High Seas Hitchhiker and Sea Turtle Symbiont: Global Genetics and Ecological Interactions of Oceanic Crabs

HIGH SEAS HITCHHIKER AND SEA TURTLE SYMBIONT: GLOBAL GENETICS AND ECOLOGICAL INTERACTIONS OF OCEANIC CRABS

JOSEPH BRYCE PFALLER

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2016

To my family, Michael Pfaller, Beverly Ringenberg and Steve Pfaller

ACKNOWLEDGMENTS

First and foremost, I want to thank my parents (Michael Pfaller and Beverly

Ringenberg) and brother (Steve Pfaller) for the love and support that they have given me during my life and throughout my career. I want to thank my mentors Karen

Bjorndal and Alan Bolten, as well as my other committee members (Gustav Paulay,

Jane Brockmann, Harvey Lillywhite and Marc Branham), for their support and guidance.

I also want to thank my non-academic family and friends who have supported me during the course of my dissertation: Rebecca Pfaller, Michael and JoAnn Frick, Dylan

Klempner, Chris and Bernie Machen, Jon and Linda Newmire, Ellen Maddux, Andre

Villanueva, Ariadna Arnau and Kim Sash. I want to thank my academic “family”:

Michael Frick, Kristina Williams, Michael Gil, Marvin Morales, Adam Payton, Kenny

Wray, Pierson Hill, Nathanael Herrera, Debbi Hannibal, Kimberly Reich, Luciano

Soares, Mark Sandfoss, Hannah Vander Zanden, Patricia Zárate, Robert Johnson,

Marco Garcia-Cruz, Melania Lopez-Castro, Peter Eliazar, Mariela Pajuelo, Alexandra

Gulick, Greg Erickson, Paul Gignac, Leslie Babonis, Nathaniel Evans, Jason Fuller and

Jamie Price.

For Chapter 2, my coauthors and I would like to thank all organizations and people that collected crabs for this study (alphabetic order): Joanna Alfaro-Shigueto

(ProDelphinus and Centre for Ecology and Conservation, University of Exeter), George

Balazs (Marine Turtle Research Program, National Ocean and Atmospheric

Administration and National Marine Fisheries Service, Pacific Islands Fisheries Science

Center), Alan Bolten (Archie Carr Center for Sea Turtle Research, University of Florida),

Michael Bresette (Inwater Research Group), Geremy Cliff (KwaZulu-Natal Shark

Board), Nathaniel Evans (Florida Museum of Natural History, University of Florida),

Angela Formia (Wildlife Conservation Society and Partenariat pour les Tortues Marines du Gabon), Bruno Giffoni (Fundação Pró-TAMAR), Michael Gil (University of Florida),

Takashi Ishihara (Sea Turtle Association of Japan and Suma Aqualife Park), Stephen

Keable (Australia Museum), Harvey Lillywhite (University of Florida), Jeffery Mangel

(ProDelphinus and Centre for Ecology and Conservation, University of Exeter), Adolfo

Marco (Estación Biológica de Doñana), Dimitris Margaritoulis (ARCHLEON, The Sea

Turtle Society of Greece), Gustav Paulay (Florida Museum of Natural History, University of Florida), Hoyt Peckham (Center for Ocean Solutions, Stanford University), Joseph

Poupin (Institut de Recherche de l'Ecole Navale), Bob Prince (Western Australian

Marine Turtle Project), Michele Schärer (University of Puerto Rico), and John Starmer

(Florida Museum of Natural History, University of Florida). I would also like to that the

PADI Foundation, Sigma Xi, the Society of Integrative and Comparative Biology and the

University of Florida Department of Biology for funding this research.

For Chapter 3, my coauthors and I would specifically like to thank all organizations and people that either supported or participated in crab collections in

Japan (Sea Turtle Association of Japan, H Fukuie, S Watanabe, K Kawano, K Saito, M

Hara, S Yamashita, K Hashimoto, K Ebisui and Y Yasuoka), Mexico (David & Lucille

Packard Foundation, US Fish & Wildlife Service, ProPeninsula, The Ocean Foundation,

Equipo ProCaguama, Grupo Tortuguero, V de la Toba, J Lucero, JM Rodríguez Barón),

Peru (Darwin Sustainable Artisanal Fisheries Initiative – Peru), and Brazil (Projecto

TAMAR/ICMBio, CEN Consulim). I would also to thank Martin Thiel and one anonymous reviewer for helpful comments.

For Chapter 4, my coauthor and I would like to thank the Sea Education

Association and the Plastics @ SEA 2012 Expedition for samples and logistical support.

I would also like to thank Alana Palau (University of Florida), Luciano Soares (University of Florida), Flávia Ribeiro (Fundação Pró-TAMAR), Thaís Pires (Fundação Pró-

TAMAR), Cecília Baptistotte (Fundação Pró-TAMAR) and Victoria Ternullo (Loggerhead

Marine Life Center) for assistance in data collection.

For Chapter 5, my coauthors and I would like to thank all organizations and people that either supported or participated in crab collections in Japan (Sea Turtle

Association of Japan, H Fukuie, S Watanabe, K Kawano, K Saito, M Hara, S

Yamashita, K Hashimoto, K Ebisui and Y Yasuoka), Hawaii, Samoa, Baja California

Sur, México (David & Lucille Packard Foundation, US Fish & Wildlife Service,

ProPeninsula, The Ocean Foundation, Equipo ProCaguama, Grupo Tortuguero V de la

Toba, J Lucero, JM Rodríguez Barón), Mexico and Central America in 2003 (NMFS

Southwest Fisheries Science Center and NOAA, and all scientific and support personnel working on the cruises), and Peru (Darwin Sustainable Artisanal Fisheries

Initiative – Peru). I also thank Jake Ferguson (University of Florida, Department of

Biology) for statistical advice.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 10

LIST OF FIGURES ...... 12

ABSTRACT ...... 14

CHAPTER

1 OCEANIC CRABS: SO EXCELLENT A FISHE ...... 16

Introduction ...... 16 A Brief Natural History of Oceanic Crabs ...... 17 Why Study Oceanic Crabs? ...... 23 Outline ...... 26 Chapter 1. Oceanic Crabs: So Excellent a Fishe ...... 27 Chapter 2. Hitchhiking the High Seas: Global Genomics of Oceanic Crabs ..... 27 Chapter 3. Social Monogamy in Planes major, a Facultative Symbiont of Loggerhead Sea Turtles ...... 27 Chapter 4. Sea Turtle Symbiosis Facilitates Social Monogamy in Oceanic Crabs ...... 27 Chapter 5. Hitchhikers Reveal Cryptic Host Behavior: New Insights from the Association Between Planes major and Sea Turtles in the Pacific Ocean .... 28 Chapter 6. Conclusions and Future Directions ...... 28

2 HITCHHIKING THE HIGH SEAS: GLOBAL GENOMICS OF OCEANIC CRABS ... 30

Introduction ...... 30 Methods ...... 33 Taxon Sampling and Justification ...... 33 DNA Extraction ...... 35 COI Amplification, Sequencing and Analyses ...... 36 Creation and Sequencing of RAD Libraries ...... 37 Processing of Sequenced RAD Tags ...... 38 Individual and Population Clustering ...... 40 Population Genomic Analyses ...... 41 Results ...... 42 COI Phylogenetic Analysis and Morphology ...... 42 RAD Libraries and Processing ...... 44 Clustering of Individuals and Populations ...... 46 Population Genomic Analyses ...... 48 Discussion ...... 52 Species-Level Relationships – Is Species Diversity Low? ...... 52

Population Structure – Is Genetic Differentiation Weak? ...... 61 Conclusions ...... 66

3 SOCIAL MONOGAMY IN PLANES MAJOR, A FACULTATIVE SYMBIONT OF LOGGERHEAD SEA TURTLES ...... 83

Introduction ...... 83 Methods ...... 87 Collection of Crabs ...... 87 Testing Hypotheses for Social Monogamy ...... 89 Testing Hypotheses for Long-Term Pairing ...... 89 Results ...... 90 Testing Hypotheses for Social Monogamy ...... 91 Testing Hypotheses for Long-Term Pairing ...... 91 Discussion ...... 93 Is Planes major Socially Monogamous on Caretta caretta? ...... 93 Is Social Monogamy in Planes major Long-Term? ...... 95

4 SEA TURTLE SYMBIOSIS FACILITATES SOCIAL MONOGAMY IN OCEANIC CRABS ...... 111

Introduction ...... 111 Methods ...... 113 Flotsam Data ...... 113 Sea Turtle Data ...... 113 Statistical Analyses ...... 114 Results ...... 114 Discussion ...... 115

5 HITCHHIKERS REVEAL CRYPTIC HOST BEHAVIOR: NEW INSIGHTS FROM THE ASSOCIATION BETWEEN PLANES MAJOR AND SEA TURTLES IN THE PACIFIC OCEAN ...... 127

Introduction ...... 127 Methods ...... 130 Results ...... 132 Discussion ...... 134 Do Turtles Display Variable/Flexible Epipelagic-Neritic Transitions? ...... 134 Do Turtles Display Similar Surface-Dwelling Behavior in Epipelagic Habitats? ...... 138 Caveats ...... 140 Perspectives ...... 141

6 CONCLUSIONS AND FUTURE DIRECTIONS ...... 151

High Seas Hitchhiker ...... 151 Monogamous Symbiont ...... 155 Oceanic Indicator ...... 158

LIST OF REFERENCES ...... 163

BIOGRAPHICAL SKETCH ...... 180

LIST OF TABLES

Table page

1-1 Summary of all known references to Planes minutus and Planes major on sea turtles in the Atlantic and Indian oceans, and Mediterranean Sea ...... 29

2-1 Sample sizes by putative species, region, habitat and genetic analysis ...... 67

2-2 Summary statistics of restriction-site associated DNA-sequencing data processing ...... 68

2-3 Pairwise comparison of genetic distance and associated P-values, observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 1 ...... 69

2-4 Pairwise comparison of genetic distance and associated P-values, observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 2 ...... 70

2-5 Pairwise comparison of genetic distance and associated P-values, observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 3 ...... 71

2-6 Analysis of molecular variance among 11 regions ...... 72

2-7 Pairwise comparison of genetic distance and associated P-values among 11 ocean regions from AMOVA ...... 73

3-1 Population and sex distribution of Planes major associated with Caretta caretta ...... 103

3-2 Relative growth of cheliped size in females and males of Planes major ...... 104

4-1 Information used to calculate surface area of plastic items ...... 118

4-2 Characteristics of plastic flotsam ...... 119

4-3 Summary of data on the population and sex distribution of Planes crabs associated with oceanic-stage loggerhead turtles ...... 121

4-4 Summary of statistical model comparisons for log-transformed crab count data, using corrected Aikake Information Criterion ...... 122

5-1 Comparison of frequency of epibiosis among turtle species, site and habitat .. 143

5-2 Summary of 13 Fisher’s Exact Tests comparing frequency of epibiosis among turtle species, site and habitat ...... 144

5-3 Results of binomial logistic regression analyses testing the effect of turtle size on frequency of occurrence ...... 145

5-4 Summary of all known references to Planes major on sea turtles in the Pacific Ocean ...... 146

LIST OF FIGURES

Figure page

2-1 Map showing collecting locations in 13 ocean regions for each putative species ...... 74

2-2 Maximum-likelihood phylogenetic analysis of the mitochondrial gene COI for the family Grapsidae ...... 75

2-3 Venn diagram showing the distribution of SNP loci among RAD datasets ...... 76

2-4 Histogram of haplotype divergence among loci in RAD dataset 1, RAD dataset 2 and RAD dataset 3...... 77

2-5 Plots comparing statistical support for different numbers of populations in STRUCTURE and AWCLUST for RAD dataset 1, RAD dataset 2 and RAD dataset 3...... 78

2-6 Results from clustering analyses of RAD dataset 1 at K=2 and K=4 ...... 79

2-7 Results from clustering analyses of RAD dataset 2 at K=2 and K=3 ...... 80

2-8 Results from clustering analyses of RAD dataset 3 at K=2 and K=4...... 81

2-9 Heatmap showing pairwise comparisons of genetic distance and associated P-values for 11 ocean regions from AMOVA ...... 82

3-1 Planes major heterosexual pair hiding within the supracaudal space of juvenile Caretta caretta ...... 105

3-2 Population distribution of the crab Planes major, symbiotic with the loggerhead sea turtle Caretta caretta ...... 106

3-3 Male-female association pattern of Planes major found as heterosexual pairs on Caretta caretta ...... 107

3-4 Relationship between carapace width of females and males of Planes major found as heterosexual pairs within the supracaudal/inguinal space of the loggerhead turtles Caretta caretta ...... 108

3-5 Relationship between curved carapace length of loggerhead turtles Caretta caretta and carapace width of females and males of the symbiotic crab Planes major...... 109

3-6 Patterns of sexual dimorphism in Planes major ...... 110

4-1 Statistical relationships between turtle body size, and turtle surface area, and turtle refuge surface area from a size series of captive and dead loggerheads from Florida and Brazil...... 123

4-2 Diagram and equation for modeling turtle surface area as one-quarter of an oblong ellipsoid ...... 124

4-3 Diagram and equation for modeling refuge surface area of sea turtles as an isosceles right pentagon with two parallel sides...... 125

4-4 Best-fit models of the relationship between submerged surface area and refuge surface area and number of adult crabs for flotsam with mean values for sea turtles superimposed...... 126

5-1 Maps showing sites, turtle-capture locations, and presence or absence of Planes major...... 147

5-2 Size-frequency histograms showing loggerhead turtles that did not host Planes major and turtles that hosted at least one Pl. major ...... 148

5-3 Size-frequency histograms showing green turtles that did not host Planes major and turtles that hosted at least one Pl. major ...... 149

5-4 Size-frequency histograms showing olive ridley turtles that did not host Planes major and turtles that hosted at least one Pl. major ...... 150

6-1 Conceptual model of epibiosis (reproduced from Frick and Pfaller 2013) ...... 162

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

HIGH SEAS HITCHHIKER AND SEA TURTLE SYMBIONT: GLOBAL GENETICS AND ECOLOGICAL INTERACTIONS OF OCEANIC CRABS

Joseph Bryce Pfaller

May 2016

Chair: Karen A. Bjorndal Major: Zoology

Oceanic crabs of the genus Planes are common and conspicuous members of the community of organisms that occupy the surface of the open ocean. Unlike other crabs that occupy intertidal or benthic habitats, Planes crabs spend their entire lives rafting on surface-drifting oceanic flotsam or as symbionts of sea turtles. This lifestyle has important implications for the global genetic structure and ecological interactions of these crabs.

Using mitochondrial and genomic data, I test the prediction that the ability of

Planes to disperse widely, not only as pelagic larvae, but also as adults associated with oceanic flotsam and sea turtles, should limit species diversification and population differentiation. I found low species diversity and weak population differentiation among rafting Planes and Pachygrapsus laevimanus, an intertidal species found to be closely related to Planes. Global genomic analyses indicate that there is no differentiation within ocean gyres, weak differentiation between gyres and moderate differentiation between ocean basins.

Using ecological data from crabs collected on sea turtles and oceanic flotsam, I test the predictions that (1) Planes major shows long-term social monogamy on sea

turtles and (2) sea turtle symbiosis facilitates social monogamy via constraints imposed by refuge size. I found strong evidence that Pl. major is socially monogamous on sea turtles, but that pairing is not necessarily long-term. Moreover, I found that refuge area, not total area, on turtles and flotsam modulate group size and composition of adult crabs and therefore facilitates social monogamy by crabs on turtles.

Using ecological data from crabs collected from three different turtle species captured in neritic and epipelagic habitats, I use crabs as indicators of epipelagic habitat use and surface-dwelling behavior to evaluate two questions: (1) Do turtles display variable/flexible epipelagic-neritic transitions? and (2) Do turtles display similar surface- dwelling behavior in epipelagic habitats? I found evidence that loggerhead and olive ridley turtles display variable/flexible epipelagic-neritic transitions, while green turtles do not. Moreover, I found evidence that epipelagic loggerheads (Caretta caretta) tend to spend more time at or near the surface than epipelagic olive ridley (Lepidochelys olivacea) and green turtles (Chelonia mydas).

CHAPTER 1 OCEANIC CRABS: SO EXCELLENT A FISHE

Introduction

To learn anything about the natural history of a wide-ranging seafarer such as Chelonia, the main problem is to keep the animal in view. There are two times in the life of a sea turtle when a zoologist can count on making contact with it: when it hatches, and when the female goes ashore to nest. Everything else is done away off somewhere out of sight, and has to be reconstructed by deduction from fragments of observation (Carr 1967, p. 25).

So begins Chapter 2 of So Excellent A Fishe by Archie Carr, a book that summarized much of the enigmatic natural history of sea turtles as of 1967 and has inspired half a century of sea turtle biologists and conservationists since its publication, myself included. Although this dissertation does not focus on sea turtles, this conundrum of observation is applicable to another wide-ranging seafarer whose natural history is intertwined with that of sea turtles. Oceanic crabs of the genus Planes are open ocean wanderers, where the intricacies of their natural history remained hidden from biologists for centuries after their initial discovery. Ironically, it wasn’t until studies of sea turtle natural history moved beyond the nesting beach, a limitation indicated by

Carr (1967), that the world of Planes crabs was exposed. Opportunities to capture sea turtles at sea – even in the open ocean – provided the opportunity and impetus to learn more about the natural history of these crabs and their peculiar, and surprisingly frequent, association with sea turtles. To date, studies by sea turtle biologists seeking to understand the role that sea turtles play in the lives of these crabs, and others inspired by the eccentricities of a wandering seafarer, have revealed many biological mysteries of Planes crabs. However, like many secrets of the ocean, other mysteries still remain ‘away off somewhere out of sight.’

A Brief Natural History of Oceanic Crabs

Oceanic crabs of the genus Planes have captured the interest of sailors and biologist alike. On September 17, 1492, Christopher Columbus and his crew, while in the middle of the North Atlantic during their first voyage to the New World, noted the presence of small crabs rafting aboard mats of “weeds” and speculated incorrectly “that these crabs were sure signs of land” (Markham 1893 via Chace 1951). Unbeknownst to

Columbus, his ‘discovery’ of the New World was preceded by the discovery of what were likely Planes crabs. Two centuries later, the aptly named ‘Columbus’ crab (Planes minutus) was described by Sloane (1725) and formally named by Linnaeus (1758).

Linnaeus (1747) also described a similar crab in the Pacific, which eventually became

Planes major (MacLeay 1838). Nearly two centuries after that, Rathbun (1914) described a third species, Planes marinus, rafting west of Baja California. After some taxonomic shuffling and numerous attempts to erect new species, these three species are still recognized today (Ng et al. 2008).

In the seminal work on Planes crabs, Fenner Chace (1951) compiled scattered references to Planes throughout the literature, and reevaluated the morphological status and biogeography of each species from museum specimens. In this work, Planes marinus was deemed to be sufficiently distinct from other Planes to be moved to

Pachygrapsus, but was later returned to Planes upon review of additional specimens

(Chace 1966). Despite quantitative rigor, the morphological characters identified by

Chace (1951) to differentiate the three Planes species are quite subtle, often overlapping, and highly dependent on the body size. Consequently, none of the characters appear to be truly diagnostic. The carapace of Pl. marinus tends to be more quadrate, slightly wider than long and laterally striated, while the carapace of Pl.

minutus and Pl. major tends to be laterally convex, not striated, and either approximately equal in width and length (Pl. minutus) or slightly longer than wide (Pl. major). However, the ranges of the ratio of carapace length:width overlap: Pl. marinus

(1:1.07-1.16), Pl. minutus (1:0.91-1.12) and Pl. major (1:0.86-1.06). Additionally, discontinuities in the morphology of walking limbs were emphasized by Chace (1951) and subsequently employed by others. Chace (1951) stated that the walking limbs of

Pl. marinus tend to be short, “not noticeably flattened,” and without a natatory fringe, while the walking limbs of Pl. minutus and Pl. major are longer, flatter and with a natatory fringe. However, the proportion of length of the three distal segments of the second limb to carapace length overlaps between Pl. marinus (0.77-0.99), Pl. minutus

(0.83-1.07) and Pl. major (0.68-0.89). Moreover, the absence of a natatory fringe in Pl. marinus was part of the justification for moving this species to Pachygrapsus (Chace

1951), but the presence of natatory fringes on additional specimens of Pl. marinus prompted its return to Planes (Chace 1966).

Other external traits related to cheliped shape and male abdomen and gonopod morphology show subtle differences between the three species, but nothing that appears to be diagnostic. More recently, Frick et al. (2011) compared the morphology of the masticatory structures of Planes and found many conserved traits, as well as some consistent differences among species in the zygocardiac and urocardiac ossicles and cardio-pyloric valves. However, it is difficult to judge whether these differences are related to ecology (i.e., geographic variation or phenotypic plasticity) or ancestry (i.e., common descent and thus associated with speciation events). Clearly, the extent of intraspecific variation found within a fairly conserved overall morphology has made

identifying diagnostic morphological traits difficult in Planes. Alternatively, of course, the lack of diagnostic morphological differences may be the result of genetic exchange and the lack of true species boundaries.

Recent phylogenetic analyses of the family Grapsidae suggest that the genus

Planes is paraphyletic due to the well-supported inclusion of Pachygrapsus laevimanus

(Schubart 2011; Ip et al. 2015), which is an intertidal species found across a narrow band of the South Pacific from Australia to Rapa Island (Poupin et al. 2005). Among the

Planes species, Pa. laevimanus is morphologically more similar to Pl. marinus in that the carapace tends to be quadrate and slightly broader than long, and laterally striated.

While Pl. marinus has been shuffled between Planes and Pachygrapsus (Chace 1951,

1966), the affinity between Pa. laevimanus and Planes was not recognized until genetic data were analyzed. The discovery of this puzzling relationship, as well as other unexpected affinities between Pachygrapsus and other grapsid species (Schubart 2011;

Ip et al. 2015), has left the taxonomy of this family in disarray.

Planes crabs are found throughout the subtropical and tropical seas of the world.

Chace (1951) more or less arbitrarily called specimens collected in the Atlantic and

Indian oceans Pl. minutus and specimens collected in the Pacific Ocean Pl. major.

Over time, the subtle morphological differences between Pl. minutus and Pl. major described by Chace (1951) were found mostly between individuals in the North and

South Atlantic (Juanicó 1976; Manning and Holthuis 1981), leading to another rather arbitrary designation for specimens collected in the North Atlantic being called Pl. minutus and specimens from everywhere outside the North Atlantic being called Pl. major. Planes marinus is found far less frequently and tends to be sympatric with Pl.

major (Chace 1951). However, Pl. marinus were recently collected from flotsam near the island of St. Martin in the Caribbean (Florida Museum of Natural History), and a reexamination of specimens collected from loggerhead turtles in Florida (Frick et al.

2006) revealed one specimen of Pl. marinus (Pfaller unpublished data). Consequently,

Pl. marinus appears to be globally distributed. Although the nonpolar distribution of

Planes likely reflects its inability to survive exceedingly cold temperatures, crabs are occasionally found in cold temperate areas outside of their normal range (Dell 1963;

O’Riordan 1976; Spivak and Bas 1999).

Planes are considered obligate rafters of surface-drifting oceanic flotsam and oceanic animals. Inanimate substrata include logs, leaves, volcanic pumice, derelict buoys and other anthropogenic debris (Chace 1951; Juanicó 1976; Dellinger et al.

1997; Spivak and Bas 1999; Frick et al. 2004; Pons et al. 2011; Gil and Pfaller 2016).

Living hosts include algae (Sargassum spp.), hydrozoans (Velella spp.), gastropods

(Janthina spp.), sea snakes (Hydrophis [Pelamis] platurus), and four species of sea turtles (Chace 1951; Dellinger et al. 1997; Frick et al. 2011; Pfaller et al. 2012, 2014a). I have compiled all known references and data on Planes-sea turtle interactions in Tables

1-1 and 5-4. Planes minutus is found frequently on loggerhead turtles (Caretta caretta) in the North Atlantic and Mediterranean Sea (Dellinger et al. 1997; Frick et al. 2004;

Casale et al. 2004, 2012). Planes major is found frequently on loggerhead and olive ridley turtles (Lepidochelys olivacea) and less frequently on hawksbill (Eretmochelys imbricata) and green turtles (Chelonia mydas) in the South Atlantic and Pacific oceans

(Carranza et al. 2003; Frick et al. 2011; Bugoni et al. 2007). Planes marinus has only been found associated with L. olivacea in the eastern tropical Pacific Ocean (Frick et al.

2011). Although there is evidence that sea turtles represent higher-quality substrata, especially for adult crabs (see below; Dellinger et al. 1997; Frick et al. 2004), no study has attempted to evaluate habitat preferences or selectivity in Planes or the degree of switching among hosts or substrata

A study by Davenport (1992) revealed several interesting morphological and behavioral adaptations to rafting in Planes. The walking limbs in Planes are flattened, rotated into the ‘advanced’ locomotory orientation for forward swimming, and complete with a natatory fringe of plumose hairs on the leading edge of each limb to increase the propulsive area. Although swimming was found to be energetically efficient (crabs do not incur oxygen debt while swimming), crabs show somewhat limited swimming endurance, remaining at the surface for only 35-45 min before sinking. This is different than swimming crabs (Portunidae), which can swim at the surface for long periods of time (Hartnoll 1971). Strongly hooked dactyls at the terminus of each leg are likely important for maintaining contact with bobbing rafts and swimming sea turtles. The

Velcro-like properties of these appendages are even apparent in preserved museum specimens. Davenport (1992) also found that crabs in the laboratory rarely stray more than 5 cm from floating substrata to catch food and they do so with a unique locomotory behavior. Crabs ‘rev-up’ with leg pairs 1-3 to reach working speed while leg pair 4 is still attached to the substratum, then detach, rapidly accelerate toward a food item and grab it, then return quickly to the substratum. Crabs will attack any potential food item that comes within ~5 cm of their raft. Diets of crabs on flotsam include small fishes and squid, euphausiids, isopods, sea skaters (Insecta), algae, barnacles and other Planes

(Davenport 1992; Frick et al. 2004). Crabs held in the laboratory were also found to

‘store’ prey items that were too large to be ingested in one meal, which might be an adaptation in response to the unpredictability of foraging opportunities in the open ocean.

Much of the natural history of Planes is known from specimens collected as symbionts of sea turtles. In some oceanic habitats, crabs are found on turtles at quite high frequencies (e.g., 82%, Dellinger et al. 1997; 83%, Carranza et al. 2003; 100%,

Frick et al. 2011). Crabs on turtles are often found hiding in the supracaudal, and sometimes inguinal, spaces of host turtles (Dellinger et al. 1997), but dietary data suggest that crabs use the whole body of host turtles as a foraging platform (Frick et al.

2004, 2011) and are not coprophagous as suggested by Crane (1937). Crabs will feed on other epibionts living on the turtles, as well as fragments of dietary items consumed by turtles that drift within reach while turtles are feeding (Davenport 1992; Frick et al.

2004, 2011). In this ‘cleaning’ association, crabs associated with turtles tend to have greater dietary diversity than crabs associated with flotsam in the same area (Davenport

1992; Frick et al. 2004). Moreover, crabs associated with turtles tend to be larger, more frequently ovigerous, and have less limb damage than crabs on flotsam (Dellinger et al.

1997; Frick et al. 2004), suggesting that turtles are a higher quality substrata, especially for adult crabs. The social structure of crabs tends to be different on different substrata: solitary adults or heterosexual pairs of adults on turtles and large demographically mixed groupings on flotsam (Dellinger et al. 1997; Frick et al. 2004, 2006, 2011; Pons et al. 2011). Dellinger et al. (1997) hypothesized that the total surface area provided by turtles and flotsam dictate the social structure of crabs on different substrata, such that one male and one female crab can defend the relatively small area provided by turtles.

However, no rigorous analysis comparing the surface area of turtles and flotsam and associated crab groups has been carried out to test this or any alternative hypotheses.

Why Study Oceanic Crabs?

The original impetus for this study was based primarily out of biological curiosity following the discovery of megalopal-stage crab larvae rafting on the bodies of pelagic sea snakes (Pfaller et al. 2012). However, there are several aspects of the biology of oceanic crabs that make studying them of interest to a broad range of biologists. Of these, I have selected three as the primary foci of this dissertation.

First, Planes are globally distributed and display vast dispersal potential. In regards to Planes, Chace (1951) stated that ‘a species believed to be identical and common in all the warmer seas of the world soon loses interest.’ To this, I could not disagree more. Such an animal would be usual and very much worthy of further investigation. In general, animal populations are structured by their ability and propensity to successfully disperse. Thus, understanding the consequences of dispersal ability is fundamental to our understanding of population and community ecology, as well as the origin and maintenance of biological diversity (Treml et al. 2015).

For Planes, the ability to disperse not only as pelagic larvae but also as adults associated with oceanic flotsam and sea turtles should vastly increase their dispersal potential, leading to limited population differentiation and reduced opportunities of diversification. On the other end of the spectrum, animals with limited dispersal ability should have strong population differentiation and enhanced opportunities for diversification. Because estimating connectivity is more effective and less prone to error when there is significant genetic structure among populations (i.e., when connectivity is low), the literature is somewhat biased towards positive examples

(Hedgecock et al. 2007). Moreover, our understanding of the population biology and connectivity in animals with high connectivity (e.g., Planes) is not only limited, but might also be somewhat biased by a lack of genetic resolution provided by traditional genotypic markers used in the past. In light of recent advances in high throughput sequencing technologies that allow genome-wide genetic variation to be incorporated into population genetic analyses of non-model organisms (Reitzel et al. 2013), there is now an opportunity to address this gap in knowledge. Studying the species-level relationships and global population structure of Planes crabs using these new technologies will provide unprecedented resolution into the systematics and global biogeography of this group, and contribute valuable information to our understanding of how theoretical dispersal potential relates to actual population differentiation and diversification among marine organisms.

Second, the lineage of crabs leading to Planes has undergone a significant ecological shift in habitat (intertidal to rafting) that has necessitated concomitant changes in morphology and behavior. Such transitions provide an opportunity to assess how different ecological forces drive evolutionary adaptations. In particular, the evolution of different animal mating strategies is ultimately determined by specific ecological factors that dictate the spatial and temporal distribution of available resources and mates (Emlen and Oring 1977). Because oceanic flotsam and sea turtles tend to be variable in size and mostly unpredictable in time and space, the adoption of different group characteristics and mating strategies in Planes provides a means to better understand how resource characteristics structure animal groups and mating strategies.

Baeza and Thiel (2007) outline a general framework for understanding how host

characteristics and ecology affect the mating system and social behavior of symbiotic crustaceans. Under this theoretical framework, reproductive strategies of symbiotic crustaceans can be predicted based on four parameters: (1) host body size relative to symbiont body size, (2) host structural complexity, (3) host abundance, and (4) the risk of mortality for symbionts away from hosts. Planes is expected to exhibit social monogamy (defined here as the cohabitation of heterosexual pairs over some period of time) and long-term heterosexual pairing when associated with sea turtles because turtle hosts are relatively small in body size and the supracaudal space is structurally simple, and are found in relatively low abundance in habitats where the risk of mortality for symbionts (e.g., predation and exhaustion) away from hosts is likely high. Because this theoretical framework was developed primarily for obligate symbionts living in or on benthic macro-invertebrates, testing this hypothesis in Planes would represent a novel test of theory in a host-vertebrate, symbiont-invertebrate system. Additionally, because

Planes is a facultative symbiont that also inhabits oceanic flotsam that can be highly variable in size and complexity, there is an opportunity to test how refuge area affects group characteristics and the adoption of different mating systems by comparing group size and composition of Planes on different substrata. Such information would broadly inform our understanding of how ecological factors contribute to the evolution of different animal mating systems.

Third, Planes are symbiotic with sea turtles, a group that is endangered because their cryptic lifestyle makes populations difficult to study and monitor. Studies that incorporate information from habitat-specific ecological interactions (e.g., interactions with oceanic Planes) can reveal valuable insights into the patterns of habitat use and

behavior of sea turtles (Frick and Pfaller 2013) and inform their conservation. Because

Planes likely colonize turtles in epipelagic habitats (upper 200 m in areas with >200 m depth) and likely dissociate with turtles when turtles transition to neritic habitats

(Dellinger et al. 1997; Frick et al. 2006), the presence of crabs on neritic turtles strongly suggests that those turtles recently occupied epipelagic habitats. Thus, differences among neritic turtles (species or size classes) in the frequency of crab occurrence should reflect differences in epipelagic to neritic transitions. Moreover, because Planes likely colonize and persist on epipelagic turtles that spend a significant proportion of time at or near the surface (Davenport 1992; Dellinger et al. 1997), differences among epipelagic turtles in the frequency of crab occurrence should reflect differences in epipelagic surface-dwelling behavior. Because epibiosis necessitates spatial overlap between the habitats occupied by hosts and free-living populations of epibionts, researchers can use the presence of epibionts to identify the habitats that the host has recently occupied (Frick and Pfaller 2013). This approach may prove to be informative for other sea turtle populations, as well as other marine vertebrates. In addition to epibionts associated with epipelagic versus neritic/benthic habitat use, other epibiotic assemblages that reflect dichotomies in habitat use in the aquatic environment (e.g., freshwater versus marine, polar versus equatorial) may reveal important information on cryptic host movements and behavior.

Outline

The chapters to follow address the following topics:

Chapter 1. Oceanic Crabs: So Excellent a Fishe

Chapter 2. Hitchhiking the High Seas: Global Genomics of Oceanic Crabs

With Adam Payton, Stuart McDaniel and Karen Bjorndal as coauthors, I used mitochondrial and genomic data to test the hypothesis that because crabs have the ability to disperse widely as pelagic larvae and as adults associated with oceanic flotsam and sea turtles, I should find low species diversity and weak population differentiation within this group.

Chapter 3. Social Monogamy in Planes major, a Facultative Symbiont of Loggerhead Sea Turtles

With Joanna Alfaro-Shigueto, Bruno Giffoni, Takashi Ishihara, Jeffery Mangel, Hoyt

Peckham, Karen Bjorndal and Antonio Baeza as coauthors, I used ecological data from crabs collected from loggerhead turtles to test the hypothesis that because of the host and environmental characteristics of this system, I should find that crabs on turtles display social monogamy and that heterosexual pairing is long-term (published article,

Pfaller et al. 2014b).

Chapter 4. Sea Turtle Symbiosis Facilitates Social Monogamy in Oceanic Crabs

With Michael Gil as a coauthor, I used ecological data from crabs collected from sea turtles and flotsam to test the hypothesis that if refuge size is a fundamental predictor of group size and composition in this system, then I should find that refuge area is a better predictor of adult crab number than total area for both flotsam and turtles, and that flotsam and turtles with similar refuge area host a similar number and composition of adult crabs.

Chapter 5. Hitchhikers Reveal Cryptic Host Behavior: New Insights from the Association Between Planes major and Sea Turtles in the Pacific Ocean

With Joanna Alfaro-Shigueto, George Balazs, Takashi Ishihara, Kerry Kopitsky, Jeffery

Mangel, Hoyt Peckham, Alan Bolten and Karen Bjorndal as coauthors, I used crabs collected from three turtle species in the Pacific Ocean as indicators of epipelagic habitat use and surface-dwelling behavior to evaluate two questions: Do turtles display variable/flexible epipelagic-neritic transitions, and do turtles display similar surface- dwelling behavior in epipelagic habitat (published article, Pfaller et al. 2014a).

Chapter 6. Conclusions and Future Directions

Table 1-1. Summary of all known references to Planes minutus and Planes major on sea turtles in the Atlantic and Indian oceans, and Mediterranean Sea

Site Habitat Species Stage N F0 (%) Reference Atlantic Ocean USA, Maine N CC A 1 - Chace (1951) USA, Florida B CC A 138 2 Caine (1986) USA, Georgia B CC A 2 3.1 Frick et al. (1998) USA, Georgia B CC A 21 - Frick et al. (2000) USA, Florida N CC S,A 594 3.7 Frick et al. (2006) USA, Rhode Is. N EIa A 1 - Chace (1951) Jamaica - UNK - 1 - Browne (1789) Puerto Rico N EI S 105 3 Schärer (2001) France N EIa - 1 - Couch and Bate (1878) Azores E CC J 17 - Frick et al. (2003) Azores E CC J 38 - Frick et al. (2004) Madeira E CC J 3 - Davenport (1994) Madeira E,N CC S,A 128 82 Dellinger et al. (1997) Madeira E EI - 1 - Murray (1895) Canary Islands E,S CC J,S 167 12 Liria Loza (2011) Cape Verde B CC A 41 2.4 Liria Loza et al. (2008) Cape Verde B CC A 4 - Pfaller unpublished data Gabon E,B LO A 2 - Pfaller unpublished data Brazil E CC J,A 18 - Bugoni et al. (2007) Brazil E CM J 1 - Bugoni et al. (2007) Brazil E EI J 1 - Bugoni et al. (2007) Uruguay N CC S,A 18 83 Carranza et al. (2003) Uruguay E CC J >72 - Pons et al. (2011) Mediterranean Sea Western E CC - 2 - Chevreux and de Guerne (1893) Western S CC J 13 - Badillo et al. (2001) Western S CC J 104 >10 Domènech et al. (2015) Eastern - CC - 2 - Roux (1828) Eastern N CC - 1 - Milne-Edwards and Bouvier (1899) Eastern N,B CC J,S 107 - Gramentz (1988) Eastern E,N,B,S CC J,S,A 2359 27 Casale et al. (2004) Eastern E,N CC J,S 117 20 Casale et al. (2012) Indian Ocean South Africa E CC A 1 - Pfaller unpublished data South Africa E CM ? 1 - Pfaller unpublished data Oman N CC A 12 0 Pfaller unpublished data W. Australia N EI - 1 - McMillan (1968) Notes. F0, Frequency of epibiosis. Habitat: E, epipelagic; N, neritic; B, nesting beach; S, stranding. Species: CC, Caretta caretta; CM, Chelonia mydas; EI, Eretmochelys imbricata; LO, Lepidochelys olivacea; UNK, unknown turtle species. Stage: J, juvenile; S, subadult; A, adult. Dashes, insufficient data for determining habitat, stage or F0. a Based on species distributions, these turtles were likely C. caretta not E. imbricata

CHAPTER 2 HITCHHIKING THE HIGH SEAS: GLOBAL GENOMICS OF OCEANIC CRABS

Introduction

Population differentiation and ultimately diversification depend in large part on the ability and propensity of organisms to successfully disperse (Palumbi 1994, 2003).

In general, organisms with high dispersal ability (i.e., those capable of long-distance dispersal) are predicted to have wide geographic ranges and weak population differentiation or structure. In theory, high levels of gene flow among distant populations would tend to stymy local adaptation or random drift (Avise 2000; Goetze 2005), leading to reduced opportunities for diversification. Conversely, organisms with weak dispersal ability (i.e., those without dispersive life stages or propagules) are predicted to have restricted geographic ranges, high degrees of population differentiation and structure, and enhanced opportunities for local adaptation and subsequent diversification

(Palumbi 1994, 2003; Avise 2000). Thus, understanding the consequences of dispersal ability is fundamental to our understanding of population and community ecology, as well as the origin and maintenance of biological diversity (McPeek and Holt 1992;

Lenormand 2002; Treml et al. 2015).

Among marine animals, considerable emphasis has been given to the role that pelagic larval duration (PLD) plays in the dispersal and connectivity of populations

(Faubry and Barber 2012). In short, because adults of many marine animals are non- dispersive – often benthic and sessile or sedentary – dispersal is primarily restricted to the vagility of pelagic larvae. As a result, PLD is often (Cowen et al. 2000; Cowen and

Sponaugle 2009), but not always (Bradbury et al. 2008; Weersing and Toonen 2009), correlated with the geographic range and degree of population differentiation (e.g., FST)

of a species. However, in some marine animals, long-distance dispersal is not limited to the movements of pelagic larvae. In particular, the communities of animals associated with the ocean’s air-water interface (termed Neuston: Naumann 1917 via Marshall and

Burchardt 2005) also disperse as juveniles and adults while drifting or rafting at the surface of the open ocean. In these animals, dispersal by pelagic larvae (if present) is then augmented considerably by the vast dispersal potential of adults and juveniles, which can use large ocean currents to disperse across ocean basins and perhaps further (Thiel and Haye 2006). In theory, the heightened dispersal ability of these animals should lead to extremely wide geographic ranges, little to no population differentiation or structure, and limited opportunities for local adaptation and diversification. Nevertheless, unlike many members of benthic communities, we know far less with respect to how theoretical dispersal potential relates to actual population differentiation and diversification within the neustonic community.

Marine organisms with large populations and high dispersal ability often pose specific challenges for those interested in estimating population connectivity and structure. Because tracking dispersal of individuals within the vast ocean is often logistically impossible, genetic data have become the primary tool by which researchers assess population connectivity at large spatial scales (Hedgecock et al. 2007; Lowe and

Allendorf 2010). However, because estimating connectivity is more effective and less prone to error when there is significant genetic structure among populations (i.e., when connectivity is low), the literature is somewhat biased towards positive examples

(Hedgecock et al. 2007). When population connectivity is high and populations are large (e.g., in neustonic animals), it becomes difficult to accurately estimate

contemporary population structure with limited genetic resolution offered by traditional genotypic markers (Goetze 2005; McCormack et al. 2013; Benestan et al. 2015).

Recent advances in high-throughput sequencing technologies have allowed genome- wide genetic variation to be incorporated in population genetic analyses of non-model organisms (Reitzel et al. 2013), providing unprecedented genetic resolution into the population biology and connectivity of previously enigmatic groups of organisms.

In this study, we use next-generation sequencing of genome-wide restriction-site associated DNA tags (RADseq), as well as more traditional mitochondrial DNA sequence data, to investigate the species-level relationships and global population structure of Planes crabs, a common and conspicuous member of neustonic communities worldwide. Unlike intertidal grapsid crabs, which rely almost exclusively on multi-staged pelagic larvae for long-distance dispersal and genetic exchange (Anger

1995), Planes also disperse as adults and juveniles while rafting on surface-drifting oceanic debris or flotsam, and as facultative symbionts of oceanic-stage sea turtles

(Chace 1951). Traditional genetic analyses of intertidal grapsid crabs show high levels of gene flow along widely dispersed latitudinal gradients, but low transoceanic gene flow, indicating that large ocean basins represent significant barriers to pelagic larval dispersal (Schubart et al. 2005; Cassone and Boulding 2006). Moreover, restricted transoceanic gene flow between sister species has been identified as a potential mechanism leading to diversification within the family Grapsidae (Schubart 2011). For

Planes, the ability to disperse, not only as pelagic larvae, but also as adults associated with oceanic flotsam and sea turtles should vastly increase their dispersal range and facilitate transoceanic, if not near global, genetic exchange. For this reason, we predict

that we will find low species diversity within this group of crabs and weak genetic differentiation among widely separated populations worldwide.

Crabs of the genus Planes are found throughout the temperate and tropical oceans of the world (Chace 1951), and there are three currently recognized species:

Planes minutus (N. Atlantic and Mediterranean), Planes major (worldwide, except N.

Atlantic), and Planes marinus (worldwide, except N. Atlantic) (Chace 1951; Ng et al.

2008). While Pl. marinus is morphologically distinct, Pl. minutus and Pl. major show very subtle morphological differences (Chace 1951). Recent phylogenetic analysis of the family Grapsidae suggest that the genus Planes is actually paraphyletic due to the well-supported inclusion of a fourth putative species, Pachygrapsus laevimanus

(Schubart 2011; Ip et al. 2015), which is an intertidal species found across a narrow band of the South Pacific from Australia to Rapa Island (Poupin et al. 2005). An analysis of the global species diversity and population-level differentiation of Planes and

Pachygrapsus laevimanus has never been conducted. The goal of this study was to conduct such an analysis with nearly complete taxonomic and geographic sampling, and genomic-level resolution.

Methods

Taxon Sampling and Justification

Crab specimens were collected or obtained from 27 sites within 13 broad ocean regions corresponding to the East and West sides of each major ocean gyre, the central

Pacific and Mediterranean Sea (Figure 2-1; groupings were not based on specific biogeographic boundaries). Each specimen was given an a priori species designation based on external morphology, habitat and/or geography following Chace (1951) and

Poupin et al. (2005): (1) Pa. laevimanus and Pl. marinus were separated by habitat

(intertidal versus rafting, respectively), relative carapace length:width (1:>1.2 versus

1:<1.2, respectively) and setation patterns on the walking limbs (no natatory fringe versus natatory fringe, respectively), (2) Pl. marinus and Pl. minutus/Pl. major were separated by the shape and striated-ness of carapace (quadrate and distinctly striated versus round and not distinctly striated, respectively) and relative carapace length:width

(1:>1.1 versus 1:<1.1, respectively), and (3) Pl. minutus and Pl. major were separated by geography (North Atlantic and Mediterranean Sea versus South Atlantic, Indian and

Pacific, respectively). Pachygrapsus laevimanus specimens were collected intertidally amongst rocks at three sites across its known geographic range (Poupin et al. 2005).

Planes specimens were collected from surface-drifting oceanic debris and sea turtles

(Caretta caretta, Chelonia mydas, Eretmochelys imbricata and Lepidochelys olivacea) at 24 sites within all 13 ocean regions. For Pa. laevimanus and Pl. marinus, only representative samples (1-5 individuals) were obtained from each site. For Pl. minutus and Pl. major, representative samples were collected from sites within each ocean region and larger samples (>10 individuals) were collected from some sites when possible (Table 2-1). Most specimens were collected specifically for this study from

2010 to 2013; however, specimens from unsampled regions or species were also acquired from earlier collections, some of which were initially collected as early as 1995.

Specimens were preserved and stored in 70-95% ethanol prior to DNA extraction.

Because the duration of storage (0.5-18 yrs) and therefore the extent of DNA degradation were highly variable among specimens, some DNA samples failed during either mitochondrial or genomic analyses. Consequently, the same set of individuals was not used in both analyses (Table 2-1).

Prior to evaluating genetic patterns within Planes and Pa. laevimanus using genomic RADseq methods (see below), we first conducted a single-locus mitochondrial

DNA (mtDNA) phylogenetic analysis within the family Grapsidae to (1) evaluate statistical support for the relationship between Pa. laevimanus and Planes (as in

Schubart 2011; Ip et al. 2015), but with wider geographic and taxonomic sampling, (2) evaluate mtDNA phylogenetic patterns within Planes/Pa. laevimanus, and (3) quantify and compare the degree of intraspecific genetic variation in mtDNA between clades/species within Planes/Pa. laevimanus and other grapsid species. Mitochondrial sequence data for Pa. laevimanus and Planes were collected specifically for this study

(see below), while mtDNA sequence data from 168 specimens representing 19 other grapsid species were provided by the Florida Museum of Natural History, University of

Florida. These included Geograpsus crinipes (N = 14), Geograpsus grayi (N = 6),

Geograpsus lividus (N = 3), Geograpsus stormi (N = 3), Goniopsis cruentata (N = 6),

Grapsus albolineatus (N = 18), Grapsus fourmanoiri (N = 2), Grapsus grapsus (N = 3),

Grapsus longitarsus (N = 4), Grapsus tenuicrustatus (N = 11), Leptograpsus variegatus

(N = 5), Metopograpsus frontalis (N = 20), Metopograpsus latifrons (N = 3),

Metopograpsus oceanicus (N = 4), Metopograpsus thukuhar (N = 24), Pachygrapsus minutus (N = 8), Pachygrapsus planifrons (N = 12), Pachygrapsus plicatus (N = 14), and

Pachygrapsus transversus (N = 8). Genomic DNA extractions and mtDNA amplification and sequencing of these museum specimens followed the same methodologies that were used for Planes and Pa. laevimanus in this study (see below).

DNA Extraction

Genomic DNA was extracted from either leg muscle tissue or whole/partial leg samples using one of three methods: (1) DNAzol (muscle tissue samples; Molecular

Research Center, Inc. Cincinnati, OH USA), (2) phenol-chloroform-isoamyl alcohol

(whole or partial leg samples), or (3) Zymo Research Genomic DNA Microprep kits

(whole or partial leg samples; Zymo Research Corporation, Irvine, CA USA). For each method, extractions were performed according to the manufacturer’s recommendations.

For (2) and (3), extractions were performed following powderization in a Geno/Grinder

2010 (Spex SamplePrep, Metuchen, NJ USA).

COI Amplification, Sequencing and Analyses

A 650 bp barcoding fragment of the mitochondrial cytochrome c oxidase subunit I

(COI) gene was amplified using degenerate universal Metazoan primers

(forward/reverse: dgLCO/dgHCO) and polymerase chain reaction (PCR) protocol described in Evans and Paulay (2012). Individual 25 µL reactions (including a negative control) contained 10x buffer, 10mM dNTP, 50 mM MgCl2, 10 µM of each primer, 5 units/µL NEB OneTaq Hot Start 2x Master Mix (New England Biolabs, Ipswich, MA

USA), 1 µL DNA, and a sufficient volume of PCR grade H2O to yield a final volume of 25

µL. PCR products were checked on a 1.4% agarose gel to verify the amplification of fragments of appropriate size (500-700 bp). Samples that did not amplify or contained fragments of incorrect size were either re-amplified or excluded from COI sequencing and analyses. Samples containing PCR products of appropriate size were cleaned of unconsumed PCR reagents with ExoSap-IT (Affymetrix, Santa Clara, CA USA) and

Sanger sequenced bidirectionally using BigDye v3.0 chemistry on an AB 3130xl Genetic

Analyzer or AB 3730xl DNA Analyzer (Applied Biosystems, Life Technologies

Corporation, Carlsbad, CA USA).

Forward and reverse reads were assembled using Sequencher v4.10.1 (Gene

Codes Corporation Ann Arbor, MI USA) and manually checked for ambiguous and

erroneous base calls. Resulting high-quality COI sequences for Pa. laevimanus (N = 6),

Pl. marinus (N = 11), Pl. minutus (N = 30) and Pl. major (N = 38) (Table 2-1) were combined with COI sequences from 168 individuals representing 19 other grapsid species (see above), and a multiple sequence alignment was generated using MUSCLE

(Edgar 2004) as implemented in SeaView v 4.5.3 (Gouy et al. 2010). A maximum likelihood (ML) phylogenetic analysis was carried out in RAxML v 8.0.0 (Stamatakis

2014) using the GTRGAMMA model, 1000 bootstrap replicates, and the rapid bootstrap and ML tree search algorithm (option -f a). Within each clade/species supported with at least 60% bootstrap support, nucleotide diversity (π) was estimated in Arlequin v 3.5

(Excoffier and Lischer 2010).

Creation and Sequencing of RAD Libraries

Genomic DNA quality was checked on a 0.7% agarose gel to ensure that the majority of DNA fragments was of high molecular weight. Samples that did not meet this criterion were excluded from RAD library development because degraded DNA has been shown to dramatically reduce the ability to recover comparable loci among individuals (Graham et al. 2015). RAD libraries were prepared following the double- digest (ddRAD) protocol described by Parchman et al. (2012) and Peterson et al.

(2012). Genomic DNA was digested with two restriction enzymes, EcoRI and MseI

(New England Biolabs, Ipswich, MA USA) for 6 hours at 37°C. Individual double- stranded sequence adaptors with unique inline barcodes (8-10 bp with 3 bp difference between any two barcodes) were ligated to the 5’ EcoRI overhang of the digested DNA and a universal 3’ adaptor was ligated to the MseI overhang with T4 DNA ligase (New

England Biolabs, Ipswich, MA USA) for 6 hours at 16°C. PCR was used to incorporate

Illumina flowcell binding sequences and sequencing priming sites to the adaptor ligated

DNA fragments using iProof High-Fidelity DNA polymerase (Bio-Rad Hercules, CA

USA) with 55°C annealing temperatures and 20 cycles. Barcoded PCR products from each individual (6 µL) were pooled, then gel size selected for fragment ranging from

250-450 bp and sequenced on an Illumina HiSeq 2000 1x100 (Illumina, San Diego, CA

USA). Size selection and sequencing were performed by the Interdisciplinary Center for

Biotechnology Research at the University of Florida.

Processing of Sequenced RAD Tags

All data processing and analyses were performed on HiPerGator, the University of Florida’s High Performance Research Computing Cluster. Quality filtering of RAD sequences was performed using FASTQ-quality-filter in the FASTX-Toolkit (Gordon and

Hannon 2010), where reads were evaluated for a Phred quality threshold of 20 on 90% of nucleotides in the read. Using FASTQ-barcode-splitter and FASTQ-trimmer, high- quality reads were demultiplexed and trimmed by removing barcodes and restriction enzyme cut sites, resulting in fragments of 84 bp in length. Sequence alignment, SNP discovery, and genotyping were performed in STACKS v. 1.21 (Catchen et al. 2011,

2013). The initial module in the STACKS pipeline, ustacks, was run separately for each individual allowing a minimum identical read depth of 2 (-m 2), a maximum nucleotide distance for comparing identical stacks (i.e., allele detection) of 2 (-M 2), the –r and –d parameters invoked, and all other parameters set to default. The cstacks module of

STACKS, which performs a comparison of loci found within multiple individuals to establish a reference catalog of all possible loci and associated alleles, was run allowing a maximum of 2 nucleotide mismatches between loci of different individuals (–n 2). To decrease computational demands and reduce the cataloging of tens of thousands of low frequency loci, cstacks was run with a reduced set of individuals (N = 35) with

representatives from all species and all populations, but not every individual sample.

The sstacks module of STACKS was then run to compare every individual against the catalog of reference loci and establish homologous loci among all individuals. In the populations module of STACKS, we retained RAD loci with a minimum stacks depth of three (-m 3) and minor allele frequency of 0.05 (-a 0.05), and SNPs that were genotyped in at least 70% of the individuals (-r 0.7). Other combinations of parameter values were tested for each module, and those described above were found to generate the best compromise between dataset size (i.e., number of loci) and information content.

We first generated a RAD loci dataset for all individuals regardless of putative species or ocean region of sampling (RAD dataset 1). Based on the results of this all- inclusive dataset and the detection of putative hybrid individuals (see below), we then generated two less inclusive datasets to test for additional fine-scale or hierarchical genetic structuring: RAD dataset 2 (all non-hybrid Pa. laevimanus and Pl. marinus) and

RAD dataset 3 (all non-hybrid Pl. minutus and Pl. major). While each of the three RAD loci datasets was assembled using the same parameter settings in each module of

STACKS (ustacks: –M 2 –m 2; cstacks: –N 2 –n 2; populations: -m 2 –a 0.05 –r 0.7), each was tailored to allow for optimal analysis of population structuring at their respective scale. For example, the all-inclusive dataset (RAD dataset 1) will contain some loci that have SNPs that are fixed between groups at a large scale, but at smaller scales, such as within group/species, those loci will be uninformative because they contain no polymorphism. Similarly, there will be loci that are polymorphic within one group/species at smaller scales that may be absent entirely from other groups/species

such that when a minimum number of individuals needed to retain a loci is enforced that locus may be excluded. The evolutionary processes acting at these different scales necessitate the use of different datasets, each tailored to addressing the questions at that scale. RAD loci datasets were exported in various file formats from STACKS for subsequent statistical analyses.

Individual and Population Clustering

We first inferred the number of species and/or population clusters using parametric and non-parametric clustering methods as implemented in the programs

STRUCTURE v. 2.3.4 (Pritchard et al. 2000) and AWCLUST v. 3.0 (Gao and Starmer

2008), respectively. Both programs provide a means of evaluating different values for

K, the number of putative genetic clusters (often species or populations), but AWCLUST is robust to small sample sizes within putative populations (<10 individuals) and violations of demographic assumptions of Hardy-Weinberg and linkage equilibrium (Gao and Starmer 2008; Deejai et al. 2010). STRUCTURE and AWCLUST analyses were first performed on the all-inclusive dataset comprising all putative species and ocean regions (RAD dataset 1), and then separately on the less inclusive datasets constructed based on the results of the first analysis (RAD datasets 2 and 3). In STRUCTURE, we used 10,000 burn-in iterations followed by another 40,000 Markov chain Monte Carlo

(MCMC) steps assuming an admixture model with correlated allele frequencies and including no prior information on sampling location. Five replicates for each value of K were run. Results from STRUCTURE were processed and visualized using the software STRUCTURE HARVESTER v. 0.06.94 (Earl and vonHolt 2012) and different values for K were evaluated by visually comparing the log likelihood probability (L(K); mean +/- standard deviation) of each model and applying the deltaK method of Evanno

et al. (2005). Based on estimated ancestry coefficients calculated in STRUCTURE, each individual was assigned to one putative species or population cluster at each value of K. In AWCLUST, pairwise allele sharing distance (ASD) matrices were generated between all individuals in each dataset and multidimensional scaling plots were constructed to visualize putative clusters and identify outliers. Gap statistics were calculated and compared for each value of K following 100 null simulations, and each individual was assigned to one putative species or population cluster based on hierarchical clustering plots. In both STRUCTURE and AWCLUST, we tested values of

K between 1-10 for RAD dataset 1 and values of K between 1-5 and 1-8 for RAD datasets 2 and 3, respectively. At each value of K, we compared the composition of individuals within clusters to quantify the congruence between STRUCTURE and

AWCLUST assignments, and to identify common and erroneous clusters based on putative species designations and geography.

Population Genomic Analyses

For each RAD dataset, the genetic diversity within clusters and genetic differentiation between clusters detected in STRUCTURE and AWCLUST was estimated by calculating population genetic statistics (pairwise genetic distance [FST], observed and expected heterozygosity, and number of private alleles) in the program

Arlequin. When the optimal value of K was not clear-cut within and/or between

STRUCTURE and AWCLUST analyses, we calculated population genomic statistics within and among clusters at multiple values of K. Significant differences in FST values among clusters were determined by running 1000 permutations in the program Arlequin.

For RAD dataset 3, we performed additional fine-scale analyses based on the results of genetic clustering. A hierarchical analysis of molecular variance (AMOVA;

Excoffier et al. 1992) was performed to test how genetic variation is partitioned within and among ocean basins and regions. Regional designations for the AMOVA follow

Table 2-1 and Figure 2-1, except for the Southeast and Southwest Atlantic (SEA and

SWA) and Southeast and Southwest Indian (SEI and SWI), which were grouped together into South Atlantic (SA) and Indian (IND), respectively, due to sample sizes of one in SWA and SWI. Significant differences in genetic differentiation (FST) were evaluated at an alpha value of 0.0009 (post-hoc Bonferroni correction for 55 comparisons). In addition, we conducted two standard Mantel tests in GENALEX v. 6.5

(Peakall and Smouse 2012) to test for correlations between genetic distance (FST) and log-transformed geographic distance among sites within the North Atlantic and Pacific oceans (i.e., test for isolation-by-distance). Geographic distance matrices (in km) were generated using Universal Transverse Mercator coordinates of each sampling location.

Results

COI Phylogenetic Analysis and Morphology

A maximum likelihood phylogenetic analysis of 253 COI sequences from 23 grapsid species (including Pa. laevimanus and three putative Planes species) resulted in consistently high bootstrap support (>87%) for the monophyly of most species (Figure

2-2). In this analysis, Pa. laevimanus and Planes form a single clade that is distinct from other grapsid species with high bootstrap support (98%), which is consistent with the paraphyly of Planes due to the well-supported inclusion of Pa. laevimanus as found by Schubart (2011) and Ip et al. (2015). Within the family as a whole, we found little to no phylogenetic structure among genera or species after collapsing nodes with weak bootstrap support (<60%).

Within the clade that unites Pa. laevimanus and Planes, we found one strongly supported polytomy (bootstrap = 87%) nested within a larger polytomy (Figure 2-2).

Individuals within the nested clade (Group 2) shared seven mtDNA SNPs that were not found within individuals outside the nested clade (Group 1). There was no statistical support for sub-structuring with respect to putative species designations or geography within either group. Group 1 comprised mostly individuals of Pa. laevimanus (N = 6) and Pl. marinus (N = 10), but also seven Pl. minutus individuals from the Northeast

Atlantic (NEA). Group 2 comprised mostly individuals of Pl. minutus (N = 23) and Pl. major (N = 38), but also one Pl. marinus individual from the Northwest Atlantic (NWA).

The nucleotide diversity within each Planes/Pa. laevimanus group (π = 0.009) and within the entire Planes/Pa. laevimanus clade (π = 0.044) was similar to or less than that of other grapsid species (π = 0.0-0.074) at this mtDNA locus (Figure 2-2).

Subsequent RADseq analyses (see below) revealed the presence of genetic hybrids between Group 1 and Group 2, the two groups detected within the COI phylogenetic analysis. These individuals are identified in Figure 2-2 as “RAD hybrids.”

Within Group 1, all Pl. marinus from the NWA (N = 2) and Pl. minutus from the NEA (N

= 7) were identified as hybrids in the RADseq analysis. Within Group 2, the single Pl. marinus from the NWA was identified as a hybrid and hybrids were identified among Pl. minutus in the NWA (1 out of 11 sequenced individuals) and NEA (2 out of 10 sequenced individuals). Two RAD hybrids from NEA were not sequenced for COI.

Excluding hybrids from the COI phylogeny, Group 1 comprised only Pa. laevimanus and

Pl. marinus, and Group 2 comprised only Pl. minutus and Pl. major.

In terms of overall morphology, Pachygrapsus laevimanus individuals from both

Australia and Rapa Island were clearly different from Pl. marinus in having a more quadrate carapace (i.e., not laterally convex) that is distinctly wider than long

(length:width, 1:>1.2) and walking limbs that lack natatory fringes. Planes marinus and

Planes minutus/major were differentiated by subtle differences in carapace dimensions

(length:width, 1:>1.1 versus 1:<1.1, respectively), shape (quadrate versus round, respectively) and striated-ness (striated versus not striated, respectively). These differences were more clear and distinctive in the Pacific and Indian oceans (i.e., between Pl. marinus and Pl. major sensu stricto), but were less distinctive in the North

Atlantic. The carapaces of Planes minutus/major in the North Atlantic were slightly less convex laterally than elsewhere, and the dimensions and shape approached that of Pl. marinus. However, whether or not the carapace was striated remained distinctive between Pl. marinus and Pl. minutus/Pl. major. Individuals identified as “RAD hybrids” were morphologically intermediate between Pl. marinus and Pl. minutus/Pl. major in the

North Atlantic: carapace only slightly wider than long (length:width, 1:1.0-1.1) and lightly striated in some individuals.

RAD Libraries and Processing

We sequenced RAD fragments for 152 barcoded individuals in two lanes on the

Illumina HiSeq 2000 platform, generating 297 million raw reads. After quality control steps, we retained 215 million reads from 145 individuals: 6 Pa. laevimanus, 11 Pl. marinus, 52 Pl. minutus and 76 Pl. major (Table 2-1). The mean number of quality filtered reads per individual was 903,186 and of these an average of 718,070 were utilized by STACKS for loci discovery and allele calling, resulting in a average of

114,409 loci per individual with an average read depth of 6.0 (Table 2-2). We

assembled RAD datasets de novo first for the full dataset comprising all putative species and ocean regions (RAD dataset 1; N = 145) and then separately for two distinct subsets of individuals detected in STRUCTURE and AWCLUST in the full analysis. After the exclusion of putative hybrid individuals identified in the full analysis

(see below), the two subsets comprised (1) only Pa. laevimanus and Pl. marinus individuals (RAD dataset 2; N = 14) and (2) only Pl. minutus and Pl. major individuals

(RAD dataset 3; N = 116).

For the three RAD datasets, we recovered the following numbers of loci: RAD dataset 1 = 1108 loci; RAD dataset 2 = 3314 loci; RAD dataset 3 = 1288 loci. Datasets contained different sets of loci because we retained only those loci that there were present in at least 70% of the individuals in each dataset. Overlap of loci between the three datasets was compared to determine the number of unique loci that were contributing to the analysis at each scale/species group (Figure 2-3). Moreover, we plotted the distributions of haplotype divergence (ΦST) to confirm that the composition of loci in each dataset contained variation at the desired evolutionary scale. Because the probability of recovering shared loci declines as the divergence between the groups being compared increases, datasets with more slowly evolving/mutating loci (high ΦST values) capture deeper divergences and datasets with more rapidly evolving/mutating loci (low ΦST values) capture more recent patterns. RAD dataset 1 contained a high proportion of loci with high ΦST values (Figure 2-4A), suggesting that this dataset was appropriate for evaluating species-level relationships, but less appropriate for identifying population-level structure. RAD datasets 2 and 3 contained a substantially higher proportion of loci with low ΦST values (Figures 2-4B, 2-4C), suggesting that these

datasets were appropriate for assessing very recent divergences and population-level patterns.

Clustering of Individuals and Populations

The optimal values for K, the number of putative species or population clusters, in each dataset was not clear-cut within and between STRUCTURE and AWCLUST analyses. Therefore, instead of selecting and analyzing just one seemingly optimal value of K, we analyzed the results from a range of possible K values within each dataset taking into account statistical support, as well as the consistency of clustering assignments among individuals and the formation of geographically erroneous clusters.

For RAD dataset 1, which comprised all putative species and ocean regions, we found support for K=2 in STRUCTURE (Figures 2-5A, 2-5B) and K=4 in AWCLUST

(Figure 2-5C). We also found support for K=6 in AWCLUST, but there was weak support and high variation in L(K) for K=6 in STRUCTURE, as well as the formation of an erroneous sixth cluster in both analyses (i.e., the cluster was composed of a small number of seemingly random individuals from many geographic locations). At K=2, there was high congruence (≥95%) between STRUCTURE and AWCLUST assignments (i.e., the same individuals grouped in similar clusters in both analyses) and most individuals segregated into two putative species clusters: one comprised of mostly

Pa. laevimanus/Pl. marinus and another comprised of mostly Pl. minutus/Pl. major

(Figure 2-6A). However, there were 15 individuals – all from either the NEA or NWA – that showed significant admixture between the two putative species clusters (ancestry coefficients = 5-95%; Figure 2-6A), suggesting that the genomic composition of these individuals may be the result of recent hybridization. At K=3 (not shown) and K=4

(Figure 2-6B), which both showed high congruence (≥95%) between STRUCTURE and

AWCLUST assignments, these 15 putative hybrid individuals formed a distinct cluster.

Recent hybridization between the two putative species clusters identified in this analysis is corroborated by mtDNA data. Thirteen of the 15 putative RAD hybrids are shown in the COI phylogenetic analysis (Figure 2-2): nine carried the Pa. laevimanus/Pl. marinus or Group 1 mitochondrial genome, four carried the Pl. minutus/Pl. major or Group 2 mitochondrial genome, and two were not sequenced for COI. At K=4 (Figure 2-6B), the

Pl. minutus/Pl. major cluster showed further segregation (with considerable admixture) corresponding primarily to different ocean basins and a priori species designations (see

RAD dataset 3 analysis below).

For RAD dataset 2, which comprised non-hybrid Pa. laevimanus (N = 6) and Pl. marinus (N = 8), we found support for K=2 and K=3 in both STRUCTURE (Figures 2-

5D, 2-5E) and AWCLUST (Figure 2-5F) and 100% congruence between STRUCTURE and AWCLUST assignments. At K=2 (Figure 2-7A), intertidal Pa. laevimanus clearly segregated from rafting Pl. marinus with almost no admixture. At K=3 (Figure 2-7B),

Pa. laevimanus remained distinct, while Pl. marinus segregated geographically into

Indian (N = 3) and Pacific Ocean (N = 5) clusters with some admixture.

For RAD dataset 3, which comprised non-hybrid Pl. minutus (N = 39) and Pl. major (N = 77), we found support for K=2 and K=4 in STRUCTURE (Figures 2-5G, 2-

5H) and AWCLUST (Figure 2-5I) and high congruence (>95%) between STRUCTURE and AWCLUST assignments at all values of K. We also found support for K=6 in

STRUCTURE, but there was little support for K=6 in AWCLUST, as well as the formation of an erroneous sixth cluster in both analyses. At K=2 (Figure 2-8A), most individuals segregated into two geographic clusters with some admixture: one

comprised mostly of individuals from the North Atlantic (including the Mediterranean

Sea) and another comprised mostly of individuals from the Pacific, with no consistent segregation of individuals from the South Atlantic and Indian oceans into either cluster

(ancestry coefficients = 40-60%). At K=4 (Figure 2-8B), individuals segregated roughly into four geographic clusters: (1) North Atlantic and Mediterranean Sea, (2) South

Atlantic and Indian oceans, (3) East and Central Pacific and (4) West Pacific. There was considerable overlap and admixture between the two Pacific clusters, with individuals from both clusters being found in each of the three regions in the North

Pacific (Figure 2-8B).

For RAD dataset 3, we tested the consistency of the data to yield support for the observed biogeographic pattern by arranging the loci from greatest to least variable then binning the loci into separate data files with each subsequent file containing less variable loci. The primary analyses were then rerun, allowing us to evaluate the proportion of loci that are supporting the same patterns as observed from the full data sets. Across all data sets, from highly variable to less variable, we consistently found support for K=4 and the aforementioned biogeographic clusters. Moreover, we tested for additional fine-scale or hierarchical genetic clustering within the North Atlantic Ocean

(RAD dataset 3, Cluster 1) and Pacific Ocean (RAD dataset 3, Clusters 3 and 4) using

STRUCTURE and AWCLUST, but found no significant support for any additional sub- structuring in either analysis.

Population Genomic Analyses

For each RAD dataset, all pairwise comparisons of genetic differentiation (FST) between clusters identified in STRUCTURE and AWCLUST were highly significant (P- value < 0.001). In RAD dataset 1 at K=2, the cluster comprising Pa. laevimanus, Pl.

marinus and three putative hybrids was differentiated from the cluster comprising Pl. minutus, Pl. major and 12 putative hybrids (FST = 0.683) (Table 2-3A). At K=3 (not shown in Figure 2-4), the cluster comprising 15 putative hybrids was differentiated from the cluster comprising only Pa. laevimanus and Pl. marinus and the cluster comprising only Pl. minutus and Pl. major, but to a greater degree in the latter (FST = 0.244 vs.

0.400) (Table 2-3B). Also at K=3, the two non-hybrid clusters were strongly differentiated from each other (FST = 0.727) (Table 2-3B). At K=4, again, the cluster comprising 15 putative hybrids was differentiated from the cluster comprising only Pa. laevimanus and Pl. marinus (FST = 0.244) and both clusters comprising Pl. minutus and

Pl. major (FST = 0.358 and 0.419) (Table 2-3C). Among the three non-hybrid clusters, the cluster comprising only Pa. laevimanus and Pl. marinus was strongly differentiated from both clusters comprising Pl. minutus and Pl. major (FST = 0.728 and 0.763), which in turn were only weakly differentiated from each other (FST = 0.099) (Table 2-3C).

Collectively, the results from RAD dataset 1 suggest the presence of two species groups and a zone of hybridization between the two in the North Atlantic, as well as additional ocean-specific differentiation within the Pl. minutus/Pl. major species group that was also consistent with a priori species designations.

In RAD dataset 2 (all hybrids excluded) at K=2, the cluster comprising intertidal

Pa. laevimanus was differentiated from the cluster comprising rafting Pl. marinus (FST =

0.261) (Table 2-4A). At K=3, the cluster comprising Pa. laevimanus was differentiated from the cluster comprising Pl. marinus from the North Pacific and the cluster comprising Pl. marinus from the Indian Ocean, but to a greater degree in the latter pairwise comparison (FST = 0.272 vs. 0.387) (Table 2-4B). The two clusters comprising

Pl. marinus from the North Pacific and Indian oceans were less differentiated from each other (FST = 0.215) than either was from the cluster comprising Pa. laevimanus (FST =

0.272 and 0.387, respectively) (Table 2-4B). Collectively, the results from RAD dataset

2 suggest that intertidal Pa. laevimanus are distinct from rafting Pl. marinus, which differentiate further by ocean basin (Pacific versus Indian).

In RAD dataset 3 (all hybrids excluded) at K=2, the cluster comprising mostly individuals from the North Atlantic (including individuals from the Mediterranean Sea, and some individuals from the South Atlantic and Indian oceans) was differentiated from the cluster comprising individuals from the South Atlantic, Indian and Pacific oceans

(FST = 0.122) (Table 2-5A). At K=3, the cluster comprising individuals from the South

Atlantic and Indian oceans (including one individual from the NWA) was approximately equally differentiated from clusters comprising individuals from the North Atlantic and

Pacific oceans (FST = 0.086 and 0.079, respectively), which were more differentiated from each other than either was from the South Atlantic/Indian cluster (FST = 0.140)

(Table 2-5B). At K=4, the cluster comprising individuals from the South Atlantic and

Indian oceans was approximately equally differentiated from the cluster comprised of individuals from the North Atlantic (FST = 0.086) and the two clusters of individuals from the Pacific (FST = 0.088 and 0.080) (Table 2-5C). Also at K=4, the cluster comprised of individuals from the North Atlantic was approximately equally differentiated from the two clusters of individuals from the Pacific (roughly East/Central and West Pacific; FST =

0.156 and 0.123, respectively), which in turn were only weakly differentiated from each other (FST = 0.038) (Table 2-5C). This analysis from RAD dataset 3 shows an overall pattern of relatively low, but consistent, genetic differentiation (FST = 0.038-0.156)

between major ocean regions with evidence for subtle differentiation between Pl. minutus (cluster 1) and Pl. major that was comparable to additional differentiation within

Pl. major (clusters 2-4).

Additionally, for RAD dataset 3 an analysis of molecular variance (AMOVA) across 11 regions in three ocean basins showed that the majority of genetic variation was found among individuals within regions (87%) and that there was considerably more genetic variation between oceans (11%) than among regions within oceans (2%)

(Table 2-6). Levels of genetic differentiation (FST) among the regions designated in the

AMOVA (Table 2-7; Figure 2-9) were generally consistent with the patterns found in the clustering analyses (with accompanying inter-cluster FST values) (Table 2-5). Pairwise comparisons with significant differences at a Bonferroni-corrected alpha value of 0.0009 mostly correspond to the extent of genetic differentiation, except when sample sizes from both regions were less than three (Figure 2-9). In general, individuals from the three North Atlantic regions (NWA, NEA, MED) showed low differentiation from each other (FST = -0.03-0.05), but showed high differentiation when compared to individuals from the Pacific regions (NWP, SWP, NCP, SCP, NEP, SEP) (FST = 0.09-0.23).

Individuals from the South Atlantic (SA) and Indian (IND) regions showed low differentiation from each other (FST = -0.037), but showed low to moderately high differentiation when compared to the North Atlantic (FST = -0.03-0.15) and moderately low to moderately high differentiation when compared to the Pacific regions (FST = 0.05-

0.15). Among the Pacific regions, individuals in the Central and East Pacific (NCP,

SCP, NEP, SEP) regions showed low to moderately low differentiation from each other

(FST = -0.03-0.09), but low to moderately high differentiation when compared to the

West Pacific regions (NWP, SWP) (FST = -0.03-0.15). Lastly, we found no correlation between genetic and geographic distance within either the North Atlantic Ocean (Mantel test: y = -0.0112x + 0.03; r2 = 0.334; P-value = 0.17) or Pacific Ocean (Mantel test: y =

0.014x - 0.04; r2 = 0.008; P-value = 0.21). Collectively, the results of RAD dataset 3 suggest that Pl. minutus and Pl. major likely actually a single, globally distributed species that shows some geographic structure with weak genetic differentiation among widely separated aggregations and no additional species diversity.

Discussion

In this study, we used mitochondrial and genome-wide SNP data to evaluate the species-level relationships among rafting Planes and intertidal Pachygrapsus laevimanus, and assess the degree of population differentiation among widely separated populations worldwide. In theory, the ability of Planes to disperse, not only as pelagic larvae, but also as adults and juveniles associated with oceanic flotsam and sea turtles should vastly increase their dispersal potential and facilitate transoceanic, if not near global, genetic exchange. Therefore, we predicted that there would be low species diversity within this group and weak differentiation among widely separated populations. The results of this study provide unprecedented resolution into the systematics and global biogeography of this group and contribute valuable information to our understanding of how theoretical dispersal potential relates to actual population differentiation and diversification among marine organisms.

Species-Level Relationships – Is Species Diversity Low?

Our results suggest that species diversity in this group is relatively low. We did not find evidence that there are any additional undescribed or cryptic species within either our mtDNA or genomic RADseq analyses. In fact, our results suggest that

species diversity within this clade may actually be lower than previously thought based solely on morphology (Ng et al. 2008).

In our mtDNA analysis, we found strong support for the reciprocal monophyly of this group (bootstrap support = 98%), which confirms the paraphyly of the genus Planes due to the well-supported inclusion of Pa. laevimanus (as suggested by Schubart 2011 and Ip et al. 2015), only with greater taxonomic and geographic sampling. While we found statistical support for the presence of two closely related species groups (Pa. laevimanus/Pl. marinus and Pl. minutus/Pl. major), we found no statistical support for the phylogenetic separation of Pa. laevimanus from Pl. marinus or Pl. minutus from Pl. major in our COI data (Figure 2-2). Within both species groups, the nucleotide diversity in COI was very low (π = 0.009) and was not consistent with any phylogenetic differences among a priori species designations. Moreover, the COI nucleotide diversity within the clade as a whole was low (π = 0.04) and comparable to that of other grapsid species (π = 0.0-0.074). While these mtDNA results draw into question that presence of even two species, it is clear that these data provide insufficient resolution for assessing the species-level relationships within this group.

Results from our RADseq analyses, along with inferences from morphological evidence, provided a more complete picture. In RAD dataset 1 (Figure 2-6; Table 2-3B,

2-3C), the large proportion of loci with high haplotype divergence (Figure 2-4) suggests that this dataset captured genetic resolution at a deeper evolutionary scale that is more indicative of species-level patterns. We found strong statistical support for the distinction between the clusters comprised of only Pa. laevimanus and Pl. marinus individuals and only Pl. minutus and Pl. major individuals (FST = 0.727) after hybrid

individuals were removed (see below). We also found some statistical support for the distinction between Pl. minutus and Pl. major at K=4 in RAD dataset 1. However, there was considerable admixture and weak genetic differentiation between these putative species in RAD dataset 1 (FST = 0.099) and results from RAD dataset 3 indicate that Pl. minutus in the North Atlantic is not more differentiated from any Pl. major cluster than

Pl. major clusters were from each other (see below). Moreover, the lack of diagnostic morphological differences between these two putative species (Chace 1951) is consistent with these genetic patterns. Although we did not find support for the distinction between Pa. laevimanus and Pl. marinus in RAD dataset 1, we found strong support for this distinction in RAD dataset 2 when a more exclusive set of RAD loci were analyzed (FST = 0.261 at K=2; FST = 0.272 and 0.387 at K=3), which was consistent with distinct morphological differences between these two putative species (Chace 1951;

Poupin et al. 2005). Collectively, based on our new RADseq data, as well as previous morphological evidence, our interpretations are that (1) Pa. laevimanus and Pl. marinus are two closely related species, (2) Pl. minutus and Pl. major are one species with some population-level genetic variation among widely separated geographic locations (see below), and (3) Pl. marinus and Pl. minutus/Pl. major have a zone of secondary genetic exchange (i.e., hybridization) in the North Atlantic Ocean.

Despite plausible theoretical expectations for the effect that high dispersal potential should have on diversification (Palumbi 1994, 2003; Avise 2000), patterns of diversification among planktonic and neustonic organisms are quite variable. Our results for Planes are consistent with theoretical predictions, in which the combination of long-distance dispersal by pelagic larvae and potentially worldwide dispersal of rafting

adults and juveniles appears to have limited diversification within the group. Other surface- and subsurface-dwelling oceanic animals show patterns that are both consistent and contradictory to theoretical expectations. Two independent lineages of sea skaters (genus Halobates; Insecta) have subsequently speciated following their colonization of the open ocean, although species diversity has remained fairly low

(Anderson et al. 2000; Damgaard et al. 2000). The amphipod Caprella andreae, which, like Planes, is an obligate associate of surface-drifting oceanic flotsam and sea turtles, shows high diversity and cryptic speciation across a relatively small geographic area compared to Planes (Cabezas et al. 2013). Oceanic nudibranchs of the genus Glaucus, which float upside-down just below the surface (i.e., hyponeustonic), display different diversification patterns between sister species: Glaucus atlanticus is cosmopolitan and shows no evidence for cryptic diversification, while Glaucus marginatus is restricted to the Indo-Pacific and has diversified into four distinct lineages (Churchill et al. 2013).

Interestingly, in the latter, genetic diversification has occurred with a parallel divergence in reproductive morphology, providing a mechanism for reproductive isolation (Churchill et al. 2013). Lastly, cosmopolitan oceanic copepods (e.g., Pleuromamma abdominalis) tend to show extensive cryptic diversity and high rates of endemism (Goetze 2003; Hirai et al. 2015). While examples of diversification patterns among planktonic and neustonic animals are relatively few compared to neritic taxa, there appears to be no ubiquitous pattern for their diversification and only some patterns are consistent with theoretical expectations based on dispersal potential. It is clear that while the capacity for long- distance dispersal likely plays an important role in limiting opportunities for local adaptation and diversification (as well as extinction), the mechanisms leading to

speciation in the open ocean are far more complex and might also involve behavioral changes associated with selection of different temperature and salinity profiles in the pelagic environment (Palumbi 1994, 2003; Knowlton 2000).

Our all-inclusive SNP dataset (RAD dataset 1), as well as mtDNA sequence data, supported the species-level distinction between Pa. laevimanus/Pl. marinus and

Pl. minutus/Pl. major (see above), but no further support for the separation of Pa. laevimanus from Pl. marinus. However, when we tested for finer-scale differentiation by tailoring the SNP datasets to optimize the genetic resolution within each species group, we found strong statistical support for differentiation between Pa. laevimanus and Pl. marinus in RAD dataset 2 (Table 2-4; Figure 2-5). From a logistical standpoint, this indicates that the polymorphic loci that differentiate Pa. laevimanus from Pl. marinus were excluded in the all-inclusive dataset; likely when we enforced a minimum percentage of individuals needed to retain a locus (-r in the populations module of

STACKS). From a biological standpoint, this suggests that intertidal Pa. laevimanus and rafting Pl. marinus are different, but this separation has yet to lead to any discernable differentiation in the mitochondrial genome or at least at the COI locus - a fast-evolving locus often used as a species-level barcode (Evans and Paulay 2012).

Our interpretation of these results is that there has been a recent and rapid speciation event separating intertidal Pa. laevimanus from rafting Pl. marinus, and likely concomitant selection for traits associated with different habitats: wider carapace and the lack of natatory fringes in Pa. laevimanus (similar to other intertidal Pachygrapsus species) and narrower carapace and natatory fringes in Pl. marinus (similar to other

Planes species). That said, this current dataset is small and we lack geographic

overlap among our samples from the two putative species (i.e., Pl. marinus from the southwest Pacific). Clearly, a more complete sampling regime is needed to evaluate the species-level relationships between Pa. laevimanus and Pl. marinus. Greater genetic resolution between these putative species will likely evoke many interesting ecological questions in regards to larval recruitment, morphological adaptation and habitat-specific differences in reproductive and social behavior.

Our all-inclusive SNP dataset (RAD dataset 1), we also found some statistical support for the distinction between Pl. minutus and Pl. major at K=4. However, there was considerable admixture and weak genetic differentiation between these putative species clusters (FST = 0.099) and results from RAD dataset 3 indicate Pl. minutus in the North Atlantic is not more differentiated from any Pl. major cluster than Pl. major clusters were from each other (see below). These results suggest that Pl. minutus and

Pl. major are likely a single, globally distributed species with some degree of population structure among widely separated geographic locations (see below). The highly subtle morphological differences between Pl. minutus and Pl. major described by Chace

(1951) and applied by subsequent authors (e.g., Manning and Holthuis 1981; Prado and de Melo 2002; Pons et al. 2011; this study) may simply be related to subtle geographic variation in body size and concomitant allometric changes in traits related to limb length

(as in Chace 1951), or regional variation or phenotypic plasticity in traits related to masticatory structures (as in Frick et al. 2011). Alternatively, morphological differences detected in North Atlantic individuals in the past may result from the inclusion of hybrid individuals during morphological comparisons. Many of the traits that were thought to differentiate Planes species tend to place Pl. minutus intermediate between Pl. marinus

and Pl. major, suggesting that genetic exchange between Pl. marinus and Pl. minutus in the North Atlantic might also lead to subtle morphological changes that have confused the taxonomy. Nevertheless, more work is needed to thoroughly quantify the amount of morphological variation in this cosmopolitan species group and evaluate potential factors related to subtle geographic differences in morphology (e.g., population genetics, geography, substrata/host, diet).

In addition to the confusion that hybrid individuals may have added to morphological distinctions in the past, the detection of hybrid individuals between Pl. marinus and Pl. minutus/Pl. major in the North Atlantic is interesting for two other reasons. First, Pl. marinus is not known to occur in the North Atlantic (Chace 1951).

Although we did not detect any non-hybrid individuals that were morphologically characterized as Pl. marinus (i.e., pure Pl. marinus), hybridization would necessitate the occurrence of pure Pl. marinus somewhere in this region or occasional dispersal events from its known range in the South Atlantic (Spivak and Bas 1999; Chace 1966).

Second, the occurrence of hybridization only in one region, the North Atlantic, invokes questions regarding reproductive isolating mechanisms across the rest of the sympatric range of Pl. marinus and Pl. minutus/Pl. major, which includes the rest of the temperate and tropical oceans of the world. Of particular interest are those instances where Pl. marinus and Pl. minutus/Pl. major share the same raft (Pfaller unpublished data) or the same turtle (Frick et al. 2011). Future studies should focus on identifying morphological, physiological and behavioral traits that both facilitate hybridization in the

North Atlantic and deter hybridization elsewhere.

The well-supported inclusion of Pa. laevimanus from both Australia and Rapa

Island within the Planes clade in our COI phylogenetic analysis confirms the paraphyly of Planes as found by Schubart (2011) and Ip et al. (2015), and necessitates taxonomic revision. Clearly, Pa. laevimanus is actually Planes laevimanus. Morphological similarity between Planes, especially Pl. marinus, and Pachygrapsus has led to taxonomic confusion in the past (Chace 1951, 1966). However, among the 14 currently described species of Pachygrapsus, Pa. laevimanus has never been linked to Planes until genetic data were analyzed (Schubart 2011; Ip et al. 2015; this study). That said, a comparison of the highly variable male gonopod morphologies among Pachygrapsus

(Figure 15 in Poupin et al. 2005) and the rather conserved gonopod morphologies found within Planes (Figure 2 in Chace 1951) clearly shows the affinity between Pa. laevimanus and Planes. This similarity has not been recognized in the literature. To date, the use of external morphology and traditional genotypic markers (Cuesta et al.

1997, 2011; Cuesta and Schubart 1999; Guerao et al. 2001; Schubart et al. 2002, 2005,

2006; Schubart 2011; Ip et al. 2015; this study) have failed to produce a resolved phylogenetic hypothesis for the family Grapsidae. Consequently, many taxonomic and evolutionary questions remain unresolved. Future morphological analyses should include male gonopod characteristics. However, robust phylogenetic resolution within this difficult family will likely require the application of high throughput sequencing technologies, including phylogenetic applications of RADseq (e.g., Jones et al. 2013;

Wagner et al. 2013), for future phylogenetic analyses.

Despite the lack of phylogenetic resolution within Grapsidae, the presence of obligate rafting species (i.e., Planes spp.) nested within a family composed entirely of

species that are either supratidal or intertidal (Anger 1995) elicits questions regarding the origin and evolution of rafting. How has the lineage leading to Planes transitioned from an intertidal to rafting lifestyle? What morphological adaptations are associated with this transition? Future studies that focus on the ecology and evolution of Pa. laevimanus and Pl. marinus may be particularly informative, as these putative species appear to have made this transition recently and rapidly. The morphological differences between the two may be related to favorable traits associated with the occupation of different habitats (e.g., natatory fringes in rafting Pl. marinus). However, we have evidence that the evolution of rafting in Planes may be a paedomorphic behavior.

Megalopal-stage crab larvae collected from flotsam and surface-drifting sea snakes

(Hydrophis [Pelamis] platura) off the Pacific coast of Costa Rica (Pfaller et al. 2012) were later identified using genetic barcodes to represent at least three intertidal grapsid species (Pachygrapsus socius, Grapsus grapsus and Goniopsis sp.; Pfaller unpublished data). Despite the diversity of intertidal and benthic crabs in the region (Vargas and

Wehrtmann 2009), graspid megalopae were found far more frequently than any other family in this pelagic associations – small numbers of Portunidae and Plagusiidae were found as well (Pfaller et al. 2012). These data, along with other observations of rafting by grapsid megalopae (e.g., Donlan and Nelson 2003), suggest that rafting on flotsam may be an adaptation associated with recruitment to adult habitats in this family (i.e., larvae cling to floating debris that then floats into adult habitats). The presence of larval-rafting behavior among intertidal grapsids provides a plausible mechanism for the evolution of an obligate rafting species from an intertidal ancestor.

Population Structure – Is Genetic Differentiation Weak?

Our results suggest that genetic differentiation among widely separated geographic location is weak in this group. While we did not find statistical support for global panmixis, global patterns of population structure are primarily at the level of major ocean basins and gyres, and genetic indices indicate recent and/or ongoing gene flow throughout the temperate and tropical oceans of the world. Our results also show the locations of important corridors and barriers to rafting dispersal at a global scale.

In our mtDNA data, we found no evidence of genetic differentiation among geographic regions within either species group (Figure 2-2). In fact, what little COI nucleotide diversity did exist within each species group (π = 0.009) was random, showing no tendency towards geographic structure. These results demonstrate the potential limitations of traditional genotypic markers (e.g., target gene sequence data) for estimating contemporary population structure, especially in organisms with large, highly connected populations. Because estimating population structure among small, isolated populations is more clear-cut and less prone to error, the literature is somewhat biased towards positive examples (Goetze 2005; McCormack et al. 2013; Benestan et al. 2015), in which traditional genotypic markers have provided sufficient genetic resolution to elucidate patterns in population structure. However, exemplified by our

COI data, traditional genotypic markers may not contain sufficient genetic resolution to evaluate population-level hypotheses. Consequently, estimating population structure is more difficult and requires greater genetic resolution to avoid erroneous conclusions

(e.g., false global or ocean panmixis). Past studies that detected little to no genetic differentiation among large populations of widely dispersed neustronic and planktonic animals [copepods (Bucklin and Kocher 1996; Bucklin et al. 1996, 2000), euphausiids

(e.g., Zane et al. 1998; Zane and Patarnello 2000; Jarman et al. 2002), squid (Sands et al. 2003) and nudibranchs (Churchill et al. 2014)] may have failed to detect finer-scale population structure because the genetic markers available at the time contained insufficient genetic variation to assess differences. For this reason, our understanding of the population biology and connectivity in these animals is not only limited, but might also be somewhat biased. Recent advances in high-throughput sequencing technologies have allowed genome-wide genetic variation to be incorporated into population genetic analyses of non-model organisms (Reitzel et al. 2013). Our study is the first to use this technology in a planktonic or neustonic organism and demonstrates the value and future promise of such tools for estimating very subtle, but significant, population structure in organisms with large, highly connected populations.

Evaluating population-level differentiation among globally distributed aggregations of Pl. minutus and Pl. major was the primary impetus for this study. By tailoring our SNP dataset to optimize the genetic resolution within this species group

(RAD dataset 3), we were able to test for finer differentiation at the appropriate intraspecific scale. We found evidence of population differentiation – but not necessarily strict isolation – primarily at the scale of major ocean basins and almost no evidence for differentiation within distinct ocean gyres (Tables 2-5, 2-6, 2-7; Figures 2-6,

2-7). Similar patterns were found for Pl. marinus. Although we did not find support for global panmixis, pairwise genetic distances among clusters (in STRUCTURE and

AWCLUST) and regions (in AMOVA) were consistently very low (FST = -0.03-0.23). At this global scale, such weak differentiation suggests that Planes populations behave somewhat similar to ubiquitous microbial populations, in which individuals are so

abundant that continuous large-scale dispersal sustains their global distribution and inhibits biogeographic structure (Finlay 2002). While Planes populations do show genetic discontinuities based on geographic boundaries (see below), further analyses testing the likelihood of various biogeographic hypotheses (e.g., global, ocean and gyre panmixis) are needed to evaluate support for different patterns.

The nonpolar distribution of Planes likely reflects its inability to survive cold temperatures (Chace 1951; Spivak and Bas 1999), thereby limiting dispersal across regions below their thermal minimum (e.g., the Arctic and Southern oceans). Clustering patterns and pairwise genetic distances (FST) show clear, albeit weak, differentiation between individuals in the Atlantic and Pacific oceans, indicating that the polar waters at

Cape Horn (southern South America) can indeed limit dispersal. However, our results indicate that the cold, but not polar, waters around Cape of Good Hope (southern

Africa) do not prevent dispersal. Instead, the junction of the Angulas and Benguela currents at the Cape of Good Hope appears to act as somewhat of dispersal corridor, as individuals in the South Atlantic and Indian oceans form a distinct cluster with little to no genetic differentiation between oceans (i.e., low FST). These patterns are only partially consistent with genetic patterns of other neustonic organisms: Glaucus nudibranchs and Halobates sea skaters show restricted dispersal across both Cape

Horn (Atlantic-Pacific disjunction) and the Cape of Good Hope (Atlantic-Indian disjunction) (Anderson et al. 2000; Churchill et al. 2014). Planes may simply have a lower thermal tolerance, thereby allowing dispersal between the Atlantic and Indian oceans. However, Planes may also be able to successfully navigate the currents in the retroflection region at the Cape of Good Hope while associated with sea turtles. Both

loggerhead and green turtles show genetic connectivity across this biogeographic boundary (Bourjea et al. 2007; Shamblin et al. 2014), providing a potential dispersal vector for Planes that is unavailable to other neustonic animals that are not known to associate with sea turtles (e.g., Glaucus nudibranchs and Halobates sea skaters). That said, in other areas of the world, turtle-mediated dispersal would be difficult to evaluate relative to rafting dispersal because turtle migratory routes are mostly restricted to large ocean currents, in which case any turtle-mediated genetic signal would be confounded by the raft-mediated genetic signal.

Other genetic discontinuities among globally distributed aggregations of Pl. minutus/Pl. major were weak, but highly significant, and were not always associated with physical barriers to dispersal (i.e., continental landmasses). Subtle differentiation between individuals in the Indian Ocean from those in the Pacific Ocean indicates that the Indonesian Archipelago represents a weak dispersal barrier. The absence of major ocean currents passing through the archipelago and the presence of thousands of islands likely limits the frequency and success of potential dispersal events across this boundary. The Indonesian Archipelago appears to be a strong dispersal barrier structuring populations of Halobates micans (Anderson et al. 2000), but does not result in any detectable genetic differentiation in populations of Glaucus atlanticus (Churchill et al. 2014). We also found evidence for weak genetic differentiation between the North and South Atlantic gyres, and very weak, but significant, differentiation between two clusters of individuals in the Pacific Ocean. In the absence of prominent physical barriers, evidence for genetic discontinuity within a species becomes difficult to explain

(Lowe and Allendorf 2010). In the Atlantic, the prevailing currents at the gyre

boundaries may have a tendency to deflect flotsam back into their respective gyres, therefore reducing the frequency of dispersal by Planes between these gyres. In the

Pacific Ocean, the two genetic clusters correspond loosely to West and East/Central groupings, although there is considerable geographic overlap between the two genetic clusters across the North Pacific. However, we did not find evidence for isolation-by- distance among samples in the Pacific Ocean, thus the mechanisms underlying this genetic discontinuity remain enigmatic.

Our results indicate that the prevailing currents along major ocean gyres tend to homogenize transoceanic aggregations of Planes crabs. Intertidal grapsid crabs that rely exclusively on multi-staged pelagic larvae for long-distance dispersal tend to show little to no transoceanic connectivity (Schubart et al. 2005; Cassone and Boulding

2006), indicating that large distances across ocean gyres represent significant barriers to pelagic larval dispersal. Therefore, our results are consistent with the prediction that pelagic larval dispersal is augmented considerably by dispersal of adult Planes associated with oceanic flotsam and sea turtles, such that wide ocean gyres do not limit connectivity between Planes aggregations. Gyre panmixis is also found in other neustonic animals (Anderson et al. 2000; Churchill et al. 2014). One interesting exception is Caprella andreae, which shows strong population differentiation across the

North Atlantic gyre (Cabezas et al. 2013), despite being an obligate associate of surface-drifting oceanic flotsam and sea turtles just like Planes. An important distinction that may contribute to this difference is that C. andreae has direct development and extended parental care (Thiel 2003), while Planes spp. are broadcast spawners with planktonic development. The fidelity of C. andreae to given raft or turtle over multiple

generations may facilitate greater genetic differentiation across the gyre (Cabezas et al.

2013). Differences in behavior and reproductive tactics among neustonic organisms are important factors to consider when interpreting patterns in population connectivity.

Conclusions

In this study, we used mitochondrial and genomic data from oceanic crabs to test hypotheses about how dispersal potential affects species-level diversification and population differentiation in the marine environment. Our results were consistent with theoretical predictions that the ability of oceanic crabs to disperse widely as pelagic larvae and as adults associated with oceanic flotsam and sea turtles, should limit their diversification and differentiation. We found low species diversity and only weak evidence of population structure among widely separated populations worldwide.

Moreover, our results highlight probable barriers and corridors to rafting dispersal at a global scale. Understanding global patterns of species diversification and population differentiation in oceanic crabs has added to the small, but growing, literature on this topic for neustronic animals. Additionally, our study demonstrates the value and future promise of genomic tools for estimating species diversity and contemporary population structure of these enigmatic organisms, as well as other organisms in which biological inferences are constrained by tools that provide insufficient genetic resolution.

Table 2-1. Sample sizes by putative species, region, habitat and genetic analysis

Number of individuals Putative species designations (Chace 1951; Poupin et al. 2005) Regiona Habitat COI only RAD only COI and RAD Total Pachygrapsus laevimanus SWP Intertidal 0 2 4 6 Pachygrapsus laevimanusb SCP Intertidal 2 0 0 2 Planes marinus NWA Turtle/Flotsam 0 0 3 3 Planes marinus SWI Flotsam 0 0 1 1 Planes marinus SEI Flotsam 0 0 2 2 Planes marinus NWP Flotsam 0 0 1 1 Planes marinus NCP Flotsam 3 3 1 7 Planes minutus NWA Turtle 4 11 7 22 Planes minutus NEA Turtle 4 16 13 33 Planes minutus MED Turtle 1 4 1 6 Planes major SWA Turtle 0 0 2 2 Planes major SEA Turtle 1 0 1 2 Planes major SWI Turtle/Flotsam 0 6 1 7 Planes major SEI Flotsam 0 0 1 1 Planes major NWP Turtle 1 18 3 22 Planes major NCP Turtle/Flotsam 3 7 6 16 Planes major NEP Turtle/Flotsam 0 7 7 14 Planes major SWP Flotsam 0 0 2 2 Planes major SCP Turtle/Flotsam 1 0 3 4 Planes major SEP Turtle/Flotsam 3 9 3 15 Total 24 84 61 169 a See Figure 2-1 for specific collection sites within each region and region abbreviations. b Old specimens, DNA too degraded for RADseq.

Table 2-2. Summary statistics of restriction-site associated DNA-sequencing (RADseq) data processing

Raw Reads Utilized Reads Loci/Individual Read Putative species (in millions) (in millions) (in millions) Depth/Locus designations (Chace 1951; Poupin et al. 2005) N Mean Min Max Mean Min Max Mean Min Max Mean s.d. Pachygrapsus laevimanus 6 0.71 0.46 0.86 0.51 0.37 0.67 0.11 0.07 0.13 4.6 28.7 Planes marinus 11 0.81 0.32 1.32 0.65 0.23 1.10 0.11 0.06 0.18 5.3 44.2 Planes minutus 52 1.04 0.33 1.56 0.89 0.21 1.34 0.12 0.04 0.17 6.4 67.2 Planes major 76 0.95 0.18 1.54 0.77 0.09 1.42 0.11 0.02 0.16 6.4 67.4 Total 145 0.90 0.31 1.34 0.72 0.21 1.15 0.11 0.04 0.16 6.0 51.0

Table 2-3. Pairwise comparison of genetic distance (FST; below diagonal) and associated P-values (above diagonal), observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 1

Clusters Ho (se) He (se) Pr A (Figure 2-6A) 1 2 Cluster 1 – Pa. laevimanus + Pl. marinus + 3 “RAD hybrids” - <0.0001 Cluster 2 – Pl. minutus + Pl. major + 12 “RAD hybrids” 0.683 -

B (not shown in Figure 2-6) 1 2 3 Cluster 1 – Pa. laevimanus + Pl. marinus - <0.0001 <0.0001 Cluster 2 – “RAD hybrids” 0.244 - <0.0001 Cluster 3 – Pl. minutus + Pl. major 0.727 0.400 -

C (Figure 2-6B) 1 2 3 4 0.068 0.127 Cluster 1 – Pa. laevimanus + Pl. marinus - <0.0001 <0.0001 <0.0001 14 (0.021) (0.032) 0.253 0.309 Cluster 2 – “RAD hybrids” 0.244 - <0.0001 <0.0001 0 (0.041) (0.032) 0.095 0.102 Cluster 3 – Pl. minutus 0.728 0.358 - <0.0001 5 (0.030) (0.021) 0.086 0.090 Cluster 4 – Pl. minutus + Pl. major 0.763 0.419 0.099 - 14 (0.032) (0.021)

Notes. Ho, observed heterozygosity; He, expected heterozygosity; Pr, number of private alleles

Table 2-4. Pairwise comparison of genetic distance (FST; below diagonal) and associated P-values (above diagonal), observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 2

Clusters Ho (se) He (se) Pr A (Figure 2-7A) 1 2 Cluster 1 – Pa. laevimanus (intertidal) - <0.0001 Cluster 2 – Pl. marinus (rafting) 0.261 -

B (Figure 2-7B) 1 2 3 0.076 0.147 Cluster 1 – Pa. laevimanus (intertidal) - <0.0001 <0.0001 1196 (0.003) (0.003) 0.084 0.133 Cluster 2 – Pl. marinus (rafting; N. Pacific) 0.272 - <0.0001 657 (0.003) (0.003) 0.113 0.207 Cluster 3 – Pl. marinus (rafting; Indian) 0.387 0.215 - 750 (0.004) (0.004)

Notes. Ho, observed heterozygosity; He, expected heterozygosity; Pr, number of private alleles

Table 2-5. Pairwise comparison of genetic distance (FST; below diagonal) and associated P-values (above diagonal), observed and expected heterozygosity and number of private alleles among clusters identified in RAD dataset 3

Clusters Ho (se) He (se) Pr A (Figure 2-8A) 1 2 Cluster 1 – Pl. minutus (NWA, NEA, MED) + Pl. major (SWA, SWI) - <0.0001 Cluster 2 – Pl. minutus (NWA) + Pl. major (SEA, SWI, SEI, Pacific) 0.122 -

B (not shown in Figure 2-8) 1 2 3 Cluster 1 – Pl. minutus (NWA, NEA, MED) - <0.0001 <0.0001 Cluster 2 – Pl. minutus (NWA) + Pl. major (SWA, SEA, SWI, SEI) 0.086 - <0.0001 Cluster 3 – Pl. major (Pacific) 0.140 0.079 -

C (Figure 2-8B) 1 2 3 4 0.188 0.213 Cluster 1 – Pl. minutus (NWA, NEA, MED) - <0.0001 <0.0001 <0.0001 29 (0.006) (0.004) 0.179 0.197 Cluster 2 – Pl. minutus (NWA) + Pl. major (SWA, SEA, SWI, SEI) 0.086 - <0.0001 <0.0001 2 (0.006) (0.005) 0.174 0.198 Cluster 3 – Pl. major (NWP, NCP, SCP, NEP, SEP) 0.156 0.088 - <0.0001 5 (0.006) (0.004) 0.191 0.207 Cluster 4 – Pl. major (NWP, SWP, NCP, NEP) 0.123 0.080 0.038 - 0 (0.006) (0.004)

Notes. Ho, observed heterozygosity; He, expected heterozygosity; Pr, number of private alleles

Table 2-6. Analysis of molecular variance (AMOVA) among 11 regions

Degrees of Percentage Sources of variation freedom Variance of variation P-value Between oceans 3 9.54 11.11 0.0019 Among regions within oceans 7 1.58 1.83 <0.0001 Among individuals within regions 221 74.80 87.06 <0.0001 Total 231 85.92 – –

Table 2-7. Pairwise comparison of genetic distance (FST; below diagonal) and associated P-values (above diagonal) among 11 ocean regions from AMOVA MED NWA NEA SAa INDb NWP SWP NCP SCP NEP SEP Mediterranean Sea (MED) – 0.546 0.061 0.017 0.001 <0.001 0.043 <0.001 0.028 <0.001 <0.001 Northwest Atlantic (NWA) -0.010 – 0.838 0.007 <0.001 <0.001 0.008 <0.001 0.001 <0.001 <0.001 Northeast Atlantic (NEA) 0.007 -0.009 – 0.003 <0.001 <0.001 0.011 <0.001 0.002 <0.001 <0.001 South Atlantic (SAa) 0.046 0.047 0.071 – 0.825 <0.001 0.185 0.003 0.107 0.002 0.006 Indian (INDb) 0.040 0.085 0.095 -0.037 – <0.001 0.013 <0.001 0.010 <0.001 <0.001 Northwest Pacific (NWP) 0.127 0.133 0.132 0.076 0.088 – 0.007 0.005 <0.001 <0.001 <0.001 Southwest Pacific (SWP) 0.151 0.128 0.151 0.093 0.093 0.090 – 0.008 0.196 0.018 0.009 North Central Pacific (NCP) 0.118 0.126 0.135 0.063 0.074 0.010 0.110 – 0.006 0.090 <0.001 South Central Pacific (SCP) 0.198 0.224 0.230 0.138 0.156 0.109 0.188 0.081 – 0.012 0.017 Northeast Pacific (NEP) 0.135 0.140 0.150 0.075 0.077 0.032 0.154 0.005 0.071 – 0.471 Southeast Pacific (SEP) 0.142 0.158 0.166 0.082 0.105 0.109 0.150 0.015 0.066 -0.005 – a Includes SWA and SEI b Includes SWI and SEI

Figure 2-1. Map showing collecting locations in 13 ocean regions (and regional abbreviations) for each putative species. Black boxes indicate how sampling sites were grouped into broad ocean regions and do not represent biogeographic boundaries.

Figure 2-2. Maximum-likelihood phylogenetic analysis of the mitochondrial gene COI for the family Grapsidae. Numbers at nodes indicate bootstrap support values and nodes with <60% bootstrap support are collapsed. Numbers at tips indicate sample sizes and estimates of nucleotide diversity within each clade. Uninformative, short branches within Group 1 and 2 are not shown to combine sequences by sampling locations.

Figure 2-3. Venn diagram showing the distribution of SNP loci among RAD datasets.

Figure 2-4. Histogram of haplotype divergence among loci in A) RAD dataset 1, B) RAD dataset 2, and C) RAD dataset 3.

Figure 2-5. Plots comparing statistical support for different numbers of populations (K) in STRUCTURE (likelihood [L[K]] and change in likelihood [delta K]) and AWCLUST (gap statistics) for A-C) RAD dataset 1, D-F) RAD dataset 2, and H-J) RAD dataset 3.

Figure 2-6. Results from clustering analyses of RAD dataset 1 at A) K=2 and B) K=4. Multidimensional scaling plots (from AWCLUST) show individuals distributed along three principal coordinate axes with different colors indicating different putative clusters and different icon shapes indicating different putative species (small gray dots indicated the positions of each point along each pair of axes). STRUCTURE bar plots show the proportion of the genome of each individual (x-axis) that originates from each putative cluster and black bars separate individuals into different clusters at each value of K. Colored bars above STRUCTURE bar plots show the consensus cluster assignment for each individual and pie charts show the putative species composition of each cluster at each value of K.

Figure 2-7. Results from clustering analyses of RAD dataset 2 at A) K=2 and B) K=3. Multidimensional scaling plots (from AWCLUST) show individuals distributed along three principal coordinate axes with different colors indicating different putative clusters and different icon shapes indicating different putative species (small gray dots indicated the positions of each point along each pair of axes). STRUCTURE bar plots show the proportion of the genome of each individual (x-axis) that originates from each putative cluster and black bars separate individuals into different clusters at each value of K. Labels below STRUCTURE bar plots show the putative species designation and geographic region of each individual. Colored bars above STRUCTURE bar plots show the consensus cluster assignment for each individual and pie charts on the map show the composition of individuals from different putative clusters in different geographic locations.

Figure 2-8. Results from clustering analyses of RAD dataset 3 at A) K=2 and B) K=4. Multidimensional scaling plots (from AWCLUST) show individuals distributed along three principal coordinate axes with different colors indicating different putative clusters and different icon shapes indicating different putative species (small gray dots indicated the positions of each point along each pair of axes). STRUCTURE bar plots show the proportion of the genome of each individual (x-axis) that originates from each putative cluster and black bars separate individuals into different clusters at each value of K. Labels below STRUCTURE bar plots show the putative species designation and geographic region of each individual. Colored bars above STRUCTURE bar plots show the consensus cluster assignment for each individual and pie charts on the map show the composition of individuals from each putative cluster in different geographic locations.

Figure 2-9. Heatmap showing pairwise comparisons of genetic distance (FST; below diagonal) and associated P-values (above diagonal) for 11 ocean regions from AMOVA.

CHAPTER 3 SOCIAL MONOGAMY IN PLANES MAJOR, A FACULTATIVE SYMBIONT OF LOGGERHEAD SEA TURTLES

Introduction

The evolution of different animal mating systems is ultimately determined by specific ecological factors that dictate the spatial and temporal distribution of available resources and mates (Emlen and Oring 1977). For symbiotic crustaceans that live in or on distinct host species, these ecological factors are defined in large part by the morphology and ecology of their hosts (Thiel and Baeza 2001). Baeza and Thiel (2007) outline a general framework for understanding how host characteristics and ecology affect the mating system and social behavior of symbiotic crustaceans. Under this theoretical framework, reproductive strategies of symbiotic crustaceans can be predicted based on four parameters: (1) host relative body size, (2) host structural complexity, (3) host abundance, and (4) the risk of mortality for symbionts away from hosts. These characteristics are considered critical in controlling the frequency of host switching and the capacity for host monopolization, and therefore the adoption of different mating systems (e.g., monogamy, pure polygamy or various forms of polygyny and polyandry). Studying how host characteristics and ecology affect the mating systems of symbiotic crustaceans offers an opportunity to understand how ecological factors contribute to the evolution of different animal mating systems.

Among other mating systems, Baeza and Thiel (2007) outline a clear set of conditions for when symbiotic crustaceans should be socially monogamous and form long-lasting heterosexual pairs. Social monogamy should be favored when hosts are relatively small in body size and structurally simple, and when hosts have relatively low abundance in habitats where the risk of mortality for symbionts (e.g., predation) away

from hosts is high. These conditions constrain movements among hosts, making host monopolization the favored behavior for both male and female symbionts due to host scarcity and the value that hosts offer as refugia (Baeza and Thiel 2007). Because spatial constraints allow only a small number of individuals (e.g., two) to cohabitate in or on the same host, both males and females maximize their reproductive behavior by cohabitating with a member of the opposite sex (Baeza 2008). Under these circumstances, resources (i.e., hosts) and mates tend to be distributed more uniformly across a dangerous environment, which makes it difficult for individuals to monopolize multiple mates or roam among hosts in search of additional mates (Baeza and Thiel

2007). Thus, symbionts inhabiting small, simple, sparse hosts in habitats where mortality risk is high away from hosts should tend to remain with an individual host and heterosexual partner for extended periods of time and adopt a monogamous mating system (Baeza and Thiel 2007). Studies on the mating strategies of symbiotic crustaceans that consider the morphology and ecology of their hosts mostly support this hypothesis (Baeza 2008, 2010; Thiel and Baeza 2001). However, other studies have found that some symbiotic crustaceans inhabiting small, simple, and sparse hosts are not strictly monogamous and display some degree of male promiscuity (Baeza et al.

2011). Additional empirical studies are needed to test the consistency and generality of these theoretical predictions.

In this study, we test the hypothesis of Baeza and Thiel (2007) that symbiotic crustaceans living in association with small, simple, sparse hosts in habitats where there is a high risk of mortality away from hosts exhibit monogamy and long-lasting heterosexual pairing. We test this hypothesis in the mating system of the flotsam crab

(Planes major) and its facultative association with loggerhead sea turtles (Caretta caretta). Planes crabs also live on pelagic flotsam and jetsam, but sea turtles represent higher quality substrata (Dellinger et al. 1997; Frick et al. 2004). Turtle hosts are relatively large in body size compared to their crab symbionts (C. caretta = 32-94 cm curved carapace length – this study; Pl. major = 8.3-26.8 mm carapace width – this study). However, Planes crabs are almost exclusively found hiding within the supracaudal and inguinal space of host turtles (Figure 3-1; Dellinger et al. 1997; Pfaller et al. 2014b), making the specific area inhabited by crabs relatively small and structurally simple. Moreover, although host turtles may concentrate at oceanic convergent zones (Polovina et al. 2000, 2004), they tend to be relatively sparse in the marine environment (0.58-0.75 turtles km-2 – Seminoff et al. 2014), especially compared to other hosts of symbiotic crustaceans (59,000 and 200,000 hosts km-2 – extrapolated from Baeza et al. 2011 and Peiró et al. 2012, respectively). Because crabs also show strong reluctance to stray from rafts and have limited swimming endurance (Davenport

1992), mortality risk for crabs off hosts is also assumed to be high. These factors should limit the ability of crabs to switch among turtles in search of additional sexual partners. In theory, the monopolization of such discrete, sparse and valuable resources

(i.e., the supracaudal and inguinal space of sea turtles) should favor monogamy with long-term heterosexual pairing (Baeza 2008; Baeza and Thiel 2007). In agreement with this prediction, Pl. major is frequently found in male-female pairs on loggerheads

(Carranza et al. 2003; Frick et al. 2011; Pons et al. 2011), and congeneric Planes minutus associated with loggerheads in the North Atlantic Ocean are found in male- female pairs more often than expected by chance (Dellinger et al. 1997). However,

there is no detailed study that tests the specific hypotheses needed to determine whether Pl. major displays social monogamy on sea turtles or whether heterosexual pairing by Planes crabs on sea turtles is long-term. Because our understanding of how host traits (i.e., relative body size, morphology, and abundance) influence the reproductive strategies of symbiotic crustaceans comes primarily from studies of symbionts living in or on benthic macro-invertebrates, this study represents a novel test of theory in a host-vertebrate, symbiont-invertebrate system.

If Pl. major are socially monogamous on C. caretta, then we expect to find that

(1) the population distribution of crabs on turtles is non-random, (2) two crabs/turtle are found more often than expected by chance, and (3) the sex distribution of crabs in pairs is non-random with male-female pairs being found more often than expected by chance.

Moreover, if heterosexual pairing is long-term, as opposed to one-time or serial monogamy, then we expect to find that (1) males pair with females regardless of their reproductive state (e.g., the presence/absence of eggs and egg developmental stage),

(2) male-female pairs display size-assortative pairing, (3) crab body size is positively correlated with host turtle body size, and (4) crabs display little to no sexual dimorphism in body size and weaponry (e.g., chelipeds used for intra-sexual aggression) (Baeza and Thiel 2007; Thiel and Baeza 2001). Support for (1) would indicate that males do not abandon females after copulation in order to roam in search of other receptive females (Diesel 1986, 1988; van der Meeren 1994). Support for (2) would indicate that pairs have grown under similar space- and resource-related constraints for long periods of time (Adams et al. 1985; Baeza 1999, 2008). Support for (3) would indicate that growth rates of crabs are related to, or constrained by, growth rates of host turtles over

time such that crabs remain on the same host, presumably with the same partner, for long periods of time (Baeza 2008). Support for (4) would indicate that selection for larger body size and weaponry in males is relaxed due to the rarity of host switching and male-male competition (Baeza 2008; Baeza and Thiel 2007; Shuster and Wade

2003). Collectively, support for these characteristics would represent a strong indication of a socially monogamous mating system in which heterosexual pairing is long-term

(Baeza 2008, 2010; Baeza and Thiel 2003; Knowlton 1980). These predictions are frequently tested when evaluating the mating strategies of symbiotic crustaceans

(Baeza 2008; Baeza et al. 2011, 2013; De Bruyn et al. 2009; Peiró et al. 2012).

Methods

Collection of Crabs

Individuals of Planes major were collected from loggerhead sea turtles (Caretta caretta) at four different localities: (1) Japan, along the east coast of Muroto on the island of Shikoku (33.28°N, 134.15°E), (2) Mexico, off Isla Magdalena on the Pacific coast of Baja California Sur (25.1–25.3°N, 112.2–112.5°W), (3) Peru, offshore along the central and southern coast (12–18.3°S, 72–80°W), and (4) Brazil, along the southern coast (27–34°S, 44–51°W) to approximately 1000 km offshore to the Rio Grande Rise

(31°S, 34.5°W). In Japan, turtles were incidentally captured by large pound net fisheries between 4 November 2010 and 24 November 2011. In Mexico, turtles were captured by hand from a small fishing boat between 3 July 2011 and 21 October 2011.

In Peru, turtles were incidentally captured by artisanal longline fisheries between 10

January 2011 and 20 January 2012. In Brazil, turtles were incidentally captured by longline fisheries between 5 July 2004 and 11 July 2006. All turtles were removed from nets and longlines within 12 hours of initial capture. Once onboard, all turtles (dead or

alive) were thoroughly inspected for Pl. major within 5 minutes (especially within the supracaudal and inguinal space of turtles), except in Japan where turtles were inspected within 3 hours. All observed crabs from each turtle were captured by hand and placed immediately in separate containers of 75-95% ethanol, or frozen and subsequently transferred to ethanol. Turtles were also measured for curved carapace length (CCL) using a flexible measuring tape (precision = 0.5 mm).

In the laboratory, all crab specimens were counted, sexed based on external characters (primarily relative abdomen width; wide in females, narrow in males, indistinguishable in juveniles – Hartnoll 1978, 1982), and measured for carapace width

(CW), cheliped length (CL) and cheliped height (CH) to the nearest 0.01 mm using

Vernier calipers. Each female crab was identified as either ovigerous or non-ovigerous based on the presence or absence of eggs underneath the abdomen. When present, each egg mass was removed and photographed under a stereomicroscope and the embryos were classified based on the following characters (Hartnoll 1963): stage I, embryos with uniformly distributed yolk and no eyespots; stage II, embryos with yolk clustered and visible, but without well-developed eyes; stage III, embryos with well- developed eyes, free abdomens, and thoracic appendages; stage IV, hatching or empty eggs. For each individual host turtle, we had information on host body size (CCL; cm), number of crabs, sex of the crabs (adult male, adult female or juvenile), body and cheliped size of the crabs (CW, CL and CH; mm), egg-carrying state for female crabs

(ovigerous or non-ovigerous), and egg stage for ovigerous female crabs (stage I, II, III or IV).

Testing Hypotheses for Social Monogamy

We tested whether the population distribution of crabs on turtles differed significantly from a random distribution by comparing the observed distribution (number of crabs per turtle) with either a Poisson distribution or a truncated Poisson distribution.

We employed a truncated Poisson distribution when the number of turtles harboring zero crabs was unknown (Plackett 1953). This was the case for Brazil and for all localities pooled. A Chi-square test of goodness of fit was used to test for significant differences between the observed and expected (null) distributions (Sokal and Rohlf

1981). We tested whether 2 crabs/turtle was observed more often than expected by chance alone using a Chi-square test of goodness of fit.

For crabs found in pairs on a single host, we tested whether the distribution of male and female crabs differed significantly from a random distribution by comparing the observed distribution of the sexes within pairs (i.e., ♂:♂, ♂:♀, and ♀:♀ ) with an expected binomial distribution. A Chi-square test of goodness of fit was used to test for significant differences between the observed and expected (null) distributions (Sokal and Rohlf 1981). These statistical procedures were carried out for each sampling locality separately and for all sampling localities pooled.

Testing Hypotheses for Long-Term Pairing

Data from separate sampling localities were pooled for the following statistical procedures. We used Chi-square tests of independence to test whether the presence of eggs or the developmental stage of the embryos carried by females affected the occurrence of males on the same turtle. We used reduced major axis (RMA) regression to test whether male and female crabs found in pairs display size-assortative pairing with respect to body size (CW, mm). We used RMA regressions to test for correlations

between turtle body size (CCL, cm) and crab body size (CW, mm) for both male and female crabs.

We used a t-test to examine differences in body size (CW, mm) between male and female crabs. In decapod crustaceans, the chelipeds serve as weapons during intra-sexual interactions (Warner 1970). We examined whether cheliped size increased linearly with body size in male and female crabs. Using the allometric model y = axb

(Hartnoll 1978, 1982), we examined the scaling relationships between carapace width

(independent variable) and cheliped length and height (CL and CH, respectively; dependent variables). The slope b of the log-log RMA regression represents the rate of exponential increase (b>1) or decrease (b<1) of each measurement relative to body size (CW, mm) of crabs. To determine if the relationships deviated from linearity, t-tests were used to test if the estimated slope b deviated from the expected slope of unity

(Sokal and Rohlf 1981). If the structures grow more or less than proportionately with a unit increase in body size of crabs, then the slope should be greater or less than unity, respectively (Hartnoll 1978). Lastly, we used analysis of covariance (ANCOVA) to test for differences between males and females in the slope of these scaling relationships.

Results

A total of 178 crabs (78 males, 91 females and 9 juveniles) was collected from

111 loggerhead turtles (Caretta caretta) captured at the four different localities (Table 3-

1). The number of crabs per turtle varied between 1 and 4 with a mean of 1.60 ± 0.66

(s.d.). A total of 149 turtles was found without crabs (Table 3-1). Turtle densities, estimated by dividing the number of turtles captured by the total area (km-2) encompassed by the widest GPS locations in each locality, ranged from 0.03 turtles km-

2 (Mexico) to 7.9 x 10-4 turtles km-2 (Brazil).

Testing Hypotheses for Social Monogamy

The population distribution of Pl. major on loggerheads was significantly different

2 from a random distribution in Japan (X 2 = 14.30, P = 0.0008) (Figure 3-2A) and Mexico

2 (X 3 = 16.03, P = 0.018) (Figure 3-2B), and not significantly different from a random

2 2 distribution in Peru (X 4 = 7.48, P = 0.11) (Figure 3-2C) and Brazil (X 4 = 8.29, P =

0.082) (Figure 3-2D). For all localities pooled, the population distribution differed

2 significantly from a random distribution (Chi-square test of goodness of fit: X 3 = 25.49,

P = 0.0001) (Figure 3-2E). We found turtles hosting two crabs more often than expected by chance in each locality separately (Chi-square test of goodness of fit:

2 2 2 Japan, X 1 = 6.30, P = 0.013; Mexico, X 1 = 6.72, P = 0.0095; Peru, X 1 = 6.67, P =

2 0.01; Brazil, X 1 = 4.66, P = 0.044) (Figure 3-2A-D, respectively) and for all localities

2 pooled (Chi-square test of goodness of fit: X 1 = 9.25, P = 0.0024) (Figure 3-2E).

For crabs found in pairs, heterosexual pairs were found more frequently than expected by chance in each locality separately (Chi-square test of goodness of fit:

2 2 Japan [Figure 3-3A], X 2 = 10.0, P = 0.0067; Mexico [Figure 3-3B], X 2 = 6.53, P =

2 2 0.038; Peru [Figure 3-3C], X 2 = 10.0, P = 0.0067; Brazil [Figure 3-3D], X 2 = 15.17, P =

2 0.005) and for all localities pooled (Chi-square test of goodness of fit: X 2 = 39.47, P <

0.0001) (Figure 3-3E). Collectively, these results support the hypothesis that Pl. major display a socially monogamous mating system on C. caretta.

Testing Hypotheses for Long-Term Pairing

Of the 45 females found in heterosexual pairs, 25 (55.5%) were ovigerous

(females carrying stage I, II, II, and IV = 9, 10, 5, and 1). Of the 28 solitary females, 13

(46%) were ovigerous (Females carrying stage I, II, III, and IV = 4, 5, 3, and 1). The proportion of paired versus solitary females that were ovigerous was not significantly

different (X2 = 0.58, df = 1, P = 0.44). The proportion of females carrying eggs at each developmental stage did not differ between solitary and paired females (Chi-square test of independence: X2 = 1.42, df = 3, P = 0.70). Thus, males were paired with females randomly with respect to the presence/absence of eggs and egg developmental stage.

We found a weak positive, but non-significant, correlation in body size between males and females forming heterosexual pairs (RMA regression: r2 = 0.073, t-test; t =

1.84, df = 1,43, P = 0.073) (Figure 3-4), indicating a lack of size-assortative pairing.

Turtle body size ranged from 32 cm CCL to 93.5 cm CCL (mean = 61 cm CCL), and crab body size ranged from 8.3 mm CW to 26.8 mm CW for females (mean = 17.8 mm

CW) and 8.4 mm CW to 23.9 mm CW for males (mean = 16.3 mm CW). We found no correlation between turtle size and female crab size (RMA regression: r2 = 0.002, t-test; t = 0.44, df = 1,84, P = 0.66) and a weak, but positive, statistically significant correlation between turtle size and male crab size (RMA regression: r2 = 0.068, t-test; t = 2.27, df =

1,71, P = 0.03) (Figure 3-5).

We found a significant difference in CW between males and females

(males

(ANCOVA: CL, interaction term F-value = 7.05, df = 1, P = 0.009; CH, F-value = 4.4, df

= 1, P = 0.037) (Table 3-1), indicating sexual dimorphism in weaponry.

Discussion

Is Planes major Socially Monogamous on Caretta caretta?

We hypothesized that if Pl. major is socially monogamous on C. caretta, then we would find (1) the population distribution of crabs on turtles is non-random, (2) two crabs/turtle are found more often than expected by chance, and (3) the sex distribution of crabs in pairs to be non-random with male-female pairs being found more often than expected by chance. Our results are consistent with this hypothesis: the overall population and sex distributions were non-random and crabs inhabited host turtles as heterosexual pairs more frequently than expected by chance. These observations are consistent with theoretical considerations that explain how environmental conditions

(e.g., host characteristics and ecology) affect the mating systems of symbiotic crustaceans (Baeza and Thiel 2007) and with past observations of Pl. minutus on C. caretta in the North Atlantic Ocean (Dellinger et al. 1997). Baeza and Thiel (2007) argue that a monogamous mating system should be adaptive under the environmental conditions found when Pl. major live on C. caretta: (1) the supracaudal/inguinal spaces on turtles are defendable resources (functionally small and structurally simple refuges),

(2) turtles tend to be sparsely distributed in the marine environment, even in foraging

‘hotspots’ (0.58-0.75 turtles km-2 – Seminoff et al. 2014), and (3) turtles offer safe refuges in habitats where the mortality risk for crabs away from host turtles is likely high

(e.g., from predation, limited swimming endurance, low substrata availability –

Davenport 1992; Hamner 1995; Shanks 1983). Under these conditions, host turtles rarely support more than two crabs, and likely because both male and female crabs maximize their reproductive behavior by cohabitating with a member of the opposite sex

(Baeza 2008), we find male-female pairs almost exclusively. These conditions should

also constrain movements among hosts, such that once socially monogamous pairs are formed they should tend to remain together on a given host turtle for extended periods of time (see next section).

In agreement with the theoretical predictions outlined by Baeza and Thiel (2007), social monogamy has been reported in numerous other symbiotic crustaceans (e.g., crabs, shrimps and isopods) that inhabit small, simple, sparse hosts (e.g., cnidarians, echinoderms, bivalves, sponges, and ascidians) in tropical environments where the predation risk for small crustaceans off hosts is presumed to be high (see references in

Thiel and Baeza 2001). Our results also support these predictions, but in a markedly different habitat and type of host. Planes crabs are primarily oceanic (occurring in water masses with depths >200 m) and surface dwelling, where their survival, growth and reproduction depend on the availability of floating substrata. Because these valuable resources are generally sparse in the open ocean, symbiosis with sea turtles is likely a highly valuable strategy. Sea turtles may even represent higher quality substrata than inanimate flotsam (Dellinger et al. 1997; Frick et al. 2004). Thus, as in other symbiotic crustaceans, associations with hosts that are both scarce and highly valuable make host guarding or host monopolization an adaptive behavior (Baeza and Thiel 2007).

The subtle difference between this turtle-crab system and other host-symbiont systems is that other sources of mortality (e.g., swimming exhaustion and low host or substrata availability) – in addition to predation – may be important for understanding the mating strategies of symbiotic crustaceans. Therefore, we argue that the theoretical predictions outlined by Baeza and Thiel (2007) can be made more general by considering all sources of mortality away from hosts, not solely predation pressure.

Host scarcity and risk of mortality alone do not necessarily lead to social monogamy. The size and complexity of symbiotic hosts must also constrain the number of symbionts, such that monopolization of a given host is energetically feasible for only a small number of symbionts (e.g., two) (Baeza and Thiel 2007). Symbiotic crustaceans associated with relatively large and morphologically complex hosts often live in large structured or unstructured groups and display polygynous mating systems (Thiel and

Baeza 2001). Interestingly, Planes crabs that colonize flotsam often live in large, seemingly unstructured aggregations (Dellinger et al. 1997; Frick et al. 2004), while crabs that colonize turtles live either solitarily or in heterosexual pairs. In this facultative system, both turtle and flotsam characteristics (i.e., size, complexity and abundance) might affect the mating behavior of the crabs, such that mating strategies are context dependent and can change depending on the characteristics of the ‘host’ (Baeza and

Thiel 2003). If turtles are just the right size and complexity for two crabs and both males and females benefit by cohabitating with a member of the opposite sex, then symbiosis facilitates social monogamy. More work is needed to test this hypothesis, but it provides a possible mechanism for the origin of obligate symbioses in which symbionts are socially monogamous.

Is Social Monogamy in Planes major Long-Term?

We hypothesized that if social monogamy in Pl. major is long-term, then we would find (1) males pair with females regardless of their reproductive state, (2) male- female pairs display size-assortative pairing, (3) crab body size is positively correlated with host turtle body size, and (4) crabs display little to no sexual dimorphism in body size and weaponry. Collectively, the results based on the hypotheses we tested were inconclusive with respect to whether heterosexual pairing is long-term. Instead, we

suspect that the duration of pairing is variable. This is somewhat inconsistent with the theoretical considerations outlined by Baeza and Thiel (2007). Baeza and Thiel (2007) argue that long-term monogamous pairing should be adaptive under the environmental conditions found when Pl. major live on C. caretta (see above). While we did find some evidence to suggest that heterosexual pairing in Pl. major is somewhat extended, we also found other lines of reasoning to suggest that pairing in Pl. major is not necessarily long-term and may involve some degree of host switching and intra-sexual (mostly male-male) competition. Our results also suggest that the line of questioning frequently employed when evaluating the mating strategies of symbiotic crustaceans (Baeza 2008;

Baeza et al. 2011, 2013; Peiró et al. 2012) needs to be expanded to accommodate a greater diversity of symbiotic interactions (e.g., vertebrate-host, crustacean-symbiont symbioses and facultative associations).

Our first hypothesis was that if heterosexual pairing is extended and not one-time or serial monogamy, then we should find that males pair with females regardless of their reproductive state. Our results are consistent with this hypothesis: males cohabit with females regardless of ovigerous state or stage of developing eggs. In promiscuous and polygamous species, heterosexual pairing is truncated and males are found with receptive females (e.g., carrying no eggs or early-stage eggs) more often than expected by chance alone (Austinixa aidae – Peiró et al. 2012). Males in these systems abandon females shortly after copulation and roam in search of other receptive females (Diesel

1986, 1988; van der Meeren 1994). Conversely, in monogamous species with extended pairing, males cohabit with females independent of their reproductive condition (Pontonia sp. – Aucoin and Himmelman 2010; Pinnixa transversalis – Baeza

1999; Pontonia marginata – Baeza 2008). Theoretically, this might be because roaming among hosts in search of other receptive females is too risky (i.e., either physically impossible or too dangerous). This pattern of pairing is consistent with what we found in Pl. major in this study. However, we also found a relatively high percentage of solitary females brooding eggs (46%). Although sperm storage has never been studied in Planes crabs, other grapsid crabs do not tend to have extended sperm storage

(Rodgers et al. 2011). Thus, if we assume that egg brooding is an indication that female Pl. major in this study had recently cohabitated and mated with a male, then one explanation for the presence of solitary ovigerous females is that males occasionally abandon females after copulation, presumably to colonize different hosts in search of additional mating opportunities. Short-term monogamy with some degree of male promiscuity and roaming has been described in other symbiotic crustaceans (Pontonia mexicana – Baeza et al. 2011; Alpheus armatus – Knowlton 1980). Similar roaming behavior by female Pl. major would be undetectable in this study because abandoned males do not show signs of recent cohabitation with the opposite sex (i.e., males are never ovigerous). Based on this line of reasoning, we cannot infer whether or not solitary male Pl. major in this study had recently cohabitated with a female. However, we found similar numbers of solitary male and female crabs (25 and 28, respectively), which suggests that roaming behavior may be similar between the sexes. If males tend to abandon females after pairing temporarily and roam in search of other females, then we might have found more solitary females than solitary males. Nevertheless, our data suggest that pairing in Pl. major is not always long-term and may involve some degree of switching among hosts.

Our second hypothesis was that if heterosexual pairing is extended, then we should find that male-female pairs display size-assortative pairing. Our results are not consistent with this hypothesis: body size of male and female crabs found in pairs was not correlated. Size-assortative pairing is expected in species that form long-term monogamous pairs because individuals in pairs would have grown under similar space- and resource-related constraints for long periods of time (Baeza 2008). Indeed, size- assortative pairing has been reported for various other long-term monogamous free- living and symbiotic crustaceans (Adams et al. 1985; Baeza 1999, 2008; Mathews

2002). In the symbiotic and monogamous crustaceans Pontonia margarita and Pinnixa tranversalis male size explains 63.8% and 77.6% of variation in female size, respectively (Baeza 1999, 2008). In contrast, male body size explains only 7.3% of the variation in female body size in Pl. major (this study) and 0.3% of the variation in female body size in Pl. minutus (Dellinger et al. 1997). This weak correlation suggests that male and female crabs do not cohabitate on the same turtle for long periods of time.

Size-assortative pairing is often weak or absent in monogamous species in which pairing is not extended and males (and/or females) switch hosts in search of additional mates (Pontonia mexicana – Baeza et al. 2011). Switching among hosts followed by random re-pairing would disrupt any size-assortative pattern, which supports the idea that monogamy is not always long-term when Pl. major associate with C. caretta.

Our third hypothesis was that if heterosexual pairing is extended, then we should find a positive correlation between the body sizes of crabs and host turtles. Our results are not consistent with this hypothesis: body size of female crabs and body size of host turtles were not correlated, and body size of male crab was only weakly correlated with

body size of host turtles. If male and/or female crabs stay with the same sexual partner and, thus the same individual host for long periods of time, a tight correlation between the body sizes of crabs and host turtles may have been found, as has been reported in various other monogamous symbiotic crustaceans (Adams et al. 1985; Baeza 1999,

2008). Such strong correlations between host and symbiont body size have been explained in terms of growth limitations experienced by symbiotic individuals that, in turn, are driven by the growth rate of their hosts (Baeza 2008). Conversely, a weak or non-existent correlation between host and symbiont size (as found in Pl. major) is usually reported for species in which males and/or females switch among hosts rather frequently (e.g., Liopetrolisthes mitra – Baeza and Thiel 2000; Thiel et al. 2003). The lack of size-assortative pairing between crabs in male-female pairs (see above) and between crabs and host turtles suggests that Pl. major may have relatively short-term associations with their individual hosts, and consequently, with the other crab inhabiting the same host. Alternatively, the weak correlation between host and symbiont body size in this system could also result from substantial differences in relative growth rate and lifespan between turtles and crabs. Because host turtles grow much more slowly

(juvenile C. caretta = 10-29% yr -1 – estimated from Bjorndal et al. 2003; grapsid crab =

46-64% yr -1 – estimated from Flores and Paula 2002) and live much longer (C. caretta

= 47-62 yrs. – Dodd 1988; grapsid crab = 2-4 yr. – Flores and Paula 2002) than symbiotic crabs, any correlation between host and symbiont size might be unperceivable. Thus, the relationship between body size of host and symbiont may be less informative for understanding the mating system of symbiotic invertebrates living in association with long-lived, vertebrate hosts.

Our fourth hypothesis was that if heterosexual pairing is extended, then we should find that crabs display little to no sexual dimorphism in body size or weaponry.

Our results were partially consistent with this hypothesis: males were on average smaller than females, but males have larger chelipeds than females for a given body size. Reverse sexual dimorphism in body size (females > males), as displayed by Pl. major, is found in other monogamous symbiotic crustaceans (e.g., Pontonia sp. –

Aucoin and Himmelman 2010; Pontonia margarita – Baeza 2008; Orthotheres tuboe –

Sakai 1969), while conventional sexual dimorphism in body size (females < males) is common among symbiotic crustaceans that display various polygynous mating systems

(Asakura 2009; Baeza and Thiel 2007). In theory, reverse sexual dimorphism reflects relaxed selection for larger body size in males because competitive interactions among males are infrequent (Baeza and Thiel 2007; Emlen and Oring 1977; Shuster and Wade

2003). This pattern supports the idea that host switching and male-male competition are infrequent and that heterosexual pairing is somewhat extended in Pl. major.

However, Pl. major also displays conventional sexual dimorphism in weaponry. Larger chelipeds in males (relative to females) suggest that males compete for and/or defend receptive females and hosts (turtles and/or flotsam) via overt aggression. Indeed, Pl. major in this study were never found in male-male pairs, but were occasionally found in female-female pairs. If this additional investment in weaponry is an indication of the frequency of agonistic interactions between males, then males likely exhibit some degree of searching and competition for females. In this situation, heterosexual pairing would be necessarily truncated (Baeza and Thiel 2007). Reverse sexual dimorphism in body size in combination with conventional sexual dimorphism in weaponry are

100

characteristics of other monogamous symbiotic crustaceans in which heterosexual pairing is not long-term (Pontonia mexicana – Baeza et al. 2011). Interestingly, females also display positive allometry in cheliped size, albeit to a lesser extent than males.

This suggests that females also participate in agonistic interactions with other crabs when securing and defending hosts and/or mates, at least temporality. This is different from other socially monogamous species, in which females allocate relatively fewer resources to weaponry with increasing body size (i.e., negative allometry in cheliped size) (e.g., Pontonia margarita – Baeza 2008; Pontonia mexicana – Baeza et al. 2011).

The observed patterns of the sexual dimorphism and allometry suggest that host switching and competition among crabs are not infrequent, which supports the idea that social monogamy in Pl. major is not necessarily long-term.

Collectively, our results suggest that Pl. major do not exclusively exhibit either extended, long-term monogamy or short-term, serial monogamy when associated with

C. caretta. Instead, our results suggest that the duration of pairing is likely variable. As outlined above, social monogamy in Pl. major was hypothesized to be a function of the size, complexity and abundance of host turtles (Baeza and Thiel 2007). While our data do not support long-term monogamy in Pl. major as predicted by theory (Baeza and

Thiel 2007), these characteristics likely have important consequences for the duration of pairing and the frequency of host switching. Because turtles are highly vagile, the relative proximity of alternative substrata – other turtles or flotsam – may vary greatly over time and space. Movements among turtles are likely very rare, as turtles tend to be sparsely distributed in the marine environment (0.03-7.9 x 10-4 turtles km-2 – this study), even in foraging ‘hotspots’ (0.58-0.75 turtles km-2 – Seminoff et al. 2014).

101

Instead, crabs may colonize or abandon turtles opportunistically when alternative substrata are in close proximity (e.g., when turtles forage along convergent zones that concentrate flotsam – Polovina et al. 2000, 2004). However, crabs likely do not actively abandon turtles when alternatives are unavailable or inaccessible. Under these conditions, the duration of monogamous pairing on host turtles may be highly variable.

Nevertheless, the process by which crabs detect turtles, assess the presence or absence of potential mates or competitors, and ultimately decide to colonize or abandon a given host turtle is entirely unknown. Results from our study suggest that crabs may colonize turtles solitarily or cohabitate with members of the opposite sex regardless of body size or reproductive state. However, more work is needed to understand the details of these interactions. Future studies should focus on quantifying the degree and direction of host switching (in the field and in the laboratory) to better understand the factors that affect the duration of monogamous pairing when Pl. major associate with C. caretta.

102

Table 3-1. Population and sex distribution of Planes major associated with Caretta caretta Turtle data Crab data

Locality N1 N0 N ♀ ♂ ♂:♀ ♀:♀ j : ♀ j : j ♀:♀:♀ ♂:♀:♀ ♂:♂:♀ ♂:♀: j : j Japan 27 116 37 6 11 10 0 0 0 0 0 0 0 Mexico 25 31 37 7 6 9 1 1 1 0 0 0 0 Peru 18 2 32 4 2 10 0 0 0 1 0 1 0 Brazila 41 - 72 11 6 16 1 2 0 0 2 1 2 Total 111 - 178 28 25 45 2 3 1 1 2 2 2 a The number of turtles without crabs in Brazil was not quantified due to logistical limitations. Notes. N1, number of turtles with at least one crab; N0, number of turtles with zero crabs; j, juvenile crab

103

Table 3-2. Relative growth of cheliped size (length and height) in females and males of Planes major Independent Lower Upper Isometric Growth Sexual variables N r2 Intercept Slope limit limit prediction P-value type dimorphisma Cheliped length Females 89 0.97 -0.27 1.07 1.04 1.11 1.0 0.004 P ♀ < ♂ Males 77 0.95 -0.35 1.18 1.12 1.24 1.0 <0.001 P Cheliped height Females 89 0.93 -0.63 1.14 1.07 1.21 1.0 0.005 P Males 77 0.92 -0.71 1.26 1.18 1.35 1.0 <0.001 P ♀ < ♂ a ANCOVA was used to test for differences in slope between males and females. See text for details. Significance level (alpha=0.05). P-values were corrected using modified t-tests to reflect differences from isometric predictions. Growth type: P = positive allometry.

104

Figure 3-1. Planes major heterosexual pair hiding within the supracaudal space of juvenile Caretta caretta (tail pulled aside to show crabs). Photo courtesy of Ricardo Santos.

105

Figure 3-2. Population distribution of the crab Planes major, symbiotic with the loggerhead sea turtle Caretta caretta in (A) Japan, (B) Mexico, (C) Peru, (D) Brazil, and (E) all four locations pooled. Observed frequency of crabs on turtles differs significantly from the expected random distribution (Poisson or truncated Poisson) for turtles in Japan and Mexico, and for the pooled data. Two crabs are found more often than expected by chance for turtles in each locality separately and for the pooled data. Sample sizes indicate numbers of turtles.

106

Figure 3-3. Male-female association pattern of Planes major found as heterosexual pairs on Caretta caretta in (A) Japan, (B) Mexico, (C) Peru, (D) Brazil, and (E) all four locations pooled. Observed frequency of heterosexual pairs differs significantly from the expected Binomial random distribution. Sample size indicates number of pairs.

107

Figure 3-4. Relationship between carapace width of females and males of Planes major found as heterosexual pairs within the supracaudal/inguinal space of the loggerhead turtles Caretta caretta. Data from all four sampling sites were pooled.

108

Figure 3-5. Relationship between curved carapace length (CCL) of loggerhead turtles Caretta caretta and carapace width of A) females and B) males of the symbiotic crab Planes major. There is a weak positive correlation between host turtle size and male crab size, but not female crab size. Data from all four sampling sites was pooled.

109

Figure 3-6. Patterns of sexual dimorphism in Planes major. A) Size frequency distribution of body size and B) relative growth of cheliped length as a function of carapace width in males (black bars and circles) and females (white bars and circle).

110

CHAPTER 4 SEA TURTLE SYMBIOSIS FACILITATES SOCIAL MONOGAMY IN OCEANIC CRABS

Introduction

Animal populations show different group characteristics depending in large part on the abundance and spatial distribution of resources (e.g., food, shelter and mates) and the capacity for individuals to monopolize limited resources (Emlen and Oring 1977;

Shuster and Wade 2003). When resources are abundant and dispersed (i.e., not limited), the per-individual value of monopolizing a resource is low and resource- defense behaviors become unnecessary. In these situations, groups (if present) should be large and inclusive. Conversely, when resources are discrete and sparse, the per- individual value of monopolizing a resource is high and individuals adopt resource- defense behaviors (e.g., herding or territoriality) to maintain access to resources. In these situations, groups should be smaller and more exclusive. Because group size and composition affect fundamental aspects of an animal’s biology, including its mating strategy (Emlen and Oring 1977), it is essential to understand how resource characteristics structure animal groups, a topic that remains poorly understood.

For animals that require discrete refuges (e.g., burrows, cavities or symbiotic hosts) for protection, mate attraction and/or successful reproduction, the ability (or inability) of individuals to monopolize refuges via competitive exclusion should depend on the size of the refuge relative to the size of the refuge-dwelling animal. Indeed, group characteristics and mating strategies of different species of obligate symbiotic crustaceans that live on or within distinct host species vary depending on the relative size and complexity of their host: large, complex hosts tend to support large, unstructured groups of symbionts, while small, simple hosts tend to support solitary

111

individuals or heterosexual pairs (Baeza and Thiel 2007). Facultative symbioses in which symbionts also use non-living hosts as refuges offer an opportunity to test how variation in refuge size influences differences in group size and composition within a species. We predict that regardless of the substrate type, groups should be larger and less exclusive when refuges are relatively large, and smaller and more exclusive when refuges are relatively small because refuge size dictates the capacity for refuge monopolization by certain individuals (Baeza et al. 2010).

We tested this prediction using the oceanic crab (Planes spp.), a facultative symbiont of loggerhead sea turtles (Caretta caretta). Planes crabs are commonly found on surface-drifting oceanic debris or flotsam, where large, demographically mixed groups seek refuge amongst colonies of stalked barnacles (Dellinger et al. 1997).

However, crabs that live on sea turtles seek refuge within the supracaudal space of host turtles (Dellinger et al. 1997; Pfaller et al. 2014a), where adult males and females form exclusive pairs, at least temporarily (Pfaller et al. 2014b). Positive allometry in weaponry (chelipeds; Warner 1970) suggests that crabs physically compete for refuges and/or mates on both substrata (Pfaller et al. 2014b). In this system, group size and composition may depend on the total available surface area of living and non-living substrata, as hypothesized by Dellinger et al. (1997). Alternatively, if refuge size is a fundamental predictor of group size and composition in this system, then we expect to find that (1) refuge area is a better predictor of adult crab number than total area for both flotsam and turtles, and (2) flotsam and turtles with similar refuge area host a similar number (i.e., 1-2) and composition (i.e., male-female pairs) of the adult crabs.

Evaluating these predictions will not only allow us to test how resource characteristics

112

structure animal groups, but also how symbiosis can facilitate the adoption of specific mating strategies.

Methods

Flotsam Data

Plastic flotsam was collected opportunistically along a 2,300 km transect through the North Pacific Subtropical Gyre. Once onboard, each item was inspected for crabs, which were collected and preserved in ethanol (70%). Each item was then characterized by general shape, and total surface area (TSA) was estimated using standard geometric equations (Table 4-1). For each item, submerged surface area

(SSAF) and ‘refuge’ surface area (RSAF) were estimated by multiplying TSA by the proportion (estimated visually) of TSA that was submerged underwater while floating and the proportion of TSA that was covered in stalked barnacles (Lepas spp.), respectively. In the laboratory, crabs were assigned to gender and maturity following

Dellinger et al. (1997).

Sea Turtle Data

We extracted data on the association between Planes crabs and 270 loggerhead turtles from the primary literature (Dellinger et al. 1997; Frick et al. 2004; Pons et al.

2011; Pfaller et al. 2014a,b), including turtle body size (curved carapace length, CCL), number of crabs per turtle, and the sex distribution of crabs in pairs. To convert linear size measurements of turtles to surface area, we quantified statistical relationships between CCL and (1) turtle submerged surface area (SSAT; comparable to SSAF), and

(2) turtle refuge surface area (RSAT) from a size series of captive and dead loggerheads

(Figure 4-1). SSAT was estimated by modeling the general shape of a turtle body

(excluding head and flippers) as one-quarter of an oblong ellipsoid (Figure 4-2), while

113

RSAT was estimated by measuring the linear dimensions of the supracaudal cavity and modeling this space as an isosceles right pentagon with two parallel sides (Figure 4-3) – this planar area was confirmed via casts and dissections of dead turtles.

Statistical Analyses

Data were log-transformed (using +1 correction) to meet statistical model assumptions of normality and homoscedasticity (crab count data) and to improve coverage (SSA and RSA). We used the corrected Aikake Information Criterion (AICc) to compare the relative fits of the data to models with one or both predictors and with different functional forms. We compared the relative strength of SSAF and RSAF as predictors of adult crab number by calculating standardized partial regression coefficients in a multiple regression model. Lastly, we used non-parametric bootstrapping (10,000 random draws with replacement) to compare the sampled distribution of the number of adult crabs found on turtles versus that of flotsam within the ranges of SSAT and RSAT.

Results

We collected 33 pieces of plastic flotsam, ranging from 16-67,749 cm2 TSA, 8-

2 2 14,905 cm SSAF (Figure 4-4A), and 0-2,608 cm RSAF (Table 4-2; Figure 4-4B). We found 0-130 adult crabs per plastic item and, when only two adult crabs were present, male-female pairs were found on seven out of eight items (Table 4-2; Figure 4-4).

Model comparisons indicated that the best-fit model was a quadratic function that included RSAF as the only predictor of adult crab number (Table 4-3). Standardized partial regression coefficients indicated that RSAF (0.61) was a better predictor of adult crab number than SSAF (0.26).

114

From data on the association between Planes crabs and loggerhead turtles (N =

2 2 270) (Table 4-3), we calculated averages for SSAT (4,090 cm ; range = 457-8,395 cm ),

2 2 RSAT (34 cm ; range = 4-66 cm ), and adult crabs per turtle (1.48; range = 1-4). For crabs found in pairs, male-female pairs were found far more frequently than expected

2 by chance (chi-square test of goodness of fit: X 2 = 94.1, P < 0.001).

Within the ranges for SSAT and RSAT, the SSAF and RSAF of plastic flotsam supported significantly more adult crabs than did turtles (bootstrapped 95% CI ≠ zero).

However, the difference in the number of crabs between flotsam and turtles was significantly different for SSA vs. RSA, with much greater values for SSA (mean difference = -9.04; 95% CI = -17.93, -2.80) than RSA (mean difference = -0.56; 95% CI

= -0.99, -0.11). Thus, the refuge surface area provided by plastic flotsam (RSAF) and turtles (RSAT) supported a similar number and composition of adult crabs.

Discussion

Our results support the prediction that refuge area, not total area, dictates group size and composition of Planes crabs living on both plastic flotsam and sea turtles, which is contrary to the hypothesis proposed by Dellinger et al. (1997). Refuge size of flotsam varied considerably (0-2,608 cm2), and we found a correspondingly wide number of adult crabs per plastic item (0-130), which was strongly correlated with increases in refuge area. As refuge area increases, the capacity of individual crabs to monopolize a given refuge area (or parts of a large refuge area) via competitive exclusion likely becomes too energetically “expensive”, allowing groups to become larger and more demographically mixed. However, the refuge area provided by smaller colonies of barnacles and turtle hosts of all sizes (RSA < 80 cm2) (Figure 4-4) is apparently sufficiently small to make refuge monopolization energetically feasible by

115

smaller, more exclusive groups of crabs (i.e., male-female pairs). These results support our prediction that group size and composition of refuge-dwelling animals are modulated by refuge size and the capacity for refuge monopolization by certain individuals (Baeza and Thiel 2007; Baeza et al. 2010).

Group size and composition affect fundamental aspects of an animal’s biology, including the adoption of different mating strategies (Emlen and Oring 1977). Although our data do not allow us to determine the specific mating strategies of crabs living in large groups on flotsam, mating strategies are likely less exclusive (e.g., polygamy or polygynandry). Because large colonies of stalked barnacles support large numbers of adult crabs, individuals (especially males) can increase their reproductive success by seeking additional mating opportunities without leaving the protection of the refuge area.

In these situations, the advantages of increased mating opportunities and reduced predation risk are likely balanced by the frequency of agonistic interactions, including territorial confrontations and cannibalism (Dellinger et al. 1997; Frick et al. 2004).

Conversely, individuals that occupy small refuges (e.g., small colonies of stalked barnacles or the supracaudal space of sea turtles) display social monogamy, in which males and females appear to form exclusive pairs, at least temporarily. In theory, this is because (1) such refuges can be monopolized by a small number of adult crabs (i.e., two) and both males and females maximize their reproductive success by cohabitating with a member of the opposite sex and (2) because roaming amongst small refuges

(i.e., different pieces of flotsam or turtles) in search of additional mating opportunities is likely to be risky (Baeza and Thiel 2007; Pfaller et al. 2014b). In this system, the conditions that promote different mating strategies vary depending on refuge size, and

116

the refuge area provided by the supracaudal space of sea turtles appears to lie within the size range of available refuges that can be monopolized by two crabs (Figure 4-4B).

Therefore, sea turtle symbiosis likely promotes social monogamy in Planes crabs due to energetic constraints imposed by refuge size and not due to refuge- or host-specific mating strategies (Pfaller et al. 2014b).

These results shed light on the evolution of obligate symbioses in which crustacean symbionts display specific mating strategies (Baeza and Thiel 2007).

Because evolution acts within the confines of intraspecific variation (Darwin 1859), it is most parsimonious to assume that obligate symbioses arose from ancestors that displayed facultative associations with either multiple living hosts or with both living and non-living substrata. As in Planes, facultative symbionts occupy a wider range of substrata and display more variation is group size and composition. Over time, if the conditions found on a particular living host consistently promoted certain group characteristics and a specific mating strategy, then selection for particular morphological and behavioral traits specific to the conditions on that host could lead to the origin of an obligate symbiosis in which the symbiont displays only that mating strategy. Because obligate associations in which symbionts display a specific mating strategy are common among symbiotic crustaceans (Baeza and Thiel 2007), this mechanism in which obligate symbioses emerge from more variable facultative associations may have been important for the diversification of marine crustaceans.

117

Table 4-1. Information used to calculate surface area of plastic items. The shape of the plastic item or the shape of components of the item determined the metrics measured to calculate surface area from standard geometric equations Shape Metric(s) measured Surface area calculation Example Cylinder width (=2r), height 2π + 2πrh bottle, rope, boat bumper Ellipsoid length (=2a), width (=2b), height (=2c) 4π((ab1.6 + ac1.6 + bc1.6)/3)1/1.6 buoy Rectangular length, width, height 2(lw + lh + wh) flat siding, insulation prism Sphere circumference (=2πr) 4πr2 buoy, toy ball

118

Table 4-2. Characteristics of plastic flotsam (N = 33)

Latitude Longitude TSA SSAF RSAF No. of adult No. Date (N) (W) Item Shape† (cm2) (cm2) (cm2) crabs 1 12-Oct-12 32º 59.3' 131º 38.1' flat piece rect. prism 16.04 8.02 0.80 0 2 14-Oct-12 33º 43.5' 133º 28.8' buoy sphere 2247.15 764.03 269.66 1 3 14-Oct-12 33º 39.3' 133º 26.2' toy ball sphere 191.16 28.67 19.12 2* 4 14-Oct-12 33º 39.3' 133º 26.2' flat piece rect. prism 50.00 25.00 0.00 0 5 14-Oct-12 33º 39.3' 133º 26.2' bottle cap cylinder 21.00 6.93 0.00 0 6 16-Oct-12 33º 34.3' 135º 26.6' buoy sphere 4066.56 1016.64 609.98 10 7 16-Oct-12 33º 33.8' 135º 25.9' flat piece rect. prism 453.00 203.85 181.20 8 8 16-Oct-12 33º 33.8' 135º 25.9' flat piece rect. prism 536.00 198.32 10.72 3 9 16-Oct-12 33º 33.8' 135º 25.9' bottle cylinder 638.99 146.97 51.12 2* 10 16-Oct-12 33º 33.8' 135º 25.9' bottle cylinder 645.96 161.97 129.19 1 11 16-Oct-12 33º 33.8' 135º 25.9' bottle cylinder 1069.54 534.77 53.48 3 12 22-Oct-12 31º 19.8' 140º 20.4' buoy sphere 1838.91 551.67 91.95 5 13 22-Oct-12 31º 19.8' 140º 20.4' buoy sphere 2141.40 963.63 471.11 55 14 22-Oct-12 31º 19.8' 140º 20.4' flat piece rect. prism 382.00 176.40 57.30 3 15 22-Oct-12 31º 19.8' 140º 20.4' flat piece rect. prism 1528.00 752.00 61.12 2* 16 23-Oct-12 30º 12.0' 140º 41.8' boat bumper cylinder 67748.64 14904.70 11856.01 130 17 23-Oct-12 30º 15.8' 140º 40.8' buoy sphere 2695.54 1078.22 32.35 1 18 24-Oct-12 30º 7.0' 141º 11.0' flat piece rect. prism 584.00 192.72 46.72 1 19 24-Oct-12 30º 7.0' 141º 11.0' buoy sphere 2637.26 923.04 237.35 9 20 24-Oct-12 30º 7.0' 141º 11.0' flat piece rect. prism 572.27 143.07 11.44 2* 21 24-Oct-12 30º 7.0' 141º 11.0' buoy sphere 199.04 59.71 29.86 3 22 24-Oct-12 30º 7.0' 141º 11.0' flat piece rect. prism 920.00 579.60 0.00 2*

119

Table 4-2. Continued

Latitude Longitude SSAF RSAF No. of adult No. Date (N) (W) Item Shape† TSA (cm2) (cm2) (cm2) crabs 23 24-Oct-12 30º 9.3' 141º 12.7' toy ball sphere 894.59 232.59 35.78 1 24 24-Oct-12 30º 9.3' 141º 12.7' bottle cylinder 604.06 132.89 60.41 2* 25 24-Oct-12 30º 8.6' 141º 18.8' buoy sphere 3293.85 1152.85 115.28 8 26 24-Oct-12 30º 3.4' 145º 3.4' buoy sphere 2141.40 1563.22 385.45 15 27 28-Oct-12 30º 24.0' 145º 45.8' rope cylinder 2211.74 1282.81 256.56 2 28 28-Oct-12 30º 24.0' 145º 45.8' flat piece rect. prism 180.00 99.00 0.00 1 29 28-Oct-12 30º 24.0' 145º 45.8' flat piece rect. prism 220.00 110.00 0.00 0 30 28-Oct-12 30º 24.0' 145º 45.8' flat piece rect. prism 411.34 205.67 0.00 0 31 28-Oct-12 30º 24.0' 145º 45.8' flat piece rect. prism 269.02 134.51 0.00 0 32 31-Oct-12 27º 0.0' 146º 46.9' buoy ellipsoid 25518.78 5103.76 2551.88 24 33 31-Oct-12 26º 59.0' 146º 47.4' buoy sphere 2038.22 1019.11 305.73 2* † See Table 4-2. * Male-female pairs.

120

Table 4-3. Summary of data on the population and sex distribution of Planes crabs associated with oceanic-stage loggerhead turtles Number of adult crabs per turtle Average number Turtle data 1 2 3 4 of adult crabs Locality N Mean CCL (range) ♀:♀ ♂:♀ ♂:♂ per turtle Azores1* 38 34.5 (13.6-59.6) 23 0 12 0 3 0 1.47 Madeira2* 105 36.9 (20.3-62.1) 47 1 49 1 1 0 1.45 Brazil3 41 59.3 (33.0-70.0) 19 1 18 0 3 0 1.61 Uruguay4 68 54.0 (42.0-70.0) 41 1 21 1 3 1 1.47 Peru3,5 18 50.4 (35.0-65.0) 6 0 10 0 2 0 1.78 Total 270 47.0 (13.6-70.0) 136 3 110† 2 12 1 1.48 * References provide insufficient data to determine exact numbers. Numbers of adult crabs per turtle may be one or two points off. † For crabs found in pairs, male-female pairs were observed more often than expected by chance (chi-square test of goodness of 2 fit against an expected binomial distribution: X 2 = 94.1, P < 0.001). References (see full citations in main article): 1Frick et al. (2004); 2Dellinger et al. (1997); 3Pfaller et al. (2014b); 4Pons et al. (2011); 5Pfaller et al. (2014a).

121

Table 4-4. Summary of statistical model comparisons for log-transformed crab count data, using corrected Aikake Information Criterion (AICC) Models AICc 1 (null) 51.56123 RSA 18.77812 RSA^2* 9.948136 RSA^3 16.25964 RSA^4 23.68437 RSA+RSA^2 12.31017 RSA+RSA^3 11.98614 RSA+RSA^4 12.27484 RSA+RSA^2+RSA^3 14.77947 RSA+RSA^2+RSA^3+RSA^4 17.3465 SSA 29.10212 SSA^2 24.03768 SSA^3 22.05682 SSA^4 22.75315 SSA+SSA^2 25.2763 SSA+SSA^3 24.6577 SSA+SSA^4 24.35054 SSA+SSA^2+SSA^3 26.95384 SSA+SSA^2+SSA^3+SSA^4 29.76898 RSA+SSA 18.59622 Notes. Lower values indicate better fit to the data (best-fit model indicated by asterisk). Standardized partial regression coefficients (beta weights) for this model show that the effect of RSA on adult crab number was 0.61, while the effect of SSA on adult crab number was 0.26.

122

Figure 4-1. Statistical relationships between turtle body size (CCL, curved carapace length), and turtle surface area (SSAT, comparable to SSAF; closed circles), and turtle refuge surface area (RSAT; open circles) from a size series of captive and dead loggerheads from Florida and Brazil (N = 23). Regression equations were used to estimate SSAT and RSAT from CCL data published in the primary literature.

123

Figure 4-2. Diagram and equation for modeling turtle surface area (SSAT) as one-quarter of an oblong ellipsoid. The head and flippers are excluded from this calculation because crabs likely do not utilize these surfaces of the turtle due to the dangers of being eaten by the turtle or becoming dislodged while the turtle is swimming.

124

Figure 4-3. Diagram and equation for modeling refuge surface area of sea turtles (RSAT) as an isosceles right pentagon with two parallel sides. A and B are curved measurements taken by pressing a flexible measuring tape along the wall of the supracaudal cavity. C is a curved measurement taken along at the junction of the last two marginal scutes and the skin of the supracaudal cavity.

125

Figure 4-4. Best-fit models (±95% CI) of the relationship between A) submerged surface area and B) refuge surface area and number of adult crabs for flotsam (white and black circles), with mean values for sea turtles superimposed (grey circles; error bars indicate the range of values).

126

CHAPTER 5 HITCHHIKERS REVEAL CRYPTIC HOST BEHAVIOR: NEW INSIGHTS FROM THE ASSOCIATION BETWEEN PLANES MAJOR AND SEA TURTLES IN THE PACIFIC OCEAN

Introduction

Large, highly mobile marine vertebrates are inherently difficult to study and monitor, which makes them vulnerable to overexploitation (Heppell et al. 1999). Long life spans, delayed sexual maturity, and wide spatiotemporal patterns of habitat use prevent direct monitoring of individuals during most life stages, especially as juveniles.

An interdisciplinary approach is needed to better understand the ecology of marine vertebrates during cryptic life stages (Jones and Seminoff 2013). New advances in molecular and satellite-tracking technologies have helped reveal critical information on migratory behavior and patterns of habitat use of these elusive and often threatened organisms (Graham et al. 2012; Saba 2013). However, such tools remain expensive and, in the case of satellite tracking, the small number of transmitters that are deployed and the short duration of deployment relative to life span limit our ability to infer population-wide and long-term patterns. Studies that incorporate information from habitat-specific ecological interactions (e.g., with predators, prey, parasites, and commensals) can reveal valuable insights into the behavioral patterns related to habitat use and help inform the conservation of marine vertebrates.

The external surfaces of marine vertebrates are often colonized by a variety of marine plants and animals. This phenomenon, termed epibiosis, results when a host supports one or more colonizers, called epibionts (Wahl and Mark 1999). Most epibionts are unspecialized organisms normally found associated with inanimate structures in the surrounding marine environment (i.e., “free living”). Epibiosis

127

necessitates spatial overlap between the geographic ranges and habitats of the hosts and free-living populations of potential epibionts (Frick and Pfaller 2013). Thus, the epibionts associated with a given host should reflect the characteristic assemblage of plants and animals that occupy the regions and habitats where the host spends time.

Because marine vertebrates often use different geographic regions and habitats during different behaviors or life stages, the presence of particular epibiont taxa with more limited distributions can reveal information about cryptic host behavior (Frick and Pfaller

2013).

Studies of the associations between sea turtles and their epibionts have provided important information on their migratory behavior (Killingley and Lutcavage 1983; Caine

1986; Eckert and Eckert 1988) and patterns of habitat use (Pfaller et al. 2008; Reich et al. 2010; Hosono and Minami 2011). For example, the foraging grounds and migratory corridors of nesting loggerhead turtles (Caretta caretta) have been inferred from the presence of particular epibiont taxa that are geographically restricted to areas away from the nesting beaches (Caine 1986; Pfaller et al. 2008). The most common dichotomy that has been used to infer inter- and intraspecific differences in habitat use among sea turtles is the presence of particular epibiont species associated with turtles occupying either epipelagic (upper 200 m in areas with >200 m depth) or neritic/benthic habitats (Limpus and Limpus 2003; Reich et al. 2010). The presence of epipelagic organisms such as pedunculate barnacles of the genera Lepas and Conchoderma and grapsid crabs of the genus Planes on sea turtles indicates that turtles are occupying or have recently occupied epipelagic habitats, which provides valuable insights into cryptic migratory behaviors and patterns of habitat use of sea turtle populations.

128

Studies of Planes crabs associated with sea turtles represent the most detailed information to date on sea turtle-epibiont symbiosis (Frick and Pfaller 2013). Planes crabs may settle on turtles as megalopal-stage larvae or colonize turtles after initially settling on flotsam. Initial colonization likely occurs while epipelagic turtles are resting or swimming slowly at the surface of the ocean. Crabs may become disassociated from turtles when turtles remain submerged for long periods of time, when turtles enter neritic waters, or when turtles come ashore to nest. Associations between Planes minutus and oceanic-stage loggerheads in the North Atlantic Ocean are common – 82% of turtles host crabs (Dellinger et al. 1997) – and crab dietary data suggest a cleaning association

(Davenport 1994; Frick et al. 2000, 2004). More recently, similar symbiotic associations have been described for Planes major on sea turtles in the South Atlantic (Carranza et al. 2003; Bugoni et al. 2007; Pons et al. 2011) and Pacific oceans (Barceló et al. 2008;

Frick et al. 2011). Because differences in the frequency of Planes crabs on sea turtles suggest inter- and intraspecific differences in the use of epipelagic habitats, quantitative surveys for crabs on turtles can provide information on cryptic behavioral traits of sea turtle populations.

Patterns of habitat use of sea turtles in the Pacific Ocean tend to be similar to those in the Atlantic Ocean (Musick and Limpus 1997). All species, excluding the flatback turtle (Natator depressus), use epipelagic habitats for juvenile development

(Musick and Limpus 1997). In general, leatherback (Dermochelys coriacea) and olive ridley (Lepidochelys olivacea) turtles tend to remain epipelagic throughout adulthood

(Plotkin 2010; Saba 2013), while loggerhead (Caretta caretta), green (Chelonia mydas) and hawksbill (Eretmochelys imbricata) turtles tend to transition to neritic habitats as

129

juveniles and remain neritic as adults (Musick and Limpus 1997). Recently, however, several studies have shown that alternative patterns of habitat use exist among turtles in the Pacific Ocean. Satellite telemetry and stable isotope analyses of loggerhead and green turtles in the western North Pacific indicate that adult turtles use both epipelagic and neritic habitats (Hatase et al. 2002, 2006, 2010). Moreover, juvenile loggerheads in the North Pacific are known to occupy either epipelagic habitats in the central North

Pacific or neritic habitats along the Baja California peninsula, where there are fundamental differences in the distribution, abundance, and quality of prey (Parker et al.

2005; Peckham et al. 2011). Because the existence of alternative strategies has emerged, more studies are needed to assess population-wide variation in patterns of habitat use and behavior among different sea turtle species and among different life stages.

In this study, we examined the interactions between Pl. major and sea turtles at five sites in the Pacific Ocean. We quantified the effects of turtle species, turtle size and habitat (neritic or epipelagic) on the frequency of these associations. These results were then integrated with a review of all known records of Pl. major-sea turtle interactions in the Pacific Ocean to evaluate two primary questions: (1) Do turtles display variable/flexible epipelagic-neritic transitions? and (2) Do turtles display similar surface-dwelling behavior in epipelagic habitats?

Methods

Associations between Planes major and sea turtles in the Pacific Ocean were surveyed at five sites (Figure 5-1A): (1) Japan, in neritic habitats along the east coast of

Muroto on the island of Shikoku (33.28°N, 134.15°E) (Figure 5-1B); (2) Hawaii, on Oahu

(21.47°N, 157.98°W) and within the surrounding epipelagic habitat (14.42–28.62°N,

130

140.10–166.82°W) (Figure 5-1C); (3) Samoa, within the surrounding epipelagic habitat

(7.13–16.47°S, 166.2–171.32°W) (Figure 5-1D); (4) Mexico and Central America, in neritic and epipelagic habitats from Baja California Sur, Mexico (25.85°N, 111.97°W) to the Nicoya Peninsula, Costa Rica (10°N, 85.42°W) to 1800 km offshore (8.52°N,

118.25°W) (Figure 5-1E); and (5) Peru, in neritic habitats in Sechura Bay (5.75°S,

80.95°W) and epipelagic habitats along the entire coast from Illescas Peninsula in the north (5.96°S, 81.05°W) to Ilo in the south (18.20°S, 73.03°W) (Figure 5-1F).

In Japan, neritic turtles incidentally captured by large pound net fisheries between 4 November 2010 and 24 November 2011 were inspected for crabs within 0.5-

3 hours of capture. In Hawaii and Samoa, epipelagic turtles recovered dead in longline fisheries or found stranded on beaches in Oahu between 29 October 2010 and 14

February 2013 were initially frozen and inspected for crabs up to four weeks post- mortem. In Mexico and Central America, turtles captured by hand from a small boat between 8 August 2003 and 16 November 2003 and between 3 July 2011 and 21

October 2011 were inspected for crabs within 5-10 min of capture. In Peru, turtles incidentally captured by artisanal gillnet (neritic) and longline fisheries (epipelagic) between 10 January 2011 and 31 May 2012 were inspected for crabs within 5-10 min of capture. All turtles were removed from nets and longlines within 12 hours of entanglement or hooking, while turtles captured by hand in Mexico and Central America were removed from the water within 30 seconds. All turtles were measured for body size (curved carapace length, CCL). When observed, Pl. major specimens were captured by hand and placed immediately in 75-95% ethanol, or frozen and subsequently transferred to 75-95% ethanol, for future studies. In Mexico and Central

131

America, ocean depth (m) was measured during each capture event and turtles were separated by habitat type: neritic (<150 m depth) and epipelagic (>350 m depth).

Turtles captured between 150-350 m depth were not included in this study.

Frequency of epibiosis (F0) was derived by dividing the number of turtles hosting

Pl. major by the number of turtles surveyed. We determined F0 for each turtle species in each site separately, for each turtle species for each habitat type (neritic and epipelagic), for each site for all applicable turtle species combined, and for each turtle species regardless of site/habitat. We used contingency tables and Fisher’s Exact

Tests to test for differences in F0 (1) among turtle species within each habitat, (2) between habitats within each turtle species, and (3) among turtle species regardless of habitat (Bonferroni correction for 13 Fisher’s Exact Tests: corrected alpha = 0.004).

Contingency tables and Fisher’s Exact Tests were used in place of logistic regression analyses because these methods generate more robust results for unbalanced sampling designs and when expected values are less than 10 (Hirji et al. 1991). To test if turtle size affected F0, we performed binomial logistic regressions for each turtle species from each site. All statistical analyses were performed in R for Windows v.

3.0.0 (R Development Core Team 2008).

Results

We surveyed 584 turtles for the presence of Pl. major and found 169 (F0 =

28.9%) hosting at least one crab. All crabs were identified as Pl. major following Chace

(1951), and most crabs were found as singletons or in heterosexual pairs regardless of turtle species, site or habitat. The frequency of epibiosis for the three primary turtle species – loggerhead turtles (Caretta caretta), green turtles (Chelonia mydas), and olive ridley turtles (Lepidochelys olivacea) – varied among species, sites and habitats (Table

132

5-1). We surveyed two hawksbill turtles (Eretmochelys imbricata) and neither turtle hosted crabs (1 in Japan and 1 in Peru). We surveyed one leatherback turtle

(Dermochelys coriacea) in Hawaii and it did not host crabs. Because of small sample sizes (N < 5), we excluded some data from statistical analyses: one hawksbill in Japan, two green turtles and one leatherback in Hawaii, and two olive ridleys and one hawksbill in Peru.

We found significant differences in F0 (1) among turtle species within each habitat, (2) between habitats within each turtle species, and (3) among turtle species regardless of habitat (Tables 5-1, 5-2). In neritic habitats and when data from both habitats were combined, loggerheads and olive ridleys hosted Pl. major at the same frequency, but more frequently than green turtles. In epipelagic habitats, loggerheads hosted Pl. major more frequently than green turtles and olive ridleys, which host Pl. major with the same frequency. Turtles in epipelagic habitats hosted Pl. major more frequently than turtles in neritic habitats for each turtle species separately and for all turtle species combined.

We found a significant negative effect of turtle size on F0 for neritic loggerheads in Mexico (Table 5-3; Figure 5-2B). We found no significant effect of turtle size on F0 for neritic loggerheads in Japan (Figure 5-2A), epipelagic green turtles in Peru (Figure 5-

3C), epipelagic olive ridleys in Hawaii (Figure 5-4A), neritic olive ridleys in Mexico and

Central America (Figure 5-4B), and epipelagic olive ridleys in Mexico and Central

America (Table 5-3; Figure 5-4C). The effect of turtle size on F0 was not tested for (1) epipelagic loggerheads in Peru because F0 was 100% on turtles with CCL data (Figure

5-2C), (2) neritic green turtles in Japan because F0 was 0% (Figure 5-3A), (3) epipelagic

133

green turtles in Samoa because only one turtle hosted Pl. major (Figure 5-3B), and (4) neritic green turtles in Peru because only one turtle hosted Pl. major (Figure 5-3C)

(Table 5-3).

Discussion

In this study, we quantified the interactions between Pl. major and sea turtles in epipelagic and neritic habitats in the Pacific Ocean. Several of these associations represented previously undescribed and/or un-quantified interactions. We integrated these findings with those from previous studies on Pl. major-sea turtle interactions in the

Pacific Ocean (Table 5-4) to gain new insights into cryptic habitat-use patterns and behavior of sea turtles in this region. Because inter- and intraspecific differences in the frequency of epibiosis by Pl. major (F0) suggest differences in epipelagic habitat use and behavior, we used F0 data to address our two questions: (1) Do turtles display variable/flexible epipelagic-neritic transitions? and (2) Do turtles display similar surface- dwelling behavior in epipelagic habitats?

Do Turtles Display Variable/Flexible Epipelagic-Neritic Transitions?

Most sea turtles exhibit ontogenetic habitat shifts, in which juvenile turtles transition from being primarily epipelagic to being primarily neritic (Musick and Limpus

1997). When turtles transition from epipelagic to neritic habitats, Planes crabs – and other epipelagic epibiota – are transported away from their optimal habitat and into areas with different physiological conditions and higher densities of predators. Such transitions also involve characteristic changes in the diving behavior of turtles, in which epipelagic turtles transition to a more benthic existence (Bolten 2003). These changes in behavior likely have a strong detrimental effect on the association between crabs and their host turtles. Our data support this premise: F0 in epipelagic habitats (55.2%) was

134

significantly greater than F0 in neritic habitats (17.8%). Based on this assumption, the presence of crabs on neritic turtles strongly suggests that turtles recently occupied epipelagic habitats. Differences in F0 should therefore reflect differences in epipelagic- neritic transitions. Three hypotheses emerge to explain differences in F0 among neritic turtles: (1) if turtles transition unidirectionally (epipelagic -» neritic) at small body sizes, then neritic turtles should show relatively high F0 at small body sizes and relatively low

(or 0) F0 at larger body sizes; (2) if turtles transition unidirectionally (epipelagic -» neritic) at variable body sizes, then neritic turtles should show relatively high F0 across all sizes;

(3) if turtles transition bi-directionally (epipelagic «-» neritic), then neritic turtles should show relatively high F0 across all sizes. Because hypotheses (2) and (3) are indistinguishable using data on F0, we can combine these two hypotheses into one: if turtles display variable/flexible epipelagic-neritic transitions, then neritic turtles should show relatively high F0 across all sizes. Support for this hypothesis would suggest that species and/or populations of turtles exhibit a high use of epipelagic habitats. We can explore these two hypotheses (1 vs. 2/3) using the F0 data to distinguish differences in epipelagic-neritic transitions among different turtle species in different locations.

Loggerhead turtles surveyed in neritic habitats off Japan were found hosting crabs relatively frequently (F0 = 21.1%) across a wide range of body sizes (60–100 cm

CCL). These subadult and adult turtles are thought to exhibit unidirectional transitions to neritic habitats at a relatively narrow range of body sizes (60–80 cm CCL – modified from Ishihara et al. 2011). However, our data suggest that the transition to neritic habitats is more variable and/or flexible. Moreover, F0 for neritic loggerheads in Japan was considerably higher than for neritic aggregations in the Atlantic Ocean (<5% – Frick

135

et al. 1998, 2006), suggesting a greater use of epipelagic habitats by neritic-stage loggerheads in Japan. Previous studies applying satellite telemetry and stable isotope analysis suggest that adult loggerheads in Japan display persistent alternative foraging strategies (oceanic vs. neritic) (Hatase et al. 2002, 2010). However, our results based on F0 suggest that this dichotomy in habitat use is less clear-cut and some individual turtles may actually use both foraging areas. Thus, our data support the hypothesis that

(at least some) subadult and adult loggerheads in Japan display variable/flexible epipelagic-neritic transitions.

Loggerheads surveyed in neritic habitats in Mexico were mostly juvenile turtles

(39-86 cm CCL – modified from Nichols et al. 2000; this study) and were found hosting crabs frequently (44.6%), especially within the lower portion of the observed range of sizes (30–65 cm CCL). Dietary and demographic data for juvenile loggerheads (40–80 cm CCL) in the eastern North Pacific suggest that turtles occupy either epipelagic habitats in the central North Pacific or neritic habitats along Mexico’s Baja California peninsula (Parker et al. 2005; Peckham et al. 2011). Our results suggest that turtles that transition to neritic habitats may transition unidirectionally or bidirectionally, but tend to do so less frequently with increasing body size. Those that remain epipelagic may host crabs frequently (e.g., epipelagic juvenile loggerheads off Peru, F0 = 93% – this study), but were not surveyed in this study. Nevertheless, our data support the hypothesis that (at least some) juvenile loggerheads in Mexico display variable/flexible epipelagic-neritic transitions.

Green turtles surveyed in neritic habitats in Japan and Peru almost invariably did not host crabs (F0 = 0% and 1% in Japan and Peru, respectively). This result is in stark

136

contrast to neritic loggerhead (F0 = 27.6%; see above) and olive ridley (F0 = 26.2%; see below) turtles. That green turtles in epipelagic habitats hosted crabs frequently (38.5%), albeit less frequently than epipelagic loggerheads (F0 = 92.9%) suggests that green turtles that transition to neritic habitats tend to do so unidirectionally at small body sizes and remain there throughout development. Because very few small neritic green turtles

(<45 cm CCL) were surveyed, we cannot evaluate whether F0 is indeed high on these small post-epipelagic turtles. It is plausible that small green turtles transition to shallow protected foraging areas where crabs are lost, then move to more high-energy coastal waters where they were captured and surveyed in this study. Such behavior has been described for green turtles in the eastern Pacific Ocean (Seminoff et al. 2003).

Demographic studies of green turtles in Peru indicate that turtles are primarily neritic

(Alfaro-Shigueto et al. 2011), but data from fisheries bycatch and satellite telemetry also indicate some use of epipelagic habitat (Seminoff et al. 2007). Interestingly, green turtles in Peru were found hosting crabs frequently in epipelagic habitats (F0 = 60.0%) and rarely in neritic habitats (F0 = 1%). Because turtles were similar in size, this suggests that these turtles may exhibit persistent alternative foraging strategies: epipelagic versus neritic. More work is needed to understand the consistency of this dichotomy in habitat use. Nevertheless, our results support the hypothesis that green turtles in neritic habitats do not display variable/flexible epipelagic-neritic transitions.

Olive ridleys surveyed in neritic habitats along Mexico and Central America were mostly adults (>60 cm CCL) and were found hosting crabs frequently (F0 = 26.2%), though not as frequently as reported in Barceló et al. (2008) (F0 = 50%; Table 5-4).

While there is a paucity of data on the life-history patterns of olive ridleys (Jones and

137

Seminoff 2013), recent studies that apply satellite telemetry and demographic data have shown that both juveniles and adults in the eastern Pacific Ocean are highly epipelagic and nomadic (Work and Balazs 2010; Plotkin 2010). Our data suggest that some adult olive ridleys (>60 cm CCL) make forays into neritic habitats, likely during foraging bouts.

This result is consistent with results from dietary analyses of olive ridleys in the eastern

Pacific Ocean (Kopitsky et al. 2005). Because epipelagic olive ridleys hosted crabs frequently (F0 = 50%), turtles may remain in neritic habitats for somewhat extended periods of time – long enough for F0 to decline from 50% to 26% – before returning to epipelagic habitats. Alternatively, neritic forays that involve nesting events may be relatively short in duration, but highly detrimental to associations with Pl. major (see F0 for nesting olive ridleys in Table 5-4). That a large proportion of the olive ridleys surveyed in Mexico and Central America in this study were surveyed during the nesting season (July to December) lends support to this idea. Our results support the hypothesis that olive ridleys in eastern North Pacific Ocean display variable/flexible epipelagic-neritic transitions.

Do Turtles Display Similar Surface-Dwelling Behavior in Epipelagic Habitats?

As ‘free-living’ individuals, Planes crabs colonize flotsam as megalopal-stage larvae and spend the remainder of their lives rafting at the surface in the open ocean

(Chace 1951). For this reason, crabs likely colonize turtles at the surface and remain associated with turtles that spend a significant proportion of time at or near the surface.

Conversely, crabs may abandon, avoid, and/or infrequently colonize turtles that remain submerged for long periods of time. Based on these assumptions, interspecific differences in F0 among epipelagic turtles suggest differences in epipelagic surface- dwelling behavior. Two alternative hypotheses emerge that might explain differences in

138

F0 among epipelagic turtles: (1) if turtles display similar surface-dwelling behavior in epipelagic habitats, then F0 should be similar among species and (2) if turtles display different surface-dwelling behavior in epipelagic habitats, then F0 should be different among species. Additionally, F0 should be proportional to the amount of time spent at or near the surface, such that turtles with high F0 spend more time at or near the surface than turtles with low F0.

We surveyed epipelagic turtles for Pl. major at four different sites (Hawaii,

Samoa, Mexico/Central America and Peru). We found that epipelagic loggerheads hosted crabs (F0 = 92.9%) more frequently than epipelagic olive ridleys (F0 = 50%) and green turtles (F0 = 38.5%). Although we surveyed epipelagic loggerheads in only one location in this study (Peru), F0 for epipelagic loggerheads in the North and South

Atlantic Ocean are also high (F0 = 82% – Dellinger et al. 1997; F0 = 83% – Carranza et al. 2003), which supports the consistency of this pattern. High F0 in epipelagic loggerheads suggests that these turtles spend a large proportion of their time at or near the surface. This is consistent with the dive patterns of epipelagic loggerheads, in which >80% of dives occur to depths less than 5 m and 40% of time is spent within 1 m of the surface (Polovina et al. 2003, 2004; Howell et al. 2010). Dietary data indicate that these turtles forage primarily on surface-dwelling prey, including organisms associated with flotsam (e.g., Planes crabs and Lepas barnacles – Parker et al. 2005; Peckham et al. 2011). This foraging behavior would provide frequent opportunities of Pl. major to colonize epipelagic loggerheads, and, once colonized, the tendency of these turtles to spend time at the surface would likely facilitate the persistence of these associations.

Both behavioral characteristics might contribute to higher F0 in epipelagic loggerheads.

139

In contrast, comparably lower F0 values for epipelagic olive ridleys and green turtles suggest that these turtles display different surface-dwelling behavior than epipelagic loggerheads and tend to spend less time at or near the surface. Differences in F0 between loggerheads and olive ridleys are consistent with data from depth-time recorders, which show that epipelagic olive ridleys spend less time at the surface and make deeper dives than epipelagic loggerheads (Polovina et al. 2003, 2004). Dietary data also support this pattern: epipelagic olive ridleys consume more subsurface prey

(e.g., pyrosomes and salps) than epipelagic loggerheads (Kopitsky et al. 2005; Polovina et al. 2004). Although comparable depth-time and dietary data are not available for epipelagic green turtles, epipelagic green turtles in the North Pacific Ocean are known to conduct ‘resting’ dives, in which turtles appear to obtain neutral buoyancy at ~35-40 m depth and remain there for some period of time (Hays et al. 2001; Rice and Balazs

2008). Subsurface diving behavior displayed by epipelagic olive ridleys and green turtles might deter Pl. major from colonizing and/or persisting on these turtles in epipelagic habitats, which would contribute to lower F0. Moreover, if these turtles do not frequently associate with flotsam in epipelagic habitats, as loggerheads do, then there may be fewer opportunities for colonization by Pl. major. Our data support the hypothesis that epipelagic loggerheads display different surface-dwelling behavior than epipelagic olive ridleys and green turtles, in which loggerheads spend more time at or near the surface. However, more work is needed to better understand the mechanisms driving the observed differences in F0 among epipelagic turtles.

Caveats

Methodological differences at different sampling sites (see Methods) may have affected F0. Turtles that spent more time within capture gear or out of the water before

140

being inspected for crabs (Japan, Hawaii and Samoa) may have had F0 values that were biased low. However, the effect of these potential biases would not change our interpretations of the results, and therefore our primary insights and conclusions would remain the same. (1) If F0 was slightly higher for both neritic turtles in Japan and epipelagic turtles in Hawaii and Samoa, then F0 for epipelagic and neritic turtles overall would each be slightly higher, maintaining the general pattern. (2) If F0 was slightly higher for neritic turtles in Japan, then support for variable/flexible epipelagic-neritic transitions among neritic loggerheads would be strengthened and support for unidirectional epipelagic-neritic transitions among neritic green turtles would be somewhat weakened, but not sufficiently to change our interpretation. (3) If F0 were higher for epipelagic olive ridleys and green turtles in Hawaii and Samoa, respectively, then support for the frequency of surface-dwelling behavior would increase for both species, but F0 for epipelagic loggerheads would still be considerably higher, maintaining our initial interpretation.

Perspectives

In this study, we used the occurrence of epipelagic epibionts to infer cryptic habitat-use patterns and behavior of sea turtles in the Pacific Ocean. Because epibiosis necessitates spatial overlap between the habitats occupied by hosts and free-living populations of potential epibionts, researchers can use the presence of particular epibiont species with more limited habitat distributions to identify the habitats that the host has recently occupied (Frick and Pfaller 2013). This approach may prove to be informative for other sea turtle populations, as well as other marine vertebrates. In addition to epibionts associated with epipelagic versus neritic/benthic habitat use, other epibiotic assemblages that reflect habitat-use dichotomies in the aquatic environment

141

(e.g., freshwater versus marine, polar versus equatorial) may reveal important information on cryptic host movements and behavior. Studies on the epibionts of marine vertebrates represent a time- and cost-effective method to infer more population-wide patterns, especially when more expensive technologies are unavailable. Future studies that utilize more advanced technologies could integrate epibiont data to elucidate a more detailed and complete picture of the ecological interactions, as well as habitat-use patterns, of large, highly mobile marine vertebrates.

142

Table 5-1. Comparison of frequency of epibiosis among turtle species, site and habitat Frequency of epibiosis (%) Site Habitat CC CM LO Total Japan N 21.1 (147) 0.0 (47) – 16.0 (194) Hawaii E – – 33.3 (15) 33.3 (15) Samoa E – 9.1 (11) – 9.1 (11) Mexico & C. N 44.6 (56)† – 26.2 (61) 35.0 (117) America Mexico & C. E – – 52.3 (105) 52.3 (105) America Peru N – 1.0 (99) – 1.0 (99) Peru E 92.9 (28) 60.0 (15) – 81.4 (43) All sites N 27.6 (203) a/a 0.7 (146) a/b 26.2 (61) a/a 17.8 (410) a/- All sites E 92.9 (28) b/a 38.5 (26) b/b 50.0 (120) b/b 55.2 (174) b/- Total N & E 35.5 (231) -/a 6.4 (172) -/b 42.0 (181) -/a 29.0 (584) † All turtles captured in 2011 off Baja California Sur, Mexico (see inset on Figure 5-1E). Notes. Neritic (N), epipelagic (E), Caretta caretta (CC), Chelonia mydas (CM), Lepidochelys olivacea (LO). Samples sizes are in parentheses. Superscripts before slash indicate significant differences between habitats and superscripts after slash indicate significant differences between turtle species. Different superscript letters indicate values that are significantly different from each other. Bonferroni correction for 13 Fisher’s Exact tests of independence: corrected alpha = 0.004 (see Table 5-2 for statistical details).

143

Table 5-2. Summary of 13 Fisher’s Exact Tests comparing frequency of epibiosis among turtle species, site and habitat Comparisons Odds ratio (95% CI) df P-value Sig. CC Neritic CC Epipelagic 0.03 (0.00-0.13) 1 < 0.0001 S CM Neritic CM Epipelagic 0.01 (0.00-0.09) 1 < 0.0001 S LO Neritic LO Epipelagic 0.36 (0.17-0.73) 1 0.002 S Neritic Epipelagic 0.18 (0.12-0.27) 1 < 0.0001 S CC Neritic CM Neritic 54.8 (9.2-2210.3) 1 < 0.0001 S CC Neritic LO Neritic 1.07 (0.54-2.20) 1 0.87 NS CM Neritic LO Neritic 0.02 (0.00-0.13) 1 < 0.0001 S CC Epipelagic CM Epipelagic 19.4 (3.6-204.2) 1 < 0.0001 S CC Epipelagic LO Epipelagic 12.8 (3.0-116.5) 1 < 0.0001 S CM Epipelagic LO Epipelagic 0.63 (0.23-1.61) 1 0.62 NS CC CM 8.02 (4.05-17.4) 1 < 0.0001 S CC LO 0.76 (0.50-1.16) 1 0.19 NS CM LO 0.10 (0.04-0.19) 1 < 0.0001 S Notes. Neritic (N), epipelagic (E), Caretta caretta (CC), Chelonia mydas (CM), Lepidochelys olivacea (LO), not significant (NS), significant (S).

144

Table 5-3. Results of binomial logistic regression analyses testing the effect of turtle size on frequency of occurrence Binomial logistic regression Species Site Habitat Slope Intercept z-value df P-value Sig. CC Japan N -0.02 0.046 -0.65 138 0.51 NS CC Mexico N -0.09 5.34 -2.42 54 0.02 S CC Peru E did not test (see text) CM Japan N did not test (see text) CM Samoa E did not test (see text) CM Peru N did not test (see text) CM Peru E -0.03 1.81 -0.29 12 0.77 NS LO Hawaii E -0.76 -49.2 1.78 13 0.08 NS LO Mexico & N -0.08 4.45 -1.07 59 0.28 NS C. America LO Mexico & E -0.01 0.92 -0.87 102 0.38 NS C. America Notes. Neritic (N), epipelagic (E), Caretta caretta (CC), Chelonia mydas (CM), Lepidochelys olivacea (LO), not significant (NS), significant (S).

145

Table 5-4. Summary of all known references to Planes major on sea turtles in the Pacific Ocean

Site Habitat Species Stage N F0 (%) Reference Japan N UNK - 1 - Sakai (1939) via Chace (1951) Japan N CC S,A 147 21 this study Japan N CM J,S,A 47 0 this study Japan N EI J 1 0 this study Australia N CC S,A >100 - Limpus and Limpus (2003) Samoa E CM J 11 9 this study Samoa E EI J 1 - Van Houtan, unpubl data Hawaii E,D CM J 2 50 this study Hawaii E DC J 1 0 this study Hawaii E,D LO J,S,A 15 33 this study California N CC J 1 - Guess (1981) California N CM - 1 - Wicksten and Behrens (2000) California N LO A 1 - Hubbs (1977) Mexico N CC J,S 87 43 Barceló et al. (2008) Mexico N CC J,S 56 45 this study Mexico N CM - >1 - Faxon (1895) via Chace (1951) Mexico N CM - 1 - Crane (1937) Mexico B CM A 6 0 Lazo-Wasem et al. (2011) Mexico N EI - 1 - Steinbeck and Ricketts (1941) Mexico B LO A 44 9 Hernández-Vásquez and Valadez-González (1998) Mexico B LO A 12 - Angulo-Lozano et al. (2007) Mexico N LO S,A 14 50 Barceló et al. (2008) Mexico B LO A 124 3 Lazo-Wasem et al. (2011) Mexico & N LO J,S,A 61 26 this study C. America Mexico & E LO J,S,A 108 52 this study C. America Galápagos E CM - 1 - Rathbun (1902) via Chace (1951) Peru E CC J 28 93 this study Peru N,E CM J,S,A 416 - Brown and Brown (1995) Peru E CM J,S 15 60 this study Peru N CM J,S,A 99 1 this study Peru N EI J 1 0 this study Peru - LO - 1 - Schweigger (1964) Peru E LO A 2 100 this study Chile D LO - 5 20 Miranda and Moreno (2002)

Notes. Frequency of epibiosis (F0), epipelagic (E), neritic (N), nesting beach (B), stranding (D), Caretta caretta (CC), Chelonia mydas (CM), Dermochelys coriacea (DC), Eretmochelys imbricata (EI), Lepidochelys olivacea (LO), unknown turtle species (UNK), juvenile (J), subadult (S), adult (A). Dashes indicate insufficient data for determining stage, habitat, or F0.

146

Figure 5-1. Maps showing sites, turtle-capture locations, and presence (open icons) or absence (closed icons) of Planes major. A) Pacific Ocean showing the five sampling areas in the present study. B) The island of Shikoku, Japan, with inset showing Cape Muroto and the locations of three pound nets (stars) where neritic turtles were captured. C) Hawaii, with epipelagic turtle-capture locations by the Hawaiian longline fisheries and turtle strandings on Oahu. D) Samoa, with epipelagic turtle-capture locations by Samoan longline fisheries. E) Mexico and Central America, with neritic and epipelagic turtle-capture locations along coast and offshore 1,800 km and inset showing turtle-capture locations off the West coast of Baja California Sur, Mexico. F) Peru, with epipelagic turtle-capture locations along the coast and inset showing turtle- capture locations within Sechura Bay. Circles, Caretta caretta; triangles, Lepidochelys olivacea; diamonds, Chelonia mydas; squares, Eretmochelys imbricata. Maps were created using seaturtle.org Maptool.

147

Figure 5-2. Size-frequency histograms showing loggerhead turtles (Caretta caretta) that did not host Planes major (black bars) and turtles that hosted at least one Pl. major (white bars) from A) neritic habitats in Japan, B) neritic habitats in Mexico, and C) epipelagic habitats in Peru. Turtles on the border of two size increments were placed in the larger increment. Asterisks indicate when a significant effect of turtle size on crab presence was detected (alpha = 0.05).

148

Figure 5-3. Size-frequency histograms showing green turtles (Chelonia mydas) that did not host Planes major (black bars) and turtles that hosted at least one Pl. major (white bars) from A) neritic habitats in Japan, B) epipelagic habitats near Samoa, C) neritic habitats in Peru, and D) epipelagic habitats off Peru. Turtles on the border of two size increments were placed in the larger increment.

149

Figure 5-4. Size-frequency histograms showing olive ridley turtles (Lepidochelys olivacea) that did not host Planes major (black bars) and turtles that hosted at least one Pl. major (white bars) from A) epipelagic habitats near Hawaii, B) neritic habitats along Mexico and Central America and C) epipelagic habitats along Mexico and Central America. Turtles on the border of two size increments were placed in the larger increment

150

CHAPTER 6 CONCLUSIONS AND FUTURE DIRECTIONS

High Seas Hitchhiker

Understanding global patterns of species diversification and population differentiation in oceanic crabs has added to the small, but growing, literature on this topic for planktonic and neustronic animals. These animals include some of the most numerous and valuable species in the ocean (Goetze 2003), and understanding their species diversity and population connectivity is fundamental to their conservation and the conservation of marine ecosystems. To date, we know surprisingly little with respect to these fundamental aspects of these important animals because of their inaccessibility in the open ocean and because past researchers have been constrained by tools that provided insufficient biological resolution.

Marine organisms with large populations and high dispersal ability have posed specific challenges for those interested in estimating population connectivity and structure. Because tracking dispersal of individuals within the vast ocean is often logistically impossible, genetic data have become the primary tool by which researchers assess population connectivity at large spatial scales (Hedgecock et al. 2007; Lowe and

(Hedgecock et al. 2007). When population connectivity is high and populations are large (e.g., in planktonic and neustonic animals), it becomes difficult to accurately estimate contemporary population structure with limited genetic resolution offered by traditional genotypic markers (Goetze 2005; McCormack et al. 2013; Benestan et al.

151

2015). For this reason, our understanding of the population biology and connectivity in these animals is not only limited, but might also be somewhat biased. Recent advances in high-throughput sequencing technologies have allowed genome-wide genetic variation to be incorporated into population genetic analyses of non-model organisms

(Reitzel et al. 2013). Our study is the first to use this technology in a planktonic or neustonic organism, and demonstrates the value and future promise of such tools for estimating contemporary population structure in organisms with large, highly connected populations. Studies that utilize new high-throughput sequencing technologies on other planktonic and neustonic animals will gain an unprecedented level of biological resolution and allow future researchers to test previously unanswerable questions in regards to diversification and differentiation in the open ocean.

Although oceanic crabs do not hold the ecological importance of many planktonic species (Goetze 2003), understanding their global genetic patterns can reveal how theoretical dispersal potential actually relates to population differentiation and diversification at a global scale. In Chapter 2, my coauthors and I tested the hypotheses that because Planes can disperse widely as planktonic larvae and as adults associated with oceanic flotsam and sea turtles, we should find (1) low species diversity within this group and (2) weak geographic differentiation among widely separated populations. These hypotheses were based on the premise that the ability to disperse widely will tend to homogenize distant populations via high connectivity (Hedgecock et al. 2007; Lowe and Allendorf 2010), leading to reduced opportunities for genetic segregation and ultimately speciation. We found evidence in our mtDNA and genomic datasets to support both predictions. However, as this is the first attempt to resolve

152

genetic patterns within this group, these analyses evoke many new questions that require continued research.

We found low species diversity among rafting Planes and intertidal

Pachygrapsus laevimanus. Our data confirm the paraphyly of the genus Planes due to the well-supported inclusion of Pa. laevimanus (as suggested by Schubart 2011; Ip et al. 2015), but also suggest that species diversity within this clade may actually be lower than previously thought (Ng et al. 2008). Our results suggest that there are three closely-related (i.e., recently-diverged) species and a zone of secondary genetic exchange (i.e., hybridization) in the North Atlantic Ocean. Intertidal Pa. laevimanus and rafting Planes marinus are likely two recently and rapidly diverged species, and Planes minutus and Planes major are likely a single, globally-distributed, obligate-rafting species. Nevertheless, more work is needed to (1) understand whether Pa. laevimanus and Pl. marinus from the same geographic area (southwest Pacific) are distinct and evaluate the factors associated with changes in morphology between species/habitats,

(2) evaluate potential factors (e.g., population genetics, geography, substrata/host, diet, etc.) related to subtle morphological differences within Pl. minutus/Pl. major, and (3) identify morphological, physiological or behavioral traits between Pl. marinus and Pl. minutus/Pl. major that both facilitate hybridization in the North Atlantic and deter hybridization across the rest of the temperate and tropical oceans of the world.

We found evidence of weak differentiation among globally distributed populations for Pl. minutus/Pl. major. Population differentiation occurred primarily at the scale of major ocean basins and we found almost no evidence for differentiation within distinct ocean gyres. Clustering patterns and pairwise genetic distances highlight probable

153

barriers (Cape Horn, Indonesian Archipelago and Atlantic equatorial currents) and corridors (Cape of Good Hope, Strait of Gibraltar and major ocean gyres) to rafting dispersal at a global scale. Although we did not find support for global panmixis, pairwise genetic distances among clusters and regions were consistently very low (FST

= -0.03-0.23). At this global scale, such weak differentiation among Planes populations suggests that continuous large-scale dispersal may sustain their global distribution and inhibit biogeographic differentiation. While Planes populations do show genetic discontinuities based on geographic boundaries, further analyses testing the likelihood of various biogeographic hypotheses (e.g., global, ocean or gyre panmixis) are needed to evaluate support for different patterns. Moreover, the role that sea turtles play as a dispersal vector across potential biogeographic boundaries requires further testing and likely a more balanced sampling design that includes samples from both turtles and flotsam in areas adjacent to the boundaries.

Our results are consistent with theoretical predications, in which the vast dispersal potential of Planes appears to have stymied local adaptation and limited diversification within the group, and led to limited population differentiation among widely separated aggregations worldwide. However, there appears to be no ubiquitous pattern for diversification or differentiation in the open ocean and only some patterns are consistent with theoretical expectations based on the dispersal potential and persistent biogeographic boundaries (Anderson et al. 2000; Goetze 2003; Cabezas et al. 2013;

Churchill et al. 2013, 2014; Hirai et al. 2015). It is clear that while the capacity for long- distance dispersal likely plays an important role in limiting opportunities diversification

(as well as extinction) and differentiation, the mechanisms leading to speciation and

154

population structure in the open ocean are far more complex (Palumbi 1994, 2003;

Knowlton 2000). Future studies that utilize high throughput sequencing technologies will provide the best opportunity to elucidate common patterns.

Monogamous Symbiont

Symbiotic crustaceans offer an opportunity to study how various ecological factors contribute to the adoption of different mating strategies, a topic that captures the attention of biologist and non-biologists alike. Because symbiotic crustaceans live in or on distinct host species, the ecological factors that dictate mating strategies are defined in large part by the morphology and ecology of their hosts (Thiel and Baeza 2001).

Baeza and Thiel (2007) outline a general framework for understanding how host characteristics and ecology affect the mating system and social behavior of symbiotic crustaceans. Under this theoretical framework, reproductive strategies of symbiotic crustaceans can be predicted based on four parameters: (1) host body size relative to symbiont body size, (2) host structural complexity, (3) host abundance, and (4) the risk of mortality for symbionts away from hosts. In particular, social monogamy should be favored when hosts are relatively small in body size and structurally simple, and when hosts have relatively low abundance in habitats where the risk of mortality for symbionts away from hosts is high. Moreover, we argue that similar factors affect the reproductive biology and social behavior of any refuge-dwelling animals and therefore these predictions can be applied more broadly.

In Chapter 3, my coauthors and I tested this hypothesis in Planes major and its facultative association with loggerhead sea turtles (Caretta caretta). We predicted that

Pl. major would display social monogamy and long-term heterosexual pairing on turtles because (1) the supracaudal/inguinal spaces on turtles are defendable resources

155

(functionally small and structurally simple refuges), (2) turtles tend to be sparsely distributed in the marine environment, even in foraging ‘hotspots’ (0.58-0.75 turtles km-2

– Seminoff et al. 2014), and (3) turtles offer safe refuges in habitats where the mortality risk for crabs away from host turtles is likely high (e.g., from predation, limited swimming endurance, low substrata availability – Davenport 1992; Hamner 1995; Shanks 1983).

Our results supported the hypothesis that Pl. major is socially monogamous on turtles, which was consistent with theoretical expectations (Baeza and Thiel 2007) and with past observations of Pl. minutus associated with turtles (Dellinger et al. 1997).

However, our results did not support the hypothesis that social monogamy in Pl. major is always long-term. Instead, our results suggest that the duration of social monogamy in Pl. major is likely variable and may involve some degree of host switching and intra- sexual (mostly male-male) competition. Because our understanding of how host traits

(i.e., relative body size, morphology, and abundance) influence the reproductive strategies of symbiotic crustaceans comes primarily from studies of symbionts living in or on benthic macro-invertebrates in tropical habitats, our study represented a novel test of theory in a markedly different habitat and type of host. Our results suggest that the line of questioning frequently employed when evaluating the mating strategies of symbiotic crustaceans (Baeza 2008; Baeza et al. 2011, 2013; Peiró et al. 2012) needs to be expanded to account for a greater diversity of symbiotic interactions (e.g., vertebrate-host, crustacean-symbiont symbioses and facultative associations).

In Chapter 4, my coauthor and I tested the hypothesis that Planes display social monogamy on turtles because the refuge area on turtles (i.e., the supracaudal space) is a valuable and limited resource that can be monopolized by small numbers of

156

individuals. We predicted that refuge area would be a better predictor of adult crab number than total area for both flotsam and turtles, and that flotsam and turtles with similar refuge area would host a similar number and composition of the adult crabs (i.e., male-female pairs). We found support for both predictions, which suggests that refuge area, not total area, dictates group size and composition in this system and that sea turtle symbiosis likely promotes social monogamy in Planes crabs due to energetic constraints imposed by refuge size. Evaluating these predictions not only informed our understanding for how resource characteristics structure animal groups, but also how symbiosis can facilitate the adoption of specific mating strategies. We proposed that facultative symbioses in which there is intraspecific variation in mating strategies among different substrata, but living hosts tend to facilitate just one strategy (as in Planes), provide a mechanism for understanding the evolution of obligate symbioses in which symbionts display specific mating strategies (Baeza and Thiel 2007)

Much like the enigmatic lives of sea turtles (Carr 1967), the oceanic proclivity of

Planes crabs makes understanding their basic natural history characteristics elusive.

To date, the mating strategies, social behavior and movement patterns of Planes have primarily been deduced indirectly from data on grouping patterns and assumptions based on morphology. Laboratory-based studies in which crab groupings can be manipulated and observed would be particularly informative with respect to agonistic and competitive interactions, mate and host/substrata fidelity, and the structuring of monogamous pairs (whether by mate guarding or territorial cooperation; Mathews

2002). Moreover, genetic studies that determine the number of males contributing to egg masses of crabs on turtles and flotsam (i.e., parentage analysis) would shed light

157

on habitat-specific differences in reproductive tactics and reveal whether social monogamy also entails reproductive/genetic monogamy. Lastly, a more rigorous and systematic (i.e., less haphazard) approach to studying these crabs in the wild, especially on flotsam, would go a long way towards revealing their biological mysteries.

Oceanic Indicator

Studies that incorporate information from habitat-specific ecological interactions

(e.g., with predators, prey, parasites, and commensals) can reveal valuable insights into the cryptic patterns of habitat use and behavior of marine vertebrates and inform their conservation. In Chapter 5, my coauthors and I used commensal Planes crabs as indicators of recent epipelagic habitat use and surface-dwelling behavior to gain insights into the biology of three species of sea turtle in the Pacific Ocean. Our results suggest that loggerhead and olive ridley turtles display variable/flexible epipelagic-neritic transitions, while green turtles tend to transition unidirectional at small body sizes.

Moreover, our results suggest that epipelagic loggerheads tend to spend more time at or near the surface than epipelagic olive ridley and green turtles. These results supported previous studies based on data from more advanced technologies (e.g., satellite tracking, time-depth profiling and stable isotope analysis), but also suggested previously unknown patterns.

Sea turtles use epipelagic habitats for juvenile development (Musick and Limpus

1997). In general, leatherback and olive ridley turtles tend to remain epipelagic throughout adulthood (Plotkin 2010; Saba 2013), while loggerhead, green and hawksbill turtles tend to transition to neritic habitats as juveniles and remain neritic as adults

(Musick and Limpus 1997). Recent studies using satellite telemetry and stable isotope analyses indicate that alternative patterns of habitat use exist. Large juvenile and adult

158

loggerheads in the North Pacific Ocean use both epipelagic and neritic habitats, and are thought to either transition unidirectional to neritic habitats or maintain persistent alternative foraging strategies (Hatase et al. 2010; Ishihara et al. 2011; Peckham et al.

2011). However, our data suggest that habitat-use patterns in loggerheads are not so clear-cut and are likely more variable and flexible than previously thought. Our data from neritic green turtles support the idea that green turtles transition to neritic habitats unidirectionally at small body sizes and remain there throughout development (Musick and Limpus 1997). However, we found green turtles in epipelagic habitats off the coast of Peru that hosted crabs frequently, suggesting the possibility of persistent alternative foraging strategies in this area. Although satellite-tracking data suggest that olive ridleys are highly epipelagic and nomadic (Work and Balazs 2010; Plotkin 2010), our data are consistent with dietary analyses (Koptisky et al. 2005) that suggest that some adult olive ridleys (>60 cm CCL) make forays into neritic habitats. Understanding the habitat-use patterns in sea turtle populations are critical for understanding the factors that drive their life history characteristics, and ultimately their productivity and resiliency, which are paramount to their long-term conservation.

Diving behavior in sea turtles is difficult to study and datasets are often limited to small number of animals. Using Planes as an indicator of differences in surface- dwelling behavior was a time- and cost-effective method to infer more population-wide patterns. Our results were consistent with differences in dive profiles and dietary analyses of epipelagic loggerhead, olive ridley and green turtles (Hays et al. 2001;

Polovina et al. 2003, 2004; Koptisky et al. 2005; Rice and Balazs 2008). Understanding behavioral differences among turtle species in epipelagic habitats is important for

159

identifying species-specific conservation measures to mitigate turtle mortality in pelagic fisheries.

In Chapter 5, my coauthors and I applied elements of the Conceptual Model of

Epibiosis (Frick and Pfaller 2013) to learn valuable insights into the cryptic patterns of habitat use and behavior of sea turtles. As part of my contribution to a book chapter entitled Sea Turtle Epibiosis published in The Biology of Sea Turtles, Volume III

(Wyneken et al. 2013), I developed a conceptual framework to better understand and learn from epibiotic interactions. The conceptual model outlines three hierarchical factors inherent to epibiotic interactions: (1) geographic overlap (Figure 6-1A), (2) ecological or habitat overlap (Figure 6-1B), and (3) the balance of costs and benefits to hosts and epibionts that dictate the likelihood of epibiosis once in close proximity

(Figure 6-1C). Epibiosis first necessitates spatial overlap between the geographic ranges and habitats of the hosts and ‘free-living’ populations of potential epibionts.

Because sea turtles and other marine vertebrates often use different geographic regions and habitat types during different behaviors or life stages, the epibionts associated with a given host should reflect the characteristic assemblage of plants and animals that occupy the regions and habitats where the host spends time. This information can reveal cryptic habitat-use patterns and behaviors of the hosts. Once in close proximity, there is a complex balance of costs and benefits for host turtles and potential epibionts that ultimately determine the likelihood of epibiosis. Figure 6-1C displays a 2D likelihood surface in which each axis represents a continuum from high benefit to high cost. The various positions of different hosts and epibionts along these cost-benefit axes depend on the net cost or benefit experienced during epibiosis. Because the

160

relative costs and benefits are different for different turtle-epibiont pairs, some associations are more likely and therefore more frequent [quadrant I (potential mutualism); top left of quadrants II and IV (commensalism)] and other are less likely and therefore less frequent (quadrant III; bottom left of quadrants II and IV (potential parasitism)]. Because the factors that affect epibiotic interactions—as displayed in this conceptual model—are inherent to the biology of the species involved, we can learn about the ecology and evolution of these species by studying epibiosis. This conceptual framework will hopefully allow researchers to better understand the factors that affect their particular epibiotic systems and more easily decipher the important biological information that can be gleaned from studying epibiosis in sea turtles.

161

Figure 6-1. Conceptual model of epibiosis (reproduced from Frick and Pfaller 2013). A,B) Venn diagrams showing the geographic and ecological overlap between hosts and epibionts, respectively. C) Two-dimensional surface showing the likelihood of epibiosis based on the balance of costs and benefits to hosts and epibionts (Bepibiont, benefit to the epibiont; Bhost, benefit to the host; Chost, cost to the host; Cepibiont, cost to the epibiont).

162

LIST OF REFERENCES

Adams J, Edwards AJ, Emberton H (1985) Sexual size dimorphism and assortative mating in the obligate coral commensal Trapezia ferruginea Latreille (Deacpoda, Xanthidae). Crustaceana 48:188–194

Alfaro-Shigueto J, Mangel JC, Bernedo F, Dutton PH, Seminoff JA, Godley BJ (2011) Small-scale fisheries of Peru: a major sink for marine turtles in the Pacific. J Appl Ecol 48:1432–1440

Anderson NM, Cheng L, Damgaard J, Sperling FAH (2000) Mitochondrial DNA sequences variation and phylogeography of oceanic insects (Hemiptera: Gerridae: Halobates spp.). Mar Biol 136:421–430

Anger K (1995) The conquest of freshwater and land by marine crabs: adaptations in life-history patterns and larval bioenergetics. J Exp Mar Biol Ecol 193:119–145

Angulo-Lozano L, Nava-Duran PE, Frick MG (2007) Epibionts of olive ridley turtles nesting at Playa Ceuta, Sinaloa, Mexico. Marine Turtle Newsletter 118:13–14

Asakura A (2009) The evolution of mating systems in decapod crustaceans. In: Martin JW, Crandall KA, Felder DL (eds) Decapod Crustacean Phylogenetics. CRC Press, Boca Raton, pp 121–182

Aucoin S, Himmelman JH (2010) A first report on the shrimp Pontonia sp. and other potential symbionts in the mantle cavity of the penshell Pinna carnea in the Dominican Republic. Symbiosis 50:135–141

Avise JC (2000) Phylogeography: the history and formation of species. Harvard University Press, Massachusetts

Badillo FJ, Aznar FJ, Tomás J, Raga JA (2001) Epibiont fauna of Caretta caretta in the Spanish Mediterranean. Proc First Mediterranean Conference on Marine Turtles. Rome.

Baeza JA (1999) Indicators of monogamy in the commensal crab Pinnixa tranversalis (Milne Edwards & Lucas) (Decapoda: Brachyura: Pinnotheridae): population distribution, male-female association and sexual dimorphism. Rev Biol Mar Oceanogr 34:303–313

Baeza JA (2008) Social monogamy in the shrimp Pontonia margarita, a symbiont of Pinctada mazatlantica, in the topical eastern Pacific coast. Mar Biol 153:387–395

Baeza JA (2010) The symbiotic lifestyle and its evolutionary consequences: social monogamy and sex allocation in the hermaphroditic shrimp Lysmata pederseni. Naturwissenschaften 97:729–741

163

Baeza JA, Bolaños JA, Hernandez JE, Lira C, López R (2011) Monogamy does not last long in Pontonia mexicana, a symbiotic shrimp of the amber pen-shell Pinnea carnea from the southeastern Caribbean Sea. J Exp Mar Biol Ecol 407:41–47

Baeza JA, Farías NE, Luppi TA, Spivak ED (2010) Refuge size, group living and symbiosis: testing the “resource economic monopolization” hypothesis with the shrimp Betaeus lilianae and description of its partnership with the crab Platyxanthus crenulatus. J Exp Mar Biol Ecol 389:85–92

Baeza JA, Ritson-Williams R, Fuentes MS (2013) Sexual and mating system in a caridean shrimp symbiotic with the winged pearl oyster in the Coral Triangle. J Zool 289:172–181

Baeza JA, Thiel M (2000) Host use pattern and life history of Liopetrolisthes mitra, an associate of the black sea urchin Tetrapygus niger. J Mar Biol Assoc UK 80:39– 45

Baeza JA, Thiel M (2003) Predicting territorial behavior in symbiotic crabs using host characteristics: a comparative study and proposal of a model. Mar Biol 142:93– 100

Baeza JA, Thiel M (2007) The mating system of symbiotic crustaceans. A conceptual model based on optimality and ecological constraints. In: Duffy JE, Thiel M (eds) Evolutionary ecology of social and sexual systems: crustaceans as model organisms. Oxford University Press, Oxford, pp 249–267

Barceló C, Peckham SH, Marinovic B (2008) What do hitchhikers eat? The diet of Planes cyaneus and their association with loggerhead and olive ridley turtles off the Pacific coast of Baja California Sur, Mexico. NOAA Tech Memo NMFS- SEFSC-569, p 118

Benestan L, Gosselin T, Perrier C, Sainte-Marie B, Rochette R, Bernatchez L (2015) RAD genotyping reveals fine-scale genetic structuring and provides powerful population assignment in a widely distributed marine species, the American lobster (Homarus americanus). Mol Ecol 24:3299–3315

Bjorndal KA, Bolten AB, Dellinger T, Delgado C, Martins HR (2003) Compensatory growth in oceanic loggerhead sea turtles: response to a stochastic environment. Ecology 84:1237–1249

Bolten AB (2003) Variation in sea turtle life history patterns: neritic vs. oceanic developmental stages. In: Lutz PL, Musick JA, Wyneken J (eds) The biology of the sea turtles, volume II. CRC Press, Boca Raton, pp 243–257

Bourjea J, Lapégue S, Gagnevin L, Broderick D, Mortimer JA, Ciccione S, Roos D, Taquet C, Grizel H (2007) Phylogeography of the green turtle, Chelonia mydas, in the Southwest Indian Ocean. Mol Ecol 16:175-186

164

Bradbury IR, Laurel B, Snelgrove PVR, Bentzen P, Campana SE (2008) Global patterns in marine dispersal estimates: the influence of geography, taxonomic category and life history. Proc R Soc Lond B 275:1803–1809

Brown CH, Brown WM (1995) Status of sea turtles in the southeastern Pacific: emphasis on Perú. In: Bjorndal KA (ed) Biology and conservation of sea turtles. Smithsonian Institution Press, Washington, DC, pp 235–240

Browne P (1789) The civil and natural history of Jamaica. London.

Bucklin A, Astthorsson OS, Gislason A, Allen LD, Smolenack SB, Wiebe PH (2000) Population genetic variation in Calanus finmarchicus in Icelandic waters: preliminary evidence of genetic differentiation between Atlantic and Arctic populations. ICES J Mar Sci 57:1592–1604

Bucklin A, Kocher TD (1996) Source regions for recruitment of Calanus finmarchicus to Georges Bank: evidence for molecular population genetic analysis of mtDNA. Deep-Sea Res 43:1665–1681

Bucklin A, La Jeunesse TC, Curry E, Wallinga J, Garrison K (1996) Molecular diversity of the copepod, Nannocalanus minor: genetic evidence of species and population structure in the North Atlantic Ocean. J Mar Res 54:285–310

Bugoni L, Fiedler FN, Monteiro DS, Estima SC (2007) Eretmochelys imbricata (hawksbill), Chelonia mydas (green), and Caretta caretta (loggerhead) Seaturtle. Epizoans. Herpetol Rev 38:199–200

Cabezas MP, Navarro-Barranco C, Ros M, Guerra-García JM (2013) Long-distance dispersal, low connectivity and molecular evidence of a new cryptic species in the obligate rafter Caprella andreae Mayer, 1890 (Crustacea: Amphipoda: Caprellidae). Helgol Mar Res 67:483–497

Caine EA (1986) Carapace epibionts of nesting loggerhead sea turtles: Atlantic coast of USA. J Exp Mar Biol Ecol 95:15–26

Carr AF (1967) So excellent a fishe. The Natural History Press, New York

Carranza A, Domingo A, Verdi A, Forselledo R, Estrades A (2003) First report of an association between Planes cyaneus (Decapoda: Grapsidae) and loggerhead sea turtles in the southwestern Atlantic Ocean. Marine Turtle Newsletter 102:5–7

Casale P, D’Addario M, Freggi D, Argano R (2012) Barnalces (Cirripedia, Thoracica) and associated epibionts from sea turtles in the central Mediterranean. Crustaceana 85:533–549

Casale P, Freggi D, Basso R, Argano R (2004) Epibiotic barnacles and crabs as indicators of Caretta caretta distribution and movements in the Mediterranean Sea. J Mar Biol Assoc UK 84:1005–1006

165

Cassone BJ, Boulding EG (2006) Genetic structure and phylogeography of the lined shore crab, Pachygrapsus crassipes, along the northeastern and western Pacific coasts. Mar Biol 149:213–226

Catchen J, Amores A, Hohenlohe P, Cresko W, Postlethwait J (2011) Stacks: building and genotyping loci de novo from short-read sequences. G3: Genes, Genomes, Genetics 1:171–182

Catchen J, Hohenlohe P, Bassham S, Amores A, Cresko W (2013) Stacks: an analysis tool set for population genomics. Mol Ecol 22:3124–3140

Chace FA (1951) The oceanic crabs of the genera Planes and Pachygrapsus. Proc US Natl Mus 101:65–103

Chace FA (1966) Decapod crustaceans from St, Helena Island, South Atlantic. Smithsonian Press, Washington DC

Chevreux E, de Guerne J (1893) Crustacés et cirrhipèpes commensaux des tortues marines de la Méditerranée. Comtes Rendus Academy of Science, Paris 116:443–446

Churchill CKC, Alejandrino A, Valdés A, Foighil DÓ (2013) Parallel changes in genital morphology delineate cryptic diversification of planktonic nudibranchs. Proc R Soc Lond B 280:20131224

Churchill CKC, Valdés A, Foighil DÓ (2013) Afro-Eurasia and the Americas present barriers to gene flow for the cosmopolitan neustonic nudibranchs Glaucus atlanticus. Mar Biol 161:899–910

Couch J Bate CS (1878) Crustacea: a Cornish fauna. J Royal Institute of Cornwall 19:451–522

Cowen RK, Lwiza KMM, Spongaugle S, Paris CB, Olson DB (2000) Connectivity of marine populations: open or closed? Science 287:857–859

Cowen RK, Sponaugle S (2009) Larval dispersal and marine population connectivity. Annu Rev Mar Sci 1:443–466

Coyne JA, Orr HA (2004) Speciation. Sinauer Associates, Inc., Connecticut

Crane J (1937) The Templeton Crocker Expedition, III: Brachygnathous crabs from the Gulf of California and the west coast of Lower California. Zoologica 22:47–48

Cuesta JA, González-Gordillo JI, Rodríguez A (1997) First zoeal stages of Grapsus adscensionis (Osbeck) and Planes minutus (Linnaeus) (Brachyura: Grapsidae) described from laboratory hatched material, with notes on larval characters of the Grapsinae. J Nat Hist 31:887–900

166

Cuesta JA, Guerao G, Schubart CD, Anger K (2011) Morphology and growth of the larval stages of Geograpsus lividus (Crustacea, Brachyura), with the descriptions of new larval characters for the Grapsidae and an undescribed setation pattern in extended developments. Acta Zool-Stockholm 92:225–240

Cuesta JA, Schubart CD (1999) First zoeal stages of Geograpsus lividus and Goniopsis pulchra from Panama confirm consistent larval characters for the subfamily Grapsinae (Crustacea: Brachyura: Grapsidae). Ophelia 51:163–176

Damgaard J, Anderson NM, Cheng L, Sperling FAH (2000) Phylogeny of sea skaters, Halobates Eschscholtz (Hemiptera, Gerridae), based of my DNA sequence and morphology. Zool J Linn Soc 130:511–526

Davenport J (1992) Observations on the ecology, behavior, swimming mechanism and energetics of the neustonic grapsid crab, Planes minutus. J Mar Biol Assoc UK 77:611–620

Davenport J (1994) A cleaning association between the oceanic crab Planes minutus and the loggerhead sea turtle Caretta caretta. J Mar Biol Assoc UK 74:735–737

Darwin C (1859) On the origin of species by natural selection. J. Murray, London

De Bruyn C, Rigaud T, David B, De Ridder C (2009) Symbiosis between the pea crab Dissodactylus primitivus and its echinoid host Meoma ventricosa: potential consequences for the crab mating system. Mar Ecol Prog Ser 375:173–183:

Deejai P, Assawamakin A, Wangkumhang P, Poomputsa K, Tongsima S (2010) On assigning individuals from cryptic population structures to optimal predicted subpopulations: an empirical evaluation of non-parametric population structure analysis techniques. Computational Systems- Biology and Bioinformatics, Communications in Computer and Information Science 115:58–70

Dell RK (1963) Pachygrapsus marinus (Rathbun), a new crab for New Zealand waters. J Roy Soc New Zeal 3:179–180

Dellinger T, Davenport J, Wirtz P (1997) Comparisons of social structure of Columbus crabs living on loggerhead sea turtles and inanimate flotsam. J Mar Assoc UK 77:185–194

Dent EA, vonHolt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv Gen Res 4:359–361

Diesel R (1986) Optimal mate searching strategy in the symbiotic spider crab Inachus phalangium (Decapoda). Ethology 72:311–328

Diesel R (1988) Male-female association in the spider crab Inachus phalangium: the influence of female reproductive stage and size. J Crustacean Biol 8:63–69

167

Dodd CK Jr (1988) Synopsis of the biological data on the Loggerhead Sea Turtle Caretta caretta (Linnaeus 1758). US Fish Wildl Serv, Biol Rep 88

Domènech F, Badillo FJ, Tomás J, Raga JA, Aznar FJ (2014) Epibiont communities of loggerhead turtles (Caretta caretta) in the western Mediterranean: influence of geographic and ecological factors. J Mar Biol Assoc UK 95:851-861

Donlan CJ, Nelson PA (2003) Observations of invertebrate colonized flotsam in the eastern tropical Pacific, with a discussion of rafting. B Mar Sci 72:231–240

Eckert KL, Eckert SA (1988) Pre-reproductive movements of leatherback sea turtles (Dermochelys coriacea) nesting in the Caribbean. Copeia 1988:400–406

Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32:1792–1797

Emlen ST, Oring LW (1977) Ecology, sexual selection and the evolution of mating systems. Science 197:215–223

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

Evans N, Paulay G (2012) Barcoding methods for invertebrates. Methods Mol Biol 858:47–77

Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5; a new series of programs to perform population genetic analyses under Linux and Windows. Mol Ecol Res 10:564–567

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

Faurby S, Barber PH (2012) Theoretical limits to the correlation between pelagic larval duration and population genetic structure. Mol Ecol 21:3419–3432

Faxon W (1895) Reports on an exploration off the west coasts of Mexico, Central and South America, and off the Galapagos Islands, in charge of Alexander Agassiz, by the U.S. Fish Commission steamer, Albatross, during 1891, Lieut-Commander Z.L. Tanner, U.S.N., commanding, 15: The stalk-eyed Crustacea. Mem Mus Comp Zool 18:1–292

Finlay BJ (2002) Global dispersal of free-living microbial eukaryote species. Science 296:1061–1063

Flores AAV, Paula J (2002) Population dynamics of the shore crab Pachygrapsus marmoratus (Brachyura: Grapsidae) in the central Portuguese coast. J Mar Biol Assoc UK 82:229–241

168

Frick MG, Kopitsky K, Bolten AB, Bjorndal KA, Martins HR (2011) Sympatry in grapsoid crabs (genera Planes and Plagusia) from olive ridley sea turtles (Lepidochelys olivacea), with descriptions of crab diets and masticatory structures. Mar Biol 158:1699–1708

Frick MG, Pfaller JB (2013) Sea turtle epibiosis. In: Wyneken J, Lohmann KJ, Musick JA (eds) The biology of sea turtles, volume III. CRC Press, Boca Raton, pp 399–426

Frick MG, Ross A, Williams KL, Bolten AB, Bjorndal KA, Martins HR (2003) Epibiotic associates of oceanic-stage loggerhead turtles from the southeastern North Atlantic. Marine Turtle Newsletter 101:18–20

Frick MG, Williams KL, Bresette M, Singewald DA, Herren RM (2006) On the occurrence of Columbus crabs (Planes minutus) from loggerhead turtles in Florida, USA. Marine Turtle Newsletter 114:12–14

Frick MG, Williams KL, Bolten AB, Bjorndal KA, Martins HR (2004) Diet and fecundity of Columbus crabs, Planes minutus, associated with oceanic-stage loggerhead sea turtles, Caretta caretta, and inanimate flotsam. J Crustacean Biol 24:350–355

Frick MG, Williams KL, Robinson M (1998) Epibionts associated with nesting loggerhead sea turtles (Caretta caretta) in Georgia, USA. Herpetol Rev 29:211– 214

Frick MG, Williams KL, Veljacic D (2000) Additional evidence supporting a cleaning association between epibiotic crabs and sea turtles: How will the harvest of Sargassum seaweed impact this relationship? Marine Turtle Newsletter 20:11–13

Gao X, Starmer JD (2008) AWclust: point-and-click software for non-parametric population structure analysis. BMC Bioinformatics 9:77

Gil MA, Pfaller JB (2016) Oceanic barnacles act as foundation species on plastic debris: implications for marine dispersal. Scientific Reports 6:19987

Goetze E (2003) Cryptic speciation on the high seas: global phylogenetics of the copepod family Eucalanidae. Proc R Soc Lond B 270:2321–2331

Goetze E (2005) Global population genetic structure and biogeography of the oceanic copepods Eucalanus hyalinus and E. spinifer. Evolution 59:2378–2398

Gordon A, Hannon GJ (2010) Fastx-toolkit. FASTQ/A short-reads pre-processing tools. Unpublished http://hannonlab. cshl. edu/fastx_ toolkit

Gouy M, Guindon S, Gascuel O (2010) SeaView version 4: a multiplatform graphical user interface for sequence alignment and phylogenetic tree building. Mol Biol Evol 27:221–224

169

Graham CF, Glenn TC, McArthur AG, Boreham DR, Kieran T, Lance S, Manzon RG, Martino JA, Pierson T, Rogers SM, Wilson JY, Somers CM (2015) Impacts of degraded DNA on restriction enzyme associated DNA sequencing (RADseq). Mol Ecol Res 15:1304–1315

Graham RT, Witt MJ, Castellanos DW, Remolina F, Maxwell S, Godley BJ, Hawkes LA (2012) Satellite tracking of manta rays highlights challenges to their conservation. PLoS ONE 7:e36834

Gramentz D (1988) Prevalent epibiont sites on Caretta caretta in the Mediterranean Sea. Naturalista Siciliano 12:33–46

Guerao G, Schubart CD, Cuesta JA (2001) The first zoeal stages of Grapsus grapsus (Linnaeus) and Geograpsus lividus (Milne Edwards) (Decapoda, Brachyura, Grapsidae) from the western Atlantic. Nauplius 9:111–121

Guess RC (1981) A pacific loggerhead captured off California’s northern Channel Islands. Herpetol Rev 12:15

Hamner WM (1995) Predation, cover, and convergent evolution in epipelagic oceans. Mar Fresh Behav Physiol 26:71–89

Hartnoll RG (1963) The biology of manx spider crabs. Proc Zool Soc Lond 141:423–496

Hartnoll RG (1971) The occurrence, methods and significance of swimming in the Brachyura. Anim Behav 19:34–50

Hartnoll RG (1978) The determination of relative growth in Crustacea. Crustaceana 34:281–292

Hartnoll RG (1982) Growth. In: Abele LG (ed) The biology of crustacea 2. Academic Press, New York, pp 111–196

Hatase H, Omuta K, Tsukamoto K (2010) Oceanic residents, neritic migrants: a possible mechanism underlying foraging dichotomy in adult female loggerhead turtles (Caretta caretta). Mar Biol 157:1337–1342

Hatase H, Sato K, Yamaguchi M, Takahashi K, Tsukamoto K (2006) Individual variation in feeding habitat use by adult female green sea turtles (Chelonia mydas): are they obligately neritic herbivores? Oecologia 146:52–64

Hatase H, Takai N, Matsuzawa Y, Sakamoto W, Omuta K, Goto K, Arai N, Fujiwara T (2002) Size-related differences in feeding habitat use of adult female loggerhead turtles Caretta caretta around Japan determined by stable isotope analyses and satellite telemetry. Mar Ecol Prog Ser 233:273–281

170

Hays GC, Åkesson S, Broderick AC, Glen F, Godley BJ, Luschi DJ, Martin C, Metcalfe JD, Papi F (2001) The diving behavior of green turtles undertaking oceanic migration to and from Ascension Island: dive durations, dive profiles and depth distribution. J Exp Biol 204:4093–4098

Hedgecock D, Barber PH, Edmands S (2007) Genetic approaches to measuring connectivity. Oceanography 20:70–79

Heppell SS, Crowder LB, Menzel TR (1999) Life table analysis of long-lived marine species with implications for conservation and management. In: Musick JA (ed) Life in the slow lane: ecology and conservation of long-lived marine animals. American Fisheries Society, Bethesda, pp 137–148

Hernández-Vásquez S, Valadez-González C (1998) Observaciones de los epizoarios encontrados sobre la tortuga golfina Lepidochelys olivacea en La Gloria, Jalisco, México. Cienc Mar 24:119–125

Hirai, J, Tsuda A, Goetze E (2015) Extensive genetic diversity and endemism across the global range of the oceanic copepod Pleuromamma abdominalis. Prog Oceangr 138:77–90

Hirji KF, Tan SJ, Elashoff RM (1991) A quasi‐exact test for comparing two binomial proportions. Stat Med 10:1137–1153

Hosono T, Minami H (2011) Stable isotope analysis of epibiotic caprellids (Amphipoda) on loggerhead turtles provides evidence of turtle’s feeding history. In: Asakura A (ed) New frontiers in crustacean biology: proceedings of the TCS summer meeting. Koninklijke Brill NV, Leiden, pp 299–309

Howell EA, Dutton PH, Polovina JJ, Bailey H, Parker DM, Balazs GH (2010) Oceanographic influence on the dive behavior of juvenile loggerhead turtles (Caretta caretta) in the North Pacific Ocean. Mar Biol 157:1011–1026

Hubbs CL (1977) First record of mating of ridley turtles in California, with notes on commensals, characters, and systematics. Calif Fish Game 63:263–267

Ip BH, Schubart CD, Tsang LM, Chu KH (2015) Phylogeny of the shore crab family Grapsidae (Decapoda: Brachyura: Thoracotremata) based on a multilocus approach. Zool J Linn Soc Lond 174:217–227

Ishihara T, Kamezaki N, Matsuzawa Y, Iwamoto F, Oshika T, Miyagata Y, Ebisui C, Yamashita S (2011) Reentry of juvenile and subadult loggerhead turtles into natal waters of Japan. Curr Herpetol 30:63–68

Jarman SN, Elliott NG, McMinn A (2002) Genetic differentiation in the Antarctic coastal krill Euphausia crystallorophias. Heredity 88:280–287

171

Johannesson K, Johannesson B, Lundgren U (1995) Strong natural selection causes microscale allozyme variation in a marine snail. Proc Natl Acad Sci USA 92:2602–2606

Jones TT, Seminoff JA (2013) Feeding biology: advances from field-based observations, physiological studies, and molecular techniques. In: Wyneken J, Lohmann KJ, Musick JA (eds) The biology of sea turtles, volume III. CRC Press, Boca Raton, pp 211–247

Jones JC, Fan S, Franchini P, Schartl M, Meyer A (2013) The evolutionary history of Xiphophorus fish and their sexually selected sword: a genome-wide approach using restriction site-associated DNA sequencing. Mol Ecol 22:2986–3001

Juanicó M (1976) Sobre la distribución geográfica de Planes minutus (L., 1758) y P. cyaneus Dana, 1852 (Crustacea-Decapoda) en la región centro sur de Chile. Rev Biol Mar Oceanog 37:145–146

Killingley JS, Lutcavage M (1983) Loggerhead turtle movements reconstructed from 18O and 13C profiles from commensal barnacle shells. Estuar Coast Shelf S 16:345–349

Knowlton N (1980) Sexual selection and dimorphism in two demes of a symbiotic, pair- bonding snapping shrimp. Evolution 34:161–173

Knowlton N (2000) Molecular genetic analyses of species boundaries in the sea. Hydrobiologia 420:73–90

Kopitsky KL, Pitman RL, Dutton PH (2005) Aspects of olive ridley feeding ecology in the eastern tropical Pacific. NOAA Tech Memo. NMFS-SEFSC-528, pp 217

Lazo-Wasem E, Pinuo T, Peña de Niz A, Feuerstein A (2011) Epibionts associated with the nesting turtles Lepidochelys olivacea and Chelonia mydas in Jalisco, Mexico: a review and field guide. Bull Peabody Mus Nat Hist. 52:221–240

Lenormand T (2002) Gene flow and the limits of natural selection. Trends Ecol Evo 17:183–189

Limpus CJ, Limpus DJ (2003) Biology of the loggerhead turtle in western South Pacific Ocean foraging areas. In: Bolten AB, Witherington BE (eds) Biology and conservation of the loggerhead sea turtle. Smithsonian Institution Press, Washington, DC, pp 93–113

Linnaeus C (1747) Wästgosta-Resa, Pä Riksens Hogloflige Ständers Befallning Forrättad är 1746. Stockholm

Linnaeus C (1758) Systema naturae, ed. 10. Stockholm

172

Liria Loza A (2011) Ecosistemas errantes: epibiontes como indicadores biogeográficos de tortugas marinas de Canarias. PhD Thesis. Universidad de Las Palmas de Gran Canaria

Liria Loza A, López-Jurado LF, Calabuig Miranda P (2008) Comparative study of the epibionts on the pelagic and mature female loggerhead turtles on the Canary and Cape Verde Islands. NOAA Technical Memorandum NMFS-SEFSC-567

Lowe WH, Allendorf FW (2010) What can genetics tell us about population connectivity? Mol Ecol 19:3038–3051

MacLeay WS (1838) Illustrations of the Annulosa of South Africa. In: Smith A (ed) Illustrations of the Zoology of South Africa; Consisting Chiefly of Figures and Descriptions of the Objects of Natural History Collected During an Expedition into the Interior of South Africa, in the Years 1834, 1835, and 1836. Invertebratae, volume 5. A. Smith, Elder and Co., London

Manning RB, Holthuis LB (1981) West African Brachyuran Crabs (Crustacea: Decapoda). Smithsonian Institution Press, Washington DC.

Markham CR (1893) The journal of Christopher Columbus (during his first voyage, 149293), and documents relating the voyages of John Cabot and Gasper Corte Real. Hakluty Society, London

Marshall HG, Burchardt L (2005) Neuston: its definition with a historical review regarding this concept and community structure. Arch Hydrobiol 164:429–448

Mathews LM (2002) Territorial cooperation and social monogamy: factors affecting intersexual interactions in pair-living snapping shrimp. Anim Behav 63:767–777

McCormack JE, Hird SM, Zellmer AJ, Carstens BC Brumfield RT (2013) Applications of next-generation sequencing to phylogeography and phylogenetics. Mol Phylogenet Evol 66:526–538

McMillan RP (1968) The crab Planes minutus associating with a turtle. The Western Australian Naturalist 10:194

McPeek MA, Holt RD (1992) The evolution of dispersal in spatially and temporally varying environments. Am Nat 140:1010–1027

Milne-Edwards A, Bouvier EL (1899) Crustacés Décapodes provenant des campagnes de l’Hirondelle (Supplement) et de la Princesse-Alice (1891-1897). Résultats des Campagnes Scientifi ques accomplies sur son Yacht par Albert 1er Prince Souverain de Monaco, 13: 1–106

Miranda L, Moreno RA (2002) Epibiontes de Lepidochelys olivacea (Eschscholtz, 1829) (Reptilia: Testdinata: Cheloniidae) en la región centro sur de Chile. Rev Biol Mar Oceanog 37:145–146

173

Murray J (1895) Report of the scientific results of the voyage of H. M. S. Challenger during the years 1872—76, A summary of the scientific results. Edinburgh and Dublin, London

Musick JA, Limpus CJ (1997) Habitat utilization and migration in juvenile sea turtles. In: Lutz PL, Musick JA (eds). The biology of sea turtles. CRC Press, Boca Raton, pp 137–163

Naumann E (1917) Beiträge zur Kenntnis des Teichnannoplanktons. II. Über das Neustron des Süßwassers. Biologisches Zentralblatt 37:98-106

Ng PKL, Guinot D, Davie PJ (2008) Systema brachyurorum: part 1. An annotated checklist of extant brachyuran crabs of the world. Raffles B Zool 17:1–286

Nichols WJ, Resendiz A, Mayoral-Russeau C (2000) Biology and conservation of loggerhead turtles (Caretta caretta) in Baja California, Mexico. NOAA Tech Memo NMFS-SEFSC-443, pp 169–171

O’Riordan CE (1976) Marine fauna notes from the National Museum of Ireland – II. The Irish Naturalists’ Journal 18:305–307

Palumbi SR (1994) Genetic divergence, reproductive isolation, and marine speciation. Annu Rev Ecol Syst 25:547–572

Palumbi SR (2003) Population genetics, demographic connectivity, and the design of marine reserves. Ecol Appl 13:146–158

Parchman TL, Gompert Z, Mudge J, Schilkey FD, Benkman CW, Buerkle CA (2012) Genome-wide association genetics of an adaptive trait in lodgepole pine. Mol Ecol 21:2991–3005

Parker DM, Cooke WJ, Balazs GH (2005) Diet of oceanic loggerhead sea turtles (Caretta caretta) in the central North Pacific. Fish Bull 103:142–152

Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research – an update. Bioinformatics 28:2537– 2539

Peckham SH, Maldonado-Diaz D, Tremblay Y, Ochoa R, Polovina J, Balazs G, Dutton PH, Nichols WJ (2011) Demographic implications of alternative foraging strategies in juvenile loggerhead turtles Caretta caretta of the North Pacific Ocean. Mar Ecol Prog Ser 425:269–280

Peiró DF, Baeza JA, Mantelatto FL (2012) Host-use pattern and sexual dimorphism reveals the mating system of the symbiotic pea crab Austinixa aidae (Crustacea: Brachyura: Pinnotheridae). J Mar Biol Assoc UK 93:715–723

174

Peterson BK, Weber JN, Kay EH, Fisher HS, Hoekstra HE (2012) Double digest RADseq: an inexpensive method for de novo SNP discovery and genotyping in model and non-model species. PLoS ONE 7:e37135

Pfaller JB, Alfaro-Shigueto J, Balazs GH, Ishihara T, Kopitsky K, Mangel JC, Peckham SH, Bolten AB, Bjorndal KA (2014a) Hitchhikers reveal cryptic host behavior: new insights from the association between Planes major and sea turtles in the Pacific Ocean. Mar Biol 161:2167–2178

Pfaller JB, Alfaro-Shigueto J, Giffoni B, Ishihara T, Mangel JC, Peckham SH, Bjorndal KA, Baeza JA (2014b) Social monogamy in the crab Planes major, a facultative symbiont of loggerhead sea turtles. J Exp Mar Biol Ecol 461:124–132

Pfaller JB, Frick MG, Brischoux F, Sheehy III CM, Lillywhite HB (2012) Marine snake epibiosis: a review and first report of decapods associated with Pelamis platurus. Integr Comp Biol 52:296–310

Pfaller JB, Frick MG, Reich KJ, Williams KL, Bjorndal KA (2008) Carapace epibionts of loggerhead turtles (Caretta caretta) nesting at Canaveral National Seashore, Florida. J Nat Hist 42:1095–1102

Plackett RL (1953) The truncated Poisson distribution. Biometrics 9:485–488

Plotkin PT (2010) Nomadic behaviour of the highly migratory olive ridley sea turtle Lepidochelys olivacea in the eastern tropical Pacific Ocean. Endang Species Res 13:33–40

Polovina JJ, Balazs GH, Howell EA, Parker DM, Seki MP, Dutton PH (2004) Forage and migration habitat of loggerhead (Caretta caretta) and olive ridley (Lepidochelys olivacea) sea turtles in the central North Pacific Ocean. Fish Oceanogr 13:36–51

Polovina JJ, Howell E, Parker DM, Balazs GH (2003) Dive-depth distribution of loggerhead (Caretta caretta) and olive ridley (Lepidochelys olivacea) sea turtles in the central North Pacific: might deep longline sets catch fewer turtles? Fish Bull 101:189–193

Polovina JJ, Kobayashi DR, Parker DM, Seki MP, Balazs GH (2000) Turtles on the edge: movements of loggerhead turtles (Caretta caretta) along oceanic fronts, spanning longline fishing ground in the central North Pacific, 1997-1998. Fish Oceanogr 9:71–82

Pons M, Verdi A, Domingo A (2011) The pelagic crab Planes cyaneus (Dana, 1851) (Decapoda, Brachyura, Grapsidae) in the southwestern Atlantic Ocean in association with loggerhead sea turtles and buoys. Crustaceana 84:425–434

Poupin J, Davie PJF, Cexus J-C (2005) A revision of the genus Pachygrapsus Randall, 1840 (Crustacea: Decapoda: Brachyura, Grapsidae), with special reference to the southwest Pacific species. Zootaxa 1015

175

Prado A, de Melo GAS (2002) Le genre Planes Bowdich (Decapoda, Grapsidae) sur Les Côtes Brésiliennes (The Genus Planes Bowdich (Decapoda, Grapsidae) along the Brazilian Coast). Crustaceana 75:579–595

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155:945–959

R Development Core Team (2008) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.r- project.org.

Rathbun MJ (1902) Papers from the Hopkins Stanford Galapagos Expedition, 18981899, VIII: Brachyura and Macura. Proc Wash Acad Sci 4:275–292

Rathbun MJ (1914) New genera and species of American brachyrhynchus crabs. Proc US Nat Mus 47:117–129

Reich KJ, Bjorndal KA, Frick MG, Witherington BE, Johnson C, Bolten AB (2010) Polymodal foraging in adult female loggerheads (Caretta caretta). Mar Biol 157:651–663

Reitzel AM, Herrera S, Layden MJ, Martindale MQ, Shank TM (2013) Going where traditional markers have not gone before: utility of and promise for RAD sequencing in marine invertebrate phylogeography and population genomics. Mol Ecol 22:2953–2970

Rice MR, Balazs GH (2008) Diving behavior of the Hawaiian green turtle (Chelonia mydas) during oceanic migrations. J Exp Mar Biol Ecol 356:121–127

Rodgers PJ, Reaka ML, Hines AH (2011) A comparative analysis of sperm storage and mating strategies in five species of brachyuran crabs. Mar Biol 158:1733–1742

Roux P (1828) Crustacés de la Méditerranée et de son littoral. Paris and Marseille

Saba VS (2013) Oceanic habits and habitats: Dermochelys coriacea. In: Wyneken J, Lohmann KJ, Musick JA (eds) The biology of sea turtles, volume III. CRC Press, Boca Raton, pp 163–188

Sakai T (1939) Studies on the crabs of Japan, IV: Brachygnatha, Brachyrhyncha. Yokendo, LTD, Tokyo, pp 365–741

Sakai T (1969) Two new genera and twenty-two new species of crabs from Japan. Proc Biol Soc Wash 82:243–280

Sands CJ, Jarman SN, Jackson GD (2003) Genetic differentiation in the squid Moroteuthis ingens inferred from RAPD analysis. Polar Biol 26:166–170

176

Schärer MT (2001) A survey of the epibiota of Eretmochelys imbricata (Testudines: Cheloniidae) of Mona Island, Puerto Rico. Rev Biol Trop 51:87–89

Schmidt PS, Rand DM (2001) Adaptive maintenance of genetic polymorphism in an intertidal barnacle: habitat- and life-stage-specific survivorship of Mpi genotypes. Evolution 55:1336–1344

Schubart CD (2011) Reconstruction of phylogenetic relationships within Grapsidae (Crustacea: Brachyura) and comparison of trans-isthmian versus amphi-atlantic gene flow on mtDNA. Zool Anz 250:472–478

Schubart CD, Cannicci S, Vannini M, Fratini S (2006) Molecular phylogeny of grapsoid crabs (Decapoda, Brachyura) and allies based on two mitochondrial genes and a proposal for refraining from current superfamily classification. J Zool Syst Evol Res 44:193–199

Schubart CD, Cuesta JA, Felder DL (2002) Glyptograpsidae, a new brachyuran family from Central America: larval and adult morphology, and a molecular phylogeny of the Grapsoidea. J Crustacean Biol 22:28–44

Schubart CD, Cuesta JA, Felder DL (2005) Phylogeography of Pachygrapsus transversus (Gibbes, 1850): the effect of the American continent and the Atlantic Ocean as gene flow barriers and recognition of Pachygrapsus socius Stimpson 1871 as a valid species. Nauplius 13:99–113

Schweigger E (1964) El Litoral Peruano. Grafica Morsom SA, Lima

Seminoff JA, Eguchi T, Carretta J, Allen CD, Prosperi D, Rangel R, Gilpatrick JW, Forney K, Peckham SH (2014) Loggerhead sea turtle abundance at a foraging hotspot in the eastern Pacific Ocean: implications for at-sea conservation. Endang Species Res 24:207–220

Seminoff JA, Jones TT, Resendiz A, Nichols WJ, Chaloupka MY (2003) Monitoring green turtles (Chelonia mydas) at a coastal foraging area in Baja California, Mexico: multiple indices describe population status. J Mar Biol Assoc UK 83:1355–1362

Seminoff JA, Zárate P, Coyne MC, Foley DG, Parker D, Lyon BN, Dutton PH (2007) Post-nesting migrations of Galápagos green turtles Chelonia mydas in relation to oceanographic conditions: integrating satellite telemetry with remote sensed ocean data. Endang Species Res 4:57–72

Shamblin BM, Bolten AB, Abreu-Grobois FA, Bjorndal KA, Cardona L, Carreras C, Clusa M, Monsón-Argüello C, Nairn CJ, Nielsen JT, Nel R, Soares LS, Stewart KR, Vilaça, Türkozan O, Yilmaz C, Dutton PH (2014) Geographic patterns of genetic variation in a broadly distributed marine vertebrate: New insights into loggerhead turtle stock structure from expanded mitochondrial DNA sequences. PLoS ONE 9:e85956

177

Shanks AL (1983) Surface slicks associated with tidally forced internal waves may transport pelagic larvae of benthic invertebrates and fishes shoreward. Mar Ecol Prog Ser 13:311–315

Shuster SM, Wade MJ (2003) Mating systems and strategies. Princeton University Press, New Jersey

Sloane H (1725) A voyage to the islands Madera, Barbados, Nieves, St. Christopher’s, and Jamaica. London

Sokal RR, Rohlf FJ (1981) Biometry. W.H. Freeman and Co., New York

Spivak ED, Bas CC (1999) First finding of the pelagic crab Planes marinus (Decapoda: Grapsidae) in the southwestern Atlantic. J Crustacean Biol 19:72–76

Stamatakis A (2014) RAxML version 8: a tool for phylogenetic analysis and post- analysis of large datasets. Bioinformatics 2014:btu033

Steinbeck J, Ricketts EF (1941) Sea of Cortez. Viking Press, New York

Thiel M (2003) Extended parental care in crustaceans – an update. Rev Chil Hist Nat 76:205–218

Thiel M, Baeza JA (2001) Factors affecting the social behaviour of symbiotic crustaceans living symbiotically with other marine invertebrates: a modeling approach. Symbiosis 30:163–190

Thiel M, Haye PA (2006) The ecology of rafting in the marine environment. III. Biogeographical and evolutionary consequences. Oceanogr Mar Biol 44:323–429

Thiel M, Zander A, Valdivia N, Baeza JA, Rueffler C (2003) Host fidelity of a symbiotic porcellanid crab: the importance of host characteristics. J Zool 261:353–362

Treml EA, Ford JR, Black KP, Swearer SE (2015) Identifying the key biophysical drivers, connectivity outcomes, and metapopulation consequences of larval dispersal in the sea. Movement Ecol 3:17 van der Meeren GI (1994) Sex- and size-dependent mating tactics in a natural population of shore crabs Carcinus maenas (Crustacea: Brachyura). J Anim Ecol 63:307–314

Vargas R, Wehrtmann IS (2009) Decapod crustaceans. In: Wehrtmann IS, Cortés J (eds) Marine biodiversity of Costa Rica, Central America. Springer, Berlin, pp 209–228

178

Wagner CE, Keller I, Wittmer S, Selz OM, Mwaiko S, Greuter L, Sivasundar A, Seehausen O (2013) Genome-wide RAD sequence data provide unprecedented resolution of species boundaries and relationships in the Lake Victoria cichlid adaptive radiation. Mol Ecol 22:787–798

Wahl M, Mark O (1999) The predominately facultative nature of epibiosis: experimental and observational evidence. Mar Ecol Prog Ser 187:59–66

Warner GF (1970) Behavior of two species of grapsid crabs during intraspecific interactions. Behaviour 36:9–19

Weersing K, Toonen RJ (2009) Population genetics, larval dispersal, and connectivity in marine systems. Mar Ecol Prog Ser 393:1–12

Wicksten MK, Behrens MD (2000) New record of the pelagic crab Planes cyaneus in California (Brachyura: Grapsidae). SCAMIT Newsletter 19:7

Work TM, Balazs GH (2010) Pathology and distribution of sea turtles landed as bycatch in the Hawaii-based North Pacific pelagic longline fishery. J Wildlife Dis 46:422- 432

Wyneken J, Lohmann KJ, Musick JA (2013) The Biology of Sea Turtles, Volume III. CRC Press, Boca Raton

Zane L, Ostellari L, Maccatrozzo L, Bargelloni L, Battaglia B, Patarnello T (1998) Molecular evidence for genetic subdivision of Antarctic krill (Euphausia superba Dana) populations. Proc R Soc Lond B 265:2387-2391

Zane L, Patarnello T (2000) Krill: a possible model for investigating the effects of ocean currents on the genetic structure of pelagic invertebrates. Can J Fish Aquat Sci 57:16-23

179

BIOGRAPHICAL SKETCH

Joseph Bryce Pfaller was born in St. Louis, Missouri and raised in Iowa City,

Iowa and to a lesser extent in Portland, Oregon. Joseph completed a Bachelor of

Science in zoology from the University of Florida in 2005 and worked as a Research

Associate between the Archie Carr Center for Sea Turtle Research (University of

Florida) and the Queensland Environmental Protection Agency from 2005-6. Joseph completed a Master of Science in biology from Florida State University in 2009 working under the supervision of Dr. Gregory Erickson studying functional morphology and feeding biomechanics in vertebrates. In 2009, Joseph began pursuing a Doctor of

Philosophy in zoology at the University of Florida under the supervision of Dr. Karen

Bjorndal. Since 2011, Joseph has been the Research Director of the Caretta Research

Project based in Savannah, Georgia, a non-profit research and conservation program dedicated to studying and conserving loggerhead sea turtles on the Wassaw National

Wildlife Refuge. Joseph completed his doctorate degree in May 2016.

180