Patterns of Evolution in Gobies (Teleostei: Gobiidae): a Multi-Scale Phylogenetic Investigation

Home , Acanthogobius, Acentrogobius, Amblygobius, Amphiprioninae, Asterropteryx, Astrabe

PATTERNS OF EVOLUTION IN GOBIES (TELEOSTEI: GOBIIDAE): A MULTI-SCALE PHYLOGENETIC INVESTIGATION

A Dissertation

LUKE MICHAEL TORNABENE

BS, Hofstra University, 2007 MS, Texas A&M University-Corpus Christi, 2010

Submitted in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

MARINE BIOLOGY

Texas A&M University-Corpus Christi Corpus Christi, Texas

December 2014

PATTERNS OF EVOLUTION IN GOBIES (TELEOSTEI: GOBIIDAE): A MULTI-SCALE PHYLOGENETIC INVESTIGATION

A Dissertation

LUKE MICHAEL TORNABENE

This dissertation meets the standards for scope and quality of Texas A&M University-Corpus Christi and is hereby approved.

Frank L. Pezold, PhD Chris Bird, PhD Chair Committee Member

Kevin W. Conway, PhD James D. Hogan, PhD Committee Member Committee Member

Lea-Der Chen, PhD Graduate Faculty Representative

December 2014

ABSTRACT

The family of fishes commonly known as gobies (Teleostei: Gobiidae) is one of the most diverse lineages of vertebrates in the world. With more than 1700 species of gobies spread among more than 200 genera, gobies are the most species-rich family of marine fishes. Gobies can be found in nearly every aquatic habitat on earth, and are often the most diverse and numerically abundant fishes in tropical and subtropical habitats, especially coral reefs. Their remarkable taxonomic, morphological and ecological diversity make them an ideal model group for studying the processes driving taxonomic and phenotypic diversification in aquatic vertebrates. Unfortunately the phylogenetic relationships of many groups of gobies are poorly resolved, obscuring our understanding of the evolution of their ecological diversity. This dissertation is a multi-scale phylogenetic study that aims to clarify phylogenetic relationships across the Gobiidae and demonstrate the utility of this family for studies of macroevolution and speciation at multiple evolutionary timescales.

In the first chapter, I present a DNA sequence matrix derived from two nuclear genes to help resolve intergeneric level phylogenetic relationships with the Gobiidae. My study is the first to use data from conserved nuclear loci to infer relationships across the Gobioidei, and the results provide strong support for the monophyly of, and interrelationships between, several ecologically divergent clades. Specifically, I show that gobies are asymmetrically divided into two clades, one of which contains primarily marine species and the other comprises mostly estuarine or freshwater taxa.

In the second chapter, I focus on the evolution of microhabitat association and morphology in one of the most diverse lineages of gobies, the reef-associated genus Eviota. Eviota species

have invaded novel microhabitats multiple times throughout their evolutionary history, often occurring independently of diagnostic morphological changes in pectoral-fin ray branching and arrangement of sensory cephalic lateralis pores. The combination of historical ecological flexibility coupled with resilience to local extinction events may explain the elevated extant biodiversity in Eviota.

Lastly, in my third chapter, I use Eviota as a model for studying fine-scale speciation in the

Coral Triangle, a marine biodiversity hotspot in the Western Pacific Ocean. A phylogeographic analysis of two species complexes that have diverged within the Coral Triangle provides strong support for the hypothesis that the Coral Triangle serves as a ‘center of origin’ or cradle of new species. Specifically, I demonstrate that a combination of biotic and abiotic factors may be contributing to rapid speciation both in allopatry and sympatry within the last 1.5 million years.

The presence of recently diverged cryptic species in the Coral Triangle implies that our current estimates of biodiversity in this marine hotspot are severely underestimated.

Ultimately, this dissertation demonstrates that gobies and other ecologically diverse clades of fishes serve as excellent model groups for studying the processes driving taxonomic and phenotypic diversification in marine species at a variety of spatial and temporal scales. This project will serve as a foundation for future studies that aim to use more comprehensive genomic datasets to address questions regarding drivers of speciation and ecological diversification in gobiid fishes.

DEDICATION

I dedicate this dissertation to my parents Elizabeth M. Tornabene and Michael F.

Tornabene. My mother has been a lifeline in times of struggle and a guiding light when I am lost at sea. My father has been an endless source of encouragement, and the unforgettable times we shared on the ocean led to my infatuation with the sea and the mysteries that lie beneath its surface. None of my successes in life would be possible without their love and encouragement.

vii

ACKNOWLEDGMENTS This dissertation could not have been completed without the encouragement and support of many amazing relatives, friends, and faculty members. I sincerely thank my adviser, mentor, and dear friend Dr. Frank Pezold. Dr. Pezold and I met in 2006 when I was undergraduate attending a conference in New Orleans. We immediately began chatting about goby systematics and after eight years the topics of our conversations haven’t changed much. Dr. Pezold has educated, motivated, supported and directed me as I wandered through the maze of academia, and I am incredibly grateful for his wisdom and guidance. I also thank my committee members

Drs. Chris Bird, Derek Hogan and Kevin Conway for their support and insight throughout the program.

I am extremely grateful for the countless individuals outside of Texas A&M system that helped make this dissertation possible. I thank my former mentor and good friend Dr. James

Van Tassell, who spent many hours each winter and summer pushing me to achieve my goals.

Drs. Mark Erdmann, Carole Baldwin, and Chenhong Li have also contributed to my success at

TAMUCC and for that I am in their debt.

My dear friends and colleagues Ryan Chabarria and Sharon Furriness contributed not only scientific expertise but also their friendship and encouragement. I thank Dr. Gabby Ahmadia,

David Boseto, Samantha Valdez and Dr. Yongjiu Chen for their major contributions to various chapters of this dissertation. I also thank all of the current and former members of the Fish

Systematics and Conservation Lab for creating an incredible research environment. I am thankful for the brotherhood of my dear friends Judd Curtis, Keith Johnson, Matthew

Magnusson, Scott Large, Frank Kelly, Jon Anderson and Tom Jerhada, who helped me through these final years.

I am incredibly grateful for my loving family. My parents Elizabeth and Michael have

viii

been my biggest supporters through this journey and I could not ask for more loving, selfless parents. My brother Paul and sister Robyn have been caring and loving siblings and have taught me much about perseverance and overcoming adversity. Lastly, my support network of family in New York and Florida consists of over a dozen aunts and uncles, cousins, grandparents and nephews, all of which are excited to see me accomplish this great achievement.

Funding for many aspects of my research was provided by NSF-OISE 0080699 and OISE-

0553910 to Frank Pezold. I also received two Lerner Gray Awards from the American Museum of Natural History, two grants from the Smithsonian Institute Schultz Fund for Ichthyological

Research, and several scholarships and grants from Texas A&M University – Corpus Christi that helped make this research possible. The three chapters here have been adapted for this dissertation with permission from publishers of the journals Molecular Phylogenetics &

Evolution and Systematics & Biodiversity.

TABLE OF CONTENTS CONTENTS PAGE

ABSTRACT ...... v

DEDICATION ...... vii

ACKNOWLEDGMENTS ...... viii

TABLE OF CONTENTS ...... x

LIST OF FIGURES ...... xiv

LIST OF TABLES ...... xv

INTRODUCTION ...... 1

CHAPTER I: Gobies are deeply divided: phylogenetic evidence from nuclear DNA (Teleostei:

Gobioidei: Gobiidae) ...... 7

Abstract ...... 7

1. Introduction ...... 8

2. Materials and methods ...... 16

3. Results ...... 20

3. 1 Gobiidae ...... 24

3. 2 Basal relationships ...... 25

4. Discussion ...... 26

4. 1 The deep divide in gobies ...... 26

4. 2 Relationships within ‘Gobionellidae’ sensu Thacker (2009) ...... 27

4. 3 Relationships within ‘Gobiidae’ sensu Thacker (2009) ...... 31

4.4 Relationships of basal gobioids ...... 34

5. Conclusions and areas of in need of future research ...... 36

Acknowledgements ...... 37

References ...... 38

CHAPTER II – Evolution of microhabitat association and morphology in a diverse group of cryptobenthic coral reef fishes (Teleostei: Gobiidae: Eviota) ...... 46

Abstract ...... 46

1. Introduction ...... 47

1.1. Ecological drivers of goby evolution ...... 47

1.2. Indo-pacific dwarfgobies ...... 49

2. Methods ...... 51

2.1. Collecting specimens and characterizing habitat association ...... 51

2.2. Specimen identification and morphological character coding ...... 54

2.3. DNA sequencing ...... 55

2.4. Phylogenetic analysis ...... 56

2.5. Habitat and character mapping ...... 57

3. Results ...... 58

3.1. Phylogenetic analysis ...... 58

3.2. Microhabitat association and morphological character evolution ...... 59

4. Discussion ...... 64

4.1. Microhabitat association ...... 64

4.2. Morphological character evolution ...... 65

4.3. Diversification in stages ...... 67

4.4. Dwarfgoby diversity, speciation and extinction ...... 69

5. Conclusions ...... 70

Acknowledgments ...... 71

References ...... 72

CHAPTER III – Support for a ‘Center of Origin’ in the Coral Triangle: cryptic diversity, recent speciation, and local endemism in a diverse lineage of reef fishes (Gobiidae: Eviota)...... 79

Abstract ...... 79

1. Introduction ...... 80

1.1 Biodiversity within the Coral Triangle: origin, accumulation or overlap? ...... 80

1.2 Gobies and speciation within a center of origin ...... 82

1.3 Tips of the Eviota tree: the E. nigriventris and E. bifasciata species complexes ...... 84

2. Methods ...... 86

3. Results ...... 88

4. Discussion ...... 95

4. 1 Eviota nigriventris complex ...... 97

4.2 Eviota bifasciata complex...... 99

4.3 Conclusions ...... 103

Acknowledgements ...... 105

References ...... 106

SUPPLEMENTARY MATERIAL ...... 122

SUMMARY AND FUTURE RESEARCH ...... 125

xii

BIOGRAPHICAL SKETCH ...... 128

xiii

LIST OF FIGURES

FIGURES PAGE

Figure 1: Synopsis of gobioid family-group relationships...... 13

Figure 2: Bayesian phylogeny of concatenated dataset...... 21

Figure 3: Bayesian phylogeny of RAG1 dataset...... 22

Figure 4: Bayesian phylogeny of rhodopsin dataset...... 23

Figure 5: Typical arrangement of Eviota sensory pore pattern...... 55

Figure 6: Prelimnary Bayesian phylogeny of Eviota based on the concatenated dataset ...... 59

Figure 7: Chronogram of Eviota species showing microhabitat association...... 61

Figure 8: Chronogram of Eviota species showing sensory pore patterns...... 62

Figure 9: Chronogram of Eviota species showing pectoral fin branching...... 63

Figure 10: Localities of samples and geograhic ranges of species complexes ...... 86

Figure 11: Bayesian phylogeny of Eviota based on concatenated datasets ...... 90

Figure 12: Time-calibrated phylogeny of Eviota based on COI dataset ...... 91

Figure 13: Time-calibrated phylogeny and median-joining haplotype network of the Eviota

nigriventris complex based on COI dataset...... 93

Figure 14: Time-calibrated phylogeny and median-joining haplotype network of the Eviota

bifasciata complex based on COI dataset ...... 94

Figure 15: Coral Triangle during glacial-maxima with sea-levels 120 m below present levels ....96

xiv

LIST OF TABLES

TABLES PAGE

Table 1: Molecular studies on gobies using nuclear loci...... 16

Table 2: New sequences generated from this study...... 19

Table 3: Specimens examined for microhabitat analysis...... 53

Table 4: Designation of microhabitat associations of Eviota species based on the percent of the

total number of specimens collected from each microhabitat...... 60

Table 5: Specimens used in phylogeography study...... 122

Table 6: Distance matrix showing % identical sites across samples in the E. nigriventris

complex...... 123

Table 7: Distance matrix showing % identical sites across samples in the E. bifasciata

complex…...... 124

INTRODUCTION

Why have some lineages of organisms persisted for millions of years virtually unchanged while others have diversified into a seemingly endless variety of forms? The uneven distribution of phenotypic and taxonomic diversity we observe in nature is a phenomenon that has interested evolutionary biologists and naturalists since they first began naming and classifying organisms.

Indeed some groups of organisms appear to possess a greater capacity for morphological, behavioral and taxonomic diversification, superficially seeming to be better ‘adapted for adaptation’. An example of such a group can be seen at the base of the phylogeny of fishes.

More than 420 million years ago an ancestral aquatic vertebrate began diverging into two species that would ultimately represent the ancestors to clades comprising the bony fishes and cartilaginous fishes. Today bony fishes comprise more than 28,000 extant species, while there are only about 1050 living species of cartilaginous fishes. Determining what factors promote or restrict phenotypic evolution, ecological differentiation, reproductive isolation, and ultimately speciation is a central theme in the field of biology. By examining patterns of evolution in exceptionally diverse clades of organisms, such as bony fishes, we can develop and test hypotheses regarding the drivers of speciation and differentiation.

Unfortunately, in many cases the phylogenetic relationships of species-rich clades are difficult to untangle due to intrinsic issues like the presence of rapid periods of explosive radiations, homoplasy caused by parallel morphological adaptations, or logistic issues such as the inability to adequately sample enough taxa or survey ample phylogenetically informative characters. All of these scenarios have contributed to confusion in the phylogenetic relationships of bony fishes. Recent molecular phylogenetic studies have attempted to clarify relationships of several of the enigmatic clades across bony fishes using increasingly broader (more taxa) and

deeper (more genes) sampling strategies (e.g. Near et al., 2012; Betancur-R et al., 2013), and have largely succeeded in this regard. However, relationships within most of the exceptionally- diverse lineages of bony fishes are still poorly-resolved. One such group is the family Gobiidae.

The Gobiidae (sensu Gill and Mooi, 2012), commonly known as gobies, comprise more than 1700 valid species and are among the most diverse families of vertebrates on earth. Most species are small, rarely exceeding a few cm in total length, and are cryptobenthic. Gobies are dominant components of the ichthyofauna of subtropical and tropical marine and estuarine habitats as well as freshwater stream communities on tropical oceanic islands. Several factors have contributed phylogenetic uncertainty within the family. Traditional phylogenetic inference based on morphological characters is challenging due to repeated parallel trends of evolution via miniaturization, leading to rampant reduction and character loss that ultimately obscure our understanding of homology and homoplasy (Van Tassell et al., 2011). On the contrary, some authors argue that phylogenetically informative morphological characters are abundant but simply have not been surveyed extensively across gobioids (Gill and Mooi, 2012; but see

Birdsong et al., 1988, Pezold, 1993). More recently molecular phylogenetic studies have improved our understanding of gobioid interrelationships and have supported many morphologically-based classifications but also refuted the monophyly of groups named in others

(see Ruber and Agorreta, 2011 for a review). Nevertheless many of the relationships within gobies have yet to be resolved by molecular data, despite multiple attempts over the last two decades. Reasons may be largely driven by inadequate taxon sampling, as some entire clades of gobies (e.g. the amblyopines) or large geographic faunal regions (e.g. African gobioids) are poorly sampled. Most studies to date have also relied on sequence data from mitochondrial genes that may not be appropriate for resolving deeper phylogenetic questions. Lastly, several

studies demonstrate that gobies underwent a short period of explosive diversification early in their evolutionary history, causing a star-burst pattern of radiation with many short internodes that could be difficult to unravel regardless of the molecular tools employed (e.g. Thacker and

Roje, 2011).

The taxonomic, ecological, morphological and behavioral diversity exhibited by gobies makes them an ideal group for studying patterns of speciation and macroevolution in fishes and marine species in general. Many lineages of gobies fall into distinct clades that have unique ecological characteristics (e.g. monophyletic groups representing shrimp gobies, mudskippers, cleaner gobies, burrowing eel gobies), indicating that habitat or behavioral differentiation played a role in goby diversification. However, the interrelationships between these ecologically- divergent clades are still largely unresolved (Thacker and Roje, 2011; Ruber and Agoretta,

2011), thus obscuring our understanding of the chain of evolutionary processes shaping the gobioid tree of life. The few studies to date that explicitly investigated the role of ecology in goby diversification focused on less inclusive subsets of gobies, and demonstrated that some clades clearly fit a model of explosive adaptive radiation driven by ecological divergence and microhabitat partitioning (e.g. Ruber et al., 2003, Yamada et al., 2009). Furthermore, after gobies diverged into these ecologically-distinct clades, speciation has been constant or perhaps even increasing as we approach the present (Thacker, 2009). Few studies have investigated the role of abiotic factors such as geology, plate tectonics and paleoceanography in driving this more recent stage of diversification (e.g. Pliocene or Pleistocene). Thus, gobies represent an understudied but promising model group for evolutionary studies investigating a multitude of biotic and abiotic mechanisms of diversification across a number of evolutionary time scales.

In this dissertation I present a multi-scale phylogenetic investigation of the Gobiidae that sheds light onto the patterns and processes behind speciation in gobies. Specifically, this study is divided into three distinct chapters, each focusing on a unique aspect of goby evolution. The first chapter addresses phylogenetic relationships among the major lineages of gobies. This study is the first to use DNA sequence data from conserved nuclear loci to infer relationships across the Gobioidei, and provides strong support for the monophyly of, and interrelationships between, several ecologically divergent clades. My second chapter focuses on the evolution of one the most diverse lineages of gobies, the reef-associated genus Eviota. The genus Eviota, commonly known as dwarfgobies, comprises more than 90 species of fishes that occur on a variety of microhabitats (sand, rubble, coral, etc.) on Indo-Pacific coral reefs. This chapter investigates the role of historical transitions in microhabitat association in producing morphological changes and exceptional taxonomic diversity within the genus.

My third and final chapter also focuses on the genus Eviota, and specifically investigates the timing of speciation events in two species complexes within the genus. In this chapter I use phylogeography of the Eviota bifasciata and E. nigriventris complexes to demonstrate that a variety of abiotic and biotic factors facilitate sympatric and allopatric speciation within the Coral

Triangle, a marine biodiversity hotspot. By using dwarfgobies as a model group for studying speciation at fine geographic scales, I show that the Coral Triangle is indeed a ‘center of origin’ that actively produces new lineages and serves as a source of biodiversity for adjacent regions.

References

Near, T.J., Eytan, R.I., Dornburg, A., Kuhn, K.L., Moore, J.A., Davis, M.P., Wainwright, P.C.,

Friedman, M., Smith, W.L., 2012. Resolution of ray-finned fish phylogeny and timing of

diversification. P Natl Acad Sci USA 109, 13698-13703.

Betancur, R.R., Broughton, R.E., Wiley, E.O., Carpenter, K., Lopez, J.A., Li, C., Holcroft, N.I.,

Arcila, D., Sanciangco, M., Cureton Ii, J.C., Zhang, F., Buser, T., Campbell, M.A.,

Ballesteros, J.A., Roa-Varon, A., Willis, S., Borden, W.C., Rowley, T., Reneau, P.C.,

Hough, D.J., Lu, G., Grande, T., Arratia, G., Orti, G., 2013. The tree of life and a new

classification of bony fishes. PLoS currents 5.

Gill, A., Mooi, R., 2012. Thalasseleotrididae, new family of marine gobioid fishes from New

Zealand and temperate Australia, with a revised definition of its sister taxon, the

Gobiidae (Teleostei: Acanthomorpha). Zootaxa 3266, 41-52.

Birdsong, R., Murdy, E., Pezold, F., 1988. A study of the vertebral column and median fin

osteology in gobioid fishes with comments on gobioid relationships. Bulletin of Marine

Science 42, 174-214.

Pezold, F., 1993. Evidence for a Monophyletic Gobiinae. Copeia 3, 634-643.

Rüber, L., Agorreta, A., 2011. Molecular systematics of gobioid fishes. In: Patzner, R.A., Van

Tassel, J.L., Kovacic, M., Kapoor, B. (Eds.), Biology of Gobies. Science Publishers,

Enfield, NH, pp. 23-50.

Thacker, C.E., Roje, D.M., 2011. Phylogeny of Gobiidae and identification of gobiid lineages.

Syst Biodivers 9, 329-347.

Rüber, L., Van Tassell, J.L., Zardoya, R., 2003. Rapid speciation and ecological divergence in

the American seven spined gobies (Gobiidae: Gobiosomatini) inferred from a molecular

phylogeny. Evolution 57, 1584-1598.

Yamada, T., Sugiyama, T., Tamaki, N., Kawakita, A., Kato, M., 2009. Adaptive radiation of

gobies in the interstitial habitats of gravel beaches accompanied by body elongation and

excessive vertebral segmentation. Bmc Evolutionary Biology 9.

Thacker, C.E., 2009. Phylogeny of Gobioidei and placement within Acanthomorpha, with a new classification and investigation of diversification and character evolution. Copeia 1, 93-104.

CHAPTER I: Gobies are deeply divided: phylogenetic evidence from nuclear DNA (Teleostei: Gobioidei: Gobiidae)

Abstract

Gobies (Gobiidae sensu Gill & Mooi, 2012) are one of the most diverse families of vertebrates, and comprise over 1700 species of marine, brackish and freshwater fishes.

Phylogenetic studies based on morphological characters and mtDNA have suggested that goby diversity is asymmetrically split between a speciose clade of predominantly marine species, and a less rich, but ecologically diverse, clade comprising predominantly freshwater and brackish species. This study is the first to explore this deep divide in gobies and their relationships at the family level using phylogenetic data from nuclear genes (RAG1, rhodopsin). Our results confirm the split within the Gobiidae, and agree with prior molecular studies on the inclusion of the following taxa within the two goby clades: (i) the more diverse of the two clades of gobies (the

‘Gobiidae’ sensu stricto of Thacker, 2009) comprises the gobiines, microdesmines, ptereleotrines and kraemeriines; (ii) the less diverse of the two gobiid clades (‘Gobionellidae’ sensu Thacker

2009) includes the gobionellines, oxudercines, amblyopines, sicydiines, as well as the European sand gobies. Some relationships within the two major gobiid clades remain unclear. Specifically, there remains confusion regarding the monophyly and interrelationships between the northern

Pacific gobionellines, the Mugilogobius group gobionellines, and the European sand gobies.

Additionally, within Thacker’s (2009) Gobiidae sensu stricto, there are several well-supported groups (e.g. the wormfishes and dartfishes, the Coral Gobies, the Gobiosomatini), yet relationships among these groups are still poorly resolved despite the use of data from two conserved nuclear genes. Future phylogenetic analyses of gobies will benefit greatly from taxon sampling that includes groups that have been historically under-represented in molecular studies

(e.g. European sand gobies, northern Pacific gobionellines, African species), as well as deeper genetic sampling including large numbers of independent loci from throughout the genome (i.e. a phylogenomic approach).

1. Introduction

Gobies are predominantly small benthic fishes forming one of the largest vertebrate families

(Gobiidae sensu Gill & Mooi, 2012). There are over 1700 valid species of Gobiidae in over 260 genera (Keith & Lord, 2011; Murdy, 2011a, 2011b; Pezold, 2011; Thacker, 2011; Thacker &

Roje, 2011; Eschmeyer, 2013). They are abundant components of the ichthyofauna of tropical and subtropical marine and estuarine waters and dominate freshwater stream fish communities on tropical oceanic islands. Because of their generally small size and an evolutionary tendency towards reduction in morphology, our understanding of their relationships and actual diversity has been obscured. Despite repeated patterns of reduction and loss, phylogenetically informative characters are abundant in gobies, and our understanding of their relationships may perhaps be limited by the lack of comprehensive morphological surveys rather than a lack of characters (Gill

&Mooi, 2012). Over the last 50 years however, considerable progress has been made through several substantive detailed morphological studies and more recently through molecular systematic studies (see Van Tassell et al., 2011; Rüber & Agorreta, 2011 for reviews).

One of the recurring patterns recovered among some of these analyses has been the identification of two large groups of gobies that have been associated with the subfamilies

Gobiinae and Gobionellinae (e.g. Miller, 1973; Pezold, 1993; Thacker, 2009). Though varying in details of membership, species historically associated with the Gobiinae are predominantly marine species whereas species historically included in the Gobionellinae are typically

associated with estuaries and freshwater (Thacker, 2009). Miller (1973) largely distinguished the two subfamilies by the number of epural bones – one in the Gobiinae and two in the

Gobionellinae. Distinctive groups such as the amphibious oxudercines (mudskippers) and the amblyopines (eel gobies) were lumped within the Gobionellinae, and the amphidromous sicydiines were included with the Gobiinae based upon the shared state of the epural number.

Miller’s intent was to focus on the recognition of basal groupings, distinctive features typifying mudskippers and eel gobies were regarded as superficial specializations for particular niches and not informative of greater relationships, although they were recognized as ‘tribes’.

Contrary to Miller (1973), Birdsong et al. (1988) separated gobioids into 32 smaller phenetic groups based upon combinations of different post-cranial axial osteological characters.

However, the largest group they recognized, termed the Priolepis group, included 54 genera comprising many of the species of Miller’s (1973) Gobiinae. Notably, members of this group dominate Indo-Pacific coral reefs and are characterized by a single epural and a 3-22110 first dorsal fin pterygiophore insertion pattern. The formula indicates that the series of pterygiophores begins in the third interneural space, with two pterygiophores in the third interneural space, two in the fourth, one in the fifth, one in the sixth and none in the seventh. Ten genera were placed in a Gobionellus group distinguished from the Priolepis group by possession of two epurals and a

3-12210 pterygiophore insertion pattern.

An examination of the suspensorium in a broad sampling of gobioid fishes (Harrison,

1989) also revealed specializations in the shape and relationship of the palatine, pterygoid and quadrate bones distinguishing Gobionellinae taxa from Gobiinae. A Ctenogobius lineage containing six genera, and an Oxyurichthys lineage with nine genera from the subfamilies

Gobionellinae, Oxudercinae and Amblyopinae were recognized. A close relationship between

the Gobionellinae (including amblyopines and oxudercines) and Sicydiinae was proposed based upon the structure of the palatine, a transverse suborbital neuromast pattern, first dorsal fin pterygiophore insertion pattern and an unossified scapula. The same palatine condition was also observed in several genera at that time included in the Gobiinae – Mugilogobius,

Stigmatogobius, Tamanka and Chlamydogobius, all of which are now recognized as gobionellines (Larson, 2001).

In a study of the cephalic lateralis system of gobioid fishes, Pezold (1993) revealed that the Gobiinae possess a single canal pore in the anterior interorbital region. This was recognized as a derived condition in addition to the single epural for the subfamily. By contrast,

Gobionellinae are characterized by a plesiomorphic paired pore condition. In addition to the

Gobionellus group of Birdsong et al. (1988) and the Ctenogobius lineage of Harrison (1989),

Gobionellinae were recognized to include the Acanthogobius, Astrabe and Chasmichthys groups of Birdsong et al. (1988) which were proposed as a single monophyletic northern Pacific group within that subfamily (Pezold, 1993). The Amblyopinae, Oxudercinae and Sicydiinae were recognized as distinct monophyletic subfamilies. Within the Gobionellinae, Larson (2001) recognized two additional groups – the Mugilogobius group and a Stenogobius group each of which she proposed as monophyletic. In contrast to the Gobiinae, there has been no proposed synapomorphy for the Gobionellinae, and the details of its relationship to the Oxudercinae,

Amblyopinae and Sicydiinae are unclear.

Analyses of mitochondrial gene sequences largely correspond with observations from morphological studies in suggestions of phylogenetic affinities and the presence of a deep divide within the Gobiidae. An early analysis of partial sequences of cytochrome b for 28 species found one cluster containing representative amblyopine and oxudercine species associated with three

northern Pacific gobionelline species (Akihito et al., 2000). An analysis of complete sequences of the 12s rRNA gene (Wang et al., 2001) revealed two monophyletic groups of gobies – one clade included species that were historically included in the Gobionellinae, Sicydiinae and

Oxudercinae and the other clade consisted of Gobiinae and Microdesmidae species. Thacker

(2003) examined complete sequences of three other mitochondrial genes (ND1, ND2 and COI) for 67 species in 51 genera. The resulting phylogeny revealed two major clades of gobies – one including the Gobionellinae, Kraemeriidae, Amblyopinae, Sicydiinae and Oxudercinae and the other Gobiinae, Ptereleotridae and Microdesmidae. In a subsequent analysis, Thacker (2009) added another mitochondrial gene and increased the number of included species to 107. It revealed the same large split with sicydiines, amblyopines and oxudercines aligned with gobionellines (the clade then recognized as the Gobionellidae), and gobiines in a clade with the

Schindleriidae, Microdesmidae and Ptereleotridae (recognized by Thacker (2009) as the

Gobiidae sensu stricto). There were several significant differences in detail between the two studies: the Kraemeriidae were recovered within the Gobiinae clade in 2009; several groups experienced changes in phylogenetic status – either shifting from paraphyly or polyphyly in the first study to monophyly in the second study (e.g. Microdesmidae, Ptereleotridae) or vice versa

(e.g. non-Stenogobius group gobionellines); and unlike the first study, gobiines were not divided into New World and Old World clades. Chakrabarty et al. (2012) reanalysed Thacker’s (2009) dataset and added several basal troglobitic gobioid species, however the relationships within the

Gobiidae remained largely unchanged from Thacker (2009). More recently Thacker (2013) reexamined relationships between Pomatoschistus and their European sand goby allies, and the gobionellines, oxudercines, sicydiines and amblyopines using mtDNA and morphological characters from the suspensorium. This study provided additional support for the deep divide in

gobies and presented morphological and molecular evidence that sand gobies are more closely related to gobionellines, amblyopines, sicydiines and oxudercines than they are to gobiines

(Thacker, 2013). Each of the mtDNA studies above differ not only in the extent of taxon sampling and the genes sequenced, but also frequently differ in the method of phylogenetic inference (i.e. Bayesian inference, maximum parsimony, maximum likelihood). To account for these differences and present a synopsis of some of their results, Agorreta & Rüber (2012) conducted a standardized reanalysis of several mtDNA datasets, the results of which reaffirmed the deep divide between a clade of gobiines, microdesmines, ptereleotrines, schindleriines and kraemeriines, and a clade of gobionellines, oxudercines, sicydiines and amblyopines (Fig. 1).

Figure 1 Synopsis of gobioid family-group relationships based on the most comprehensive mitochondrial dataset (modified from Agorreta& Rüber, 2012) and our concatenated nuclear dataset. Closed circles under branches of mtDNA tree indicate Bayesian posterior probabilities ≥

0.95. The * above branches of the nDNA tree depict strong support (≥ 0.85 posterior probability) for both concatenated and RAG1 datasets, whereas the * below branches depict strong support from both concatenated and rhodopsin datasets. Terminal branches comprised of groups of genera within the Gobiinae are identified by common names given by Thacker & Roje (2011).

The ‘?’ after the clade of Tiny Banded gobies indicates that this group was only monophyletic in the RAG1 analysis.

Compared with studies on other diverse groups of fishes (i.e. cyprinids, cichlids, labrids), phylogenetic studies on gobies have been slow to embrace the use of nuclear loci in deep-scale phylogenetic studies (Rüber & Agorreta, 2011). Protein-coding nuclear loci on the whole have slower mutation rates than mtDNA and are less likely to suffer from codon saturation and loss of

phylogenetic signal due to extensive homoplasy, thus making them more suitable for deep-scale phylogenetic analyses. On the other hand, nuclear loci that are too conserved may not have enough polymorphisms to resolve periods of rapid diversification that occurred in the past. Both points are illustrated in a phylogenetic study of all hypothesized orders of osteichthyan fishes using 20 nuclear genes and one mitochondrial gene (Betancur-R et al., 2013). The 1410 ingroup species included 41 gobioid fishes (following the classification of Gill & Mooi (2012), not that presented by Betancur-R et al., 2013). Because the analysis was directed at higher taxonomic levels, within-family representation was neither broad nor deep. The Gobiidae were represented by 37 species, five of which represented gobionellines, one was an oxudercine, six were microdesmines and the remaining 25 were gobiines. Greatest resolution of the osteichthyan phylogeny was found at the ordinal level or higher. Within the Gobiidae the five gobionellines and oxudercine formed one clade and the gobiines and microdesmines formed a sister clade, consistent with mtDNA topologies. Thirteen genera were represented by more than one species, but only six were recovered as monophyletic. Assuming correct identifications at the genus level, the poor return for some of the genera, at least, (e.g. Amblygobius, Eviota) may be due to an insensitivity of the genes to recent diversification.

Of the eight most prominent molecular phylogenetic studies addressing familial and sub- familial relationships of gobioids (Akihito et al., 2000; Wang et al., 2001; Thacker, 2003, 2009,

2013; Thacker & Hardman, 2005; Thacker & Roje, 2011; Agorreta & Rüber, 2012), only one study by Thacker & Roje (2011) utilized slow-evolving nuclear genes to infer phylogenetic relationships. The latter study focused on the intergeneric relationships within the Gobiidae sensu Thacker (2009) – i.e. the Gobiinae, Microdesmidae, Kraemeriidae, Schindleriidae,

Ptereleotridae and did not explicitly address the deep divide within the Gobiidae sensu Gill &

Mooi (2012) or other relationships to basal gobioids. Fourteen additional studies have used nuclear loci (often in combination with mtDNA) to infer phylogenetic relationships within genera or between closely related genera of gobies (Table 1). A total of 12 nuclear loci have been sequenced in these studies. The most widely used nuclear loci, and subsequently the nuclear loci for which the most sequences are available in GenBank, are the Recombination Activating Gene

I (RAG1) and rhodopsin (Rüber & Agorreta, 2011). The purpose of the present study is to offer a new phylogenetic analysis of family-level relationships within the Gobiidae using an independent dataset generated from the two nuclear genes RAG1 and rhodopsin. We present a phylogenetic hypothesis generated from a concatenated matrix of 77 newly sequenced terminal taxa plus 14 taxa for which both RAG1 and rhodopsin sequences are deposited in GenBank. We also generate individual gene trees for these two loci featuring expanded datasets for each gene that include representatives from every major goby group for which rhodopsin or RAG1 sequences are available in GenBank. In the discussion of the results of this study hereafter, we follow the most recent classification of the Gobiidae offered by Gill &Mooi (2012). For species within the Gobiidae however, we also follow Gill & Mooi (2012) in informally referencing some subfamilies and families recognized in the classification of Nelson (2006), e.g. kraemeriines, microdesmines, gobionellines, amblyopines and oxudercines, to help facilitate discussion of the various sub-groups, understanding that some may be paraphyletic or polyphyletic. We follow the classification of Milyeringidae as resurrected by Chakrabarty (2010). For other basal gobioids we also informally reference the family-level classification of Thacker (2009; Eleotridae,

Butidae, Odontobutidae) as eleotrids, butids, odontobutids, recognizing that these families have not been formally diagnosed by morphological characters (Pezold, 2011).

Table 1 Molecular studies on gobies using nuclear loci.

2. Materials and methods

Rhodopsin and RAG1 were sequenced for 77 specimens (Table 2). For all species sequenced in this study, total genomic DNA was extracted from ethanol-preserved tissue samples using the

Qiagen ® DNAeasy Blood and Tissue Kit (Qiagen, Valencia, California). Rhodopsin was targeted using primers from Chen et al. (2003): Rh 193F, and either Rh 1073R or Rh 1039R.

RAG1 was targeted using the primers Rag1 2533F from López et al. (2004) and Rag1Ra from

Tornabene & Pezold (2011). Several specimens failed to amplify using these RAG1 primers, therefore a slightly shorter segment of RAG1 was amplified for these specimens using combinations of these original primers and primers designed from conserved regions slightly downstream from the original primers based on an alignment of existing gobioid RAG1 sequences. The internal primers are GnelR1F (5’ GATCTBGAGGAGGACATYRTGG 3’) and

GdaeR1R (5’ GCTCTCCASACRGGYTTCATYT 3’). Both genes were amplified via PCR using

GoTaq® Hotstart Master Mix (Promega, Madison, Wisconsin) with the following thermal profile: 2 minutes at 95◦C, followed by 35 cycles of 40 seconds at 95◦C, 40 seconds at 52–54◦C, and 90 seconds at 72◦C, followed by a single extra extension period of 5 minutes at 72◦C.

Amplification of PCR products was confirmed via gel electrophoresis using 1.5% agarose gel stained with SYBR® Green fluorescent dye. PCR purification and DNA sequencing was performed by Molecular Cloning Labs (MCLABS, San Francisco, CA).

To supplement our taxon sampling for our combined analysis, we included 14 representative species from goby groups that were poorly represented in our existing dataset and had both RAG1 and Rhodopsin sequences deposited in GenBank (marked with * on Fig. 2). For the individual gene-trees, we further expanded our single-gene datasets by including representative species that had sequences of either RAG1 or rhodopsin in GenBank, but not both.

For many genera and species there were multiple RAG1 or Rhodopsin sequences available in

GenBank (e.g. Elacatinus, Luciogobius), and in these cases we limited our representation to 1–3 individuals from each genus. Sequence assembly and alignment were performed using the program Sequencher ver. 4.8 (Gene Codes, Ann Arbor, MI) and alignments were double- checked by eye.

Best fitting substitution models were determined for each gene using the program jModelTest ver. 0.1.1 (Posada, 2008) based on Akaike Information Criterion (AIC; Akaike,

1973) scores. The appropriate base substitution models as determined by AIC scores were

HKY+Γ for RAG1, and GTR+I+ Γ for rhodopsin (Hasegawa et al., 1985; Tevaré, 1986; Reeves,

1992; Yang, 1994). Phylogeny was inferred for each dataset and the concatenated dataset (each gene receiving its own substitution model) using Bayesian methods in the program MrBayes ver.

3.2 (Ronquist et al., 2012). Two parallel Metropolis-coupled Markov Chain Monte Carlo runs were run for 20 000 000 generations for the concatenated dataset (10 000 000 for individual gene analyses), sampling trees every 1000 generations. All trees were rooted with Apogon, as recent phylogenetic analyses have hypothesized a close relationship between apogonids and gobioids

(Thacker, 2009; Near et al., 2012; Wainwright et al., 2012; Betancur-R et al., 2013). Stationarity of each MCMC run was assessed using the program Tracer ver. 1.5 (Rambaut & Drummond,

2007), and the parallel MCMCs were considered to have reached convergence when the average standard deviation of split frequencies for each analysis approached 0.01. All MCMC runs reach stationarity by 500 000 generations. To further confirm that each parallel MCMC run converged on trees with similar topologies, a maximum clade-credibility tree was generated from post- burnin trees of each run using the program TreeAnnotator (available at http://beast.bio.ed.ac.uk/TreeAnnotator) and compared with one another. Once it was visually confirmed that each run converged on a similar tree topology, the trees from both parallel runs were combined and a consensus tree was generated from the total sample of post-burnin trees.

Table 2 New sequences generated from this study.

3. Results

Our concatenated analysis was derived from a matrix consisting of 1665 aligned base pairs (bp) including 735 bp of the rhodopsin gene and 930 bp of RAG1. A total of 75 species of Gobiidae in 55 genera, and 14 species in 11 outgroup genera including an apogonid, a rhyacichthyid, a milyeringid, an odontobutid, several eleotrids and several butids were represented. Independent analysis of the rhodopsin gene alone included all species present in the concatenated alignment plus 62 additional sequences retrieved from GenBank, representing a total of 137 species in 89 genera of Gobiidae. Similar analysis of the RAG1 gene was performed using all species from the concatenated dataset plus 19 additional Gobiidae species from Gen- Bank, representing a total of

94 species in 65 genera of Gobiidae. GenBank accession numbers for all new sequences and corresponding voucher catalog numbers are given in Table 2.

Figure 2 Bayesian phylogeny of concatenated dataset. Support values are posterior probabilities.

Relationships with posterior probabilities less than .50 are collapsed. The * indicates taxa for which sequences came from GenBank.

Figure 3 Bayesian phylogeny of RAG1 dataset. Support values are posterior probabilities.

Relationships with posterior probabilities less than .50 are collapsed.

Figure 4A and 4B (below) Bayesian phylogeny of rhodopsin dataset. Support values are posterior probabilities. Relationships with posterior probabilities less than 0.50 are collapsed.

3. 1 Gobiidae

The hypothesis resulting from the concatenated dataset is shown in Fig. 2. The Gobiidae comprises two deeply divergent groups corresponding to the Gobionellidae and Gobiidae sensu

Thacker 2009 (henceforth Thacker’s ‘Gobionellidae’ and Thacker’s ‘Gobiidae’, respectively, to distinguish from the classification of Gill &Mooi, 2012). This division is seen in our combined analysis as well as in our individual gene trees (Figs 3, 4). Well-resolved clades within the

‘Gobionellidae’ from our combined dataset include: a clade containing Pomatoschistus, species from the Mugilogobius group (Larson, 2001), and species from the northern Pacific group

(Pezold, 1993); an amblyopine, oxudercine clade; and a Stenogobius group, sicydiine clade.

Similarly, each independent gene tree recovered a ‘Gobionellidae’ clade that includes a paraphyletic Stenogobius group and monophyletic sicydiine clade, which is in turn sister to a clade containing oxudercines and amblyopines. The clades at the base of the ‘Gobionellidae’ recovered by our single gene analyses differ in detail (see Relationships within ‘Gobionellidae’ sensu Thacker (2009) section below), but generally include Mugilogobius group gobionellines, northern Pacific gobionellines, and European sand gobies (Pomatoschistus and allies).

Within the clade consistent with Thacker’s ‘Gobiidae’, deep resolution is also lacking.

However, three major clades are well supported in our combined analysis (1.0 posterior probability): (i) a clade composed of Eviota, Gobiodon and Pleurosicya; (ii) a clade comprising

Bathygobius, Glossogobius, Schismatogobius, Istigobius, Acentrogobius, Drombus,

Favonigobius, Oplopomus and Exyrias; (iii) and a large clade containing Callogobius,

Lophogobius, Coryphopterus, Gladiogobius, Asterropteryx, Ptereleotris, Microdesmus,

Lythrypnus, Priolepis, Zosterisessor, Gobius, a monophyletic Benthophilini, a monophyletic

Gobiosomatini, Amblyogobius and Valenciennea. Support values for clades within Thacker’s

‘Gobiidae’ vary considerably between the two gene trees. Some groups were strongly supported in both trees, including a monophyletic Benthophilini, a monophyletic Gobiosomatini and the clade containing the coral gobies Gobiodon, Eviota and Pleurosicya. Both gene trees also contained a clade comprised of ptereleotrines and microdesmines. Representative kraemeriines were recovered within Thacker’s Gobiidae in the rhodopsin analysis, however kraemeriines were not included in the RAG1 or concatenated analyses.

3. 2 Basal relationships

Strong resolution was obtained for several basal nodes in the concatenated phylogeny, an exception being the recovery of a Rhyacichthys/Perccottus clade as part of a trichotomy

including the apogonid outgroup representative and a clade containing all other gobioid fishes.

Results from the two single gene analyses (Figs 3, 4) differed somewhat from one another. It is obvious from the support values of basal relationships that the topology and support at the basal portion of the concatenated tree was driven largely by information from RAG 1 (Fig. 3). For rhodopsin a basal polytomy was returned containing Apogon, a clade including the eleotrids, butids and gobiids, and a clade holding rhyacichthids, Perccottus and Milyeringa (Fig. 4). In the larger gobioid clade, our concatenated and RAG1 trees show Milyeringa basal to all other taxa and eleotrids sister to a clade containing reciprocally monophyletic butids and gobiids.

4. Discussion

4. 1 The deep divide in gobies

A deep divide within the family Gobiidae has long been suggested by several studies that presented independent evidence from morphological and molecular characters. Miller’s (1973) illumination of differences in the number of epurals was perhaps the first indication of the deep divide. Birdsong et al.’s (1988) exploration of the insertion patterns between dorsal fin pterygiophores and underlying vertebrae, and Harrison’s (1989) recognition of specialized characters regarding the palatine, pterygoid, and quadrate increased our understanding of which side of the deep divide various goby genera fell on. Pezold’s (1993) recognition that the combination of a shared single anterior interorbital pore and a single epural bone were found in a large number of gobies provided additional evidence for the deep divide in gobies and presented a hypothesis for a monophyletic Gobiinae. While each of these studies support a divide amongst gobies, understanding exactly which taxa belong on each side of the divide has been problematic due to the extensive plasticity and homoplasy in gobiid morphology – likely caused by the

tendency in gobies to evolve via reduction and simplification. Recent studies of mtDNA have also recognized the deep divide in gobies and have offered new hypotheses regarding the inclusion of various gobiid taxa in the two major goby clades (e.g. Thacker, 2003, 2009, 2013;

Agorreta & Rüber, 2012).

The present study complements the above studies by offering an independent hypothesis addressing the divide amongst gobies using nuclear genes. The results of our analyses agree in many regards with results from past phylogenetic hypotheses based on mtDNA (e.g. Thacker,

2009; Agorreta & Rüber, 2012), and in some regards with previous morphology-driven hypotheses (e.g. Miller, 1973; Birdsong et al., 1988; Harrison, 1989; Pezold, 1993); however some finer details of the relationships differ between studies. The following paragraphs discuss our results in the context of these previous hypotheses and point out areas where additional research is needed.

4. 2 Relationships within ‘Gobionellidae’ sensu Thacker (2009)

The most recent hypotheses based on mtDNA (Thacker, 2009; Agorreta & Rüber, 2012;

Thacker, 2013) identified a clade of Gobiidae that consisted of gobiines, microdesmines, ptereleotrines, kraemeriines, Schindleria, as well as a clade containing gobionellines, sicydiines, ambylopines, sand gobies and oxudercines. Within the less diverse of the two goby clades

(‘Gobionellidae’ sensu Thacker, 2009), the details of relationships vary between studies. Most of the discord lies in the basal relationships within the clade. Thacker’s (2009) hypothesis showed that the first clade to branch off in her ‘Gobionellidae’ contained Pandaka, Gobiopterus and

Acanthogobius. This was followed by a clade containing Mugilogobius, Gillichthys,

Typhlogobius, Eucyclogobius and Chaenogobius, all of which were sister to a clade containing oxudercines/amblyopines/sicydiines/Stenogobius group gobionellines. Support values for these

basal nodes in Thacker’s (2009) study were low (0.58 and 0.60 Bayesian posterior probability).

Agorreta & Rüber (2012) arrived at slightly different results. In their analysis, Mugilogobius instead was included in the first clade of the ‘Gobionellidae’ (which also included

Pomatoschistus), and not closely allied with Gillichthys, Typhlogobius, Eucyclogobius or

Chaenogobius; the latter genera formed their own distinct clade sister to the oxudercines/amblyopines/sicydiines/Stenogobius group gobionellines (Agorreta & Rüber, 2012, supplementary Fig. S12 (see supplemental material online). The analysis of Thacker (2013) found that the first clade of the ‘Gobionellidae’ contained Pomatoschistus, Pandaka and

Gobiopterus (named the Mugilogobius lineage), however Mugilogobius was not included in the molecular dataset. Results from our study also illustrate the uncertainty at the base of the

‘Gobionellidae’ clade. Our concatenated tree shows a well-supported clade with Pomatoschistus being sister to a clade containing Mugilogobius, Redigobius, Tridentiger and Acanthogobius with very low support (Fig. 2), and on both of our individual gene trees Pomatoschistus (and other

European sand gobies in the rhodopsin tree) form a polytomy with Redigobius, Mugilogobius and a monophyletic northern Pacific gobionelline group (Figs 3, 4). The sand goby genera

Pomatoschistus, Knipowitschia, Economidichthys, Gobiusculus were originally diagnosed by

McKay & Miller (1997). McKay & Miller (1997) analyzed head pore and papillae patterns and characters from the axial skeleton and suspensorium, and hypothesized that sand gobies were closely related to the European gobies Buenia, Lebetus, Deltentosteus, as well as Nesogobius (a genus whose classification has been uncertain; Larson, 2001; Pezold, 2011; Thacker & Roje,

2011; Thacker, 2013). While several members of European sand gobies were long considered to be gobiines in studies using both morphology and molecular data (Pezold, 1993; Larson, 2001;

Neilson & Stepien, 2009a; but see Pezold, 1993 regarding Knipowitschia), it is apparent from

this study and others (e.g. Agorreta & Rüber, 2012; Thacker, 2013) that Pomatoschistus, and perhaps Knipowitschia, Economidichthys and Gobiusculus, are more closely related to gobionellines. Perhaps the only firm conclusion that can be made from this study and past studies is that Pomatoschistus and their sand goby allies, the Mugilogobius group of gobionellines, and the northern Pacific group of gobionellines all form a clade within the ‘Gobionellidae’, and the monophyly and inter-relationships of each of these groups remains unclear.

There is considerable agreement between our study and those of Thacker (2009, 2013) and Agorreta & Rüber (2012) regarding the remaining relationships within Thacker’s

‘Gobionellidae’. Our results show a sister relationship between the oxudercines (mudskippers) and ambylopines (eel gobies), with this clade being sister to sicydiines and the paraphyletic

Stenogobius group of gobionellines. Both Thacker (2009, 2013) and Agorreta & Rüber (2012) recovered a similar relationship; however the amblyopines were nested within the oxudercines in those studies. We were unable to test the monophyly of oxudercines relative to amblyopines as our study did not include the genera that have recovered oxudercines as paraphyletic in past research (i.e. Boleophthalmus, Pseudapocryptes, Scartelaos). Our study also agreed with both

Thacker (2009, 2013) and Agorreta& Rüber (2012) in recovering a monophyletic Sicydiinae, with this clade being closely related to Awaous, Stenogobius and the remaining Stenogobius group gobionellines (Ctenogobius, Gnatholepis, Evorthodus, Oligolepis, Oxyurichthys,

Gobioides, Gobionellus), although the inclusion of Gobioides and Gobionellus in the latter clade was not well supported in our rhodopsin tree (Fig. 4).

The phylogenetic hypotheses from our study and others for the ‘Gobionellidae’ show similarities to several morphological hypotheses. Most notably, Harrison’s (1989) examination of the palatopterygoquadrate complex and other characters support monophyly for a clade

containing his ‘Ctenogobius lineage’ (Ctenogobius, Evorthodus, Gobioides, Oligolepis,

Gobionellus, Gnatholepis), an ‘Oxyurichthys lineage’ (Oxyurichthys, amblyopines, oxudercines),

Stenogobius, Awaous, and the sicydiines. These groups form a monophyletic clade in our study and others (Thacker, 2009; Agorreta & Rüber, 2012), however the relationships within this clade differ between Harrison’s (1989) hypothesis and the consensus from the molecular studies.

Harrison’s hypothesis (Harrison, 1989, fig. 11), which depicts ((Awaous, sicydiines)(Stenogobius(‘Ctenogobius lineage’, Oxyurichthys lineage))), could be rearranged as

(‘Oxyurichthys lineage’ (‘Ctenogobius lineage’ (Stenogobius (Awaous, sicydiines)))). This latter arrangement requires one additional step in comparison to the original topology, a reversal in the form of the shortening of papillae row d on the branch leading to (Stenogobius(Awaous,

Sicydiines)). This new arrangement closely resembles the molecular consensus for these groups, which differ mainly in that the genus Oxyurichthys is allied with genera comprising Harrison’s

‘Ctenogobius lineage’ in the molecular hypotheses rather than with the amblyopines/oxudercines, and Awaous and Stenogobius are sister in the molecular hypotheses rather than forming a grade with sicydiines.

Data from morphological studies support the monophyly of the Mugilogobius group and northern Pacific group gobionellines (Birdsong et al., 1988; Pezold, 1993; Larson, 2001), and these hypotheses cannot be rejected by our results. However, comparing hypotheses on how these groups are related to the remainder of the ‘Gobionellidae’ is problematic, as most morphological studies on gobionellines did not include sicydiines, amblyopines and oxudercines

(e.g. Larson, 2001; Pezold, 2004; but see Parenti & Thomas, 1998), and most molecular hypotheses that include sicydiines, amblyopines and oxudercines have only sparse representation from the Mugilogobius and northern pacific groups (e.g. this study; Thacker, 2009, 2013).

4. 3 Relationships within ‘Gobiidae’ sensu Thacker (2009) The more diverse of the two clades of gobies (i.e. the ‘Gobiidae’ sensu Thacker, 2009) has been shown by previous analyses of mtDNA (i.e. Thacker 2009; Agorreta & Rüber, 2012) as well as combined analysis of mtDNA and nuclear genes (Thacker & Roje, 2011) to include gobiines, kraemeriines, ptereleotrines, microdesmines and Schindleria. Our results agree with these studies regarding the inclusion of gobiines, ptereleotrines, kraemeriines and microdesmines in this clade, however we did not include Schindleria in our combined analysis as no nuclear genes were available for Schindleria in GenBank for individual gene analyses.

In the most comprehensive phylogeny of Thacker’s ‘Gobiidae’ to date, Thacker & Roje

(2011) identified 13 distinct lineages of gobiines with each lineage being characterized by combinations of genetic data (mtDNA and nuclear genes), shared morphological characters, shared ecological traits, or biogeographical patterns. Relationships between the 13 groups in their study were poorly resolved. Some of these groups are supported here, however other groups of

Thacker & Roje (2011) that were represented by fewer taxa in their study do not hold when additional taxa are analyzed here. In our figures we highlight several of these groups to facilitate comparison with the results of Thacker & Roje (2011). Specifically, results from our concatenated tree (Fig. 2) provide additional support for the monophyly of the lineages referred to by Thacker & Roje (2011) as the Coral Gobies, Ponto-Caspian gobies (Benthophilinae of

Neilson & Stepien, 2009a), Burrowing Paired Gobies, Crested Gobies, Inshore Gobies and

American Seven-spined Gobies (Gobiosomatini). We also recovered strong support for a monophyletic group containing the Wormfishes and Dartfishes (microdesmines, ptereleotrines; a relationship weakly supported by Thacker & Roje, 2011). Our taxon sampling does not allow us to test the reciprocal monophyly of ptereleotrines and microdesmines with our concatenated dataset. Microdesmines are resolved as a monophyletic group on the rhodopsin tree, but

monophyly of ptereleotrines could not be established (Fig. 4). Morphological studies also support the monophyly of microdesmines (Thacker, 2000; Gill & Mooi, 2010), with the latter study providing 10 synapomorphies for microdesmines. Some of these synapomorphies are present in Schindleria and may indicate a close relationship between the two groups (Gill &

Mooi, 2010).

Thacker & Roje (2011) also identified a clade of Indo- Pacific taxa referred to as the

Lagoon Gobies which received very low support in their analysis. Our concatenated hypothesis supports monophyly of a clade that includes many of Thacker & Roje’s (2011) Lagoon Gobies, but also contains Drombus (listed as a member of the Flapheaded Gobies by Thacker & Roje,

2011). Gladiogobius, a genus considered to belong to the Lagoon Gobies (Thacker & Roje,

2011), is instead allied with Asterropteryx in all three trees here. These results are not surprising, as neither Gladiogobius nor Drombus were included in the molecular phylogenetic analysis of

Thacker & Roje (2011). The Flapheaded Gobies (10 genera) were represented by a single genus

(Callogobius) in that study, and the Lagoon Gobies (24 genera) were represented by only nine genera, with statistical support for the latter group being very weak (Thacker & Roje, 2011). Our combined analysis did not support a monophyletic Tiny Banded Gobies group (Priolepis,

Lythrypnus and allies; Thacker & Roje, 2011), however this group was monophyletic on our

RAG1 gene tree but with very low posterior probability (Fig. 3). Our study also was the first molecular study to include Schismatogobius. Schismatogobius was previously considered to be a derived gobionelline (Pezold, 1993, 2011; Larson, 2001), however our RAG1 and concatenated phylogenies showed a close relationship to Inshore Gobies (Bathygobius, Glossogobius) and the rhodopsin phylogeny indicated an affinity with Kraemeria (Figs 2–4).

Most of the GenBank sequences for the rhodopsin dataset came from the study by

Thacker & Roje (2011), and unsurprisingly our rhodopsin analysis recovered many of the same lineages of gobies within the ‘Gobiidae’ clade as their study (e.g. monophyletic Reef Shrimp

Gobies and Silt Shrimp Gobies clades). Fewer genera were recovered as monophyletic in the rhodopsin phylogeny than in the RAG1 or the combined phylogenies; 14 of 37 genera (38%) were returned as either paraphyletic or polyphyletic in the rhodopsin analysis compared with six of 22 (27%) in the RAG 1 analysis. This may be due to the fact we were attempting to resolve the relationships of the largest of our three datasets (in terms of taxa) with the least amount of nucleotide data. Alternatively, the greater number of genera represented by multiple species in the rhodopsin analysis may have affected these percentages. Lastly, the convergence of species through similarities derived due to selection by shared photic conditions of their habitat

(Larmuseau et al., 2010) cannot be ruled out as a potential reason for conflict between RAG1 and rhodopsin gene trees. Some relationships from our rhodopsin tree were surprising, including

Asterropteryx semipunctatus being nested within Fusigobius rather than with the two other

Asterropteryx representatives, and Lythrypnus elasson being nested deep within Priolepis rather than with L. zebra (Fig. 4). Both A. semipunctatus and L. elasson were GenBank sequences, thus we cannot rule out voucher misidentification as a source of these unexpected results.

No morphological studies have explicitly attempted to infer the phylogenetic relationships between all, or even most, of the genera that comprise the ‘Gobiidae’ sensu

Thacker (2009). The post-cranial osteological survey of these taxa by Birdsong et al. (1988) phenetically recognized several groups that belong to this clade. Some groups identified by

Birdsong et al. (1988) agree well with clades in our study (e.g. Gobiosoma group).Most of the taxa comprising Thacker’s ‘Gobiidae’ were either assigned to the diverse Priolepis group (54

genera), the Gobius group (five genera), or the Bathygobius group (eight genera)—none of which are supported as monophyletic by our results. These groups were distinguished from one another by vertebral counts – 26 for Priolepis group, 27 for Bathygobius group and 28 for

Gobius group members, reflecting differences in the number of precaudal vertebrae. Thus it appears that precaudal vertebrae number is phylogenetically uninformative within Thacker’s

‘Gobiidae’. The shared states of 3-22110 pterygiophore insertion pattern and a single epural do unite all of these genera, as well as the single anterior interorbital pore proposed as a synapomorphy by Pezold (1993).

4.4 Relationships of basal gobioids

Resolving basal gobioid relationships was not the focus of this study; however several observations can be made from our results and those of past studies. None of our datasets were able to resolve a monophyletic Gobioidei. This may suggest that either the outgroup is too distant, and or that the two genes used here are not conservative enough for resolution this deep in the phylogeny. Monophyly of gobioids is strongly supported by Betancur-R et al., 2013 who used 21 nuclear genes, and by Thacker (2009). The phylogenetic position of Rhyacichthys was poorly resolved here. Our concatenated and RAG1 analyses (Figs 2, 3) suggest a sister relationship between Rhyacichthys and odontobutids (represented by Perccottus in our analyses), with Milyeringa being resolved sister to the remaining gobioids (butids, eleotrids and gobiids).

The rhodopsin analysis (Fig. 4), however, recovered odontobutids sister to Milyeringa, with that clade being sister to rhyacichthyids. The results from our concatenated phylogeny for basal gobioid relationships are identical to those of the RAG1 analysis. Hypothesized relationships derived from mtDNA are also unclear. In the first molecular study focusing on basal gobioids,

Thacker & Hardman (2005) recovered Rhyacichthys as the lone basal gobioid, followed by a

clade of Milyeringa plus odontobutids. Thacker (2009) later recovered an unresolved trichotomy consisting of Rhyacichthys, Milyeringa and odontobutids. Comparisons of our results to the reanalysis by Agorreta & Rüber (2012) are problematic, because their study intentionally constrained Rhyacichthys to be the outgroup. Chakrabarty et al. (2012) reanalysed Thacker’s

(2009) dataset, but with different outgroups and with addition of the troglobitic sleeper

Typhleotris. Their results suggest that Rhyacichthys and odontobutids are sister to one another (in agreement with this study), but relationships between this clade, eleotrids and the blind sleepers

(Milyeringa, Typhleotris) are not well-resolved. Perhaps most importantly, Chakrabarty et al.

(2012) recovered two major clades of gobioids, one comprising rhyacichthyids, odontobutids, milyeringids and eleotrids, which were all sister to a clade comprising butids and the Gobiidae.

This arrangement differs from all previous molecular or morphological hypotheses of basal gobioid relationships. Larson et al. (2013) recommended against using the name Milyeringidae, citing the lack of data from nuclear genes as one of the primary reasons why they retained

Milyeringa in Eleotridae. While the exact placement of Milyeringa varies somewhat across studies, there is little evidence from mtDNA or nuclear genes that Milyeringa belongs to the

Eleotridae. Thus, we retain Milyeringidae until additional evidence refutes the monophyly of this group. The conflicting hypotheses between our study and previous molecular hypotheses for the relationships among basal gobioids can be attributed to several factors: (i) incomplete and incongruent taxon sampling of basal gobioid groups across studies; (ii) outgroup choice; and (iii) the use of fast-evolving mitochondrial genes (i.e. COI, Cyt-b, etc.) which are more subject to homoplasy when attempting to resolve deep phylogenetic relationships of taxa on long branches.

Lastly, the phylogenetic position of the Thalasseleotridae (Thalasseleotris, Grahamichthys) remains to be tested with molecular data. Gill & Mooi (2012) hypothesized that a monophyletic

Thalasseleotridae was sister to the Gobiidae based on the specialized shapes of the interhyal, posterior ceratohyal and urohyal, coupled with the absence of pharyngobranchial 4 and the absence of a dorsal postcleithrum. Representatives of Thalasseleotridae were not available for this study.

5. Conclusions and areas of in need of future research

This study presents another independent line of evidence supporting the division of gobiid diversity into two large clades, with one clade being more taxonomically diverse and extremely successful on reefs and coastal marine habitats, and the other clade being more ecologically divergent, but occurring primarily in estuarine and freshwater habitats. Our understanding of the relationships within these clades has steadily increased, and through this study and others we are beginning to see stability and clarity in some of the major relationships. The relationships between the sicydiines, oxudercines, amblyopines and gobionellines (the Stenogobius group) are becoming apparent, and represent one of the more stable limbs on the gobiid tree. Similarly, within Thacker’s ‘Gobiidae’ we continue to see many repeated patterns across studies, including the monophyly of many of Thacker & Roje’s (2011) 13 gobiid lineages (e.g. Gobiosomatini,

Benthophilini, Inshore Gobies, Coral Gobies, etc.). Nevertheless, as the backbone of the gobiid phylogeny becomes clearer, several key branches still require further investigation. For groups such as the basal members of Thacker’s ‘Gobionellidae’, i.e. the northern Pacific gobionellines plus the Mugilogobius group and European sand gobies, the solution to better phylogenetic resolution may simply be increased taxon sampling. For other poorly resolved relationships, such as the ‘bush’ atop Thacker’s ‘Gobiidae’, additional taxon sampling may represent only part of the solution, and achieving better resolution for such a diverse clade may also require the addition of sequence data from many more independent loci or novel suites of morphological

characters. A phylogenomic approach that takes advantage of high throughput sequencing to efficiently target large numbers of conserved protein-coding loci for hundreds of taxa (e.g. Li et al., 2013) represents a promising avenue for future work on gobies, as well as other taxonomically rich groups of fishes. This increase in the rate at which we can acquire substantial sequence data for numerous taxa will help alleviate the bottleneck at the sequencing step of molecular phylogenetics, but will create temporary bottlenecks in the steps involving data management, alignment and analysis. New developments in bioinformatics will undoubtedly target some of these obstacles with data, but ultimately the availability of fresh, high-quality tissue samples will be a limiting factor in massive phylogenomic analyses. This will place renewed emphasis on field collection and international collaborations – both of which will undoubtedly be required in the immediate future if we wish to tackle the remaining phylogenetic uncertainties in gobies. Lastly, while much emphasis is placed on efforts to increase the quantity and quality of DNA sequence data available for resolving phylogenetic uncertainties, there remains an equally pressing need for detailed explorations of new suites of morphological characters that will be phylogenetically informative. Through such studies, we can begin to move towards a new and stable classification of gobies that recognizes monophyletic groups that are ideally diagnosed by morphological characters. A recent example of this is the study by Gill &

Mooi (2012). Their investigation of new suites of morphological characters resulted in the discovery of two morphological synapomorphies for Gobiidae (the shape of the proximal heads of the pelvic-fin rays and the development of a prominent process on ceratobranchial 5) that are in agreement with results from recent molecular studies.

Acknowledgements

R. Chabarria helped collection specimens, assisted with laboratory work, and provided helpful comments throughout this study. D. Boseto, B. Lynch, G. Ahmadia and several other colleagues at Texas A&M University – Corpus Christi also helped collect many of the specimens. We thank

P. Chakrabarty, K. Piller and J. Van Tassell for contributing specimens and tissues. We gratefully acknowledge M.A. Fish at California Department of Fish & Wildlife for his help in acquiring specimens used in this study. We also thank B. Brown, B. Schelly and R. Arrindell at the American Museum of Natural History, D. Catania at California Academy of Sciences, M.

Sabaj-Perez at the Academy of Natural Sciences in Philadelphia, R. Robins at University of

Florida, and J. Williams and D. Pitassy at the National Museum of Natural History –

Smithsonian Institution for their help with loaning and depositing museum material. Special thanks to L. Rüber, A. Agorreta and two anonymous reviewers for their helpful comments on an earlier version of this manuscript. This project was funded in part by NSF-OISE 0080699 and

OISE-0553910 to FP.

References

Agorreta, A., Rüber, L., 2012. A standardized reanalysis of molecular phylogenetic hypotheses

of Gobioidei. Syst Biodivers 10, 375-390.

Akaike, H., 1973. Information theory as an extension of the maximum likelihood principle. In:

Petrov, B., Csaki, F. (Eds.), Second International Symposium of Information Theory.

Akademiai Kiado, Budapest, Hungary, pp. 267-281.

Akihito, P., Iwata, A., Kobayashi, T., Ikeo, K., Imanishi, T., Ono, H., Umehara, Y., Hamamatsu,

C., Sugiyama, K., Ikeda, Y., Sakamoto, K., Fumihito, A., Ohno, S., Gojobori, T., 2000.

Evolutionary aspects of gobioid fishes based upon a phylogenetic analysis of

mitochondrial cytochrome b genes. Gene 259, 5-15.

Betancur, R.R., Broughton, R.E., Wiley, E.O., Carpenter, K., Lopez, J.A., Li, C., Holcroft, N.I.,

Arcila, D., Sanciangco, M., Cureton Ii, J.C., Zhang, F., Buser, T., Campbell, M.A.,

Ballesteros, J.A., Roa-Varon, A., Willis, S., Borden, W.C., Rowley, T., Reneau, P.C.,

Hough, D.J., Lu, G., Grande, T., Arratia, G., Orti, G., 2013. The tree of life and a new

classification of bony fishes. PLoS currents 5.

Birdsong, R., Murdy, E., Pezold, F., 1988. A study of the vertebral column and median fin

osteology in gobioid fishes with comments on gobioid relationships. Bulletin of Marine

Science 42, 174-214.

Chakrabarty, P., 2010. Status and phylogeny of Milyeringidae (Teleostei: Gobiiformes) with the

description of a new blind cave-fish from Australia, Milyeringa brooksi, n. sp. Zootaxa

2557, 19-28.

Chakrabarty, P., Davis, M., Sparks, J., 2012. The first record of a trans-oceanic sister-group

relationship between obligate vertebrate troglobites. PLoS ONE 7, e44083.

Chen, W.J., Bonillo, C., Lecointre, G., 2003. Repeatability of clades as a criterion of reliability: a

case study for molecular phylogeny of Acanthomorpha (Teleostei) with larger number of

taxa. Molecular Phylogenetics and Evolution 26, 262-288.

Dawson, M.N., Staton, J., Jacobs, D.K., 2001. Phylogeography of the tidewater goby,

Eucyclogobius newberryi (Teleostei, Gobiidae), in coastal California. Evolution 55,

1167-1179.

Eschemeyer, W.N. 2013. Catalog of Fishes. California Academy of Sciences.

[http://research.calacademy.org/research/ichthyology/catalog/fishcatmain.asp, accessed

10 May 2013].

Gill, A., Mooi, R., 2010. Character evidence for the monophyly of the Microdesminae, with

comments on relationships to Schindleria (Teleostei: Gobioidei: Gobiidae). Zootaxa

2442, 51-59.

Gill, A., Mooi, R., 2012. Thalasseleotrididae, new family of marine gobioid fishes from New

Zealand and temperate Australia, with a revised definition of its sister taxon, the

Gobiidae (Teleostei: Acanthomorpha). Zootaxa 3266, 41-52.

Harrison, I., 1989. Specialization of the gobioid palatopterygoquadrate complex and its relevance

to gobioid systematics. J Nat Hist 23, 325-353.

Hasegawa, M., Kishino, H., Yano, T., 1985. Dating of the human-ape splitting by a molecular

clock of mitochondrial DNA. Journal of Molecular Evolution 22, 160-174.

Huyse, T., Van Houdt, J., Volckaert, F.A.M., 2004. Paleoclimatic history and vicariant

speciation in the "sand goby" group (Gobiidae, Teleostei). Molecular Phylogenetics and

Evolution 32, 324-336.

Keith, P., Lord, C., 2011. Systematics of Sicydiinae. In: Patzner, R.A., Van Tassel, J.L.,

Kovacic, M., Kapoor, B. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH,

pp. 119-128.

Keith, P., Lord, C., Lorion, J., Watanabe, S., Tsukamoto, K., Couloux, A., Dettae, A., 2011.

Phylogeny and biogeography of Sicydiinae (Teleostei: Gobiidae) inferred from

mitochondrial and nuclear genes. Mar Biol 158, 311-326.

Larmuseau, M.H.D., Raeymaekers, J.A.M., Ruddick, K.G., Van Houdt, J.K.J., Volckaert,

F.A.M., 2009. To see in different seas: spatial variation in the rhodopsin gene of the sand

goby (Pomatoschistus minutus). Molecular Ecology 18, 4227-4239.

Larmuseau, M.H.D., Vancampenhout, K., Raeymaekers, J.A.M., Van Houdt, J.K.J., Volckaert,

F.A.M., 2010. Differential modes of selection on the rhodopsin gene in coastal Baltic and

North Sea populations of the sand goby, Pomatoschistus minutus. Molecular Ecology 19,

2256-2268.

Larson, H.K., 2001. A revision of the gobiid fish genus Mugilogobius (Teleostei: Gobioidei) and

its systematic placement. Western Australian Museum, Perth, W.A.

Larson, H.K., Foster, R., Humphreys, W., Stevens, M., 2013. A new species of the blind cave

gudgeon Milyeringa (Pisces: Gobioidei, Eleotridae) from Barrow Island, Western

Australia, with a redescription of M. veritas Whitley. Zootaxa 3616, 135-150.

Li, C., Hofreieter, M., Straube, N., Corrigan, S., Naylor, G.J.P., 2013. Capturing protein-coding

genes across highly divergent species. BioTechniques 54, 321-326.

López, J., Chen, W., ORTÍ, G., Wood, R., 2004. Esociform phylogeny. Copeia 2004, 449-464.

Maxfield, J.M., Van Tassel, J.L., St Mary, C.M., Joyeux, J.C., Crow, K.D., 2012. Extreme

gender flexibility: Using a phylogenetic framework to infer the evolution of variation in

sex allocation, phylogeography, and speciation in a genus of bidirectional sex changing

fishes (Lythrypnus, Gobiidae). Molecular Phylogenetics and Evolution 64, 416-427.

Mckay, S., Miller, P., 1997. The affinities of European sand gobies (Teleostei: Gobiidae). J Nat

Hist 31, 1457-1482.

Miller, P., 1973. The osteology and adaptive features of Rhyacichthys aspro (Teleostie:

Gobioidei) and the classification of gobioid fishes. Journal of Zoology 171, 397-434.

Murdy, E., 2011. Systematics of Amblyopinae. In: Patzner, R.A., Van Tassel, J.L., Kovacic, M.,

Kapoor, B. (Eds.), The Biology of Gobies. Science Publishes, Enfield, NH, pp. 107-118.

Murdy, E., 2011. Systematics of Oxudercinae. In: Patzner, R.A., Van Tassel, J.L., Kovacic, M.,

Kapoor, B. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH, pp. 99-106.

Near, T.J., Eytan, R.I., Dornburg, A., Kuhn, K.L., Moore, J.A., Davis, M.P., Wainwright, P.C.,

Friedman, M., Smith, W.L., 2012. Resolution of ray-finned fish phylogeny and timing of

diversification. P Natl Acad Sci USA 109, 13698-13703.

Neilson, M.E., Stepien, C.A., 2009. Escape from the Ponto-Caspian: Evolution and

biogeography of an endemic goby species flock (Benthophilinae: Gobiidae: Teleostei).

Molecular Phylogenetics and Evolution 52, 84-102.

Neilson, M.E., Stepien, C.A., 2009. Evolution and phylogeography of the tubenose goby genus

Proterorhinus (Gobiidae: Teleostei): evidence for new cryptic species (vol 96, pg 664,

2009). Biological Journal of the Linnean Society 97, 708-708.

Nelson, J., 2006. Fishes of the World. 4th ed. John Wiley & Sons, Hoboken.

Parenti, L., Thomas, K., 1998. Pharyngeal jaw morphology and homology in sicydiine gobies

(Teleostei: Gobiidae) and allies. Bulletin of Marine Sciences 237, 257-274.

Pezold, F., 1993. Evidence for a Monophyletic Gobiinae. Copeia 3, 634-643.

Pezold, F., 2004. Phylogenetic analysis of the genus Gobionellus (Teleostei: Gobiidae). Copeia

2004, 260-280.

Pezold, F., 2011. Systematics of the Family Gobionellidae. In: Patzner, R., Van Tassell, J.L.,

Kovacic, M., Kapoor, B. (Eds.), The Biology of Gobies. CRC Press, Oxford & IBH

Publishing Co., New Delhi.

Posada, D., 2008. jModelTest: phylogenetic model averaging. Mol Biol Evol 25, 1253-1256.

Rambaut, A., Drummond, A. 2007. Tracer ver. 1.5. [http://tree.bio.ed.ac.uk/software/tracer/,

accessed 18 May 2013].

Reeves, J., 1992. Heterogeneity in the substitution process of amino acid site of proteins coded

for by mitochondrial DNA. Journal of Molecular Evolution 35, 17-31.

Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., Darling, A., Hohna, S., Larget, B.,

Liu, L., Suchard, M.A., Huelsenbeck, J.P., 2012. MrBayes 3.2: Efficient Bayesian

phylogenetic inference and model choice across a large model space. Syst Biol 61, 539-

542.

Rüber, L., Agorreta, A., 2011. Molecular systematics of gobioid fishes. In: Patzner, R.A., Van

Tassel, J.L., Kovacic, M., Kapoor, B. (Eds.), Biology of Gobies. Science Publishers,

Enfield, NH, pp. 23-50.

Taylor, M., Hellberg, M., 2005. Marine radiations at small geographic scales: speciation in

Neotropical reef gobies (Elacatinus). Evolution 59, 374-385.

Taylor, M., Hellberg, M., 2006. Comparative phylogeography in a genus of coral reef fishes:

biogeographical and genetical concordance in the Caribbean. Molecular Ecology 15, 695-

707.

Tevaré, S., 1986. Some probabilistic and statistical problems in the analysis of DNA sequences.

Lectures on Mathematics in the Life Sciences 17, 57-86.

Thacker, C., 2000. Phylogeny of the wormfishes (Teleostei : Gobioidei : Microdesmidae).

Copeia, 940-957.

Thacker, C.E., 2003. Molecular phylogeny of the gobioid fishes (Teleostei: Perciformes:

Gobioidei). Molecular Phylogenetics and Evolution 26, 354-368.

Thacker, C.E., 2009. Phylogeny of Gobioidei and placement within Acanthomorpha, with a new

classification and investigation of diversification and character evolution. Copeia 1, 93-

104.

Thacker, C., Hardman, M., 2005. Molecular phylogeny of basal gobioid fishes: Rhyacichthyidae,

Odontobutidae, Xenisthmidae, Eleotridae (Teleostei: Perciformes: Gobioidei). Molecular

Phylogenetics and Evolution 37, 858-871.

Thacker, C.E., 2011. Systematics of Gobiidae. In: Patzner, R.A., Van Tassel, J.L., Kovacic, M.,

Kapoor, B. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH, pp. 129-138.

Thacker, C.E., 2013. Phylogenetic placement of the European sand gobies in Gobionellidae and

characterization of gobionellid lineages (Gobiiformes: Gobioidei). Zootaxa 3619, 369-

382.

Thacker, C.E., Roje, D.M., 2011. Phylogeny of Gobiidae and identification of gobiid lineages.

Syst Biodivers 9, 329-347.

Thacker, C.E., Thompson, A.R., Roje, D.M., 2011. Phylogeny and evolution of Indo-Pacific

shrimp-associated gobies (Gobiiformes: Gobiidae). Molecular Phylogenetics and

Evolution 59, 168-176.

Tornabene, L., Ahmadia, G., Berumen, M., Smith, D., Jompa, J., Pezold, F., 2013. Evolution of

microhabitat association and morphology in a diverse group of cryptobenthic coral reef

fishes (Teleostei: Gobiidae: Eviota). Molecular Phylogenetics and Evolution 66, 391-400.

Tornabene, L., Pezold, F., 2011. Phylogenetic analysis of Western Atlantic Bathygobius

(Teleostei: Gobiidae). Zootaxa 3042, 27-36.

Van Tassell, J.L., Tornabene, L., Taylor, M., 2011. A History of Gobioid Morphological

Systematics. In: Patzner, R., Van Tassell, J.L., Kovacic, M., Kapoor, B. (Eds.), The

Biology of Gobies. Science Publishers, Enfield, NH, pp. 3-22.

Wainwright, P.C., Smith, W.L., Price, S.A., Tang, K.L., Sparks, J.S., Ferry, L.A., Kuhn, K.L.,

Eytan, R.I., Near, T.J., 2012. The Evolution of Pharyngognathy: A Phylogenetic and

Functional Appraisal of the Pharyngeal Jaw Key Innovation in Labroid Fishes and

Beyond. Systematic Biology 61, 1001-1027.

Wang, H.-Y., Tsai, M.-P., Dean, J., Lee, S.-C., 2001. Molecular phylogeny of gobioid fishes

(Perciformes: Gobioidei) based on mitochondrial 12s rRNA sequences. Molecular

Phylogenetics and Evolution 20, 390-408.

Yamada, T., Sugiyama, T., Tamaki, N., Kawakita, A., Kato, M., 2009. Adaptive radiation of

gobies in the interstitial habitats of gravel beaches accompanied by body elongation and

excessive vertebral segmentation. Bmc Evolutionary Biology 9.

Yang, Z., 1994. Maximum likelihood phylogenetic estimation from DNA sequences with

variable rates over sites: approximate methods. Journal of Molecular Evolution 39, 306-

314.

CHAPTER II – Evolution of microhabitat association and morphology in a diverse group of cryptobenthic coral reef fishes (Teleostei: Gobiidae: Eviota)

Abstract

Gobies (Teleostei: Gobiidae) are an extremely diverse and widely distributed group and are the second most species rich family of vertebrates. Ecological drivers are key to the evolutionary success of the Gobiidae. However, ecological and phylogenetic data are lacking for many diverse genera of gobies. Our study investigated the evolution of microhabitat association across the phylogeny of 18 species of dwarfgobies (genus Eviota), an abundant and diverse group of coral reef fishes. We also explore the evolution of pectoral fin-ray branching and sensory head pores to determine the relationship between morphological evolution and microhabitat shifts. Our results demonstrate that Eviota species switched multiple times from a facultative hard-coral association to inhabiting rubble or mixed sand/rubble habitat. We found no obvious relationship between microhabitat shifts and changes in pectoral fin-ray branching or reduction in sensory pores, with the latter character being highly homoplasious throughout the genus. The relative flexibility in coral-association in Eviota combined with the ability to move into non-coral habitats suggests a genetic capacity for ecological release in contrast to the strict obligate coral-dwelling relationship commonly observed in closely related coral gobies, thus promoting co-existence through fine scale niche partitioning. The variation in microhabitat association may facilitate opportunistic ecological speciation, and species persistence in the face of environmental change. This increased speciation opportunity, in concert with a high resilience to extinction, may explain the exceptionally high diversity seen in Eviota compared to related genera in the family.

1. Introduction

The group of fishes known as gobies (family Gobiidae sensu Gill and Mooi, 2012) is one of the most diverse families of marine fishes in the world, and is surpassed only by the fish family

Cyprinidae in being the most species-rich family of vertebrates. The family Gobiidae is formally diagnosed by the number of branchiostegals present and the unique osteology of their ceratobranchials and pelvic fin-ray hemitrich (Gill and Mooi, 2012), however most gobies are commonly recognized by their small size, cryptobenthic nature, and unique pelvic fin upon which they often perch. There are more than 1600 species of gobies across over 200 genera

(Thacker, 2003) with representatives in virtually all coastal marine, brackish, tidal freshwater and oceanic isle freshwater habitats with the exception of the polar seas. The diversity seen in gobies is made possible by extensive adaptations within the family, which has enabled finescale ecological niche utilization and microhabitat specialization (e.g. Yamada et al., 2009). Owing to their remarkable taxonomic and ecological diversity, gobies have been a model group for studies on rapid speciation and adaptive radiation (Rüber et al., 2003; Taylor and Hellberg, 2005;

Yamada et al., 2009).

1.1. Ecological drivers of goby evolution

Molecular phylogenies of gobies have greatly increased our understanding of the patterns of evolution and the distribution of diversity within the family Gobiidae. The most comprehensive phylogeny to date showed that the family Gobiidae comprises two large monophyletic groups across which species richness is asymmetrically distributed (Thacker, 2009). It has been suggested that the disproportionately large number of species in the more diverse of the two clades (recognized as Gobiidae sensu stricto by Thacker (2009)) may be driven by a habitat switch from freshwater to the marine environment, creating both ecological and spatial

opportunities for subsequent speciation in a new, expansive and complex environment (Thacker,

2009). A more recent analysis of the Gobiidae sensu Thacker (2009) revealed it to comprise 13 distinct lineages, each united by a combination of morphological, molecular, biogeographic, ecological, or behavioral characteristics (i.e. coral gobies, lagoon gobies, reef-associated shrimp gobies, etc.; Thacker and Roje, 2011).

Ties between ecological shifts and species diversification are obvious within several subgroups of gobies. Neogobiines in the subfamily Benthophilinae (Ponto-Caspian &

Mediterranean lineage of Thacker and Roje (2011)) achieved remarkable diversity via a shift from saltwater to freshwater, promoting subsequent allopatric speciation by isolation of freshwater basins (Neilson and Stepien, 2009). Speciation within mudskippers (genera

Boleophthalmus, Oxuderces, Periophthalmodon, Periophthalmus, Scartelaous and Zappa) was facilitated by differentiation into several ecological guilds characterized by differences in salinity, water quality and habitat terrestriality (Polgar et al., 2010). Interstitial gobies of Japan

(genus Luciogobius) underwent adaptive radiation and morphological diversification driven largely by ecological differentiation into interstitial habitats with differing sediment properties

(Yamada et al., 2009). Recent speciation within the neon gobies of the Caribbean genus

Elacatinus occurred in distinct stages, with several stages being driven in part by ecology. Early in their evolutionary history, neon gobies first segregated into groups occurring on sponge versus coral microhabitat, followed by differentiation into different feeding strategies, colors and morphologies (Taylor and Hellberg, 2005; Colin, 2010). Thus, studies of diversification within and among goby genera suggest that ecological plasticity in gobies has occurred throughout the evolutionary history of the group and is not limited to ancient timelines.

1.2. Indo-pacific dwarfgobies

The coral gobies (Thacker and Roje, 2011) comprise 13 genera, most of which occur on coral reefs and have obligate or facultative relationships with live coral (Herler, 2007; Herler et al.,

2009; Thacker and Roje, 2011). While most genera of coral gobies include fewer than 20 species, a notable exception is the genus Eviota, commonly known as dwarfgobies. There are over 60 valid species of Eviota and many more species have yet to be described (Gill and Jewett,

2004; Shibukawa and Suzuki, 2005; Herler et al., 2009; Froese and Pauly, 2012). In addition to their exceptional species richness, the rapid life-cycle, high fecundity and overall abundance of

Eviota make them an important component of the trophic webs on coral reefs (Depczynski and

Bellwood, 2003, 2005, 2006). Past studies on coral goby habitat associations suggested that, in comparison to several closely related genera of obligate coral dwelling species (Gobiodon,

Pleurosicya), some species of Eviota have less stringent associations with live coral and do not possess species-specific coral associations (Herler and Hilgers, 2005). Prior studies featured only a few Eviota species, and little is known about the ecology of the vast majority of the genus.

Eviota species are miniscule in size (typically 10–20 mm in standard length) and reliable diagnostic morphological characters are scarce, likely due to the high degree of reduction, simplification or loss associated with miniaturization. The presence or absence of branching pectoral fin rays and sensory pore patterns on the head are frequently used to diagnose species of

Eviota (Lachner and Karnella, 1980). Pectoral fin morphology is highly specialized in gobiids, as these fins are used primarily for propulsion (Adriaens et al., 1993). In addition to swimming, many benthic fishes also utilize pectoral fin rays for maintaining position against a current, or gripping and manipulating the substrate (Webb, 1989; Brandstätter et al., 1990; Taft, 2011).

Thus, the presence of branched versus unbranched pectoral rays may alter the way Eviota species

perform these actions and could potentially relate to the microhabitat with which they associate.

Sensory pores of gobies are an extension of the lateral line system, which has long been known to be adapted to specific habitats or lifestyles of fishes (Coombs et al., 1988). Accessory lateral- line systems (sensory pores, sensory papillae) in gobies have been widely used as taxonomic characters, however variation (often via reduction) in these systems both within and between species may be related to differences in microhabitat characteristics such as turbidity, flow and location in the water column (Ahnelt, 1995; Ahnelt et al., 1995, 2004; Ahnelt and Scattolin,

2003; Ahnelt and Bohacek, 2004; Stelbrink and Freyhof, 2006; Asaoka et al., 2011).

The purpose of the present study is to investigate the evolution of microhabitat association in Eviota species. Additionally, we examine the evolution of sensory pores and pectoral fin-ray branching to explore the relationship between morphology and microhabitat use, and ultimately evaluate the relative role of both microhabitat and morphology in the diversification of Eviota. We infer phylogenetic relationships of 18 species of Eviota collected from three regions of the Pacific Ocean and the Red Sea using molecular data from mitochondrial and nuclear DNA. The phylogenetic structure will serve as a framework to explore the association between species habitats and morphology across the evolutionary history of the group. Through this approach, we will be able to investigate the following questions: (i) are

Eviota species that occur in similar habitats closely related to one another, or has the use of specific microhabitats evolved several times independently; (ii) are patterns of pectoral fin-ray branching and sensory pores homologous in Eviota; (iii) is the evolution of microhabitat association correlated with the evolution of pectoral fin-ray branching and sensory pore patterns;

(iv) and did Eviota diversification occur in ‘‘stages’’, similar to that of the Atlantic neon gobies

Elacatinus, with each stage of speciation being characterized either by a unique ecological shift or a morphological change (Streelman and Danley, 2003; Taylor and Hellberg, 2005).

2. Methods

2.1. Collecting specimens and characterizing habitat association

Fishes were collected from four regions throughout the geographic range of Eviota species:

Saudi Arabia (Red Sea), French Polynesia (Moorea), Indonesia (Hoga, southeast Sulawesi), and

Micronesia (Pohnpei). All collections were from shallow fringing and patch reefs (<10 m). Past studies on habitat specificity of coral gobies demonstrated that unlike species of Gobiodon,

Bryaninops and Pleurosicya, the few Eviota species observed did not show obvious species- specific coral preferences (Herler and Hilgers, 2005; Herler, 2007). Therefore, our collections did not target specific species of coral, but instead focused on three broad microhabitats of the reef: sand, rubble and all hard coral. We sampled 106 hard coral, 57 rubble, and 63 sand microhabitats for a total of 229 sample events; 64 from Indonesia, 84 from Moorea, 63 from

Pohnpei, and 18 from the Red Sea. Microhabitats were sampled using a circular fine mesh net (1 m diameter) covered with a plastic sheet weighted on the edges. The plastic sheet was used to contain fish anesthetic, a 100 mL mixture of a four parts ethanol to one part clove oil. Anesthetic was dispersed within the sample area using two 60 mL syringes. The anesthetic remained in the net for 1 min, then the plastic sheet and net were slowly removed and divers collected anesthetized fishes for 5–10 min. Fishes in crevices and holes were dislodged with forceps and a dental pick. Specimens were placed in ice slurry to preserve pigmentation for photographs to help in taxonomic identification. Specimens were then preserved whole in 95% ethanol.

A species was determined to be associated with a specific habitat type when at least 60% of the total number of specimens of that species were collected from that habitat. If species were not observed to have associations to one microhabitat type, than they were considered associates of multiple microhabitats and categorized accordingly.

Table 3 Specimens examined for microhabitat analysis. DII/A = rays of second dorsal and anal

fins.Table 1 Specimens examined. Acronyms for catalog numbers: AMNH, American Museum of Natural History; CAS, California Academy of Sciences; MZB Museum Zoologicum Bogoriense. DII/A = number of second dorsal and anal fin rays.

Genbank Accession Species Country Latitude Longitude Numbers Catalog number Notes on identification

Eviota French Polynesia JX483989, 5°28'28.04"S 123°45'29.84"E AMNH 256486 albolineata (Moorea) JX483966

9/8 elements in DII/A; 5th pelvic ray ≈ 40% of 4th ray; Eviota cf. JX483984, Micronesia (Pohnpei) 6°57'55.79"N 158°18'22.74"E CAS 234536 pore pattern 2; pectoral rays unbranched; but lacks spilota JX483970 diagnostic spot on pectoral base present in E. spilota

Eviota Red Sea JX483977, 8/9 elements in DII/A; 5th pelvic ray ≈ 10% of 4th ray; pore Saudi Arabia (Red Sea) 21°29'37.06"N 39° 8'34.43"E CAS 234525 sp. 1 JX483972 pattern 1; pectoral rays branched

Eviota Red Sea JX483997, 9/9 elements in DII/A; 5th pelvic ray ≈ 10% of 4th ray; pore Saudi Arabia (Red Sea) 21°29'37.06"N 39° 8'34.43"E CAS 234526 sp. 2 JX483975 pattern 1; pectoral rays branched

Eviota Red Sea JX483978, 9/9 elements in DII/A; 5th pelvic ray ≈ 20-30% of 4th ray; Saudi Arabia (Red Sea) 21°29'37.06"N 39° 8'34.43"E CAS 234527 sp. 3 JX483969 pore pattern 3; pectoral rays heavily damaged

Resembles E. guttata in all counts, measurements, and Eviota cf. JX483993, Saudi Arabia (Red Sea) 21°29'37.06"N 39° 8'34.43"E CAS 234528 pigment patterns but possesses pore pattern 3 vs. pattern 1 guttata JX483963 as in E. guttata

Eviota French Polynesia JX483953, 5°28'28.04"S 123°45'29.84"E CAS 234535 epiphanes (Moorea) JX483988

JX483998, Eviota guttata Indonesia 5°28'28.04"S 123°45'29.84"E MZB 29011 JX483955

French Polynesia JX483979, Eviota infulata 17°28'58.17"S 149°48'56.36"W AMNH 256487 (Moorea) JX483959

Eviota JX483983, Micronesia (Pohnpei) 6°57'55.79"N 158°18'22.74"E CAS 234529 lachdeberei JX483968

Eviota JX483985, Micronesia (Pohnpei) 6°58'10.56"N 158° 9'15.12"E CAS 234530 punctulata JX483954

Eviota JX483996, Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20912 punctulata JX483956

Eviota JX483986, Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20913 queenslandica JX483961

JX483994, Eviota sigillata Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20914 JX483965

JX483995, Eviota sigillata Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20915 JX483971

JX483976, Eviota spilota Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20916 JX483967

Eviota JX483987, Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20917 winterbottomi JX483974

9/9 elements in DII/A; pelvic fin broken; pore pattern 2; Eviota Pohnpei JX483980, Micronesia (Pohnpei) 6°59'30.33"N 158°14'30.36"E CAS 234531 pectoral rays branched; prominent spot on caudal peduncle; sp. 1 JX483973 no vertical bars on cheek

Eviota JX483990, Micronesia (Pohnpei) 6°58'10.56"N 158° 9'15.12"E CAS 234532 melasma JX483960

Eviota JX483982, Micronesia (Pohnpei) 6°59'30.33"N 158°14'30.36"E CAS 234533 shimadai JX483964

Bryaninops French Polynesia JX483991, 17°28'31.46"S 149°48'27.71"W AMNH 256488 ridens (Moorea) JX483957

Gobiodon JX483992, Indonesia 5°28'28.04"S 123°45'29.84"E MZB 20918 unicolor JX483958

Asterropteryx JX483981, Micronesia (Pohnpei) 6°59'30.33"N 158°14'30.36"E CAS 234534 ensifera JX483962

2.2. Specimen identification and morphological character coding

Specimens of Eviota are notoriously difficult to identify partially because of their diminutive size

(typically 10–20 mm), the large number of undescribed species, and the lack of a comprehensive published dichotomous key. As mentioned above, there are a very limited number of morphological characters useful for diagnosing species. An unpublished draft of a dichotomous key for all valid Eviota species plus several undescribed species was graciously provided by

Richard Winterbottom and Dave Greenfield to supplement original species descriptions and assist in our identifications. Nevertheless, some species in our phylogeny did not perfectly fit the current species descriptions and may represent additional undescribed species. All specimens in this study are listed in Table 3 along with comments on identifications. GenBank Accession numbers and museum catalog numbers are also provided.

The morphology of pectoral fin-rays in Eviota fall into two general categories; branched and unbranched. Branched fin-rays typically begin as a single rigid ray at the base of the fin and bifurcate into two or more branches towards the distal tips of the ray. The number of branched rays is variable among species, and most branched rays are located in the central portion of the fin, whereas the rays on the dorsal and ventral portions of the rays usually show less branching.

For our analysis, species were coded as ‘‘branched’’ if any of the rays on the fin were branched, or ‘‘unbranched’’ if all rays were unbranched. Coding of sensory pore patterns follows terminology of Lachner and Karnella (1980), with pattern 1 being a ‘‘complete’’ pattern with all head pores present, and all other patterns having one or several pores absent or fused (Fig. 5).

Figure 5 Typical arrangement of sensory pore pattern I. Pore pattern 2 is missing the IT pore.

Pore pattern 3 is missing the IT and PITO pores, and has an enlarged or paired AITO pore. Pore pattern 4 is missing the IT pore and the NA pores. For definition of abbreviations, see Lachner and Karnella (1980). Figure modified from Lachner and Karnella (1980).

2.3. DNA sequencing

Total genomic DNA was extracted from ethanol-preserved tissue samples using the Qiagen

#DNAeasy Blood and Tissue Kit (Qiagen, Valencia, California). Partial mitochondrial cytochrome oxidase I (COI) sequences were targeted using the primers GOBYL6468 and

GOBYH7696 (Thacker, 2003). A segment of the nuclear gene protease III (Ptr) was targeted using the primers PtrF2 and PtrR2 (Yamada et al., 2009). Heterozygous sites in the nuclear data were coded with IUPAC ambiguity codes. Segments of COI were amplified via PCR using

GoTaq# Hotstart Master Mix (Promega, Madison, Wisconsin) with the following thermal profile: 2 min at 95 °C, followed by 35 cycles of 40 s at 95 °C, 40 s at 52–54 °C, and 90 s at 72

°C, followed by a single extra extension period of 5 min at 72 °C. Thermal profile for PCR of the

Ptr gene was identical to that of the COI with the exception of an annealing temperature of 55

°C. Amplified PCR products were verified via gel electrophoresis using 2% agarose gel stained with SYBR® Green fluorescent dye. Purification of PCR products and DNA sequencing was performed by Molecular Cloning Labs (MCLABS, San Francisco, California). Sequences were assembled and aligned using the program Sequencher ver.4.8 (Gene Codes, Ann Arbor,

Michigan) and alignments were checked by eye. Three outgroups were used in the analysis:

Bryaninops ridens, Gobiodon unicolor, and Asterropteryx ensifera. Both Bryaninops and

Gobiodon belong to a clade of coral reef gobies hypothesized to be sister to Eviota (Herler et al.,

2009; Thacker and Roje, 2011). Asterropteryx is more distantly related to Eviota but belongs to the same major gobiid clade (Gobiidae sensu Thacker, 2009; Gobiinae sensu Pezold, 1993).

2.4. Phylogenetic analysis

Best fitting substitution models were determined for each gene using the program jModelTest ver.0.1.1 (Posada, 2008) based on Akaike Information Criterion (AIC) scores. Phylogeny was inferred for each dataset independently as well as a concatenated matrix using Bayesian methods in the program MrBayes ver.3.2 (Ronquist et al., 2012). For each analysis, two parallel

Metropolis-coupled Markov Chain Monte Carlo runs were generated for 10,000,000 iterations with a sampling frequency of 1000 iterations. Stationarity of each MCMC run was assessed using the program Tracer ver. 1.5 (Rambaut and Drummond, 2007), and the parallel MCMCs were considered to have reached convergence when the average standard deviation of split frequencies for each analysis approached 0.01. Preliminary MrBayes analyses indicated that the default branch length priors implemented in MrBayes caused MCMCs to fail, with runs frequently converging on local optima sampling trees with unrealistically long branches – an increasingly common phenomenon observed in Bayesian phylogenetics (Brown et al., 2010;

Marshall, 2010). This was remedied by setting the branch length prior to an unconstrained-

exponential prior with a mean of 0.01 (default mean in MrBayes is 0.1; Marshall, 2010). To further confirm that each parallel MCMC run converged on trees with similar topologies, a maximum-clade-credibility tree was generated from each set of post-burnin trees using the program TreeAnnotator (available at http://beast.bio.ed.ac.uk/TreeAnnotator) and compared to one another. Once it was visually confirmed that each run converged on a similar tree topology, the trees from both parallel runs were combined and a maximum-clade-credibility tree was generated from the total sample of post-burnin trees (Fig. 6).

2.5. Habitat and character mapping

To visualize branch lengths in units of time, an ultrametric phylogeny was generated using

BEAST ver. 1.7.0 (Drummond et al., 2012). Molecular data were partitioned by gene with each gene receiving its own substitution model as determined by the results of the model test.

Outgroups were removed from the alignment for the BEAST analysis. Each partition was assigned an uncorrelated-lognormal relaxed molecular clock model and a yule speciation prior was used for the analysis. The tree root height prior was arbitrarily set to 1.0 so that the units of time across the tree were relative to the age of the entire tree rather than absolute years. The

BEAST analysis was run for 10,000,000 iterations with a sampling frequency of 1000 generations. Output from the MCMC was analyzed using Tracer to determine that stationarity was reached, insure proper mixing, and to assess burnin. Two additional BEAST runs were performed to confirm that analyses converged on similar trees. A maximum-clade-credibility tree was generated from our post-burnin trees using TreeAnnotator.

Ancestral state reconstructions for habitat associations were estimated for nodes on the ultrametric phylogeny using maximum- likelihood in the software language R (R Development

Core Team, 2012) through the function ‘‘ace’’ in the APE package (Paradis et al., 2004). The

same method was used to estimate the likelihood of ancestral character states for pectoral fin branching and sensory pore patterns. For all ancestral state estimations, three different models of character state change were fit to our phylogenetic data: an equal-rates model, where all rates of change between states are equal; a symmetrical model, where the rate of change from state 1 →

2 = 2→1, from 1→3 = 3→1, and from 2→3 = 3→2; and a model where all rates of character state changes were different. The fit of each model was evaluated using AIC scores and models were compared using ANOVA.

3. Results

3.1. Phylogenetic analysis

We successfully sequenced 1089 bp of COI and 607 bp of Ptr for a total of 1696 bp. The

MCMCs for all Bayesian analyses reached stationarity by 300,000 iterations. The single-gene analysis of the Ptr dataset yielded a topology identical to the concatenated phylogeny (Fig. 6).

Ancestral relationships on COI gene tree were poorly resolved with many long branches forming a polytomy at the base of the phylogeny, however the few relationships that were well supported in the COI tree were identical to those of the Ptr gene tree and the concatenated phylogeny. The

Bayesian phylogeny of the concatenated dataset (Fig. 6) showed strong support for the monophyly of Eviota. Species within the genus were divided into two well-supported deeply divergent clades with each clade containing species from all four collection localities. Bayesian posterior probabilities at nodes were generally high across the tree with one exception; the position of the branch containing Eviota Pohnpei sp. 1 + Eviota Red Sea sp. 1 was poorly resolved. The inclusion of this lineage within the larger of the two clades of Eviota is well- supported, however its position within this group is either sister to all other species in this clade, or sister to a clade containing E. melasma, E. guttata, E. cf. guttata, and E. albolineata (as shown

in the ultrametric phylogenies, Figs. 7-9). Relationships within the smaller of the two major

Eviota clades are fully resolved with high posterior probability support values.

Figure 6 Bayesian phylogeny of Eviota based on the concatenated dataset. Support values at nodes are Bayesian posterior probabilities. Scale bar units = expected substitutions/site.

3.2. Microhabitat association and morphological character evolution

We collected a total of 374 specimens of Eviota comprising 18 species from 229 sampling events across our four localities (Table 4). Habitat associations of species from the Red Sea were based on only one individual for each species and additional collections are obviously needed to corroborate their associations with hard coral. Nevertheless, most species in our study demonstrated a strong preference for one microhabitat type. The exceptions are E. sigillata and

E. shimadai, which were evenly distributed across both rubble and sand microhabitats, and

therefore they were deemed to be associated with both habitats. Myers (1999) also notes that E. sigillata is commonly found over sand. When collected over rubble, both species were more common when there was also some sand present amongst the rubble as opposed to when there was only a deep layer of rubble (Ahmadia pers. observation). We therefore coded the combination of rubble and sand as a third microhabitat type.

Table 4 Designation of microhabitat associations of Eviota based on the percent of the total number of specimens collected from each microhabitat. Bold font indicates the dominant microhabitat type for each species and their percent occurrence on this particular microhabitat.

Percent occurrence

Microhabitat Total No.

Species Location Hard Coral Rubble Sand Association Specimens Eviota punctulata Indonesia 80% 20% 0% Hard Coral 5 Eviota queenslandica Indonesia 14% 81% 5% Rubble 80 Eviota sigillata Indonesia 0% 47% 53% Rubble/Sand 30 Eviota winterbottomi Indonesia 100% 0% 0% Hard Coral 22 Eviota guttata Indonesia 90% 10% 0% Hard Coral 10 Eviota spilota Indonesia 25% 75% 0% Rubble 4 Eviota Moorea sp. 1 Moorea 10% 85% 5% Rubble 20 Eviota infulata Moorea 67% 0% 33% Hard Coral 3 Eviota cf. spilota Pohnpei 5% 95% 0% Rubble 21 Eviota lachdeberei Pohnpei 16% 84% 0% Rubble 25 Eviota melasma Pohnpei 100% 0% 0% Hard Coral 4 Eviota Pohnpei sp.1 Pohnpei 26% 74% 0% Rubble 19 Eviota punctulata Pohnpei 61% 31% 7% Hard Coral 54 Eviota shimadai Pohnpei 0% 58% 42% Rubble/Sand 68 Eviota Red Sea sp.1 Red Sea 100% 0% 0% Hard Coral 1 Eviota Red Sea sp.2 Red Sea 100% 0% 0% Hard Coral 1 Eviota Red Sea sp.3 Red Sea 100% 0% 0% Hard Coral 1 Eviota cf. guttata Red Sea 100% 0% 0% Hard Coral 1

Figure 7 Chronogram of Eviota species showing distribution of microhabitat association. Pies at nodes are probabilities of ancestral character states. Scale bar is in units of time relative to the root age.

Figure 8 Chronogram of Eviota species showing distribution of sensory pore patterns. Pies at nodes are probabilities of ancestral character states. Scale bar is in units of time relative to the root age.

Figure 9 Chronogram of Eviota species showing distribution of pectoral fin branching. Pies at nodes are probabilities of ancestral character states. Scale bar is in units of time relative to the root age.

There was no significant difference between the fit of the equal-rates models, the symmetrical models and the all-rates-different models for any of the ancestral character state reconstructions (ANOVA, p > 0.05; delta AIC < 4), thus we used the simpler equal-rates model for our reconstructions. Ancestral state reconstruction for habitat association indicated that the common ancestor to all Eviota species was most likely a coral-associate (Fig. 7). Throughout the history of Eviota there were multiple independent habitat switches from coral to rubble or from coral to rubble/sand. There were no clear instances of reversals from rubble or sand/rubble back to coral, however several nodes have high likelihoods for more than one habitat association and thus the precise instance on the phylogeny of habitat transitions is ambiguous in some cases.

Ancestral state reconstruction for sensory pore patterns showed that most ancestral nodes had similar likelihoods for all four pore patterns, and that changes in pore patterns were common throughout the evolutionary history of Eviota (Fig. 4). Analysis of pectoral fin-ray branching patterns indicated that all species with and without branched pectoral fin-rays were reciprocally monophyletic, and that this character diagnoses the two main clades of Eviota in our phylogeny

(Fig. 5), hereafter referred to as the branched clade and the unbranched clade.

4. Discussion

4.1. Microhabitat association

Many Eviota species display consistent associations with a single type of microhabitat on the reef. The strength of this association varies between species. For example, E. winterbottomi was found only on hard coral, whereas E. guttata displayed a strong association with hard coral but was also found on rubble in one instance. Past studies report the latter species from coral rock, which we did not sample due to the scarcity of this habitat at most of our sites (Herler and

Hilgers, 2005; Herler, 2007). Other species displayed strong associations for rubble, but would

occasionally be found in a hard coral or sand collection (e.g., Eviota Moorea sp. 1, E. lachdeberei, E. queenslandica).

Ancestral state reconstruction of microhabitat associations revealed that the common ancestor of all Eviota was likely a hard coral associate. This is consistent with the fact that the sister groups to Eviota (Gobiodon, Pleurosicya, Bryaninops) are all obligate coral dwellers.

Interestingly, Eviota species have invaded rubble habitats several times throughout their evolutionary history. As many as three independent habitat switches from coral to rubble occurred in the branched clade, the most recent of which occurred between E. winterbottomi and

E. queenslandica. Within the unbranched clade both sand/rubble species likely descended from a sand/rubble common ancestor early in the history of the clade. Similarly, the rubble-associated species of unbranched clade, E. spilota, E. cf. spilota and E. lachdeberei, all likely descended from a rubble-associated ancestor. Eviota infulata and Eviota Red Sea sp. 3 are both represented by few individuals in our samplings, but our collections indicate that both occur on hard coral and likely descend from a coral-associated ancestor.

4.2. Morphological character evolution

The two major clades of Eviota recovered in our analysis correspond with the division of species with branched versus unbranched pectoral fin-rays (Fig. 9). The maximum likelihood reconstruction of ancestral states could not unambiguously resolve the state of the common ancestor of all Eviota; however, if we consider that most gobies have branched pectoral fin-rays

(including members of our outgroup genera Bryaninops, Gobiodon and other species of the

‘‘coral gobies’’ sensu Thacker and Roje, 2011), the most parsimonious explanation is that the ancestor of Eviota had branched rays. There are no examples of reversals in this character in any of the species in our study, despite multiple switches in habitat association. Sensory pore

patterns, unlike pectoral fin-ray branching, displayed a large amount of homoplasy with each pattern appearing multiple times independently across the phylogeny. This is consistent with observations by Greenfield and Randall (2010), who also questioned the homology of sensory pore patterns, pointing out that E. shimadai and E. sigillata are morphologically similar and may be closely related, yet the two are placed in different ‘‘groups’’ based on the absence of a single pore in E. shimadai. Virtually all internal nodes on the phylogeny had ambiguous sensory pore character states, and there was no obvious relationship between a particular pore pattern and a specific microhabitat type.

Shifts in Eviota microhabitat associations occur independently of changes in morphology.

Reduction of head canals has been observed in several groups of gobies that have invaded novel habitats (Miller, 2004; Kovačić, 2005; Stelbrink and Freyhof, 2006). These reductions may be associated with a release of predation pressures in the novel habitat, as predator detection is a major function of the accessory lateralis system (Stelbring and Freyhof, 2006). The incongruence between pore reduction and shifts in microhabitat association in Eviota may suggest that the selective forces driving sensory pore pattern evolution act independently of the microhabitats in which Eviota species occur. On the other hand, sensory pore loss or fusion in Eviota may simply be a byproduct of overall body size reduction and associated fusion of cranial elements, thus evolving in a manner that appears to be random with respect to selective forces from different microhabitats.

The fact that Eviota species have been able to repeatedly invade novel habitats in the absence of significant changes to pectoral fin-ray morphology indicates that the shape of individual rays is less important in dwarfgobies than in other cryptobenthic fishes. Gobies primarily use their specialized pectoral fins for swimming (Adriaens et al., 1993). A pectoral fin

with highly-branched rays could potentially increase propulsion by increasing the surface area of the fin, however most branching in many Eviota occurs mainly at the distal tips of the median fin rays and the increase in surface area may not be significant enough to gain any appreciable locomotive benefit. Other cryptobenthic fishes, such as some blennies and sculpins, use specialized unbranched pectoral fin-rays to grip or manipulate the substrate and support the weight of the body (Webb, 1989; Brandstätter et al., 1990; Taft, 2011). This may not be the case in Eviota. If the simple, unbranched fin rays in Eviota were adaptations for specific substrates, we would expect the form of rays to vary in species from different microhabitats. Furthermore, the structural adaptations in goby pectoral girdles that allow for strong fin adduction also limit maneuvering and manipulating the fin itself, making it unlikely that unbranched rays of Eviota are adaptations for interacting with the substrate (Adriaens et al., 1993). Species of Eviota also possess highly modified pelvic fins relative to other coral gobies, which also contact the substrate. Among species of Eviota, there are differences in the shape, size, and branching of pelvic fin rays, and it is possible that variation in this character may also affect microhabitat association. However, there is substantial variation within species in these pelvic fin-ray characters, and incorporating and summarizing this within- species variation in a way that facilitates a meaningful comparison across the phylogeny is problematic.

4.3. Diversification in stages

Taylor and Hellberg (2005) suggested that adaptive radiations occurred in stages in the reef gobies Elacatinus and Tigrigobius/Risor. In Atlantic Elacatinus, initial radiations resulted in a clade of sponge-dwelling gobies and a clade of cleaner-gobies (Taylor and Hellberg, 2005). A subsequent stage of speciation within each clade was driven largely by color differences and allopatric speciation. Additionally, segregation in habitat depth may have also played a part in

speciation within sponge-dwelling Elacatinus by facilitating reproductive isolation (Colin, 1975,

2010). We see similar evidence of staged radiations in Eviota. However, unlike Elacatinus,

Eviota species appear to have initially diverged in morphology first, forming distinct clades with branched versus unbranched pectoral fin-rays. Subsequent stages of divergence have been ecologically driven. In the unbranched clade, species are sorted by habitats forming clades of sand/rubble, coral, and rubble species early in the evolutionary history of the clade, whereas shifts from coral to rubble in the branched clade occurred multiple times throughout the history of the clade.

Allopatric speciation cannot be rejected as an explanation for speciation near the tips of the phylogeny. Although some pairs of closely related species within each clade currently have partially overlapping distributions (i.e. E. shimadai and E. sigillata both are reported from

Pohnpei and Indonesia; Greenfield and Randall, 2010), the ages of divergence between many of these species pairs are not recent and we cannot reject the combination of initial reproductive isolation and subsequent secondary contact (via range expansions) as the most parsimonious explanation for these distribution patterns. It is worth noting that the sympatric or parapatric species pairs with similar habitat associations (i.e. E. shimadai and E. sigillata) were not collected together at any site in our study. This may be the result of one species being locally displaced due to interspecific competition, including possible bathymetric partitioning of habitat

– a pattern observed in putative Hawaiian sister species E. susanae and E. rubra (Greenfield and

Randall, 1999) as well as in Gobiodon sp. 1 and G. reticulatus on host-coral Acropora samoensis

(Dirnwöber and Herler, 2007). Alternatively, this could also be the result of a local competitive lottery model scenario (Sale, 1978; Munday, 2004a).

4.4. Dwarfgoby diversity, speciation and extinction

The repeated habitat transitions observed in Eviota may be a major factor contributing to the high species richness of the genus. Interspecific competition among gobies for live coral may be significantly higher than for sand or rubble habitats, as many species of gobies on coral reefs have obligatory habitat associations with specific types of coral. Entire genera may be composed of obligate coral species, including Gobiodon, Pleurosicya, and Bryaninops – all of which are closely related to Eviota (Herler, 2007; Thacker and Roje, 2011). If we parsimoniously assume that coral-association arose in the common ancestor to the aforementioned genera, then the ability to shift from a strict species-specific or morphotype-specific (e.g., branching versus massive versus foliose) coral association to a more broad, generalist coral association could be considered an ecological release. Additional ecological opportunities would exist for species that are free from coral or coral-rock obligations entirely, and can occur in rubble or sand – habitats that are typically inhabited by fewer small cryptobenthic fishes than live coral (Ahmadia et al.,

2012). Eviota species that are found on live coral often occur on many different species of live coral as well as coral rock (Herler, 2007), suggesting that the gobies are attracted to a range of complex three-dimensional structures rather than a single organism. Although rubble (or a mixture of rubble and sand) may not be as structurally complex as coral, Eviota species are small enough to capitalize on smaller scale habitat complexities that characterize rubble-sand habitats and thus may be released from the need to occupy larger three-dimensional structures.

The high number of species of Eviota relative to other coral gobies may be explained in part by their ability to repeatedly colonize habitats that are less preferred by other reef fishes throughout their evolutionary history, thus occupying new niches on the reef and facilitating further speciation. However, the number of extant species in a group is not only a function of

speciation, but is also determined by the relative rate of extinction. Extinction risk in coral reef fishes depends on several factors including their response to changes in habitat quality (Munday,

2004b; Pratchett et al., 2006, 2008; Graham et al., 2011). Species with strict microhabitat requirements or narrow diets are more vulnerable to natural disturbances (Munday, 2004b;

Pratchett et al., 2006). For example, obligate coral dwelling Gobiodon species currently experience both local and global extinction risks due to habitat loss caused primarily by coral bleaching (Munday, 2004b; Bellwood et al., 2012). Entire cryptobenthic fish assemblages can change rapidly as a result of massive coral loss, and these assemblage changes can persist for many years after corals recover (Bellwood et al., 2006, 2012). In such scenarios, Eviota species have actually been shown to increase in abundance (Bellwood et al., 2006). Having the ability to inhabit a variety of different types of coral or coral rock would be advantageous in bleaching events, where some types of corals are impacted more than others. Additionally, for rubble or sand associated species, habitat availability would be unaffected by most threats to coral reefs, and in some scenarios (e.g., cyclones) disturbances may even increase the amount of rubble habitat available for species. Beyond benefitting from their habitat associations, some species of

Eviota have been shown to have very wide dietary preferences, high reproductive output, fast life-cycles, and evidence of hermaphroditism – all of which may contribute to low extinction relative to other closely related coral reef gobies (Depczynski and Bellwood, 2003, 2005, 2006;

Cole, 1990).

5. Conclusions

This study represents the first phylogenetic analysis and most comprehensive ecological survey of the genus Eviota – one of the most diverse, yet poorly studied groups of reef fishes. Our molecular phylogeny serves as a framework to identify potential ecological drivers of this

diversity, and investigate the relationship between microhabitat and morphology. This study has several important findings that increase our understanding of gobiid diversity, habitat use of cryptobenthic reef fish, and the relationships between morphological evolution and microhabitat association in reef fishes:

 Most Eviota species are closely associated with a particular type of habitat or substrate on

reefs, although these preferences are less strict than that of other coral gobies.

 The common ancestor of all Eviota species was likely a coral associated species, and

habitat shifts from coral to rubble, or from coral to rubble/sand occurred many times

independently throughout the evolutionary history of Eviota.

 Shifts in microhabitat association occurred independent of changes in pectoral fin-ray

branching or sensory pore patterns, the latter of which was highly homoplasious in

Eviota.

 The relative flexibility of coral association in Eviota combined with the repeated shifts

into non-coral habitats may serve as an ecological release from the strict obligate coral-

dwelling relationship seen in closely related species, promoting subsequent speciation in

new vacant or underutilized niches. Increased speciation opportunity coupled with

resilience to extinction may explain the exceptionally high species-richness in Eviota.

Acknowledgments

We thank Jocelyn Curtis-Quick, Dan Lazell, Abi Powell, Iwan, Pippa Mansell, Laura Sheard,

Conservation Society of Pohnpei, and Brian Lynch and students from the College of Micronesia for assistance in the field. Mike Cavazos, Tim Harlow, Andrew Layman, and Elizabeth Hinkle assisted with lab work. Ryan Chabarria and Sharon Furiness assisted with lab work and contributed helpful discussion. We thank David Greenfield for assisting with identifications of

some species and he and Rick Winterbottom for providing preliminary dichotomous keys for

Eviota species. We gratefully acknowledge the support of the staff at the Hoga Marine Research

Center, Universitas Hasanuddin (UNHAS), the Wakatobi Government, the Tamana National

Wakatobi, the State Ministry of Research and Technology (RISTEK), the KAUST Coastal and

Marine Resources Core Lab, and the staff of the Berkley Gump Station in Moorea. Barbara

Brown at AMNH, Dave Catania at CAS, and Renny Kurnia Hadiaty at MZB provided assistance with depositing voucher specimens. The first and second authors are indebted to E.V. Ohta and her family members for their selfless contributions that made this project possible. Funding for field work was provided by Operation Wallacea, and by NSF OISE-0553910 to FP.

References

Adriaens, D., Decleyre, D., Verraes, W., 1993. Morphology of the pectoral girdle in

Pomatoschistus lozanoi de Buen, 1923 (Gobiidae), in relation to pectoral fin adduction.

Bel. J. Zool. 123, 135–157.

Ahmadia, G.N., Pezold, F.L., Smith, D.J., 2012. Cryptobenthic fish biodiversity and microhabitat

use in healthy and degraded coral reefs in SE Sulawesi, Indonesia. Mar. Biodiv. 42, 433-

442.

Ahnelt, H., Bohacek, V., 2004. The lateral line system of two sympatric eastern Pacific gobiid

fishes of the genus Lythrypnus (Teleostei: Gobiidae). Bull. Mar. Sci. 74, 31–51.

Ahnelt, H., Goschl, J., Dawson, M.N., Jacobs, D.K., 2004. Geographical variation in the cephalic

lateral line canals of Eucyclogobius newberryi (Teleostei, Gobiidae) and its comparison

with molecular phylogeography. Folia-Zool. 53, 385–398.

Ahnelt, H., 1995. Two new species of Knipowitschia from Western Anatolia (Turkey). Mitt.

Hamb. Zool. Mus. Inst. 92, 155–167.

Ahnelt, H., Bianco, P., Schwammer, H., 1995. Systematics and zoogeography of Knipowitschia

caucasica (Teleostei: Gobiidae) based on new records from the Aegean Anatolian area.

Ichthyol. Explor. Freshw. 6, 49–60.

Ahnelt, H., Scattolin, G., 2003. The lateral line system of a blind goby, Typhlogobius

calidorniensis Steindachner, 1879 (Teleostei:Gobiidae). Annalen des Naturhistorisches

Museums in Wien Serie B Botanik und Zoologie, 11–25.

Asaoka, R., Nakae, M., Sasaki, K., 2011. Description and innervation of the lateral line system in

two gobioids, Odontobutis obscura and Pterogobius elapoides (Teleostei: Perciformes).

Ichthyol. Res. 58, 51–61.

Bellwood, D.R., Hoey, A.S., Ackerman, J.L., Depczynski, M., 2006. Coral bleaching, reef fish

community phase shifts and the resilience of coral reefs. Global Change Biol. 12, 1587–

1594.

Bellwood, D.R., Baird, A.H., Depczynski, M., Gonzalez-Cabello, A., Hoey, A.S., Lefevre, C.D.,

Tanner, J.K., 2012. Coral recovery may not herald the return of fishes on damaged coral

reefs. Oecologia 170, 567-573.

Brandstätter, R., Misof, B., Pazmandi, C., Wagner, G., 1990. Microanatomy of the pectoral fin in

blennies (Blenniini, Blennioidea, Teleostei). J. Fish Biol. 37, 729– 743.

Brown, J.M., Hedtke, S.M., Lemmon, A.R., Lemmon, E.M., 2010. When trees grow too long:

investigating the causes of highly inaccurate Bayesian branch-length estimates. Syst.

Biol. 59, 145–161.

Cole, K.S., 1990. Patterns of gonad structure in hermaphroditic gobies (Teleostei, Gobiidae).

Environ. Biol. Fishes 28, 125–142.

Colin, P.L., 1975. Neon Gobies. T.F.H. Publications, Neptune City, NJ.

Colin, P.L., 2010. Fishes as living tracers of connectivity in the tropical western North Atlantic:

I. Distribution of the neon gobies, genus Elacatinus (Pisces: Gobiidae). Zootaxa, 36–52.

Coombs, S., Janseen, J., Webb, J.F., 1988. Diversity of lateral line systems: evolutionary and

functional considerations. In: Atema, J., Fay, R., Popper, A., Tovolga, W. (Eds.), Sensory

Biology of Aquatic Animals. Springer, New York, pp. 553–593.

Depczynski, M., Bellwood, D.R., 2003. The role of cryptobenthic reef fishes in coral reef

trophodynamics. Mar. Ecol.-Prog. Ser. 256, 183–191.

Depczynski, M., Bellwood, D.R., 2005. Shortest recorded vertebrate lifespan found in a coral

reef fish. Curr. Biol. 15, R288-R289.

Depczynski, M., Bellwood, D.R., 2006. Extremes, plasticity, and invariance in vertebrate life

history traits: Insights from coral reef fishes. Ecology 87, 3119– 3127.

Dirnwöber, M., Herler, J., 2007. Microhabitat specialization and ecological consequences for

coral gobies of the genus Gobiodon in the Gulf of Aqaba, northern Red Sea. Mar. Ecol.-

Prog. Ser. 342, 265–275.

Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with

BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969-1973.

Froese, R., Pauly, D., 2012. FishBase: World Wide Web Electronic Publication.

(accessed 04.04.12).

Gill, A., Jewett, S., 2004. Eviota hoesei and E. readerae, new species of fish from the southwest

pacific, with comments on the identity of E. corneliae Fricke (Perciformes: Gobiidae).

Rec. Aust. Mus. 56, 235–240.

Gill, A., Mooi, R., 2012. Thalasseleotrididae, new family of marine gobioid fishes from New

Zealand and temperate Australia, with a revised definition of its sister taxon, the

Gobiidae (Teleostei: Acanthomorpha). Zootaxa 3266, 41–52.

Graham, N.A.J., Chabanet, P., Evans, R.D., Jennings, S., Letourneur, Y., MacNeil, M.A.,

McClanahan, T.R., Ohman, M.C., Polunin, N.V.C., Wilson, S.K., 2011. Extinction

vulnerability of coral reef fishes. Ecol. Lett. 14, 341–348.

Greenfield, D., Randall, J., 1999. Two new Eviota species from the Hawaiian Islands (Teleostei:

Gobiidae). Copeia 1999, 439–446.

Greenfield, D., Randall, J., 2010. Four new gobiid fishes of the genus Eviota from the Western

Pacific, with clarification of Eviota guttata and Eviota albolineata (Teleostei: Gobiidae).

Proc. Calif. Acad. Sci. 61, 269–289.

Herler, J., 2007. Microhabitats and ecomorphology of coral- and coral rock associated gobiid

fish (Teleostei: Gobiidae) in the northern Red Sea. Mar. Ecol.-Evol. Persp. 28, 82–94.

Herler, J., Hilgers, H., 2005. A synopsis of coral and coral-rock associated gobies (Pisces:

Gobiidae) from the Gulf of Aqaba, northern Red Sea. Aqua Int. J. Ichthyol. 10, 103–132.

Herler, J., Koblmuller, S., Sturmbauer, C., 2009. Phylogenetic relationships of coralassociated

gobies (Teleostei, Gobiidae) from the Red Sea based on mitochondrial DNA data. Mar.

Biol. 156, 725–739.

Kovačić, M., 2005. Morphological variability of Pomatoschistus canestrinii (Gobiidae), with the

reduction of squamation and head canals. Cybium 29, 373–379.

Lachner, E., Karnella, S., 1980. Fishes of the Indo-Pacific genus Eviota with descriptions of

eight new species (Teleostei: Gobiidae). Smithsonian Contrib. Zool. 315, 1–127.

Marshall, D.C., 2010. Cryptic failure of partitioned Bayesian phylogenetic analyses: lost in the

land of long trees. Syst. Biol. 59, 108–117.

Miller, P., 2004. The Freshwater Fishes of Europe. vol. 8/II. Gobiidae 2. Aula-Verlag,

Wiebelsheim.

Munday, P.L., 2004a. Competitive coexistence of coral-dwelling fishes; the lottery hypothesis

revisited. Ecology 85, 623–628.

Munday, P.L., 2004b. Habitat loss, resource specialization, and extinction on coral reefs. Global

Change Biol. 10, 1642–1647.

Myers, R.F., 1999. Micronesian Reef Fishes, 3rd ed. Coral Graphics, Barrigada.

Neilson, M.E., Stepien, C.A., 2009. Escape from the Ponto-Caspian: evolution and biogeography

of an endemic goby species flock (Benthophilinae: Gobiidae: Teleostei). Mol.

Phylogenet. Evol. 52, 84–102.

Paradis, E., Claude, J., Strimmer, K., 2004. APE: Analyses of Phylogenetics and Evolution in R

language. Bioinformatics 20, 289–290.

Pezold, F., 1993. Evidence for a monophyletic Gobiinae. Copeia 3, 634–643.

Polgar, G., Sacchetti, A., Galli, P., 2010. Differentiation and adaptive radiation of amphibious

gobies (Gobiidae: Oxudercinae) in semi-terrestrial habitats. J. Fish Biol. 77, 1645–1664.

Posada, D., 2008. JModelTest: phylogenetic model averaging. Mol. Biol. Evol. 25, 1253–1256.

Pratchett, M., Baird, A.H., McCowan, D., Coker, D., Cole, A., Wilson, S., 2008. Protracted

declines in coral cover and fish abundance following climate induced coral bleaching on

the Great Barrier Reef. In: Proceedings of the 11th International Coral Reef, Symposium,

pp. 1309–1313.

Pratchett, M., Wilson, S., Baird, A., 2006. Declines in the abundance of Chaetodon

butterflyfishes (Chaetodontidae) following extensive coral depletion. J. Fish Biol. 69,

1269–1280.

R Development Core Team, 2012. R: A Language and Environment for Statistical Computing. R

Foundation for Statistical Computing, Vienna, Austria. .

Rambaut, A., Drummond, A.J., 2007. Tracer v1.4. .

Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., Darling, A., Hohna, S., Larget, B.,

Liu, L., Suchard, M.A., Huelsenbeck, J.P., 2012. MrBayes 3.2: efficient Bayesian

phylogenetic inference and model choice across a large model space. Syst. Biol. 61, 539–

542.

Rüber, L., Van Tassell, J.L., Zardoya, R., 2003. Rapid speciation and ecological divergence in

the American seven spined gobies (Gobiidae: Gobiosomatini) inferred from a molecular

phylogeny. Evolution 57, 1584–1598.

Sale, P., 1978. Coexistence of coral reef fishes—a lottery for living space. Environ. Biol. Fishes

3, 85–102.

Shibukawa, K., Suzuki, T., 2005. Two new species of the gobiid fish genus Eviota (Teleostei,

Perciformes, Gobioidei) from the Western Pacific. Bull. Natl. Sci. Mus. Ser. A 31, 65–76.

Stelbrink, B., Freyhof, J., 2006. Reduction of scales and head canals in Pomatoschistus

canestrinii (Ninni, 1883) (Teleostei, Gobiidae). Verhandlungen der Gesellschaft für

Ichthyologie 5, 71–77.

Streelman, J.T., Danley, P.D., 2003. The stages of vertebrate evolutionary radiation. Trends

Ecol. Evol. 18, 126–131.

Taft, N., 2011. Functional implications of variation in pectoral fin ray morphology between

fishes with different patterns of pectoral fin use. J. Morphol. 272, 1142–1152.

Taylor, M., Hellberg, M., 2005. Marine radiations at small geographic scales: speciation in

Neotropical reef gobies (Elacatinus). Evolution 59, 374–385.

Thacker, C.E., 2003. Molecular phylogeny of the gobioid fishes (Teleostei: Perciformes:

Gobioidei). Mol. Phylogenet. Evol. 26, 354–368.

Thacker, C.E., 2009. Phylogeny of Gobioidei and placement within Acanthomorpha, with a new

classification and investigation of diversification and character evolution. Copeia 2009,

93–104.

Thacker, C.E., Roje, D.M., 2011. Phylogeny of Gobiidae and identification of gobiid lineages.

Syst. Biodivers. 9, 329–347.

Webb, P., 1989. Station-holding by three species of benthic fishes. J. Exp. Biol. 145, 303–320.

Yamada, T., Sugiyama, T., Tamaki, N., Kawakita, A., Kato, M., 2009. Adaptive radiation of

gobies in the interstitial habitats of gravel beaches accompanied by body elongation and

excessive vertebral segmentation. BMC Evol. Biol. 9.

CHAPTER III – Support for a ‘Center of Origin’ in the Coral Triangle: cryptic diversity, recent speciation, and local endemism in a diverse lineage of reef fishes (Gobiidae: Eviota).

Abstract

The Coral Triangle is widely regarded as the richest marine biodiversity hot-spot in the world. One factor that has been proposed to explain elevated species-richness within the Coral

Triangle is a high rate of in situ speciation within the region itself. Dwarfgobies (Gobiidae:

Eviota) are a diverse genus of diminutive cryptobenthic reef fishes with limited dispersal ability, and life histories and ecologies that increase potential for speciation. We use molecular phylogenetic and biogeographic data from two clades of Eviota species to examine patterns, processes and timing associated with species origination within the Coral Triangle. Sequence data from mitochondrial and nuclear DNA were used to generate molecular phylogenies and median-joining haplotype networks for the genus Eviota, with emphasis on the E. nigriventris and E. bifasciata complexes – two species groups with distributions centered in the Coral

Triangle. The E. nigriventris and E. bifasciata complexes both contain multiple genetically distinct, geographically restricted color morphs indicative of recently-diverged species originating within the Coral Triangle. Relaxed molecular-clock dating estimates indicate that most speciation events occurred within the Pleistocene, and the geographic pattern of genetic breaks between species corresponds well with similar breaks in other marine fishes and sessile invertebrates. Regional isolation due to sea-level fluctuations may explain some speciation events in these species groups, yet other species formed with no evidence of physical isolation.

The timing of diversification events and present day distributions of Eviota species within the

Coral Triangle suggest that both allopatric speciation (driven by ephemeral and/or ‘soft’ physical barriers to gene flow) and sympatric speciation (driven by niche partitioning and assortative mating) may be driving diversification at local scales within the Coral Triangle. The presence of multiple young, highly-endemic cryptic species of Eviota within the Coral Triangle suggests that

(i) the Coral Triangle is indeed a “cradle” of reef fish biodiversity and that (ii) our current approximations of reef fish diversity in the region may be significantly underestimated.

1. Introduction

1.1 Biodiversity within the Coral Triangle: origin, accumulation or overlap?

The world’s richest marine biodiversity hot-spot is located in a region known as the Coral

Triangle (Allen and Werner, 2002; Veron et al., 2009; Carpenter et al., 2011). The Coral

Triangle (CT), also known as the Indo-Malay Archipelago or the Indo-Australian Archipelago, is a region of islands in the Western Pacific that includes the Philippines, Indonesia, Malaysian

Borneo, Timor-Leste, Papua New Guinea and the Solomon Islands (Veron et al., 2009). The area is home to over 600 species of corals and over 2600 species of reef fishes (Veron et al.,

2009; Allen and Erdmann, 2012). Species richness across a wide array of marine organisms declines with distance (both in latitude and longitude) from the CT (Stehli and Wells, 1971;

Veron, 1995; Briggs, 1999a; Mora et al., 2003). Many hypotheses have been proposed to explain the exceptional species-richness within this region, including: (1) the center of origin hypothesis (Ekman, 1953; Briggs, 1999a; Briggs, 1999b), which proposes that most species originate within the CT, likely due to elevated rates of speciation within the region, and that the abundance of new species originating within the CT provides a source of biodiversity for surrounding areas; (2) the center of accumulation hypothesis, which states that speciation occurs

primarily in multiple peripheral regions, and fauna accumulate in the CT via dispersal of individual species (Ladd, 1960) or of entire assemblages following tectonic events (Remington,

1968; McKenna, 1973; Santini and Winterbottom, 2002); and (3) the center of overlap hypothesis (Woodland, 1983), which proposes that high biodiversity in the CT is due to its position straddling the boundary between the Indian Ocean and the Pacific Ocean, with each ocean having its own distinct fauna forming via vicariance as the two basins were separated, and fauna subsequently overlap in the CT as species expanded their ranges.

Empirical evidence in support of each of these hypotheses has begun to emerge over the last two decades, most recently in the form of molecular phylogeography studies (see Carpenter et al., 2011; Briggs and Bowen, 2013; Gaither and Rocha, 2013 for reviews). However, no single hypothesis has stood out as a front-runner in explaining most of the biodiversity in the CT.

Instead, the processes proposed above are not mutually exclusive and may act in concert to produce exceptional biodiversity (Mironov, 2006; Halas and Winterbottom, 2009; Barber, 2009;

Bowen et al., 2013). Species richness in the CT can also be explained in part by the availability of stable coral reef habitat in this region, especially during periods of quaternary glacial cycles, thus serving as an evolutionary refuge for species (Bellwood and Hughes, 2001; Pellissier et al.,

2014). Pellisier et al. (2014) demonstrated a strong correlation between present day reef fish diversity and historical proximity to coral reef refugia during glacial cycles. Thus, in periods of dramatic climate change and habitat loss, the CT may still function as both a cradle of speciation as well as a repository for peripheral species.

The thousands of species of fishes that contribute to the diversity within the CT exhibit a variety of mating systems, locomotive behaviors, larval morphologies and dispersal capabilities, habitat requirements, life spans, reproductive output, and parental care regimes (Sale, 1993,

2002; Deloach, 1999; Leis and Carson-Ewart, 2000; Hixon, 2011). Each of these factors directly effects the potential of a lineage to diversify in a given place and time. Thus, one should not reasonably expect a single model to fit well when comparing a broad range of reef fish families and genera that differ dramatically in their ability to respond to a particular driver of speciation.

On the other hand, one may expect that species that share similar ecological and life history characteristics would show similar patterns of evolution and biogeography, even if they belong to vastly different phylogenetic lineages. Thus, rather than searching for a single hypothesis explaining diversity across all organisms in the CT, it may be more informative to ask whether a given hypothesis is more likely to explain biogeographic patterns for a particular suite of organisms that possess certain ecological and life history characteristics. Conversely, by studying organisms exhibiting life history traits conducive to a particular mode or pattern of speciation we can examine evolutionary dynamics associated with the speciation of clades and ultimately the factors contributing to species richness within a region. Here we offer a model group of coral reef fishes for study of the processes and events associated with the origination of species within the CT.

1.2 Gobies and speciation within a center of origin

Empirical evidence suggests that sympatric, parapatric, and allopatric speciation are all responsible for the diversity within the CT (Briggs, 1999c; Briggs, 2005; Bowen et al., 2013). If the CT indeed produces species at elevated rates relative to peripheral regions, these processes must be operating within very small geographic scales and within rapid timeframes. Thus, evidence for the center of origin hypothesis is likely to come from species with characteristics that facilitate allopatric speciation even within a localized region where barriers to gene flow are ephemeral or ‘soft’ (e.g. periodic sea-level fluctuations due to glaciations; major oceanographic

surface currents). Characteristics of such species would include extremely limited adult mobility, limited larval dispersal ability, fragmented populations and short life cycles with high turnover, enabling them to form distinct species after only short periods of physical reproductive isolation.

On the other hand, speciation in sympatry is thought to occur very rapidly, perhaps even faster than what is required for allopatric populations (Landry et al., 2003; Coyne and Orr, 2004).

Species that have increased potential for rapid, ecologically-driven sympatric speciation within the CT would have characteristics that facilitate the co-existence of closely related species via the exploitation of underutilized niches in a complex, heterogeneous reef environment, often in the face of intense competition from other distantly related reef-associated species. These characteristics facilitating both allopatric and/or sympatric speciation are present in the family

Gobiidae (gobies), making them an excellent model group for elucidating processes contributing to speciation within the CT.

Gobies are one of the most diverse families of vertebrates in the world. With more than

1700 species spread across more than 200 genera, gobies make up a dominant fraction of the total fish diversity in a broad range of ecosystems including coral reefs (Murdy, 2011a; Murdy,

2011b; Keith and Lord, 2011; Patzner et al., 2011; Pezold, 2011; Thacker, 2011; Eschmeyer,

2014; Thacker and Roje, 2011; Herler et al., 2011; Ahmadia et al., 2012).

One of the most diverse clades of coral-associated gobies is the genus Eviota. Commonly referred to as dwarfgobies, species of Eviota are miniscule in size (frequently less than 20 mm total length) and occur on a broad range of microhabitats including a variety of hard corals, coral rock, rubble, and sand (Herler and Hilgers, 2005; Herler et al., 2007; Herler et al., 2011;

Ahmadia et al., 2012; Tornabene et al., 2013a). With over 90 valid species of Eviota described to date, dwarfgobies represent one of the most evolutionarily successful lineages of coral reef

fishes. Their remarkable diversity may be explained by repeated habitat shifts from a strict obligate coral-dwelling lifestyle into sand or rubble habitats, promoting subsequent speciation in novel and underutilized niches (Tornabene et al., 2013a).

Evolutionary time scales may also be accelerated in dwarfgobies relative to other reef fish. The maximum life span of Eviota sigillata is 59 days and is the shortest known life cycle of any vertebrate (Depczynski and Bellwood, 2005, 2006). Short generation times (47-74 days) and high mortality result in rapid turnover of populations (Depczynski and Bellwood, 2005, 2006).

Reproduction in dwarfgobies also likely plays a role in their evolutionary potential. Mating in

Eviota occurs in pairs and is characterized by a series of well-choreographed, species-specific spawning interactions prior to deposition and fertilization of eggs (Sunobe, 1998; Sunobe and

Nakazono, 1999). Females often chose males with stronger secondary sexual traits (e.g. dorsal fin length), and males spend significant effort guarding nests after fertilization (Sunobe and

Nakazono, 1999; Sekiya and Karino, 2004; Karino and Arai, 2006). As a result, home ranges for mating pairs are very small, and thus their small adult size, localized spawning, and short pelagic larval duration of 24-26 days produce populations that may be genetically and demographically isolated. The combination of isolation and high turnover may increase the rates at which dwarfgoby populations respond to natural selection and are subjected to genetic drift.

1.3 Tips of the Eviota tree: the E. nigriventris and E. bifasciata species complexes

The Eviota nigriventris species complex is an ideal group for studying recent speciation within the CT. For 78 years E. nigriventris was considered to be a single species until recent studies indicated the presence of a complex containing at least three species with distinct coloration (Greenfield and Randall, 2011; Greenfield and Tornabene, 2014). Similarly, E.

bifasciata was considered a single species until studies suggested the presence of a species complex that includes at least two other taxa from the CT (Allen, 2001; Allen et al., 2013). The distributions of both of these species complex are centered in the CT and do not extend considerably into the Indian Ocean or Central Pacific Ocean. Additional un-named color morphs in both complexes exist in the CT and peripheral regions. Therefore, these species complexes represent two independent instances where recent speciation events may have occurred within the CT, thus making them excellent groups to examine the factors generating biodiversity within the region. Furthermore, a comprehensive genetic survey of dwarfgobies in this region may reveal additional cryptic species in other clades of Eviota, ultimately resulting in a better understanding of the breadth of the diversity in one of the world’s most diverse lineages of marine fishes.

The present study uses a molecular phylogenetic approach to study diversification at the tips of the dwarfgoby phylogeny. Here we present a time-calibrated molecular phylogeny for 133 specimens of Eviota from within the Coral Triangle, as well as external localities in Fiji,

Pohnpei, Moorea, the Red Sea, Saipan, and Seychelles. By focusing our taxon sampling efforts on the E. nigriventris and E. bifasciata species complexes, we test several hypotheses regarding speciation in Eviota and diversification in the CT, including: (i) regional color morphs within the

E. nigriventris and E. bifasciata complexes reflect genetic differences associated with distinct species; (ii) the timing of speciation events and the present day biogeography of the E. nigriventris and E. bifasciata complexes within the CT are consistent with the center of origin hypothesis; and (iii) contemporary species boundaries in Eviota correspond to well-documented genetic breaks found in other species within the CT.

Figure 10 Localities of samples used in this study. The range of the E. nigriventris complex is outlined in a solid line, and the range of the E. bifasciata species complex is shaded. The Coral

Triangle, as delineated by Veron et al. (2009), is outlined by a dotted line.

2. Methods

We collected specimens of Eviota from throughout the Indo-Pacific region and additional samples were utilized from museum collections (Fig. 10; Supplementary Material Table 5). Most specimens were photographed underwater or immediately following collection to help confirm identification and record color patterns. Fish were collected with hand-nets and a solution of clove oil anesthetic or ichthyocide. Specimens were stored whole in 95% ethanol prior to DNA extraction with a Qiagen ®DNAeasy Blood and Tissue Kit (Qiagen, Valencia, California).

Mitochondrial cytochrome c oxidase I (COI), and the nuclear gene protease III (Ptr) were amplified via PCR using primers from Thacker (2003; GOBYL6468, GOBYH7696), Ward et al.

(2005; FishF1-5, FishR1-5), and Yamada et al. (2009; PtrF2, PtrR2), with GoTaq® Hotstart

Master Mix (Promega, Madison, Wisconsin). The thermal profile consisted of 2 min at 95 °C, followed by 35 cycles of 40s at 95 °C, 40s at 52-55 °C, and 90s at 72 °C. Purification of PCR products and sequencing was performed by Beckman Coulter Genomics (Danvers,

Massachusetts). The nuclear gene Ptr was included to help resolve relationships towards the root

of the phylogeny, and very few phylogenetically informative sites were found between closely- related species at this locus. Sequences were assembled and aligned in Geneious R6 (Biomatters

Ltd., available at http://www.geneious.com), and alignments were double checked by eye.

Several specimens used in this study were sequenced in previous studies using similar protocols

(Tornabene et al., 2013a; Tornabene et al., 2013b; Greenfield and Tornabene, 2014). Nexus files for sequence alignments are provided as Supplementary Material.

Phylogeny of Eviota was inferred using Bayesian methods on the concatenated dataset partitioned by gene. Prior to concatenation, COI and Ptr alignments were analyzed in jModeltest ver.0.1.1 (Posada, 2008) to determine the best fitting substitution models based on Akaike

Information Criterion (AIC) scores. The COI and Ptr alignments were then concatenated and phylogeny was then inferred using MrBayes ver.3.2, with each gene receiving its own partition and substitution model. For each MrBayes analysis, two parallel Metropolis-coupled Markov

Chain Monte Carlo runs were generated for 10,000,000 iterations with a sampling frequency of

1000 iterations. Stationarity and adequate mixing of each MCMC run was determined using the program Tracer ver.1.5 (Rambaut and Drummond, 2007). Branch length priors for MrBayes runs were set to an unconstrained-exponential prior with a mean of 0.01 (default mean in

MrBayes is 0.1) to remedy the common problem of MCMC chains converging on local optima with unrealistically long branch lengths (Brown et al., 2010; Marshall, 2010). Outgroups included the two Coral Goby genera Gobiodon and Bryaninops, and the more distantly related

Asterropteryx. Within the E. nigriventris and E. bifasciata complexes, Network ver. 4.6.1.2

(available at http://www.fluxus-engineering.com/sharenet.htm) was used to construct median- joining haplotype networks from the COI datasets (Bandelt et al., 1999). Alignments for median-joining analysis were truncated both in sequence length and in numbers of samples

(individuals with large amounts of missing data were excluded) resulting in 1094 bp for the E. nigriventris group and 577 bp for the E. bifasciata group. Genetic distance matrices for these datasets are listed in Supplementary Material Tables 6 and 7.

Divergence times were estimated from the COI dataset using Bayesian methods in the program BEAST ver.1.7 (Drummond et al., 2012). The fossil record for gobiids is depauperate and accurately identified fossils that would be appropriate for dating our tree do not exist.

Therefore, our phylogeny was calibrated using COI mutation rate estimates calculated from the divergence between Evorthodus lyricus and Evorthodus minutus (approximately 0.048 mutations/site per 106 years), a geminate-pair of gobies separated by the Isthmus of Panama approximately 2.8 ma. Evorthodus sequences used for estimating mutation rate were public data at the Barcode of Life Data System (Ratnasingham and Hebert, 2013; http://www.boldsystems.org/, accessed 10 April 2014). An uncorrelated lognormal relaxed molecular-clock prior was used for the analysis, as our data deviated significantly from a strict global molecular clock model (p < 0.05; likelihood ratio test). Mutation rates were drawn from a prior with a normal distribution centered on a mean of 0.05 with a standard deviation of 0.01.

The topology of the MrBayes multi-locus tree was used as starting tree for the BEAST analysis.

3. Results

The monophyly of the Eviota nigriventris complex and the E. bifasciata complex were both well-supported, with the former being sister to the E. lachdeberei complex, and the latter being sister to a clade containing an undescribed species from Saipan and Pohnpei (Figs. 2, 3). The precision of age estimations decreased towards the root of the phylogeny, where the median age estimate of the genus Eviota was 10.63 Ma with a wide 95% credible interval of 6.5-16.8 Ma.

Ranges for age estimates were considerably narrower towards the tips of the phylogeny. The E. nigriventris complex and E. bifasciata complex are both less than 2.5 million years old, and most species on our tree have origins from the late Miocene through the late Pliocene (Fig. 12).

However, in groups where taxon sampling was more robust, divergence times between recently- evolved lineages are of Pleistocene origin (< 0.5 Ma in the E. nigriventris complex, <1.5 Ma in the E. bifasciata complex).

Figure 11 Bayesian phylogeny of Eviota based on concatenated COI and Ptr datasets. Node values are Bayesian posterior probabilities. Nodes with values less than 0.50 are collapsed into polytomies. Eviota nigriventris and E. bifasciata complexes are collapsed as black triangles.

Figure 12 Time-calibrated phylogeny of Eviota based on COI dataset. Units for the x-axis are

106 years. Node values represent the minimum and maximum values of the 95% highest- posterior-density intervals (credible intervals) of age estimates. Tips of the E. bifasciata and E. nigriventris complexes are collapsed (black triangles). Asterisks indicate relationships that are not supported in concatenated dataset.

The Eviota nigriventris complex includes six reciprocally monophyletic lineages (Fig.

13); E. nigriventris, from the Banda Sea, Sulawesi, Halmahera, Cendrawasih Bay, and Raja

Ampat; E. dorsogilva, from the type locality in Fiji; E. brahmi, from the type locality in New

Hanover, Papua New Guinea; E. dorsopurpurea, from the type locality in Milne Bay, Papua

New Guinea; and two lineages that are similar to E. dorsogilva in morphology but are genetically distinct and have unique coloration, from Ferguson Island and New Hanover Island, respectively.

The first split within this complex occurs around 1.25 Ma (0.71-2.08 Ma 95% HPD interval) with the divergence of a clade that includes E. dorsopurpurea and E. brahmi, the two of which then separate within 60,000 to 230,000 years ago. Subsequent splits within the E. nigriventris group occurred around 830 Ka (460 Ka-1.37 Ma), 330 Ka (170-560 Ka) and 210 Ka (90-380

Ka).

The Eviota bifasciata complex also contains six lineages (Fig. 14); E. pamae, from the type locality in the Kei Islands; E. raja, from the type locality in Raja Ampat; three lineages that are identified as E. bifasciata, with one from Cendrawasih Bay, a second from Bali and

Anambas, and a third from Milne Bay, Papua New Guinea; and a unique eastern Indonesian color morph from the Banda Sea, Sulawesi, Halmahera and Raja Ampat. The five major speciation events within this complex occurred in rapid succession within a very narrow time frame (approximately 1 million years). The rooted phylogeny depicts the eastern Indonesia/Palau clade as the first lineage to diverge from the common ancestor. However, the median-joining haplotype network depicts six missing haplotypes at the center of the network that differ by less than 10 mutations. It is from these central haplotypes that each of the six major lineages diverge dramatically (12-37 mutations).

Figure 13 Time-calibrated phylogeny and median-joining haplotype network of the Eviota nigriventris complex based on COI dataset. Blue bars at nodes are 95% highest-posterior- density intervals (credible intervals) for age estimates. Closed circles on the map are localities where samples from each lineage were collected. Open circles represent localities where a particular lineage/color morph was observed but not collected. Size of circles on haplotype network represents the number of individuals with that haplotype. Black circles represent missing haplotypes. Lines connecting haplotypes are one mutational step unless otherwise denoted by tick marks or numbers.

Figure 14 Time-calibrated phylogeny and median-joining haplotype network of the Eviota bifasciata complex based on COI dataset. Blue bars at nodes are 95% highest-posterior-density intervals (credible intervals) for age estimates. Closed circles on the map are localities where samples from each lineage were collected. Open circles represent localities where a particular lineage/color morph was observed but not collected. Size of circles on haplotype network represents the number of individuals with that haplotype. Black circles represent missing haplotypes. Lines connecting haplotypes are one mutational step unless otherwise denoted by tick marks or numbers.

4. Discussion

Renema et al. (2008) demonstrated the CT as a marine hotspot formed approximately 23 Ma with the collision of Australia with the Pacific arc and Southeast Asian margin. They summarize data from the fossil record and from dated molecular phylogenies to suggest that many groups that exist in the CT today have origins that predate the origin of the current CT hotspot, and that most of the recent diversification within the region pre-dates Pleistocene sea-level fluctuations

(occurring mainly in the Pliocene or earlier). Our analysis of Eviota reveals that the genus as a whole is young, forming in the mid- to late Miocene after the CT hotspot was established. Most of the branching events on the phylogeny are in the late Miocene to early Pliocene, which is in agreement with the synopsis of Renema et al. (2008). However, species groups that have been sampled from multiple localities (e.g. E. nigriventris complex, E. bifasciata complex, E. guttata complex, E. lachdeberei, E. atriventris) reveal multiple isolated genetic lineages (species) corresponding to regional color morphs that diverged more recently in late Pliocene through the

Pleistocene. This pattern suggests that the lineages originating in the Miocene to Pliocene may have stem ages that are old, but likely represent cryptic species complexes comprised of recently-diverged lineages that have yet to be sequenced throughout their range, thus inflating the estimated ages of individual species. This would mirror the pattern found in the gobiid genus

Trimma¸ where nearly all morphologically defined ‘species’, when sequenced from multiple locations, represent complexes of cryptic lineages (Winterbottom et al., 2014). If this pattern of widespread cryptic diversity holds true in Eviota, the number of species in the genus may be well over 120-130 species, considering there are currently 91 described species (most based on morphology or color alone) and perhaps a dozen others that await description (authors’ unpublished data; David Greenfield, pers. comm.).

Figure 15 Coral Triangle during glacial-maxima with sea-levels 120 m below current levels.

Abbreviations: IT – Indonesian Throughflow; HE – Halmahera Eddy; SEC – South Equatorial

Current; SEC/NGCC – South Equatorial Current/New Guinea Coastal Current. Sea-level map modified from Voris (2000). Currents based on Godfrey (1996) and Schiller et al. (2008).

The present-day distributions of the E. bifasciata and E. nigriventris complexes are centered in the CT (Fig. 10). Both species groups extend slightly beyond the CT in the north to the Ryukyu Islands and Palau, as well as to the east off Australia, Vanuatu, New Caledonia and

Fiji (E. nigriventris complex only). Thus, the ranges of the groups themselves suggest an origin within the CT, as neither group occurs well into the Indian Ocean or the Central Pacific Ocean.

Within each species group, there is strong agreement between genetic divergence, coloration and geography. Each group comprises at least six genetic lineages, with most lineages possessing unique geographic distributions and color morphs. These data indicate the presence of recent speciation occurring both within the CT, and between the CT and bordering localities.

4. 1 Eviota nigriventris complex

The most recent speciation even in the Eviota nigriventris complex is between E. brahmi and E. dorsopurpurea. Eviota brahmi occurs in the Bismarck Archipelago, the Solomon Islands, and

Milne Bay (Papua New Guinea), although tissues were only available from the Bismarck

Archipelago. Eviota brahmi co-occurs with its sister species E. dorsopurpurea in Milne Bay, where the latter species is endemic. The sea-level fluctuations during the Pleistocene are unlikely to have caused the split between this species pair, as the waterways surrounding Papua

New Guinea, the Bismarck Archipelago, and the Solomon Islands remained open during that period. The relatively widespread occurrence of E. brahmi suggests that ocean currents such as the South Equatorial Current/New Guinean Coastal Current (SEC/NGCC), which passes between Milne Bay and the Bismarck Archipelago, may not be a barrier of dispersal between these localities. Thus, the recent divergence of E. dorsopurpurea represents a potential case of sympatric speciation within Milne Bay. In Milne Bay, both species have been observed in close proximity, but in distinct groups segregating between different species of coral (Greenfield and

Randall, 2011). Thus, a combination of microhabitat partitioning and assortative mating between

E. brahmi and E. dorsopurpurea may be responsible for creating and or maintaining species boundaries. Habitat partitioning has also been observed between other closely related sympatric species of Eviota (e.g. E. rubra and E. susanae in Hawaii: Greenfield and Randall, 1999), as well as in sympatric pairs of Caribbean neon gobies that possess identical coloration (Colin 1975), and in sympatric sister-species of coral-dwelling Gobiodon (Munday et al., 2004), with the latter example representing one of the strongest cases for sympatric speciation in the sea. The shallow genetic divergence between the E. brahmi and E. dorsopurpurea is striking given the dramatic color differences between the two, and is even less than the intraspecific variation seen in other

Eviota lineages. Nonetheless, the rooted phylogeny recovers the species as reciprocally monophyletic, and this appears to be a situation where divergence in color pattern evolved faster than mutations could accumulate in the mitochondrial genome – a pattern also observed in

Centropyge (Gaither et al., 2014).

The divergence between Eviota nigriventris and a genetically-distinct undescribed color morph from southeast Papua New Guinea (Eviota cf. dorsogilva, Ferguson Island) corresponds with similar genetic breaks between Papua New Guinea and eastern Indonesia populations of nautilus, giant clams, stomatopods, and anemonefish (Wray, 1995; Benzie and Williams, 1997;

Barber et al., 2002; Barber et al., 2006; Timm et al., 2008; Huelsken et al., 2013). Concordant genetic breaks in this region may be maintained by the eastward reflection of the SEC/NGCC in the region north of the Bird’s Head, West Papua that forms the Halmahera Eddy, an oceanographic feature that may present a barrier for dispersal of pelagic larvae. The last split within this group separates E. dorsogilva (Fiji) from a genetically distinct color morph in New

Hanover, Papua New Guinea. This split has also been observed in five species of reef fish,

Pomacentrus moluccensis, Amphiprion melanopus, Chrysiptera talboti, Cirrhilabrus punctatus and Labroides dimidiatus (Drew et al., 2008). Four of those species also possess distinct multiple color morphs that correspond with a Fiji/Indo-West Pacific genetic break.

Several lineages in the Eviota nigriventris complex have small distributions and appear to be endemic to small regions within the Coral Triangle or in the periphery. The notable exception is Eviota nigriventris, which occurs throughout eastern Indonesia. This species may also be more widespread than what is shown here, as the genetic identity of Eviota nigriventris in the NW portion of its range (Java, Sumatra, Philippines, Ryukyu Islands) is unknown. This species complex fits well with the ‘peripheral budding’ model proposed by Hodge et al. (2012, Fig 4c).

Under this model, a large widespread parent species repeatedly produces a series of ‘buds’, or smaller isolated species via series of peripatric, parapatric or sympatric speciation events. These budded species can be local endemic species within the range of the larger species, or may be located on the periphery of the parent species. The phylogeographic pattern left by this process shows the widespread species as being very young on the phylogeny, despite it being the ancestral ‘parent’ species, as its sister species on the tree will always be the most recently formed

‘bud’. In the case of the E. nigriventris complex, the eastern Indonesian E. nigriventris may have been the parent population repeatedly ‘budding’ species to the east (and perhaps to the NW as well), the first of which led to the E. brahmi/E. dorsopurpurea lineage, the second led to the

E. dorsogilva (Fiji)/E. cf. dorsogilva (New Hanover) lineage, and the last to the E. cf. dorsogilva

(Ferguson Island) lineage.

4.2 Eviota bifasciata complex

The Eviota bifasciata complex contains six reciprocally monophyletic genetic lineages in the

Coral Triangle (Fig. 14). Two of the six lineages in this complex have been described as distinct species; E. pamae from the Kei Islands, and E. raja from Raja Ampat. Another lineage occurs in western Indonesia (Bali, Anambas). This species has also been observed in Malaysia (Sabah), and near the E. bifasciata type locality in the Philippines, thus we tentatively consider this the

‘true’ Eviota bifasciata. However, this lineage possesses similar coloration to two other genetically distinct lineages labeled ‘Eviota cf. bifasciata’, one occurring in Cendrawasih Bay and a second in Milne Bay, the latter represented by a single specimen. The last lineage, also labeled ‘Eviota cf. bifasciata’, occurs in eastern Indonesia (Sulawesi, Halmahera, Banda Sea,

Raja Ampat). This lineage has unique coloration and has also been observed in Palau. The five major speciation events within this complex occurred in rapid succession within a very narrow

time frame (approximately 1 million years). The rooted phylogeny depicts the eastern

Indonesia/Palau clade as the first lineage to diverge from the common ancestor. However, the median-joining haplotype network depicts six missing haplotypes at the center of the network that differ by less than 10 mutations. It is from these central haplotypes that each of the six major lineages diverge dramatically (12-37 mutations).

The biogeography of the Eviota bifasciata complex agrees well with the paleoceanographic history of the region during Pliocene-Pleistocene sea level fluctuations.

During glacial periods beginning in the Pliocene, sea level in the CT dropped as polar ice caps formed. Drops in sea level reached as low as 120 m below current levels in the Pleistocene (Fig.

15; Voris, 2000). A sea level drop of 30-50 m effectively isolates the South China Sea from the

Java Sea to the south via the exposure of the Sunda Shelf, and a drop of 120 m results in near- complete isolation of the South China Sea from the Celebes Sea to the east via the exposure and expansion of the Palawan Islands and Sulu Archipelago. This initial separation likely restricted gene flow and initiated speciation between the western Indonesian clade of E. bifasciata and populations to the east. Sea level rise in subsequent interglacial periods would allow for this species to expand its range south into the Java Sea, and east into the Celebes Sea. Interestingly, the Bali sample of E. bifasciata shows strong genetic divergence from the Anambas population, despite having similar coloration. It is likely that, after an initial southward range expansion, this

Java Sea population was temporarily re-isolated from the South China Sea during subsequent glacial cycles with the re-emersion of the Sunda Shelf. More samples and photographs of the

Java Sea population could help test this hypothesis and clarify the status of the Bali E. bifasciata sample. Present day gene flow between western Indonesian E. bifasciata and species to the east may be impeded by the Indonesian Throughflow, which passes 1-20 million m3 s-1 of water

100

through the Makassar Strait between Borneo and Sulawesi and exits the CT to the south via the

Lombok Strait or to the east around the tip of Timor-Leste (Fig. 15; Godfrey, 1996; Schiller et al., 2008). This apparent genetic break between the South China Sea/western Indonesia and eastern Indonesia has been observed in the scad Decapterus russelli, the damselfish Chrysiptera rex, two species of anemonefish in the genus Amphiprion, and seahorses of the genus

Hippocampus (Perrin and Borsa, 2001; Lourie et al., 2005; Timm et al., 2008; Drew et al., 2010).

A sea level drop of 40-120 m also further constricts the mouth of Cendrawasih Bay to an opening less than 100 km wide via the expansion of Yapen and Miosnum Islands (Fig. 15). This isolation may have been sufficient to restrict gene flow and result in the emergence of a unique

Cendrawasih Bay haplotype. The presence of a distinct haplotype in Cendrawasih Bay is emerging as a repeated pattern in several groups of marine organisms including the sea star

Protoreaster nodosus, anemonefishes, boring giant clams, and the stomatopod Haptosquilla pulchella (Barber et al., 2006; Timm et al., 2008; DeBoer et al., 2008; Crandall et al., 2008).

Allen and Erdmann (2006) described an endemic wrasse from Cendrawasih Bay, moreover noting that other anomalous color morphs of reef fishes occur only within the bay. More recently, Allen and Erdmann (2012) and Greenfield and Erdmann (2014) list a total of 14 coral reef fish species (and mention a number of hard coral and mantis shrimp species) that are considered restricted range endemics of Cendrawasih Bay, including the dwarfgoby E. tetha. It has been hypothesized that the Tosem Block of the South Caroline Arc moved across the opening of the bay in the late Pliocene, resulting in an early period of isolation (Allen and

Erdmann, 2006; 2012; Hill and Hall, 2003). The north coast of Papua receives a constant supply of surface water from eastern Papua New Guinea via the SEC/NGCC; however, flow into

Cendrawasih Bay is minimal as most of the water passes north of the entrance and reflects

101

eastward via the Halmahera Eddy (Schiller et al., 2008). Thus, it is likely that the complex history of tectonic movements, sea level fluctuations, and oceanographic currents

(NGCC/Halmahera Eddy) all contribute to the isolation and observed endemism within

Cendrawasih Bay. The Cendrawasih Bay haplotype is sister to a single specimen from Milne

Bay, Papua New Guinea, differing by 27 mutations (Fig. 14). The Milne Bay population may show the same color pattern as the Cendrawasih Bay clade, however our only observation of

Milne Bay E. bifasciata comes from a single photograph, and additional genetic samples and photos from Papua New Guinea are needed to help determine whether these haplotypes represent multiple species.

Eviota pamae has the most restricted range within the E. bifasciata complex, occurring only in the Kei Islands. The Kei Islands lie on the west side of ‘Lydekker’s Line’, which is recognized as the biogeographic divide between the Indo-Malay region and the Australia-New

Guinea region (Simpson, 1977). No other member of the E. bifasciata complex has been observed in the Kei Islands, however the eastern Indonesian E. cf. bifasciata color morph was observed north of the Kei Islands at Kaimana, West Papua. One possibility is that E. pamae is a relic of a more widespread Sahul Shelf lineage that is now restricted to the Kei Islands after sea level rise, due to the lack of suitable coral habitat along most of the south coast of Papua. This seems unlikely as the species does not occur in the Aru Islands, which are home to reef-fish fauna typically associated with the Australia-New Guinea province (Simpson, 1977; M.V.

Erdmann, pers. obs.). The mechanisms explaining the divergence of E. pamae and subsequent maintenance of genetic isolation remain obscure. Nevertheless, this species represents a clear example of recent speciation occurring at small geographic distances within the CT.

102

The other highly endemic species within the E. bifasciata complex is E. raja, which occurs in the Raja Ampat Islands. This species co-occurs with the eastern Indonesian lineage of

Eviota. cf. bifasciata throughout the Raja Ampat Islands, and both species have been observed cohabitating the same coral head (M.V. Erdmann, pers. obs.). Thus, niche partitioning via habitat preference does not appear to be occurring between these two lineages, and there are no known occurrences of intermediate color morphs that would suggest hybridization occurs in this contact zone. As with E. pamae, the mechanism promoting speciation of the highly endemic E. raja remains elusive, but nonetheless represents an intriguing example of recent speciation and endemism within the CT.

4.3 Conclusions

In summary, concordance between color morphs, geography, and genetics indicate the presence of multiple highly-restricted endemic species these two species complexes from Milne Bay, New

Hanover, Ferguson, Kei Islands, Raja Ampat, and Cendrawasih Bay. Major branching events occurred within the Pleistocene, and the patterns suggest speciation occurred in allopatry, parapatry and sympatry within the CT. On one hand, the timing of speciation events and present-day species distributions suggest that sea level fluctuations and subsequent isolation of water bodies likely played a part in the speciation process of some groups, yet other lineages appear to have diverged in the absence of hard physical barriers to gene flow. This suggests that a combination of both physical and biological factors is responsible for driving dwarfgoby speciation within the CT. The pelagic larval duration (PLD) for Eviota species is 24-26 days, which is not exceptionally short when compared to other fishes such as damselfishes and blennioids (Depczynski and Bellwood, 2006; Wellington and Victor, 1989; Riginos and Victor,

2001). Biophysical modeling suggests that regional self-recruitment may be high within the CT

103

and may explain regional genetic structuring in individuals that are capable of settling after 15-

30 days in the water column (Kool et al., 2011). Several studies have demonstrated that PLD is not a strong predictor of true dispersal capability, range size or levels of genetic structure (Victor and Wellington, 2000; Imron et al., 2007; Weersing and Toonen, 2009; Mora et al., 2012; Sotka,

2012; Luiz et al., 2013; but see Riginos and Victor, 2001). Instead, larval behavior, adult characteristics, or environmental traits may be more important (Woodson and McManus, 2007;

Sotka, 2012; Luiz et al., 2013). Larvae that actively forage may disperse hundreds of km less than previous dispersal estimates, as species may actively avoid high-flow offshore with less primary productivity (Woodson and McManus, 2007). This may be the case for Eviota larvae, who exhaust yolk-sac resources in the first two days after hatching and must immediately begin foraging or risk starvation by days three or four (Sunobe and Nakazono, 1987).

In the event of possible secondary contact between Eviota lineages, interspecific competition for habitat could be a strong driver of ecological partitioning (Hobbs and Munday,

2004; Munday, 2004; Munday et al., 2001). We see some evidence of this in the sympatric contact zone of E. brahmi and E. dorsopurpurea in Milne Bay, where closely related species show strong niche partitioning and segregate into distinct groups of conspecifics, but not in

Eviota raja and the eastern Indonesian E. cf. bifasciata which frequently co-occur over a single coral head. Mating between color morphs of Eviota may also be prevented by their species- specific pre-mating behaviors and females that are highly selective for males with the strongest secondary sex characteristics (Sunobe 1998; Sunobe and Nakazono, 1999; Sekiya and Karino,

2004; Karino and Arai, 2006).

Lastly, the age of speciation events in our study contradict Bellwood and Meyer’s (2009) observation that CT endemics have origins dating back 4-25 Ma and likely represent the ‘last

104

stand’ for previously wide-spread species that have seen their ranges shrink over time. The discrepancy between our dates and those of Bellwood and Meyer (2009) are likely due to differences in the completeness of taxon sampling, and previous studies relying mostly on taxonomically recognized species (ignoring cryptic species), potentially inflating node ages

(Rocha and Bowen, 2008). Nonetheless, our findings highlight the importance of the CT for generating new biodiversity, especially in groups with limited dispersal potential, that display evidence of local niche partitioning, and that have potential for assortative mating. The presence of numerous cryptic lineages across the Eviota phylogeny suggests that previous estimates or species richness in Eviota, and of cryptobenthic reef fishes in the CT in general, may be significantly underestimated. This places increased responsibility on taxonomists to integrate this phylogeographic information into their naming and circumscribing evolutionarily distinct lineages; named species are far more likely to benefit from international conservation legislation and planning tools (Mace, 2004).

Acknowledgements

We thank David Greenfield of the California Academy of Sciences for many valuable discussions on Eviota taxonomy. Gerry Allen kindly contributed several specimens from Papua

New Guinea and information on species distributions and habitat. We thank Gabby Ahmadia,

Jocelyn Curtis Quick and the staff of Conservation International's Indonesia marine program for helping acquire many specimens used in this study. Dave Catania at California Academy of

Sciences, and Leo Smith and Andy Bentley at University of Kansas, Biodiversity Institute provided curatorial assistance. Funding for this project was provided by the American Museum of Natural History Lerner Gray Award (LT), the Louis Stokes Alliances for Minority

105

Participation program (SV), the Paine Family Trust (ME), the Walton Family Foundation (ME), the Henry Foundation (ME) and NSF OISE-0553910 (FP).

References

Ahmadia, G.N., Pezold, F., Smith, D.J., 2012. Cryptobenthic fish biodiversity and microhabitat

use in healthy and degraded coral reefs in SE Sulawesi, Indonesia. Mar. Biodivers. 42, 433–

442.

Allen, G.R., Erdmann, M. V, 2006. Cirrhilabrus cenderawasih, a new wrasse (Pisces: Labridae)

from Papua, Indonesia. Aqua, Int. J. Ichthyol. 11, 125–131.

Allen, G.R., Erdmann, M. V, 2012. Reef fishes of the East Indies. Tropical Reef Research, Perth.

Allen, G.R., 2001. Description of two new gobies (Eviota, Gobiidae) from Indonesian seas.

Aqua, Int. J. Ichthyol. 4, 125–130.

Allen, G.R., Werner, T.B., 2002. Coral reef fish assessment in the “coral triangle” of

southeastern Asia. Environ. Biol. Fishes 65, 209–214.

Allen, G.R., Brooks, W.M., Erdmann, M. V, 2013. Eviota pamae, a new species of coral reef

goby (Gobiidae) from Indonesian seas. Aqua, Int. J. Ichthyol. 19, 79–84.

Bandelt, H.J., Forster, P., Rohl, A., 1999. Median-joining networks for inferring intraspecific

phylogenies. Mol. Biol. Evol. 16, 37–48. Barber, P.H., 2009. The challenge of

understanding the Coral Triangle biodiversity hotspot. J. Biogeogr. 36, 1845–1846.

106

Barber, P.H., Erdmann, M. V, Palumbi, S.R., 2006. Comparative phylogeography of three

codistributed stomatopods: Origins and timing of regional lineage diversification in the

coral triangle. Evolution (N. Y). 60, 1825–1839.

Barber, P.H., Palumbi, S.R., Erdmann, M. V, Moosa, M.K., 2002. Sharp genetic breaks among

populations of Haptosquilla pulchella (Stomatopoda) indicate limits to larval transport:

patterns, causes, and consequences. Mol. Ecol. 11, 659–674. doi:DOI 10.1046/j.1365-

294X.2002.01468.x

Bellwood, D.R., Hughes, T.P., 2001. The state of coral reef science - Response. Science (80-. ).

293, 1997.

Bellwood, D.R., Meyer, C.P., 2009. Searching for heat in a marine biodiversity hotspot. J.

Biogeogr. 36, 569–576.

Bellwood, D.R., Hoey, A.S., Ackerman, J.L., Depczynski, M., 2006. Coral bleaching, reef fish

community phase shifts and the resilience of coral reefs. Glob. Chang. Biol. 12, 1587–1594.

doi:DOI 10.1111/j.1365-2486.2006.01204.x

Benzie, J.A.H., Williams, S.T., 1997. Genetic structure of giant clam (Tridacna maxima)

populations in the west Pacific is not consistent with dispersal by present-day ocean

currents. Evolution (N. Y). 51, 768–783. doi:Doi 10.2307/2411153

Bowen, B.W., Rocha, L.A., Toonen, R.J., Karl, S.A., 2013. The origins of tropical marine

biodiversity. Trends Ecol. Evol. 28, 359–366.

107

Briggs, J.C., 1999a. Extinction and replacement in the Indo-West Pacific Ocean. J. Biogeogr. 26,

777–783.

Briggs, J.C., 1999b. Coincident biogeographic patterns: Indo-West Pacific Ocean. Evolution (N.

Y). 53, 326–335.

Briggs, J.C., 1999. Modes of speciation: Marine Indo-West Pacific. Bull. Mar. Sci. 65, 645–656.

Briggs, J.C., 2005. The marine East Indies: Diversity and speciation. J. Biogeogr. 32, 1517–

1522.

Briggs, J.C., Bowen, B.W., 2013. Marine shelf habitat: biogeography and evolution. J. Biogeogr.

40, 1023–1035.

Brown, J.M., Hedtke, S.M., Lemmon, A.R., Lemmon, E.M., 2010. When trees grow too long:

Investigating the causes of highly inaccurate Bayesian branch-length estimates. Syst. Biol.

59, 145–161. doi:DOI 10.1093/sysbio/syp081

Carpenter, K.E., Barber, P.H., Crandall, E.D., Ablan-Lagman, M.C.A., Ambariyanto, Mahardika,

G.N., Manjaji-Matsumoto, B.M., Juinio-Menez, M.A., Santos, M.D., Starger, C.J., Toha,

A.M.A., 2011. Comparative phylogeography of the Coral Triangle and implications for

marine management. J. Mar. Biol. 201, 1–14.

Cole, K.S., 1990. Patterns of gonad structure in hermaphroditic gobies (Teleostei, Gobiidae).

Environ. Biol. Fishes 28, 125–142.

108

Colin, P.L., 1975. The neon gobies. Comparative biology of the gobies of the genus Gobiosoma,

subgenus Elacatinus (Pisces: Gobiidae) in the Tropical Western Atlantic. T.F.H.

Publications, Neptune City.

Coyne, J.A., Orr, H.A., 2004. Speciation. Sinauer Associates, Sunderland.

Crandall, E.D., Jones, M.E., Munoz, M.M., Akinronbi, B., Erdmann, M. V, Barber, P.H., 2008.

Comparative phylogeography of two seastars and their ectosymbionts within the Coral

Triangle. Mol. Ecol. 17, 5276–5290.

DeBoer, T.S., Subia, M.D., Ambariyanto, Erdmann, M. V, Kovitvongsa, K., Barber, P.H., 2008.

Phylogeography and limited genetic connectivity in the endangered boring giant clam

across the Coral Triangle. Conserv. Biol. 22, 1255–1266. doi:DOI 10.1111/j.1523-

1739.2008.00983.x

Deloach, N., 1999. Reef fish behavior: Florida, Caribbean, Bahamas. New World Publications,

Inc., Jacksonville, FL.

Depczynski, J., Bellwood, D.R., 2004. Microhabitat utilization patters in cryptobenthic reef fish

communities. Mar. Biol. 145, 455–463.

Depczynski, M., Bellwood, D.R., 2003. The role of cryptobenthic reef fishes in coral reef

trophodynamics. Mar. Ecol. Ser. 256, 183–191.

Depczynski, M., Bellwood, D.R., 2005. Shortest recorded vertebrate lifespan found in a coral

reef fish. Curr. Biol. 15, R288–R289.

109

Depczynski, M., Bellwood, D.R., 2006. Extremes, plasticity, and invariance in vertebrate life

history traits: Insights from coral reef fishes. Ecology 87, 3119–3127.

Drew, J., Allen, G.R., Kaufman, L., Barber, P.H., 2008. Endemism and regional color and

genetic differences in five putatively cosmopolitan reef fishes. Conserv. Biol. 22, 965–975.

doi:DOI 10.1111/j.1523-1739.2008.01011.x

Drew, J.A., Allen, G.R., Erdmann, M. V, 2010. Congruence between mitochondrial genes and

color morphs in a coral reef fish: population variability in the Indo-Pacific damselfish

Chrysiptera rex (Snyder, 1909). Coral Reefs 29, 439–444. doi:DOI 10.1007/s00338-010-

0586-5

Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012. Bayesian phylogenetics with

BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973. doi:DOI

10.1093/molbev/mss075

Ekman, S., 1953. Zoogeography of the sea. Sidgwick and Jackson, London.

Eschmeyer, W.N. (2014). Catalog of fishes: genera, species, references.

(http://research.calacademy.org/research/ichthyology/catalog/fishcatmain.asp). Electronic

version accessed 10 March 2014.

Gaither, M.R., Rocha, L.A., 2013. Origins of species richness in the Indo-Malay-Philippine

biodiversity hotspot: evidence for the centre of overlap hypothesis. J. Biogeogr. 40, 1638–

1648.

110

Gaither, M.R., Schultz, J.K., Bellwood, D.R., Pyle, R.L., DiBattista, J.D., Rocha, L.A., Bowen,

B.W., 2014. Evolution of pygmy angelfishes: Recent divergences, introgression, and the

usefulness of color in taxonomy. Molecular Phylogenetics and Evolution 74, 38-47.

Godfrey, J.S., 1996. The effect of the Indonesian Throughflow on ocean circulation and heat

exchange with the atmosphere: A review. J. Geophys. Res. 101, 12217–12237. doi:Doi

10.1029/95jc03860

Graham, N.A.J., Chabanet, P., Evans, R.D., Jennings, S., Letourneur, Y., MacNeil, M.A.,

McClanahan, T.R., Ohman, M.C., Polunin, N.V.C., Wilson, S.K., 2011. Extinction

vulnerability of coral reef fishes. Ecol. Lett. 14, 341–348. doi:DOI 10.1111/j.1461-

0248.2011.01592.x

Greenfield, D.W., Erdmann, M. V, 2014. Eviota tetha, a new species of dwarfgoby from

Cenderawasih Bay, West Papua, Indonesia (Teleostei: Gobiidae). J. Ocean Sci. Found. 11,

23–31.

Greenfield, D.W., Randall, J.E., 1999. Two new Eviota species from the Hawaiian Islands

(Teleostei : Gobiidae). Copeia 439–446.

Greenfield, D.W., Randall, J.E., 2011. Two new Indo-Pacific species in the Eviota nigriventris

complex (Teleostei: Gobiidae). Zootaxa 54–66.

Greenfield, D.W., Tornabene, L., 2014. Eviota brahmi n.sp. from Papua New Guinea, with a

redescription of Eviota nigriventris (Teleostei: Gobiidae). Zootaxa 3793, 133–146.

111

Halas, D., Winterbottom, R., 2009. A phylogenetic test of multiple proposals for the origins of

the East Indies coral reef biota. J. Biogeogr. 36, 1847–1860.

Herler, J., Hilgers, H., 2005. A synopsis of coral and coral-rock associated gobies (Pisces:

Gobiidae) from the Gulf of Aqaba, northern Red Sea. Aqua, Int. J. Ichthyol. 10, 103–132.

Herler, J., 2007. Microhabitats and ecomorphology of coral- and coral rock-associated gobiid

fish (Teleostei : Gobiidae) in the northern Red Sea. Mar. Ecol. Evol. Perspect. 28, 82–94.

Herler, J., Munday, P.L., Hernaman, V., 2011. Gobies on Coral Reefs, in: Patzner, R.A., Van

Tassel, J.L., Kovacic, M., Kapoor, B.G. (Eds.), The Biology of Gobies. Science Publishers,

Enfield, NH.

Hill, K.C., Hall, R., 2003. Mesozoic-Cenozoic evolution of Australia’s New Guinea margin in a

west Pacific context, in: Hillis, R.R., Müler, R.D. (Eds.), Evolution and dynamics of the

Australian Plate. Geological Society of Australia Special Publication 22 and Geological

Society of America Special Paper 372, pp. 265–290. .

Hixon, M.A., 2011. 60 Years of Coral Reef Fish Ecology: Past, present, future. Bull. Mar. Sci.

87, 727–765. doi:DOI 10.5343/bms.2010.1055

Hobbs, J.P.A., Munday, P.L., 2004. Intraspecific competition controls spatial distribution and

social organisation of the coral-dwelling goby Gobiodon histrio. Mar. Ecol. Prog. Ser. 278,

253–259. doi:Doi 10.3354/Meps278253

Hodge, J.R., Read, C.I., van Herwerden, L., Bellwood, D.R., 2012. The role of peripheral

endemism in species diversification: Evidence from the coral reef fish genus Anampses

112

(Family: Labridae). Mol. Phylogenet. Evol. 62, 653–663. doi:DOI

10.1016/j.ympev.2011.11.007

Huelsken, T., Keyse, J., Liggins, L., Penny, S., Treml, E.A., Riginos, C., 2013. A novel

widespread cryptic species and phylogeographic patterns within several giant clam species

(Cardiidae: Tridacna) from the Indo-Pacific Ocean. PLoS One 8. doi:ARTN e80858 DOI

10.1371/journal.pone.0080858

Imron, Jeffrey, B., Hale, P., Degnan, B.M., Degnan, S.M., 2007. Pleistocene isolation and recent

gene flow in Haliotis asinina, an Indo-Pacific vetigastropod with limited dispersal capacity.

Mol. Ecol. 16, 289–304. doi:DOI 10.1111/j.1365-294X.2006.03141.x

Karino, K., Arai, R., 2006. Effect of clutch size on male egg-fanning behavior and hatching

success in the goby, Eviota prasina (Klunzinger). J. Exp. Mar. Bio. Ecol. 334, 43–50.

Keith, P., Lord, C., 2011. Systematics of Sicydiinae, in: Patzner, R.A., Van Tassel, J.L., Kovacic,

M., Kapoor, B.G. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH, pp. 119–

128.

Kool, J.T., Paris, C.B., Barber, P.H., Cowen, R.K., 2011. Connectivity and the development of

population genetic structure in Indo-West Pacific coral reef communities. Glob. Ecol.

Biogeogr. 20, 695–706. doi:DOI 10.1111/j.1466-8238.2010.00637.x

Ladd, H.S., 1960. Origin of the Pacific island molluscan fauna. Am. J. Sci. 258, 137–150.

113

Landry, C., Geyer, L.B., Arakaki, Y., Uehara, T., Palumbi, S.R., 2003. Recent speciation in the

Indo-West Pacific: Rapid evolution of gamete recognition and sperm morphology in cryptic

species of sea urchin. Proc. R. Soc. Biol. Sci. Ser. B 270, 1839–1847.

Lourie, S.A., Green, D.M., Vincent, A.C.J., 2005. Dispersal, habitat differences, and

comparative phylogeography of Southeast Asian seahorses (Syngnathidae : Hippocampus).

Mol. Ecol. 14, 1073–1094. doi:DOI 10.1111/j.1365-294X.2005.02464.x

Luiz, O.J., Allen, A.P., Robertson, D.R., Floeter, S.R., Kulbicki, M., Vigliola, L., Becheler, R.,

Madin, J.S., 2013. Adult and larval traits as determinants of geographic range size among

tropical reef fishes. Proc. Natl. Acad. Sci. U. S. A. 110, 16498–16502. doi:DOI

10.1073/pnas.1304074110

Mace, G.M., 2004. The role of taxonomy in species conservation. Philos. Trans. R. Soc. London

B Biol. Sci. 359, 711–719.

Marshall, D.C., 2010. Cryptic Failure of Partitioned Bayesian Phylogenetic Analyses: Lost in the

Land of Long Trees. Syst. Biol. 59, 108–117. doi:DOI 10.1093/sysbio/syp080

McKenna, M.C., 1973. Sweepstakes, filters, corridors, Noah’s Arks, and beached Viking funeral

ships in palaeogeography., in: Tarling, D.H., Runcorn, S.K. (Eds.), Implications of

Continental Drift to the Earth Sciences. Academic Press, New York, pp. 295–308.

Mironov, 2006. Centers of marine fauna redistribution. Entomol. Rev. 86, S32–S44.

Mora, C., Chittaro, P.M., Sale, P.F., Kritzer, J.P., Ludsin, S.A., 2003. Patterns and processes in

reef fish diversity. Nature 421, 933–936. doi:Doi 10.1038/Nature01393

114

Mora, C., Treml, E.A., Roberts, J., Crosby, K., Roy, D., Tittensor, D.P., 2012. High connectivity

among habitats precludes the relationship between dispersal and range size in tropical reef

fishes. Ecography (Cop.). 35, 89–96. doi:DOI 10.1111/j.1600-0587.2011.06874.x

Munday, P.L., 2004a. Habitat loss, resource specialization, and extinction on coral reefs. Glob.

Chang. Biol. 10, 1642–1647. doi:DOI 10.1111/j.1365-2486.2004.00839.x

Munday, P.L., 2004b. Competitive coexistence of coral-dwelling fishes; the lottery hypothesis

revisited. Ecology 85, 623–628.

Munday, P.L., Jones, G.P., Caley, M.J., 2001. Interspecific competition and coexistence in a

guild of coral-dwelling fishes. Ecology 82, 2177–2189. doi:Doi 10.1890/0012-

9658(2001)082[2177:Icacia]2.0.Co;2

Munday, P.L., van Herwerden, L., Dudgeon, C.L., 2004. Evidence for sympatric speciation by

host shift in the sea. Curr. Biol. 14, 1498–1504. doi:DOI 10.1016/j.cub.2004.08.029

Murdy, E.O., 2011a. Systematics of Amblyopinae, in: Patzner, R.A., Van Tassel, J.L., Kovacic,

M., Kapoor, B.G. (Eds.), The Biology of Gobies. Science Publishes, Enfield, NH, pp. 107–

118.

Murdy, E.O., 2011b. Systematics of Oxudercinae, in: Patzner, R.A., Van Tassel, J.L., Kovacic,

M., Kapoor, B.G. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH, pp. 99–

106.

115

Patzner, R.A., Van Tassell, J.L., Kovačić, M., Kapoor, B.G. (Eds.), 2011. The Biology of

Gobies, 1st ed. Science Publishers ; DIstributed by CRC Press, Enfield, NH Boca Raton,

FL.

Pellissier, L., Leprieur, F., Parravicini, V., Cowman, P.F., Kulbicki, M., Litsios, G., Olsen, S.M.,

Wisz, M.S., Bellwood, D.R., Mouillot, D., 2014. Quaternary coral reef refugia preserved

fish diversity. Science (80-. ). 344, 1016–1019.

Perrin, C., Borsa, P., 2001. Mitochondrial DNA analysis of the geographic structure of Indian

scad mackerel in the Indo-Malay archipelago. J. Fish Biol. 59, 1421–1426.

Pezold, F., 2011. Systematics of the family Gobionellidae, in: Patzner, R.A., Van Tassell, J.L.,

Kovacic, M., Kapoor, B.G. (Eds.), The Biology of Gobies. CRC Press, Oxfod and IBH

Publishing Co., New Delhi.

Posada, D., 2008. jModelTest: phylogenetic model averaging. Mol Biol Evol 25, 1253–1256.

doi:msn083 [pii] 10.1093/molbev/msn083

Pratchett, M.S., Baird, A.H., McCowan, D.M., Coker, D.J., Cole, A.J., Wilson, S.K., 2008.

Protracted declines in coral cover and fish abundance following climate-induced coral

bleaching on the Great Barrier Reef. Proc. 11th Int. Coral Reef Symp. 1309–1313.

Pratchett, M.S., Wilson, S.K., Baird, A.H., 2006. Declines in the abundance of Chaetodon

butterflyfishes (Chaetodontidae) following extensive coral depletion. J. Fish Biol. 69, 1269–

1280.

116

Rambaut, A., and Drummond, A.J. 2007 Tracer version 1.5. Available at

http://tree.bio.ed.ac.uk/software/tracer/.

Ratnasingham, S. and Hebert, P.D.N. 2013. A DNA-based registry for all animal apecies: the

Barcode Index Number (BIN) System. PloS ONE 8 (3): e66213.

DOI:10.1371/journal.pone.0066213

Riginos, C., Victor, B.C., 2001. Larval spatial distributions and other early life-history

characteristics predict genetic differentiation in eastern Pacific blennioid fishes. Proc. R.

Soc. London Ser. B-Biological Sci. 268, 1931–1936.

Remington, C.L., 1968. Suture-zones of hybrid interaction between recently joined biotas. Evol.

Biol. 2, 321–428.

Renema, W., Bellwood, D.R., Braga, J.C., Bromfield, K., Hall, R., Johnson, K.G., Lunt, P.,

Meyer, C.P., McMonagle, L.B., Morley, R.J., O’Dea, A., Todd, J.A., Wesselingh, F.P.,

Wilson, M.E.J., Pandolfi, J.M., 2008. Hopping hotspots: Global shifts in marine

biodiversity. Science (80-. ). 321, 654–657. doi:DOI 10.1126/science.1155674

Rocha, L.A., Bowen, B.W. (2008) Speciation in coral-reef fishes. Journal of Fish Biology, 72,

1101-1121.

Sale, P.F., 1993. The ecology of fishes on coral reefs. Academic Press, San Diego, CA.

Sale, P.F., 2002. Coral Reef Fishes: dynamics and diversity in a complex ecosystem.

Santini, F., Winterbottom, R., 2002. Historical biogeography of Indo-western Pacific coral reef

biota: is the Indonesian region a centre of origin? J. Biogeogr. 29, 189–205.

117

Schiller, A., Oke, P.R., Brassington, G., Entel, M., Fiedler, R., Griffin, D.A., Mansbridge, J. V,

2008. Eddy-resolving ocean circulation in the Asian-Australian region inferred from an

ocean reanalysis effort. Prog. Oceanogr. 76, 334–365. doi:DOI

10.1016/j.pocean.2008.01.003

Sekiya, Y., Karino, K., 2004. Female mate preference in goby Eviota prasina: Do secondary

sexual traits influence female choice? Zool. Sci. 21, 859–863.

Simpson, G.G., 1977. Too many lines; the limits of the Oriental and Australian Zoogeographic

regions. Proc. Am. Philos. Soc. 121, 107–120.

Sotka, E.E., 2012. Natural selection, larval dispersal, and the geography of phenotype in the sea.

Integr. Comp. Biol. 52, 538–545.

Stehli, F.G., Wells, J.W., 1971. Diversity and age patterns in hermatypic corals. Syst. Biol. 20,

115–126.

Sunobe, T., 1998. Reproductive behavior in six species of Eviota (Gobiidae) in aquaria. Ichthyol.

Res. 45, 409–412.

Sunobe, T., Nakazono, 1999. Alternative mating tacitics in the gobiid fish, Eviota prasina.

Ichthyol. Res. 46, 212–215.

Sunobe, T., Nakazono, A., 1987. Embryonic development and larvae of genus Eviota (Pisces:

Gobiidae) I. Evota abax and E. storthynx. J. Fac. Agric. Kyushu Univ. 31, 287–295.

118

Thacker, C.E., 2003. Molecular phylogeny of the gobioid fishes (Teleostei: Perciformes:

Gobioidei). Mol. Phylogenet. Evol. 26, 354–368.

Thacker, C.E., 2011. Systematics of Gobiidae, in: Patzner, R.A., Van Tassel, J.L., Kovacic, M.,

Kapoor, B.G. (Eds.), The Biology of Gobies. Science Publishers, Enfield, NH, pp. 129–138.

Thacker, C.E., Roje, D.M., 2011. Phylogeny of Gobiidae and identification of gobiid lineages.

Syst. Biodivers. 9, 329–347. doi:Doi 10.1080/14772000.2011.629011

Timm, J., Figiel, M., Kochzius, M., 2008. Contrasting patterns in species boundaries and

evolution of anemonefishes (Amphiprioninae, Pomacentridae) in the centre of marine

biodiversity. Mol. Phylogenet. Evol. 49, 268–276. doi:DOI 10.1016/j.ympev.2008.04.024

Tornabene, L., Ahmadia, G.N., Berumen, M.L., Smith, D.J., Jompa, J., Pezold, F., 2013a.

Evolution of microhabitat association and morphology in a diverse group of cryptobenthic

coral reef fishes (Teleostei: Gobiidae: Eviota). Mol. Phylogenet. Evol. 66, 391–400.

Tornabene, L., Ahmadia, G.N., Williams, J.T., 2013b. Four new species of dwarfgobies

(Teleostei: Gobiidae: Eviota) from the Austral, Gambier, Marquesas and Society

Archipelagos, French Polynesia. Syst. Biodivers. 11, 363–380. doi:Doi

10.1080/14772000.2013.819822

Veron, J.E.N., 1995. Corals in space and time. Cornell University Press, Ithaca.

Veron, J.E.N., Devantier, L.M., Turak, E., Green, A., Kininmonth, S., Stafford-Smith, M.,

Peterson, N., 2009. Delineating the Coral Triangle. Galaxea, J. Coral Reef Stud. 11, 91–

100.

119

Victor, B.C., Wellington, G.M., 2000. Endemism and the pelagic larval duration of reef fishes in

the eastern Pacific Ocean. Mar. Ecol. Prog. Ser. 205, 241–248. doi:Doi

10.3354/Meps205241

Voris, H.K., 2000. Maps of Pleistocene sea levels in Southeast Asia: Shorelines, river systems

and time durations. J. Biogeogr. 27, 1153–1167.

Ward, R.D., Zemlak, T.S., Innes, B.H., Last, P.R., Hebert, P.D.N., 2005. DNA barcoding

Australia’s fish species. Philos. Trans. R. Soc. London B Biol. Sci. 360, 1847–1857.

Weersing, K., Toonen, R.J., 2009. Population genetics, larval dispersal, and connectivity in

marine systems. Mar. Ecol. Prog. Ser. 393, 1–12. doi:Doi 10.3354/Meps08287

Wellington, G.M., Victor, B.C., 1989. Planktonic larval duration of one hundred species of

Pacfiic and Atlantic damselfishes (Pomacentridae). Mar. Biol. 101, 557–567.

Winterbottom, R., Hanner, R.H., Burridge, M., Zur, M., 2014. A cornucopia of cryptic species -

a DNA barcode analysis of the gobiid fish genus Trimma (Percomorpha, Gobiiformes).

Zookeys 79–111. doi:DOI 10.3897/zookeys.381.6445

Woodland, D.J., 1983. Zoogeography of the Siganidae (Pisces): an interpreation of distribution

and richness patterns. Bull. Mar. Sci. 33.

Woodson, C.B., McManus, M.A., 2007. Foraging behavior can influence dispersal of marine

organisms. Limnol. Oceanogr. 52, 2701–2709. doi:DOI 10.4319/lo.2007.52.6.2701

120

Wray, C.G., Landman, N.H., Bonacum, J., 1995. Genetic-divergence and geographic

diversification in Nautilus. Paleobiology 21, 220–228.

Yamada, T., Sugiyama, T., Tamaki, N., Kawakita, A., Kato, M., 2009. Adaptive radiation of

gobies in the interstitial habitats of gravel beaches accompanied by body elongation and

excessive vertebral segmentation. Bmc Evol. Biol. 9. doi:Artn 145 Doi 10.1186/1471-2148-

9-145

121

SUPPLEMENTARY MATERIAL

Table 5 Specimens used in phylogeography study.

Genus Species Sample Voucher Genbank COI Genbank Ptr Eviota albolineata T5695 SAIAB 86191 KP013294 KP013374 Eviota albolineata n/a AMNH 256486 JX483989 JX483966 Eviota ancora AC1 CAS 237493 KP013213 KP013299 Eviota atriventris EA4 CAS 237494 KP013228 KP013313 Eviota atriventris EA5 CAS 237494 KP013229 KP013314 Eviota atriventris EA6 CAS 237494 KP013230 KP013315 Eviota atriventris EA7 CAS 237494 KP013231 KP013316 Eviota atriventris NI1 CAS 237495 KP013282 KP013364 Eviota bifasciata EB13 CAS 237497 KP013234 KP013319 Eviota bifasciata EB3 CAS 237496 KP013240 KP013325 Eviota bifasciata EB4 CAS 237497 KP013241 KP013326 Eviota bifasciata EB5 CAS 237497 KP013242 KP013327 Eviota bifasciata EB6 CAS 237497 KP013243 KP013328 Eviota bifasciata EB7 CAS 237497 KP013244 N/A Eviota brahmi NS1 CAS 236622 KJ439732 KJ439705 Eviota brahmi NS2 CAS 236622 KJ439733 KJ439706 Eviota brahmi NS3 CAS 236622 KJ439734 KJ439707 Eviota brahmi NS4 CAS 236622 KJ439735 KJ439708 Eviota brahmi NS5 CAS 236622 KJ439736 KJ439709 Eviota brahmi NS7 CAS 236622 KJ439737 KJ439710 Eviota brahmi NS8 CAS 236622 KP013285 N/A Eviota brahmi NS9 CAS 236622 KJ439738 KJ439711 Eviota cf. atriventris EA1 WAM P. 34290-001 KP013226 KP013311 Eviota cf. atriventris EA2 WAM P. 34290-001 KP013227 KP013312 Eviota cf. bifasciata DP4 WAM P. 34287-001 KP013225 KP013310 Eviota cf. bifasciata EB1 CAS 237498 KP013232 KP013317 Eviota cf. bifasciata EB10 CAS 237500 KP013233 KP013318 Eviota cf. bifasciata EB16 CAS 237501 KP013235 KP013320 Eviota cf. bifasciata EB17 CAS 237501 KP013236 KP013321 Eviota cf. bifasciata EB18 CAS 237501 KP013237 KP013322 Eviota cf. bifasciata EB19 CAS 237501 KP013238 KP013323 Eviota cf. bifasciata EB2 CAS 237498 KP013239 KP013324 Eviota cf. bifasciata EB8 CAS 237499 KP013245 N/A Eviota cf. bifasciata EB9 CAS 237499 KP013298 KP013329 Eviota cf. bifasciata NV10 CAS 237511 KP013286 KP013366 Eviota cf. dorsogilva DG12 WAM P. 34289-001 KP013214 KP013300 Eviota cf. dorsogilva DG13 WAM P. 34289-001 KP013215 KP013301 Eviota cf. dorsogilva DG14 WAM P. 34289-001 KP013216 KP013302 Eviota cf. dorsogilva DG15 WAM P. 34289-001 KP013217 KP013303 Eviota cf. dorsogilva DG16 WAM P. 34288-001 KP013218 KP013304 Eviota cf. dorsogilva DG17 WAM P. 34288-001 KP013219 KP013305 Eviota cf. dorsogilva DG18 WAM P. 34288-001 KP013220 KP013306 Eviota cf. dorsogilva DG19 WAM P. 34288-001 KP013221 KP013307 Eviota cf. dorsogilva DG4 WAM P. 34289-001 KP013222 KP013308 Eviota cf. dorsogilva DG5 WAM P. 34289-001 KP013223 N/A Eviota cf. dorsogilva DG8 WAM P. 34289-001 KP013224 KP013309

122

Table 6 Distance matrix showing % identical sites across samples in the E. nigriventris complex.

Colors for sample names correspond to clades in Fig. 13.

93.7 93.7 93.8 93.7 93.9 94.0 94.0 97.1 97.2 97.2 97.1 97.1 97.1 97.3 98.7 98.9 99.4 99.2 99.2 99.3 99.3 99.2 98.6 98.6 98.4 92.2 92.0 92.2 92.3 92.3 92.4 92.6 92.7 92.6 92.6 92.5 92.6 92.5 DG2 NV9 93.7 93.7 93.8 93.7 93.9 94.0 94.0 97.1 97.2 97.2 97.1 97.1 97.1 97.3 98.8 99.4 99.8 99.7 99.7 99.7 99.7 99.2 98.7 98.7 98.6 92.3 92.1 92.3 92.3 92.3 92.4 92.4 92.5 92.6 92.6 92.5 92.6 92.5 93.7 93.7 93.8 93.7 93.9 94.0 94.0 97.1 97.2 97.2 97.1 97.1 97.1 97.3 98.9 99.5 99.9 99.7 99.7 99.8 99.7 99.3 98.8 98.8 98.6 92.2 92.0 92.2 92.3 92.3 92.4 92.4 92.5 92.6 92.6 92.5 92.6 92.5 NV13 NV3 93.7 93.7 93.8 93.7 93.9 94.0 94.0 97.1 97.2 97.2 97.1 97.1 97.1 97.3 98.9 99.5 99.9 99.7 99.9 99.8 99.7 99.3 98.8 98.8 98.6 92.2 92.0 92.2 92.3 92.3 92.4 92.4 92.5 92.6 92.6 92.5 92.6 92.5 93.7 93.7 93.8 93.7 93.9 94.0 94.0 97.0 97.1 97.1 97.0 97.0 97.0 97.2 98.8 99.4 99.8 99.6 99.9 99.7 99.7 99.2 98.7 98.7 98.5 92.2 92.0 92.1 92.2 92.2 92.4 92.4 92.5 92.6 92.6 92.5 92.6 92.5 DG3 NV8 93.6 93.6 93.7 93.6 93.8 93.9 93.9 97.2 97.3 97.3 97.2 97.2 97.2 97.3 98.8 99.4 99.8 99.6 99.7 99.7 99.7 99.2 98.7 98.7 98.5 92.5 92.3 92.5 92.6 92.6 92.7 92.7 92.8 92.9 92.9 92.8 92.9 92.8 NV2 93.8 93.8 93.9 93.8 94.0 94.1 94.1 97.2 97.3 97.3 97.2 97.2 97.2 97.3 99.0 99.5 99.8 99.8 99.9 99.9 99.8 99.4 98.9 98.9 98.7 92.3 92.1 92.3 92.4 92.4 92.5 92.5 92.6 92.7 92.7 92.6 92.7 92.6 93.5 93.5 93.6 93.5 93.7 93.8 93.8 96.7 96.8 96.8 96.7 96.7 96.7 96.9 98.5 99.5 99.4 99.4 99.5 99.5 99.4 98.9 98.4 98.4 98.3 92.2 92.0 92.2 92.3 92.3 92.4 92.4 92.5 92.6 92.6 92.5 92.6 92.5 DG1 NV5 93.9 93.9 94.0 93.9 94.1 94.0 94.0 97.6 97.7 97.7 97.6 97.6 97.6 97.8 98.5 99.0 98.8 98.8 98.9 98.9 98.8 98.7 99.4 99.4 99.2 92.2 92.0 92.2 92.3 92.3 92.4 92.6 92.7 92.6 92.6 92.5 92.6 92.5 93.6 93.6 93.7 93.6 93.8 93.5 93.5 99.6 99.7 99.7 99.6 99.6 99.6 97.8 96.9 97.3 97.3 97.2 97.3 97.3 97.3 97.3 97.5 97.5 97.3 92.3 92.1 92.1 92.2 92.2 92.3 92.5 92.6 92.5 92.5 92.4 92.5 92.4 DG4 93.4 93.4 93.5 93.4 93.6 93.5 93.5 99.8 99.9 99.9 99.8 99.8 99.6 97.6 96.7 97.2 97.2 97.0 97.1 97.1 97.1 97.1 97.5 97.5 97.3 92.1 92.0 92.0 92.0 92.0 92.1 92.3 92.4 92.3 92.3 92.2 92.3 92.2 DG8 93.4 93.4 93.5 93.4 93.6 93.5 93.5 99.8 99.9 99.9 99.8 99.8 99.6 97.6 96.7 97.2 97.2 97.0 97.1 97.1 97.1 97.1 97.5 97.5 97.3 92.1 92.0 92.0 92.0 92.0 92.1 92.3 92.4 92.3 92.3 92.2 92.3 92.2 DG5 93.4 93.4 93.5 93.4 93.6 93.5 93.5 99.8 99.9 99.9 99.8 99.8 99.6 97.6 96.7 97.2 97.2 97.0 97.1 97.1 97.1 97.1 97.5 97.5 97.3 92.1 92.0 92.0 92.0 92.0 92.1 92.3 92.4 92.3 92.3 92.2 92.3 92.2 DG12 complex. Colors for sample names correspond to clades on Figure 4. Figure on clades to correspond names sample for Colors complex. 93.5 93.5 93.6 93.5 93.7 93.6 93.6 99.9 99.9 99.9 99.9 99.7 97.7 96.8 97.3 97.3 97.1 97.2 97.2 97.2 97.2 97.6 97.6 97.4 92.2 92.0 92.0 92.1 92.1 92.2 92.4 92.5 92.4 92.4 92.3 92.4 92.3 100.0 DG15 93.5 93.5 93.6 93.5 93.7 93.6 93.6 99.9 99.9 99.9 99.9 99.7 97.7 96.8 97.3 97.3 97.1 97.2 97.2 97.2 97.2 97.6 97.6 97.4 92.2 92.0 92.0 92.1 92.1 92.2 92.4 92.5 92.4 92.4 92.3 92.4 92.3 100.0 DG14 E. nigriventris E. nigriventris 93.6 93.6 93.7 93.6 93.8 93.7 93.7 99.9 99.9 99.8 99.8 99.8 99.6 97.6 96.7 97.2 97.2 97.0 97.1 97.1 97.1 97.1 97.5 97.5 97.3 92.3 92.1 92.1 92.2 92.2 92.3 92.5 92.6 92.5 92.5 92.4 92.5 92.4 DG13 FD3 98.2 98.2 98.3 98.2 98.2 99.8 93.7 93.6 93.6 93.5 93.5 93.5 93.5 94.0 93.8 94.1 93.9 94.0 94.0 94.0 94.0 94.0 93.7 93.7 93.6 92.8 92.6 92.5 92.6 92.6 93.0 92.9 93.0 92.9 92.9 93.0 93.1 93.0 FD2 98.2 98.2 98.3 98.2 98.2 99.8 93.7 93.6 93.6 93.5 93.5 93.5 93.5 94.0 93.8 94.1 93.9 94.0 94.0 94.0 94.0 94.0 93.7 93.7 93.6 92.8 92.6 92.5 92.6 92.6 93.0 92.9 93.0 92.9 92.9 93.0 93.1 93.0 99.8 99.8 99.9 99.8 98.2 98.2 93.8 93.7 93.7 93.6 93.6 93.6 93.8 94.1 93.7 94.0 93.8 93.9 93.9 93.9 93.9 93.9 93.6 93.6 93.7 92.8 92.8 92.7 92.8 92.8 93.1 93.1 93.1 93.1 93.1 93.1 93.2 93.1 DG16 99.8 99.8 99.9 99.8 98.2 98.2 93.6 93.5 93.5 93.4 93.4 93.4 93.6 93.9 93.5 93.8 93.6 93.7 93.7 93.7 93.7 93.7 93.4 93.4 93.5 92.6 92.6 92.5 92.6 92.6 93.0 92.9 93.0 92.9 92.9 93.0 93.1 93.0 DG19 99.9 99.9 99.9 99.9 98.3 98.3 93.7 93.6 93.6 93.5 93.5 93.5 93.7 94.0 93.6 93.9 93.7 93.8 93.8 93.8 93.8 93.8 93.5 93.5 93.6 92.7 92.7 92.6 92.7 92.7 93.1 93.0 93.1 93.0 93.0 93.1 93.1 93.1 DG18 NS6 99.8 99.9 99.8 99.8 98.2 98.2 93.6 93.5 93.5 93.4 93.4 93.4 93.6 93.9 93.5 93.8 93.6 93.7 93.7 93.7 93.7 93.7 93.4 93.4 93.5 92.6 92.6 92.5 92.6 92.6 93.0 92.9 93.0 92.9 92.9 93.0 93.1 93.0 99.8 99.9 99.8 99.8 98.2 98.2 93.6 93.5 93.5 93.4 93.4 93.4 93.6 93.9 93.5 93.8 93.6 93.7 93.7 93.7 93.7 93.7 93.4 93.4 93.5 92.6 92.6 92.5 92.6 92.6 93.0 92.9 93.0 92.9 92.9 93.0 93.1 93.0 DG17 FD2 FD3 NV5 NV2 NV8 NV3 NV9 NV1 NV7 NV4 NS6 DP1 DP3 DP2 DP6 DP5 NS3 NS1 NS4 NS2 NS7 NS9 NS5 NS8 DG5 DG8 DG4 DG1 DG3 DG2 NV13 DG17 DG18 DG19 DG16 DG13 DG14 DG15 DG12 Sample Distance matrix showing % identical sites across samples in the the in samples across sites identical % showing matrix Distance Eviota brahmi Eviota brahmi Eviota brahmi Eviota brahmi Eviota brahmi Eviota brahmi Eviota brahmi Eviota brahmi Eviota Eviota dorsogilva Eviota dorsogilva Eviota Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota nigriventris Eviota Eviota cf. dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota dorsogilva cf. Eviota

Eviota dorsopurpurea Eviota dorsopurpurea Eviota dorsopurpurea Eviota dorsopurpurea Eviota dorsopurpurea Eviota Supplementary Table S2. Table Supplementary

123

Table 7 Distance matrix showing % identical sites across samples in the E. bifasciata complex.

Colors for sample names correspond to clades on Fig. 14.

91.3 91.3 91.2 91.5 91.2 92.2 92.5 92.5 92.5 92.5 92.5 92.7 92.0 92.5 92.5 92.5 92.4 92.2 91.9 91.9 91.9 91.9 91.9 91.7 99.8 99.8 NV10 EB2 91.2 91.2 91.0 91.3 91.0 92.1 92.4 92.4 92.4 92.4 92.4 92.5 91.9 92.4 92.4 92.4 92.2 92.0 91.7 91.7 91.7 91.7 91.7 91.5 99.7 99.8 EB1 91.2 91.2 91.2 91.3 91.3 92.1 92.4 92.4 92.4 92.4 92.4 92.5 91.9 92.4 92.4 92.4 92.2 92.4 91.7 91.7 91.7 91.7 91.7 91.5 99.7 99.8 EP5 92.0 92.0 91.7 91.9 92.2 92.2 92.5 92.5 92.5 92.5 92.5 92.4 92.4 92.5 92.5 92.5 92.4 92.5 99.8 99.8 99.8 99.7 99.7 91.5 91.5 91.7 EP1 92.0 92.0 91.7 91.9 92.0 92.2 92.5 92.5 92.5 92.5 92.5 92.4 92.4 92.5 92.5 92.5 92.4 92.5 99.8 99.8 99.8 99.7 99.7 91.7 91.7 91.9 EP3 92.1 92.1 91.8 91.9 91.9 92.3 92.6 92.6 92.6 92.6 92.6 92.5 92.5 92.6 92.6 92.6 92.5 92.6 99.9 99.9 99.9 99.7 99.7 91.7 91.7 91.9 EP4 92.2 92.2 91.9 92.0 92.0 92.4 92.7 92.7 92.7 92.7 92.7 92.5 92.5 92.7 92.7 92.7 92.5 92.7 99.9 99.8 99.8 91.7 91.7 91.9 100.0 100.0 EP2 92.2 92.2 91.9 92.0 92.0 92.4 92.7 92.7 92.7 92.7 92.7 92.5 92.5 92.7 92.7 92.7 92.5 92.7 99.9 99.8 99.8 91.7 91.7 91.9 100.0 100.0 EP6 92.2 92.2 91.9 92.0 92.0 92.4 92.7 92.7 92.7 92.7 92.7 92.5 92.5 92.7 92.7 92.7 92.5 92.7 99.9 99.8 99.8 91.7 91.7 91.9 100.0 100.0 EB3 92.4 92.4 92.0 92.2 92.9 93.8 93.8 93.8 93.8 93.8 93.8 93.6 93.6 97.4 97.2 97.2 97.1 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.0 92.2 EB6 92.7 92.7 92.2 92.5 92.5 94.0 93.9 93.9 93.9 93.9 93.9 93.8 93.8 99.7 99.8 99.8 97.1 92.5 92.5 92.5 92.5 92.4 92.4 92.2 92.2 92.4 EB5 92.9 92.9 92.4 92.7 92.7 94.2 94.1 94.1 94.1 94.1 94.1 93.9 93.9 99.8 99.8 97.2 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 EB4 92.9 92.9 92.4 92.7 92.7 94.2 94.1 94.1 94.1 94.1 94.1 93.9 93.9 99.8 99.8 97.2 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 EB7 92.9 92.9 92.4 92.7 92.7 94.2 94.1 94.1 94.1 94.1 94.1 93.9 93.9 99.8 99.8 99.7 97.4 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 ER1 93.8 93.8 93.4 93.6 93.2 99.2 99.5 99.5 99.5 99.5 99.5 99.3 93.9 93.9 93.9 93.8 93.6 92.5 92.5 92.5 92.5 92.4 92.4 91.9 91.9 92.0 ER6 93.9 93.9 93.6 93.8 93.4 99.5 99.8 99.8 99.8 99.8 99.8 99.3 93.9 93.9 93.9 93.8 93.6 92.5 92.5 92.5 92.5 92.4 92.4 92.5 92.5 92.7 complex. Colors for sample names correspond to clades on Figure 5. Figure on clades to correspond names sample for Colors complex. ER3 94.1 94.1 93.8 93.9 93.6 99.7 99.8 99.5 94.1 94.1 94.1 93.9 93.8 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 100.0 100.0 100.0 ER4 94.1 94.1 93.8 93.9 93.6 99.7 99.8 99.5 94.1 94.1 94.1 93.9 93.8 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 100.0 100.0 100.0 ER9 94.1 94.1 93.8 93.9 93.6 99.7 99.8 99.5 94.1 94.1 94.1 93.9 93.8 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 E. bifasciata 100.0 100.0 100.0 100.0 ER2 94.1 94.1 93.8 93.9 93.6 99.7 99.8 99.5 94.1 94.1 94.1 93.9 93.8 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 100.0 100.0 100.0 ER7 94.1 94.1 93.8 93.9 93.6 99.7 99.8 99.5 94.1 94.1 94.1 93.9 93.8 92.7 92.7 92.7 92.6 92.5 92.5 92.4 92.4 92.5 100.0 100.0 100.0 100.0 ER5 94.2 94.2 93.8 94.0 93.6 99.7 99.7 99.7 99.7 99.7 99.5 99.2 94.2 94.2 94.2 94.0 93.8 92.4 92.4 92.4 92.3 92.2 92.2 92.1 92.1 92.2 DP4 95.3 95.3 95.0 95.1 93.6 93.6 93.6 93.6 93.6 93.6 93.4 93.2 92.7 92.7 92.7 92.5 92.9 92.0 92.0 92.0 91.9 92.0 92.2 91.3 91.0 91.2 99.8 99.8 99.3 95.1 94.0 93.9 93.9 93.9 93.9 93.9 93.8 93.6 92.7 92.7 92.7 92.5 92.2 92.0 92.0 92.0 91.9 91.9 91.9 91.3 91.3 91.5 EB10 99.5 99.5 99.3 95.0 93.8 93.8 93.8 93.8 93.8 93.8 93.6 93.4 92.4 92.4 92.4 92.2 92.0 91.9 91.9 91.9 91.8 91.7 91.7 91.2 91.0 91.2 EB16 99.5 99.8 95.3 94.2 94.1 94.1 94.1 94.1 94.1 93.9 93.8 92.9 92.9 92.9 92.7 92.4 92.2 92.2 92.2 92.1 92.0 92.0 91.2 91.2 91.3 EB18 100.0 99.5 99.8 95.3 94.2 94.1 94.1 94.1 94.1 94.1 93.9 93.8 92.9 92.9 92.9 92.7 92.4 92.2 92.2 92.2 92.1 92.0 92.0 91.2 91.2 91.3 EB17 100.0 EP6 EP2 EP4 EP3 EP1 EP5 DP4 ER5 ER7 ER2 ER9 ER4 ER3 ER6 ER1 EB7 EB4 EB5 EB6 EB3 EB1 EB2 NV10 EB17 EB18 EB16 EB10 Distance matrix showing % identical sites across samples in the the in samples across sites identical % showing matrix Distance Sample Eviota raja Eviota raja Eviota raja Eviota raja Eviota raja Eviota raja Eviota raja Eviota raja Eviota Eviota pamae Eviota pamae Eviota pamae Eviota pamae Eviota pamae Eviota pamae Eviota Eviota bifasciata Eviota bifasciata Eviota bifasciata Eviota bifasciata Eviota bifasciata Eviota

Eviota cf. bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota bifasciata cf. Eviota Supplementary Table S3. Table Supplementary

124

SUMMARY AND FUTURE RESEARCH

This dissertation provides a multi-tier investigation of gobiid evolution. In the first chapter I used a novel set of markers to help clarify sub-familial and intergeneric relationships within the Gobiidae. My results provided a third independent line of evidence (morphology, mtDNA, and now nDNA) demonstrating that gobiid diversity is deeply divided into two ecologically divergent clades. I then investigated more fine-scale phylogenetic relationships within the genus Eviota, an abundant and diverse lineage of coral reef fishes. Here I demonstrated how lineages have switched reef microhabitats several times throughout their evolutionary history, and frequently do so independent of changes in morphology. These results suggested that variation in microhabitat preference combined with resilience to extinction may contribute to the remarkable extant species-richness in Eviota. Lastly, my third chapter demonstrated how gobies can be used to address major issues in biogeography. Here I addressed questions regarding the sources of biodiversity in the marine hotspot known as the Coral

Triangle. Through a phylogeographic survey of the Eviota nigriventris and E. bifasciata complexes, I showed that rapid diversification has occurred at very fine geographic scales within the last 1.5 million years via both allopatric and sympatric speciation, supporting the ‘Center of

Origin’ hypothesis.

Collectively, my chapters demonstrate that gobies and other ecologically diverse clades of fishes serve as excellent model groups for studying the processes driving taxonomic and phenotypic diversification in marine species at a variety of spatial and temporal scales. While the last two chapters here focus on miniature reef-associated species, similar evolutionary questions regarding the evolution of phenotype exist for unique groups such as the mudskippers

(Oxudercinae), rock-climbing gobies (Sicydiinae), and the phenotypically diverse American

125

seven-spined gobies (Gobiosomatini). All of these lineages likely represent instances where explosive periods of speciation were driven largely by a combination of niche diversification, morphological adaptation, and complex geologic and paleoceanographic changes.

Once we elucidate the phylogeny and general patterns of phenotypic evolution in these groups, we can shift our focus in one of several directions. One possibility is to look towards the present at pairs of sympatric sister species that are in the early stages of speciation and niche divergence (e.g. Eviota brahmi and E. dorsopurpurea), and search for fine-scale signatures of natural selection across the genome. By scanning the entire genome using an efficient sub sampling strategy such as RAD-Seq (Baird et al., 2008; Hohenlohe et al., 2010), we can quickly identify regions of the genome that are significantly more divergent than background Fst levels across the genome. We can then explore these outlier loci and the genomic regions linked to them (also known as ‘islands of divergence’ or ‘speciation genes’) to begin answering questions on the genomic basis speciation in the face of gene-flow.

In addition to looking at the tips of the gobiid phylogeny, future studies can also turn attention to the past to investigate the branches of the goby tree of life that are still difficult to resolve. One promising avenue of research is to develop and assemble large datasets comprised of new phylogenomic markers that possess the phylogenetically informative characters required to resolve problematic nodes. New methods have recently been developed to do this such as targeted gene-capture (also known as hybrid gene capture, or sequence capture) which couples next-generation sequencing technology with mRNA capture probes to capture, amplify, and sequence tens of thousands of loci (including protein coding regions) across genomes of widely- divergent, non-model organisms (Faircloth et al., 2012; Li et al., 2013). In addition to providing phylogenetically informative sites for resolving problematic nodes, sequencing tens of thousands

126

of protein-coding loci also allows us to identify sites, genes, or gene families that are evolving in patterns that differ significantly from trees generated by putatively neutral or non-coding regions, and thus may have undergone periods of natural selection. By combining these sequence-capture tools with genome-scans from RAD-Seq data, we can paint a genome-wide view of phenotypic evolution and speciation in gobies, one of the world’s most successful lineages of vertebrates.

References

Baird, N.A., Etter, P.D., Atwood, T.S., Currey, M.C., Shiver, A.L., Lewis, Z.A., Selker, E.U.,

Cresko, W.A., Johnson, E.A., 2008. Rapid SNP Discovery and Genetic Mapping Using

Sequenced RAD Markers. PLoS One 3.

Hohenlohe, P.A., Bassham, S., Etter, P.D., Stiffler, N., Johnson, E.A., Cresko, W.A., 2010.

Population Genomics of Parallel Adaptation in Threespine Stickleback using Sequenced

RAD Tags. Plos Genet 6.

Faircloth, B.C., McCormack, J.E., Crawford, N.G., Harvey, M.G., Brumfield, R.T., Glenn, T.C.,

2012. Ultraconserved Elements Anchor Thousands of Genetic Markers Spanning

Multiple Evolutionary Timescales. Systematic Biology 61, 717-726.

Li, C., Hofreieter, M., Straube, N., Corrigan, S., Naylor, G.J.P., 2013. Capturing protein-coding

genes across highly divergent species. BioTechniques 54, 321-326.

127

BIOGRAPHICAL SKETCH

Luke Michael Tornabene received his Bachelor of Science degree in Biology at Hofstra

University in Hempstead, New York. As an undergraduate he became interested in systematics of fishes working with Dr. James Van Tassell of the American Museum of Natural History. His projects included a review of Atlantic species the goby genus Bollmannia, the description of a new species of Microgobius from Pacific Panama, and helping Dr. Van Tassell and Dr. Ross

Robertson of the Smithsonian Tropical Research Institute with their Shorefishes of the Greater

Carribbean: Online Information System project.

Upon graduating, Luke entered the Master’s of Science program in Biology at Texas

A&M University – Corpus Christi under advisement of Dr. Frank Pezold. His thesis projects focused on the systematics of the gobiid genus Bathygobius in the Western Atlantic and Eastern

Pacific oceans. While obtaining his Master’s degree, Luke continued to publish several systematic papers with Dr. Van Tassell and colleagues at the Smithsonian National Museum of

Natural History.

After receiving his Master’s degree in 2010 Luke continued his work with Dr. Pezold in the Marine Biology PhD program at Texas A&M University – Corpus Christi. As a PhD student, Luke’s research interests expanded from taxonomy and systematics to broader evolutionary themes including macroevolution, evolutionary ecology, speciation, phylogeography and phylogenomics. His geographic area of interest also shifted from Caribbean and Eastern Pacific faunal provinces to the Indo-Pacific region.

128

Luke has a passion for field work and has conducted research throughout the continental

United States, Puerto Rico, Ecuador and the Galapagos Islands, Costa Rica, Panama, Barbados,

Indonesia and French Polynesia. To date Luke has published 10 peer reviewed journal articles including the dissertation chapters herein, and co-authored a book chapter. He has been an instructor for several courses including intro biology lab, professional skills, and ichthyology lab.

Luke is an avid saltwater angler and in his spare time he can be found fishing from his kayak or perched on the rocks at one of the local jetties.

129