genome appears to evolve more rapidly. López- Legentil and Pawlik (2009) found four COI haplotypes in Bahamian populations of the giant barrel sponge Xestospongia muta and DeBiasse et al. (Chapter 3) found significant genetic population structure across the Caribbean in C. vaginalis based upon the distribution of 19 COI haplotypes. Variation in COI has been sufficient to reveal cryptic

47 species (Chapter 3; Wulff 2006; Blanquer & Uriz 2007; Pöppe et al. 2010; Xavier et al. 2010). In a phylogenetic context, data from multiple loci including COI show many Haplosclerid genera are paraphyletic (Redmond et al. 2011). Despite the fact that the genera Callyspongia and Haliclona are considered clearly defined based on morphology (Hooper et al. 2002), the phylogenetic trees of Redmond et al. (2011) place these taxa in two divergent clades, neither of them corresponding to a presently defined genus. We also recovered these two divergent clades in our phylogenies.

Unusual mitochondrial molecular evolution, lack of robust morphological characters, and uncertain phylogenetic relationships from the phylum to genus level (Nosenko et al. 2013) present challenges to species identification in sponges but the data we have collected here, combined with those from previous studies, support the hypothesis that C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis represent one evolutionarily significant unit. Despite the paraphyly of Callyspongia, these taxa form a clade to the exclusion of other closely related Haliclona species (Figure 4.3). Indels and segregation of SNPs in 28S support the species status of C. fallax, C. tenerrima, and C. plicifera but not of C. armigera, C. ?eschrichtii, C. longissima, or C. vaginalis. Indels in ribosomal genes can provide synapomorhies in the marine Haplosclerids (Redmond & McCormack 2008) and 28S shows strong phylogenetic signal for distinguishing species in this order (Redmond et al. 2011). Finally, the rapidly evolving nuclear protein coding gene regions mep and fil (DeBiasse et al. 2014) also support the hypothesis that C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis belong to one lineage.

4.4.2. Conflicts between morphological and molecular species boundaries in the Porifera

Mismatches between morphological species definitions and molecular data are common in sponges and fall across a wide spectrum of discordance. Many examples exist where divergent taxa have identical morphology or where a single lineage is represented by many different morphotypes. Phylogenetic studies based on molecular markers, particularly those examining cosmopolitan species, reveal deeply divided lineages among taxa that are morphologically indistinguishable (Reveillaud et al. 2010; reviewed in Xavier et al. 2010). A phylogeographic investigation across the range of C. vaginalis uncovered a morphologically cryptic species in Central America (Chapter 3). Other examples demonstrate that morphological variation in sponges does not correspond to genetic lineages. In the excavating sponge genus Cliona, multiple growth forms occur within genetically divergent lineages, suggesting these morphologies represent developmental stages or adaptive phenotypes, not different species (Xavier et al. 2010; de Paula et al. 2012; Escobar et al. 2012). Andreakis et al. (2012) found cryptic genetic lineages in the elephant ear sponge Ianthella basta did not correspond to color difference among individuals. Similar to our results for Callyspongia, four different morphotypes in the orange icing sponge Mycale laevis formed a monophyletic group with no genetic differentiation among them (Loh et al. 2012). Such a pattern has also been noted for several coral species as well (Forsman et al. 2010; Budd et al. 2012; Prada et al. 2014a).

48 Morphological differences in the absence of genetic divergence may be the result of biotic and abiotic factors. For example, the different growth forms in M. laevis mentioned above are caused by differences in predation pressure (Loh & Pawlik 2009). Water energy influences morphology in the sponges Halichondria panacea (Palumbi 1984, 1986) and Cliona celata (Bell et al. 2002) and wave action has been implicated in causing the fan shaped morphology of C. vaginalis, tends to be found in areas of high surge (VP Richards, personal communication; Humann & DeLoach 2003). Fine scale microhabitat data were not collected as part of this study but subtle environmental differences may shape differences in morphology among the Callyspongia taxa examined here.

We found C. armigera and C. vaginalis to be genetically indistinguishable but previous work showed significant differences in growth and reproductive strategy between these taxa (Leong & Pawlik 2010; Leong & Pawlik 2011). Callyspongia vaginalis relies on recruitment through larval dispersal while C. armigera propagates primarily through asexual fragmentation. When larvae are present in C. armigera they are smaller than those found in C. vaginalis (670 µm versus 830 µm). Leong and Pawlik (2011) found Callyspongia vaginalis was reproductively active during the summer months with a secondary peak of activity in December but seasonality was difficult to determine in C. armigera; of 60 individuals surveyed from May to October, only two were found to be reproductively active. Comparisons of growth rates suggested that C. vaginalis invests more resources in sexual reproduction while C. armigera invests more in growth (Leong & Pawlik 2010). Although we detected no evidence of genetic differentiation between C. vaginalis and C. armigera, contrasting reproductive strategies may drive, or be driven by, isolation at genetic loci other than those surveyed here. More data are needed to determine whether differences in reproductive strategy are indirectly caused by differences in morphology (i.e. as a rope sponge C. armigera is more prone to fragmentation than the branch tube sponge C. vaginalis and devotes less tissue to larval brooding) or if reproductive differences have a genetic basis.

4.4.3. Hidden genetic divergence in Callyspongia: islands of divergence?

Our survey did not reveal genetic differentiation among C. armigera, C. ?eschrichtii, C. longissima, and C. vaginalis, but we cannot rule out the possibility that there are areas of divergence in other regions of the genome, perhaps at loci that determine phenotype or reproductive isolation. Such islands of divergence have been documented in other species and are believed to be the first step in the process of speciation with gene flow (Wu 2001; Feder et al. 2012). For example, less than 0.5% of the genome is genetically differentiated between marine and freshwater threespine stickleback ecotypes and divergent regions are associated with ecologically adapted traits (Jones et al. 2012). Martin et al. (2013) found genomic differential was strongest near loci determining divergent wing patterns in Heliconius butterflies. In the passerine bird species Manacus candei and M. vitellinus regions associated with reproductive isolation were scattered across the genome (Parchman et al. 2013). Renaut et al. (2013) demonstrated that genomic islands of divergence were the result of reduced recombination in Helianthus sunflowers. A genome wide examination of Callyspongia in conjunction with analysis of microhabitat data would shed light on whether differences in morphology are driven by

49 environment or by divergent regions of the genome other than the ones surveyed here.

Sponges present a challenging, yet fascinating, system for testing hypotheses of species boundaries. The potential to untangle the combined influences of molecular evolution, environment, and life history on the evolution of morphology and phylogeny will undoubtedly increase as more genomic data become available for this group (Riesgo et al. 2014).

50 Chapter 5. Conclusions

Sponges are an ecologically and evolutionarily diverse and important group yet there have been few data available about how well morphological and genetic parameters define sponge species as evolutionarily independent entities. The broad goals for my dissertation were to determine how genetic variation is partitioned among species in the genus Callyspongia and among individuals within the species C. vaginalis. Using data collected from existing mitochondrial DNA primers and nuclear gene regions I developed with next generation sequencing methods, I tested competing models of population history to explain intraspecific patterns of genetic diversity and geographic subdivision in C. vaginalis and tested how well morphology corresponds to evolutionarily distinct lineages.

In Chapter 2 I found discordant patterns of genetic diversity and population structure in the mitochondrial and nuclear genomes in C. vaginalis. Using neutrality tests and coalescent simulations in a hypothesis-testing framework, I showed this mito-nuclear discordance was not the result of selection on mtDNA or incomplete lineage sorting of nucDNA but was consistent with past changes in population size or sperm-mediated gene flow. I found the summary statistics ΦST, deep coalescent events (DCEs), and s were superior in revealing incomplete lineage sorting than theta and pi because their values better trace the accumulation of sequence differences among populations (ΦST) and lineages (DCEs and s). Some authors have downplayed the existence of mito-nuclear discordance in marine systems (Bowen et al. 2014), but examples where nuclear and mitochondrial loci reveal different genetic patterns and historical demographic events do exist (Pardini et al. 2001; Eytan & Hellberg 2010; DeBiasse et al. 2014). Observations of mito-nuclear discordance will likely become more common as the number of studies using sufficient sequence data from both genomes increases. As this happens, researchers should move towards testing hypotheses about the drivers of discordance (Toews & Brelsford 2012), rather than ascribing them to the popular explanation of the day (as has been the case for lineage sorting). To facilitate this shift, I developed a procedure by which researchers working on any taxon can statistically test whether mito-nuclear discordance is the result of incomplete lineage sorting using the summary statistics most informative for their data.

Interestingly, I found that patterns of mito-nuclear discordance can vary within species. For example, along the Florida reef tract, there was discordance between the mitochondrial and nuclear DNA. However, at large spatial scales spanning the Caribbean, genetic patterns of subdivision in the mitochondrial and nuclear genomes of C. vaginalis were largely concordant and corresponded to geography. These results suggest that processes acting across different spatial or temporal scales can influence the agreement between patterns of subdivision in the mitochondrial and nuclear genomes within a species. Across species I found phylogeographic concordance for the sponge and three of its invertebrate commensals. All taxa were divided west-east, with a break at the Straits of Florida. The sponge and two species of amphipod showed secondary substructure within the eastern region and Central America was strongly isolated from the remainder

51 of the Caribbean for the sponge and the brittle star. Taken together, these results warn against making generalizations about the concordance (or discordance) of genetic data i) between genomes within species, ii) among individuals within species, or iii) among co- distributed taxa without explicit explorations of such patterns on a case-by-case basis.

Model-based species delimitation methods revealed evolutionarily distinct lineages are not concordant with current morphological species designations in Callyspongia. This discordance was multi-faceted. For example, while I identified a morphologically indistinguishable, yet genetically divergent lineage in C. vaginalis in Central America (Chapter 3), four taxa with different morphologies (C. armigera, C. longissima, C. ?eschrichtii and C. vaginalis) shared alleles and genetic distances within these taxa overlapped with distances among them. Model-based methods of species delimitation supported collapsing these four into a single evolutionary lineage. Single-locus gene trees showed C. fallax, C. tenerrima, and C. plicifera were reciprocally monophyletic and genetic distance among these species exceed those within taxa, suggesting good species do exist within the genus Callyspongia (Chapter 4). These results highlight the complex relationship between morphology and evolutionary lineage that is likely prevalent in across the Porifera and in many other understudied coral reel taxa with simple or plastic morphological characters (Prada et al. 2014a).

In its entirety, my dissertation provides data on what defines a sponge species and identifies factors responsible for shaping genetic diversity within and among sponges at the population and species levels. My results motivate several questions for future research, including whether morphological differences in the C. vaginalis clade are the result of microhabitat differences shaping phenotypic plasticity and whether islands of isolation, perhaps at loci underlying sponge shape or reproductive isolation, exist in genomic regions other than the ones surveyed here. As the development of genomic resources for sponges continues, I feel the Porifera will be an excellent model for testing questions about ecology and evolution in coral reef invertebrates specifically and animal taxa generally.

52 References

Anderson DR (2008) Model based inference in the life sciences: a primer on evidence Springer.

Andrade CA, Barton ED (2000) Eddy development and motion in the Caribbean Sea. Journal of Geophysical Research: Oceans 105, 26191-26201.

Andras JP, Kirk NL, Drew Harvell C (2011) Range-wide population genetic structure of Symbiodinium associated with the Caribbean sea fan coral, Gorgonia ventalina. Molecular Ecology 20, 2525-2542.

Andras JP, Rypien KL, Harvell CD (2013) Range-wide population genetic structure of the Caribbean sea fan coral, Gorgonia ventalina. Molecular Ecology 22, 56-73.

Andreakis N, Luter HM, Webster NS (2012) Cryptic speciation and phylogeographic relationships in the elephant ear sponge Ianthella basta (Porifera, Ianthellidae) from northern Australia. Zoological Journal of the Linnean Society 166, 225-235.

Barber PH, Palumbi SR, Erdmann MV, Moosa MK (2000) Biogeography: a marine Wallace's line? Nature 406, 692-693.

Baringer MON, Larsen JC (2001) Sixteen years of Florida Current transport at 27 N. Geophysical Research Letters 28, 3179-3182.

Barnes DK, Bell JJ (2002) Coastal sponge communities of the West Indian Ocean: morphological richness and diversity. African Journal of Ecology 40, 350-359.

Baums I, Johnson M, Devlin-Durante M, Miller M (2010) Host population genetic structure and zooxanthellae diversity of two reef-building coral species along the Florida Reef Tract and wider Caribbean. Coral Reefs 29, 835-842.

Baums IB, Miller MW, Hellberg ME (2005) Regionally isolated populations of an imperiled Caribbean coral, Acropora palmata. Molecular Ecology 14, 1377-1390.

Baums IB, Paris CB, Chérubin LM (2006) A bio-oceanographic filter to larval dispersal in a reef-building coral. Limnology and Oceanography 51, 1969-1981.

Bell J, Barnes D, Turner J (2002) The importance of micro and macro morphological variation in the adaptation of a sublittoral demosponge to current extremes. Marine Biology 140, 75-81.

Bell JJ, Davy SK, Jones T, Taylor MW, Webster NS (2013) Could some coral reefs become sponge reefs as our climate changes? Global Change Biology 19, 2613- 2624.

53 Bentlage B, Wörheide G (2007) Low genetic structuring among Pericharax heteroraphis (Porifera: Calcarea) populations from the Great Barrier Reef (Australia), revealed by analysis of nrDNA and nuclear intron sequences. Coral Reefs 26, 807-816.

Bergquist P (1978) Sponges Hutchinson and Company, London.

Betancur-R R, Acero A, Duque-Caro H, Santos SR (2010) Phylogenetic and morphologic analyses of a coastal fish reveals a marine biogeographic break of terrestrial origin in the southern Caribbean. PLoS One 5, e11566.

Bickford D, Lohman DJ, Sodhi NS, et al. (2007) Cryptic species as a window on diversity and conservation. Trends in Ecology & Evolution 22, 148-155.

Birky CW (2001) The inheritance of genes in mitochondria and chloroplasts: laws, mechanisms, and models. Annual Review of Genetics 35, 125-148.

Bishop J (1998) Fertilization in the sea: are the hazards of broadcast spawning avoided when free-spawned sperm fertilize retained eggs? Proceedings of the Royal Society of London. Series B: Biological Sciences 265, 725-731.

Bishop J, Pemberton A (2006) The third way: spermcast mating in sessile marine invertebrates. Integrative and Comparative Biology 46, 398-406.

Blanquer A, Uriz M (2007) Cryptic speciation in marine sponges evidenced by mitochondrial and nuclear genes: a phylogenetic approach. Molecular Phylogenetics and Evolution 45, 392-397.

Blanquer A, Uriz MJ (2010) Population genetics at three spatial scales of a rare sponge living in fragmented habitats. BMC Evolutionary Biology 10, 13.

Blanquer A, Uriz MJ, Caujapé-Castells J (2009) Small-scale spatial genetic structure in Scopalina lophyropoda, an encrusting sponge with philopatric larval dispersal and frequent fission and fusion events. Marine Ecology Progress Series 380, 95-102.

Bowen BW, Shanker K, Yasuda N, et al. (2014) Phylogeography unplugged: comparative surveys in the genomic era. Bulletin of Marine Science 90, 13-46.

Brazeau DA, Sammarco PW, Gleason DF (2005) A multi-locus genetic assignment technique to assess sources of Agaricia agaricites larvae on coral reefs. Marine Biology 147, 1141-1148.

Briggs JC (1984) Freshwater fishes and biogeography of Central America and the Antilles. Systematic Ziology 33, 428-435.

Brown WM, George M, Wilson AC (1979) Rapid evolution of animal mitochondrial DNA. Proceedings of the National Academy of Sciences 76, 1967-1971.

54 Budd AF, Nunes FL, Weil E, Pandolfi JM (2012) Polymorphism in a common Atlantic reef coral (Montastraea cavernosa) and its long-term evolutionary implications. Evolutionary Ecology 26, 265-290.

Butler I, Hunt J, Herrnkind W, et al. (1995) Cascading disturbances in Florida Bay, USA: cyanobacteria blooms, sponge mortality, and implications for juvenile spiny lobsters Panulirus argus. Marine Ecology Progress Series 129, 119-125.

Campbell C (1965) The Santa Marta wrench fault of Colombia and its regional setting. In: Caribbean Geological Conference IV, pp. 247-260, Trinidad.

Carlin J, Robertson DR, Bowen BW (2003) Ancient divergences and recent connections in two tropical Atlantic reef fishes Epinephelus adscensionis and Rypticus saponaceous (Percoidei: Serranidae). Marine Biology 143, 1057-1069.

Carstens BC, Dewey TA (2010) Species delimitation using a combined coalescent and information-theoretic approach: an example from North American Myotis bats. Systematic Biology 59, 400-414.

Carstens BC, Satler JD (2013) The carnivorous plant described as Sarracenia alata contains two cryptic species. Biological Journal of the Linnean Society 109, 737- 746.

Cheviron ZA, Brumfield RT (2009) Migration-selection balance and local adaptation of mitochondrial haplotypes in rufous-collared sparrows (Zonotrichia capensis) along an elevational gradient. Evolution 63, 1593-1605.

Clement M, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Molecular Ecology 9, 1657-1659.

Colin PL (1995) Surface currents in Exuma Sound, Bahamas and adjacent areas with reference to potential larval transport. Bulletin of Marine Science 56, 48-57.

Colin PL (1996) Longevity of some coral reef fish spawning aggregations. Copeia 1996, 189-192.

Colin PL (2002) A new species of sponge-dwelling Elacatinus (Pisces: Gobiidae) from the western Caribbean. Zootaxa 106, 1-7.

Collette BB (1983) Two new species of coral toadfishes, family Batrachoididae, genus Sanopus, from Yucatan, Mexico, and Belize. Proceedings of the Biological Society of Washington 96, 719-724.

Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science 199, 1302- 1310.

55 Cowart J, Henkel T, McMurray S, Pawlik J (2006) Sponge orange band (SOB): a pathogenic-like condition of the giant barrel sponge, Xestospongia muta. Coral Reefs 25, 513-513.

Cowen R, Paris C, Srinivasan A (2006) Scaling of connectivity in marine populations. Science 311, 522-527.

Crandall ED, Jones ME, Munoz MM, et al. (2008) Comparative phylogeography of two seastars and their ectosymbionts within the Coral Triangle. Molecular Ecology 17, 5276-5290.

Dailianis T, Tsigenopoulos C, Dounas C, Voultsiadou E (2011) Genetic diversity of the imperilled bath sponge Spongia officinalis Linnaeus, 1759 across the Mediterranean Sea: patterns of population differentiation and implications for taxonomy and conservation. Molecular Ecology 20, 3757-3772.

Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9, 772-772. de Goeij JM, van Oevelen D, Vermeij MJ, et al. (2013) Surviving in a marine desert: the sponge loop retains resources within coral reefs. Science 342, 108-110. de Paula TS, Zilberberg C, Hajdu E, Lôbo-Hajdu G (2012) Morphology and molecules on opposite sides of the diversity gradient: Four cryptic species of the Cliona celata (Porifera, Demospongiae) complex in South America revealed by mitochondrial and nuclear markers. Molecular Phylogenetics and Evolution 62, 529-541.

DeBiasse M, Nelson B, Hellberg M (2014) Evaluating summary statistics used to test for incomplete lineage sorting: mito-nuclear discordance in the reef sponge Callyspongia vaginalis. Molecular Ecology 23, 225-238.

DeBiasse M, Richards V, Shivji M (2010) Genetic assessment of connectivity in the common reef sponge, Callyspongia vaginalis (Demospongiae: Haplosclerida) reveals high population structure along the Florida reef tract. Coral Reefs 29, 47- 55.

Degnan JH, Rosenberg NA (2009) Gene tree discordance, phylogenetic inference and the multispecies coalescent. Trends in Ecology & Evolution 24, 332-340.

DeSalle R, Egan MG, Siddall M (2005) The unholy trinity: taxonomy, species delimitation and DNA barcoding. Philosophical Transactions of the Royal Society B: Biological Sciences 360, 1905-1916.

Doney SC, Fabry VJ, Feely RA, Kleypas JA (2009) Ocean acidification: the other CO2 problem. Annual Review of Marine Science 1, 169-192.

56 Doney SC, Schimel DS (2007) Carbon and climate system coupling on timescales from the Precambrian to the Anthropocene*. Annual Review of the Environment and Resources 32, 31-66.

Drummond A, Ashton B, Buxton S, et al. (2012) Geneious v4.5.

Duchassaing P, Michelotti G (1864) Spongiaires de la mer Caraïbe. Natuurkundige verhandelingen van de Hollandsche maatschappij der wetenschappen te Haarlem.

Dulvy NK, Freckleton RP, Polunin NV (2004) Coral reef cascades and the indirect effects of predator removal by exploitation. Ecology Letters 7, 410-416.

Duminil J, Kenfack D, Viscosi V, Grumiau L, Hardy OJ (2011) Testing species delimitation in sympatric species complexes: The case of an African tropical tree, Carapa spp. (Meliaceae). Molecular Phylogenetics and Evolution 62, 275-285.

Duran S, Giribet G, Turon X (2004a) Phylogeographical history of the sponge Crambe crambe (Porifera, Poecilosclerida): range expansion and recent invasion of the Macaronesian islands from the Mediterranean Sea. Molecular Ecology 13, 109- 122.

Duran S, Pascual M, Estoup A, Turon X (2004b) Strong population structure in the marine sponge Crambe crambe (Poecilosclerida) as revealed by microsatellite markers. Molecular Ecology 13, 511-522.

Duran S, Pascual M, Turon X (2004c) Low levels of genetic variation in mtDNA sequences over the western Mediterranean and Atlantic range of the sponge Crambe crambe (Poecilosclerida). Marine Biology 144, 31-35.

Duran S, Rützler K (2006) Ecological speciation in a Caribbean marine sponge. Molecular Phylogenetics and Evolution 40, 292-297.

Earl DA, vonHoldt BM (2011) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4, 1-3.

Edwards DL, Knowles LL (2014) Species detection and individual assignment in species delimitation: can integrative data increase efficacy? Proceedings of the Royal Society B: Biological Sciences 281.

Ence DD, Carstens BC (2011) SpedeSTEM: a rapid and accurate method for species delimitation. Molecular Ecology Resources 11, 473-480.

Erpenbeck D, Duran S, Rützler K, et al. (2007) Towards a DNA taxonomy of Caribbean Demosponges: a gene tree reconstructed from partial mitochondrial CO1 gene sequences supports previous rDNA phylogenies and provides a new perspective

57 on the systematics of Demospongiae. Journal of the Marine Biological Association of the United Kingdom 87, 1563-1570.

Erpenbeck D, Hooper J, Wörheide G (2006) CO1 phylogenies in diploblasts and the 'Barcoding of Life'- are we sequencing a suboptimal partition? Molecular Ecology Notes 6, 550-553.

Erwin PM, Thacker RW (2007) Phylogenetic analyses of marine sponges within the order Verongida: a comparison of morphological and molecular data. Invertebrate Biology 126, 220-234.

Escobar D, Zea S, Sánchez JA (2012) Phylogenetic relationships among the Caribbean members of the Cliona viridis complex (Porifera, Demospongiae, Hadromerida) using nuclear and mitochondrial DNA sequences. Molecular Phylogenetics and Evolution 64, 271-284.

Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14, 2611-2620.

Evans WJ (2012) Distribution and abundance of Ophiothrix suensonii on sponge vs. non- sponge habitat at Whale Shoals patch reef, Belize. Papers and Publications: Interdisciplinary Journal of Undergraduate Research 1, 1-9.

Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0): an integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1, 47.

Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131, 479-491.

Eyre-Walker A (2002) Changing effective population size and the McDonald-Kreitman test. Genetics 162, 2017-2024.

Eytan RI, Hayes M, Arbour-Reily P, Miller M, Hellberg ME (2009) Nuclear sequences reveal mid-range isolation of an imperilled deep-water coral population. Molecular Ecology 18, 2375-2389.

Eytan RI, Hellberg ME (2010) Nuclear and mitochondrial sequence data reveal and conceal different demographic histories and population genetic processes in Caribbean reef fishes. Evolution 64, 3380-3397.

Fay JC, Wu CI (1999) A human population bottleneck can account for the discordance between patterns of mitochondrial versus nuclear DNA variation. Molecular Biology and Evolution 16, 1003-1005.

58 Feder JL, Egan SP, Nosil P (2012) The genomics of speciation-with-gene-flow. Trends in Genetics 28, 342-350.

Flot JF (2007) Champuru 1.0: a computer software for unraveling mixtures of two DNA sequences of unequal lengths. Molecular Ecology Notes 7, 974-977.

Fogarty ND, Vollmer SV, Levitan DR (2012) Weak prezygotic isolating mechanisms in threatened Caribbean Acropora corals. PLoS One 7, e30486.

Folmer O, Black M, Hoeh W, Lutz R, Vrjenhoek R (1994) DNA primers for amplification of mitochondrial cytochrome C oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 294- 299.

Forsman ZH, Concepcion GT, Haverkort RD, et al. (2010) Ecomorph or endangered Coral? DNA and microstructure reveal Hawaiian species complexes: Montipora dilatata/flabellata/turgescens & M. patula/verrilli. PLoS One 5, e15021.

Funk DJ, Omland KE (2003) Species-level paraphyly and polyphyly: frequency, causes, and consequences, with insights from animal mitochondrial DNA. Annual Review of Ecology, Evolution, and Systematics 34, 397-423.

Galindo HM, Olson DB, Palumbi SR (2006) Seascape genetics: a coupled oceanographic-genetic model predicts population structure of Caribbean corals. Current Biology 16, 1622-1626.

Giles EC, Kamke J, Moitinho-Silva L, et al. (2012) Bacterial community profiles in low microbial abundance sponges. FEMS Microbiology Ecology 83, 232-241.

Glynn PW (1996) Coral reef bleaching: facts, hypotheses and implications. Global Change Biology 2, 495-509.

Graham SW, Olmstead RG, Barrett SC (2002) Rooting phylogenetic trees with distant outgroups: a case study from the commelinoid monocots. Molecular Biology and Evolution 19, 1769-1781.

Grosberg RK (1991) Sperm-mediated gene flow and the genetic structure of a population of the colonial ascidian Botryllus schlosseri. Evolution 45, 130-142.

Hale MC, McCormick CR, Jackson JR, DeWoody JA (2009) Next-generation pyrosequencing of gonad transcriptomes in the polyploid lake sturgeon (Acipenser fulvescens): the relative merits of normalization and rarefaction in gene discovery. BMC Genomics 10, 203.

Halpern BS, Walbridge S, Selkoe KA, et al. (2008) A global map of human impact on marine ecosystems. Science 319, 948-952.

59 Hamby RK, Zimmer EA (1992) Ribosomal RNA as a phylogenetic tool in plant systematics. In: Molecular systematics of plants, pp. 50-91. Springer.

Hare MP (2001) Prospects for nuclear gene phylogeography. Trends in Ecology & Evolution 16, 700-706.

Harrington RC, Near TJ (2011) Phylogenetic and Coalescent Strategies of Species Delimitation in Snubnose Darters (Percidae: Etheostoma). Systematic Biology 61, 63-79.

Harrison PL, Babcock RC, Bull GD, et al. (1984) Mass spawning in tropical reef corals. Science 223, 1186-1189.

Hart MW, Sunday J (2007) Things fall apart: biological species form unconnected parsimony networks. Biology Letters 3, 509-512.

Heled J, Drummond A (2010) Bayesian inference of species trees from multilocus data. Molecular Biology and Evolution 27, 570-580.

Heled J, Drummond AJ (2008) Bayesian inference of population size history from multiple loci. BMC Evolutionary Biology 8, 289.

Hellberg M (2006) No variation and low synonymous substitution rates in coral mtDNA despite high nuclear variation. BMC Evolutionary Biology 6, 24.

Hellberg ME (2009) Gene flow and isolation among populations of marine animals. Annual Review of Ecology, Evolution, and Systematics 40, 291-310.

Hemond EM, Vollmer SV (2010) Genetic diversity and connectivity in the threatened staghorn coral (Acropora cervicornis) in Florida. PLoS One 5, e8652.

Henkel T, Pawlik J (2005) Habitat use by sponge-dwelling brittlestars. Marine Biology 146, 301-313.

Heyman WD, Kjerfve B (2000) The Gulf of Honduras. In: Coastal Ecosystems of Latin America (eds. Seeliger U, Kjerfve B). Elsevier.

Hickerson MJ, Cunningham CW (2005) Contrasting quaternary histories in an ecologically divergent sister pair of low-dispersing intertidal fish (Xiphister) revealed by multilocus DNA analysis. Evolution 59, 344-360.

Hickerson MJ, Stahl E, Takebayashi N (2007) msBayes: pipeline for testing comparative phylogeographic histories using hierarchical approximate Bayesian computation. BMC Bioinformatics 8, 268.

Hillis DM, Dixon MT (1991) Ribosomal DNA: molecular evolution and phylogenetic inference. Quarterly Review of Biology 66, 411-453.

60 Hird S, Kubatko L, Carstens B (2010) Rapid and accurate species tree estimation for phylogeographic investigations using replicated subsampling. Molecular Phylogenetics and Evolution 57, 888-898.

Hoegh-Guldberg O, Bruno JF (2010) The impact of climate change on the world's marine ecosystems. Science 328, 1523-1528.

Hooper JNA, Van Soest RWM, Willenz P (2002) Systema Porifera: a guide to the classification of sponges Kluwer Academic/Plenum Publishers.

Huang D, Meier R, Todd PA, Chou LM (2008) Slow mitochondrial COI sequence evolution at the base of the metazoan tree and its implications for DNA barcoding. Journal of Molecular Evolution 66, 167-174.

Huang W, Takebayashi N, Qi Y, Hickerson MJ (2011) MTML-msBayes: approximate Bayesian comparative phylogeographic inference from multiple taxa and multiple loci with rate heterogeneity. BMC Bioinformatics 12, 1.

Hudson RR (2002) Generating samples under a Wright-Fisher neutral model of genetic variation. Bioinformatics 18, 337-338.

Hudson RR, Kreitman M, Aguadé M (1987) A test of neutral molecular evolution based on nucleotide data. Genetics 116, 153-159.

Huelsenbeck JP, Andolfatto P, Huelsenbeck ET (2011) Structurama: Bayesian inference of population structure. Evolutionary Bioinformatics Online 7, 55.

Humann P, DeLoach N (2003) Reef Creature Identification New World Publications Inc., Jacksonville, Florida.

Hurst GDD, Jiggins FM (2005) Problems with mitochondrial DNA as a marker in population, phylogeographic and phylogenetic studies: the effects of inherited symbionts. Proceedings of the Royal Society B: Biological Sciences 272, 1525- 1534.

Irwin DE (2002) Phylogeographic breaks without geographic barriers to gene flow. Evolution 56, 2383-2394.

Itskovich V, Belikov S, Efremova S, et al. (2007) Phylogenetic relationships between freshwater and marine Haplosclerida (Porifera, Demospongiae) based on the full length 18S rRNA and partial COXI gene sequences. Porifera Research: Biodiversity, Innovation and Sustainability. Museu Nacional, Rio de Janeiro, 383- 391.

Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23, 1801-1806.

61 Johnson GD, Brothers EB (1989) Acanthemblemaria paula, a new diminutive Chaenopsid (Pisces, Blennioidei) from Belize, with comments on life-history. Proceedings of the Biological Society of Washington 102, 1018-1030.

Jones FC, Grabherr MG, Chan YF, et al. (2012) The genomic basis of adaptive evolution in threespine sticklebacks. Nature 484, 55-61.

Jones G, Milicich M, Emslie M, Lunow C (1999) Self-recruitment in a coral reef fish population. Nature 402, 802-804.

Kayal E, Lavrov DV (2008) The mitochondrial genome of Hydra oligactis (Cnidaria, Hydrozoa) sheds new light on animal mtDNA evolution and cnidarian phylogeny. Gene 410, 177-186.

Klautau M, Russo CAM, Lazoski C, et al. (1999) Does cosmopolitanism result from overconservative systematics? A case study using the marine sponge Chondrilla nucula. Evolution 53, 1414-1422.

Knowles LL (2009) Statistical phylogeography. Annual Review of Ecology, Evolution, and Systematics 40, 593-612.

Laboy-Nieves EN, Klein E, Conde JE, et al. (2001) Mass mortality of tropical marine communities in Morrocoy, Venezuela. Bulletin of Marine Science 68, 163-179.

Lasker HR, Gutierrez-Rodriguez C, Bala K, Hannes A, Bilewitch JP (2008) Male reproductive success during spawning events of the octocoral Pseudopterogorgia elisabethae. Marine Ecology Progress Series 367, 153-161.

Lavrov DV, Wang X, Kelly M (2008) Reconstructing ordinal relationships in the Demospongiae using mitochondrial genomic data. Molecular Phylogenetics and Evolution 49, 111-124.

Leaché AD, Fujita MK (2010) Bayesian species delimitation in West African forest geckos (Hemidactylus fasciatus). Proceedings of the Royal Society B: Biological Sciences 277, 3071-3077.

Leal MC, Madeira C, Brandão CA, Puga J, Calado R (2012) Bioprospecting of marine invertebrates for new natural products- a chemical and zoogeographical perspective. Molecules 17, 9842-9854.

Lee T, Ó Foighil D (2004) Hidden Floridian biodiversity: mitochondrial and nuclear gene trees reveal four cryptic species within the scorched mussel, Brachidontes exustus, species complex. Molecular Ecology 13, 3527-3542.

Leong W, Pawlik JR (2010) Fragments or propagules? Reproductive tradeoffs among Callyspongia spp. from Florida coral reefs. Oikos 119, 1417-1422.

62 Leong W, Pawlik JR (2011) Comparison of reproductive patterns among 7 Caribbean sponge species does not reveal a resource trade-off with chemical defenses. Journal of Experimental Marine Biology and Ecology 401, 80-84.

Levitan DR, Petersen C (1995) Sperm limitation in the sea. Trends in Ecology & Evolution 10, 228-231.

Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451-1452.

Lindquist N, Bolser R, Laing K (1997) Timing of larval release by two Caribbean demosponges. Marine Ecology Progress Series 155, 309-313.

Loh T-L, López-Legentil S, Song B, Pawlik JR (2012) Phenotypic variability in the Caribbean orange icing sponge Mycale laevis (Demospongiae: Poecilosclerida). Hydrobiologia 687, 205-217.

Loh T-L, Pawlik JR (2009) Bitten down to size: Fish predation determines growth form of the Caribbean coral reef sponge Mycale laevis. Journal of Experimental Marine Biology and Ecology 374, 45-50.

López-Legentil S, Erwin PM, Henkel TP, Loh TL, Pawlik JR (2010) Phenotypic plasticity in the Caribbean sponge Callyspongia vaginalis (Porifera: Haplosclerida). Scientia Marina 74, 445-453.

López-Legentil S, Pawlik J (2009) Genetic structure of the Caribbean giant barrel sponge Xestospongia muta using the I3-M11 partition of COI. Coral Reefs 28, 157-165.

Losos JB, Jackman TR, Larson A, de Queiroz K, Rodríguez-Schettino L (1998) Contingency and determinism in replicated adaptive radiations of island lizards. Science 279, 2115-2118.

Ludeman DA, Farrar N, Riesgo A, Paps J, Leys SP (2014) Evolutionary origins of sensation in metazoans: functional evidence for a new sensory organ in sponges. BMC Evolutionary Biology 14, 3.

Maddison WP, Maddison DR (2010) Mesquite: a modular system for evolutionary analysis.

Maldonado M (2006) The ecology of the sponge larva. Canadian Journal of Zoology 84, 175-194.

Maritz K, Calcino A, Fahey B, Degnan B, Degnan SM (2010) Remarkable consistency of larval release in the spermcast-mating demosponge Amphimedon queenslandica (Hooper and van Soest). Open Marine Biology Journal 4, 57-64.

63 Martin SH, Dasmahapatra KK, Nadeau NJ, et al. (2013) Genome-wide evidence for speciation with gene flow in Heliconius butterflies. Genome Research 23, 1817- 1828.

McDonald JH, Kreitman M (1991) Adaptive protein evolution at the Adh locus in Drosophila. Nature 351, 652-654.

McKay BD, Zink RM (2010) The causes of mitochondrial DNA gene tree paraphyly in birds. Molecular Phylogenetics and Evolution 54, 647-650.

McMullin ER, Hourdez S, Schaeffer SW, Fisher CR (2003) Phylogeny and biogeography of deep sea vestimentiferan tubeworms and their bacterial symbionts. Symbiosis 34, 1-41.

Meiklejohn CD, Montooth KL, Rand DM (2007) Positive and negative selection on the mitochondrial genome. Trends in Genetics 23, 259-263.

Meirmans PG, Van Tienderen PH (2004) GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes 4, 792-794.

Miller K, Alvarez B, Battershill C, Northcote P, Parthasarathy H (2001) Genetic, morphological, and chemical divergence in the sponge genus Latrunculia (Porifera: Demospongiae) from New Zealand. Marine Biology 139, 235-250.

Milne I, Wright F, Rowe G, et al. (2004) TOPALi: software for automatic identification of recombinant sequences within DNA multiple alignments. Bioinformatics 20, 1806-1807.

Miloslavich P, Díaz JM, Klein E, et al. (2010) Marine biodiversity in the Caribbean: regional estimates and distribution patterns. PLoS One 5, e11916.

Mindell DP, Honeycutt RL (1990) Ribosomal RNA in vertebrates: evolution and phylogenetic applications. Annual Review of Ecology and Systematics, 541-566.

Mladenov PV (1983) Breeding patterns of three species of Caribbean brittle stars (Echinodermata: Ophiuroidea). Bulletin of Marine Science 33, 363-372.

Moberg F, Folke C (1999) Ecological goods and services of coral reef ecosystems. Ecological Economics 29, 215-233.

Morrison CL, Rios R, Emmett Duffy J (2004) Phylogenetic evidence for an ancient rapid radiation of Caribbean sponge-dwelling snapping shrimps (Synalpheus). Molecular Phylogenetics and Evolution 30, 563-581.

Mumby PJ, Broad K, Brumbaugh DR, et al. (2008) Coral reef habitats as surrogates of species, ecological functions, and ecosystem services. Conservation Biology 22, 941-951.

64 Nagelkerken I, Van der Velde G, Gorissen M, et al. (2000) Importance of mangroves, seagrass beds and the shallow coral reef as a nursery for important coral reef fishes, using a visual census technique. Estuarine, Coastal and Shelf Science 51, 31-44.

Neigel J, Domingo A, Stake J (2007) DNA barcoding as a tool for coral reef conservation. Coral Reefs 26, 487-499.

Nielsen R (2001) Statistical tests of selective neutrality in the age of genomics. Heredity 86, 641-647.

Nosenko T, Schreiber F, Adamska M, et al. (2013) Deep metazoan phylogeny: when different genes tell different stories. Molecular Phylogenetics and Evolution 67, 223-233.

Nunney L (1993) The influence of mating system and overlapping generations on effective population size. Evolution 47, 1329-1341.

Oaks JR (2014) An Improved Approximate-Bayesian Model-choice Method for Estimating Shared Evolutionary History. arXiv preprint arXiv:1402.6303.

Odum HT, Odum EP (1955) Trophic structure and productivity of a windward coral reef community on Eniwetok Atoll. Ecological Monographs, 291-320.

Oliver J, Babcock R (1992) Aspects of the fertilization ecology of broadcast spawning corals: sperm dilution effects and in situ measurements of fertilization. The Biological Bulletin 183, 409-417.

Orozco-Terwengel P, Nagy ZT, Vieites DR, Vences M, Louis Jr E (2008) Phylogeography and phylogenetic relationships of Malagasy tree and ground boas. Biological Journal of the Linnean Society 95, 640-652.

Orr JC, Fabry VJ, Aumont O, et al. (2005) Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms. Nature 437, 681-686.

Palumbi SR (1984) Tactics of acclimation: morphological changes of sponges in an unpredictable environment. Science 225, 1478-1480.

Palumbi SR (1986) How body plans limit acclimation: responses of a demosponge to wave force. Ecology 67, 208-214.

Pandolfi JM, Bradbury RH, Sala E, et al. (2003) Global trajectories of the long-term decline of coral reef ecosystems. Science 301, 955-958.

Paradis E (2010) pegas: an R package for population genetics with an integrated-modular approach. Bioinformatics 26, 419-420.

65 Parchman T, Gompert Z, Braun M, et al. (2013) The genomic consequences of adaptive divergence and reproductive isolation between species of manakins. Molecular Ecology 22, 3304-3317.

Pardini AT, Jones CS, Noble LR, et al. (2001) Sex-biased dispersal of great white sharks. Nature 412, 139-140.

Parker MA, Doyle JL, Doyle JJ (2004) Comparative phylogeography of Amphicarpaea legumes and their root-nodule symbionts in Japan and North America. Journal of Biogeography 31, 425-434.

Paulay G (1997) Diversity and distribution of reef organisms. In: Life and death of coral reefs (ed. Birkeland C), pp. 298-353. Chapman and Hall, New York.

Peters JL, Bolender KA, Pearce JM (2012) Behavioural vs. molecular sources of conflict between nuclear and mitochondrial DNA: the role of male-biased dispersal in a Holarctic sea duck. Molecular Ecology 21, 3562-3575.

Philippe H, Derelle R, Lopez P, et al. (2009) Phylogenomics revives traditional views on deep animal relationships. Current Biology 19, 706-712.

Pinzon JH, LaJeunesse T (2011) Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Molecular Ecology 20, 311-325.

Pond SLK, Frost SDW (2005) Datamonkey: rapid detection of selective pressure on individual sites of codon alignments. Bioinformatics 21, 2531-2533.

Pond SLK, Posada D, Gravenor MB, Woelk CH, Frost SDW (2006) GARD: a genetic algorithm for recombination detection. Bioinformatics 22, 3096-3098.

Pöppe J, Sutcliffe P, Hooper JN, Wörheide G, Erpenbeck D (2010) COI barcoding reveals new clades and radiation patterns of Indo-Pacific sponges of the family Irciniidae (Demospongiae: Dictyoceratida). PLoS One 5, e9950.

Posada D (2008) jModelTest: phylogenetic model averaging. Molecular Biology and Evolution 25, 1253-1256.

Prada C, DeBiasse M, Neigel J, et al. (2014a) Genetic species delineation among branching Caribbean Porites corals. Coral Reefs in press.

Prada C, Hellberg ME (2013) Long prereproductive selection and divergence by depth in a Caribbean candelabrum coral. Proceedings of the National Academy of Sciences 110, 3961-3966.

Prada C, McIlroy S, Beltrán D, et al. (2014b) Cryptic diversity hides host and habitat specialization in a gorgonian-algal symbiosis. Molecular Ecology in press.

66 Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155, 945.

Raleigh J, Redmond NE, Delahan E, et al. (2007) Mitochondrial Cytochrome oxidase 1 phylogeny supports alternative taxonomic scheme for the marine Haplosclerida. Journal of the Marine Biological Association of the United Kingdom 87, 1577- 1584.

Rambaut A, Grassly NC (1997) Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Computer applications in the biosciences: CABIOS 13, 235-238.

Rauchenberger M (1988) Historical biogeography of poeciliid fishes in the Caribbean. Systematic Zoology 37, 356-365.

Redmond NE, McCormack GP (2008) Large expansion segments in 18S rDNA support a new sponge clade (Class Demospongiae, Order Haplosclerida). Molecular Phylogenetics and Evolution 47, 1090-1099.

Redmond NE, Morrow CC, Thacker RW, et al. (2013) Phylogeny and systematics of Demospongiae in light of new small-subunit rRibosomal DNA (18S) sequences. Integrative and Comparative Biology 53, 388-415.

Redmond NE, Raleigh J, van Soest RWM, et al. (2011) Phylogenetic relationships of the marine Haplosclerida (Phylum Porifera) employing ribosomal (28S rRNA) and mitochondrial (COX1, NAD1) gene sequence data. PLoS One 6, e24344.

Renaut S, Grassa C, Yeaman S, et al. (2013) Genomic islands of divergence are not affected by geography of speciation in sunflowers. Nature Communications 4, 1827.

Reveillaud J, Remerie T, van Soest R, et al. (2010) Species boundaries and phylogenetic relationships between Atlanto-Mediterranean shallow-water and deep-sea coral associated Hexadella species (Porifera, Ianthellidae). Molecular Phylogenetics and Evolution 56, 104-114.

Ribeiro SM, Omena EP, Muricy G (2003) Macrofauna associated to Mycale microsigmatosa (Porifera, Demospongiae) in Rio de Janeiro State, SE Brazil. Estuarine, Coastal and Shelf Science 57, 951-959.

Richards VP (2010) Genetic connectivity and microevolution of reef associated taxa, Nova Southeastern University.

Richards VP, Stanhope MJ, Shivji MS (2012) Island endemism, morphological stasis, and possible cryptic speciation in two coral reef, commensal Leucothoid amphipod species throughout Florida and the Caribbean. Biodiversity and Conservation 21, 343-361.

67 Richards VP, Thomas JD, Stanhope MJ, Shivji MS (2007) Genetic connectivity in the Florida reef system: comparative phylogeography of commensal invertebrates with contrasting reproductive strategies. Molecular Ecology 16, 139-157.

Richardson PL (2005) Caribbean Current and eddies as observed by surface drifters. Deep Sea Research Part II: Topical Studies in Oceanography 52, 429-463.

Riesgo A, Farrar N, Windsor PJ, Giribet G, Leys SP (2014) The analysis of eight transcriptomes from all Poriferan classes reveals surprising genetic complexity in sponges. Molecular Biology and Evolution 31, 1102-1120.

Rittmeyer EN, Austin CC (2012) The effects of sampling on delimiting species from multi-locus sequence data. Molecular Phylogenetics and Evolution 65, 451-463.

Roberts CM, McClean CJ, Veron JE, et al. (2002) Marine biodiversity hotspots and conservation priorities for tropical reefs. Science 295, 1280-1284.

Roe AD, Rice AV, Coltman DW, Cooke JE, Sperling FA (2011) Comparative phylogeography, genetic differentiation and contrasting reproductive modes in three fungal symbionts of a multipartite bark beetle symbiosis. Molecular Ecology 20, 584-600.

Romano SL, Palumbi SR (1996) Evolution of scleractinian corals inferred from molecular systematics. Science 271, 640-641.

Rosenberg NA (2003) The shapes of neutral gene genealogies in two species: probabilities of monophyly, paraphyly, and polyphyly in a coalescent model. Evolution 57, 1465-1477.

Rosenberg NA (2004) DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes 4, 137-138.

Rozas J, Sánchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19, 2496-2497.

Rützler K (1978) Sponges in coral reefs. In: Monographs on oceanographic methodology. Coral reefs: research methods (eds. Stodart DR, Johanes RE), pp. 209-313. UNESECO.

Satler JD, Carstens BC, Hedin M (2013) Multilocus species delimitation in a complex of morphologically conserved trapdoor spiders (Mygalomorphae, Antrodiaetidae, Aliatypus). Systematic Biology 62, 805-823.

Sawyer SA, Hartl DL (1992) Population genetics of polymorphism and divergence. Genetics 132, 1161-1176.

68 Schmitt S, Tsai P, Bell J, et al. (2011) Assessing the complex sponge microbiota: core, variable and species-specific bacterial communities in marine sponges. The ISME Journal 6, 564-576.

Schott FA, Lee TN, Zantopp R (1988) Variability of structure and transport of the Florida Current in the period range of days to seasonal. Journal of Physical Oceanography 18, 1209-1230.

Schröder H, Efremova S, Itskovich V, et al. (2003) Molecular phylogeny of the freshwater sponges in Lake Baikal. Journal of Zoological Systematics and Evolutionary Research 41, 80-86.

Shearer T, Van Oppen M, Romano S, Wörheide G (2002) Slow mitochondrial DNA sequence evolution in the Anthozoa (Cnidaria). Molecular Ecology 11, 2475- 2487.

Srivastava M, Simakov O, Chapman J, et al. (2010) The Amphimedon queenslandica genome and the evolution of animal complexity. Nature 466, 720-726.

Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for haplotype reconstruction from population data. The American Journal of Human Genetics 68, 978-989.

Stone GN, Lohse K, Nicholls JA, et al. (2012) Reconstructing community assembly in time and space reveals enemy escape in a western Palearctic insect community. Current Biology 22, 532-537.

Sumaila UR, Marsden AD, Watson R, Pauly D (2007) A global ex-vessel fish price database: construction and applications. Journal of Bioeconomics 9, 39-51.

Swofford D (2003) PAUP*: Phylogenetic Analysis Using Parsimony (* and Other Methods), Version 4.0 b10. Sinauer, Sunderland, MA.

Tajima F (1989a) The effect of change in population size on DNA polymorphism. Genetics 123, 597-601.

Tajima F (1989b) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123, 585-595.

Tamura K, Peterson D, Peterson N, et al. (2011) MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Molecular Biology and Evolution 28, 2731-2739.

Taylor MS, Hellberg ME (2003) Genetic evidence for local retention of pelagic larvae in a Caribbean reef fish. Science 299, 107-109.

Taylor MS, Hellberg ME (2005) Marine radiations at small geographic scales: speciation in neotropical reef gobies (Elacatinus). Evolution 59, 374-385.

69 Taylor MS, Hellberg ME (2006) Comparative phylogeography in a genus of coral reef fishes: biogeographic and genetic concordance in the Caribbean. Molecular Ecology 15, 695-707.

Taylor MW, Radax R, Steger D, Wagner M (2007) Sponge-associated microorganisms: evolution, ecology, and biotechnological potential. Microbiology and Molecular Biology Reviews 71, 295-347.

Thacker RW, Hill AL, Hill MS, et al. (2013) Nearly complete 28S rRNA gene sequences confirm new hypotheses of sponge evolution. Integrative and Comparative Biology 53, 373-387.

Thomas JD, Klebba KN (2006) Studies of commensal leucothoid amphipods: two new sponge-inhabiting species from south Florida and the western Caribbean. Journal of Crustacean Biology 26, 13-22.

Thompson AR, Thacker CE, Shaw EY (2005) Phylogeography of marine mutualists: parallel patterns of genetic structure between obligate goby and shrimp partners. Molecular Ecology 14, 3557-3572.

Thornhill DJ, Xiang Y, Fitt WK, Santos SR (2009) Reef endemism, host specificity and temporal stability in populations of symbiotic dinoflagellates from two ecologically dominant Caribbean corals. PLoS One 4, e6262.

Toews DPL, Brelsford A (2012) The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology 21, 3907-3930.

Tschanz CM, Marvin RF, Cruz J, Mehnert HH, Cebula GT (1974) Geologic evolution of the Sierra Nevada de Santa Marta, northeastern Colombia. Geological Society of America Bulletin 85, 273-284.

Turner W, Spector S, Gardiner N, et al. (2003) Remote sensing for biodiversity science and conservation. Trends in Ecology & Evolution 18, 306-314. van der Strate HJ, Boele-Bos SA, Olsen JL, Van De Zande L, Stam WT (2002) Phylogeogeographic studies in the tropical seaweed Cladophoropsis membranacea (Chlorophyta, Ulvophyceae) reveal a cryptic species complex. Journal of Phycology 38, 572-582.

Van Soest RWM, Boury-Esnault N, Hooper JNA, et al. (2014) World Porifera Database. www.marinespecies.org/porifera.

Voigt O, Eichmann V, Wörheide G (2012) First evaluation of mitochondrial DNA as a marker for phylogeographic studies of Calcarea: a case study from Leucetta chagosensis. Hydrobiologia 687, 101-106.

Vollmer SV, Palumbi SR (2002) Hybridization and the evolution of reef coral diversity. Science 296, 2023-2025.

70 Vollmer SV, Palumbi SR (2007) Restricted gene flow in the Caribbean staghorn coral Acropora cervicornis: implications for the recovery of endangered reefs. Journal of Heredity 98, 40-50.

Webster NS, Taylor MW, Behnam F, et al. (2009) Deep sequencing reveals exceptional diversity and modes of transmission for bacterial sponge symbionts. Environmental Microbiology 12, 2070-2082.

Weisz JB, Lindquist N, Martens CS (2008) Do associated microbial abundances impact marine demosponge pumping rates and tissue densities? Oecologia 155, 367-376.

Whalan S, De Nys R, Smith-Keune C, et al. (2008) Low genetic variability within and among populations of the brooding sponge Rhopaloeides odorabile on the central Great Barrier Reef. Aquatic Biology 3, 111-119.

White AT, Vogt HP, Arin T (2000) Philippine coral reefs under threat: the economic losses caused by reef destruction. Marine Pollution Bulletin 40, 598-605.

Willis BL, van Oppen MJ, Miller DJ, Vollmer SV, Ayre DJ (2006) The role of hybridization in the evolution of reef corals. Annual Review of Ecology, Evolution, and Systematics 37, 489-517.

Wilson AB, Eigenmann Veraguth I (2010) The impact of Pleistocene glaciation across the range of a widespread European coastal species. Molecular Ecology 19, 4535- 4553.

Wolfe KH, Li W-H, Sharp PM (1987) Rates of nucleotide substitution vary greatly among plant mitochondrial, chloroplast, and nuclear DNAs. Proceedings of the National Academy of Sciences 84, 9054-9058.

Wörheide G (2006) Low variation in partial cytochrome oxidase subunit I (COI) mitochondrial sequences in the coralline demosponge Astrosclera willeyana across the Indo-Pacific. Marine Biology 148, 907-912.

Wörheide G, Epp LS, Macis L (2008) Deep genetic divergences among Indo-Pacific populations of the coral reef sponge Leucetta chagosensis (Leucettidae): Founder effects, vicariance, or both? BMC Evolutionary Biology 8, 24.

Worm B, Barbier EB, Beaumont N, et al. (2006) Impacts of biodiversity loss on ocean ecosystem services. Science 314, 787-790.

Wu C-I (2001) The genic view of the process of speciation. Journal of Evolutionary Biology 14, 851-865.

Wulff J (1995) Effects of a hurricane on survival and orientation of large erect coral reef sponges. Coral Reefs 14, 55-61.

71 Wulff JL (2006) Sponge systematics by starfish: predators distinguish cryptic sympatric species of Caribbean fire sponges, Tedania ignis and Tedania klausi n. sp.(Demospongiae, Poecilosclerida). The Biological Bulletin 211, 83-94.

Xavier J, Rachello-Dolmen P, Parra-Velandia F, et al. (2010) Molecular evidence of cryptic speciation in the "cosmopolitan," excavating sponge Cliona celata (Porifera, Clionaidae). Molecular Phylogenetics and Evolution 56, 13-20.

Yang Z, Rannala B (2010) Bayesian species delimitation using multilocus sequence data. Proceedings of the National Academy of Sciences 107, 9264.

Yund PO (1998) The effect of sperm competition on male gain curves in a colonial marine invertebrate. Ecology 79, 328-339.

Yund PO, Murdock K, Johnson SL (2007) Spatial distribution of ascidian sperm: two- dimensional patterns and short vs. time-integrated assays. Marine Ecology Progress Series 341, 103-109.

Zea S, Henkel TP, Pawlik JR (2009) The Sponge Guide: a picture guide to Caribbean sponges. www.spongeguide.org.

Zink RM, Barrowclough GF (2008) Mitochondrial DNA under siege in avian phylogeography. Molecular Ecology 17, 2107-2121.

72 Appendix A. Supplementary Text, Tables, and Figures for Chapter 2.

Generation of novel nuclear markers

Total RNA was extracted from a single individual using TRIzol (Invitrogen). cDNA was produced using the Invitrogen SuperscriptII and ClonTech SMART RT. cDNA was run on an agarose gel to check fragment size (ranged from 300 bp to 4 kb) and cleaned using a Strategene purification kit. The cDNA was quantified, sonicated, and sequenced on the 454 Life Sciences platform at the UCLA Genotyping and Sequencing Core. The 1/16th partial run resulted in 33,278 total reads averaging approximately 300 bp in length. These reads were assembled into 825 contigs averaging 692 bp using Newbler v2.3. There were 24,217 singletons.

We identified contigs using blastx. Primers were designed for approximately 20 contigs with a match to the Amphimedon queenslandica (sponge), Nematostella vectinsis (sea anemone), or Strongylocentrotus purpuratus (sea urchin) complete genome sequences. Primers were screened for single band amplification, target amplicon size, ease of sequencing, and presence of only two alleles per individual. Primers for six genes showed consistent PCR and sequencing results and had sequence variation among individuals (Appendix A, Table 1).

PCR amplification and sequencing of nucDNA loci

PCR amplifications were conducted in 25-µl reactions consisting of 2.5 µl of 10x buffer, 10 µM of dNTPs and each primer, and 0.25 units of One TaqTM DNA polymerase (New England Biolabs Inc.). Amplifications were performed in a Bio-Rad T100 thermocyler under the following conditions: an initial denaturation cycle of 3 minutes at 95°C, 2 minutes annealing at 50°C and 2 minutes extension at 72°C, followed by 38 cycles of 30 seconds at 95°C, 45 seconds at 50°C and 45 seconds at 72°C with a final extension cycle at 72°C for 10 minutes. Samples were sequenced in the forward and reverse direction using BigDye chemistry v3.1 on an ABI 3130XL at the Louisiana State University Genomics Facility. Forward and reverse sequences were aligned and edited using Geneious v4.5.5 (Drummond et al. 2012).

Parameters for tests of recombination in nuclear genes

Each gene region was tested for intra-locus recombination using GARD and SBP as implemented in Hy-Phy (Pond & Frost 2005; Pond et al. 2006) and with the DSS method implemented in TOPALi v2 (Milne et al. 2004). Recombination was not detected in any of the gene regions.

Program settings for estimation of population parameters in BEAST and *BEAST

In BEAST, we constructed a COI gene tree using the constant size coalescent tree prior and a strict molecular with the clock rate set to 1.0. We used a strict molecular clock

73 because the marginal posterior distribution of the standard deviation of the uncorrelated log-normal relaxed clock abutted 1.0 in preliminary runs. We used a random starting tree, allowing the root of the tree to be one of the parameters that BEAST estimates. Two independent MCMC analyses were run for 10 million steps, sampling every 1000 steps. Convergence on the posterior distribution was determined by viewing the log files in Tracer v1.5. All parameters had effective sample size (ESS) values greater than 300. Treefiles were combined in LogCombiner v1.7.1 with a 10% burnin for each file and the maximum clade credibility (MCC) tree for the combined tree file was calculated in TreeAnnotator v1.7.1. In the resulting COI gene tree, mtDNA clades 1 and 2 were sister and mtDNA clade 3 was basal. Because alternative rootings of the tree could affect parameter estimates of population size and lineage split times, we independently confirmed the (1,2), 3) relationship of the C. vaginalis mtDNA clades found in the original BEAST analysis. To do this, we estimated a new tree in BEAST using an individual from each haplotype within each of the three C. vaginalis mtDNA clades and the closely related outgroup Amphimedon queenslandica (in the same order as Callyspongia). The COI sequence for A. queenslandica was obtained from the complete mitochondrial genome for this species available on GenBank (NC_008944.1). In BEAST we estimated the tree using a coalescent tree prior, a lognormal relaxed (uncorrelated) clock, and a random starting tree. We ran two independent MCMC analyses for 10 million steps, sampling every 10,000 steps. Convergence was determined by viewing the log files in Tracer and all parameters had ESS values greater than 400. Treefiles were combined in LogCombiner v1.7.1 with a 10% burnin for each file and the maximum clade credibility (MCC) tree for the combined tree file was calculated in TreeAnnotator. The resulting tree (Appendix A, Figure 2) has posterior probabilities of at least 0.90 for nodes that determine the relationship among C. vaginalis mtDNA clades 1, 2, and 3.

The split time for mitochondrial lineages 1 and 2 (τ1-2) and the split time for the most recent common ancestor of lineages 1 and 2 from lineage 3 (τ1/2 – 3) for the slow mitochondrial substitution rate were obtained by dividing the node heights from the resultant gene tree by 5.0 x 10-10 substitutions per site per year, with generation time assumed to be one year. The split times for τ1-2 and τ1/2 – 3 for the fast rate were obtained by dividing the node heights by 5.0 x 10-9 substitutions per site per year (Appendix A, Table 2).

Effective population sizes (Ne) for each lineage were estimated in *BEAST. We conducted two independent runs with each mitochondrial substitution rate and a strict clock. The MCMC analyses were run for 50 million steps, sampling every 1000 steps. All parameters had ESS values greater than 300. Convergence was assessed as stated above. We used a script (starbeast_demog_log.py) from biopy, a Python package written by Joseph Heled (http://code.google.com/p/biopy/), to generate a new log file for each *BEAST run, viewable in Tracer, that follows the size for each population encountered during the run: the Ne for lineage 1, 2, and 3 (Ne 1, Ne 2, Ne 3), the Ne for the MRCA of lineages 1 and 2 (Ne 1 – 2), and the Ne for the MRCA of lineages 1, 2, and 3 (Ne 1/2 – 3). The new log file reports a value for the beginning and end of each branch of the species tree that corresponds to the number of alleles when the substitution rate is specified in substitutions per site per generation when generation time is one year. We averaged the

74 number of alleles reported at the end of each branch for the two runs using each rate and doubled the Ne of each lineage estimated from COI to obtain effective sizes that correspond to the nuclear gene loci (Appendix A, Table 2).

We estimated the substitution rates for the nuclear genes relative to COI in BEAST. We used a strict clock, set the clock rate for COI to 1.0 and allowed the rate to be estimated for the nuclear genes. Two independent MCMC analyses were run for 50 million steps, sampling every 1000 steps. All parameters had ESS values greater than 300. Convergence was assessed as stated above. The log files were combined with a 10% burnin for each file and the substitution rates for each nuclear gene relative to COI were recorded. We multiplied the relative nuclear rates by the slow and fast mitochondrial rates to get an estimate of the substitution rate for each nuclear gene (Appendix A, Table 2). This nuclear substitution rate was used in the calculation of theta required for the ms simulations (Hudson 2002).

To make our test of the null hypothesis conservative, in our simulations we used parameter values for population size and population split time that would be most likely to produce a signal of ILS in the simulated data. Values for the population split times were taken from the lower end of the 95% confidence interval (more recent splits) and values for population size were taken from the upper end of the 95% confidence interval (larger population size) (Appendix A, Table 2).

75 Table 1 Primer sequences and expected amplicon size for nuclear gene markers used in this study. ______

Gene name Primer sequence (5’ – 3’) Amplicon size (Abbreviation) ______

Catalase F: GATGATACACATTTTCATCGTCTTGG 520 bp (cata) R: ACGCTCCTGAATCTTTGGACTAGC

Cirhin F: TCTGTCTATATGATGAGTTACGATGC 230 bp (cir) R: CGTCACTTAAAAACTTTACGCTCC

Cathepsin F: TGAAGCTAAGAATGGCAAGTGTCG 215 bp (cps) R: GCCAGTACACCATGGTCAAGACG

Elongation factor-1 F: GCTTGGVTATTCTCCTGTTYTGG 270 bp alpha (ef) R: TTCHACRCACATGGGCTTGGACGG

Filamin F: CAGTACAATATAAGCATGGTCCCTCG 230 bp (fil) R: GTATAGGCTCACACAGTCCTTCCC

Macrophage F: CTTTCTCTATAAGCACTAATATTGG 210 bp expressed protein R: AGTAGTCATAGCAAATGAGAGAGG (mep) ______

76 Table 2 Demographic parameters estimated in BEAST and *BEAST from slow and fast mitochondrial substitution rates. Ne, effective size; HPD, highest posterior density ______

Slow mtDNA rate (5.0x10-10 substitutions/site/year) ______

Lineage split times (median values and HPD in millions of years before present):

1 from 2: 15.0 (7.2 – 24.6) 1/2 from 3: 22.8 (12.0 – 35.6)

Lineage effective size (mean values and HPD for number of mitochondrial alleles):

Ne1: 810,000 (43,000 – 1,920,000) Ne 2: 990,000 (44,000 – 2,300,000) Ne 3: 880,000 (77,000 – 2,000,000) Ne 1/2: 660,000 (10,400 – 2,200,000) Ne A: 1,140,000 (36,000 – 3,000,000)

Nuclear substitution rates (substitutions/site/year)

cata: 1.14x10-9 cir: 7.90x10-10 cps: 1.34x10-9 ef: 1.48x10-9 fil: 3.57x10-9 mep: 1.81x10-9 ______

Fast mtDNA rate (5.0x10-9 substitutions/site/year) ______

Lineage split times (median values and HPD in millions of years before present):

1 from 2: 1.5 (0.72 – 2.46) 1/2 from 3: 2.28 (1.22 – 3.56)

Lineage effective size (mean values and HPD for number of mitochondrial alleles):

Ne 1: 81,000 (3,300 – 189,000) Ne 2: 102,000 (4,000 – 230,000) Ne 3: 88,000 (7,300 – 200,000) Ne 1/2: 63,000 (735 – 183,000) Ne A: 115,000 (5,000 – 300,000)

Nuclear substitution rates (substitutions/site/year)

cata: 1.14x10-8 cir: 7.90x10-9 cps: 1.34x10-8 ef: 1.48x10-8 fil: 3.57x10-8 mep: 1.81x10-8 ______

77 Table 3 Results of hierarchical analyses of molecular variance (AMOVA) from Genodive (genotype data). Groups were defined as mitochondrial lineages or geographic locations. Two datasets were analyzed: A) only individuals sequenced for all six nuclear genes and B) all individuals sequenced for cir, mep, and fil. Bolded values are significant ______

Dataset: Genodive_A Groups: mtDNA lineages

Source of variation % variance Φ statistic P value ______

Among groups -0.10 ΦCT = -0.001 not reported

Among populations 4.60 ΦSC = 0.046 0.014 within groups

Within populations 95.5 ΦST = 0.045 0.017 ______

Dataset: Genodive_A Groups: Geographic locations

Source of variation % variance Φ statistic P value ______

Among groups 3.70 ΦCT = 0.037 not reported

Among populations 1.50 ΦSC = 0.015 0.258 within groups

Within populations 94.9 ΦST = 0.051 0.001 ______

Dataset: Genodive_B Groups: mtDNA lineages

Source of variation % variance Φ statistic P value ______

Among groups -1.00 ΦCT = 0.054 not reported

Among populations 6.40 ΦSC = 0.064 0.006 within groups

Within populations 94.6 ΦST = 0.054 0.015

78 (Table 3 continued) ______

Dataset: Genodive_B Groups: Geographic locations

Source of variation % variance Φ statistic P value ______

Among groups 5.30 ΦCT = 0.053 not reported

Among populations 1.20 ΦSC = 0.013 0.279 within groups

Within populations 93.5 ΦST = 0.065 0.001 ______

a t a d

f o

b o r p

n L

. t s e

f o

n a e M

K

Figure 1 Graph of the mean log likelihood score from 20 iterations for each value of K run in STRUCTURE.

79

Figure 2 COI gene tree estimated in BEAST to confirm the relationships among the C. vaginalis mtDNA clades. Posterior probabilities are listed at the nodes of the tree.

80 (mtDNA) (nDNA)

NeA (nDNA) o1/2-3 ACGTCATCCTAGTAG ACGTCATCCTAGTAG Ne1/2 ACGTCATCCTAGTAG o1-2 ACGTCATCCTAGTAG Ne3 Ne2 Ne1

1. 2. 3.

500

400

NeA s e e

r 300

1/2-3 t

o f o

. 200 o N Ne1/2 o1-2 100

Ne3 Ne2 Ne1 0 1 2 3 4 5 6 7 8 9 10 No. of DCEs 4. 5. 6.

Figure 3 Representation of the steps in the parametric bootstrap analysis. 1). Simulate 1000 coalescent trees matching demographic history of COI. Ne, effective population size; τ, divergence time; 2). Simulate nuclear sequence data on coalescent trees; 3). Estimate gene trees from nuclear sequence data; 4). Constrain simulated gene trees to fit demographic history of COI; 5). Count number of deep coalescent events (DCEs) between nuclear and COI trees; 6). Build null distribution of number of DCEs.

81 fast 80 fast fast

slow slow 6 slow 1.77 0.009 0.104 0.6 5 60 4 0.4 40 3 Density 2 0.2 20 1 0 0 0.0

0 1 2 3 4 5 0.00 0.01 0.02 0.03 0.04 0.05 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST

fast fast slow slow 0.25 0.06 12 30 0.05 0.20 0.04 0.15 Density 0.03 0.10 0.02 0.05 0.01 0.00 0.00

-5 0 5 10 0 10 20 30 40

DCE S Figure 4 Density plots showing the distribution of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and a model of nucleotide substitution matching the Cirhin gene. The black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. The red arrow indicates the value calculated for the empirical data.

82 fast 50 fast fast

slow slow 5 slow 1.51 0.025 -0.033 40 0.6 4 30 3 0.4 Density 20 2 0.2 10 1 0 0 0.0

0 1 2 3 4 0.00 0.01 0.02 0.03 0.04 0.05 0.06 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST

fast fast slow slow 0.06

0.15 16 35 0.05 0.04 0.10 Density 0.03 0.02 0.05 0.01 0.00 0.00

-5 0 5 10 15 20 0 10 20 30 40 DCE S

Figure 5 Density plots showing the distribution of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and a model of nucleotide substitution matching the Cathepsin gene. The black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. The red arrow indicates the value calculated for the empirical data.

83 7 fast fast fast slow slow slow 0.6

1.24 0.007 6 0.158 60 0.5 5 0.4 4 40 Density 0.3 3 0.2 2 20 1 0.1 0 0 0.0

0 1 2 3 4 5 0.00 0.01 0.02 0.03 0.04 0.05 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST 0.30 fast fast slow slow 0.06 13 27 0.25 0.05 0.20 0.04 0.15 Density 0.03 0.10 0.02 0.05 0.01 0.00 0.00

-5 0 5 10 15 0 10 20 30 40 DCE S

Figure 6 Density plots showing the distribution of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and a model of nucleotide substitution matching the Elongation factor 1 alpha gene. The black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. The red arrow indicates the value calculated for the empirical data.

84 0.6

fast 35 fast fast

slow slow 5 slow 2.03 0.184 0.100 0.5 30 4 25 0.4 20 3 0.3 Density 15 2 0.2 10 1 0.1 5 0 0 0.0

0 1 2 3 4 5 0.00 0.02 0.04 0.06 0.08 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST 0.08 fast fast slow slow 0.12 29 31 0.10 0.06 0.08 0.04 Density 0.06 0.04 0.02 0.02 0.00 0.00

-5 0 5 10 15 20 25 30 0 10 20 30 40 DCE S

Figure 7 Density plots showing the distribution of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and a model of nucleotide substitution matching the Filamin gene. The black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. The red arrow indicates the value calculated for the empirical data.

85 7 35 0.6 fast fast fast slow slow slow

1.97 0.023 6 -0.023 30 0.5 5 25 0.4 4 20 0.3 Density 3 15 0.2 2 10 0.1 1 5 0 0 0.0

0 1 2 3 4 5 0.00 0.02 0.04 0.06 0.08 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Theta Nucleotide Diversity PhiST 0.15 fast 0.07 fast slow slow 24 32 0.06 0.05 0.10 0.04 Density 0.03 0.05 0.02 0.01 0.00 0.00

-5 0 5 10 15 20 25 0 10 20 30 40

DCE S

Figure 8 Density plots showing the distribution of summary statistics calculated for datasets simulated with population parameters matching the mtDNA demographic scenario and a model of nucleotide substitution matching the Macrophage expressed protein gene. The black and grey curves represent values for datasets simulated assuming a fast and slow substitution rate, respectively. The red arrow indicates the value calculated for the empirical data.

86 Appendix B. Supplementary Tables and Figures for Chapter 3.

Table 1 Accession numbers for previously published sequences used in this study ______

Species Locus Accession No. Reference ______

Callyspongia vaginalis COI GQ304583-4646, DeBiasse et al. (2010) GQ304677-6734

catalase HG001142-1246 DeBiasse et al. (2104)

cirhin HG000786-0907 DeBiasse et al. (2104)

cathespin HG000670-0785 DeBiasse et al. (2104)

elongation HG001030-1141 DeBiasse et al. (2104) factor 1 alpha

filamin HG000908-1029 DeBiasse et al. (2104)

macrophage HG737726-7846 DeBiasse et al. (2104) expressed protein

Leucothoe ashleyae COI EF053411-3423 Richards et al. (2007)

JQ004957-4968 Richards et al. (2012)

Leucothoe kensleyi COI EF053456-3503 Richards et al. (2007)

JQ004939-4956 Richards et al. (2012)

Ophiothrix suensonii COI (pending) Richards (2010)

______

87 Table 2 Genetic diversity indices for Callyspongia vaginalis for each gene by sampling location. n, sample size (alleles); bp, base pairs; S, segregating sites; h, haplotypes; Hd, haplotype diversity; π, nucleotide diversity. Bold Tajima’s D values are significant ______

Location n bp S h Hd π Tajima’s D ______mtCOI Key Largo✚ 17 511 18 3 0.669 0.0143 2.1103 Long Key✚ 14 511 18 3 0.641 0.0137 1.6144 Marquesas Keys✚ 14 511 18 3 0.659 0.0150 2.1341 Dry Tortugas✚ 16 511 18 3 0.633 0.0133 1.6176 Glover’s Reef 33 511 16 4 0.280 0.0036 -1.8795 Utila 19 511 13 2 0.199 0.0051 -1.1215 Roatan 32 511 12 2 0.175 0.0041 -0.9451 Bocas del Toro 24 511 7 2 0.228 0.0104 -0.4614 Curacao 30 511 11 2 0.515 0.0007 3.3472 St. Croix 28 511 0 1 0.000 0.0000 NA Vieques 30 511 0 1 0.000 0.0000 NA Crooked Island 23 511 1 2 0.087 0.0002 -1.1610 Bimini 29 511 14 4 0.658 0.0107 1.4485 All Locations 308 511 25 9 0.733 0.0118 0.8302 cata Key Largo* 30 346 13 10 0.770 0.0099 -0.0765 Long Key* 26 346 13 8 0.776 0.0077 -0.7390 Marquesas Keys* 20 346 17 10 0.758 0.0131 -0.1949 Dry Tortugas* 30 346 13 9 0.851 0.0095 0.0175 Glover’s Reef 20 346 9 4 0.432 0.0040 -1.5482 Utila 5 346 14 5 0.667 0.0127 -0.2367 Roatan 4 346 13 4 1.000 0.0217 0.5837 Bocas del Toro 18 346 4 2 0.209 0.0024 -0.8412 Curacao 26 346 15 6 0.649 0.0117 0.1045 St. Croix 16 346 8 3 0.425 0.0082 0.6192 Vieques 14 346 9 3 0.473 0.0091 0.4093 Crooked Island 8 346 3 3 0.464 0.0032 -0.1774 All Locations 242 346 25 36 0.887 0.0153 0.7590 cir Key Largo* 34 159 7 6 0.729 0.0090 -0.0748 Long Key* 28 159 7 6 0.635 0.0064 -0.9755 Marquesas Keys* 28 159 7 6 0.730 0.0094 -0.5035 Dry Tortugas* 32 159 10 9 0.833 0.0121 -0.4323 Glover’s Reef 22 159 8 6 0.801 0.0123 -0.3493 Utila 20 159 5 5 0.758 0.0075 -0.5916

88 (Table 2 continued) ______

Location n bp S h Hd π Tajima’s D ______cir (continued) Roatan 20 159 5 6 0.758 0.0068 -0.7173 Bocas del Toro 18 159 5 4 0.595 0.0064 -0.9594 Curacao 28 159 5 4 0.587 0.0055 -0.9018 St. Croix 18 159 7 7 0.843 0.0096 -0.8542 Vieques 12 159 5 5 0.758 0.0084 -0.7177 Crooked Island 8 159 2 2 0.250 0.0031 -1.3101 Bimini 34 159 9 9 0.836 0.1031 -1.0044 All Locations 304 159 17 20 0.767 0.0092 -1.3166 cps Key Largo* 34 113 7 6 0.754 0.0261 2.0813 Long Key* 26 113 7 7 0.763 0.0277 2.1556 Marquesas Keys* 28 113 7 6 0.733 0.0231 -0.6803 Dry Tortugas* 28 113 7 7 0.741 0.0255 1.8119 Glover’s Reef 20 113 7 7 0.826 0.0248 1.3780 Utila 22 113 9 8 0.818 0.0210 -0.1311 Roatan 20 113 7 9 0.900 0.0255 1.5095 Bocas del Toro 18 113 7 5 0.706 0.2325 0.9876 Curacao 28 113 7 6 0.730 0.0205 0.8619 St. Croix 16 113 2 4 0.525 0.0058 0.2007 Vieques 14 113 3 5 0.593 0.0101 0.6474 Crooked Island 8 113 7 5 0.983 0.0278 0.7811 Bimini 34 113 7 6 0.793 0.0176 0.4678 All Locations 298 113 9 13 0.773 0.0218 1.5584 ef Key Largo* 28 161 3 6 0.726 0.0064 0.8728 Long Key* 26 161 6 8 0.843 0.0110 0.3764 Marquesas Keys* 26 161 7 6 0.794 0.0145 0.8309 Dry Tortugas* 28 161 6 6 0.698 0.0071 -0.7663 Glover’s Reef 22 161 4 4 0.260 0.0226 -1.8776 Utila 22 161 6 6 0.693 0.0081 -0.6538 Roatan 20 161 3 3 0.484 0.0054 0.0554 Bocas del Toro 14 161 2 2 0.495 0.0061 1.5537 Curacao 26 161 3 4 0.286 0.0023 -1.3110 St. Croix 18 161 3 4 0.542 0.0038 -0.8196 Vieques 12 161 1 2 0.409 0.0025 0.5406 Crooked Island 8 161 1 2 0.429 0.0027 0.3350 Bimini 32 161 4 4 0.542 0.0045 -0.6899

89 (Table 2 continued) ______

Location n bp S h Hd π Tajima’s D ______ef (continued) All Locations 288 161 13 19 0.752 0.0084 -0.8349 fil Key Largo* 34 127 12 14 0.884 0.0240 0.7571 Long Key* 28 127 7 9 0.839 0.0256 2.4217 Marquesas Keys* 28 127 10 11 0.749 0.0199 -0.0509 Dry Tortugas* 32 127 10 8 0.792 0.0234 0.6120 Glover’s Reef 22 127 10 8 0.602 0.0149 -1.2909 Utila 22 127 6 5 0.407 0.0068 -1.4688 Roatan 20 127 3 3 0.426 0.0051 -0.6261 Bocas del Toro 18 127 6 3 0.523 0.0136 -0.0357 Curacao 26 127 9 10 0.822 0.0247 1.0585 St. Croix 18 127 8 7 0.850 0.0271 1.6714 Vieques 12 127 7 6 0.758 0.0180 -0.0508 Crooked Island 8 127 5 5 0.857 0.0205 1.5962 Bimini 30 127 9 13 0.864 0.0232 0.9301 All Locations 300 127 16 35 0.848 0.0239 -0.1188 mep Key Largo* 34 91 10 8 0.816 0.0351 0.9391 Long Key* 28 91 10 8 0.812 0.0297 0.1677 Marquesas Keys* 28 91 9 6 0.759 0.0274 0.2448 Dry Tortugas* 32 91 9 7 0.786 0.0201 -0.5622 Glover’s Reef 22 91 8 8 0.892 0.0360 1.6128 Utila 20 91 8 5 0.711 0.0246 -0.0192 Roatan 20 91 7 6 0.805 0.0327 1.6674 Bocas del Toro 18 91 4 4 0.575 0.0170 0.9789 Curacao 26 91 12 8 0.785 0.0261 -0.8271 St. Croix 16 91 5 5 0.833 0.0207 0.8234 Vieques 14 91 3 4 0.780 0.0153 1.4676 Crooked Island 8 91 6 5 0.857 0.0283 0.5185 Bimini 30 91 4 6 0.715 0.0134 0.5366 All Locations 296 91 10 11 0.835 0.0204 1.0192 ______✚DeBiasse et al. (2010); *DeBiasse et al. (2014)

90 Probability of assignment 0.65 Probability of assignment 0.35 Key Largo Long Key Bimini Bimini Marquesas Keys Florida Florida Dry Tortugas Crooked Island Crooked Island Veracruz Veracruz Vieques Glover’s Reef Glover’s Reef Vieques Utila Utila Roatan St. Croix Roatan St. Croix

Curacao Curacao Bocas del Toro Bocas del Toro

A B

Probability of assignment 0.85

Florida Bimini

Crooked Island Veracruz

Glover’s Reef Vieques Utila Roatan St. Croix

Curacao Bocas del Toro

C

Figure 1 Maps showing the distribution of genetic clusters inferred from STRUCTURAMA including all individuals sampled based on (A, B) the nucDNA alone and on (C) the nucDNA and mtDNA combined. The sector area in the pie chart at each sampling location is proportional to number of individuals in a particular cluster at that site.

91

Figure 2 Unrooted statistical parsimony networks showing the relationships among alleles of the cata gene. Circles represent alleles with size proportional to frequency of the sequence. Small black circles represent possible, but not sampled alleles and lines between circles represent one mutational change between sequences. Divergent individuals from Central America had alleles that failed to join the network at the 95% confidence level (black shading).

92 L(K) (mean +- SD) DeltaK = mean(|L”(K)|) / sd(L(K))

-3600 15 -3700 a t

a -3800 d

f o -3900 10 b o K

r a p t

-4000 l n e L

D . t

s -4100 e

05 f o

n -4200 a e

M -4300 00 -4400

1 2 3 4 5 6 7 8 9 10 11 12 13 14 2 3 4 5 6 7 8 9 10 11 12 13 14 K K

Utila Roatan Bimini Veracruz Curacao St. CroixVieques Key Largo Long Key Dry Tortugas Glover’s Reef Bocas del Toro Crooked Island Marquesas Keys

Figure 3 Population assignment information for all individuals based on the nucDNA only. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with peaks at K = 2 and 7. Below: bar plots for K=2 through K=7 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

93 Figure 4 Population assignment information for all individuals based on the nucDNA and mtDNA combined. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with peaks at K = 2, 7, and 10. Bottom: bar plots for K = 2 through K = 10 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

94 CAW NC geo

L(K) (mean +- SD) DeltaK = mean(|L”(K)|) / sd(L(K))

-3800 40 a t -4000 a d

30 f o

b -4200 o K r a p t

l 20 n -4400 e L

D . t s e

f -4600 o 10 n a e -4800 M 00 -5000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 2 3 4 5 6 7 8 9 10 11 12 13 14 K K

Utila Roatan Bimini Veracruz Curacao St. CroixVieques Key Largo Long Key Dry Tortugas Glover’s Reef Bocas del Toro Crooked Island Marquesas Keys

95 Figure 5 Population assignment information for individuals based on the nucDNA only and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with peaks at K = 2. Bottom: bar plot for K = 2 through 8 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

96 Utila Roatan Bimini Curacao St. Croix Vieques Key Largo Long Key Dry Tortugas Glover’s Reef Bocas del Toro Crooked Island Marquesas Keys

97 Figure 6 Population assignment information for individuals based on the nucDNA and mtDNA combined and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with a peak at K = 2. Bottom: bar plot for K = 2 through K = 6 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

98 Utila Bimini Roatan Curacao Vieques Long Key St. Croix Key Largo Marquesas Crooked Is. Dry Tortugas Glover’s Reef Bocas del Toro

99

Utila Florida Bimini Curacao St. Croix Vieques

Crooked Island

Figure 7 Population assignment information for individuals assigned to the “east” cluster in Figure 5 based on the nucDNA only and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with a peak at K = 3. Below: bar plot for K = 3 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

100 Florida Bimini Curacao St. Croix Vieques Crooked Is. Bocas del Toro

Figure 8 Population assignment information for individuals assigned to the “east” cluster in Figure 6 based on the nucDNA and mtDNA combined and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with a peak at K = 4. Below: bar plots for K = 2 through 4 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

101

L(K) (mean +- SD) DeltaK = mean(|L”(K)|) / sd(L(K)) 35 a t a

d 30

-2100 f o 25 b

o -2150 K r

a p 20

t l n

-2200 e L

. 15 D t s e -2250 10 f o

n 05

a -2300 e 00 M

1 2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 K K

Utila Roatan Bimini Key Largo Long Key Dry Tortugas Glover’s Reef Bocas del Toro Marquesas Keys

Figure 9 Population assignment information for individuals assigned to the “west” cluster in Figure 5 based on the nucDNA only and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with a peak at K = 3. Below: bar plots for K = 3 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

nucDNA

102 Utila Roatan Bimini Long Key Key Largo Marquesas Dry Tortugas Glover’s Reef Bocas del Toro

Figure 10 Population assignment information for individuals assigned to the “west” cluster in Figure 6 based on the nucDNA and mtDNA combined and excluding divergent individuals from Bocas, Utila, and Veracruz. Top left: graph showing the mean likelihood of each K tested. Top right: graph showing ΔK with a peak at K = 2. Below: bar plots for K = 2 through 4 where each individual is represented by a vertical line, with the colored portion of the line representing the proportion of that individual’s genome originating from each inferred clusters.

103 West y l n ) O 0

. 1 A ( N 1

D = c

u K n ) 9 9 . 0 (

2

=

K nucDNA + mtDNA

Utila Roatan Bimini Long Key Key Largo Marquesas Dry Tortugas Glover’s Reef Bocas del Toro East nucDNA and mtDNA East nucDNA only ) ) 9 5 6 . 7 . 0 0 (

(

1

2

=

=

K K ) ) 5 2 2 2 . . 0 0 ( (

3 2

= =

K K

Florida Bimini Florida Bimini Curacao St. Croix Vieques Curacao St. Croix Vieques Crooked Is. Crooked Is. Bocas del Toro

Figure 11 Population assignment information estimated in STRUCTURAMA for individuals of the “west” and “east” clusters (excluding divergent individuals from Bocas del Toro, Utila, and Veracruz) based on the nucDNA only and on the nucDNA and mtDNA combined. Cluster number and probability of assignment is listed to the left of each bar plot.

104

Cryptic

1.0 West 0.74

Florida Florida 0.74 0.93

East East 0.70 2.0E-4

1.0 West Cryptic

Cryptic Florida 1.0 8.0E-5 0.48

Cryptic West 0.41

East East 1.0 2.0E-4 0.69

Florida East 0.45

0.93 West Florida 8.0E-5 0.47

1.0 West

Cryptic 2.0E-4

Figure 12 Species trees estimated in *BEAST. Posterior probabilities are listed at each node.

105 Appendix C. Supplementary Tables and Figures for Chapter 4.

ARM08 92 ARM53 VAG47

VAG83

ARM57

VAG24

VAG26

VAG39

VAG40

VAG42

VAG41

ESC09

ESC10 100 ESC11 LON01

ARM54

ARM55

PLI66

FAL12

FAL13 100 FAL14 FAL15

FAL16

FAL17 100 FAL18 FAL19

FAL20

FAL269

FAL270

COM01 0.04

Figure 1 Maximum likelihood tree estimated in PAUP* using the 5’ region of the COI gene. Colored boxes indicate each species corresponding to the legend in Figure 1. Support values from 100 bootstrap replicates appear at the nodes when greater than 50. The tree is rooted with Amphimedon compressa (accession number NC_010201).

106 VAG47

VAG83

VAG41

ESC11

ESC09

ESC10

VAG26

VAG24

VAG39

VAG40 100 VAG42 ARM57

ARM07

ARM54

ARM55

ARM56

ESC02

ESC03

ESC04

LON01

ARM08

ARM53 96 PLI65 PLI64

PLI68 93 TEN32 FAL12

FAL13

FAL14

FAL15

FAL16

FAL17

FAL18

88 FAL19

FAL20

FAL21

FAL270

COM01 0.04

Figure 2 Maximum likelihood tree estimated in PAUP* using the 3’ region of the COI gene. Colored boxes indicate each species corresponding to the legend in Figure 1. Support values from 100 bootstrap replicates appear at the nodes when greater than 50. The tree is rooted with Amphimedon compressa (accession number NC_010201).

107 LON01

ARM07

ARM08

ARM53

ARM54

ARM55

ARM56

97 ARM57

ESC09

ESC11

VAG83

VAG42

VAG39

VAG26

VAG24

VAG41

FAL269

FAL270 100 FAL12 FAL13

FAL18

FAL19

FAL14

FAL15

FAL16

FAL17

FAL20 100 FAL21 PLI68

74 PLI64

PLI67

TEN32

85 TEN34

TEN260

COM01

0.05

Figure 3 Maximum likelihood tree estimated in PAUP* using the D1 region of the 28S gene. Colored boxes indicate each species corresponding to the legend in Figure 1. Support values from 100 bootstrap replicates appear at the nodes when greater than 50. The tree is rooted with Amphimedon compressa (accession number JN178945).

108 ARM53a

ARM55b 53 VAG40b VAG83b

ESC02b

ARM56a

ARM57a

ESC02a

VAG39a

VAG42a

VAG47b

ARM53b

ARM56b

VAG83a

ARM55a

LON01a

VAG40a

ARM54b

ARM57b

ESC09a

ESC09b 79 ESC10b ESC11b

FAL12b

ESC10a

ESC11a

FAL12a

VAG26a

VAG26b

VAG39b

VAG42b

VAG47a

ARM54a

LON01b

0.0060

Figure 4 Maximum likelihood tree estimated in PAUP* using a portion of the filamin gene. Colored boxes indicate each species corresponding to the legend in Figure 1. No bootstrap support values were greater than 50. The tree is rooted at its midpoint.

109 ARM07a

ARM07b

ARM56a

ARM53a

LON01a

LON01b

VAG24a

VAG39a

ARM08a

ARM08b

ARM53b

ARM55b

ARM57b

VAG24b

VAG39b

VAG42a

VAG42b

ARM54a

ARM54b

VAG40b

ARM55a

ARM56b

ESC09a

ESC09b

ESC10a

ESC10b

ESC11a

ESC11b

VAG83a

VAG83b

FAL14

ARM57a

VAG40a

VAG41a

VAG41b

VAG47a

VAG47b

QUE01

0.01

Figure 5 Maximum likelihood tree estimated in PAUP* using a portion of the macrophage expressed protein gene. Colored boxes indicate each species corresponding to the legend in Figure 1. No bootstrap support values were greater than 50. The tree is rooted with Amphimedon queenslandica (accession number GCF_000090795).

110 Appendix D. Permission to Reproduce Chapter 2.

 5LJKWVOLQN3ULQWDEOH/LFHQVH

-2+1:,/(<$1'6216/,&(16( 7(506$1'&21',7,216 -XQ

7KLVLVD/LFHQVH$JUHHPHQWEHWZHHQ0HOLVVD'H%LDVVH 

$OOSD\PHQWVPXVWEHPDGHLQIXOOWR&&&)RUSD\PHQWLQVWUXFWLRQVSOHDVHVHH LQIRUPDWLRQOLVWHGDWWKHERWWRPRIWKLVIRUP

/LFHQVH1XPEHU  /LFHQVHGDWH -XQ

/LFHQVHGFRQWHQWSXEOLVKHU -RKQ:LOH\DQG6RQV

/LFHQVHGFRQWHQWSXEOLFDWLRQ 0ROHFXODU(FRORJ\ /LFHQVHGFRQWHQWWLWOH (YDOXDWLQJVXPPDU\VWDWLVWLFVXVHGWRWHVWIRULQFRPSOHWHOLQHDJH VRUWLQJPLWRQXFOHDUGLVFRUGDQFHLQWKHUHHIVSRQJH&DOO\VSRQJLD YDJLQDOLV /LFHQVHGFRS\ULJKWOLQH ‹-RKQ:LOH\ 6RQV/WG /LFHQVHGFRQWHQWDXWKRU 0HOLVVD%'H%LDVVH%UDGOH\-1HOVRQ0LFKDHO(+HOOEHUJ

/LFHQVHGFRQWHQWGDWH 1RY 6WDUWSDJH 

(QGSDJH  7\SHRIXVH 'LVVHUWDWLRQ7KHVLV

5HTXHVWRUW\SH $XWKRURIWKLV:LOH\DUWLFOH )RUPDW 3ULQWDQGHOHFWURQLF 3RUWLRQ )XOODUWLFOH

:LOO\RXEHWUDQVODWLQJ" 1R 7LWOHRI\RXUWKHVLV 02'(/%$6('7(6762)+,6725,&$/'(02*5$3+<$1'63(&,(6 GLVVHUWDWLRQ '(/,0,7$7,21,17+(&$5,%%($1&25$/5(()6321*( &$//<6321*,$ ([SHFWHGFRPSOHWLRQGDWH -XQ

([SHFWHGVL]H QXPEHURI  SDJHV 7RWDO 86' 7HUPVDQG&RQGLWLRQV 7(506$1'&21',7,216

7KLVFRS\ULJKWHGPDWHULDOLVRZQHGE\RUH[FOXVLYHO\OLFHQVHGWR-RKQ:LOH\ 6RQV,QFRU RQHRILWVJURXSFRPSDQLHV HDFKD:LOH\&RPSDQ\ RUKDQGOHGRQEHKDOIRIDVRFLHW\ZLWK KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

111  5LJKWVOLQN3ULQWDEOH/LFHQVH ZKLFKD:LOH\&RPSDQ\KDVH[FOXVLYHSXEOLVKLQJULJKWVLQUHODWLRQWRDSDUWLFXODUZRUN FROOHFWLYHO\:,/(< %\FOLFNLQJֺDFFHSWֺLQFRQQHFWLRQZLWKFRPSOHWLQJWKLVOLFHQVLQJ WUDQVDFWLRQ\RXDJUHHWKDWWKHIROORZLQJWHUPVDQGFRQGLWLRQVDSSO\WRWKLVWUDQVDFWLRQ DORQJ ZLWKWKHELOOLQJDQGSD\PHQWWHUPVDQGFRQGLWLRQVHVWDEOLVKHGE\WKH&RS\ULJKW&OHDUDQFH &HQWHU,QF &&& V%LOOLQJDQG3D\PHQWWHUPVDQGFRQGLWLRQV DWWKHWLPHWKDW\RXRSHQHG \RXU5LJKWVOLQNDFFRXQW WKHVHDUHDYDLODEOHDWDQ\WLPHDWKWWSP\DFFRXQWFRS\ULJKWFRP 

7HUPVDQG&RQGLWLRQV

7KHPDWHULDOV\RXKDYHUHTXHVWHGSHUPLVVLRQWRUHSURGXFHRUUHXVH WKH:LOH\ 0DWHULDOV DUHSURWHFWHGE\FRS\ULJKW

:LWKUHVSHFWWRWKH:LOH\0DWHULDOVDOOULJKWVDUHUHVHUYHG([FHSWDVH[SUHVVO\JUDQWHG E\WKHWHUPVRIWKHOLFHQVHQRSDUWRIWKH:LOH\0DWHULDOVPD\EHFRSLHGPRGLILHG DGDSWHG H[FHSWIRUPLQRUUHIRUPDWWLQJUHTXLUHGE\WKHQHZ3XEOLFDWLRQ WUDQVODWHG UHSURGXFHGWUDQVIHUUHGRUGLVWULEXWHGLQDQ\IRUPRUE\DQ\PHDQVDQGQRGHULYDWLYH ZRUNVPD\EHPDGHEDVHGRQWKH:LOH\0DWHULDOVZLWKRXWWKHSULRUSHUPLVVLRQRIWKH UHVSHFWLYHFRS\ULJKWRZQHU

7KH:LOH\0DWHULDOVDQGDOORIWKHLQWHOOHFWXDOSURSHUW\ULJKWVWKHUHLQVKDOODWDOOWLPHV UHPDLQWKHH[FOXVLYHSURSHUW\RI-RKQ:LOH\ 6RQV,QFWKH:LOH\&RPSDQLHVRUWKHLU UHVSHFWLYHOLFHQVRUVDQG\RXULQWHUHVWWKHUHLQLVRQO\WKDWRIKDYLQJSRVVHVVLRQRIDQG WKHULJKWWRUHSURGXFHWKH:LOH\0DWHULDOVSXUVXDQWWR6HFWLRQKHUHLQGXULQJWKH FRQWLQXDQFHRIWKLV$JUHHPHQW

1(,7+(5:,/(<125,76/,&(162560$.(6$1<:$55$17<25

KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

112  5LJKWVOLQN3ULQWDEOH/LFHQVH 5(35(6(17$7,212)$1<.,1'72<2825$1<7+,5'3$57< (;35(66,03/,('2567$78725<:,7+5(63(&7727+(0$7(5,$/6 257+($&&85$&<2)$1<,1)250$7,21&217$,1(',17+( 0$7(5,$/6,1&/8',1*:,7+287/,0,7$7,21$1<,03/,(' :$55$17<2)0(5&+$17$%,/,7<$&&85$&<6$7,6)$&725< 48$/,7<),71(66)25$3$57,&8/$5385326(86$%,/,7< ,17(*5$7,2125121,1)5,1*(0(17$1'$//68&+:$55$17,(6 $5(+(5(%<(;&/8'('%<:,/(<$1',76/,&(16256$1':$,9(' %<<28

:,/(<VKDOOKDYHWKHULJKWWRWHUPLQDWHWKLV$JUHHPHQWLPPHGLDWHO\XSRQEUHDFKRI WKLV$JUHHPHQWE\\RX

,112(9(176+$//:,/(<25,76/,&(16256%(/,$%/(72<2825 $1<27+(53$57<25$1<27+(53(562125(17,7<)25$1< 63(&,$/&216(48(17,$/,1&,'(17$/,1',5(&7(;(03/$5<25 381,7,9('$0$*(6+2:(9(5&$86('$5,6,1*2872)25,1 &211(&7,21:,7+7+('2:1/2$',1*3529,6,21,1*9,(:,1*25 86(2)7+(0$7(5,$/65(*$5'/(662)7+()2502)$&7,21 :+(7+(5)25%5($&+2)&2175$&7%5($&+2):$55$17< 72571(*/,*(1&(,1)5,1*(0(172527+(5:,6( ,1&/8',1* :,7+287/,0,7$7,21'$0$*(6%$6('21/2662)352),76'$7$ ),/(686(%86,1(6623325781,7<25&/$,062)7+,5'3$57,(6  $1':+(7+(5251277+(3$57<+$6%((1$'9,6('2)7+( 3266,%,/,7<2)68&+'$0$*(67+,6/,0,7$7,216+$//$33/< 127:,7+67$1',1*$1<)$,/85(2)(66(17,$/385326(2)$1< /,0,7('5(0('<3529,'('+(5(,1

6KRXOGDQ\SURYLVLRQRIWKLV$JUHHPHQWEHKHOGE\DFRXUWRIFRPSHWHQWMXULVGLFWLRQWR EHLOOHJDOLQYDOLGRUXQHQIRUFHDEOHWKDWSURYLVLRQVKDOOEHGHHPHGDPHQGHGWRDFKLHYH DVQHDUO\DVSRVVLEOHWKHVDPHHFRQRPLFHIIHFWDVWKHRULJLQDOSURYLVLRQDQGWKHOHJDOLW\ YDOLGLW\DQGHQIRUFHDELOLW\RIWKHUHPDLQLQJSURYLVLRQVRIWKLV$JUHHPHQWVKDOOQRWEH DIIHFWHGRULPSDLUHGWKHUHE\

7KHIDLOXUHRIHLWKHUSDUW\WRHQIRUFHDQ\WHUPRUFRQGLWLRQRIWKLV$JUHHPHQWVKDOOQRW FRQVWLWXWHDZDLYHURIHLWKHUSDUW\ VULJKWWRHQIRUFHHDFKDQGHYHU\WHUPDQGFRQGLWLRQ RIWKLV$JUHHPHQW1REUHDFKXQGHUWKLVDJUHHPHQWVKDOOEHGHHPHGZDLYHGRUH[FXVHG E\HLWKHUSDUW\XQOHVVVXFKZDLYHURUFRQVHQWLVLQZULWLQJVLJQHGE\WKHSDUW\JUDQWLQJ VXFKZDLYHURUFRQVHQW7KHZDLYHUE\RUFRQVHQWRIDSDUW\WRDEUHDFKRIDQ\ SURYLVLRQRIWKLV$JUHHPHQWVKDOOQRWRSHUDWHRUEHFRQVWUXHGDVDZDLYHURIRUFRQVHQW WRDQ\RWKHURUVXEVHTXHQWEUHDFKE\VXFKRWKHUSDUW\

7KLV$JUHHPHQWPD\QRWEHDVVLJQHG LQFOXGLQJE\RSHUDWLRQRIODZRURWKHUZLVH E\ \RXZLWKRXW:,/(< VSULRUZULWWHQFRQVHQW

$Q\IHHUHTXLUHGIRUWKLVSHUPLVVLRQVKDOOEHQRQUHIXQGDEOHDIWHUWKLUW\  GD\VIURP KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

113  5LJKWVOLQN3ULQWDEOH/LFHQVH UHFHLSWE\WKH&&&

7KHVHWHUPVDQGFRQGLWLRQVWRJHWKHUZLWK&&&ֺV%LOOLQJDQG3D\PHQWWHUPVDQG FRQGLWLRQV ZKLFKDUHLQFRUSRUDWHGKHUHLQ IRUPWKHHQWLUHDJUHHPHQWEHWZHHQ\RXDQG :,/(<FRQFHUQLQJWKLVOLFHQVLQJWUDQVDFWLRQDQG LQWKHDEVHQFHRIIUDXG VXSHUVHGHV DOOSULRUDJUHHPHQWVDQGUHSUHVHQWDWLRQVRIWKHSDUWLHVRUDORUZULWWHQ7KLV$JUHHPHQW PD\QRWEHDPHQGHGH[FHSWLQZULWLQJVLJQHGE\ERWKSDUWLHV7KLV$JUHHPHQWVKDOOEH ELQGLQJXSRQDQGLQXUHWRWKHEHQHILWRIWKHSDUWLHV VXFFHVVRUVOHJDOUHSUHVHQWDWLYHV DQGDXWKRUL]HGDVVLJQV

,QWKHHYHQWRIDQ\FRQIOLFWEHWZHHQ\RXUREOLJDWLRQVHVWDEOLVKHGE\WKHVHWHUPVDQG FRQGLWLRQVDQGWKRVHHVWDEOLVKHGE\&&&ֺV%LOOLQJDQG3D\PHQWWHUPVDQGFRQGLWLRQV WKHVHWHUPVDQGFRQGLWLRQVVKDOOSUHYDLO

:,/(<H[SUHVVO\UHVHUYHVDOOULJKWVQRWVSHFLILFDOO\JUDQWHGLQWKHFRPELQDWLRQRI L WKHOLFHQVHGHWDLOVSURYLGHGE\\RXDQGDFFHSWHGLQWKHFRXUVHRIWKLVOLFHQVLQJ LLL &&&ֺV%LOOLQJDQG3D\PHQWWHUPV LL WKHVHWHUPVDQGFRQGLWLRQVDQG WUDQVDFWLRQ DQGFRQGLWLRQV

7KLV$JUHHPHQWZLOOEHYRLGLIWKH7\SHRI8VH)RUPDW&LUFXODWLRQRU5HTXHVWRU 7\SHZDVPLVUHSUHVHQWHGGXULQJWKHOLFHQVLQJSURFHVV

7KLV$JUHHPHQWVKDOOEHJRYHUQHGE\DQGFRQVWUXHGLQDFFRUGDQFHZLWKWKHODZVRIWKH \6WDWHRI1HZ

:,/(<23(1$&&(667(506$1'&21',7,216

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

&RS\ULJKWLQDQ\UHVHDUFKDUWLFOHLQDMRXUQDOSXEOLVKHGDV2SHQ$FFHVVXQGHUD&UHDWLYH &RPPRQV/LFHQVHLVUHWDLQHGE\WKHDXWKRU V $XWKRUVJUDQW:LOH\DOLFHQVHWRSXEOLVKWKH DUWLFOHDQGLGHQWLI\LWVHOIDVWKHRULJLQDOSXEOLVKHU$XWKRUVDOVRJUDQWDQ\WKLUGSDUW\WKHULJKW WRXVHWKHDUWLFOHIUHHO\DVORQJDVLWVLQWHJULW\LVPDLQWDLQHGDQGLWVRULJLQDODXWKRUVFLWDWLRQ GHWDLOVDQGSXEOLVKHUDUHLGHQWLILHGDVIROORZV>7LWOHRI$UWLFOH$XWKRU-RXUQDO7LWOHDQG 9ROXPH,VVXH&RS\ULJKW F >\HDU@>FRS\ULJKWRZQHUDVVSHFLILHGLQWKH-RXUQDO@/LQNVWRWKH ILQDODUWLFOHRQ:LOH\ֺVZHEVLWHDUHHQFRXUDJHGZKHUHDSSOLFDEOH

KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

114  5LJKWVOLQN3ULQWDEOH/LFHQVH 7KH&UHDWLYH&RPPRQV$WWULEXWLRQ/LFHQVH

7KH&UHDWLYH&RPPRQV$WWULEXWLRQ/LFHQVH &&%< DOORZVXVHUVWRFRS\GLVWULEXWHDQG WUDQVPLWDQDUWLFOHDGDSWWKHDUWLFOHDQGPDNHFRPPHUFLDOXVHRIWKHDUWLFOH7KH&&%< OLFHQVHSHUPLWVFRPPHUFLDODQGQRQFRPPHUFLDOUHXVHRIDQRSHQDFFHVVDUWLFOHDVORQJDVWKH DXWKRULVSURSHUO\DWWULEXWHG

7KH&UHDWLYH&RPPRQV$WWULEXWLRQ/LFHQVHGRHVQRWDIIHFWWKHPRUDOULJKWVRIDXWKRUV LQFOXGLQJZLWKRXWOLPLWDWLRQWKHULJKWQRWWRKDYHWKHLUZRUNVXEMHFWHGWRGHURJDWRU\WUHDWPHQW ,WDOVRGRHVQRWDIIHFWDQ\RWKHUULJKWVKHOGE\DXWKRUVRUWKLUGSDUWLHVLQWKHDUWLFOHLQFOXGLQJ ZLWKRXWOLPLWDWLRQWKHULJKWVRISULYDF\DQGSXEOLFLW\8VHRIWKHDUWLFOHPXVWQRWDVVHUWRU LPSO\ZKHWKHULPSOLFLWO\RUH[SOLFLWO\DQ\FRQQHFWLRQZLWKHQGRUVHPHQWRUVSRQVRUVKLSRI VXFKXVHE\WKHDXWKRUSXEOLVKHURUDQ\RWKHUSDUW\DVVRFLDWHGZLWKWKHDUWLFOH

)RUDQ\UHXVHRUGLVWULEXWLRQXVHUVPXVWLQFOXGHWKHFRS\ULJKWQRWLFHDQGPDNHFOHDUWRRWKHUV WKDWWKHDUWLFOHLVPDGHDYDLODEOHXQGHUD&UHDWLYH&RPPRQV$WWULEXWLRQOLFHQVHOLQNLQJWRWKH UHOHYDQW&UHDWLYH&RPPRQVZHESDJH

7RWKHIXOOHVWH[WHQWSHUPLWWHGE\DSSOLFDEOHODZWKHDUWLFOHLVPDGHDYDLODEOHDVLVDQGZLWKRXW UHSUHVHQWDWLRQRUZDUUDQWLHVRIDQ\NLQGZKHWKHUH[SUHVVLPSOLHGVWDWXWRU\RURWKHUZLVHDQG LQFOXGLQJZLWKRXWOLPLWDWLRQZDUUDQWLHVRIWLWOHPHUFKDQWDELOLW\ILWQHVVIRUDSDUWLFXODU SXUSRVHQRQLQIULQJHPHQWDEVHQFHRIGHIHFWVDFFXUDF\RUWKHSUHVHQFHRUDEVHQFHRIHUURUV

&UHDWLYH&RPPRQV$WWULEXWLRQ1RQ&RPPHUFLDO/LFHQVH

7KH&UHDWLYH&RPPRQV$WWULEXWLRQ1RQ&RPPHUFLDO &&%<1& /LFHQVHSHUPLWVXVH GLVWULEXWLRQDQGUHSURGXFWLRQLQDQ\PHGLXPSURYLGHGWKHRULJLQDOZRUNLVSURSHUO\FLWHGDQG LVQRWXVHGIRUFRPPHUFLDOSXUSRVHV VHHEHORZ

&UHDWLYH&RPPRQV$WWULEXWLRQ1RQ&RPPHUFLDO1R'HULYV/LFHQVH

7KH&UHDWLYH&RPPRQV$WWULEXWLRQ1RQ&RPPHUFLDO1R'HULYV/LFHQVH &&%<1&1' SHUPLWVXVHGLVWULEXWLRQDQGUHSURGXFWLRQLQDQ\PHGLXPSURYLGHGWKHRULJLQDOZRUNLV SURSHUO\FLWHGLVQRWXVHGIRUFRPPHUFLDOSXUSRVHVDQGQRPRGLILFDWLRQVRUDGDSWDWLRQVDUH PDGH VHHEHORZ

8VHE\QRQFRPPHUFLDOXVHUV

)RUQRQFRPPHUFLDODQGQRQSURPRWLRQDOSXUSRVHVLQGLYLGXDOXVHUVPD\DFFHVVGRZQORDG FRS\GLVSOD\DQGUHGLVWULEXWHWRFROOHDJXHV:LOH\2SHQ$FFHVVDUWLFOHVDVZHOODVDGDSW WUDQVODWHWH[WDQGGDWDPLQHWKHFRQWHQWVXEMHFWWRWKHIROORZLQJFRQGLWLRQV

7KHDXWKRUV PRUDOULJKWVDUHQRWFRPSURPLVHG7KHVHULJKWVLQFOXGHWKHULJKWRI SDWHUQLW\ DOVRNQRZQDVDWWULEXWLRQWKHULJKWIRUWKHDXWKRUWREHLGHQWLILHGDV VXFK DQGLQWHJULW\ WKHULJKWIRUWKHDXWKRUQRWWRKDYHWKHZRUNDOWHUHGLQVXFKDZD\ WKDWWKHDXWKRU VUHSXWDWLRQRULQWHJULW\PD\EHLPSXJQHG 

:KHUHFRQWHQWLQWKHDUWLFOHLVLGHQWLILHGDVEHORQJLQJWRDWKLUGSDUW\LWLVWKHREOLJDWLRQ RIWKHXVHUWRHQVXUHWKDWDQ\UHXVHFRPSOLHVZLWKWKHFRS\ULJKWSROLFLHVRIWKHRZQHURI WKDWFRQWHQW

,IDUWLFOHFRQWHQWLVFRSLHGGRZQORDGHGRURWKHUZLVHUHXVHGIRUQRQFRPPHUFLDO

KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

115  5LJKWVOLQN3ULQWDEOH/LFHQVH UHVHDUFKDQGHGXFDWLRQSXUSRVHVDOLQNWRWKHDSSURSULDWHELEOLRJUDSKLFFLWDWLRQ DXWKRUVMRXUQDODUWLFOHWLWOHYROXPHLVVXHSDJHQXPEHUV'2,DQGWKHOLQNWRWKH GHILQLWLYHSXEOLVKHGYHUVLRQRQ:LOH\2QOLQH/LEUDU\ VKRXOGEHPDLQWDLQHG &RS\ULJKWQRWLFHVDQGGLVFODLPHUVPXVWQRWEHGHOHWHG

$Q\WUDQVODWLRQVIRUZKLFKDSULRUWUDQVODWLRQDJUHHPHQWZLWK:LOH\KDVQRWEHHQ DJUHHGPXVWSURPLQHQWO\GLVSOD\WKHVWDWHPHQW7KLVLVDQXQRIILFLDOWUDQVODWLRQRIDQ DUWLFOHWKDWDSSHDUHGLQD:LOH\SXEOLFDWLRQ7KHSXEOLVKHUKDVQRWHQGRUVHGWKLV WUDQVODWLRQ

8VHE\FRPPHUFLDOIRUSURILWRUJDQLVDWLRQV

8VHRI:LOH\2SHQ$FFHVVDUWLFOHVIRUFRPPHUFLDOSURPRWLRQDORUPDUNHWLQJSXUSRVHV UHTXLUHVIXUWKHUH[SOLFLWSHUPLVVLRQIURP:LOH\DQGZLOOEHVXEMHFWWRDIHH&RPPHUFLDO SXUSRVHVLQFOXGH

&RS\LQJRUGRZQORDGLQJRIDUWLFOHVRUOLQNLQJWRVXFKDUWLFOHVIRUIXUWKHUUHGLVWULEXWLRQ VDOHRUOLFHQVLQJ

&RS\LQJGRZQORDGLQJRUSRVWLQJE\DVLWHRUVHUYLFHWKDWLQFRUSRUDWHVDGYHUWLVLQJZLWK VXFKFRQWHQW

7KHLQFOXVLRQRULQFRUSRUDWLRQRIDUWLFOHFRQWHQWLQRWKHUZRUNVRUVHUYLFHV RWKHUWKDQ QRUPDOTXRWDWLRQVZLWKDQDSSURSULDWHFLWDWLRQ WKDWLVWKHQDYDLODEOHIRUVDOHRU OLFHQVLQJIRUDIHH IRUH[DPSOHDFRPSLODWLRQSURGXFHGIRUPDUNHWLQJSXUSRVHV LQFOXVLRQLQDVDOHVSDFN 

8VHRIDUWLFOHFRQWHQW RWKHUWKDQQRUPDOTXRWDWLRQVZLWKDSSURSULDWHFLWDWLRQ E\IRU SURILWRUJDQLVDWLRQVIRUSURPRWLRQDOSXUSRVHV

/LQNLQJWRDUWLFOHFRQWHQWLQHPDLOVUHGLVWULEXWHGIRUSURPRWLRQDOPDUNHWLQJRU HGXFDWLRQDOSXUSRVHV

8VHIRUWKHSXUSRVHVRIPRQHWDU\UHZDUGE\PHDQVRIVDOHUHVDOHOLFHQFHORDQWUDQVIHU RURWKHUIRUPRIFRPPHUFLDOH[SORLWDWLRQVXFKDVPDUNHWLQJSURGXFWV

3ULQWUHSULQWVRI:LOH\2SHQ$FFHVVDUWLFOHVFDQEHSXUFKDVHGIURP FRUSRUDWHVDOHV#ZLOH\FRP

)XUWKHUGHWDLOVFDQEHIRXQGRQ:LOH\2QOLQH/LEUDU\ KWWSRODERXWZLOH\FRP:LOH\&'$6HFWLRQLGKWPO

2WKHU7HUPVDQG&RQGLWLRQV

Y

KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

116  5LJKWVOLQN3ULQWDEOH/LFHQVH

,I\RXZRXOGOLNHWRSD\IRUWKLVOLFHQVHQRZSOHDVHUHPLWWKLVOLFHQVHDORQJZLWK\RXU SD\PHQWPDGHSD\DEOHWR&23<5,*+7&/($5$1&(&(17(5RWKHUZLVH\RXZLOOEH LQYRLFHGZLWKLQKRXUVRIWKHOLFHQVHGDWH3D\PHQWVKRXOGEHLQWKHIRUPRIDFKHFN RUPRQH\RUGHUUHIHUHQFLQJ\RXUDFFRXQWQXPEHUDQGWKLVLQYRLFHQXPEHU 2QFH\RXUHFHLYH\RXULQYRLFHIRUWKLVRUGHU\RXPD\SD\\RXULQYRLFHE\FUHGLWFDUG 3OHDVHIROORZLQVWUXFWLRQVSURYLGHGDWWKDWWLPH

0DNH3D\PHQW7R &RS\ULJKW&OHDUDQFH&HQWHU 'HSW 32%R[ %RVWRQ0$

)RUVXJJHVWLRQVRUFRPPHQWVUHJDUGLQJWKLVRUGHUFRQWDFW5LJKWV/LQN&XVWRPHU 6XSSRUWFXVWRPHUFDUH#FRS\ULJKWFRPRU WROOIUHHLQWKH86 RU 

*UDWLVOLFHQVHV UHIHUHQFLQJLQWKH7RWDOILHOG DUHIUHH3OHDVHUHWDLQWKLVSULQWDEOH OLFHQVHIRU\RXUUHIHUHQFH1RSD\PHQWLVUHTXLUHG

KWWSVVFRS\ULJKWFRP$SS'LVSDWFK6HUYOHW 

117 Vita Melissa Barrett DeBiasse was born in Las Cruces, New Mexico on December 13, 1979. She moved with her family to Cincinnati, Ohio in 1985. She attended Milford High School where she was a member of the swim, water polo, and track teams and was the sports editor and editor-in-chief of the MHS yearbook her junior and senior years, respectively.

In August 1998, Melissa enrolled at Indiana University in Bloomington, Indiana where she majored in Biology, minored in History, and was a member of the varsity women’s water polo team. She worked as an undergraduate research assistant in the lab of Dr. Keith Clay under the mentorship of Dr. Jennifer Rudgers. As an undergraduate student, she learned to SCUBA dive and decided to pursue a career as a marine scientist.

Upon graduation, Melissa worked as an intern at the Center for Shark Research at the Mote Marine Laboratory in Sarasota, Florida under the mentorship of Drs. Michelle Heupel and Jim Gelsleichter. She began her Master’s degree at Nova Southeastern University in Fort Lauderdale, Florida in August 2003 under the mentorship of Dr. Mahmood Shivji. In 2008, Melissa began her third and final degree at Louisiana State University in Baton Rouge, Louisiana in the lab of Dr. Michael Hellberg. Melissa defended her dissertation in May of 2014 and will begin a post-doctoral position with Dr. Morgan Kelly at LSU in July 2014. Geaux Tigers.

118