The Assessment of Current Biogeographic Patterns of Coral Reef

THE ASSESSMENT OF CURRENT BIOGEOGRAPHIC PATTERNS OF CORAL REEF

FISHES IN THE RED SEA BY INCORPORATING THEIR EVOLUTIONARY AND

ECOLOGICAL BACKGROUND

Dissertation by

Vanessa S. N. Robitzch Sierra

In Partial Fulfillment of the Requirements

For the Degree of Doctor of Philosophy

King Abdullah University of Science and Technology, Thuwal,

Kingdom of Saudi Arabia

©March, 2017

Vanessa S. N. Robitzch Sierra

EXAMINATION COMMITTEE PAGE

The dissertation of Vanessa S. N. Robitzch Sierra is approved by the examination committee.

Committee Chairperson: Dr. Michael Berumen Committee Members: Dr. Christian Voolstra, Dr. Timothy Ravasi, Dr. Giacomo Bernardi

ABSTRACT

THE ASSESSMENT OF CURRENT BIOGEOGRAPHIC PATTERNS OF CORAL REEF

FISHES IN THE RED SEA BY INCORPORATING THEIR EVOLUTIONARY AND

ECOLOGICAL BACKGROUND

Vanessa S. N. Robitzch Sierra

The exceptional environment of the Red Sea has lead to high rates of endemism and biodiversity. Located at the periphery of the world’s coral reefs distribution, its relatively young reefs offer an ideal opportunity to study biogeography and underlying evolutionary and ecological triggers. Here, I provide baseline information on putative seasonal recruitment patterns of reef fishes along a cross shelf gradient at an inshore, mid-shelf, and shelf-edge reef in the central Saudi

Arabian Red Sea. I propose a basic comparative model to resolve biogeographic patterns in endemic and cosmopolitan reef fishes. Therefore, I chose the genetically, biologically, and ecologically similar coral-dwelling damselfishes Dascyllus aruanus and D. marginatus as a model species-group. As a first step, basic information on the distribution, population structure, and genetic diversity is evaluated within and outside the Red Sea along most of their global distribution. Second, pelagic larval durations (PLDs) within the Red Sea environmental gradient are explored. For the aforementioned, PLDs of the only other Red Sea Dascyllus, D. trimaculatus, are included for a more comprehensive comparison. Third, to further assess ongoing pathways of connectivity and geneflow related to larval behavior and dispersal in

Red Sea reef fishes, the genetic composition and kinship of a single recruitment

4 cohort of D. aruanus arriving together at one single reef is quantified using single nuclear polymorphisms (SNPs). Genetic diversity and relatedness of the recruits are compared to that of the standing population at the settlement reef, providing insight into putative dispersal strategies and behavior of coral reef fish larvae. As a fourth component to study traits shaping biogeography, the ecology and adaptive potential of the cosmopolitan D. aruanus is described by studying morphometric-geometrics of the body structure in relation to the stomach content and prey type from specimen along the cross-shelf of the central Red Sea and at a site outside the Red

Sea, in Madagascar, and approach whether foraging strategies change depending on geographic location and environment, and if differences in diet are followed by phenotypic plasticity. Jointly, results suggest that biological responses and putative adaptive strategies are correlated with different biogeographic ranges and habitat preferences.

ACKNOWLEDGMENTS

In dedication to the amazing uniqueness of the Red Sea. It is full of explainable exceptions to entertain you for millions of years.

For the completion of this work, I thank my parents for their unconditional support, trust, love, and pride. Together with Shobhit and the rest of my family I have the strongest backbone one could ever wish for, with me, at all times. I cannot thank those people enough. With them, I felt strong, even throughout probably the most difficult stage of my life. I further thank Scott for having been a motivating, loyal, and kind companion, who showed me what hard work, endurance, discipline, and commitment really look like; four qualities without which I would not have been able to complete such an extensive work in this short period of time. It is hard to put down in words how lucky I further am in having met some of the most loving people I can call friends. I thank them very much for always being just a Skype call or text message away, no matter the time zone, geographical location, or issue. These also include my Saudi skippers, friends, and charming and protective helpers from Thuwal: Abdullah, Mohammed, Assam, Whaled, Ghazi; and Amr, Ramzi, and Haitham. It was never too hot, too early, too tiring, or too impossible to help and please me with all of their motivating energy! I am also grateful to have found a field of work, in which I can rely on colleagues and their knowledge at all times and no matter if I a stranger, student, or casual acquaintance. Among many, these include in particular two previous postdocs at KAUST and good friends, Pablo and Joey; my advisor, Dr. Michael Berumen; my former advisor, Dr. Xabier Irigoyen; my committee members: Dr. Giacomo Bernardi, Dr. Chrisitan Voolstra, and Dr. Timothy Ravasi; and the generous and enthusiastic Dr. Bruno Frédérich, Dr. Ben Victor, and Dr. David Kaplan. Last but not least, I thank Dani for giving me back the ability to perceive other wonders in life, to widen the scope of curiosities to explore and learn from, and become a finer, stronger, and more complex being.

TABLE OF CONTENTS

TITLE PAGE ...... 1

EXAMINATION COMMITTEE PAGE ...... 2

ABSTRACT ...... 3

ACKNOWLEDGMENTS ...... 5

TABLE OF CONTENTS ...... 6

LIST OF ABBREVIATIONS ...... 14

LIST OF FIGURES ...... 17

LIST OF TABLES ...... 20

CHAPTER I

GENERAL INTRODUCTION TO THE STUDY OF BIOGEOGRAPHY...... 21

1.1 BIOGEOGRAPHY AND BIODIVERSITY IN MARINE ECOSYSTEMS ...... 24

1.1.1 Evolution and Historical Biogeography of Coral Reefs ...... 25

1.1.2 Evolution and Historical Biogeography of Coral Reef Fishes ...... 26

1.1.3 Coral Reef Biogeographic Regions and Their Origin ...... 28

1.2 THE RED SEA: INTEGRATING ECOLOGICAL AND HISTORICAL BIOGEOGRAPHY TO

UNDERSTAND CURRENT CORAL REEF FISH DISTRIBUTIONS ...... 31

1.2.1 Origin and Environment ...... 31

1.2.2 Ecological Biogeography and Connectivity ...... 32

1.2.2.1 Endemism and Biodiversity ...... 32

1.2.2.2 Population Genetics and Connectivity ...... 34

1.2.2.3 Larval Dispersal and Recruitment ...... 36

1.3 DASCYLLUS ARUANUS AND D. MARGINATUS AS MODEL ORGANISMS ...... 39

1.3.1 Dascyllus aruanus (Linnaeus, 1758) ...... 41

1.3.1.1 Biogeography and Evolution ...... 41

1.3.1.2 Ecology and Biology ...... 42

1.3.2 Dascyllus marginatus (Rüppell, 1829) ...... 44

1.3.2.1 Biogeography and Evolution ...... 44

1.3.2.2 Ecology and Biology ...... 44

1.4 REFERENCES ...... 46

CHAPTER II

ANNUAL DISTRIBUTION OF RECRUITMENT OF CORAL REEF FISHES ALONG A

CROSS-SHELF GRADIENT IN THE CENTRAL RED SEA ...... 56

2.1 INTRODUCTION ...... 56

2.2 MATERIALS AND METHODS ...... 65

2.2.1 Trials and First Collections ...... 65

2.2.2 Experimental Design ...... 66

2.2.3 Collection of Biological Data ...... 67

2.2.3.1 Sampling and Biomass Assessment ...... 67

2.2.3.2 DNA Extraction, Barcoding, Identification, and Quantification ...... 68

2.2.4 Collection of Environmental Data ...... 71

2.2.5 Statistical Analysis and Correlations ...... 72

2.2.6 Modeling of Biological Response Variables ...... 73

2.3 RESULTS ...... 74

2.3.1 Trials and First Collections ...... 74

2.3.2 Experimental Design ...... 76

2.3.3 Collection of Biological Data ...... 78

2.3.3.1 Sampling and Biomass Assessment ...... 78

2.3.3.2 DNA Extraction, Barcoding, Identification, and Quantification ...... 79

2.3.4 Collection of Environmental Data ...... 82

2.3.5 Seasonality and Recruitment Peaks ...... 85

2.3.6 Modeling of Biological Response Variables ...... 87

2.3.6.1 Total Biomass (TBM) ...... 87

2.3.6.2 Total Fish Biomass (FBM) ...... 88

2.3.6.3 Total Biomass and Total Fish Biomass (TBM and FBM) ...... 89

2.3.6.4 Specific Fish Abundance (SFA) ...... 91

2.3.7 Species Composition ...... 92

2.4 DISCUSSION AND CONCLUSIONS ...... 94

2.4.1 Patterns in Recruitment of Coral Reef Fishes in the Central Red Sea .. 94

2.4.2 Modeling of Environmental Variables ...... 101

2.4.3 Biogeography and Biodiversity ...... 109

2.4.4 The Red Sea ...... 110

2.5 REFERENCES ...... 115

CHAPTER III

GENETICS, BIOLOGY, AND ECOLOGY OF THE ENDEMIC DASCYLLUS

MARGINATUS AND THE COSMOPOLITAN D. ARUANUS DAMSELFISHES IN VS.

OUTSIDE THE RED SEA: DEFINING BIOGEOGRAPHIC RANGES ...... 124

3.1 INTRODUCTION ...... 124

3.2 MATERIALS AND METHODS ...... 131

3.2.1 Study Region and Sample Collection of Dascyllus marginatus and D.

aruanus for the Generation of Single Nucleotide Polymorphism (SNP)

Libraries ...... 131

3.2.2 Single Nucleotide Polymorphism (SNP) Library Preparation and

Sequencing ...... 132

3.2.3 De-novo Assembly ...... 134

3.2.3.1 Dascyllus marginatus ...... 134

3.2.3.2 Dascyllus aruanus ...... 136

3.2.4 Population Genetics and Statistical Analyses ...... 137

3.2.4.1 Principal Coordinate Analysis (PCoA) ...... 138

3.2.4.2 Clustering Analysis ...... 138

3.2.5 Phylogenetic Analysis ...... 139

3.2.6 Assessment of Pelagic Larval Duration (PLD) ...... 141

3.2.6.1 Sample Collection ...... 141

3.2.6.2 Otolith Preparation and Pelagic Larval Duration (PLD) Assessment

...... 145

3.2.6.3 Chlorophyll a (CHLA) and Sea Surface Temperature (SST)

Correlations ...... 145

3.2.6.4 Statistical Analyses ...... 146

3.3 RESULTS ...... 148

3.3.1 De-novo Assembly ...... 148

3.3.2 Population Genetics and Statistical Analyses ...... 150

3.3.2.1 Dascyllus aruanus ...... 150

3.3.2.2 Dascyllus aruanus inside the Red Sea ...... 155

3.3.2.3 Dascyllus marginatus ...... 158

3.3.2.4 Dascyllus marginatus inside the Red Sea ...... 164

3.3.2.5 Comparative Analysis of Dascyllus aruanus and D. marginatus ...... 167

3.3.3 Phylogenetic Analysis ...... 169

3.3.4 Pelagic Larval Duration (PLD) Measurements ...... 171

3.3.4.1 Within the Red Sea ...... 172

3.3.4.2 Inside vs. Outside the Red Sea ...... 173

3.3.4.3 Environmental Correlations With Pelagic Larval Duration (PLD) in the

Red Sea ...... 175

3.3.4.4 Correlations With Latitude ...... 176

3.3.4.5 Correlations With Sea Surface Temperature (SST) ...... 177

3.3.4.6 Correlations With Chlorophyll a (CHLA) ...... 177

3.4 DISCUSSION AND CONCLUSIONS ...... 179

3.4.1 Contrasting Genetic Structures of the Widespread Dascyllus aruanus

vs. the Endemic D. marginatus ...... 180

3.4.2 Mito-Nuclear Discrepancies in the Genetic Structure of Dascyllus

marginatus of the Arabian Sea ...... 186

3.4.3 Pelagic Larval Durations (PLDs) of Dascyllus in the Red Sea ...... 190

3.5 APPENDIX ...... 194

3.5.1 Sample Collection of Dascyllus marginatus and D. aruanus Within the

Red Sea For a First Assessment of Genetic Diversity Using Nuclear

Microsatellite Markers (MSATs) ...... 194

3.5.2 Amplification and Fragment Analysis of Nuclear Microsatellite

Markers (MSATs) ...... 195

3.6 REFERENCES ...... 196

CHAPTER IV

SIBSHIP ANALYSIS AND GENETIC DIVERSITY OF ONE SINGLE RECRUITMENT

EVENT OF THE CORAL-DWELLING FISH DASCYLLUS ARUANUS ...... 205

4.1 INTRODUCTION ...... 205

4.2 MATERIALS AND METHODS ...... 214

4.2.1 Study Site and Sample Collection of Dascyllus aruanus ...... 214

4.2.2 Single Nucleotide Polymorphisms (SNPs) Library Preparation and

Sequencing ...... 215

4.2.3 De-novo Assembly ...... 216

4.2.4 Sibship Assignment, Population Structure, and Statistical Analyses 218

4.2.5 Otolith Preparation and Pelagic Larval Duration (PLD) Assessment

...... 220

4.2.6 Statistical Significance of Kinship Results and Further Data Mining 221

4.2.7 Theoretical Values on the Biology and Population Dynamics of

Dascyllus aruanus in the Red Sea ...... 224

4.3 RESULTS ...... 225

4.3.1 Raw Sequence Filtering and Assembly ...... 225

4.3.2 Population Genetic Statistics and Bayesian Clustering Analysis ...... 226

4.3.3 Sibship Assignment ...... 228

4.3.4 Pelagic Larval Duration (PLD) Measurements ...... 230

4.3.5 Statistical Significance of Kinship Results and Further Data Mining 231

4.4 DISCUSSION AND CONCLUSIONS ...... 234

4.4.1 Why Are Closely Related Larvae Travelling Together? ...... 236

4.4.2 What are the Consequences of Schooling With Kin During Pelagic

Larval Duration (PLD) for Population Dynamics? ...... 241

4.4.3 Further Data Mining ...... 244

4.5 REFERENCES ...... 246

CHAPTER V

INFLUENCES OF HABITAT LOCATION ON FORAGING AND MORPHOLOGICAL

TRAITS OF DASCYLLUS ARUANUS ...... 257

5.1 INTRODUCTION ...... 257

5.2 MATERIALS AND METHODS ...... 260

5.2.1 Study Site and Sample Collection of Dascyllus aruanus ...... 260

5.2.2 Geometric-Morphometric Analyses ...... 261

5.2.3 Assessment of Trophic Niche ...... 262

5.3 RESULTS ...... 263

5.3.1 Geometric-Morphometric Analyses ...... 263

5.3.2 Assessment of Trophic Niche ...... 268

5.4 DISCUSSION AND CONCLUSIONS ...... 268

5.5 REFERENCES ...... 272

CHAPTER VI

FINAL DISCUSSION ...... 274

6.1 BIOGEOGRAPHY ...... 274

6.2 BIODIVERSITY ...... 280

6.3 CONCLUSIONS ...... 280

6.4 REFERENCES ...... 282

APPENDICES ...... 284

7.1 LIST OF APPENDICES ...... 284

7.2 LIST OF PUBLICATIONS ...... 285

7.3 PRODUCTIVITY AND SEA SURFACE TEMPERATURE ARE CORRELATED WITH THE PELAGIC LARVAL

DURATION OF DAMSELFISHES IN THE RED SEA (2015)

7.4 HOMOGENEITY OF CORAL REEF COMMUNITIES ACROSS 8 DEGREES OF LATITUDE IN THE SAUDI

ARABIAN RED SEA (2015)

7.5 A REVIEW OF CONTEMPORARY PATTERNS OF ENDEMISM FOR SHALLOW WATER REEF FAUNA IN

THE RED SEA (2016)

7.6 CORAL BLEACHING IN SAUDI ARABIA AFFECTING BOTH THE RED SEA AND ARABIAN GULF

(2016)

LIST OF ABBREVIATIONS

Abbreviations Name/Description ADCP Aqua Doppler Current Profiler AFA Al Fahal, Mid Shelf Reef, Saudi Arabia AKA Shib Al Karrah AO Atlantic Ocean ASP Adults Standing Population ASU Abu Shosha, Inshore Reef, Saudi Arabia BL Blennies BP Base -Pairs CD Current Direction CF Cardinal Fishes CHLA Chlorophyll a Coastal Prox. Proximity to the Coast COC Island of Cocos Keeling, Australia COI Cytochrome Oxidase I CP Central Pacific CRT Coral Reef Triangle CS Current Speed CV Canonical Variable CVA Canonical Analysis of Variation Da Dascyllus aruanus ddRAD Double Digest Restriction -site Associated DNA Sequencing Depth Sampling Depths DF Damselfishes DJI Djibouti Dm Dascyllus marginatus Dt Dascyllus trimaculatus E The Average Number of Eggs per Colony EP East Pacific FBM Fish Biomass FDR False Discovery-Rate FISH Fish Abundance FL Full Likelihood FPLS Combined Full Likelihood and Pairwise-Likelihood Score Method FSB Farasan Banks GAQ Gulf of Aqaba GB Gobies GBR Great Barrier Reef H The Hatch Rate per Clutch

Hd The Percentage that Hatched Eggs per Day HWE Hardy-Weinberg-Equilibrium IA Indo-Australian Archipelago IBB Isolation by Burn -out IBD Isolation by Distance IBE Isolation by Environment IBO Isolation by Oceanography IND Indonesia IO Indian Ocean L The Trajectory of Potential Habitat for D. aruanus LAT Latitude LD Linkage Disequilibrium LD Lunar Day LONG Longitude MAD Madagascar MAG Magateen, Yemen MCE Minimal Cross -Entropy MLG Multilocus Genotypes MSAT Microsatellite Markers MUS Musandam , Oman MYA Million Years Ago NCRS North-Central Red Sea OTU Operational Taxonomic Unit P Pressure PAI Parcel Islands, China PARROT Parrot Fishes PCoA Principal Coordinate Analysis PF Parrot Fishes PI Probability Identity PL Pairwise-Likelihood PLD Pelagic Larval Duration PW Pairwise R1 Inshore Reef R2 Mid-Shelf Reef R3 Shelf-Edge Reef Rcd Number of Potential Daily Fish Recruits Produced by an Average Colony of D. aruanus During the Reproductive Season in the Red Sea RCP Recruiting Cohort Population Rp The rate at which in theory the SP assignment increases every time you half the radius providing the TRCP (Rp = SPr50/ SPr25 = 3.49/1.47= 2.37) RS Red Sea

RV Research Vessel Thuwal S The Survival Rate of the PLD SCRS South-Central Red Sea SFA Specific Fish Abundance SL Size Length SNA Shib Nazar, Shelf -Edge Reef, Saudi Arabia SNP Single Nucleotide Polymorphism SOC Socotra Island, Yemen SPr25 Estimated Number of Sibling Pairs from permutations ran TRCPrs (average = 3.49) SPr50 Estimated Number of Sibling Pairs from permutations ran TRCPrs (average = 1.47) SPRCP The number of Sibling Pairs found by COLONY within the RCP (=48 SP (SPRCP=48) SRS Southern Red Sea SST Sea Surface Temperature T Temperature TAI Taiwan TBM Total Biomass TFA Total Fish Abundance TL Lagoon of Tuléar, Madagascar TNR Total Number of Recruits TRCP Total Recruiting Cohort Population TRCPCmin TRCP that Results for the Minimum Number of Colonies Recruiting to Al Karrah According to the RCP Sibship Analysis. (rCmin = 1.066 km) TRCPr25 TRCP of Al Karrah Reef Under the Assumption that Larvae from Reefs Around a 25km Radius Recruit to it. TRCPr50 TRCP of Al Karrah Reef Under the Assumption that Larvae from reefs Around a 50km Radius Recruit to it TRCPRper Estimation of TRCP of Al Karrah Reef by Permutations TRCPrRCP TRCP from the Maximum Possible Radius (rRCP = 4.31 km) Inferred by the results from the RCP VIZ Visibility W The Average “width” of D. aruanus Habitat WP West Pacific WR Wrasses WRASSES Wrasses WTH Weather

LIST OF FIGURES

Figure Description 2.1 Sampling sites of recruiting fish larvae in the central Red Sea 2.2 Histograms of the total biomass and fish biomass at each reef location of this study 2.3 Rosette plots representing the main current directions and dominance of the current at three reefs off Thuwal, Saudi Arabia in the central Red Sea 2.4 Species-specific fish abundance (SFA) of the six most collected families within recruiting fish larvae and total number of recruits (TNR) sampled at each reef 2.5 Yearlong profiles of abundance of recruiting coral reef fishes 3.1 Sampling sites inside and outside the Red Sea (RS) for the generation of genetic data on single nucleotide polymorphisms (SNPs) of Dascyllus aruanus and D. marginatus 3.2 Sampling sites of Dascyllus marginatus, D. aruanus, and D. trimaculatus for the assessment of pelagic larval duration (PLD) inside the environmental gradient of the Red Sea (RS) as well as outside the RS 3.3 Distribution of fish sizes of Dascyllus marginatus and D. aruanus used for PLD measurements from each site of Figure 3.2 3.4 Pairwise (PW) F’ST comparisons over all loci as well as PW FSTs comparisons per locus between each of the sampling sites of Dascyllus aruanus 3.5 Genetic clustering analysis in the R package LEA for SNPs data sets of Dascyllus aruanus 3.6 Principal coordinate analysis (PCoA) for the pairwise (PW) F’ST comparisons among all samples of each sampling site of Dascyllus aruanus inside and outside the Red Sea (RS) 3.7 Principal coordinate analysis (PCoA) for the pairwise (PW) F’ST comparisons among all samples of each sampling site of Dascyllus aruanus only inside the Red Sea (RS) 3.8 Pairwise (PW) F’ST comparisons over all loci as well as PW FST comparisons per locus (colored histograms) between each of the sampling sites of Dascyllus marginatus 3.9 Genetic clustering analysis in the R package LEA for the SNPs data set of Dascyllus marginatus for all sampling sites 3.10 Principal coordinate analysis (PCoA) for the genetic distance between each sample of Dascyllus marginatus inside and outside the Red Sea (RS)

3.11 Principal coordinate analysis (PCoA) for the genetic distance between each sample of Dascyllus marginatus inside the Red Sea (RS) 3.12 Principal coordinate analysis (PCoA) for the pairwise (PW) F’ST comparisons among all samples of each sampling site of Dascyllus marginatus from inside and outisde the Red Sea (RS) 3.13 Genetic clustering analysis in the R package LEA 3.14 Phylogenetic tree based on sequences of the mitochondrial COI marker for a subset of samples from the entire sampling range of Dascyllus marginatus 3.15 Mean pelagic larval duration (PLD, in days) for Dascyllus marginatus, D. aruanus, and D. trimaculatus averaged among all study sites in the Red Sea 3.16 Mean pelagic larval durations (PLD, in days) of three Dascyllus species in the four regions of the Red Sea (RS) 3.17 Mean pelagic larval duration (PLD, in days) of Dascyllus aruanus and D. trimaculatus from locations inside the Red Sea (RS) and outside the RS (Indo-Australian Archipelago (IA) and Indian Ocean (IO)) 3.18 Validated monthly regional averages of chlorophyll a concentrations (CHLA, in mg m− 3) and sea surface temperature (SST, in °C) 3.19 Plots of Pearson correlations of pelagic larval durations (PLD, in days, y-axis) of three Dascyllus species (Dascyllus marginatus, D. aruanus, and D. trimaculatus) App. 3.1 Sampling sites inside the central and southern Red Sea (RS) for Dascyllus aruanus (represented by triangles) and D. marginatus and from Socotra and Djibouti for D. marginatus (represented by circles). 4.1 Sampling site of Dascyllus aruanus recruits in the Saudi Arabian north central Red Sea 4.2 Frequency of the pelagic larval durations (PLDs) within the recruiting cohort population (RCP; a)) and the adult standing population (ASP; b)) of Dascyllus aruanus at Al Karrah Reef (north central Red Sea, Saudi Arabia) 5.1 Sampling sites of Dascyllus aruanus form the central Red Sea (marked by a star) and Madagascar (marked by a triangle); and location of the sites in the central Red Sea along a cross-shelf gradient 5.2 Display of the location of 21 points for the homologous analysis of body shape of Dascyllus aruanus implemented in linear models 5.3 Canonical Analysis of the variation (CVA) in body shape of Dascyllus aruanus from the four sites, 3 inside the central Red Sea (AFA (red), ASU (green), and SNA (blue)) and one off the south-western coast of Madagascar (TL (purple))

5.4 Canonical Analysis of the variation (CVA) in body shape of Dascyllus aruanus from the three sites inside the central Red Sea (AFA (red), ASU (green), and SNA (blue)) 5.5 Illustration of the measurement of the diameter of eye-size and the attachment angle of the pectoral fin 5.6 Boxplot of the measurements of the attachment angle of the pectoral fins of Dascyllus aruanus from sites inside the Red Sea (AFA, ASU, and SNA populations) vs .Madagascar (TL population; x axis) 5.7 Histogram of the relative abundances of zooplankton (blue) and benthic (red) organisms present in the stomach contents of Dascyllus aruanus from sites inside the central Red Sea (AFA, ASU, SNA) and off Madagascar (TL)

LIST OF TABLES

Table Description 2.1 List of Parameters Measured and Implemented in Linear and Generalized Linear Models of the Data 2.2 Taxonomic Diversity of Recruiting Coral Reef Fish Larvae at Each of the Sampled Reefs 2.3 Visualization of the Explanatory Variables in the Best Fitting Linear and Generalized Linear Models of Larval Recruitment in the Red Sea 3.1 Sampling sites for the assessment of the pelagic larval duration (PLD) of Dascyllus trimaculatus, D. aruanus, and D. marginatus, including number of samples, reef names and coordinates 3.2 Sampling sites for the generation of genetic libraries of single nucleotide polymorphisms (SNPs) for the species Dascyllus marginatus (Dm) and D. aruanus (Da), including number of samples, site names, acronyms (Abb.), and coordinates 3.3 Pairwise (PW) F’ST comparisons over all loci of Dascyllus aruanus from sampling site inside the Red Sea (RS) only 3.4 Pairwise (PW) F’STs comparisons over all loci of Dascyllus marginatus from sampling site inside the Red Sea (RS) only 3.5 Average pelagic larval durations (PLD in days; mean ± standard deviation) of Dascyllus marginatus, Dascyllus aruanus, and Dascyllus trimaculatus based on replicate counts of daily growth increments from otoliths 4.1 Summary of Principal Population Genetic Statistics 4.2 List of Important Abbreviations and their Respective Value or Result 5.1 Number of Samples of Dascyllus aruanus per Site and Site-Specific Information

CHAPTER I

GENERAL INTRODUCTION TO THE STUDY OF BIOGEOGRAPHY

For many centuries assessing and understanding the distribution patterns of species and biodiversity has been a big challenge for ecologists, evolutionary biologists, and in a wider sense, biogeographers. Biogeography is defined as the study of the spatial distribution of species and attempts to understand the patterns of biological diversity (Lomolino et al., 2006). There are two main approaches in understanding the distribution of today’s biodiversity: one is looking at the current species habitat, ecology, and biology, and describing the boundaries that limit its geographic distribution. This can be referred to as ecological biogeography. The second approach is to look at the evolutionary history of the species or taxon. This can lead to its source of origin and ancestry, which can be either geographically or environmentally related to the present location of the species and denotes the study of historical biogeography. However, ideally, the study of biogeography should include expertise and data from a variety of sources, sciences, and time-scales, as it is a multidisciplinary field of research, which integrates concepts of ecology, evolutionary biology, geology, and physical geography.

One the one hand, historical biogeography makes use of fossil and geological data, which provide insight into the changes of continental position (tectonics), ancient climate-regimes, and evolutionary history of the Earth’s flora and fauna over hundreds of millions of years (e.g., speciation and extinction events over long

22 temporal scales). On the downside, the large range of times scales and scarce historical data allow for a wide range of errors, which are hard to correct for. For example, the location where a fossil is found is a reference point of the historical biogeography and origin of that specific taxonomic unit, and indicates where and when it existed. But fossils are rare, a single data point, and hard to replicate. Thus, the probability of one data point being within the standard historical distribution of the taxon (both geographically and temporally) is completely unpredictable.

Consequently, the age of a fossil can characterize the origin or extinction or any time period in between a taxon’s evolutionary history and for many present taxa not even a single fossil record has been found. On the plus side, the beauty and advantage of biogeographical studies today is the possibility to integrate new molecular and computational tools to improve the performance of inference with little data, and the statistical power and accuracy of the research. In this manner, and with the increased availability of different sources of data, genetic markers are currently used to verify estimations from former studies based on geological and paleontological data by dating origins of species and connections between taxonomic units. Furthermore, both data sources can be jointly calibrated using computational power to reduce the error of time-scales. Fields of application of genetic markers in historical biogeographic studies are evolutionary biology, phylogeography, and phylogenetics. In ecological biogeography these include molecular ecology, population genetics, connectivity, and adaptive physiology.

On the other hand, ecological biogeography has a very high current value for applied conservation and management purposes of species among economically relevant and highly vulnerable or rare species. For such purpose, to record presence or absence of species or taxa and to catalog them according to the geographic location helps us quantify biodiversity and identify biodiversity hotspots as well as, geographic areas with higher need for protection (also referred to as ecoregions;

Olson and Dinerstein, 2002; Archambault et al., 2010; Miloslavich et al., 2011;

Costello et al., 2010; Miloslavich et al., 2010; Ellis et al., 2011). Furthermore, the effective health of a stock, population, and/or species can be inferred from genetic data and linked to the biogeographic range and habitat availability.

Just as biogeographic studies are a multidisciplinary field, which integrates many different types of data sources and research-fields, there is still a strong segregation between ecological and historical/evolutionary biogeographic studies. The lack of overlapping data and tools to address different temporal scales within the two fields, lead to the study of either strictly evolutionary or ecological processes.

However, looking at current species biodiversity and their biogeography opens many questions regarding the delimiting factors of a species distribution, which can have both ecological and historical backgrounds. While some species are already reproductively, ecologically, and genetically well defined, there are many others under ongoing speciation and hybridization or extinction with unclear or unstable boundaries for geneflow and it is still unclear whether factors/forces that are currently limiting species from dispersing and defining their habitat have any historical significance to potentially explain evolution and origin of speciation, and

24 biogeography. Yet, historical information on species biogeographical origins, adaptation, speciation, and ecology is less accurate than parameters we can measure today. Hence, the integration of ecological data on progenitor species is crucial, widens the scope of information we have to interpret evolutionary processes, and has increasingly been addressed (Rocha et al., 2007; Riddle et al., 2008; Barber,

2009; 2011; Bowen et al., 2014). It additionally provides a holistic approach to infer the fate of the taxa under future ecological and climatic circumstances.

1.1 BIOGEOGRAPHY AND BIODIVERSITY IN MARINE ECOSYSTEMS

The more recent and currently vibrant environments for studying biodiversity are the marine habitats and particularly, coral reefs. Due to the vast ranges of, and limited access to coral reefs, the time we have to survey these underwater and remote realms is limited. Hence, we are still at the tip of the iceberg in cataloging biogeographic ranges and biodiversity for most of the coral reefs worldwide.

Beyond that it is necessary to jointly look at past and present trends in coral reef evolution, growth, and distributions of biodiversity, to allow for a comprehensive inference of future trends in the development and adaptation of coral reef ecosystems. This will improve our understanding when investigating major drivers of speciation and/or extinction in changing environments and help us integrate all our knowledge in conservation and management of biodiversity.

1.1.1 Evolution and Historical Biogeography of Coral Reefs

Coral reefs, built by Scleractinian corals, as we know them today, have an evolutionary history that goes back to over 250 MYA. However, mass extinctions, tectonic movements, temperature changes, and sea level fluctuations have affected and shaped this ecosystem drastically ever since its early origin. All of the coral genera from the Triassic are extinct and only very little records of Jurassic reef remains are found today. The late Jurassic was an important period for the rise of marine diversity after the “Great Dying” (~252 MYA). However, it was during the

Cretaceous (145-66 MYA) that the marine fauna experienced acute and fluctuating intervals of environmental pressure. Even though this period was characterized by major changes in the positions of continental landmasses, the climatic changes with extreme fluctuations in sea levels, adding or vanishing barriers for dispersal were the main drivers of the biogeographic distributions of species and sources for centers of biodiversity within coral reef ecosystems. Additionally, atmospheric and oceanic temperatures were 10-15°C higher than they are today, which widened the range for coral reefs to thrive, extending from approximately 45° N to 77° S

(reviewed in Veron, 1995, 2000).

The origin of modern coral reefs refers to a more recent evolutionary history, which initiated with the survivors of the extinctions in the end-Cretaceous and throughout the Palaeogene (67-24 MYA). Subsequently, during the Miocene (24 – 5.2 MYA), the subdivision into today’s biogeographic provinces with diverse cosmopolitan and local fauna started to diverge, which was further shaped by extinctions and

26 speciation events and highly influenced by two major glacial maxima during the

Plio-Pleistocene (reviewed in Veron, 1995, 2000). Nevertheless, while our modern reefs dominated by Scleractinian corals, have a much younger history of about 66

MY, the diversification of two of the major branches of current coral species

Acroporidae and Pocilloporidae can be dated to ages as far as the Triassic over 200

MYA, indicating a vibrant ancient history even among contemporary environments and founders of biodiversity.

1.1.2 Evolution and Historical Biogeography of Coral Reef Fishes

Even though the first fish body plan comes from a 500 MY old fossil, the origins of

Teleostei, which now accounts for 96% of all fishes, lies about 200 MYA. The largest order of fishes herein is Perciformes (the “perch-like” fishes) and the first geological record of Perciforma ever found is from the late Cretaceous (97-90 MY; Sorbini,

1979), which is in agreement with age estimates from otolithic records (Benton &

Association, 1993) . The oldest fully fossil specimen, Nardoichthyus, is a bit younger

(approx. 74 MY old, Sorbini and Bannikov, 1991). However, from bioinformatics reconstructions of phylogenetic trees the age of stem lineages of Perciforma lies somewhere between 205 (Yamanoue & Miya, 2008) and 180 MYA (Azuma &

Kumazawa, 2008) based on mitochondrial DNA (mitogenomes) and between 93-

115 MY based on a single nuclear gene (RAGI; Santini and Harmon, 2009). Trying to find the crown origins for reef fishes becomes even harder. In this case, not only the genetic and/or morphological link between the fossil and present reef fishes has to

27 be established, but it also has to be geographically and ecologically linked to ancient coral reefs. The first step towards such a discovery was undertaken after the largest and most diverse fossil assemblage for marine Perciformes was found in the Eocene deposits of Monte Bolca (Italy; approx. 50-46 MYA, Bellwood, 1996). High resemblances between many of these morphological phylogenies were compared to modern extant coral reef assemblages and provided a basis of putative origins of fish families in modern coral reefs (Bellwood, 1996). Furthermore, these assemblages indicate a high diversification of morphologies during this era and post the K/T boundary mass extinction (Friedman, 2010) and noticeable similarities to current morphologies in reef fishes (Goatley et al., 2010). All in all, from what we know today, coral reef fishes might have originated sometime in the Paleocene (66-

56 MYA), which coincides with the first continuous evolutionary development of contemporary Scleractinian coral reefs.

More than 5000 species of reef fishes depend on coral reefs globally (Goatley et al.,

2010), which equals to approximately one third of all biodiversity of marine fishes living in only about 0.01% of the world’s ocean. Biodiversity of coral reef fishes follow a latitudinal gradient with the diversity hotspots located towards lower latitudes (Ekman, 1935; Rosen, 1981). This same pattern is reflected also in terrestrial systems. However this is not necessarily related to the location of origin of species diversity. Hypotheses like the three “center(s) of” (origin (Ekman, 1953), accumulation (Ladd, 1960), and overlap (Woodland, 1983) in the Pacific biogeography; reviewed by Bellwood, 2012) and a fourth on the “center of survival”

(Hughes et al., 2002; Connolly et al., 2003), try to tackle the historical relationships of biogeographic patterns today of restricted, overlapping, and widespread species ranges among many coral reef taxa. Most recently, Cowman and Bellwood, (2013) have used three of the most species-diverse coral reef fish families to infer the historical biogeography, origin, and dispersal of the world’s coral reef fishes and explain the current global distributions of these taxa.

1.1.3 Coral Reef Biogeographic Regions and Their Origin

Most studies on marine biogeography speak of five distinct marine regions: the East

Pacific (EP), the Atlantic Ocean (AO), the Indian Ocean (IO), the central Indo-

Australian Archipelago (IA) and the Central Pacific (CP) (Cowman & Bellwood,

2013). These regions were all well connected with little to no barriers for dispersal during the Eocene, during which high gene flow with quasi-panmictic populations was very likely (Cowman & Bellwood, 2013).

The EP and AO kept a closely linked historical biogeography and cladogenesis, since the closure of the Isthmus of Panama only occurred ca. 3.1 MYA. Many families found in both regions are also families present in the Indo-Pacific, but not as much vice-versa. Reasons for this phylogenetic isolation are due to a more restricted and ancient connection to the Indo-Pacific, which ended by the late Miocene after the pantropical connection via the Tethys Sea, was lost. Additionally, the low biodiversity in the EP and AO is probably due to reduced speciation capacity and/or increased rate of extinction of lineages, as well as low potential of lineage

29 replenishment from the distant Indo-Pacific (Cowman & Bellwood, 2013). This makes the faunas of both regions fairly independent and genetically quite distinct from the ones in the Indo-Pacific regions (IO, IA, and CP) (Rocha et al., 2008; Floeter et al., 2007; Joyeux and Floeter, 2001).

The origin of many extant reef fishes is from the Eocene and geographically located in what was known as the Tethys Sea (Bellwood, 1996). However, the IA is commonly recognized as center of biodiversity for modern coral reef taxa.

Moreover, there is a well-established principle of a latitudinal gradient of species diversity in coral reefs (Ekman, 1935; Rosen, 1981) and a to some extend more controversial putative longitudinal gradient away from the IA hotspot (Palumbi et al., 1997; Briggs, 1999; Hoeksema, 2007). A similar pattern can be observed in terrestrial animals, for which the highest biodiversity is found around the tropical belt, while it decreases towards the poles and temperate areas. That makes the IA a potential “Bull’s Eye” for coral reef biodiversity (Cowman & Bellwood, 2013) and a center of accumulation, survival, origination (Roberts et al., 2002; Briggs, 2003;

Mora et al., 2003), and range expansion (Cowman & Bellwood, 2013) during a long historical period starting at least as early as the Oligocene over 33 MYA. This was validated by Cowman and Bellwood (2013) using molecular biogeographical reconstructions for the coral reef fish families: Labridae, Pomacentridae, and

Chaetodontidae, which concluded that the IA has served as a ground for species diversification over time (for at least 33 MY) thanks to stable and putatively omnipresent coral reef environments for the last ~65 MY and as a refuge for coral reef associated taxa despite of the location of origins (Hoeksema, 2007; Cowman &

Bellwood, 2013). To sum up, while ancient hotspots of marine life like the Tethys

Sea went through massive environmental changes and extinction events in the late

Eocene and during the Oligocene (less than 40 MYA), the IA environment changed to a lesser extent and taxa could either reappear secondarily from sister lineages present in the IA or could find shelter in coral reefs around the IA (Barber &

Bellwood, 2005; Cowman & Bellwood, 2011).

Other regions, like the Indian Ocean and Central Pacific rather exemplify

“macroevolutionary sinks”, as they have been recipients of species originally from the Tethys Sea (before the Miocene) and the IA (during and after). Therefore, origination in in the IO and CP only accounts for very little of the present coral reef biodiversity of which most originated in the IA (Cowman & Bellwood, 2013). The most likely reasons for this pattern in species origin are the loss of Arabian biodiversity hotspots (Renema et al., 2008), tectonic rearrangements, and global cooling (Hallam & Wignall, 1997; Allen & Armstrong, 2008) throughout the

Oligocene/Miocene, which lead to the loss of tropical zones and coral reef habitat in the IO and CP (Cowman & Bellwood, 2011). However, as research increases in the western Indo-Pacific, we are discovering more and more species and high rates of endemism especially in the Red Sea (DiBattista et al., 2015a, 2015b).

1.2 THE RED SEA: INTEGRATING ECOLOGICAL AND HISTORICAL

BIOGEOGRAPHY TO UNDERSTAND CURRENT CORAL REEF FISH DISTRIBUTIONS

1.2.1 Origin and Environment

The Red Sea is a fairly young environment, which started forming after the northeast African continental crust began to part and drift to the east around 31

MYA (Bosworth et al., 2005). The first opening of the Red Sea basin happened during the Miocene (around 24 MYA) in the north, connecting it with today’s

Mediterranean Sea. However, the connection and water flow was weak and inconstant, which lead to many complete droughts (Bosworth et al., 2005). It was not until approximately 5 MYA, that the Red Sea ultimately found a stable connection to oceanic waters through the Gulf of Aden in the south (Siddall et al.,

2003; Bosworth et al., 2005).

Currently, the Red Sea stretches 2270 km from 30 °N to 12°N of latitude. It is the northernmost tropical sea and characterized by being relatively deep (maximum depth of 2920 m), very narrow (max. width: 350 km, at around 16 °N), and at the periphery of coral reef ecosystems. Due to its geographic location it experiences high evaporation rates (approx. 2m yr-1) and has no freshwater inflow (Morcos,

1970; Sofianos et al., 2002). Thus, the only source of water is the Indian Ocean at the south, which all in all leads to its well-defined gradients of salinity (42 to 37 psu), sea-surface temperature (winter, 20–28 °C; summer, 26–32 °C), and nutrient

32 concentration (low to high) from north to south (Edwards, 1987; Sofianos & Johns,

2007; Raitsos et al., 2013; Kürten et al., 2014).

1.2.2 Ecological Biogeography and Connectivity

1.2.2.1 Endemism and Biodiversity

Biogeographic data indicate that endemism tends to be less prominent in marine ecosystems than terrestrial (Olson & Dinerstein, 2002). However, marine ecosystems near isolated islands or among patchy habitats like coral reefs and especially in enclosed seas like the Red Sea have a tendency for evident endemism

(Kelleher et al., 1995; Groombridge & Jenkins, 1996). The Red Sea for example, contains exceptional concentrations of biodiversity and endemics. Hence, it belongs to the “Global 200”, a list of 238 ecoregions worldwide, including 43 marine regions

(Olson & Dinerstein, 2002). More than 50 genera of Scleractinian corals and well over 1000 species of fishes (Lieske & Myers, 2004; Berumen et al., 2013) of which at least 14% are endemic, inhabit the Red Sea (Randall, 1994; Lieske & Myers, 2004;

DiBattista et al., 2013). Among other taxa, the ratio of endemism is even higher, with

15% in echinoderms, 33% in crustaceans, 25% of corals (Cox & Moore, 2010) and up to 50% among butterflyfishes (Roberts et al., 1992).

Areas of such high biodiversity and moreover, high endemism are not only significant for conservation but also extremely relevant as an analytic unit for the study of historical biogeography (Grehan, 1993). The Red Sea as an “area of

33 endemism” is however, interestingly very young, which makes it an even more exceptional target for biogeographic studies with an ecological emphasis.

Additionally, even though neighboring marine regions have previously been defined as “macroevolutionary sinks” (Cowman & Bellwood, 2013), the Red Sea is more likely to be a modern species incubator in case of some coral reef taxa (DiBattista et al., 2013).

Reasons for such high levels of endemism and biodiversity can be the following:

(1) Isolation: The Red Sea is relatively isolated from other coral reefs worldwide. It is the northernmost tropical sea and it is located at the periphery of the world’s coral reef distribution with only a very narrow (18 km) and relatively shallow (137 m) connection in the south (Strait of Bab al Mandab) to the Indian Ocean (Eshel et al., 1994; Bailey, 2009).

(2) Unique/exceptional environmental conditions: The Red Sea is a very unique environment. Herein, three major environmental gradients are present and driven by: geographic location, high evaporation rates (over 2m yr-1, Sofianos et al., 2002), rare fresh water input, latitudinal gradients in temperatures, and a seasonally fluctuating regime in water exchange with the Gulf of Aden in the South. Salinity and temperatures display antagonistic physicochemical gradients from north to south, with highest levels of salinity and lowest sea surface temperatures (SST) in the north, and lowest salinities and highest SST in the south. The third environmental gradient consists of nutrients and turbidity, which increase from north to south

(Raitsos et al., 2013; Kürten et al., 2014). Most of the Red Sea waters are very oligotrophic with high visibility. However, at the south, the influence of incoming

34 nutrient rich waters from the Gulf of Aden become noticeable with over a 30 fold increase in CHLA at the Farasan banks and islands (Kürten et al., 2014), constituting a completely different environment/habitat for coral reef biota. Additionally, the

Red Sea is one of the hottest and most saline marine systems harboring coral reefs.

(3) Recent geological history: Increased isolation and severity of environmental changes in recent history during the Pleistocene may also play a crucial role in Red

Sea endemism. Even though the Red Sea’s geological history is relatively young, it has been a place of many changes and climatic fluctuations in the recent geological past. Sea level fluctuations and the last glacial maxima have led to sea level drops

(SLDs) approx. 340,000 years ago (340 KYA), 270 KYA, 130 KYA and 22 KYA, with the latter two reaching approximately –120 m. These SLDs can be traced in core samples of the Red Sea basin (Siddall et al., 2003) and most likely lead to periods of increased isolation of the Red Sea faunas (reviewed in DiBattista et al., 2013, 2015).

Additionally, there are three records of aplanktonic intervals around 450 KYA, 130

KYA and 20 KYA (Siddall et al., 2003), probably related to salinities exceeding ~49 psu (Hemleben & Meischner, 1996; Rohling et al., 1998; Fenton & Geiselhart, 2000).

1.2.2.2 Population Genetics and Connectivity

The connectivity among a species’ population is defined by the transfer of individuals among demes (i.e. subpopulations) and so, involves both the transfer of genes among these as well as the replenishment of demes (Levin, 2006; Cowen et al.,

2007).

In the same manner, the levels of genetic differentiation and connectivity of a species depend on the amount of gene flow between populations. However, the required numbers of specimen exchanges (i.e. migration rate) between populations to maintain genetic connectivity within a species is lower then the one required to maintain a genetically homogeneous population. Therefore, the genetic structure of a species can be assessed at two major scales: an evolutionary scale (e.g. the origin of sister species from isolated populations as a consequence of genetic drift and

“hard” barriers to dispersal; or secondary contact and gene flow between sister species redefining the progenitor species) as opposed to a demographic or ecological scale (e.g. the genetic differentiation of population or subpopulations either by geographic distance, soft barriers to dispersal, and/or ecologically through selective mortality of specific phenotypes). For the former evolutionary scale, phylogenetic studies measure divergence and genetic exchange of much lower rates

(e.g. 1 or 2 events of genetic exchange per generation enough to stop cladogenesis) and thus, use genes evolving at the adequate pace like COI (mitochondrial DNA) for fishes. For the latter demographic/ecological scale, continuous gene flow and migration rates of several orders of magnitude are necessary to maintain genetic homogeneity and population growth (Lowe & Allendorf, 2010). Hence, measuring genetic differentiation and gene flow require faster evolving and more over, a wider set of genetic markers, typically microsatellites (non-coding tandem repeats in nuclear DNA) or single nucleotide polymorphisms (SNPs, in general of nuclear DNA origin). Even though both types of connectivity studies within a species work at different spatial and temporal time scales, both are necessary to fully assess

36 biogeographic patterns of and their implications for reef fishes ecology and evolution.

1.2.2.3 Larval Dispersal and Recruitment

The measurement of larval dispersal of marine fauna is difficult. It is even harder if it relies on the direct sampling and or tracking of the pelagic larval stage. The high numbers of tiny pelagic larva produced by most marine taxa requires high sampling effort and even still typically results in relatively low yields and coverage.

Additionally, identification of larvae to species or even family level without molecular tools is time consuming and for some taxa, even with expertise, rather impossible. Thus, the pelagic larval phase is commonly known as the “black box” of the life-cylce of fishes and an extensive insight is even more limited for the extremely diverse biota of coral reef fishes (Boehlert, 1996). However, knowing the potential dispersal of larvae is crucial for the management and maintenance of genetically diverse and well-connected marine species’ populations.

From oceanographic modeling and inference of genetic connectivity, previous researchers have concluded that several factors seem to play a crucial role in determining larval dispersal patterns. For example, currents can produce physical barriers (Baums et al., 2005, 2006; Thornhill et al., 2008) as well as powerful transport routes for pelagic larvae (Lessios et al., 1999; Muss et al., 2001; Pineda et al., 2007; Treml et al., 2008). Former calculations of dispersal of fish larva and eggs using physical oceanographic parameters regarded larvae as particle-like units

37 carried and dispersed by oceanographic currents, unable to regulate their fate and final destination (Scheltema, 1971a, 1971b, 1971c, Strathmann, 1974, 1980). Thus, connectivity and dispersal was based on number of larvae, mortality rate, and intensity of oceanographic currents or barriers to dispersal. With new molecular tools, genetic studies on populations revealed that much of the inferred dispersal and connectivity did not coincide with the actual genetic information/connectivity of the population (Thresher et al., 1989). For instance, Madduppa et al. (2014) found self-recruitment among coral reef fishes of up to 65.2% - much higher than expected. Consequentially, many studies attempted to remodel potential larval dispersal by integrating measurements of genetic connectivity among geographic regions in addition to oceanographic data and biological data like pelagic larval durations and reproductive strategies (brooder vs. broadcast spawner) (Thresher et al., 1989). However, results of correlation or discrepancies between the models, differed from species to species and impede a generalization of dispersal inference for marine organisms (e.g. Shanks, 2009; Hogan et al., 2012; Miller and Ayre, 2008;

Sponaugle et al., 2002; Cowen and Sponaugle, 2009; Cowen et al., 2006). Some recent studies have found that chemical cues (Gerlach et al., 2007; Lecchini &

Nakamura, 2013), a much stronger and efficient vertical and horizontal swimming behavior of “planktonic” larvae even against currents (Leis & Mccormick, 2002), orientated navigation (following, e.g., sunlight) (Mouritsen et al., 2013), and possibly intended travelling and/or settling with its kin (e.g. Shapiro, 1983; Buston et al.,

2009; Madduppa et al., 2014; but see Avise and Shapiro, 1986) are some of the

38 reasons for the complexity and unpredictability of larval dispersal and thus, the connectivity between populations of coral reef fishes.

Nevertheless, regardless of the potential dispersal of a larva, the genetic structure, richness, and connectivity of a species ultimately relies on successful settlement and reproduction of the recruits (also referred to as realized connectivity, Watson et al.,

2010). Furthermore, the quantification and assessment of recruitment patterns for marine organisms is also much more feasible and an efficient method to track connectivity as a result of larval dispersal: a) the number of individuals recruiting to a reef is already drastically reduced (highest mortalities during pelagic phase) and thus, calculating dispersal and potential connectivity from recruitment data is much more accurate. b) Coral reefs are relatively close to the shore and are smaller in dimension compared to the open ocean. Therefore, to survey dispersal and connectivity directly at the reefs is more efficient and target-oriented. c) By the time a larva reaches the reef and is ready to settle it is bigger and morphologically better differentiated, which makes identification and qualitative recruitment analysis less of a struggle even without molecular tools.

So far, the only information on recruitment in coral reef of the Red Sea is from the far north near Eilat (Froukh et al., 2001; Ben-Tzvi et al., 2007), which denotes the lack of significant data from most of the Red Sea. Additionally, in the south, where we have the connection to other coral reef ecosystems, the water exchange regimes for the straight of Bab El Mandab aren’t constant/change with the season. While summer: surface outflow, intermediate inflow and deeper outflow (3 layer); in winter typical estuarine 2 layer regime: shallow surface inflow and deep outflow

(maybe increasing the isolation and decreasing connectivity for some species that will have to swim against the currents to get into the Red Sea when recruiting in the summer months.

1.3 DASCYLLUS ARUANUS AND D. MARGINATUS AS MODEL ORGANISMS

Dascyllus aruanus and D. marginatus belong to the family of damselfishes

(Pomacentridae). The first paleontologic record of damselfishes is about 50 MY old

(Blot, 1980; Bellwood & Sorbini, 1996; Bellwood & Wainwright, 2002). This family of coral reef fishes is well studied (Godwin, 1995), very abundant throughout the

Indo-Pacific, and it is one of the ten most species rich among coral reef fishes

(Bellwood & Wainwright, 2002). It also covers a variety of ecological, morphological, and biological traits of many reef fishes (Elliott et al., 1999; Cooper et al., 2009), which allows to use it as model organisms in a wider sense and as an indicator species for habitat characterizations and availability. More specifically, species from the Dascyllus genus are commonly known as planktivorous coral reefs dwellers, which are highly site attached and habitat specific (Godwin, 1995). They also usually prefer habitats with conspecifics, where they preferentially settle (Shpigel &

Fishelson, 1986; Sweatman & St John, 1990; Booth, 1991, 1992; Öhman et al., 1998).

The latter facilitates sampling, replication, and experimental design. The genus diversified approximately 16 MYA and there are ten recognized differentiated

Dascyllus species divided into three to four species complexes (McCafferty et al.,

2002; Leray et al., 2010). This genus evolved probably 20 MYA and the major

40 cladogenesis of todays’ species lineages occurred around 7 to 12 MYA, while the youngest diversification is about 2 MY old (within the D. trimaculatus species complex, McCafferty et al., 2002). Dascyllus nowadays solely inhabits the broader

Indo-Pacific from the Red Sea all the way to Hawaii. Another interesting aspect of studying this genus in terms of biogeography, is that two of its species: D. aruanus and D. trimaculatus are cosmopolitan to the vast geographic range and have different habitats and body sizes but an overlapping biogeography. The other eight have either a smaller distribution range, are endemic to regions at the periphery of the genus’ geographic distribution, and/or are only present in specific archipelagos

(Godwin, 1995). Nevertheless, they share the same habitat type and body size with one of the former two cosmopolitan species, and co-occur whenever their biogeographic ranges overlap (e.g. Shpigel and Wise, 1982; Limbourn et al., 2007;

Schmitt and Holbrook, 1999). In contrast, among the endemics no overlapping biogeography is found among those with similar habitat preference and habitat range. Thus, these endemics or more restricted/specialized species often replace each other in their specific habitat, but share it with the widely distributed species

(Shpigel & Fishelson, 1986; Ormond et al., 1996; Boehlert, 1999; Schmitt &

Holbrook, 1999, 2002; Ben-Tzvi et al., 2009; Pratchett et al., 2012). To which degree the cosmopolitan and endemics share the habitat/niche, and the alongside ecological, biological and biogeographical consequences of co-occurrence are still unstudied. However, the study of this genus may provide insight into current drivers of biogeography as well as potential ecological traits (rather than simply geological forces), which have shaped cladogenesis and biogeography of modern

41 coral-dwelling fishes in evolutionary time-scales (among past, present and future scenarios).

1.3.1 Dascyllus aruanus (Linnaeus, 1758)

1.3.1.1 Biogeography and Evolution

Dascyllus aruanus has a monophyletic origin, diversified around 9.3 MYA

(McCafferty et al., 2002), and can nowadays be found throughout the Indo-Pacific

(up to Hawaii) and the Red Sea. However, just recently evidence has been published for a possible subdivision into two species with the boundary located in the Indian

Ocean (Liu et al., 2014; Borsa et al., 2015). The Indian Ocean lineage ranges from the

Red Sea to the eastern extremity of the Sunda Shelf and is described to be morphologically and also genetically (nuclear markers) different from the Pacific lineage but the two only diverge 0.6% at the cytochrome b locus (Liu et al., 2014;

Raynal et al., 2014). The origin of these putatively geminate species can be the result of vicariance due to reproductive barriers, which limited geneflow and not just due to isolation by distance (Raynal et al., 2014). The epiphets aruanus (Linnaeus, 1758) and abudafur (Forsskål) are suggested for the Pacific and Indo-Red Sea sister- species respectively (Borsa et al. 2014). However, and after some discussions among colleagues I decide to keep the two clades together and refer to them as D. aruanus, since D. abudafur is yet to be officially re-described.

There are no fossil records for D. aruanus, which makes the geographical and temporal assignment of origin and validation of diversification difficult. However, from a molecular study on historical biogeography of reef fishes, most of the contemporary Pomacentrid species seem to have diversified in the central Indo-

Australian Archipelago, presumably from sister species originally from the Tethys

Sea (Cowman & Bellwood, 2013). In this region, from the Indonesian Archipelago and south the Great Barrier Reef, it overlaps with its only sister species D. melanurus, where both are commonly found in shallow lagoons (Godwin, 1995;

Bernardi & Crane, 1999; McCafferty et al., 2002). However, and especially through the exposure of such habitat to major changes during glaciation, the species habitat and original distribution ranges must also have changed drastically (Planes et al.,

1993). This along side the wide distribution range of this species complex let speculate that it has been the origin of other species with more restricted biogeography (Bernardi & Crane, 1999).

1.3.1.2 Ecology and Biology

Dascyllus aruanus is a zooplanktivorous coral reef fish that forms colonies and resides in branching corals for its entire life-span (Fricke & Holzberg, 1974; Coates,

1982). The individuals use the entire coral colony equally as their home (Sale,

1970), though there is a strict and size-dominated hierarchy, in which all individuals within one colony differ in their total body length (personal measurements;

Gillespie, 2009). There are several PLD measurements available (except from the

Red Sea) with a mean PLD of 21.1 d (Luiz et al., 2013). Its growth rate is of ca. 5.08 mm per month during the first year, reaching putatively maturity at a size 38 mm and a size of ca. 61mm at the end of its first year. Breeding peaks have been reported from April to January for Minicoy atoll (Pillai et al., 1985a) and spawning peaks in Japan, following a semilunar spawning cycle, are from June to September

(2-4 days directly before or around new and full moon) (Mizushima et al., 2000). Its mating system is haremic protogynus, with one male defending the coral colony on which females (smaller bodied) are resident (Fricke & Holzberg, 1974; Coates,

1982; Shpigel & Fishelson, 1986; Schwarz & Smith, 1990). The number of mature males can in rare cases increase if resident to a big coral colony. Dascyllus aruanus has restricted mobility typically due to predation pressure and habitat patchiness

(Fricke, 1980; Shpigel & Fishelson, 1986). In rare cases, habitat loss can trigger movement between coral colonies (Coates, 1982; Coker et al., 2012). Dascyllus aruanus feeds during the day and does not discriminate between the type of zooplankton prey (Pillai et al., 1985a). In captivity it will also feed on any type of animal food (Pillai et al., 1985b). This opportunistic behavior is one potential reason for its high success and wide biogeographic range, and it can indicate a generalist foraging strategy with high adaptive qualities.

1.3.2 Dascyllus marginatus (Rüppell, 1829)

1.3.2.1 Biogeography and Evolution

Dascyllus marginatus has a monophyletic origin around 9.6 MYA (McCafferty et al.,

2002). It can nowadays be found throughout the Red Sea and the Gulf of Aden, reaching Socotra Island. There are no reliable records of D. marginatus nor D. aruanus from the Persian Gulf, and even though the two co-occur throughout the

Red Sea, D. aruanus does not seem to be present in the Gulf of Aden except for very shallow areas in Djibouti. Although D. marginatus is often referred to as a Red Sea endemic, it should rather be considered an “Arabian Seas endemic” as it is also present outside the Red Sea. Similar to D. aruanus, no fossil records are available for this species, hampering geographical and temporal assignment of its origin.

However, it is very likely that it did not originate in the much younger Red Sea but rather somewhere around the Arabian peninsula, presumably in an isolated habitat in which it found refuge during increased tectonic changes prior the final opening of the Red Sea 5 MYA.

1.3.2.2 Ecology and Biology

Dascyllus marginatus is alike D. aruanus, a protogynous hermaphrodite that lives in branching corals and feeds on zooplankton close to the coral colony. The ecological and biological similarities between D. marginatus and D. aruanus are such that they

45 resemble a single species (Shpigel & Fishelson, 1986) but are reproductively and genetically isolated (there are no reports of hybridization, unlike for some other

Dascyllus species; McCafferty et al., 2002). Another putative difference can be the hierarchical structure of D. marginatus, which is still size-based but less strict

(personal measurements from the Red Sea show more exceptions, where equally large specimens inhabit a single colony). The segregation within the coral colony differs between the two species, in that D. marginatus individuals seem to have a defined and constant site of the coral colony to forage and inhabit (Holzberg, 1973).

PLD measurements for D. marginatus were not available until recently (Ben-Tzvi et al., 2007). The study reported highly variable PLDs of 26.07 ±1.07 (Aug-Sep) and

20.96 ±0.66 (Oct-Nov) for specimens from just one single reef in the Gulf of Aqaba, which can be evidence for high plasticity during ontogeny in this endemic species.

Ben-Tzvi and coworkers (2007) reported no consistent recruitment in terms of intensity, duration, and timing relative to the lunar cycle but rather a positive correlation to magnitude of downwelling. However, the recorded recruitment peaks could also be designated around the full moon and total lunar eclipse in October

2004, as well as two days after full moon and three days prior new moon in October

2005 (during the October to November survey).

1.4 REFERENCES

Allen M.B. & Armstrong H.A. (2008) Arabia–Eurasia collision and the forcing of mid- Cenozoic global cooling. Palaeogeography, Palaeoclimatology. Archambault P., Snelgrove P.V.R., Fisher J. a D., Gagnon J.-M., Garbary D.J., Harvey M., Kenchington E.L., Lesage V., Levesque M., Lovejoy C., Mackas D.L., McKindsey C.W., Nelson J.R., Pepin P., Piché L., & Poulin M. (2010) From sea to sea: Canada’s three oceans of biodiversity. PloS one, 5, e12182. Avise J.C. & Shapiro D.Y. (1986) Evaluating Kinship of Newly Settled Juveniles Within Social Groups of the Coral Reef Fish Anthias squamipinnis. Evolution, 40, 1051–1059. Azuma Y. & Kumazawa Y. (2008) Mitogenomic evaluation of the historical biogeography of cichlids toward reliable dating of teleostean divergences. BMC Evolutionary Biology. Bailey R. (2009) Ecosystem geography: from ecoregions to sites. . Barber P.H. (2009) The challenge of understanding the Coral Triangle biodiversity hotspot. Journal of Biogeography, 36, 1845–1846. Barber P.H. & Bellwood D.R. (2005) Biodiversity hotspots: evolutionary origins of biodiversity in wrasses (Halichoeres: Labridae) in the Indo-Pacific and new world tropics. Molecular Phylogenetics and Evolution, 35, 235–53. Baums I.B., Miller M.W., & Hellberg M.E. (2005) Regionally isolated populations of an imperiled Caribbean coral, Acropora palmata. Molecular Ecology, 14, 1377– 1390. Baums I.B., Paris C.B., & Cherubin L.M. (2006) A bio-oceanographic filter to larval dispersal in a reef-building coral. Limnology and Oceanography, 51, 1969–1981. Bellwood D.R. (1996) The Eocene fishes of Monte Bolca: the earliest coral reef fish assemblage. Coral Reefs. Bellwood D.R. (2012) Biodiversity hotspots, evolution and coral reef biogeography. . Bellwood D.R. & Sorbini L. (1996) A review of the fossil record of the Pomacentridae (Teleostei: Labroidei) with a description of a new genus and species from the Eocene of Monte Bolca, Italy. Zoological Journal of the Linnean Society. Bellwood D.R. & Wainwright P.C. (2002) The History and Biogeography of Fishes on Coral Reels. Coral Reef Fishes Elsevier Science (USA). Ben-Tzvi O., Kiflawi M., Gildor H., & Abelson A. (2007) Possible effects of downwelling on the recruitment of coral reef fishes to the Eilat (Red Sea) coral reefs. Limnology and Oceanography, 52, 2618–2628. Ben-Tzvi O., Kiflawi M., Polak O., & Abelson A. (2009) The effect of adult aggression on habitat selection by settlers of two coral-dwelling damselfishes. PloS one, 4, e5511.

Benton M. & Association P. (1993) The fossil record 2. Bernardi G. & Crane N.L. (1999) Molecular phylogeny of the humbug damselfishes inferred from mtDNA sequences. Journal of Fish Biology, 54, 1210–1217. Berumen M.L., Hoey A.S., Bass W.H., Bouwmeester J., Catania D., Cochran J.E.M., Khalil M.T., Miyake S., Mughal M.R., Spaet J.L.Y., & Saenz-Agudelo P. (2013) The status of coral reef ecology research in the Red Sea. Coral Reefs, 32, 737–748. Blot J. (1980) La faune ichthyologique des gisements du Monte Bolca (Province de Vérone, Italie). Catalogue systématique présentant l’état actuel des recherches. Bull. Mus. natn. Hist. nat. Paris, 4. Boehlert G.W. (1996) Larval dispersal and survival in tropical reef fishes. Reef Fisheries (ed. by N.V.C. Polunin and C.M. Roberts), Springer Science and Business Media Dordrecht. Boehlert G.W. (1999) Settlement and Recruitment of Three Damselfish Species: Larval Delivery and Competition for Shelter Space. Oecologia, 118, 76–86. Booth D.J. (1991) The effects of sampling frequency on estimates of recruitment of the domino damselfish Dascyllus albisella Gill. Journal of Experimental Marine Biology and Ecology, 145, 149–159. Booth D.J. (1992) Larval settlement patterns and preferences by domino damselfish Dascyllus albisella Gill. 155, 85–104. Borsa P., Durand J.-`Dominique, Chen W.-J., Hubert N., Muths D., Mou-Tham G., & Kulbicki M. (2015) Comparative phylogeography of the western Indian Ocean reef fauna. Acta Oecologica, 1–15. Bosworth W., Huchon P., & McClay K. (2005) The Red Sea and Gulf of Aden Basins. Journal of African Earth Sciences, 43, 334–378. Bowen B.W., Shanker K., Yasuda N., Malay M.C. (Machel) D., von der Heyden S., Paulay G., Rocha L.A., Selkoe K.A., Barber P.H., Williams S.T., Lessios H.A., Crandall E.D., Bernardi G., Meyer C.P., Carpenter K.E., & Toonen R.J. (2014) Phylogeography unplugged: comparative surveys in the genomic era. Bulletin of Marine Science, 90, 13–46. Briggs J.C. (1999) Coincident biogeographic patterns: Indo-west Pacific ocean. Evolution. Briggs J.C. (2003) Marine centres of origin as evolutionary engines. Journal of Biogeography. Buston P.M., Fauvelot C., Wong M.Y.L., & Planes S. (2009) Genetic relatedness in groups of the humbug damselfish Dascyllus aruanus: small, similar-sized individuals may be close kin. Molecular Ecology, 18, 4707–4715. Coates D. (1982) Some observations on the sexuality of humbug damselfish, Dascyllus aruanus (Pisces, Pomacentridae) in the field. Zeitschrift für Tierpsychologie.

Coker D.J., Pratchett M.S., & Munday P.L. (2012) Influence of coral bleaching, coral mortality and conspecific aggression on movement and distribution of coral- dwelling fish. Journal of Experimental Marine Biology and Ecology, 414–415, 62– 68. Connolly S.R., Bellwood D.R., & Hughes T.P. (2003) Indo-Pacific biodiversity of coral reefs: deviations from a mid-domain model. Ecology. Cooper J.W., Smith L.L., & Westneat M.W. (2009) Exploring the radiation of a diverse reef fish family: phylogenetics of the damselfishes (Pomacentridae), with new classifications based on molecular analyses of all genera. Molecular phylogenetics and evolution, 52, 1–16. Costello M.J., Coll M., Danovaro R., Halpin P.N., Ojaveer H., & Miloslavich P. (2010) A census of marine biodiversity knowledge, resources, and future challenges. PloS one, 5, e12110. Cowen R.K., Gawarkiewicz G., Pineda J.J., Thorrold S.R., & Werner F.E. (2007) Population connectivity in marine systems: an overview. Oceanography, 20, 14– 21. Cowen R.K., Paris C., & Srinivasan A. (2006) Scaling of connectivity in marine populations. Science. Cowen R.K. & Sponaugle S. (2009) Larval dispersal and marine population connectivity. Annual Review Of Marine Science, 1, 443–66. Cowman P.F. & Bellwood D.R. (2011) Coral reefs as drivers of cladogenesis: expanding coral reefs, cryptic extinction events, and the development of biodiversity hotspots. Journal of Evolutionary Biology. Cowman P.F. & Bellwood D.R. (2013) The historical biogeography of coral reef fishes: global patterns of origination and dispersal. Journal of Biogeography, 40, 209–224. Cox C. & Moore P. (2010) Biogeography: an ecological and evolutionary approach. . DiBattista J.D., Berumen M.L., Gaither M.R., Rocha L.A., Eble J.A., Choat J.H., Craig M.T., Skillings D.J., & Bowen B.W. (2013) After continents divide: comparative phylogeography of reef fishes from the Red Sea and Indian Ocean. Journal of Biogeography, 40, 1170–1181. DiBattista J.D., Choat J.H., Gaither M.R., Hobbs J.-P.A., Lozano-Cortes D.F., Myers R.F., Paulay G., Rocha L.A., Toonen R.J., Westneat M.W., & Berumen M.L. (2015a) On the origin of endemic species in the Red Sea. Journal of Biogeography. DiBattista J.D., Roberts M.B., Bouwmeester J., Bowen B.W., Coker D.J., Lozano-cort D.F., Choat J.H., Gaither M.R., Hobbs J.A., Maha T., Kochzius M., Myers R.F., Paulay G., & Vanessa S.N. (2015b) A review of contemporary patterns of endemism for shallow water reef fauna in the Red Sea. Journal of Biogeography, 1–17. Edwards A.J. (1987) Key Environments: Red Sea. International Union for Conservation of Nature and Natural Resources. {IV} Series.

Ekman S. (1935) Tiergeographie des meeres. Ekman S. (1953) Zoogeography of the sea: 1-417. Sidgwick & Jackson Ltd, London, . Elliott J.K., Lougheed S.C., Bateman B., McPhee L.K., & Boag P.T. (1999) Molecular phylogenetic evidence for the evolution of specialization in anemonefishes. Proceedings. Biological sciences / The Royal Society, 266, 677–85. Ellis S.L., Incze L.S., Lawton P., Ojaveer H., MacKenzie B.R., Pitcher C.R., Shirley T.C., Eero M., Tunnell J.W., Doherty P.J., & Zeller B.M. (2011) Four Regional Marine Biodiversity Studies: Approaches and Contributions to Ecosystem-Based Management. PloS one, 6, e18997. Eshel G., Cane M.A., & Blumenthal M.B. (1994) Modes of subsurface, intermediate, and deep water renewal in the Red Sea. Journal of Geophysical Research, 99, 15941–15952. Fenton M. & Geiselhart S. (2000) Aplanktonic zones in the Red Sea. Marine and Freshwater Research. Floeter S.R., Rocha L.A., Robertson D.R., Joyeux J.C., Smith-Vaniz W.F., Wirtz P., Edwards A.J., Barreiros J.P., Ferreira C.E.L., Gasparini J.L., Brito A., Falcón J.M., Bowen B.W., & Bernardi G. (2007) Atlantic reef fish biogeography and evolution. Journal of Biogeography, 071009214220004. Fricke H.W. (1980) Control of Different Mating Systems in a Coral Reef Fish by one Environmental Factor. Animal Behaviour, 28, 561–569. Fricke H.W. & Holzberg S. (1974) Social Units and Hermaphroditism in a Pomacentrid Fish. Naturwissenschaftern, 367–368. Friedman M. (2010) Explosive morphological diversification of spiny-finned teleost fishes in the aftermath of the end-Cretaceous extinction. Proceedings of the Royal Society B. Froukh T.J., Khalaf M.A., & Disi A.M. (2001) Studies On Taxonomy And Ecology Of Some Fish Larvae From The Gulf Of Aqaba. University of Jordan. Gerlach G., Atema J., Kingsford M.J., Black K.P., & Miller-Sims V. (2007) Smelling home can prevent dispersal of reef fish larvae. Proceedings of the National Academy of Sciences of the United States of America, 104, 858–863. Gillespie A. (2009) Group Dynamics and Interactions between Two Cohabiting Damselfishes, Dascyllus aruanus and Pomacentrus moluccensis. Natural Sciences, 82–86. Goatley C., Bellwood D., & Bellwood O. (2010) Fishes on coral reefs: changing roles over the past 240 million years. Paleobiology. Godwin J. (1995) Phylogenetic and habitat influences on mating system structure in the humbug damselfishes (Dascyllus, Pomacentridae). Bulletin of Marine Science, 57, 637–652. Grehan J. (1993) Conservation biogeography and the biodiversity crisis: a global

problem in space/time. Biodiversity Letters. Groombridge B. & Jenkins M. (1996) Assessing biodiversity status and sustainability. Hallam A. & Wignall P. (1997) Mass extinctions and their aftermath. Hemleben C. & Meischner D. (1996) Three hundred eighty thousand year long stable isotope and faunal records from the Red Sea: Influence of global sea level change on hydrography. Hoeksema B.W. (2007) Delineation of the Indo-Malayan centre of maximum marine biodiversity: the Coral Triangle. Biogeography, Time, and Place: Distributions, Barriers, and Islands (ed. by W. Renema), pp. 117–178. Springer. Hogan J.D., Thiessen R.J., Sale P.F., & Heath D.D. (2012) Local retention, dispersal and fluctuating connectivity among populations of a coral reef fish. Oecologia, 168, 61–71. Holzberg S. (1973) Beobachtungen zur Ökologie und zum Sozialverhalten des Korallenbarsches Dascyllus marginatus Rüppell (Pisces; Pomacentridae) 1) 2. Zeitschrift für Tierpsychologie. Hughes T.P., Bellwood D.R., & Connolly S.R. (2002) Biodiversity hotspots, centres of endemicity, and the conservation of coral reefs. Ecology Letters, 5, 775–784. Joyeux J.-C. & Floeter S.R. (2001) Biogeography of tropical reef fishes: the South Atlantic puzzle. Journal of Biogeography. Kelleher K., Bleakley C., & Wells S. (1995) A global representative system of marine protected areas Vol. 3: Central Indian Ocean, Arabian Seas, East Africa and East Asian Seas. Kürten B., Al-Aidaroos A.M., Struck U., Khomayis H.S., Gharbawi W.Y., & Sommer U. (2014) Influence of environmental gradients on C and N stable isotope ratios in coral reef biota of the Red Sea, Saudi Arabia. Journal of Sea Research, 85, 379– 394. Ladd H. (1960) Origin of the Pacific island molluscan fauna. American Journal of Science. Lecchini D. & Nakamura Y. (2013) Use of chemical cues by coral reef animal larvae for habitat selection. Aquatic Biology, 19, 231–238. Leis J.M. & Mccormick M.I. (2002) The Biology, Behavior, and Ecology of the Pelagic, Larval Stage of Coral Reef Fishes. The Ecology of Fishes on Coral Reefs (ed. by P.F. Sale), pp. 171–200. Academic Press, Inc., San Diego, California. Leray M., Beldade R., Holbrook S.J., Schmitt R.J., Planes S., & Bernardi G. (2010) Allopatric divergence and speciation in coral reef fish: the three-spot dascyllus, Dascyllus trimaculatus, species complex. Evolution, 64, 1218–1230. Lessios H.A., Kessing B.D., Robertson D.R., & Paulay G. (1999) Phylogeography of the pantropical sea urchin Eucadaris in relation to land barriers and ocean

currents. Evolution, 53, 806–817. Levin L.A. (2006) Recent progress in understanding larval dispersal: new directions and digressions. Integrative and Comparative Biology, 46, 282–297. Lieske E. & Myers R. (2004) Korallenriff-Führer Rotes Meer: Rotes Meer bis Golf von Aden, Südoman. Franckh-Kosmos-Verlag, Stuttgart. Limbourn A.J., Jones G.P., Munday P.L., & Srinivasan M. (2007) Niche shifts and local competition between two coral reef fishes at their geographic boundary. Marine and Freshwater Research, 58, 1120–1129. Liu S.-Y.V., Chang F.-T., Borsa P., Chen W.-J., & Dai C.-F. (2014) Phylogeography of the humbug damselfish, Dascyllus aruanus (Linnaeus, 1758): evidence of Indo- Pacific vicariance and genetic differentiation of peripheral populations. Biological Journal of the Linnean Society, n/a-n/a. Lomolino M. V., Riddle B.R., & Brown J.H. (2006) Biogeography. Sinauer Associates, Inc., Sunderland, Massachusetts. Lowe W.H. & Allendorf F.W. (2010) What can genetics tell us about population connectivity? Molecular Ecology, 19, 3038–51. 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. PNAS, 110, 16498–16502. Madduppa H.H., Timm J., & Kochzius M. (2014) Interspecific, spatial and temporal variability of self-recruitment in anemonefishes. PloS one, 9, e90648. McCafferty S., Bermingham E., Quenouille B., Planes S., Hoelzer G., & Asoh K. (2002) Historical biogeography and molecular systematics of the Indo-Pacific genus Dascyllus (Teleostei: Pomacentridae). Molecular Ecology, 11, 1377–1392. Miller K.J. & Ayre D.J. (2008) Population structure is not a simple function of reproductive mode and larval type: insights from tropical corals. The Journal of animal ecology, 77, 713–724. Miloslavich P., Díaz J.M., Klein E., Alvarado J.J., Díaz C., Gobin J., Escobar-Briones E., Cruz-Motta J.J., Weil E., Cortés J., Bastidas A.C., Robertson R., Zapata F., Martín A., Castillo J., Kazandjian A., & Ortiz M. (2010) Marine biodiversity in the Caribbean: regional estimates and distribution patterns. PloS one, 5, e11916. Miloslavich P., Klein E., Díaz J.M., Hernández C.E., Bigatti G., Campos L., Artigas F., Castillo J., Penchaszadeh P.E., Neill P.E., Carranza A., Retana M. V, Díaz de Astarloa J.M., Lewis M., Yorio P., Piriz M.L., Rodríguez D., Yoneshigue-Valentin Y., Gamboa L., & Martín A. (2011) Marine biodiversity in the Atlantic and Pacific coasts of South America: knowledge and gaps. PloS one, 6, e14631. Mizushima N., Nakashima Y., & Kuwamura T. (2000) Semilunar spawning cycle of the humbug damselfish Dascyllus aruanus. Journal of Ethology, 18, 105–108. 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.

Morcos S.A. (1970) Physical and chemical oceanography of the Red Sea. Oceanography and Marine Biology, 8, 73–202. Mouritsen H., Atema J., Kingsford M.J., & Gerlach G. (2013) Sun Compass Orientation Helps Coral Reef Fish Larvae Return to Their Natal Reef. PloS one, 8, e66039. Muss A., Robertson D.R., Stepien C.A., Wirtz P., & Bowen B.W. (2001) Phylogeography of Ophioblennius: the role of ocean currents and geography in reef fish evolution. Evolution, 55, 561–572. Öhman M.C., Munday P.L., Jones G.P., & Caley M.J. (1998) Settlement strategies and distribution patterns of coral-reef fishes. Journal of Experimental Marine Biology and Ecology, 225, 219–238. Olson D.M. & Dinerstein E. (2002) The Global 200: Priority Ecoregions For Global. Annals of the Missouri Botanical Garden, 89, 199–224. Ormond R.F.G., Roberts J.M., & Jan R.-Q. (1996) Behavioural differences in microhabitat use by damselfishes (Pomacentridae): implications for reef fish biodiveristy. Journal of Experimental Marine Biology and Ecology, 202, 85–95. Palumbi S., Grabowsky G., & Duda T. (1997) Speciation and population genetic structure in tropical Pacific sea urchins. Evolution. Pillai C.S.G., Mohan M., & Koya K.K.K. (1985a) Ecology and biology of the white tailed humbug Dascyllus aruanus (Pomacentridae, Pisces) from Minicoy Atoll. Journal of the Marine Biological Association of India, 27, 113–123. Pillai C.S.G., Mohan M., & Kunhikoya K.K. (1985b) A critique on the relationship of surface area of live coral with total number of fishes as well as the biomass of fish in a coexisting system of Chromis caeruleus and Dascyllus aruanus (Pomacentridae) at Minicoy Atoll. Journal of the Marine Biological Association of India, 27, 1–8. Pineda J., Hare J.A., & Sponaugle S. (2007) Larval Transport and Dispersal in the Coastal Ocean and Consequences for Population Connectivity. Oceanography, 20, 22–39. Planes S., Bonhomme F., & Galzin R. (1993) Genetic structure of Dascyllus aruanus populations in French Polynesia. Marine Biology, 117, 665–674. Pratchett M.S., Coker D.J., Jones G.P., & Munday P.L. (2012) Specialization in habitat use by coral reef damselfishes and their susceptibility to habitat loss. Ecology and Evolution, 2, 2168–2180. Raitsos D.E., Pradhan Y., Brewin R.J.W., Stenchikov G., & Hoteit I. (2013) Remote sensing the phytoplankton seasonal succession of the Red Sea. PloS one, 8, e64909. Randall J. (1994) Twenty-two new records of fishes from the Red Sea. Fauna of Saudi Arabia, 14, 259–275. Raynal J.M., Crandall E.D., Barber P.H., Mahardika G.N., Lagman M.C., & Carpenter K.E. (2014) Basin isolation and oceanographic features influencing lineage

divergence in the humbug damselfish (Dascyllus aruanus) in the Coral Triangle. Bulletin of Marine Science, 90, 513–532. Renema W., Bellwood D.R., & Braga J.C. (2008) Hopping hotspots: global shifts in marine biodiversity. Science. Riddle B.R., Dawson M.N., Hadly E.A., Hafner D.J., Hickerson M.J., Mantooth S.J., & Yoder A.D. (2008) The role of molecular genetics in sculpting the future of integrative biogeography. Progress in Physical Geography, 32, 173–202. Roberts C., Shepherd A., & Ormond R. (1992) Large-scale variation in assemblage structure of Red Sea butterflyfishes and angelfishes. Journal of Biogeography, 239–250. Roberts C.M., McClean C.J., Veron J.E.N., Hawkins J.P., Allen G.R., McAllister D.E., Mittermeier C.G., Schueler F.W., Spalding M., Wells F., Vynne C., & Werner T.B. (2002) Marine biodiversity hotspots and conservation priorities for tropical reefs. Science (New York, N.Y.), 295, 1280–1284. Rocha L.A., Rocha C.R., Robertson D.R., & Bowen B.W. (2008) Comparative phylogeography of Atlantic reef fishes indicates both origin and accumulation of diversity in the Caribbean. BMC Evolutionary Biology, 8, 157. Rocha L. a., Craig M.T., & Bowen B.W. (2007) Phylogeography and the conservation of coral reef fishes. Coral Reefs, 26, 501–512. Rohling E., Fenton M., & Jorissen F. (1998) Magnitudes of sea-level lowstands of the past 500,000 years. Nature. Rosen B. (1981) The tropical high diversity enigma-the coral’s eye view. PL Florey. Sale P. (1970) Behaviour of the humbug fish. Australian Natural History. Santini F. & Harmon L. (2009) Did genome duplication drive the origin of teleosts? A comparative study of diversification in ray-finned fishes. BMC Evolutionary Biology, . Scheltema R. (1971a) Larval dispersal as a means of genetic exchange between geographically separated populations of shallow-water benthic marine gastropods. The biological bulletin. Scheltema R. (1971b) Dispersal of phytoplanktotrophic shipworm larvae (Bivalvia: Teredinidae) over long distances by ocean currents. Marine Biology. Scheltema R. (1971c) The dispersal of the larvae of shoal-water benthic invertebrate species over long distances by ocean currents. 4th European marine biology symposium. Cambridge. Schmitt R.J. & Holbrook S.J. (1999) Settlement and recruitment of three damselfish species: larval delivery and competition for shelter space. Oecologia, 118, 76– 86. Schmitt R.J. & Holbrook S.J. (2002) Spatial variation in concurrent settlement of three damselfishes: relationships with near-field current flow. Oecologia, 131,

391–401. Schwarz A. & Smith L. (1990) Sex change in the damselfish Dascyllus reticulatus (Richardson)(Perciformes: Pomacentridae). Bulletin of Marine Science. Shanks A.L. (2009) Pelagic Larval Duration and Dispersal Distance Revisited. Biological Bulletin, 216, 373–385. Shpigel M. & Fishelson L. (1986) Behavior and physiology of coexistence in two species of Dascyllus (Pomacentridae, Teleostei). Environmental Biology of Fishes, 17, 253–265. Shpigel M. & Wise G.S. (1982) Niche overlap among two species of coral dwelling fishes of the genus Dascyllus (Pomacentridae). Environmental Biology of Fishes, 7, 65–68. Siddall M., Rohling E.J., Almogi-Labin A., Hemleben C., Meischner D., Schmeltzer I., & Smeed D.A. (2003) Sea-level fluctuations during the last glacial cycle. Nature, 423, 853–858. Sofianos S.S. & Johns W.E. (2007) Observations of the summer Red Sea circulation. Journal of Geophysical Research, 112. Sofianos S.S., Johns W.E., & Murray S.P. (2002) Heat and freshwater budgets in the Red Sea from direct observations at Bab el Mandeb. Deep Sea Research, 49, 1323–1340. Sorbini L. (1979) Segnalazione di un plettognato Cretacico Plectocretacicus nov. General. Boll Mus Civ Stor Nat Verona, . Sorbini L. & Bannikov A. (1991) The Cretaceous fishes of Nardo. II An enigmatic spiny-rayed fish. Bolletino della. Sponaugle S., Cowen R.K., Shanks A., Morgan S.G., Leis J.M., Pineda J., Boehlert G.W., Kingsford M.J., Lindeman K.C., Grimes C., & Munro J.L. (2002) Predicting self- recruitment in marine populations: biophysical correlates and mechanisms. Bulletin of Marine Science, 70, 341–375. Strathmann R. (1974) The spread of sibling larvae of sedentary marine invertebrates. American Naturalist. Strathmann R. (1980) Why does a larva swim so long? Paleobiology. Sweatman H.P.A. & St John J. (1990) Effects of selective settlement and of aggression by residents on distribution of young recruits of two tropical damselfishes. Marine Biology, 105, 247–252. Thornhill D.J., Mahon A.R., Norenburg J.L., & Halanych K.M. (2008) Open-ocean barriers to dispersal: a test case with the Antarctic Polar Front and the ribbon worm Parborlasia corrugatus (Nemertea: Lineidae). Molecular Ecology, 17, 5104–5117. Thresher R.E., Colin P.L., & Bell L.J. (1989) Planktonic Duration, Distribution and Population Structure of Western and Central Pacific Damselfishes

(Pomacentridae). Copeia, 1989, 420–434. Treml E.A., Halpin P.N., Urban D.L., & Pratson L.F. (2008) Modeling population connectivity by ocean currents, a graph-theoretic approach for marine conservation. Landscape Ecology, 23, 19–36. Veron J.E.N. (1995) Corals in space and time: the biogeography and evolution of the Scleractinia. Veron J.E.N. (2000) Corals of the World, Volumes 1–3. Australian Institute of Marine Science. . Watson J., Mitarai S., Siegel D., Caselle J.E., Dong C., & McWilliams J. (2010) Realized and potential larval connectivity in the Southern California Bight. Marine Ecology Progress Series, 401, 31–48. Woodland D. (1983) Zoogeography of the Siganidae (Pisces): an interpretation of distribution and richness patterns. Bulletin of Marine Science. Yamanoue Y. & Miya M. (2008) A new perspective on phylogeny and evolution of tetraodontiform fishes (Pisces: Acanthopterygii) based on whole mitochondrial genome sequences: Basal. BMC Evolutionary Biology. (2011) Coral reefs: an ecosystem in transition. Springer Science and Business Media B. V.

CHAPTER II

ANNUAL DISTRIBUTION OF RECRUITMENT OF CORAL REEF FISHES ALONG A

CROSS-SHELF GRADIENT IN THE CENTRAL RED SEA

2.1 INTRODUCTION

The study of the first life stages of coral reef fishes is a difficult and intriguing field of research. Most reproductive female fish can produce larvae by the thousands every year and these larvae spend their smallest, most vulnerable, and hardest-to-track stages out in the open oceans fighting all sorts of biotic and abiotic stressors (Leis,

1991). These biotic or abiotic encounters end lethally for the majority of the larvae

(Hjort, 1914; Dahlberg, 1979; McGurk, 1986; Doherty & Williams, 1988; Robertson

& Blaber, 1993; Doherty & Fowler, 1994) and the when and where this happens is nearly impossible to predict or observe. Thus, the pelagic larval phase is commonly known as the “black box” of the life-cycle of fishes and the extremely diverse suite of coral reef fishes make extensive insight even more difficult (Boehlert, 1996). Still, studies on the reproduction and early life stages of coral reef fishes are crucial for the assessment of fish stocks (Sale, 2004) and also to understand pathways of geneflow and connectivity between populations. A plentiful, healthy, and consistent supply of recruiting larvae is essential to maintain coral reef fish populations

(Letourneur, 1996) and biodiversity, especially in regions where these are under fishing pressure and other anthropogenic stressors. Research on early life stages of fishes includes the description of spawning aggregations, larval fish behavior, and

57 modeling of dispersal and connectivity routes of pelagic larvae, as well as recruitment and settlement patterns, grounds, sites, and habitats.

Initially, coral reef fishes were thought to spawn year-round lacking clear seasonality in reproduction due to their near-equatorial distribution in comparison to temperate marine species (Qasim, 1956) and in fact, the seasonal changes in biomass of other marine biota like zoo- and phytoplankton is relatively low in the tropics (Barry et al., 1995). However, censuses of recruiting fishes carried out daily

(e.g. Brothers et al., 1983; Robertson et al., 1988; Schroeder, 1987; Williams, 1980), weekly (e.g. Milicich et al., 1992), monthly (e.g. Caselle and Warner, 1996; Meekan et al., 1993; Robertson et al., 1993), or through seasonal snap shot sampling (e.g.

Fowler et al., 1992; Williams et al., 1994) show clear peaks of recruitment, which can be as short and limited as to over 80% of recruitment happening in just 6 days in an entire year (Sponaugle & Cowen, 1994). Such seasonal cycles can, for example, echo varying breeding capabilities of adults, where seasonal energy surpluses lead to higher fecundities (Victor, 1986). Furthermore, for most coral reef fishes these peaks in reproduction tend to occur during the more favorable conditions of the summer months (Planes et al., 1993; Russell et al., 1997; Miller et al., 2000; Russ &

Abesamis, 2010; Sponaugle et al., 2012; Cure et al., 2015) in which temperature and primary production are the highest (Russell et al., 1974, 1977; Talbot et al., 1978;

Williams, 1983; Schmitt & Holbrook, 2000). These productive warm months provide optimal conditions for efficient larval growth (Wellington & Victor, 1992; Sponaugle et al., 2006). Nevertheless, the success of larvae is a complex interplay between biotic and abiotic environmental conditions, but physiological and biological

58 prerequisites are also of particular importance, especially at the limits of a species’ biogeographic range and physiological tolerance/adaptive capabilities. For example, if the temperatures are too high, the food demand also becomes higher and if the required energy for higher metabolic rates cannot be delivered in extreme oligotrophic waters, fish larvae will find themselves in stressful conditions facing potential starvation. Aside from that, extreme temperatures above (or below) tolerance ranges can affect the larvae’s behavior and hamper their development, even with only minor changes in temperature (Stevens, 1996; Rankin & Sponaugle,

2011). It is hard to verify how these factors were influencing reproduction and geneflow in geological timescales and throughout much more extreme historical changes in the environment. Or, how (ir-)relevant a single recruitment event actually is for the long-term evolutionary success and distribution of fish species.

Other drivers influencing the mortality rate, biological processes, transport, behavior, and timing of settlement of fish larvae are reproductive cycles of prey

(Lasker, 1975) and predators (Dragovich & Potthoff, 1972; Hobson & Chess, 1978;

Hamner et al., 1988), site specific physical oceanographic and atmospheric features

(Cowen, 2002) such as strength of monsoonal influenced weather conditions

(especially in the Pacific, Indo-Pacific, and Indian Ocean vs. e.g. the Caribbean,

(Robertson et al., 1990, 1999), storms, rainfall, wind strength and direction (and even particle content; Srinivasan and Jones, 2006), as well as the timing with lunar phases and tides (e.g. Sponaugle and Cowen, 1997, 1994). The influence of the moon not only directly dictates much of the reproductive behavior and patterns

(Robertson et al., 1988; Sponaugle & Cowen, 1996) but also tides, which can drive

59 larval supply, too (e.g. Leis, 1994). The strong and widely persistent influence of the moon on the biology of almost all marine species is likely due to the fact that lunar cycles have been the only parameter with consistent patterns throughout the entire evolutionary history of coral reef biota. Regardless of spawning mode (benthic vs. pelagic) or length of larval duration, lunar patterns seem to drive spawning, recruitment, and settlement (Sponaugle & Cowen, 1994, 1996, 1997; Sponaugle et al., 2011).

Most of the aforementioned potential drivers often have seasonal patterns, which can lead to different yields of recruitment even if species spawn year-round (e.g.

Labrids (Srinivasan & Jones, 2006; Fontes et al., 2010); and some Pomacentrids

(Robertson et al., 1990, 1999; Srinivasan & Jones, 2006). Seasonality close to the equator, where we find coral reefs, is highly dictated by al conditions especially in the large Indo-Pacific Ocean. While it may play a bigger role in some regions, other regions, like the Caribbean, show less or no seasonality (Luckhurst & Luckhurst,

1977; McFarland & Ogden, 1985; Robertson et al., 1990, 1999) potentially due to lower monsoonal influences or generally highly variable environmental conditions with no clear seasonal trends to adapt to. Monsoonal conditions change current regimes (e.g. Racault et al., 2015) and in isolated seas such as the Red Sea it might have even higher impact on geneflow between populations of different biogeographic provinces like between those inside vs. outside the Red Sea.

The Red Sea has only one very shallow and narrow connection at the south to the rest of the world’s coral reefs. While it was long thought that the water exchange flow followed a typical estuarine two-layer pattern (with shallow surface water

60 inflow and deep water outflow) throughout the year, it is now known that this is only the case during winter. In summer, there is a three-layer water exchange with surface water flowing out of the Red Sea (Murray & Johns, 1997; Sofianos et al.,

2002; Siddall et al., 2003; Smeed, 2004; Yao et al., 2014). In theory, this could imply fluctuating routes and barriers for the dispersal and geneflow of species and their populations outside and inside the Red Sea depending on the timing of their reproductive peaks. For example, species from outside the Red Sea may have increased isolation and decreased connectivity to the Red Sea during summer reproduction, as their larvae would have to move against outflowing surface currents to get into the Red Sea. Furthermore, the waters inside vs outside the Red

Sea have quite different properties (e.g., temperature, density, and nutrients), which could influence the survival of sensitive fish larvae. These current regimes may have been subject to major changes throughout the Red Sea’s geological history, periodically changing the odds of which species’ larvae could make it into or out of the Red Sea, shaping biogeographic patterns and regional species richness.

In general, patterns of reproductive success in coral reef fishes may exhibit annual or seasonal flux due to a large number of variables. These variables can also exhibit anomalies and even be present/absent during some years, increasing the inaccuracy of predictive models. To reduce the noise of models including such highly variable environmental parameters and several biological responses, it is helpful to distinguish between spawning, recruitment, settlement, and reproduction. Only so, more accurate inferences can be yield when assessing the establishment, reproductive success, and growth of fish populations. Peaks in spawning do not

61 necessarily lead to peaks in recruitment and due to prolonged larval durations in some species, the recruitment peaks can follow months after peaks in spawning (e.g.

Caballero-Arango et al., 2013; Fairclough, 2005). Hence, successful spawning does not guarantee successful reproduction nor recruitment and/or settlement

(Robertson et al. 1988, 1993; Robertson 1990). Seasonality in settlement can also be a response to variation in mortality rates (Luckhurst & Luckhurst, 1977) as the cohort of one coral reef fish progeny loses individuals gradually between each of the aforementioned steps of the early pelagic life. As the mortality rates in between those stages differ among each other (eggs/sperm, larvae, recruits, and juveniles), and also among species, years, seasons, and timing with environmental settings (e.g. moon phases (Johannes, 1978; Sponaugle & Cowen, 1996), weather conditions, etc.), it is thought that spawning cycles of adults are also tailored to these predictable seasonal conditions (Doherty, 1983). Thus, one way to more easily acquire accurate data on the survival and success of larvae, reproductive potential, and seasonality in each of their nursery grounds is to assess the last larval phase and hence, focus on the recruitment of fish larvae. Recruitment, as I will use the term in this thesis, refers to a phase of transition, when coral reef fishes arrive at a reef site, commonly undergo metamorphosis, are ready to settle, and change from a pelagic to a more benthic life form. Additionally, the correct identification of the fish recruits is much simpler, as they are easier to distinguish morphologically and are lower in numbers in comparison to earlier larval stages, and even genetic quantification (through barcoding) is feasible.

There are two major sources of error to consider when assessing seasonality in

62 recruitment of coral reef fishes: (1) the temporal variation due to anomalies in environmental conditions masking ‘true’ patterns and trends; (2) the accuracy in identifying the recruits themselves. The latter can be overcome by combining identification methods and crosschecking results, which will be addressed in more detail in the next section. The former is more complicated and requires large amounts of temporal data from long-term time-series sampling over multiple seasons to resolve reliable trends. Some regions might be easier to assess than others due to their geographical and oceanographic features. These might be more stable in their environmental conditions and show fewer anomalous events.

Therefore, the isolation of the Red Sea makes it an interesting study case. It has stable wind fields, practically no rainfall or fluvial influx, and very few cloudy days.

The largest oceanographic variability is seen in the southern Red Sea and arises from seasonal monsoons in the northwest Indian Ocean (e.g. Racault et al., 2015;

Raitsos et al., 2013). The environment in the center of the Red Sea is quite homogeneous and is even less prone to atmospheric anomalies due to its central location (Attada Raju, personal communication). It has relatively oligotrophic and calm waters (little fluctuations in wave action and benign weather conditions) and relatively high sea-surface temperatures throughout the year (seasonal lows around

26-28ºC).

This study focuses on coral reef fish recruitment at three reefs located along a cross- shelf gradient in the Arabian central Red Sea. Even though the reefs off Saudi Arabia have long been exploited by fishermen (Jin et al., 2012; Abu El-Regal, 2013; Spaet &

Berumen, 2015; Spaet et al., 2016), very little research has been conducted in the

63 region (Berumen et al., 2013). In general, this area is lacking data from long-term studies or extended experiments, and even baseline knowledge on biodiversity and the biology and distribution ranges of some Red Sea species is very recent. From the few publications available, we have little information on seasonal reproductive features of invertebrates like corals, sea urchins, and anemones in the central Red

Sea (Benayahu & Loya, 1983; Kramarsky-Winter & Loya, 1998; Bouwmeester et al.,

2011, 2016). For coral reef fishes, there are some reports on conspicuous spring spawning aggregations of parrotfish in the southern Red Sea (Hipposcarus harid,

March or April around the full moon; (Gladstone, 1996; Spaet, 2013). In the northern Red Sea, various fish species seem to predominantly spawn during May to

August (warmer months and mostly in mangroves and/or over sea grasses, away from the vicinity of reefs (Abu El-Regal, 2013)). Information on recruitment in the northern Red Sea is only available from the northernmost region of the Gulf of

Aqaba (Froukh et al., 2001; Ben-Tzvi et al., 2007), which is considered to differ significantly from other parts of the Red Sea regarding its environment and biota

(Sheppard & Sheppard, 1991; Roberts et al., 1992; Fine et al., 2013), most likely a result of its isolation. Therefore, assessing recruitment in the central Red Sea provides information that may be representative for a much wider biogeographic range. In terms of nursery sites, some observations made in snappers (McMahon et al., 2011) and other coral reef fishes (Abu El-Regal 2013) suggest that nursery areas differ from the habitats of the adult snappers and thus, migration after settlement may also influence connectivity, dispersal, and site-specific presence of certain Red

Sea fish species. Similar observations have been made for other reef fishes, which

64 settle into other habitats than coral reefs, like tide pools (Randall, 1961), estuaries

(Keener et al., 1988), sea grasses (Bell et al., 1987), algae (Bellwood & Choat, 1989), or mangroves (Lindeman, 1989) and settle onto coral reefs at later life stages after undergoing a post-settlement migration. Knowledge on recruitment dynamics can therefore assist the inference of settlement habitats, nursery grounds, abundance patterns, and demographic dynamics of adult populations (Doherty & Fowler, 1994) and can be used to predict future distributions of coral reef fish species as the environment and habitat availability changes. For example, another study in the central Red Sea on parentage analysis of clownfishes reported an interesting observation, as the researchers could not find any recruiting fish during the summer months (Nanninga et al., 2015). This is generally unusual for tropical fishes and does not coincide with observations mentioned from the northern Red Sea. Apart from the study on clownfishes, no other has been carried out, as far as I am aware, on the recruitment of coral reef fishes of Saudi Arabia. This chapter pursues four main goals: First, I want to explore whether Red Sea fishes have a clear recruitment peak in the year. I hypothesize that this peak will not occur during the extraordinarily high summer temperatures. Second, I will collect baseline data on recruitment patterns of coral reef fishes and its taxonomic composition in the central Red Sea.

Third, the data will be used to model potential drivers of fluxes in total biomass and fish abundance of the most prominent taxa in the samples using the best currently available metadata. Finally, I herein propose an efficient pipeline to identify fish recruits and further feed the databanks with barcodes of Red Sea fish specimens to help ease identification and catalog biodiversity.

2.2 MATERIALS AND METHODS

2.2.1 Trials and First Collections

Prior to the final experimental design, the equipment and sampling method was tested starting in 2013 using a prototype of the Bellamare collapsible LED battery- powered light trap (500-micron mesh). This was tested at the KAUST Harbor

Residential Area in 2013 and during research cruises in February (3rd – 14th), March

(4th – 7th), and June (13th – 18th) in 2014 at a wide range of reef-types along the Saudi

Arabian coasts of the Red Sea. After successful and highly biodiverse catches, a first trial using nine light traps at the three sampling locations chosen for this study was performed, not only to assess logistical feasibility of the experiment, but also to assess potential peaks/patterns in recruitment associated to lunar cycles (e.g. higher abundances of pre-settlement fishes during the darker lunar phases

(Johannes 1978; Robertson et al. 1988)). Collections took place the same way as described below everyday for 34 consecutive days starting on the 16th of October

2014. After a short break, a final trial collection took place once again during the quarter moons of the following month for five consecutive nights (26th of November to the 1st of December and from the 11th to the 14th of December 2014) to compare catches between those two periods and have a final inspection of the logistics before the onset of a year-long sampling experiment.

2.2.2 Experimental Design

a) b) c)

Depth (m) Abu Shosha -1,060 - -500

Al Fahal -499 - -200

-199 - -100 250 km Shib Nazar -99 - -50 Land -49 - -20 Central Red Sea Reef -19 - -1 Figure 2.1: Sampling sites of recruiting fish larvae in the central Red Sea. The white star on the map to the left (a) indicates the location at the Saudi Arabian coast, off Thuwal. The map in the middle (b) shows the bathymetry of the sampling region and the color-coded stars the location at each of the three reefs, where three light traps were deployed to collect the fish recruits along the cross-shelf gradient: inshore (Abu Shosha, yellow), mid-shelf (Al Fahal, green), and at the shelf-edge (Shib Nazar, blue) reefs. To the right (c) are a sketch (top) and a picture (bottom) of the light trap and the mooring system

The study focuses on the assessment of recruitment patterns of coral reef fishes in the central Red Sea conducted in a yearlong study. A set of three replicates of light traps (nine total, three per reef; Bellamare collapsible LED battery-powered-design,

500-micron mesh, Figure 2.1) were used to collect fish recruits along a cross-shelf gradient in the central Red Sea. These were set at a reef inshore, one mid-shelf, and one at the shelf-edge off the coast of Thuwal, Saudi Arabia, near KAUST (Figure 2.1).

Collection took place once per lunar month during five consecutive nights around

67 the new moon for the period of twelve months at fixed moorings and ~2m below the surface at the northern end of the sheltered side of each reef. Every day at the same time and prior to sampling, the batteries of the light traps were replaced and each light trap and mooring were cleaned. This assured the same starting conditions for each sample and was intended to avoid sampling biases due to e.g., algal growth, the presence of other recruiting organisms on the moorings, reduced light intensity, or induced chemical cues. Additionally, at each reef a current meter (2000 kHz frequency ADCP: Aquadopp Doppler Current Profiler, Nortek AS) was set up to measure environmental data such as current speed (CS) and direction (CD), temperature (T), and pressure (P). Visibility (VIZ) was measured daily during sampling by the same diver using a transect tape attached to one of the moorings, swimming away from the mooring until it is no longer visible (always in the same direction relative to the sun). This method was used instead of a Secchi disk, as the visibility was greater than the total depth of the site. Weather (WTH) conditions were also estimated by the same person each day using a 0-10 scale, defining choppiness/sea-roughness with 0 representing a flat sea and 10 a very rough/choppy sea.

2.2.3 Collection of Biological Data

2.2.3.1 Sampling and Biomass Assessment

Clupeidae, salps, and jellyfishes that occasionally landed in the trap were

68 immediately removed from the sample, as they were not considered as coral reef fishes and the weight and water content of salps and jellyfishes bias the assessment of TBM. Their presence in the sample was however recorded as well as the number of salps. Afterwards, the total number of recruits (TNR), total biomass (TBM), and the biomass constituted by coral reef fishes (FBM) were measured separately. Coral reef fish recruits collected were counted and preserved in ethanol for later identification. In a few cases when the sample was too large (approximately > 0.4 kg), a plankton splitter was used to split the sample into equal fractions (one to three times depending on the sample size) and one of the fractions was used to extrapolate the fish abundance and biomass. Prior to this procedure the sample was inspected for the presence of unique/unusual specimen that could get lost in one of the factions and/or be overrepresented if present in the faction used for measurements.

2.2.3.2 DNA Extraction, Barcoding, Identification, and Quantification

Fish larvae are tiny, delicate, and very difficult to identify. Therefore, I developed the following protocol to identify and measure the abundance of fish recruits as confidently as possible. Recruits of coral reef fish from each day of collection and each light trap were grouped morphologically using 1.5x magnification lenses

(according to the position, size and shape of the eyes, their body length and shape, pigmentation, and shape and position of their mouth). The same person did the grouping each time for every sample. A first trial was performed using specimens

69 caught during the testing of the experimental design. These specimens were morphologically grouped as previously mentioned and then all of them barcoded to assess the accuracy of the morphological grouping technique. In most cases the specimen grouped in this manner where from the same species, and never from different genera except, putatively for gobies since this family lacks much genetic data required for this assessment. This family was therefore mainly grouped at the family level (see the results and discussion sections for further details). From each morphological group, two to three specimens were randomly chosen, photographed and used to identify the species or genus of the group using the genetic mitochondrial markers 16s and COI. If all two to three specimens belonged to the same genetic lineage (species or genus) they were labeled as correctly identified. If not so, all remaining specimens in that group were sequenced as well. Whenever there was uncertainty in the morphological assignment, due to for example, the degradation of the specimens, missing body parts, or unusual pigmentations, all specimens were also barcoded. For the amplification of mtDNA, total genomic DNA was isolated from preserved tissue following the ‘‘HotSHOT” method of Meeker et al.

(2007) and stored at -20ºC. The primers used were: L2510 (5’-CGC CTG TTT ATC

AAA AAC AT-3’) and H3080 (5’-CCG GTC TGA ACT CAG ATC ACG T-3’) (Palumbi et al., 1991) for the 16s mitochondrial rDNA gene; and the “fishcocktail” primer mix

VF2_t1 (5’-TGT AAA ACG ACG GCC AGT CAA CCA ACC ACA AAG ACA TTG GCA C-3’),

FishF2_t1 (5’-TGT AAA ACG ACG GCC AGT CGA CTA ATC ATA AAG ATA TCG GCA C-3’),

FishR2_t1 (5’-CAG GAA ACA GCT ATG ACA CTT CAG GGT GAC CGA AGA ATC AGA A-

3’), FR1d_t1 (5’-CAG GAA ACA GCT ATG ACA CCT CAG GGT GTC CGA ARA AYC ARA A-

3’) with a M13F (-21) and M13R (-27) universal primer sequence (italicized;

Messing 1983) for the 5’ region of the COI gene (Ivanova et al. 2007). PCRs were carried out in 10μl reactions containing 5 μl Qiagen Multiplex PCR Kit mix, 1.25

μmol of each primer, 2.5 μl water, and 1-2 μl DNA (30–100 ng/μl) as template. COI and 16s amplifications took place using the following touchdown PCR conditions respectively: 95ºC for 15 min, 6 cycles at 94ºC for 60 s, 55ºC for 40 s (−0.5ºC /cycle),

72ºC for 60 s, followed by 29 cycles of 94ºC for 60 s, 52ºC for 40 s, 72ºC for 60 s, and one final elongation step of 72ºC for 10min for the COI and 95ºC for 15 min, 10 cycles at 94ºC for 60 s, 57ºC for 90 s (−0.2ºC /cycle), 72ºC for 60 s, followed by 13 cycles of 94ºC for 60 s, 55ºC for 90 s, 72ºC for 60 s, and one final elongation step of

72ºC for 10min for 16s. PCR products were cleaned with Illustra ExoStar 1-step

(Biotech) following the manufacturer’s protocol. Fragments were sequenced using the forward primer (M13F (-21) for COI) on a Sanger ABI 3730XL in the Bioscience

Core lab of KAUST. Sequences were aligned and edited using GENEIOUS v.8.0.3

(Gene Codes). The pictures were used for the final confirmation of correct assignments and to exclude any misidentification due to putative wrong entries in the databanks or cross contaminations within our samples. Two different sequence libraries were used for the COI barcode alignment: (1) a library generated from

BOLD sequences and (2) an unpublished Red Sea library based on specimen from the region (see Cocker et al. in review for details and sequence references). For the

16s barcode sequences, I used the BOLD database only (www.barcodinglife.org,

(Ratnasingham & Hebert, 2013)). The BLAST command line standalone algorithm

(available at NCBI, blast.ncbi.nlm.nih.gov (Altschul et al., 1990)) was used to align

71 the obtained sequences to the libraries. A total of 1110 specimens were barcoded and photographed. Taxonomic assignments of specimens for which the BLAST results were not satisfying (no match found or low scores/sequence identity) were done in two steps: (1) the assignment was conservatively kept only to the genus or family level (depending on the BLAST score), and (2) an unrooted phylogenetic tree was built with all 1110 sequences aligned in MOTHUR (version 1.34.0 (Schloss et al.,

2009)) using neighbor-joining methods (Maximum Composite Likelihood) with pairwise deletion. This tree was then used to assign the specimen to operational taxonomic units (OTUs) based on the individuals they clustered with within the tree.

Once again the assignment was confirmed with the pictorial voucher of each specimen. Additionally, all other BLAST-based assignments were double checked within the phylogenetic tree-clusters as well, to assure the clusters and alignments were comprehensive. In this manner, a robust loop of cross-referencing was developed to confidently ID fish larvae. Once specimens were assigned to an OTU, the respective specific fish abundance was quantified (SFA), which is equivalent to the number of fishes grouped into an OTU.

2.2.4 Collection of Environmental Data

Validated monthly averages of CHLA concentrations (mg m-3) were provided by D.

Raitsos for the region of the North Central Red Sea (NCRS) (following the regional delineations of Raitsos et al. (2013)). The data was obtained from 10-year high- resolution satellite remote sensing from the NASA Giovanni website

(http://oceancolor.gsfc.nasa.gov) (see Raitsos et al., (2013) for a detailed description of the data processing).

Further environmental parameters were also collected in situ during the time frame of the experiment using three current meters (ADCPs: Aquadopp Doppler Current

Profiler, Nortek AS), one for each of the three sampling locations. These were attached to the bottom of one of the moorings (8 – 14 m depths) to collect data every 10 minutes for 4 months, after which they were replaced for maintenance.

The measurements used from these instruments were: temperature (T), pressure

(P), current speed (CS), and current direction (CD). The data was then averaged for each month and each day of sampling to model the following response variables:

Total biomass (TBM), total fish biomass (FBM), total fish abundance (FBM), specific fish abundance (SFA), and the reef itself (REEF, R1-3 from inshore to shelf-edge, to explore which parameters define the specific reef/sampling site).

2.2.5 Statistical Analysis and Correlations

Inc.). Replicates within sites were tested for significant differences using an ANOVA.

Prior to modeling, the data of each variable/parameter and its distribution was explored statistically. Data were checked for homogeneity of variance using

Bartlett’s test (Barlett, 1937) using the R function bartlett.test. For pairwise comparisons, I used the Kruskal-Wallis rank sum statistic (using kruskal.test) and

Student’s t-test (using t.test). Overall comparisons were done using an AMOVA with

73 the and the aov function. All measurements were then tested for correlations (using a correlation matrix cor function and symnum to generate the matrix). Those with >

0.6 correlation were excluded from the models. Significant differences between reefs were also assessed using ANOVAs and t-tests. To avoid problems associated with non-independence of variables, scatter plots and a correlation matrix were used to detect whether any variables were co-linear prior to the multiple regression analysis.

2.2.6 Modeling of Biological Response Variables

The analysis of models was conducted in R as well. A linear model lm was used to model TBM and REEF. Count data on FBM was modeled using a general linear model glm with poisson distribution family. In both cases, a stepwise model inference was performed to choose the best fitting model using all the environmental variables available. In the case of FBM these TBM was additionally included and the results with and without its inclusion were compared and discussed. The model’s R2 and significance were inferred using the step function. The relative importance of each predicting parameter of the lms was then calculated using calc.relimp (library relaimpo) and a bootstrapping of 1000.

2.3 RESULTS

2.3.1 Trials and First Collections

The main outcome from these trials of equipment testing and sampling was the greatly efficiency of the light traps. In general, the traps had the potential to sample a wide taxonomic range and large amounts of recruits and other zooplankton, sometimes even exceeding the capacity of the light trap itself (i.e., catches sometimes exceeded the volume of the cod end by about threefold). However, the biodiversity and biomass of the samples was also highly variable. This could be due to several factors such as seasonal and lunar timing; the site, reef type, and/or habitat; depth; weather; region; distance to the reef and coast; and potentially mainly the influence of other light sources (apart from the light trap) such as coastal cities, the light of the research vessel/boat itself if sampling is performed directly off the boat, the moonlight, etc. One of the strongest biases seemed to be related to the charge and power of the battery as well as the cleanness of the light trap. Therefore, for the yearlong experiment these two factors were kept consistent for every sample. If deployed several days in a row without proper cleaning, the light trap will have significant fouling, the mesh will turn green-brownish, and the mesh will become less translucent. This probably reduces the light intensity coming out of the light trap and its ability to attract animals. The light trap itself loses transparency and might also become more obvious to fish larvae, which might therefore avoid entering it. The epiphytic growth causing the coloring of the mesh could also

75 potentially expel chemical cues, which can bias and even repel fish larvae. For those reasons, I chose to set out the light traps for maximum five consecutive nights, cleaning the mesh thoroughly after each day of sampling with high pressurized seawater provided by a pump, a diesel generator, and a thick hose (35 mm width/diameter) on board; I also used the same type of fully-charged rechargeable batteries, turning on the LED lamps at the same time of the day for each day of sampling at each site throughout the experiment. Interestingly as well, was the fact that the LED lamps were twice missing from inside the light trap. One was recovered close to the mooring at a rocky reef structure and the hook that attaches the lamp to the top of the light trap was broken and had bite marks. The second lamp could not be retrieved but one of the mesh-funnels of the light trap was folded inside-out pointing out, towards the exterior of the trap, indicating the putative escape of a bigger animal. In my opinion both of these events indicate the presence of a curious octopus that released and took the lamp. One of these events happened inshore, where I also observed an octopus near the rocky reef structure a couple days later and the other was at the mid-shelf site. Octopi are territorial and thus, I believe these could have been the individuals entering the light trap. To avoid losses due to the missing light, the hook system was replaced by a heavy-duty karabiner.

Further losses affecting the sample itself due to octopus predation on are very unlikely since the fish larvae are probably too small to be prey for an octopus.

2.3.2 Experimental Design

No significant differences were found between the replicates in terms of TBM, FBM, and SFA measurements. Therefore, for each reef all three light-trap catches (true replicates) were pooled to measure recruitment peaks, TBM, FBM, and SFA (Table

2.1).

77 Table 2.1: List of Parameters Measured and Implemented in Linear and Generalized Linear Models of the Data REE WT VIZ WRASSE BLENNIE DAY MONTH YEAR F LAT. LONG. H [m] CS CD P SST CHLA TBM [g] FBM [g] Fish GOBIES PARROT S S DAMSEL CARDINAL PIPE 09-13 JAN 2016 1 22.3055 39.0493 2.8 31.5 0.0195 221.69 10.78 26.10 0.2322 217.40 0.3271 15 10 3 0 0 0 0 0 18-22 FEB 2015 1 22.3055 39.0493 6.7 15.8 0.0250 242.70 10.84 25.28 0.2066 274.40 0.5844 1 0 0 0 0 0 0 0 19-23 MAR 2015 1 22.3055 39.0493 4.0 15.2 0.0360 164.30 11.08 26.52 0.1681 298.10 0.2310 8 4 0 1 1 0 0 0 18-21 APR 2015 1 22.3055 39.0493 2.3 27.3 0.0271 147.13 10.76 25.84 0.1166 30.63 0.0347 4 3 0 0 0 0 0 0 17-21 MAY 2015 1 22.3055 39.0493 2.0 23.0 0.0329 153.94 11.21 27.49 0.1134 64.80 0.2327 18 12 0 0 1 1 0 1 15-19 JUN 2015 1 22.3055 39.0493 4.5 28.0 0.0299 203.44 10.64 28.89 0.1398 77.50 0.4192 55 25 17 10 1 0 0 0 15-19 JUL 2015 1 22.3055 39.0493 3.1 27.9 0.0353 252.06 10.01 30.38 0.1090 87.70 0.1856 13 5 3 1 0 0 0 0 13-17 AUG 2015 1 22.3055 39.0493 2.0 28.4 0.0270 182.79 10.45 32.72 0.0912 82.00 0.0446 8 1 4 1 0 0 0 0 12-16 SEP 2015 1 22.3055 39.0493 1.9 33.9 0.0297 155.80 10.69 32.04 0.0872 60.20 0.2790 23 4 13 2 0 0 0 0 12-16 OCT 2015 1 22.3055 39.0493 0.8 24.2 0.0273 160.83 10.62 31.85 0.0940 129.40 0.7395 43 3 36 1 0 0 0 0 10-14 NOV 2015 1 22.3055 39.0493 4.6 22.3 0.0187 214.14 10.67 29.12 0.1279 78.70 0.5390 73 32 31 4 0 0 0 0 10-14 DEC 2015 1 22.3055 39.0493 5.3 30.9 0.0198 214.41 10.60 27.18 0.1818 95.90 0.4518 63 55 4 1 0 0 0 0 AVERAGE 1 22.3055 39.0493 3.3 25.7 0.0274 192.77 10.70 28.62 0.1390 124.73 0.3391 27 13 9 2 0 0 0 0

09-13 JAN 2016 2 22.3054 38.9684 2.8 26.6 0.0261 189.40 12.18 26.05 0.2322 76.60 0.0763 1 1 0 0 0 0 0 0 18-22 FEB 2015 2 22.3054 38.9684 6.7 19.4 0.0374 182.85 12.06 25.28 0.2066 60.50 0.8021 6 3 0 0 1 0 0 1 19-23 MAR 2015 2 22.3054 38.9684 4.0 23.8 0.0248 178.57 12.66 26.04 0.1681 61.00 0.1873 9 7 0 0 0 0 0 0 18-21 APR 2015 2 22.3054 38.9684 2.3 22.0 0.0187 165.36 12.93 25.53 0.1166 121.25 0.1839 8 6 0 0 1 0 0 0 17-21 MAY 2015 2 22.3054 38.9684 2.0 19.3 0.0205 179.79 13.11 26.37 0.1134 99.80 0.3310 26 15 1 0 5 0 2 0 15-19 JUN 2015 2 22.3054 38.9684 4.5 21.1 0.0461 211.64 12.52 28.26 0.1398 118.90 0.1199 27 22 3 0 1 0 0 0 15-19 JUL 2015 2 22.3054 38.9684 3.1 18.4 0.0199 196.96 12.01 30.40 0.1090 73.70 0.0608 8 4 0 0 2 0 0 0 13-17 AUG 2015 2 22.3054 38.9684 2.0 22.3 0.0173 200.05 12.75 32.20 0.0912 244.20 0.3120 103 93 6 0 0 0 0 0 12-16 SEP 2015 2 22.3054 38.9684 1.9 21.3 0.0151 183.89 13.18 31.85 0.0872 387.00 1.1294 29 17 0 0 0 0 0 0 12-16- OCT 2015 2 22.3054 38.9684 0.8 18.2 0.0149 170.99 12.81 31.58 0.0940 416.50 0.6349 139 101 29 2 5 0 2 0 10-14 NOV 2015 2 22.3054 38.9684 4.6 24.7 0.0386 163.52 12.69 29.45 0.1279 240.30 1.3814 73 49 12 1 4 1 0 0 10-14 DEC 2015 2 22.3054 38.9684 5.3 28.6 0.0339 174.55 12.27 27.34 0.1818 98.70 0.0370 6 5 0 0 0 0 0 0 AVERAGE 2 22.3054 38.9684 3.2 22.1 0.0261 183.13 12.60 28.36 0.1390 166.54 0.4380 36 27 4 0 1 0 0 0

09-13 JAN 2016 3 22.3415 38.8538 2.8 39.8 0.0409 187.14 17.05 26.24 0.2322 80.60 0.1127 6 1 1 2 0 1 0 0 18-22 FEB 2015 3 22.3415 38.8538 6.7 33.4 0.0503 173.90 17.13 25.68 0.2066 43.70 0.3725 4 0 0 2 0 0 1 0 19-23 MAR 2015 3 22.3415 38.8538 4.0 38.2 0.0575 118.22 14.72 25.82 0.1681 35.92 0.0971 9 4 0 3 1 1 0 0 18-21 APR 2015 3 22.3415 38.8538 2.3 37.8 0.0287 192.60 14.35 25.44 0.1166 45.00 0.2774 9 2 0 2 1 1 1 0 17-21 MAY 2015 3 22.3415 38.8538 2.0 34.2 0.0243 165.97 14.81 28.13 0.1134 70.58 0.3493 20 3 3 1 8 2 1 0 15-19 JUN 2015 3 22.3415 38.8538 4.5 32.2 0.0265 183.86 14.63 28.65 0.1398 58.00 0.1096 10 2 3 1 2 1 0 1 15-19 JUL 2015 3 22.3415 38.8538 3.1 29.6 0.0328 170.97 16.15 29.33 0.1090 126.50 0.1861 12 3 1 1 4 2 0 0 13-17 AUG 2015 3 22.3415 38.8538 2.0 33.6 0.0287 180.00 16.15 30.01 0.0912 50.30 0.1531 13 7 2 0 2 0 0 0 12-16 SEP 2015 3 22.3415 38.8538 1.9 33.7 0.0425 144.64 16.15 30.70 0.0872 64.40 0.2378 12 4 3 1 0 1 0 1 12-16- OCT 2015 3 22.3415 38.8538 0.8 31.0 0.0339 171.42 16.15 29.99 0.0940 121.90 1.1580 163 6 73 79 0 0 0 0 10-14 NOV 2015 3 22.3415 38.8538 4.6 40.0 0.0352 167.99 17.19 29.65 0.1279 86.70 0.3401 35 8 9 14 0 1 1 0 10-14 DEC 2015 3 22.3415 38.8538 5.3 43.7 0.0315 163.14 17.10 27.73 0.1818 67.80 0.1177 10 3 1 4 0 0 0 0 AVERAGE 3 22.3415 38.8538 3.2 35.6 0.0361 168.32 15.97 28.12 0.1390 70.95 0.2926 25 4 8 9 2 1 0 0 TOTAL ALL 3.3 27.8 0.0298 181.41 13.09 28.37 0.139 120.74 0.3566 30 14 7 4 1 0 0 0 AVERAGE Average values of environmental parameters for each sampling period (DAYs, MONTH, and YEAR), including the reef location (REEF 1: inshore; REEF 2: mid-shelf; REEF 3: shelf-edge) and geographical point (latitude (LAT.); longitude (LONG.)) where the measurements were taken, are presented in the left half of the table. The Environmental parameters are abbreviated as follows: weather (WTH); visibility (VIZ); current speed (CS), current direction (CD); pressure (P); sea surface temperature (SST); and chlorophyll a content (CHLA). Sums of biomasses (total biomass (TBM); fish biomass (FBM)), fish abundances (Fish), and species-specific fish abundances for seven fish families are found in the other half of the table (right hand). Abundances of Scaridae (parrotfishes (PARROT)) and other Labridae (wrasses (WRASSES)) are presented separately due to their high abundance and highly species diverse nature of wrasses. The overall averages and sums are given in the last row. More details to the methodology in measurements can be found in the Material and Methods section of this data chapter (under 2.2). 78

2.3.3 Collection of Biological Data

2.3.3.1 Sampling and Biomass Assessment

TBM and FBM were both the highest at the mid-shelf reef around the fall (Sep-Nov;

green bars, Figure 2.2, Table 2.1). Inshore the TBM peaks were shifted even further

into the winter months (Jan-Mar; yellow bars, Figure 2.2). The shelf-edge reef TBM

and FBM profiles generally had lower values, but showed a similar pattern to those

from the midshelf site. Abundance of Fishes 12

inshore inshore Total Biomass (TBM) Total Biomass (TBM) mid-shelf midshore 2000 2000 2000

10 shelf-edge offshore 1500 1500 1500 8 TBM TBM 1000 1000 1000 500 500

6 500 DF Total Biomass (TBM) 0 0

Fish Biomass (FBM) Fish Biomass (FBM) Months Months 4 7 6 6 6 5 5 5 2 4 4 4 FBM FBM 3 3 3 0 2 2 2 Fish Biomass (FBM) 1 1 1 Months 0 0

Months Months Figure 2.2: Histograms of the total biomass (TBM in grams, top) and fish biomass (FBM in grams; bottom) at each reef location of this study (inshore (green), mid- shelf (yellow), and shelf-edge (blue)). The values are given in grams and each value for each reef is the sum of TBM and FBM of three replicate samples for each month of a yearlong sampling (y-axis), starting in June for better visualization. Summer months are highlighted with an orange shadow. Collections from February to December took place in 2015. January was sampled in 2016.

2.3.3.2 DNA Extraction, Barcoding, Identification, and Quantification

5136 fishes were counted, from which 4855 were identified (95%). For the identification, about 1200 individuals were barcoded and aligned against our Red

Sea databank (COI library) and the currently available COI-BOLD sequences.

However, there are still many species that are either unidentified or for which sequences are not yet available. Therefore, only conservative estimates of the putative total numbers of operational taxonomic units (OTUs) are reported.

Furthermore, for modeling and statistical analysis I only considered the abundance of the six most frequently caught families: blennies (Blennioidae), gobies (Gobiidae), wrasses (Labridae excluding Scarinae), damselfishes (Pomacentridae), parrotfishes

(scarine labrids), and cardinalfishes (Apogonidae). Parrotfishes and all other wrasses were treated separately due to the high abundance of parrotfishes in the catches and the generally high biodiversity of wrasses. The estimated taxonomic composition per reef is at least (see Table 2.2, Figure 2.4): 8 genera and 8 species of labrids; 3 genera and 3 species of scarids; 11 genera and 20 species of gobies; 7 genera and 7 species of blennies; 4 genera and 8 species of damselfishes; 3 genera and 5 species of cardinalfishes; and 3 genera and 3 species of pipefishes; and at least

7 species from 7 other genera out of 1506 collected specimens (1436 identified) shelf-edge. Mid-shore the collection of 2166 (2049 identified) fishes included at least 11 genera and 20 species of gobies; 8 genera and 8 species of wrasses; 3 genera and 3 species of parrotfishes; 7 genera and 7 species of blennies, 3 genera and 4 species of cardinalfishes; 3 genera and 3 species of damselfishes; 2 genera and 80

2 species of pipefishes; and 10 species of at least 10 other genera. Catches inshore were the lowest but with 1464 (1370 identified) collected fishes, similar to the shelf-edge yield, and included at least 20 genera and 21 species of gobies, 4 genera and 4 species of parrotfishes, 11 genera and 12 species of wrasses, 7 genera and 7 species of blennies, 4 genera and 7 species of damselfishes; 2 genera and two species of pipefishes; 3 genera and 4 species of cardinalfishes, and at least 18 other species from 16 other genera. As a result, even though yield was the lowest inshore, it seemed to have the highest taxonomic richness in the catches.

Table 2.2: Taxonomic Diversity of Recruiting Coral Reef Fish Larvae at Each of the Sampled Reefs Taxonomic diversity (conservative estimates) REEF1 REEF2 REEF3 (inshore) (mid-shelf) (shelf-edge) Total Individuals 1464 2166 1506 Collected Indiv. IDed 1370 2049 1436 Genera 67 47 46 Species 75 57 61 Labrids Indiv. IDed 68 20 542 Gen. 11 8 8 Spp. 12 8 8 Scarids Indiv. IDed 490 257 483 Gen. 4 3 3 Spp. 4 3 3 Gobbiidae Indiv. IDed 696 1609 216 Gen. 20 11 11 Spp. 21 20 20 Blenniidea Indiv. IDed 13 89 93 Gen. 7 7 7 Spp. 7 7 7 Pomacentrids Indiv. IDed 14 11 49 Gen. 4 3 4 Spp. 7 3 8 Cardinalfishes Indiv. IDed 5 8 23 Gen. 3 3 3 Spp. 4 4 5 Pipefishes Indiv. IDed 6 6 12 Gen. 2 2 3 Spp. 2 2 3 Others Indiv. IDed 78 49 18 Gen. 16 10 7 Spp. 18 10 7 Total values of sampled coral reef fish recruits are given in the first row (individuals collected). These were genetically identified using the COI mtDNA marker. Identified individuals (Indiv. IDed) are also presented for each of the three sampled reef sites (inshore, mid-shelf and shelf-edge). Based on the COI barcoding a conservative estimate of the putative total number of genera and species within the sample are provided overall identified idividuals as well as for each of the seven main taxonomic units (OTUs) and/or fish families used in this study as well as for the remaining OTUs (Others).

Other interesting results include the absence of some and high abundances of other taxa. From all collection efforts using light traps and plankton tows since 2012, the only surgeonfish recruit/larvae (Acanthuridae) ever collected in the Red Sea (as far as a I am aware of from discussions with colleagues), was a sailfin tang (Zebrasoma desjardinii) at the shelf-edge reef on the 14.09.15. The high numbers of parrotfishes in the light traps can likely be linked to the type of light trap used. Similar light traps were used in the Caribbean and those collections also yielded high numbers of parrotfishes, which on the other hand were more or less absent from catches in the

GBR, where Plexiglas light traps with a different design are mainly used.

Additionally, several squid species were also found in the catches but only at mid- shelf and shelf-edge sites.

2.3.4 Collection of Environmental Data

After the collection of environmental data, daily averages of CS, CD, and T, were chosen to integrate in the predictive models. For some days of sampling we lack daily measurements of some parameters due to technical issues with the instruments. In such cases the monthly averages were used and in case these were also not available, the monthly averages from Roik et al., (2016) from the years

2012 and 2013 were chosen. Prior to merging measurements from other sources, all parameters from the different data sets were explored for correlations, for which I only found positive correlations and no significant differences between the different available data (p > 0.05) allowing little risk of biases by merging data sets. Constant 83 positive correlations between datasets also speak for the presumably more stable environmental conditions throughout the years in geographically isolated and enclosed coral reef ecosystems such as the RS, which are less exposed to the influence of anomalies in open oceans.

Statistical tests showed highly significant (p <0.001) differences between all three sites for the parameters: CS, CD, and VIZ. T did not differ significantly between sites but was consistently higher at the inshore site and had the largest range (25.28 –

32.83 ºC, avg. 28.67 ºC). The lowest T measurements with the smallest range of change were measured shelf-edge (25.33 – 30.70 ºC, avg. 28.16 ºC). Mid-shelf, T was similar to those inshore (25.28 – 32.46 ºC, avg. 28.41 ºC). Another interesting feature is that SST in this region peaked in the months of August to October

(coherent with Roik et al. (2016) and Raitsos et al. (2012)), which is by two month outside and delayed from the common summer months of the northern hemisphere

(June, July, and August).

VIZ and CS were the largest and strongest shelf-edge and also had the largest range of variation (VIZ: 24-52 m, median: 35 m; CS: 0.016 – 0.082 m/s, median: 0.033 m/s). Inshore and mid-shelf measurements were more similar with the mid-shelf having the lowest VIZ and CS (Inshore: VIZ: 13-37 m, median: 25 m; CS: 0.015 –

0.059 m/s, median: 0.027 m/s; Mid-shelf VIZ: 24-52 m, median: 21.5 m; CS: 0.013 –

0.053 m/s, median: 0.023 m/s). The predominant daily CD was to the south (Figure

2.3). The largest fluctuations in CD were at the inshore reef, which even measured currents in the opposite direction (towards the north-west). The mid-shelf reef showed the smallest fluctuations in CD (see Figure 2.3, Table 2.1). Fluctuations in 84 sea level ranged from 1 – 3 m and followed a shelf gradient trend with the largest changes at the shelf-edge site (R1: 1.1 m; R2: 1.5 m; and R3: 2.9 m). Overall, the current profilers at the shelf-edge site (R3) did not perform any measurements in the period of July to October (see Figure 2.3)

A B C R1: inshore R2: mid-shelf R3: shelf-edge

R3 R3 R3 R2 R1 R2 R1 R2 R1

10 km

Figure 2.3: Rosette plots representing the main current directions and dominance of the current at three reefs off Thuwal, Saudi Arabia in the central Red Sea. Measurements were done by three current meters (2000 kHz frequency ADCP: Aquadopp Doppler Current Profiler, Nortek AS), one per sampling site, along the cross-shelf. The increasing redness and size of the arrow represents the increase dominance/prevalence of the direction of the current flow. The plots are separated according to the periods of sampling: 2015 Mar-Jun (A); 2015 Jul-Oct (B); 2015/16 Oct-Mar (C). The map in the middle is lacking measurements for the reef at the shelf-edge due to failure of the instrument throughout Jul-Oct. The plots were generated by Colleen Campbell.

All in all, these differences in environmental parameters suggest that there might be reef specific features driving biological processes. In fact, the factor REEF seemed to be one of the highest explanatory parameters implemented in the linear models of biological response variables (TBM and FBM) and was correlated with the abiotic measurements. In other words, each reef had its unique biological and 85 environmental signature. Therefore, I decided to model the biological response discretely for each reef instead of over all sites (i.e. the entire data set).

None of the environmental measurements implemented models had a correlation above 0.6. If any, correlations were below 0.3. The removal of parameters with high correlations helps avoid trends that may hide underlying dynamics.

2.3.5 Seasonality and Recruitment Peaks

There were clear peaks in the recruitment and FBM (Figure 2.4). These were most evident in wrasses and parrotfishes, with both peaking in October and November.

Gobies and blennies also showed peaks in those months, however these also had peaks during other months. Blennies for instance also peaked in May and gobies in

December, both months outside of the summer period. However, gobies also show a peak in August at the shelf-edge site (which is also the reef with the overall lowest temperatures) and, for the Blennies, the peak in May was the dominant peak at the shelf-edge reef but were generally low in numbers. Damselfishes also recruited generally in low numbers, but steadily throughout the year without any noticeable peak or trend. Overall, and especially in the case of the wrasses, peaks in recruitment seem to be shifted towards cooler fall/winter months. This could be due to the relatively hot summers in the Red Sea in comparison to other coral reef ecosystems. In the case of the inshore reef, which also exhibits the hottest temperatures, gobies peak even later in the year, in December. However, these peaks could also be species specific artifacts as the species might be reef specific 86 DF

0 2 4 6 8 10 12

themselves (e.g.: Munday et al., 1997) making it hard to exactly infer what is driving

these cross-shelf differences in recruitment peaks. In other words, it can be a Abundance of Recruits (counts) FA

0 200 400 600 400 600 800 consequence of evolutionary biology, biogeography, physiology, the environment 200 or

the combined response to all of those, which can also act at different rates and Abundance of Fishes in Months different proportions depending on the taxon.

a) Speciﬁc Fish Abundance (SFA) Abundance of Fishes b) Total Number of Recruits (TNR) Abundance of Recruits 12 Abundance of Fishes Abundance of Fishes 8 8 8 offshore midshore inshore

6 6 inshore Months inshore 4 4 CF 4 CF mid-shelf midshore 2 2

10 shelf-edge offshore Cardinalﬁshes 0 0

Abundance of Fishes Abundance of Fishes Months Months 12 12 12 10 10 8 8 8 DF 8 6 6 DF DF 0 2 4 6 8 10 12 4 4 4 Damselﬁshes 2 2 0 0

Abundance of Fishes Abundance of Fishes Months Months 40 40 40 6 30 30 DF 20 20 BL 20 BL Blennies 10 10 Abundance of Recruits (counts) FA 0 0

Abundance of Fishes4 Abundance of Fishes Months Months0 200 400 600 400 400 600 800 200 300 300 200 200 WR 200 WR Wrasses 100 100 2 Abundance of Fishes 0 0 Abundance of Fishes Abundance of Fishes

Months Months 350 350 350 Months 300 300 250 250 250 200 200 PF PF 0 150 150 150 100 100 Parro8ishes 50 50 50 0 0 Abundance of Fishes Abundance of Fishes

Months Months 500 500 500 Months 400 400 400 300 300 300 GB GB 200 200

Gobies 200 100 100 100 Abundance of Recruits 0 0

Months Months

200 400 600 offshore midshore inshore 800 Months Figure 2.4: Graph (a) shows the specific fish abundance (SFA) of the six most collected families within recruiting fish larvae with increasing abundance from top to bottom. The abundances for each month of sampling displayed on the x-axis, are color-coded per reef site (inshore (green), mid-shelf (yellow), and shelf-edge (blue)). The number of collected fish is on the y-axis. Graph (b) to the right shows the total number of recruits (TNR) sampled at each reef, grouped per month and color-coded as in (a). Summer months are highlighted with an orange shadow.

2.3.6 Modeling of Biological Response Variables

Along the cross-shelf gradient significant differences between reefs were found for

CD, CS, and VIZ. In terms of biotic parameters reefs differ significantly from each other in TBM, and the SFA of damselfishes, wrasses, and gobies. However, they did not differ significantly from each other in terms of FBM, FBM, and the SFA of pipefishes, cardinalfishes, blennies, and parrotfishes.

The factor REEF was further modeled to assess the relevance of each feature distinguishing the reefs using a lm. The best fitting model included 6 regressors explaining 37.43% of variance in the following proportion: VIZ (59.4%), CD

(14.4%), CS (12.6%), blennies, wrasses, and cardinalfishes.

Each of the following biological response variables were also modeled for the entire collection/dataset and for each specific reef (see Table 2.3).

2.3.6.1 Total Biomass (TBM)

TBM was modeled using the following components: REEF, MTH, lunar day (LD), chlorophyll a concentrations (CHLA, average from 2003-2011) and the daily averages of: weather (WTH), VIZ, CS, CD, P, and T.

The best fitting model explained a proportion of variance of 30.73%. It included 6 regressors with the following relative importance for the main contributors: VIZ

(52%), T (18%), CS (14%), LD, CHLA, and P. At R1 the model explaining 50.72% of variance included 4 regressors: CHLA (42%), VIZ (26%), MTH (19%), and T. R2 88 included 5 regressors and explained 61.7% variance: T (41%), P (21%), CHLA

(18%), MTH and VIZ. R3 36.21% variance with 4 regressors: LD (49%), VIZ (24%),

P (14.1%), and WTH. The results from R2 suggest that it is less influenced by these

“short-term”/daily fluxes related to weather conditions, perhaps due to the more sheltered nature of this site.

2.3.6.2 Total Fish Biomass (FBM)

The total fish biomass (FBM) was modeled with the same components above but also included TBM as a predictor.

TBM was a significant for all resulting models. Over all sites the best fitting model included 3 regressors explained 16.6% of variance in the following proportion: TBM

(74%), MTH (17.2%), and WTH (8.4%). At R1, 5 regressors accounted for 41% of variance: LD (46%), TBM (22%), MTH (17%), CD, and P. At R2 3 regressors accounted for 26%: TBM (58%), WTH (26%), and CD (16%). And at R3 5 regressors accounted for 29%: TBM (60%), WTH (14%), CHLA (13%), P, and T.

Without TBM as explanatory regressor, the explanatory power of the models was much lower with the following results: Over all sites, the best fitting model included

3 regressors explaining only 7.4% of variance: MTH, WTH, and VIZ. At R1, 4 regressors explained 34.2% variance: LD, P, MTH, and CD. At R2 3 regressors explained 25.6% of variance: T, WTH, P, and CD. And at R3, 5 regressors explained

21.3% of variance: T, WTH, CHLA, P, and LD.

2.3.6.3 Total Biomass and Total Fish Biomass (TBM and FBM)

In general, it seems that daily “weather” related variables as well as temperature

(seasonal regressors) accounted for most of the variance of the TBM. When comparing the resulting models for TBM vs. FBM, one could interpret that TBM is more likely to change on a daily basis and is thus more susceptible to short-term environmental conditions such as changes related to weather. Invertebrate zooplankton and larvae (a major component of TBM) have a more planktonic nature than fish larvae and their abundance thus, more likely to be influenced by currents.

The invertebrate zooplankton and larval input may thus be more susceptible to fluctuations in current speed, visibility, and pressure. On the other hand, FBM could potentially be more stable to short-term, quick fluxes or changes in environmental conditions and be thus better explained by more seasonally driven (longer-term) environmental features like the month of sampling, temperature, and CHLA, as vertebrates are more nektonic and may have more complex swimming behavior.

90 Table 2.3: Visualization of the Explanatory Variables in the Best Fitting Linear and Generalized Linear Models of Larval Recruitment in the Red Sea

TBM FBM ALL WR PF GB BL DF CF WR PF GB BL DF CF WR PF GB BL DF CF REEF

ALL R1 R2 R3 WR PF GB BL DF CF ALL R1 R2 R3 R1 R1 R1 R1 R1 R1 R2 R2 R2 R2 R2 R2 R3 R3 R3 R3 R3 R3 w wo w wo w wo w wo

x x x x x x REEF x x 8 x 17 x x x x x x x x x x x x x x MTH x x 46 x x x x x x x x x x x x LD x 17 x 26 x 14 x x x x x x x x x x x x WTH 52 x x x x x x x x x x x x x x x x x x x x 59.4 VIZ 14 x x x x x x x x x x x x x x 12.6 CS x x 16 x x x x x x x x x x x x x 14.4 CD x x x x x x x x PR 18 x x x x x x T x x x 13 x x x x x x x x x x x x x x x CHLA 74 22 58 60 x x x x x x x x x x x x x x TBM

FBM x x x x x x x x x x x WR x x x x x x x x x x x x PF x x x x x x x x x x x x x GB x x x x x x x x x x x x x x BL x x x x x x x x x x x x DF x x x x x x x x x x x x CF 31 51 62 36 17 7 41 34 26 26 29 21 12 13 13 10 5 6 7 11 10 8 2 7 10 9 9 11 3 4 9 12 9 5 9 2 # % Abiotic (blue) and biotic (green) parameters implemented in the models are displayed per row. Green or blue colored cells indicate that the said parameter was kept in the best fitting model. The numbers inside those colored cells give the percentage that a single parameter explained of the total variance explained by the best fitting model. The total variance explained by the models is given in the last row of the table (left side of the black vertical bar). The number of parameters (#) that were kept in the best fitting model of each of the response variables (in the top row) is also given in the last row (right side of the black vertical line). Grey cells indicate that the parameter was excluded a priori and white cells, that it was not included in the best predictive model. Vertically, each column presents each of the modeled variables (Total Biomass (TBM); FBM; SPAs: wrasses (WR); parrotfishes (PF); gobies (GB); blennies (BL); damselfishes (DF); cardinalfishes (CF)); either over all sites (ALL) or per reef site along a cross-shelf gradient (inshore (R1); mid-shelf (R2); shelf-edge (R3)). For the FBM, each model was ran including ( “w”: with) or excluding (i.e., “wo”: without) TBM as a predictive parameter. 91

2.3.6.4 Specific Fish Abundance (SFA)

For the glms of the different families, I chose the same parameters as above but also included biological factors such as the presence of other fish OTUs and TBM (as a proxy for the presences of invertebrates and a potential food source for the larvae as

TBM was mainly composed of copepods). The model’s variables were therefore the following (always excluding the respective modeled fish family): TBM; reef; month; lunar day; CHLA; the daily averages of weather, visibility, current speed, current direction, temperature; and daily abundances of wrasses, parrotfishes, gobies, blennies, damselfishes, and cardinalfishes. This total of 15 regressors were used for each OTU over all sites and 14 at each reef, from which seven were categorize as abiotic and seven as biotic (as in Table 2.3). When looking at the fish abundances of each OTU at each reef, the percentage/ratio of biotic and abiotic factors explaining the abundance of one of the fish families was with 50% for each, equally distributed at the shelf-edge and inshore reefs. At the mid-shelf reef (R2), the inclusion of abiotic factors in the models of fish abundance was slightly lower (40% of the explanatory variables were categorized as abiotic), indicating that abiotic factors might have a lower impact in the more sheltered environment of R2. However, the ratios are, overall, quite equal making it difficult to draw any further conclusions apart from the fact that biotic and abiotic factors seem to be relevant predictors in the models of abundances of fish families. Furthermore, modeling the most common families (to reduce noise by the inclusion of uninformative regressors): parrotfishes, other wrasses, and gobies; the inclusion of biotic factors into the final model

92 increases to 70% inshore, 50% mid-shelf and 60% shelf-edge. Either way, any biotic parameter influencing the abundance of fish recruits (such as abundance of other recruiting taxa, TBM, and CHLA) is likely also influenced by abiotic factors with seasonally driven fluctuations, especially in tropical seas at high latitudes. Thus, any correlation between biotic factors and abundance SPA and FBM is likely a coupled response that initiated with abiotic changes (likely seasonal).

2.3.7 Species Composition

The abundance of the most commonly caught species also differed among reefs

(Figure 2.4) and are given with “>” signs in case a specimen was lost or too degraded to be identified. Within the wrasses, Cheilinus fasciatus (>396 shelf-edge) and

Stethojulis alboviatta (>74 shelf-edge; >3 mid-shelf; >8 inshore) were the most abundant, however their abundance was reef-specific. In general, most of the wrasse abundance was found at the shelf-edge reef while almost no wrasses were collected at the mid-shelf site. Cheilinus fasciatus was replaced by C. abudjubbe at the mid- shelf and inshore reefs (>1 and >18 respectively). The inshore reef also commonly had a relatively abundant presence of Oxycheilinus gigramm (>10). Parrotfishes had lower taxonomic diversity; most of the catch was represented by Chlorurus sordidus

(>297 shelf-edge; >70 mid-shelf; >304 inshore) and Hipposcarus harid (>89 shelf- edge; >36 mid-shelf; > 132 inshore). Similar to the wrasses, parrotfishes were less abundant at mid-shelf reefs. However, in comparison to the wrasses, the dominance of the two most common parrotfish species was consistent at all reefs, with C.

93 sordidus being consistently more abundant than H. harid, and the total abundance of parrotfishes consistently higher inshore than at the shelf-edge. Gobies were the most diverse and species rich family and also the hardest to identify (Coker et al. in review; Isari et al. in review). The identification was challenging morphologically and molecularly due to missing barcodes and sequence data as well as low numbers of described species (something also reported by D. Coker and S. Isari, in review).

The species composition of gobies also differed among reefs. This is to be expected as previous studies found that gobies are generally very habitat-type-specific

(Munday et al., 1997). The most abundant taxon was different among all three reefs, except for Gobiidae sp. 36, which was the most abundant at the shelf-edge and inshore site. In detail: at the shelf-edge, Gobiidae sp. 36 was the most abundant

(>46) followed by Eviota sp. (>29) and Lotilia sp. (>27). The highest overall number of gobies was found at the mid-shelf site, with Gobiidae sp. 28 most prominent

(>70), followed by Cryptocentrus sp. (>68), Vanderhorstia sp. (>50), and Gobiidae sp.

30 (>28). Inshore, the most frequently caught Gobiidae were similar to the shelf- edge site: Gobiidae sp. 36 (>329), followed by Eviota sp. (>99), and Cryptocentrus sp.

(21). Blennies were mostly from the Tripterygiidae family. The reefs differed in the numbers of genera and were represented with >5 genera at the shelf-edge site, >3 genera mid-shelf and one single Tripterygiidae sp. inshore. The inshore reef also had the lowest abundance of blennies. Damselfish catches were dominated by Chromis viridis at all sites and were the easiest to identify as they had all undergone metamorphosis and were very similar to their adult stage.

2.4 DISCUSSION AND CONCLUSIONS

2.4.1 Patterns in Recruitment of Coral Reef Fishes in the Central Red Sea

The majority of coral reef fishes have dispersive larvae. Thus, the geographic boundaries of their distribution range are usually set by constraints to establish their recruits (Luiz et al., 2013) and temperature seems to be a main limiting factor for said limits to dispersal (Kimball et al., 2004; Figueira et al., 2009; Poertner &

Peck, 2010). Temperature also strongly influences species growth, reproduction, and survival (Nakano et al., 2004; Sponaugle & Grorud-Colvert, 2006; Sponaugle et al., 2006; Biro et al., 2007; Dixson et al., 2008; Munday et al., 2008). This chapter was a first step to quantify and recruitment in the Red Sea. Additionally, basic modeling was applied to capture and categorize sources of the variance found in putative recruitment patterns. All in all, it seemed that food availability could be a strong predictors for recruitment peaks in the Red Sea. However, these peaks may be shifted towards later and cooler months of the year putatively linked to hotter sea surface temperatures highs than common for coral reefs in the Red Sea.

Additionally, species composition and peaks in recruitment are highly likely to be reef-type specific.

The big question I find is, how biological, ecological, and probably highly variable processes in early life history of coral reef fishes (driving recruitment patterns) can actually influence long-term evolutionary processes, which ultimately shape the biogeographic distribution of coral reef fishes. Worldwide, recruitment peaks are

95 commonly found during the warm summer months (e.g. GBR ~28°C (Russell et al.,

1997): based on 54 species One Tree Island over the period of three years; Fontes et al., 2010). In the Red Sea, very little is known about the influence of temperature on reproductive patterns and recruitment, in general. However, from one attempt to assess recruitment in the central Red Sea, settling recruits of a damselfish

(Amphiprion bicinctus), were only found during the winter months in January and

February (Nanninga et al., 2015). It was hypothesized it may be too hot but there was no direct data to back this up. The samples in our study barely contain

Pomacentrids, which makes a comparison to the former study hard. Contrastingly, most larval collections from the Great Barrier Reef (GBR) have high numbers of damselfishes, which is likely linked to the type of light trap used. From the few

Pomacentrids collected herein, recruitment seemed homogeneous throughout the year and from personal observations on Dascyllus aruanus, another damselfish, recruits were spotted at several reefs over the entire year. Nevertheless, and for the main families represented in our sample (parrotfishes, other labrids, and gobies) we also recorded recruitment peaks outside of the typical summer months. Instead, these peaks occurred in fall and early winter (October, November, and December).

Additionally, peaks in recruitment can be species-specific and thus, dependent on the species caught, which would also explain discrepancies between observations made at the family level.

Another feature of the central Red Sea that might partially account for the shift in recruitment is that the annual SST-highs are not within the usual summer months of the northern hemisphere (June to August) but occur rather later in the year during

August to September (Raitsos et al., 2013; Roik et al., 2016). Hence, it is recommendable to specify which are the months of SST highs and refer to those as the local summer. Nonetheless, recruitment peaks in the Red Sea do not align with the SST highs either but instead are further delayed into the slightly cooler months of October and November, in contrast to the GBR (Russell et al., (1997); Figure 2.5).

This could also be interpreted as a response to particularly high SST and an adaptive strategy for successful recruitment under exceptionally high temperatures. We also seem to measure a cross-shelf gradient in SST highs with hottest SST at inshore and lower SST highs at the shelf-edge site, which accordingly seem to increase the shift in recruitment and might be the reason why the latest peaks are found at the hotter inshore reef (see Figure 2.5).

34 34 34 34 34 34 34 34 34

32 32 32 32 32 32 32 32 32

30 30 30 30 30 30 30 30 30

28 28 28 28 28 28 28 28 28

26 26 26 26 26 26 26 26 26

24 24 24 24 24 24 24 24 24

22 22 22 22 22 22 22 22 22

20 20 20 20 20 20 20 20 20 a) 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 1 b) 3 2 4 3 5 4 6 5 7 6 8 7 9 8 10 9 11 10 12 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 1 c) 6 2 7 3 8 4 9 5 10 6 11 7 12 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12

34 34 34 34 34 34 32 32 32 32 32 32 30 30 30 30 30 30 28 28 SST 28 a) 28 28 28 26 26 26 26 26 26 34 34 34 34 34 34 Coral Reef Fish Recruitment 34 34 34 32 32 32 24 24 24 32 24 32 32 24 32 32 32 24 30 30 30 30 30 30 30 Inshore 30 30 28 28 28 28 22 22 28 28 22 28 22 28 28 22 22 26 26 26 26 26 26 26 26 26 24 24 24 24 20 20 24 24 20 24 20 24 24 20 20 22 22 22 22 1 1 2 2 3 3 4 4 5 5 22 6 6 7 7 8 8 9 10 11 12 12 22 22 1 1 2 2 3 3 4 4 5 5 6 6 22 7 7 8 8 9 9 10 11 12 22 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 10 11 12 12 20 20 20 Peaks are usually in the summer months 20 20 20 20 20 20 a) 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 1 b) 3 2 4 3 5 4 6 5 7 6 8 7 9 8 10 9 11 10 12 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 1 c) 6 2 7 3 8 4 9 5 10 6 11 7 Mid-shelf 12 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Planes et al. 1993; Russell et al. 1997; Miller et al. 2000; Abesamis & Russ 2010; Sponaugle 2012 34 34 34 34 34 35 34 32 32 32 32 32 SST[°C] 32 Shelf-edge SST 30 30 30 30 30 30 30 28 28 28 28 25 • Florida 28 28 GBR Inshore 26 b) 26 26 26 26 20 26 34 34 34 d) 34 34 34 34 34 34 • Australia GBR 34 34 34 34 32 34 32 34 34 Coral Reef Fish Recruitment 32 34 34 24 24 24 32 32 32 32 24 32 32 32 32 32 32 24 32 32 24 Series1 32 30 32 30 32 30 30 30 30 15 30 30 30 Mid-shelf 30 30 30 30 28 30 28 30 30 28 30 30 28 28 28 22 22 28 28 28 22 • Hawaii 28 28 28 28 22 26 28 26 28 28 26 22 26 28 26 26 28 26 22 26 26

26 26 26 24 24 26 24 24 26 24 24 26 24 24 24 26 26 26 10 22 22 22 24 24 24 20 24 22 24 22 22 24 22 22 22 24 20 24 24 20 20 20 20 • 20 20 20 Jamaica 20 20 20 20 20 20 22 22 22 22 22 22 22 22 22 1 1 2 2 3 3 1 2 4 3 4 4 5 5 5 6 7 8 6 6 9 10 7 7 11 12 8 8 1 9 2 3 10 4 5 6 11 7 8 12 12 9 10 11 12 1 2 1 1 3 2 1 4 3 5 4 2 6 2 5 7 6 8 3 7 3 9 8 10 9 4 11 4 10 12 11 5 12 5 6 6 1 2 7 7 3 4 5 8 8 6 7 9 9 8 9 10 10 11 12 11 12 1 2 3 4 5 1 6 2 7 3 8 4 9 5 10 6 11 7 12 8 1 9 1 10 11 2 2 12 3 3 1 4 2 4 3 4 5 5 5 6 6 7 6 8 9 7 7 10 11 12 8 8 9 1 2 10 3 4 11 5 6 7 12 12 8 9 10 11 12 a) Peaks are usually in the summer months b) Mean of Species RecruiCng c) Shelf-edge 20 20 20 20 20 20 20 20 20 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 5 a) 1 b) 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 1 c) 6 2 7 3 8 4 9 5 10 6 11 7 12 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 • Japan 34 Planes et al. 1993; Russell et al. 1997; Miller et al. 2000; Abesamis & Russ 2010; Sponaugle34 2012 34 0 34 34 34 34 34 34 GBR 32 32 0 32 2 4 6 8 10 12 14 34 34 34 32 32 32 32 32 35 32 30 30 30 32 SST[°C] 32 30 32 30 30 SST Summer 30 30 30 28 30 28 28 30 30 28 30 28 28 28 28 26 28 26 26 28 28 26 c) 25 28 Coral Reef Fish Recruitment 26 26 • Florida Inshore 26 26 24 26 24 24 26 26 24 26 24 24 Coral Reef Fish Recruitment d) 20 • 24 22 24 22 22 Australia GBR 24 24 24 22 24 22 22 Series1 20 20 15 20 20 20 Mid-shelf 22 22 22 22 22 22 20 • Hawaii 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 10 11 Peaks are usually in the summer months 12 12 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 11 12 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 10 11 12 12 20 20 20 20 10 20 20 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 10 11 Peaks are usually in the summer months 12 12 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 10 11 12 Planes et al. 1993; 1 1 2 2 3 3 Russell et al. 19974 4 5 5 6 6 ; Miller et al. 2000; 7 7 8 8 9 Abesamis10 11 & Russ 2010; 12 12 Sponaugle 2012 • Jamaica

Mean of Species RecruiCng Shelf-edge 5 Planes et al. 1993; Russell et al. 1997; Miller et al. 2000; Abesamis & Russ 2010; Sponaugle 2012 35 • Japan 0 SST[°C] SST GBR 35 0 30 2 4 6 8 10 12 14 SST[°C] SST 30 25 • Florida Inshore Summer d) 20 • 25 • Florida Inshore Australia GBR Series1 d) 15 Mid-shelf 20 • Hawaii d) • Australia GBR 10 Series1 15 Mid-shelf • Jamaica

Mean of Species RecruiCng • Hawaii Shelf-edge 5 10 • Japan 0 • Jamaica

Mean of Species RecruiCng Shelf-edge GBR 5 0 • Japan 2 4 6 8 10 12 14 0 GBR Summer 0 2 4 6 Figure8 2.10 5: 12 Yearlong 14 (months on the x-axes) profiles of abundance of recruiting Summer coral reef fishes (continuous color-coded line for each of the sampling sites (a-c): inshore (yellow), mid-shelf (green), and shelf-edge (blue)); total biomass profiles (dashed color-coded line respectively) in the central Red Sea (RS, a-c) and the site- specific sea surface temperature (SST) profiles in orange (y-axes; in ºC; all graphs (a-d)). The bottom graph (d) represents the results of a similar study displaying the mean of species recruiting (continuous black line) at a site on the Great Barrier Reef (GBR; Russell et al. (1997)). The Red arrows indicate the peaks of recruitment and annual SSTs highs. Comparing graphs the shift of the recruitment peak and the SST- High increases along the cross-shelf gradient from the shelf-edge to inshore within the RS (a-c), while the GBR has an almost perfect alignment of the SST-High and the peak of recruitment, which both occur during the summer months of the southern hemisphere (grey-shadowed vertical bar). The SST-Highs in the RS sites are further denoted by an orange shadowed vertical bar, while the respective summer months are highlighted with in grey as for the GBR.

Interestingly, however, the models tested herein (including temperature and other

parameters) generally do not support temperature as a main driver of the

abundance of fish recruits. Instead, reef type (i.e., location along the shelf) and total

biomass (TBM) seem more significant in the best fitting models.

On one hand, the fact that reef type has a high explanatory power on the total fish biomass (FBM) might point towards (a) species-specific larval behavior leading to differences in recruitment success along the cross-shelf gradient, and/or towards

(b) distinctive habitats and environmental conditions along the cross-shelf gradient giving each reef specific features to which recruiting larvae respond. The latter would lead to taxonomic differences in recruitment and species-specific abundance among reefs, which we indeed found in this study. On the other hand, the importance of TBM (which is here used as a proxy of larval fish prey) as explanatory variable can be related to the vital necessity to meet a larvae’s food demand in hot

(higher metabolic rates) and highly oligotrophic waters. Indeed, when plotting the data of TBM and FBM, the profiles and their seasonal peaks resemble those typically present in the “biological pump” of temperate seas with co-varying fluxes in abundance of phyto- and zooplankton, in which seasonally and cyclically a peak in phytoplankton (primary production) leads to a peak in zooplankton and a consequential decline in phytoplankton, which is then followed by a decline in zooplankton from which the cycle starts over again (Longhurst & Harrison, 1989;

Longhurst, 1995). These cycles are also linked to temperature profiles. Here, the profiles of TBM and FBM similarly display an increase in TBM followed by a peak in

FBM and recruitment (Figure 2.3). If we assume TBM represents food availability

(as it was mainly composed of copepods, the primary larval prey) it might explain why an increase in TBM leads to an increase in recruitment. Similar observations have been described in the “match–mismatch” hypothesis, in which larvae hatch near the peak abundance of their zooplankton prey (Cushing, 1973, 1990). In the

99 very warm Red Sea, these larvae potentially have increased metabolic rates, which might explain why ultimately TBM is the more important variable rather than temperature in the best fitting models. All in all, it is a very complex process in which the origin of zooplankton prey and the survival of recruits cannot just be dictated by one single parameter like temperature, but likely involves a cascade of physical processes providing nutrients for primary production (CHLA) along with convenient oceanographic conditions.

Another strong force that can shape oceanographic features (e.g. tides) and biological processes such as reproductive cycles is the moon (Sponaugle & Cowen,

1994, 1996). Correlations to lunar cycles might bias seasonal patterns if these seasonal patterns refer to the solar calendar. Hence, the moon may induce anomalies and shifts in peaks of recruitment or spawning from one month to another instead. In some years, for example, the preferred new moon for recruitment might fall into the month of September (late September), while the next year it might be in October (early October), a consequence solely due to the shorter duration of the lunar cycles relatively to the solar months. This might explain for instance some of the variation reported in shifts of spawning aggregations of

Hipposcarus harid parrotfishes in the southern Red Sea in which a consistent aggregation in March had one exceptional year in which it occurred in April instead

(Gladstone, 1996). In fact, similar corrections for differences between the lunar and solar calendars have been observed in Palauan fishes and in the rising of Palolo worms in Samoa and Fiji (Johannes, 1981). Thus, some anomalies in the seasonality

100 of biological processes might not even be related to fluxes in environmental, oceanographic and/or physical factors at all, but are more simply dictated by the lunar regime. Daily fluctuations in numbers of new recruits have also been reported to occur both on lunar cycles (Randall, 1961; McFarland & Ogden, 1985) as well as independent of lunar cycles (Williams, 1983). This study’s sampling occurred during the new moon of each month, which fortunately happened around the middle of the month, especially for the months during which we measured peaks of recruitment

(September to December). Thus, it would be particularly interesting to assess how these recruitment peaks might change when the lunar cycles shift toward either the very beginning or end of the months. Temperatures are monthly stable and follow the solar calendar. Hence, it would be very helpful to record lunar recruitment during different years and measure whether a shift in the lunar cycle leads towards recruitment during the later new moon of the season or the earlier one, giving us further insight into the preferences of recruiting larvae and how strongly they may be impacted by the relatively high SST of the Red Sea. Alternatively, their abundance may be more strongly related to other factors such as food availability, solar cycles, or lunar regimes. All in all, spanning a longer time series is very much needed and it might also be recommendable to decide whether one chooses the lunar rather than the solar calendar when it comes to modeling biological responses related to reproductive processes as opposed to the parameters included in the present models.

101

2.4.2 Modeling of Environmental Variables

Along the same lines, a major issue is the difficulty in discriminating between physical-oceanographic and environmental (biotic and abiotic) variables to determine and model which parameters are driving biological responses such as total biomass (TBM), fish biomass (FBM), and specific fish abundance (SFA). Even though all parameters implemented in the models of this study correlated with statistical significance, many of them are naturally linked. For example, weather conditions can directly influence surface current speed and direction and therefore also turbidity. If you include the month as a regressor as well as temperature and

CHLA content even though there is no severe statistical correlation among these in the data, it is reasonable to suppose that they show seasonal co-variations and thus their fluctuation is also co-linked to the month of sampling, another parameter implemented in our models. This makes it difficult to decide which parameters to include in a single model and how to interpret the results of very complex interaction. Therefore, I decided to choose the parameters for the initial models solely using a mathematical approach and include all parameters that were not statistically over 30% correlated (after exchange with modelers; e.g. Fernando

Caugua; James Churchill; Sabrina Vettori), despite these being arguably naturally linked or triggering one another (i.e. co-varying). Then, the best-fitting models were examined with a more biological point of view to be able to scout logical patterns in biological responses. I concluded that the most important oceanographic features for the TBM are visibility and temperature. The significance of visibility on TBM

102 could be (1) due to the fact that the invertebrates in the plankton are highly attracted by the light from the light traps; thus the catches strongly depend on the visibility of the water as the light intensity and range of radiance will drop with increasing turbidity. (2) Visibility will also depend on the source of the water coming to the reef, which can carry more or less zooplankton depending on said source. The source however, is not necessarily strictly related to the main current direction (CD), additionally explaining why ultimately visibility was more relevant than CD when predicting TBM. The reason why the source of the incoming water and the CD might not be correlated is that measurements in CD could be biased by the reef’s topographic structure, especially at sheltered reef sites. For instance, if the waters are coming from an shelf-edge source (e.g. W or NW) rather than from shore

(E orNE) they may or may not carry more zooplankton and/or be more or less turbid, changing measured visibility. However, once the waters hit the reef the current direction could be distorted depending on its speed and measured to be coming from a different direction as the water is being pushed through the reef’s structure. In particular, on calm days and at sheltered reef sites even atmospheric processes and changes in water temperature (and density) could influence the measurements of the dynamics of currents. Thus, while current direction could be a major explanatory variable in TBM we might be missing the correlation due to biases in measurements. Furthermore, another response to winds and surface waters coming from shore can be an increased input of nutrients and dust from land, solely driven by wind strength and direction, and not necessarily linked to nor changing the CD. The particles brought by the winds can (3) immediately increase

103 turbidity due to increase particulates but also could show a retarded effect as the input of nutrients will eventually fuel primary production and increase the amount of plankton in the water. An increase in phytoplankton will additionally (4) increase zooplankton, resulting in increased TBM. One could also argue, though, that depending on how much visibility is lost due to the increasing particles in the water column the efficacy of the sampling method might also decrease with decreased visibility (due to increase particulates) and result in lower TBM. The data from my study does not allow for conclusive discrimination among the various drivers of the responses captured in the study. Moreover, some processes, such as changes in the winds, can be sporadic but fortunately seem mainly to be seasonal in the Red Sea, reducing biases due to anomalies in environmental conditions and allowing for smoother models. Remotely-sensed observations measure consistent seasonal blooms of phytoplankton as well as changes in temperature, which might explain why both are seemingly important explanatory variables for seasonality of TBM in comparison to other less stable but still arguably very important parameters mentioned above.

In general, when exploring the results of TBM and FBM per reef and the percentage of explanatory power of each of the variables, it appears that usually the regressors that is the most ‘variable’ (exhibiting even daily fluctuations) at the more exposed reef sites (shelf-edge and inshore sites), are also the ones that have the highest explanatory power. While in environmentally and oceanographically more stable and protected reef-sites (mid-shelf), the more ‘stable’ regressors, which rather change with season and not on a daily basis, become more significant. For instance

104 the most ‘oceanographically stable’ and protected site was the mid-shelf reef and at this site the TBM was mostly driven by temperature (seasonal and slow/gradual changes), pressure (had the lowest variance from all sites), and CHLA

(seasonal/gradual change). The TBM of the shelf-edge reef was mostly driven by lunar day (daily change), visibility (even hourly changes are possible), and pressure

(which had the highest variances shelf-edge, among all reefs). Accordingly, the inshore reef was a mix between seasonally driven parameters such as CHLA and visibility, which had the highest range of change among all sites at this inshore reef.

As from personal observations, visibility also seemed to be related to weather conditions (sea roughness) and the source of the water flowing through the reef

(captured by noticing the wave direction); water coming from shore appeared to have lower visibility, while waters coming from shelf-edge had higher visibility. The

TBM can also be directly linked to source and direction of the incoming water itself as shelf-edge waters may carry more plankton than waters from shore. Hence, seasonal changes in oceanographic features also drove the presence and abundance of plankton at the reef. In this regard, the more nektonic fish larvae (i.e., FBM), which have more complex swimming behaviors, seem to be responding to variables prone to more abrupt/sporadic changes such as current direction and weather conditions than to seasonal parameters. Current direction and weather conditions are possibly directly affecting swimming behavior and, consequentially, the timing that fish larvae choose to approach the reef for successful recruitment. Stronger currents and rough weather could additionally make it difficult for the larvae to enter the light trap as it becomes harder to swim in a precise direction (Doherty,

105

1987; Thorrold, 1992; Anderson & Thompson, 2002; Lindquist & Shaw, 2005); the light trap might also be moving more in nights of higher wave action, making it harder to enter. For the more planktonic invertebrates this may not play such a significant role as they are more likely more prone to be passively flushed into the light trap rather than intentionally swimming into or avoiding the openings of the trap. Additionally, the more complex biology of fish recruits could also be reflected in the over all lower amounts of variance that the models were able to explain in

FBM compared to TBM. Moreover, TBM was the most significant regressor when modeling FBM, which could also be evidence of the complexity of recruitment processes of fishes, as the parameter TBM is already a very complex process/variable of its own. It also suggests that the environmental drivers triggering TBM and FBM are likely similar and likely the triggers of the food cascade in the Red Sea (the FBM might be directly feeding on the copepods constituting most of the TBM).

When interpreting the results of SFA, models were explored for each site separately due to the differences in taxonomic composition between reefs. Those differences also mirror the relevance and uniqueness of reef-specific features and how these probably dictate most of the taxa-specific biodiversity of each site. As previously mentioned for the FBM, we find similar trends regarding the reef-specific explanatory regressors when modeling SFA: for the more exposed reefs, the more variable regressors are crucial like it is the current speed shelf-edge, and the visibility inshore (see also Bergenius and McCormick, 2005; and Munday et al.,

1997). Additionally, weather was also quite significant inshore, which could be

106 explained by the fact that recruits are coming from the pelagic and are being eaten as they approach the reefs and their abundance is gradually being reduced the closer they get to shore. Thus, as soon as you get rougher weather, not only do the larvae have a harder time swimming but also their predators will have a harder time trying to catch them and/or will prefer to remain in their shelters (saving energy in highly oligotrophic waters), resulting in higher survival of larvae allowing for higher numbers of recruits at far inshore reefs under said conditions. Additionally, rough weather could also be increasing the speed of the larvae approaching the reef from the shelf-edge source and so, reduce the amount they spend in the more dangerous transition zones (from the pelagic to the reef) and as they enter the so called “wall of mouths” (the high densities of hungry coral reef biota). The source of the incoming water may also dictate the species composition and abundance of the settling larvae.

While some reef fish families can complete their larval stage near reefs and in lagoons, others cannot and are only found in the oceanic pelagic (Leis, 1994). Thus, depending on the present direction of the currents, it may or may not provide a suitable route for the recruitment of some and not other fish species. Mid-shelf, the most powerful regressor was current direction, which again is the steadier regressor in the more protected/environmental stable reef.

When assessing the abundance of each taxon per reef (i.e. parrotfishes, other wrasses, gobies, blennies, cardinalfishes and damselfishes), I categorized the regressors into environmental and biological regressors (see Table 2.3, green and blue color coded). Inshore, the regressors explaining the abundance of the different

107 taxa seemed to be less environmentally determined and biological regressors had a higher overall representation in the final model. Furthermore, while TBM was imperative in all models, for some taxa at the mid-shelf reef it was not significant.

This might be due to the overall consistently high TBM at the mid-shelf site. The reef at the shelf-edge, the presence of one taxon seemed to be more strongly associated with the presence of other taxa, which might hint towards a common set of regressors dictating recruitment among all taxa at the more exposed and pelagic shelf-edge reef. In other words, all fish recruitments, regardless of the taxon may have to behave similar in the rougher pelagic waters and thus be more synchronized in response to oceanographic/physical conditions/processes at shelf-edge reefs.

This also appears to hold true at the inshore reef, at which strong oceanographic processes may be the main source of many larvae at reefs very close to shore, which might elsewise rather be the “last resort/destination for a recruiting pelagic larva”.

Nevertheless, it is very hard to find strong trends within the SFA and thus, the conclusions above should be used with caution. One reason for the low resolution within the models of SFA can be due to the conservative identification approach, in which many species are grouped together into one family (which is then the unit of modeling). Many fish families might have very species-specific recruitment patterns that cannot be resolved/modeled if clustered together with other congeners and con-familial species. If each species within that OTU has a variance of its own, the compilation of variances may mask underlying trends and therefore reduce the explanatory power of the models. Thus, modeling just one single species at a time could potentially have higher resolution and give models with higher accuracy and

108 later on allow depicting common trends among groups of species (which may be taxonomically greatly distinct but grouped by other biological and ecological features). Studies in the GBR have suggested taxonomic consistency in patterns of inter-annual larval supply (Milicich & Doherty, 1994), which also encourages a more taxonomically exhaustive exploration of our data and the promotion of time-series data collections for the Red Sea. The complex dynamics of oceanographic settings in our study and the relatively still small amount of data may obscure the relationship between seasonal environmental factors and species-specific patterns in early life history traits of fish larvae, and magnitude of recruitment events. Furthermore, studies like Sponaugle et al., (2006) and Bergenius et al., (2005), which also used a combination of environmental variables were able to explain 13 to 38% of the variation in larval growth and settlement magnitude of the a single acanthurid

Acanthurus chirurgus, which demonstrate that both transport- and growth-related processes can influence the magnitude of recruitment of tropical reef fishes also encouraging further exploration of our data. In general, the complexity of larval recruitment demands complex and increasingly discrete modeling approaches with more detailed data for higher resolution (see also suggestions in Staaterman and

Paris, 2014). For instance, the inclusion of presence and abundance of species- specific prey and predators as well as more site-specific current patterns and directions, taking into consideration the reef structure and vicinity, could be some of the parameters that could improve the explanatory power of models.

109

2.4.3 Biogeography and Biodiversity

Summing up all the aforementioned, larvae may be more likely present or actively target certain locations, where for example conditions in the plankton like food availability may be more favorable for the survival of larvae compared to other locations; and beyond that, habitat availability may also determine the presence and taxonomic distribution of coral-dwelling fishes (Menge & Olson, 1990; Munday,

2002) At a more localized and smaller spatial scale, other factors shaping the features of each reef might play a significant role. These are for example the exposure of the site to physical oceanographic conditions (Doherty, 1987; Thorrold,

1992; Anderson et al., 2002; Lindquist & Shaw, 2005) such as wave strength, turbidity, current speed and direction, etc. These features can vary at very small scales (spatially) even within a single reef and shape localized species composition

(e.g. Gobiodon species in Munday et al., (1997)) but also at bigger spatial scales such as along the cross-shelf gradient and then additionally, within a biogeographic province as it is the case along the environmental gradient in the Red Sea. These spatial gradients can drive patterns in recruitment abundance and taxonomic composition. Contrastingly, features with temporal fluxes/variation probably only influence recruitment abundance and less so biogeography. However, if these become too abnormal/strong and/or more frequent, dispersal and connectivity between populations may be hindered or change, and re-shape biogeographic ranges in the longer-term.

110

Another issue related to studying biogeography of species is to first catalog biodiversity, i.e. to know which species are where to begin with. When assessing the

SFA in our study the lack of sequence data and genetic barcodes available for coral reef fishes hindered the identification of many OTUs. For instance, we found evidence for a huge lack of genetically identified and/or described species, especially for gobies, which is seemingly one of the most biodiverse coral reef fish family. Alone in the central Red Sea we expect to have 100+ undescribed OTUs in the Gobiidae family, for which it would have been impossible to identified most of the specimen within our samples without our Red Sea specific data bank (COI library) provided by DiBattista et al. (in Press.). Thus, the publication of our sequences and addition of barcodes to the data banks together with the identification method presented in this study will ease the cataloging of coral reef fishes even if highly cryptic and/or morphologically tedious to identify due to its life-stage.

2.4.4 The Red Sea

Coral reefs are notably extremely complex systems. Nonetheless, the Red Sea might provide a perfect ground to in situ discretely examine the influence of a variety of combinations of environmental parameters on its reef biota and complexity. It naturally displays an environmental gradient along its basin from north to south, which allows for the aforementioned type of research. Temperature, salinity, nutrient availability and thus, primary production as well as current regimes, reef-

111 scapes, and habitat availability change gradually, providing a wide set of reef-types to study taxonomic specific responses to different environments and ideally help understanding which are the most significant drivers of distribution ranges and adaptive responses of coral reefs species (see also Chapter II and Robitzch et al.

2015). Furthermore, (1) The Red Sea is enclosed, quite isolated, and shows little fluctuations in its environmental conditions throughout the years. This allows for higher confidence in the inference of trends for example in seasonality of recruitment, even with only little snap shot data and/or data from short time-series in comparison to less environmentally stable tropical seas. For instance, the Red Sea might also be less susceptible to powerful El Niño events like the one in 1997, which in Florida caused the entire shift of recruitment of Lutjanus griseus all the way from summer to winter, probably due to abnormal climatic conditions and rainfall

(Allman & Grimes, 2002). (2) The RS displays a natural and quite stable environmental gradient of salinity, SST, CHLA/productivity, and even bathymetry and reef structure from north to south. The influence of such parameters at large scales are still far from being understood (Thresher, 1991)but in the RS, the influence of each of these parameters could be assessed in situ at different levels in relatively comparable habitats of similar species composition (i.e. in a single biogeographic province). Comparative and comprehensive studies (using similar methodology, sampling techniques, and sites/reef types) along the longitudinal axis and latitudinal gradient of the RS could provide better understanding of the influence of each of those parameters on the recruitment of fish larvae and their importance in correlation to the presence of seasonality as an adaptive strategy and

112 the consequences these responses can have for the species biogeographic range. (3)

The Red Sea is a unique biogeographic region of its own and amongst the most biodiverse coral reef ecosystems in the world, from which several species occur along its entire basin. Hence, the assessment of recruitment of those RS omnipresent species at each of the environmentally distinct RS regions (as described by Raitsos et al., (2013) and partitioned in e.g. Gulf of Aqaba, Northern RS, North-Central RS,

South-Central RS, and Southern RS; see also Robitzch et al., (2015)), will enlighten which of the parameters unique to a region are more decisive for successful recruitment and whether these are taxa specific or for instance more specific to the adaptive strategies of single species (see also Chapter II and Robitzch et al., (2015)).

For example, if it is the case that temperature is most significant, then the fishes in the much cooler Gulf of Aqaba will rather display recruiting peaks in the months with SST highs, similar to those in other tropical seas (with results alike those of the

GBR). While when assessing recruitment in the hottest region of the southern RS the peaks might be even further shifted towards the more bearable SST in the regional winter and early spring. Similarly, one could approach the influence of food and habitat availability and other site-specific features along the gradient; and that for an extraordinarily wide range of taxa. The RS is not only known to be extremely hot and species rich, but also extremely oligotrophic, especially at higher latitudes as its major and most constant source of nutrients is the incoming water from the Gulf of

Aden in its southernmost region. For example, in Robitzch et al. (2015) PLDs within one species showed significant differences between regions along the RS basin probably also due to this gradient in productivity. Thus, one could also expect

113 similar regional differences in other biological processes such as recruitment patterns (FBM) and larval behavior (correlations in the models), especially in the extremes of the RS (GAQ/northern vs. southern RS). Taxa specific physiological tolerances may also determine patterns of abundance (SFA) among regions (Menge

& Olson, 1990). Another interesting fact is that as we move south to the extremes in environments of the RS, there is a major environmental break/barrier set by said differences between the southern RS and the much cooler Gulf of Aden. The geographic features of the very narrow and shallow connection between the two regions, the straight of Bab El Mandab, already presumably isolate those two significantly. However, the additional environmental differences may lead to even further restricted geneflow if one assumes that species in the Gulf of Aden might display recruitment peaks in warmer and regionally more beneficial summer months, while peaks in recruitment in the southern RS may even more radically be shifted to cooler winter and spring months, which may cause reproductive isolation or at least restrict geneflow. Furthermore, as previously mentioned, the current flow patterns and water exchange between the two areas are also different between summer and winter and flowing in the “wrong direction” as to assist/benefit geneflow: assuming that RS biota rather reproduce in winter, when the surface water layer at Bab el Mandab is flowing into the RS, while in summer, when the Gulf of Aden specimen are presumably in their reproductive highs, the surface waters are rather flowing out of the Red Sea into the Gulf of Aden. Thus, we believe that a shift in the peaks of reproductive success in the Red Sea in comparison to other coral reef ecosystems worldwide could lead to further isolation of this

114 biogeographic province and its biota. However, the uniqueness in biodiversity we are recording nowadays in the Red Sea is potentially already the result of such isolation, which probably has even been higher in the past, and will decrease as the

Red Sea continues to open in the south and environmental conditions between the

Gulf of Aden and the RS become more homogeneous. All in all, if such shifts and biological responses are generally quickly triggered by changes in the environment thanks to fast adaptive strategies, and then lead to significant differences in reproductive strategies between two populations, geneflow and biogeographic distributions of coral reef fishes may be more plastic than we expect. Current abrupt and fast changes in climate will show how fast some species may be able to adapt and how plastic biogeographic ranges really are. Therefore a continuous monitoring of said biological adaptive responses and the conduction of similar studies over longer time scales and different biogeographic provinces are highly valuable and may help predict the fate of coral reef ecosystems and biodiversity to help find strategies that ultimately lower disruptive consequences of anthropogenic induced changes. One could for example identify regions more likely to become sources or refugia of biodiversity through time and protect these more intensively.

115

2.5 REFERENCES

Abu El-Regal M. (2013) Spawning seasons, spawning grounds and nursery grounds of some Red Sea fishes. The Global Journal of Fisheries and Aquatic Research, 6, 105–125. Allman R.J.. & Grimes C.B. (2002) Temporal and spatial dynamics of spawning, settlement, and growth of gray snapper (Lutjanus griseus) from the West Florida shelf as determined from otolith microstructures. Fishery Bulletin, 100, 391–403. Altschul S.F., Gish W., Miller W., Myers E.W., & Lipman D.J. (1990) Basic Local Alignment Search Tool. Journal of Molecular Biology, 215, 403–410. Anderson E.C. & Thompson E.A. (2002) A Model-Based Method for Identifying Species Hybrids Using Multilocus Genetic Data. Genetics, 160, 1217–1229. Anderson T.W., Bartels E., Carr M.H., Bartels C.T., Shenker J.M., & Hixon M.A. (2002) Current velocity and catch efficiency in sampling settlement-stage larvae of coral-reef fishes. Fishery Bulletin, 100, 404–413. Barlett M.S. (1937) Properties of Sufficiency and Statistical Tests. Proceeding of the Royal Society of London Series A, Mathematical and Physical, 160, 268–282. Barry J.P., Baxter C.H., Sagarin R.D., & Gilman S.E. (1995) Climate-related, long-term faunal changes in a California rocky intertidal community. Science, 267, 672– 675. Bell J.D., Westoby M., & Steffe A.S. (1987) Fish larvae settling in seagrass: do they discriminate between beds of different leaf density? Journal of Experimental Marine Biology and Ecology, 111, 133–144. Bellwood D.R. & Choat J.H. (1989) A Description of the Juvenile Phase Coiour Patterns of 24 Parrotfish Species (Family Scaridae) from the Great Barrier Reef, Australia. Records of the Australian Museum, 41, 1–41. Ben-Tzvi O., Kiflawi M., Gildor H., & Abelson A. (2007) Possible effects of downwelling on the recruitment of coral reef fishes to the Eilat (Red Sea) coral reefs. Limnology and Oceanography, 52, 2618–2628. Benayahu Y. & Loya Y. (1983) Surface Brooding in the Red Sea Soft Coral Parerythropodium fulvum fulvum (Forskål, 1775). Biological Bulletin, 165, 353– 369. Bergenius M.A.J. & McCormick M. (2005) Environmental influences on larval duration, growth and magnitude of settlement of a coral reef fish. Marine Biology. Berumen M.L., Hoey A.S., Bass W.H., Bouwmeester J., Catania D., Cochran J.E.M., Khalil M.T., Miyake S., Mughal M.R., Spaet J.L.Y., & Saenz-Agudelo P. (2013) The status of coral reef ecology research in the Red Sea. Coral Reefs, 32, 737–748.

116

Biro P.A., Post J.R., & Booth D.J. (2007) Mechanisms for climate-induced mortality of fish populations in whole-lake experiments. PNAS, 104, 9715–9719. Boehlert G.W. (1996) Larval dispersal and survival in tropical reef fishes. Reef Fisheries (ed. by N.V.C. Polunin and C.M. Roberts), Springer Science and Business Media Dordrecht. Bouwmeester J., Berumen M.L., & Baird A.H. (2011) Daytime broadcast spawning of Pocillopora verrucosa on coral reefs of the central Red Sea. Galaxea, Journal of Coral Reef Studies, 13, 23–24. Bouwmeester J., Gatins R., Giles E.C., Sinclair-Taylor T.H., & Berumen M.L. (2016) Spawning of coral reef invertebrates and a second spawning season for scleractinian corals in the central Red Sea. Invertebrate Biology, 1–12. Brothers E.B., Williams D.M., & Sale P.F. (1983) Length of larval life in twelve families of fishes at “One Tree Lagoon”, Great Barrier Reef, Australia. Marine Biology, 76, 319–324. Caballero-Arango D., Brulé T., Nóh-Quiñones V., Colás-Marrufo T., & Pérez-D’\iaz E. (2013) Reproductive biology of the tiger grouper in the southern Gulf of Mexico. Transactions of the American Fisheries Society, 142, 282–299. Caselle J.E. & Warner R.R. (1996) Variability in Recruitment of Coral Reef Fishes : The Importance of Habitat at Two Spatial Scales. Ecology, 77, 2488–2504. Cowen R.K. (2002) Oceanographic Influences on Larval Dispersal and Retention and Their Consequences for Population Connectivity. Coral Reef Fishes, 149–170. Cure K., Hobbs J.-P.A., & Harvey E.S. (2015) High recruitment associated with increased sea temperatures towards the southern range edge of a Western Australian endemic reef fish Choerodon rubescens (family Labridae). Environmental Biology of Fishes, 98, 1059–1067. Cushing D.H. (1973) Transfer of Organic Matter in Different Trophic Levels: Production in the Indian Ocean and the Transfer from the Primary to the Secondary Level. The Biology of the Idian Ocean Cushing D.H. (1990) Plankton Production and Year-class Strength in Fish Populations: an Update of the Match/Mismatch Hypothesis. Advances in Marine Biology pp. 249–293. Dahlberg M.D. (1979) A Review of Survival Rates of Fish Eggs and Larvae in Relation to Impact Assessments. Marine Fisheriess. Dixson D.L., Jones G.P., Munday P.L., Planes S., Pratchett M.S., Srinivasan M., Syms C., Thorrold S.R., & Bay S. (2008) Coral reef fish smell leaves to find island homes. Proceeding of the Royal Society Biology, 275, 2831–2839. Doherty P.J. (1983) Tropical Territorial Damselfishes: Is Density Limited by Aggression or Recruitment? Ecology, 64, 176–190. Doherty P.J. (1987) Light-traps: selective but useful devices for quantifying the distributions and abundances of larval fishes. Bulletin of Marine Science, 41,

117

423–431. Doherty P.J. & Fowler T. (1994) An empirical test of recruitment limitation in a coral reef fish. Science (New York, N.Y.), 263, 935–939. Doherty P.J. & Williams D.M. (1988) The replenishment of coral reef fish populations. Oceanogr Mar Biol Annu Rev, 26, 551. Dragovich A. & Potthoff T. (1972) Comparative study of food of skipjack and yellowfin tunas off the coast of west Africa. Fishery Bulletin, 70, 1087–1110. Fairclough D. (2005) The biology of four tuskfish species (Choerodon: Labridae) in Western Australia. Murdoch University, Figueira W.F., Biro P., Booth D.J., & Valenzuela V.C. (2009) Performance of tropical fish recruiting to temperate habitats: role of ambient temperature and implications of climate change. Marine Ecology Progress Series, 384, 231–239. Fine M., Gildor H., & Genin A. (2013) A coral reef refuge in the Red Sea. Global Change Biologyhange Biology, 19, 3640–3647. Fontes J., Afonso P., Santos R.S., & Caselle J.E. (2010) Temporal variability of larval growth, size, stage duration and recruitment of a wrasse, Coris julis (Pisces: Labridae), from the Azores. Scientia Marina, 74, 721–729. Fowler A.J., Doherty P.J., & Williams D.M. (1992) Multi-scale analysis of recruitment of a coral reef fish on the Great Barrier Reef. Marine Ecology Progress Series, 82, 131–141. Froukh T.J., Khalaf M.A., & Disi A.M. (2001) Studies On Taxonomy And Ecology Of Some Fish Larvae From The Gulf Of Aqaba. University of Jordan, Gladstone W. (1996) Unique Annual Aggregation of Longnose Parrotfish (Hipposcarus harid) at Farasan Island (Saudi Arabia, Red Sea). Copeia, 2, 483– 485. Hamner W.M., Jones M.S., Carleton J.H., Hauri I.R., & Williams D.M. (1988) Zooplankton, planktivorous fish, and water currents on a windward reef face: Great Barrier Reef, Australia. Bulletin of Marine Science, 42, 459–479. Hjort J. (1914) Fluctuations in the great fisheries of northern Europe. Hobson E.S. & Chess J.R. (1978) Trophic relationships among fishes and plankton in the lagoon at Eneweak Atoll, Marshall Islands. Fishery Bulletin, 76, 133–153. Jin D., Kite-Powell H., Hoagland P., & Solow A. (2012) A Bioeconomic Analysis of Traditional Fisheries in the Red Sea. Marine Resource Economics, 27, 137–148. Johannes R.E. (1978) Reproductive strategies of coastal marine fishes in the tropics. Environmental Biology of Fishes, 3, 65–84. Johannes R.E. (1981) Words of the lagoon: fishing and marine lore in the Palau district of Micronesia. Univ of California Press. Keener P., Johnson G.D., Stender B.W., Brothers E.B., & Beatty H.R. (1988) Ingress of

118

postlarval gag, Mycteroperca microlepis (Pisces: Serranidae), through a South Carolina Barrier Island inlet. Bulletin of Marine Science, 42, 376–396. Kimball M.E., Miller J.M., Whitfield P.E., & Hare J.A. (2004) Thermal tolerance and potential distribution of invasive lionfish (Pterois volitans/miles complex) on the east coast of the United States. Marine Ecology Progress Series, 283, 269– 278. Kramarsky-Winter E. & Loya Y. (1998) Reproductive strategies of two fungiid corals from the northern Red Sea:environmental constraints? Marine Ecology Progress Series, 174, 175–182. Lasker R. (1975) Field criteria for survival of anchovy larvae: the relation between inshore chlorophyll maximum layers and successful first feeding. Fishery Bulletin, 73, 453–462. Leis J. (1991) Introduction. Dynamics of Reef Fish Populations pp. 121–122. Leis J.M. (1994) Coral sea atoll lagoons: closed nurseries for the larvae of a few coral reef fishes. Bulletin of Marine Science, 54, 206–227. Letourneur Y. (1996) Dynamics of fish communities on Reunion fringing reefs, Indian Ocean. I. Patterns of spatial distribution. Journal of Experimental Marine Biology and Ecology, 195, 1–30. Lindeman K.C. (1989) Coastal construction, larval settlement and early juvenile habitat use in grunts, snappers and other coastal fishes of southeast Florida. Bulletin of Marine Science, 44, 1068. Lindquist D.C. & Shaw R.F. (2005) Effects of current speed and turbidity on stationary light-trap catches of larval and juvenile fishes. Fishery Bulletin, 103, 438–444. Longhurst A. (1995) Seasonal cycles of pelagic production and consumption. Progress in Oceanography, 36, 77–167. Longhurst A.R. & Harrison W.G. (1989) The biological pump: Profiles of plankton production and consumption in the upper ocean. Progress in Oceanography, 22, 47–123. Luckhurst B.E. & Luckhurst K. (1977) Recruitment patterns of coral reef fishes on the fringing reef of Curacao, Netherlands Antilles. Canadian Journal of Zoology, 55, 681–689. 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. PNAS, 110, 16498–16502. McFarland W.N. & Ogden J.C. (1985) Recruitment of young coral reef fishes from the plankton. 3, 37–51. McGurk M.D. (1986) larvae: role of spatial patchiness. Marine Ecology, 34, 227–242. McMahon K.W., Berumen M.L., Mateo I., Elsdon T.S., & Thorrold S.R. (2011) Carbon

119

isotopes in otolith amino acids identify residency of juvenile snapper (Family: Lutjanidae) in coastal nurseries. Coral Reefs, 30, 1135–1145. Meekan M.M.G., Milicich M.M.J., Doherty P.P.J., & others (1993) Larval production drives temporal patterns of larval supply and recruitment of a coral reef damselfish. Marine Ecology Progress Series, 93, 217–225. Menge B.A. & Olson A.M. (1990) Role of Scale and Environmental Factors in Regulation of Community Structure. Trends in Ecology & Evolution, 5, 52–57. Milicich M.J. & Doherty P.J. (1994) Larval supply of coral reef fish populations: magnitude and synchrony of replenishment to Lizard Island, Great Barrier Reef. Marine Ecology Progress Series, 110, 121–134. Milicich M.J., Meekan M.G., & Doherty P.J. (1992) Larval supply: a good predictor of recruitment of three species of reef fish (Pomacentridae). Marine Ecology Progress Series, 86, 153–166. Miller M.W., Weil E., & Szmant A.M. (2000) Coral recruitment and juvenile mortality as structuring factors for reef benthic communities in Biscayne National Park, USA. Coral Reefs, 115–123. Munday P., Jones G.P., & Caley M.J. (1997) Habitat specialisation and the distribution and abundance of coral-dwelling gobies. Marine Ecology Progress Series, 152, 227–239. Munday P.L. (2002) Does habitat availability determine geographical-scale abundances of coral-dwelling fishes? Coral Reefs, 21, 105–116. Munday P.L., Jones G.P., Pratchett M.S., & Williams A.J. (2008) Climate change and the future for coral reef fishes. Fish and Fisheries, 9, 261–285. Murray S.P. & Johns W.E. (1997) Direct observations of seasonal exchange through the Bab el Mandab Strait. Geophysical Research Letters, 24, 2557–2560. Nakano K., Takemura A., Nakamura S., Nakano Y., & Iwama G.K. (2004) Changes in the cellular and organismal stress responses of the subtropical fish, the Indo- Pacific sergeant, Abudefduf vaigiensis, due to the 1997 – 1998 El Nino/Southern Oscillation. Environmental Biology of Fishes, 70, 321–329. Nanninga G.B., Saenz-Agudelo P., Zhan P., Hoteit I., & Berumen M.L. (2015) Not finding Nemo: limited reef-scale retention in a coral reef fish. Coral Reefs. Planes S., Bonhomme F., & Galzin R. (1993) Genetic structure of Dascyllus aruanus populations in French Polynesia. Marine Biology, 117, 665–674. Poertner H.O. & Peck M.A. (2010) Climate change effects on fishes and fisheries: Towards a cause-and-effect understanding. Journal of Fish Biology, 1–35. Qasim S.Z. (1956) Time and Duration of the Spawning Season in some Marine Teleosts in Relation to their Distribution. University College of North Wales, Bangor. Racault M.-F., Raitsos D.E., Berumen M.L., Brewin R.J.W., Platt T., Sathyendranath S., & Hoteit I. (2015) Phytoplankton phenology indices in coral reef ecosystems:

120

Application to ocean-color observations in the Red Sea. Remote Sensing of Environment, 160, 222–234. Raitsos D.E., Pradhan Y., Brewin R.J.W., Stenchikov G., & Hoteit I. (2013) Remote sensing the phytoplankton seasonal succession of the Red Sea. PloS one, 8, e64909. Randall J.E. (1961) A contribution to the biology of the convia surgeonfish of the Hawaiian Islands, Acanthurus triostegus sandvicensis. Pacific Science, 15, 215– 272. Rankin T.L. & Sponaugle S. (2011) Temperature influences selective mortality during the early life stages of a coral reef fish. PLoS ONE, 6, e16814. Ratnasingham S. & Hebert P.D.N. (2013) A DNA-Based Registry for All Animal Species: The Barcode Index Number (BIN) System. PloS one, 8. Roberts C.M.C., Shepherd A.A.R.D.A., & Ormond R.R.F.G. (1992) Large-scale variation in assemblage structure of Red Sea butterflyfishes and angelfishes. Journal of Biogeography, 19, 239–250. Robertson A.I. & Blaber S.J.M. (1993) Plankton, epibenthos and fish communities. Tropical mangrove ecosystems, 173–224. Robertson D.R., Green D.G., & Victor B.C. (1988) Temporal Coupling Of Production And Recruitment Of Larvae Of A Caribbean Reef Fish. Ecology, 69, 370–381. Robertson D.R., Petersen C.W., & Brawn J.D. (1990) Lunar reproductive cycles of benthic-brooding reef fishes: reflections of larval biology or adult biology? Ecological Monographs, 60, 311–329. Robertson D.R., Schober U.M., & Brawn J.D. (1993) Comparative varition in spawning output and juvenile recruitment of some Caribbean reef fishes. Marine Ecology Progress Series, 94, 105–113. Robertson D.R., Swearer S.E., Kaufmann K., & Brothers E.B. (1999) Settlement vs . Environmental Dynamics in a Pelagic-Spawning Reef Fish at Caribbean Panama. Ecological Monographs, 69, 195–218. Robitzch V.S.N., Lozano-Cortés D., Kandler N.M., Salas E., & Berumen M.L. (2015) Productivity and sea surface temperature are correlated with the pelagic larval duration of damsel fi shes in the Red Sea. Marine Pollution Bulletin. Roik A., Roethig T., Roder C., Ziegler M., Kremb S.G., & Voolstra C.R. (2016) Year- Long Monitoring of Physico-Chemical and Biological Variables Provide a Comparative Baseline of Coral Reef Functioning in the Central Red Sea. PloS one, 1–34. Russ G.R. & Abesamis R.A. (2010) Patterns of recruitment of coral reef fishes in a monsoonal environment. Coral Reefs, 29, 911–921. Russell B.C.. C., Anderson G.R. V., & Talbot F.H. (1997) Seasonality and Recruitment of Coral Reef Fishes. Australian Journal of Marine and Freshwater Research, 28, 521–528.

121

Russell B.C., Anderson G.R. V, & Talbot F.H. (1977) Seasonality and recruitment of coral reef fishes. Marine and Freshwater Research, 28, 521–528. Russell B.C., Talbot F.H., & Domm S. (1974) Patterns of colonization of artificial reefs by coral reef fishes. 1, 207–215. Sale P.F. (2004) Connectivity, recruitment variation, and the structure of reef fish communities. Integrative and Comparative Biology, 44, 390–399. Schloss P.D., Westcott S.L., Ryabin T., Hall J.R., Hartmann M., Hollister E.B., Lesniewski R.A., Oakley B.B., Parks D.H., Robinson C.J., Sahl J.W., Stres B., Thallinger G.G., Horn D.J. Van, & Weber C.F. (2009) Introducing mothur: Open- Source , Platform-Independent, Community-Supported Software for Describing and Comparing Microbial Communities. Applied And Environmental Microbiology, 75, 7537–7541. Schmitt R.J. & Holbrook S.J. (2000) Habitat-Limited Recruitment of Coral Reef Damselfish. Ecology, 81, 3479–3494. Schroeder R.E. (1987) Effects of patch reef size and isolation on coral reef fish recruitment. Bulletin of Entomological Research, 41, 441–451. Sheppard C.R.C. & Sheppard A.L.S. (1991) Corals and coral communities of Arabia. Siddall M., Rohling E.J., Almogi-Labin A., Hemleben C., Meischner D., Schmeltzer I., & Smeed D.A. (2003) Sea-level fluctuations during the last glacial cycle. Nature, 423, 853–858. Smeed D.A. (2004) Exchange through the Bab el Mandab. Deep Sea Research II, 51, 455–474. Sofianos S.S., Johns W.E., & Murray S.P. (2002) Heat and freshwater budgets in the Red Sea from direct observations at Bab el Mandeb. Deep Sea Research, 49, 1323–1340. Spaet J. (2013) Predictable annual aggregation of longnose parrotfish (Hipposcarus harid) in the Red Sea. Marine Biodiversity, . Spaet J.L.Y. & Berumen M.L. (2015) Fish market surveys indicate unsustainable elasmobranch fisheries in the Saudi Arabian Red Sea. Fisheries Research, 161, 356–364. Spaet J.L.Y., Nanninga G.B., & Berumen M.L. (2016) Ongoing decline of shark populations in the Eastern Red Sea. Biological Conservation, 201, 20–28. Sponaugle S., Boulay J., & Rankin T. (2011) Growth- and size-selective mortality in pelagic-larvae of a common reef fish. Aquatic Biology, 13, 263–273. Sponaugle S. & Cowen R.K. (1994) Larval durations and recruitment patterns of two Caribbean gobies (Gobiidae): contrasting early life histories in demersal spawners. Marine Biology, 120, 133–143. Sponaugle S. & Cowen R.K. (1996) Nearshore patterns of coral reef fish larval supply to Barbados, West Indies. Marine Ecology Progress Series, 133, 13–28.

122

Sponaugle S. & Cowen R.K. (1997) Early Life History Traits and Recruitment Patterns of Caribbean Wrasses (Labridae). Ecological Monographs, 67, 177– 202. Sponaugle S. & Grorud-Colvert K. (2006) Environmental variability, early life- history traits, and survival of new coral reef fish recruits. Integrative and Comparative Biology, 46, 623–633. Sponaugle S., Grorud-Colvert K., & Pinkard D. (2006) Temperature-mediated variation in early life history traits and recruitment success of the coral reef fish Thalassoma bifasciatum in the Florida Keys. Marine Ecology Progress Series, 308, 1–15. Sponaugle S., Paris C., Walter K.D., Kourafalou V., & D’Alessandro E.K. (2012) Observed and modeled larval settlement of a reef fish to the Florida Keys. Marine Ecology Progress Series, 453, 201–212. Srinivasan M. & Jones G.P. (2006) Extended breeding and recruitment periods of fishes on a low latitude coral reef. Coral Reefs, 25, 673–682. Staaterman E. & Paris C.B. (2014) Modelling larval fish navigation: the way forward. ICES Journal of Marine Science, 71, 918–924. Stevens G.C. (1996) Extending Rapoport’ s Rule to Pacific Marine Fishes. Journal of Biogeography, 23, 149–154. Talbot F.H., Russell B.C., & Anderson G.R. V (1978) Coral Reef Fish Communities: Unstable, High-Diversity Systems? Ecological Monographs, 425–440. Thorrold S.R. (1992) Evaluating the performance of light traps for sampling small fish and squid in open waters of the central Great Barrier Reef lagoon. Marine Ecology Progress Series, 89, 277–285. Thresher R.E. (1991) Geographic variability in the ecology of coral reef fishes: evidence, evolution, and possible implications. The ecology of fishes on coral reefs, 401–436. Victor B.C. (1986) Delayed Metamorphosis with Reduced Larval Growth in a Coral Reed Fish (Thalassoma bifasciatum). Canadian Journal of Fisheries and Aquatic Sciences, 43, 1208–1213. Wellington G.M. & Victor B.C. (1992) Regional differences in duration of the planktonic larval stage of reef fishes in the eastern Pacific Ocean. Marine Biology, 113, 491–498. Williams D.M. (1980) Dynamics of the pomacentrid community on small patch reefs in One Tree Lagoon (Great Barrier Reef). Bulletin of Marine Science, 30, 159– 170. Williams D.M., English S., & Milicich M.J. (1994) Annual recruitment surveys of coral reef fishes are good indicators of patterns of settlement. Bulletin of Marine Science, 54, 314–331. Williams Dm. (1983) Daily, Monthly and Yearly Variability in Recruitment of a Guild

123

of Coral Reef Fishes. Marine Ecology Progress Series, 10, 231–237. Yao F., Hoteit I., Pratt L.J., Bower A.S., Zhai P., Koehl A., & Gopalakrishnan G. (2014) Seasonal overturning circulation in the Red Sea: 1. Model validation and summer circulation. Journal of Geophysical Research: Oceans, 2238–2262.

124

CHAPTER III

GENETICS, BIOLOGY, AND ECOLOGY OF THE ENDEMIC DASCYLLUS

MARGINATUS AND THE COSMOPOLITAN D. ARUANUS DAMSELFISHES IN VS.

OUTSIDE THE RED SEA: DEFINING BIOGEOGRAPHIC RANGES1

3.1 INTRODUCTION

Many coral reef fish species occupy geographic ranges that have inconsistent and diverse geographical and environmental features. Both are crucial factors for gene flow and the maintenance or diversification of species across time and space. In marine environments, narrow connections or landmasses between oceanographic locations can be significant barriers for geneflow between the regions and lead to vicariance. In coral reef fishes, large trajectories of deep pelagic waters without shallow habitats between regions also isolate populations and may lead to differentiation of species due to isolation by distance (IBD; reviewed in Hellberg,

2007; Jones et al., 2009; Selkoe and Toonen, 2011). Environmental differences, for instance, imposed by current regimes, fluxes in temperature and water chemistry, and habitat composition may be hard to overcome, causing decreased dispersal and geneflow through processes of natural selection, too. This has more recently been used as a metric referred to as isolation by environment (IBE; e.g., Nanninga et al.,

2014; Selkoe et al., 2008; Wang and Bradburd, 2014). In this context, oceanographic features like upwelling or monsoonal driven climatic changes may act as drivers of

1 A portion of this chapter (sections 3.2.6, 3.3.4, and 3.4.3) has been published in Robitzch et al. (2015, Appendix 7.3) 125 selection for some coral reef fish species more than others even if closely related and with similar life histories (Bird et al., 2007; Gaither et al., 2010; Barber et al.,

2011; Carpenter et al., 2011; Dibattista et al., 2012; Fouquet et al., 2012).

Additionally, isolation by oceanography (IBO), which can be seen as a combination of IBE and IBD, does not necessarily follow linear patterns and varies at temporal and spatial scales. Therefore, while testing for IBO it is of importance to consider geography alongside of environmental parameters (e.g., Crispo et al., 2006;

Cushman et al., 2006; Lee and Mitchell-Olds, 2011; Nanninga et al., 2014; Wang,

2013; Wang and Summers, 2010).

IBD, IBE and/or IBO can cause limited gene flow, which eventually may lead to divergence via genetic drift (Wright, 1943; Slatkin, 1987; Riginos & Liggins, 2013).

Likewise, natural selection under differing ecological regimes can also lead to divergence and limited gene flow even if there is high dispersal between populations (Via, 2002; Nosil et al., 2009). Thus, connectivity and gene flow between populations depends on the ability of propagules of a species to overcome environmental differences, geographical barriers, and long distances (Wang &

Bradburd, 2014).

The Red Sea, which harbors thriving and continuous coral reefs along both sides of its narrow basin, is an ideal location to study the effect of a variety of environmental parameters on coral reefs (Berumen et al., 2013). Due to its shape, location, and only a single shallow and narrow connection to the Indian Ocean, it displays unique environmental gradients of temperature, salinity, productivity, and turbidity

(Raitsos et al., 2013; Racault et al., 2015). These physical and environmental 126 characteristics lead to a more or less gradual change in the reef structure or “reef- scape” along the Red Sea basin and a more abrupt environmental change from the

Red Sea into the Gulf of Aden. Together, they form the Arabian Sea. Its oceanographic, environmental, and geographical features present different types of barriers to marine biota, all in all, providing an excellent in situ lab for testing hypothesis on the previously three mentioned barriers to geneflow: IBD, IBE, and

IBO. The Red Sea has been recognized as a suitable natural location to assess how environmental characteristics affect coral reef biota (e.g., (Froukh and Kochzius,

2007; Giles et al., 2015; Lozano-Cortés and Berumen, 2016; Ngugi et al., 2012;

Roberts et al., 1992; Roberts et al., 2016; V. Robitzch et al., 2015; V. S. N. Robitzch et al., 2015; Sawall et al., 2014a, 2014b). It is also characterized for high levels of endemic species with either differing or still undefined exact biogeographic distributions, either restricted to the Red Sea or distributed around the Arabian

Peninsula. Such uncertainties on biogeographic ranges are usually due to the lack of morphological and genetic data to confirm the phylogeny of the species at different locations. Reports on cryptic species or morphological and phenotypic differences between populations of the same species often disguise accurate assignments of distribution ranges. More recent studies have thus, targeted such species with an apparent Arabian Sea distribution range, using molecular tools to assess genetic structure and putative evolutionary processes shaping biogeography of coral reef fishes (Saenz-agudelo et al., 2015; DiBattista et al., 2017) and defining distribution ranges. Herein, the main focus has been on three theoretical barriers. One, within the Red Sea (at the Farasan Banks) due to a strong shift from a deep and 127 oligotrophic northern Red Sea basin, to a relatively nutrient rich, turbid, shallower, and hotter southern Red Sea (IBE). Another, between the Red Sea and the Gulf of

Aden caused by the shallow and narrow straight of Bad El Mandab (IBD). And a third potential environmental barrier given by seasonal upwelling at the Yemeni

Coast, in the Gulf of Aden (IBO).

In this chapter, I target questions around the drivers of differing biogeographic ranges among closely related coral dwelling fishes with similar life history. I use the

Arabian Sea to comparatively test the effects of those three barriers on the distribution, dispersal, and genetic structure of a widespread versus and Arabian

Sea endemic damselfish. I further hypothesize the degree to which IBD, IBE, and IBO may explain the genetic patterns observed and whether these are likely related to natural selection, adaptation, and/or phenotypic plasticity. Lastly, I assess the effect of environmental conditions on the biology of the three Dascyllus damselfishes present in the Red Sea, using chlorophyll a (CHLA) and sea surface temperature

(SST) as environmental variables and the pelagic larval duration (PLD) of the congeneric species as the biological trait. I present these three species as a good model group because they are very similar, genetically and biologically, but have different distribution ranges and, what I refer to, as degree of ecological specialization.

The PLD of marine species has often been put in relation with their distribution range and dispersal potential (e.g. Bohonak, 1999; Doherty et al., 1995; Jablonski,

1986). However, correlations and results have been yet, ambiguous and in many cases contradictive (Weersing & Toonen, 2009). One issue masking trends in PLD 128 and for instance dispersal potential of fishes may be the plasticity in PLDs (Robitzch et al., 2015b). Intraspecific variations in larval growth and development may be due to temperature and food availability (Heath, 1992; Houde & Zastrow, 1993;

Takahashi & Watanabe, 2005; Mcleod et al., 2013; Robitzch et al., 2015b), which can be estimated from remote sensing satellite data on sea surface temperature (SST) and chlorophyll a (CHLA) concentrations respectively. However, most studies that have assessed the influence of temperature and food availability on the larval stage of fishes have focused on larval growth rate (Mcleod et al., 2015), size at settlement

(McCormick & Molony, 1995), or recruitment cohort size (Lo-Yat et al., 2011) rather than on the length of PLDs in the standing population, and still fewer of these studies target coral reef fishes. Hence, I specifically used measurements of PLD variations within the standing populations (i.e., post-settlement and adult fishes, as opposed to recruitment cohorts) of three congeneric coral reef damselfishes present in the Red Sea. These species represent a suitable model group to further assess relations between distribution ranges, biological traits, and genetic structures of widespread vs. endemic damselfishes and elucidate characteristics that may be shaping connectivity patterns and dispersal potential of the coral reef fishes species across biogeographic provinces.

The species model group is constituted by Dascyllus marginatus, D. aruanus and D. trimaculatus. The endemic species, D. marginatus, can be found inside the Red Sea along its entire basin, but mostly only at inshore and mid-shelf reefs. It can be additionally found at almost every reef in the shallower southern Red Sea and throughout the Gulf of Aden. The widespread congeneric D. aruanus, can be found 129 along the entire Red Sea, and the wide Indo-Pacifc. However, it is only sparsely present in the southern Red Sea and completely absent in the Gulf of Aden. Both of these species are biologically, morphologically, and ecologically very similar. They are zooplanktivorous, live in haremic size-based hierarchical protogynous colonies and dwell in branching corals on sandy protected patchy reefs and in lagoons, in which they also lay benthic brooding egg-clutches. Their PLD is also similar according to literature and with a duration of about three weeks, typical for damselfishes (Dataset S1, Luiz et al., 2013). The third and only other Dascyllus species present in the Red Sea, D. trimaculatus, was used to complete the species model but solely its PLD (not population genetics) was investigated. I consider D. trimaculatus as the most ecologically versatile species within this group. It can be found on exposed reef walls, sheltered backside-reefs, and lagoons; it is not dependent on live coral; and it only uses reef structure for shelter, recruitment, and as spawning habitat. While D. trimaculatus mostly recruits onto anemones in the

Red Sea, it can also recruit into branching corals, sea urchins, and other microhabitats. Once they reach juvenile stages, D. trimaculatus leave their settlement substrate, range throughout the entire reef structure, and spend most of their lifetime foraging in the water column. Intermediate on the ecological specialization ranking is D. aruanus. While D. aruanus and D. marginatus show similar habitat preferences and even co-inhabit the same coral colony, I suggest that the degree of ecological specialization is strongest in D. marginatus, which cannot be found on very shallow reef-tops and lagoons, nor can it be found on offshore reefs in most of the Red Sea. Thus, it represents the ecologically most specialized species of 130 the group. The Dascyllus species model also seems suitable to study variation in

PLDs in the context of biogeography of coral reef associated fishes. Of the three species that occur in the Red Sea, D. marginatus is an Arabian endemic while D. trimaculatus and D. aruanus are both widespread species, common in most of the

Indo-Pacific. For the assessment of genetic structure in the light of distribution range, data on single nucleotide polymorphism (SNPs) was generated to compare differences between two of the species: the widespread D. aruanus vs. the endemic

D. marginatus.

All in all, I will present 1) genetic structure of D. aruanus and D. marginatus inside the Red Sea and 2) along their entire distribution and in relation to their biogeography; 3) the influence of three potential barriers to geneflow in the Arabian

Sea on the two species; 4) influences of environmental parameters such as SST and

CHLA on the PLDs of D. aruanus and D. marginatus and D. trimaculatus (coral reef associated damselfishes with different degrees of ecological specialization) inside the Red Sea; and 5) variations in PLDs and their putative relation to biogeographic ranges among these three congeneric species.

131

3.2 MATERIALS AND METHODS

3.2.1 Study Region and Sample Collection of Dascyllus marginatus and D. aruanus for the Generation of Single Nucleotide Polymorphism (SNP) Libraries

30 30° D. aruanus IA

b4 D. marginatus

N ° IO

20 30

30° IA

10 10°

N ° IO 20

GAQ 10 D. aruanus 10° GAQGAQ

30° 30 NCRS NCRS ● SCRSSCRS D. marginatus SRSSRS

GAQGAQ

30° 30 NCRS

● NCRS 25 25° SCRSSCRS 2000 km

SRSSRS 25

25° 200080 km° 100°100 120120°

20 20°

80 ° 100°100 120120° 20 b1 20°

b3 CHLA [mg m-3] 15°

15 0.08 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2.5CHLA [mg 10 m-3] 30 15° b2 15 0.08 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2.5 10 30

200 km200 km

200 km 10°

10 200 km

10° 10 35° 40° 35 °45 ° 40 ° 50 ° 45 ° 50° Figure 3.1: Sampling sites inside and outside the Red Sea (RS) for the generation of genetic data on single nucleotide polymorphisms (SNPs) of Dascyllus aruanus and D. marginatus. Triangles represent those for the former and circles those for the latter species. Both have three sampling sites outside the Red Sea, while D. aruanus has one more (six) inside the Red Sea compared to D. marginatus (five). More information on the sampling sites can be found in Table 3.1. Chlorophyll a (CHLA, in mg m−3) concentrations of the Indo-Pacific and Red Sea are displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate environmental differences between locations. Black dashed lines represent potential barriers to dispersal and/or geneflow for coral reef species: 20 ̊N (b1), the Strait of Bab al Mandab (b2), Gulf of Aden and Oman upwelling (b3), and the land barrier of the Indo-Australian Archipelago (IA, b4) (b1-3, sensu Saenz-agudelo et al., (2015); b4, sensu Cowman and Bellwood, (2013)).

132

Between the years 2009 and 2015 D. marginatus samples were collected from nine sites along the species entire distribution range around the Arabian Peninsula, between the Gulf of Aqaba (GAQ) 28 ̊N, 34 ̊E) in the northern Red Sea (RS) and

Musandam (MUS, 12 ̊N, 43 ̊E) in the Persian Gulf, Oman (Figure 3.1). Five sites were selected inside the Red Sea, one of which is located directly at the entrance of the

Red Sea (in Djibouti, (DJI)), and three outside of the Red Sea. Congeneric D. aruanus samples were also collected during the same time period from nine sites along its biogeographic range: five inside the Red Sea, from the Gulf of Aqaba to Djibouti (at the stretch of Bab el Mandab), and four outside the Red Sea, from Madagascar

(MAD), the Parcel Islands (PAI, China), Taiwan (TAI), and Indonesia (IND; Figure

3.1). Whenever possible both species were sampled from the same reef or at the closest possible reef. In total, 160 samples of each species were used to create ddRAD libraries for the generation of SNPs data. Additionally, D. aruanus samples from an reef in the north central Red Sea (Shib Al Karrah) were included, as SNP data was available for those samples from chapter III.

3.2.2 Single Nucleotide Polymorphism (SNP) Library Preparation and Sequencing

Genomic DNA was extracted from fin or gill tissue preserved in 96% ethanol using a

Nucleospin-96 Tissue kit (Macherey-Nagel, Düren, Germany). Double digest restriction associated DNA (ddRAD) libraries were prepared using 500 ng of high quality DNA per sample following the protocol described by Peterson et al., (2012) 133 with some modifications. Briefly, the genomic DNA is digested at 37°C for three hours using the restriction enzymes SphI and MluCI (NEB) followed by a ligation step, where each sample is assigned to one of sixteen unique adaptors. Pools of sixteen individuals are combined and run on a 1.5% agarose gel, from which fragments of ~400 base-pairs (bp) are manually excised and purified using a

Zymoclean Gel DNA recovery kit. Each pool is then amplified adding a unique indexing primer for each pool according to the standard Illumina multiplexed sequencing protocol in a 50μl PCR reaction containing 25μl Kapa Hifi Hotstart Ready

Mix Taq, 20μl of pooled library DNA and 2.5μl of the universal Illumina PCR primer and an additional 2.5 μl of one of twelve unique indexing primer for each pool.

Amplifications are carried out in an Eppendorf 94-well vapo.protect Mastercycler

Pro System (Fisher Scientific) using the following protocol: initial step at 95 ̊C for 3 min, followed by ten cycles of 98 ̊C for 20 sec, 60 ̊C for 30 sec, and 72 ̊C for 30 sec, and a final step at 72 ̊C for 5 min. DNA libraries are then quantified using the high sensitivity DNA analysis kit in a 2100 Bioanalyzer (Agilent Technologies) and by running a qPCR in an ABI 7900HT fast real-time PCR system (Thermo Fisher

Scientific) using the KAPA Library Quantification Kits (Kapa Biosystems). One last time, the length (in bp) and quality of the library fragments is measured in the 2200

TapeStation (Agilent Technologies) using the High Sensitivity D1000 ScreenTape kit. Pools are subsequently combined in equimolar concentration to form a single genomic library. In total four libraries were created and run on four separate lanes of a HiSeq 2000 Illumina sequencer (each library and lane contained 80 individuals; two libraries for each, D. marginatus and D. aruanus; single end reads, 1 x 101 bp; v3 134 reagents).

3.2.3 De-novo Assembly

Sequences of 160 individuals per species (Dascyllus marginatus and D. aruanus) were demultiplexed and filtered for quality using the ‘process_radtags.pl’ pipeline in

STACKS version 1.42 (Catchen et al., 2011). Individual reads with phred-scores below 30 (average on sliding window) or with ambiguous barcodes were discarded.

After this, individuals for which less than 400 000 reads were retained were also discarded. For all remaining 141 D. marginatus and 139 plus 22 (from the central

Red Sea adult population of chapter III) D. aruanus specimens, RADSeq loci were assembled de-novo using the ‘denovo_map.pl’ pipeline in Stacks. Different parameter combinations were evaluated, which resulted in different numbers of loci but gave similar results in genetic comparisons (genetic clustering and pairwise FST among sites).

3.2.3.1 Dascyllus marginatus

For the main analyses, we used a parameter combination similar to the one recommended by Mastretta-Yanes et al., (2014): minimum read depth to create a stack (-m) = 3, number of mismatches allowed between loci within individuals (-M)

= 3, number of mismatches allowed between loci within catalog (-n) = 4. The default of (-n) = 1 but due to first trial analyses, which hinted towards high fixed loci and 135 high genetic differentiation between the populations of inside vs. outside the Red

Sea we decided to test different (-n) values, up to (-n) = 6, to avoid that same loci are treated as separate ones because they erroneously seem absent among individuals of different population (inside vs. outside the Red Sea) due to higher mutations between these two populations and as a consequence be excluded in case of low read depth/coverage or called as different alleles. Another hint towards such loss of loci due to higher number of mismatches between populations was the fact that in general a high percentage (~25%) of haplotypes were called from secondary reads, which allows per default mismatches of (M+2) = 5. In other words once the number of mismatches is increased, the number of retained loci increases as well. Therefore, it is better to increase (-n) in order to retain more loci but additionally use the (-H) option, which disables calling haplotypes from secondary reads to be conservative and avoid incoherent allele calling due to too many mismatches/mutations and calling same loci twice. Thus, combining the two options allows the match of the same loci between highly divergent populations and thus, the retention of higher number of more accurately called SNPs (instead of a posteriori recalling haplotypes from secondary reads using (M+2) = 5 mismatches). To further test and be positive that there is no critical bias in the data by increasing (-n) and implementing (-H), the number of total SNPs was assessed for the different values used for (-n) (2, 4, and 6) and mantel tests were performed among the different pairwise FSTs for with and without (-H). Indeed, the amount of SNPs in the data set increased with increasing (- n) and the mantel tests between the FSTs of the different parameter settings were all linear correlated and had similar average FSTs. Hence, there was little doubt for 136 unwanted biases. The final data set was produced using (-n) =4 in combination with

(-H), and (-t) to remove or break up highly repetitive RAD-Tags.

Following de-novo mapping, further data filtering was performed using the population component of STACKS. First, the minimum stack depth per individual was increased to (-m) = 6. Second, only those loci present in at least 85% of individuals were retained (-r) = 0.85. Third, all loci with minor allele frequencies lower than 0.05 were removed to reduce the number of false polymorphic loci due to sequencing error (--min_maf) = 0.05. Fourth, the maximum observed heterozygosity required to process a nucleotide site at a locus was set to 60% to filter loci out of Hardy-Weinberg-Equilibrium (HWE) (--max_obs_het) = 0.6. As a last step, the log likelihood (lnl) ratio of all loci was calculated using rxstacks and after looking at its distribution a log likelihood threshold of (--lnl_lim) = -5 was used to filter loci with lnl values below -5. The write_random_snp option produced a vcf file with only one single randomly chosen SNP per stack out of 1.04 Mio reads built into the catalog. The resulting vcf file was converted to other program specific input files using PGDSPIDER version 2.0.5.2 (Lischer & Excoffier, 2012) and GenoDive (Version

2.0b23) The final data set consisted of 4987 loci and 113 individuals with less than

20% missing data and loci missing in less than 15% of the population.

3.2.3.2 Dascyllus aruanus

For this species a similar approach as mentioned above was used and the same different parameter sets were tested. Therefore, a relatively high (-n) value and the 137

(-H) option were also chosen for D. aruanus, due to the wide biogeographic range of this species and the fact that several studies suggested strong population differentiation between individuals inside vs. outside the Red Sea, plus others even recommend the split of the species into two distinct lineages. The first data set consisted only of the libraries specifically designed for this chapter (without the

AKA reef). Here fore, denovomap.pl retained over 1 Mio reads in the catalog with a –

(-n) =2 parameter set. When including AKA less reads were retained (>891 t) using

(-n) = 4 but the resulting total number of SNPs in the data set was still very good.

Therefore after applying the same procedure as above, the final data set consisted of

113 individual with 7997 SNPs, using (-n) = 4 and (-H), meaning that all parameters were the same as for D. marginatus (see section above).

3.2.4 Population Genetics and Statistical Analyses

As it is crucial to have as many loci as possible but many programs cannot deal well with missing data (e.g. they omit the sample or loci for the whole analysis and/or the analysis in general becomes less accurate) I used GenoDive to fill in missing data. It uses the most common genotype within the population of that individual for which the alleles are missing. Hardy Weinberg equilibrium (HWE) and linkage disequilibrium (LD) were also tested for one site of each species using the R packages poppr, adegenet, and qvalue. Under 0.005% of the loci were out of HWE or in LD, probably due to the pre-filtering and parameter settings used in STACKS.

To address all aims of this study we performed a principal coordinate analysis 138

(PCoA) and a genetic clustering analysis. Both these analyses allowed elucidating the spatial patterns of genetic admixture along the distribution range of D. aruanus and D. marginatus and provided information about the levels of admixture between their populations inside and outside the Red Sea. In order to further assess how strong the genetic differentiation is between the sampled populations, the distribution of pairwise FSTs for each locus between populations was investigated.

This allows elucidating for example the existence of fixed loci and inferring the putative origin of the differentiation.

3.2.4.1 Principal Coordinate Analysis (PCoA)

Principal coordinate analysis (PCoA) of a genotype covariance matrix was performed in GenAlEx (6.501) for each individual sample. The PCA performed over populations (sampling sites) was based on an AMOVA using a matrix of pairwise

F’STs (also in GenAlEx, 6.501). This was used to summarize genotypic differentiation across all sampled individuals as well as populations.

3.2.4.2 Clustering Analysis

To explore genetic structure across sampling sites, a clustering analysis was performed in LEA snmf without a priori information of the geographic origin of each sample. Additionally, STRUCTURE was run under the admixture model with correlated allele frequencies (Falush et al. 2003), with a burn-in period of 200 000 139

MCMC iterations, followed by 300,000 iterations for each run. The number of K

(putative populations) ranged from one to eleven when all loci and all sites were included and ranged from one to six for analyses that only included samples from the Red Sea. Five replicate analyses were run in LEA for each value of K and three in

STRUCTURE. With the STRUCTURE results the number of clusters was inferred by comparing the ln Pr (X|K) among different values of K. The value of K for which ln Pr

(X|K) was highest or reached a plateau was selected as the most parsimonious number of populations in our sample. The ad hoc statistic ΔK (Evanno et al. 2005) was also considered. In LEA the K value with the lowest minimal cross-entropy was chosen.

3.2.5 Phylogenetic Analysis

As the analyses of the Dascyllus marginatus proceeded, very strong genetic differentiation between the populations from inside vs outside the Red Sea were found, including a remarkably high number of fixed alleles. A phylogenetic tree using the SNP data was also generated, which showed two clearly distinct genetic clusters. This lead to the assumption of potentially dealing with two separate species in the data set rather than just one. Hence, the COI gene was sequenced for two individuals of each sampling site using the universal FishF2 and FishR2 primers

(Ward et al., 2005) and PCR conditions were: 95ºC for 15 min, 6 cycles at 94ºC for

60 s, 55ºC for 40 s (−0.5ºC /cycle), 72ºC for 60 s, followed by 29 cycles of 94ºC for

60 s, 52ºC for 40 s, 72ºC for 60 s, and one final elongation step of 72ºC for 10min. 140

Furthermore, the SNPs based phylogentic tree had one sample, which seemed to potentially be a hybrid between the two clusters. Therefore, further nuclear genes were tested on 5 individuals from each cluster plus the putative hybrid to assess whether it was indeed a potential hybrid or not. The nuclear genes and primers used were the following:

In the case of D. aruanus, there have been more studies on its phylogeny suggesting the existence of two separated species, one inside the Red Sea and the western

Indian Ocean and one in the western Indo-Pacific (Borsa et al., 2014; Liu et al.,

2014). Therefore, a phylogenetic analysis similar to that done for D. marginatus seemed redundant and I focused on the SNPs data analyses only.

141

3.2.6 Assessment of Pelagic Larval Duration (PLD)

3.2.6.1 Sample Collection

V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 3

FigureFig. 1. Sampling 3. sites2 of:Dascyllus Sampling sites of marginatus, D. aruanus, and D. trimaculatusDascyllus marginatus, D. aruanus,. A total of 20 sites are grouped into four regions within the Red Sea (RS)and and twoD. trimaculatus outside the Red Sea (IA: Indo- Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- for central RS), the or sails assessment (SRS: southern RS) according of to the pelagic respective region. larval The regions were duration assigned in consensus (PLD) with those in Raitsos inside et al. (2013) the . The two environmental locations outside the Red Sea are represented by a white nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional gradient of the Red Sea (RS) as well as outside the information on exact number of samples and coordinates for each site can be found in Table 1. The chlorophyll a (CHLA,Red Sea in mg m−3). A total of 20 sites are concentrations of the Indo-Paciﬁc and Red Sea are displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations. grouped into four regions within the Red Sea and two outside the RS (IA: Indo- Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- central RS), or sails (SRS: southern RS) according to the respective region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea are represented by a white nabla and a white triangle (IO and IA, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional information on exact number of samples and coordinates for each site can be found in Table 3.1. The chlorophyll a (CHLA, in mg m−3) concentrations of the Indo-Pacific and Red Sea are displayed to visualize approximate differences between locations (NASA Giovanni website; http://oceancolor.gsfc.nasa.gov).

Fig. 2. Distribution of ﬁsh sizes (total length, in mm) of Dascyllus marginatus (left panel) and D. aruanus (right panel) used for PLD measurements from each site. Sites are given with three letter abbreviations on the x-axis (reef names can be found in Table 1). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles represent mean sizes and the whiskers display the maximum and minimum sizes of the ﬁshes.

142

Clove oil, tweezers, hand nets, and spears were used to collect juvenile and adult

specimens from the three Dascyllus species (D. aruanus, D. marginatus, and D.

trimaculatus) present in the Red Sea (Figure 3.2). Fishes of different sizes (Figure

3.3) and/or from different sampling years (Table 3.1) were collected for PLD

assessment to cover potential temporal variations as much as possible. These were

caught at 18 reef sites off the Saudi Arabian coastline and at two sites outside the

Red Sea (Indian Ocean “IO” and Indo-Australian-Archipelago “IA”; Table 3.1 and

Fig. 1. SamplingFigure sites of 3.Dascyllus2). marginatus Samples , D. aruanus were , and D. trimaculatus immediately . A total of 20 sites preserved are grouped into four in regions 90% within the ethanol. Red Sea (RS) and The two outside sampling the Red Sea (IA: Indo- Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- central RS), or sails (SRS: southern RS) according to the respective region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea are represented by a white nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional informationrange (N1600 km of coastline) covers most of the latitudinal span of the on exact number of samples and coordinates for each site can be found in Table 1. The chlorophyll a (CHLA, in mg m−3) concentrations of theRed Sea Indo-Paciﬁc. and Red Sea are displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations.

Fig. 2. DistributionFigure of fi sh3. sizes3: (total Distribution of fish sizes (total length, in mm) of length, in mm) of Dascyllus marginatus (left panel) and D. aruanus (right panel) used for PLD measurementsDascyllus mar from each site. Sitesginatus are given with three letter abbreviations on the x-axis (reef names can be found in Table 1). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles represent(left panel) and mean sizes and the whiskers displayD. aruanus the maximum and(right panel) used for PLD measurements from each site minimum sizes of the fishes. of Figure 3.2. Sites are given with three letter abbreviations on the x-axis (reef Pleasenames can be found in Table cite this article as: Robitzch, V.S.N., et al., Productivity3.1). The sites are in latitudinal order from north to and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 south from left to right, respectively along the axis. Black circles represent mean sizes and the whiskers display the maximum and minimum sizes of the fishes.

143

Whenever possible, specimens from all three species were collected from the same reef site for coherent interspecific comparisons. In case of absence of any of the species, specimens from the nearest possible site were sampled. The sampling sites were divided into four major Red Sea regions: “GAQ” (Gulf of Aqaba), “NCRS” (north- central RS), “SCRS” (south-central RS); “SRS” (southern RS) (in consensus with

Raitsos et al., (2013); Figure 3.2 and Table 3.1).

144

Table 3.1: Sampling sites for the assessment of the pelagic larval duration (PLD) of Dascyllus trimaculatus. D. aruanus, and D. marginatus, including number of samples, reef names and coordinates Location Region Site N (Dt) N (Dm) N (Da) Latitude Longitude Reef Name

RS GAQ HAQ 2 7 7 29°15'11.45"N 34°56'20.11"E Haql NCRS YRM 1 - - 23°46'21.12"N 38°16'34.02"E Yanbu-Ras Majiz RMA - 9 10 23°18'36.12"N 38°26'12.48"E Ras Masturah AKA - 3 11 22°56'15.90"N 38°45'56.58"E Shib Al Karrah SNA 2 - 12 22°20'51.96"N 38°51'09.42"E Shib Nazar FSA - 8 - 22°17'46.80"N 39°04'26.64"E inner Fsar AFA 4 - - 22°17'49.20"N 38°57'32.40"E Al Fahal SAK - - 7 21°40'16.53"N 38°50'36.60"E Shib Al Kabir AMA - - 10 22°04'25.75"N 38°46'40.26"E Abu Madafi Total 7 20 50 NSRS MAN 4 7 8 20°08'05.10"N 40°06'04.39"E Manilla Bay SAU 2 - 7 19°53'15.43"N 40°09'23.94"E Saut Reef LCG - 2 - 20°09'44.52"N 40°13'36.72"E Coast Guard Reef Total 6 9 15 SRS GHU - - 1 17°06'37.20"N 42°04'03.10"E Ghurab DAH 2 - - 16°52'22.60"N 41°26'24.50"E Dhi Dahaya ZDU 1 7 - 16°50'03.30"N 42°18'38.20"E Zahrat Durakah BAG 1 - 1 16°58'44.00"N 41°23'05.60"E Al Baglah DUR - 13 - 16°51'36.20"N 42°19'18.00"E Durakah DUM 4 - 11 16°33'50.80"N 42°03'30.60"E Dumsuq Total 8 20 13

IA OUT IND - - 7 02°11'11.04"N 118°32'55.33"E Manado Nth Sulawesi

IO OUT COC 5 - - 12°7'42.24"S 96°55'0.42"E Cocos Keeling

Total 28 56 92 The three main geographic locations are the Red Sea (RS), the Indo-Australian- Archipelago (IA), and the Indian Ocean (IO). The RS location is subdivided in four major regions (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS). The sites represent the reefs sampled for each region (three letter codes based on the reef names). The IA and IO locations are represented by one sampling site each: Mandando in North Sulawesi, Indonesia (IND) and the island of Cocos Keeling, Australia (COC) respectively. N represents the number of individuals for which pelagic larval durations (PLDs) were assessed for statistical analyses of each studied species: D. trimaculatus (Dt), D. marginatus (Dm), and D. aruanus (Da). Samples from the RS were collected in 2013, 2014, and 2015; from the IO in 2010 and 2014; and from the IA in 2003.

145

3.2.6.2 Otolith Preparation and Pelagic Larval Duration (PLD) Assessment

PLDs for all three species were estimated from daily increments up to the settlement marks of sagittal otoliths. Sagittae were extracted and mounted onto a microscopy glass slide with thermoplastic resin, then ground and polished

(following Wilson and McCormick, 1997) using aluminum oxide lapping films

(South Bay Technology, Inc.) of three different thicknesses (30, 12, and 5 μm). Three to ten pictures were taken of each polished otolith with the program AxioVision Rel.

AXIO Scope A1 microscope (200× magnification) and a Zeiss AxioCam ICc1. From the pictures, three independent PLD readings were made and if the counts deviated by less than 10% the mean count for each individual was calculated and included in the analyses. The same reader calculated PLDs for all species and lacked any information or metadata about the origin of the sample. A total of 28 D. trimaculatus,

56 D. marginatus, and 92 D. aruanus otolith readings were used for the statistical analyses (Table 3.1).

3.2.6.3 Chlorophyll a (CHLA) and Sea Surface Temperature (SST) Correlations

Two environmental parameters of the Red Sea were tested for correlations to PLD data: Chlorophyll a (CHLA) and sea surface temperature (SST). CHLA content is a good proxy for phytoplankton biomass (Håkanson et al., 2007) and is herein used as a proxy of food availability for the pelagic larva of the three studied species. 146

Validated regional averages of CHLA concentrations (mg m− 3) and SST (°C) were provided by D. Raitsos, based on Raitsos et al. (2013) for the regions NRS, NCRS,

SCRS, and SRS; and by D. Dreano for the Gulf of Aqaba (GAQ). These estimates were produced from a 10-year high-resolution data set of satellite remote-sensing CHLA estimates (see Raitsos et al. (2013) for a detailed description of data acquisition).

These estimates were subsequently used for Pearson correlations with PLD for each species

3.2.6.4 Statistical Analyses

All data were tested for normality (Kolmogorov–Smirnov test) and homoscedasticity (Levene's test) prior to analyses. The differences in PLD among the three species in the four Red Sea regions (GAQ, NCRS, SCRS, and SRS) were explored with a factorial ANOVA. The spatial variation of the PLD within each species was tested using an independent one-way ANOVA for each of the three

Dascyllus species. A post-hoc test applying the unequal N Tukey's HSD means comparisons was used to detect any significant differences within the two aforementioned analyses. Subsequently, the PLD data was correlated (Pearson correlation) with latitude, SST, and CHLA (i.e., food availability) measurements to assess the impact of the latitudinal environmental gradient of the Red Sea. Finally, the PLDs from the Red Sea sites were compared against those from locations outside the Red Sea for the two widespread species (D. trimaculatus and D. aruanus). For this analysis, data from the four Red Sea regions for each of the two species was 147 merged and compared independently with the IO (D. trimaculatus) and the IA (D. aruanus) using a t-student test. All analyses were conducted with STATISTICA 8.0

(StatSoft, Inc. 2007).

148

3.3 RESULTS

3.3.1 De-novo Assembly

Table 3.2: Sampling sites for the generation of genetic libraries of single nucleotide polymorphisms (SNPs) for the species Dascyllus marginatus (Dm) and D. aruanus (Da), including number of samples, site names, acronyms (Abb.), and coordinates Lib. Prep. Data set Lib. Prep. Data set Loc. Region Site Name Abb. Latitude Longitude N (Dm) N (Dm) N (Da) N (Da) RS Gulf of Aqaba Haql GAQ 20 6 20 13 29°15′11.45′′N 34°56′20.11′′E North RS Burqan NRS 20 20 20 16 27°54'35.8''N 35°03'55.4''E

Shib Al Karrah Yanbu AKA - - 38 22(24-2) 22°56′15.90′′N 38°45′56.58′′E

Central RS Thuwal CRS 4 4 - - 22°17′46.80′′N 39°04′26.64′′E

Manilla Bay Al Lith CRS (FSB) 16 6 20 5 20°08′05.10′′N 40°06′04.39′′E

South RS Farasan SRS 20 15 9 - 16°37'07.9"N 41°55'58.9"E

Kamaran - 17 23 15°23'15.8"N 42°33'15.0"E

Djibouti Djibouti DJI 20 9 20 16 11°43'16.03"N 43°09'53.94"E

Total 100 60 144 95

IO Gulf of Aden Magateen MAG 20 19 - - 13°24'38.3"N 46°22'36.3"E Socotra SOC 20 18 - - 12°36'17.3"N 54°21'03.1"E

Persian Gulf Musandam MUS 20 16 - - 26°13'08.9"N 56°17'02.7"E

West IO Madagascar MAD - - 20 19 23°22'59.68"S 43°38'25.70"E

Total 60 53 20 19

WP West Pacific Paracel Islands PAI - - 12 11 16°31'40.68"N 111°45'2.45"E Taiwan TAI - - 15 4 20°41′19.17′′N 116°56′36.69′′E

Indonesia IND - - 7 2 2°11'11.04"N 118°32'55.33"E

Total - - 34 17

ALL 160 113 198 131 The three main geographic locations (Loc.) are the Red Sea (RS), the Indian Ocean (IO), and the West Pacific (WP). The RS location is subdivided in seven major regions, for which only the south RS is further subdivided into the Farasan and Kamaran sector as not enough specimen for the two species were found in both sampling sites. The sites are also grouped and have a three-letter abbreviations (Abb.), which is representative of the sampling region and used throughout this chapter regions (GAQ: Gulf of Aqaba, NRS: north RS, AKA: Shib Al Karrah; CRS: central RS; FSB: Farasan Banks; SRS: south RS, and DJI: Djibouti). The IO and WP locations are also represented by sampling regions, sites and a three letter abbreviation for each site, for which sites from the WP only have D. aruanus samples as the biogeographic range of D. marginatus ends along the Gulf of Aden in the north of the WP. Abbreviations for the sites are the following IO: MAG: Magateen, SOC: Socotra Island, MUS: Musandam, MAD: Madagasar; WP: PAI: Paracel Islands (China), TAI: Taiwan, and IND: Indonesia. The total number of samples used for the generation of SNP libraries is given first and the number of samples, which were lastly included in the analysis and thus, generated good quality and enough SNPs data is given second for D. marginatus (Dm) and D. aruanus (Da) respectively.

149

The final data set consisted of 4987 loci and 113 individuals for D. marginatus; and

133(-2) individuals with 7997 SNPs for D. aruanus (Table 3.2). The data came from the final parameter set in STACKS allowing up to four mismatches, not calling secondary reads, with a minimum stack coverage of six, a heterozygosity of minimum 60%, only including loci in at least 85% of the populations, and with less than 20% missing data per individual.

150

3.3.2 Population Genetics and Statistical Analyses

3.3.2.1 Dascyllus aruanus

GAQ 01 NRS 02 AKA 03 CRS 04 SRS 05 DJI 06 MAD 07 PAI 08 TAI 09

HAQ NRS AKA CRS SRS DJI MAD PAI TAI IND NRS 02 4000 HAQ ---- 0.053 0.097 0.077 0.063 0.067 0.001 0.001 0.001 0.001 2000 NRS 0.016 ---- 0.118 0.102 0.078 0.103 0.001 0.001 0.001 0.001 0 F’ST AKA 0.012 0.007 ---- 0.188 0.149 0.148 0.001 0.001 0.001 0.001

CRS 0.026 0.018 0.012 ---- 0.101 0.128 0.001 0.001 0.001 0.001 AKA 03 4000 SRS 0.013 0.009 0.005 0.015 ---- 0.124 0.001 0.001 0.001 0.001 2000 DJI 0.015 0.009 0.005 0.017 0.006 ---- 0.001 0.001 0.001 0.001 0 MAD 0.163 0.155 0.147 0.164 0.144 0.140 ---- 0.001 0.001 0.001 CRS

PAI 0.428 0.418 0.400 0.461 0.402 0.409 0.392 ---- 0.041 0.001 04

4000 Overall Pairwise TAI 0.403 0.393 0.378 0.431 0.380 0.384 0.377 0.059 ---- 0.001 2000 0 IND 0.350 0.342 0.327 0.373 0.330 0.330 0.346 0.266 0.319 ---- SRS 4000 05 2000 0 DJI 4000 06 2000

frequency 0 Frequency MAD

4000 07 2000 0 PAI 4000 08 2000 0 TAI 4000 09 2000 0 IND 4000 10 2000 0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 Pairwise pairwiseFST per Locus Fst

Figure 3.4: Pairwise (PW) F’ST comparisons over all loci (Table in the top right corner; values below the diagonal) as well as PW FSTs comparisons per locus (colored histograms) between each of the sampling sites of Dascyllus aruanus. The sampling sites are abbreviated with three-letter codes, which can be found in Table 3.2. The different colors represent the potential different biogeographic regions along the y axis of the figure (red: inside the Red Sea (RS); green: at the connection of the Red Sea off Djibouti; in turquoise: the site of Madagascar (MAD) in the western Indian Oean (IO); and purple: the remaining sites outside the Red Sea and in the western Pacific (WP) and coral reef triangle (CRT)). The table in the top right also includes p-values for the PW F’ST comparisons above the diagonal. Significantly different F’ST comparisons are marked in bold and italic numbers below the diagonal.

151

Standardized pairwise (PW) F’STs among all individuals between sampling sites were only significantly different between the populations inside and outside the Red

Sea (including Djibouti (DJI)). The genetic differentiation between the Red Sea populations and Madagascar (MAD; Indian Ocean (IO)) was however, over a third smaller than that between the Red Sea populations and those in the Western Pacific

(WP: Parcel Islands (PAI), Taiwan (TAI), and Indonesia (IND), see table in Figure

3.4). The patterns of the distribution of PW FSTs per locus showed similar distributions among population only inside the Red Sea (Figure 3.4 green and red histograms) and also for those comparisons among populations only outside the

Red Sea (last three righter-most purple histograms of Figure 3.4), without MAD.

These PW FSTs among inside and among outside the Red Sea populations were also respectively quite low and mainly around “0” (Figure 3.4). In comparison, the distribution profiles of the histograms between those comparisons of populations inside vs outside the Red Sea are wider (with higher FST values), showing there is higher genetic differentiation (the remaining purple and blue histograms of Figure

3.4, to the left). Those last histogram profiles are also very similar and show hints for loci under selection as there are some FST-values close to “1”, evidence for fixed loci in the populations (mainly between RS+MAD vs. PAI+TAI+IND, in purple, Figure

3.4). 152

a) b) 1.0 1.0 c) 0.8 0.8 0.66 0.6 0.6 K=2 0.4 0.4 Ancestry coefficients 0.65 0.2 0.2 0.64 0.0 0.0 Entropy − Sampled individuals 1.0 0.63 1.0 0.8 0.8 Minimal Cross 0.62 0.6 0.6

K=3 0.61 0.4 0.4 Ancestry coefficients 0.60 0.2 0.2 Minimal Cross-Entropy 2 4 6 8 10 0.0 0.0 Number of ancestral populations

Sampled individuals 1.0 1.0 Number of popula

0.6 0.6 K=4 0.4 0.4 Ancestry coefficients d) 0.2 0.2 0.0 0.0

Ancestry Coeﬃcients Sampled individuals 0.66 1.0 1.0 0.8 0.8 0.65 0.6 0.6 K=5 0.64 Entropy 0.4 0.4 − Ancestry coefficients 0.63 0.2 0.2 Minimal Cross 0.62 0.0 0.0

Sampled individuals 1.0 1.0 0.61 0.8 0.8 0.60 Minimal Cross-Entropy 0.6 0.6 K=6 2 4 6 8 10 Number of ancestral populations 0.4 0.4 Ancestry coefficients Number of popula

133 Sampled131 individuals Sampled Individuals

Figure 3.5: Genetic clustering analysis in the R package LEA for SNPs data sets of Dascyllus aruanus. Section a) and b) show the barplots generated in LEA. Plots in a) are based on a data set of 133 samples, which included 2 genetically odd samples which build their own cluster in the last three plots K>3. Barplots in section b) are based on the adjusted data set of 131 samples and excluding the two distinct samples under section a). The two plots to the right under sections c) and d) display the assessment of minimal cross-entropy (MCE) to choose the most likely number of total population clusters K. Section c) is for the dataset displayed under section a) and thus, hast the lowest MCE at K=4, while section d) hast the lowest at K=3, as it is for the dataset under section b), which does not include the samples which cluster separately. KAUST11_130; KAUST11_136.

Genetic clustering analysis in LEA (R) and in STRUCTURE gave comprehensive results. I chose to display just those from LEA because the plots resembled those from STRUCTURE and due to the relatively ease to modify plots from LEA for better visualization (no further packages needed). In Figure 3.5 genetic clustering helped 153 reveal that there were two odd samples in one of the populations within the Red

Sea, which clustered separately (i.e., two distinctively colored bars to the left in the last three bar plots in Figure 3.5a, for K=4, 5, and 6; North Red Sea (NRS)). There are two possible explanations for the discrimination of these two samples: First, these are representative of a genetically different population. Second, these specimen were mistakenly taken from a different set of samples, potentially, a different but closely related species. I suggest the latter mentioned to be more likely. The samples are from a museum collection in which fin clips of D. aruanus and D. marginatus were stored together in a 96 well plate. These are also duplicates/subsamples of the initial collection. I therefore, believe it may have happened that fin clips from D. marginatus were taken instead for D. aruanus, either during subsampling or as the samples selected for this study were compiled/taken from the plate to another for shipping. This could be tested by sequencing a mitochondrial DNA marker (e.g. COI) from the DNA extractions used for the SNPs library and blasting (ncbi databank;

BLASTn) the sequence to assess the species. The fact that these two samples do not appear as a segregated cluster in the K=2 bar plots is likely due to the low number of samples. By replicating the genotypes of those two samples (n>5) and rerunning the analyses these should significantly group into one separate cluster already at K=2.

The samples were cautiously removed to build a second dataset and compare results and check for putative biases. The two samples were both from the NRS sampling site and population statistics did not change with the exclusion of the samples. However, in the inference of putative population clusters the respective results for the total number of population clusters K (K with the lowest minimal 154 cross-entropy in plots Figure 3.5d vs Figure 3.5c) showed an increase in the total number of K from three to four if the two samples were included. It is therefore highly recommended to explore the data with different methods and amend the dataset if required before running final analyses. Figure 3.5 conclusively shows that our final data set of 131 individuals putatively consists of three distinct populations.

These would be one in the Red Sea including DJI (Figure 3.5b, K=3: dark-blue cluster/bars), one in Madagascar (MAD; Figure 3.5b, K=3: light blue cluster/bars), Principal Coordinates (PCoA) and one in the Western Pacific (WP: Principal Coordinates (PCoA) Paracel Islands (PAI), Taiwan (TAI), and Principal Coordinates (PCoA) Indonesia (IND); plot in Figure 3.5b, K=3: pink cluster/bars). Principal Coordinates (PCoA) Principal Coordinates (PCoA) D. aruanus (all) Principal Coordinates (PCoA) Axis 1 2 Principal Coordinates (PCoA) 3 Principal Coordinates (PCoA) % 65.03 17.60 11.68 Principal Coordinates (PCoA) Cum % 65.03 82.63 94.31 Principal Coordinates (PCoA) Coord. 2 Principal Coordinates (PCoA) Coord. 2

Coord. 2 IND PAI TAI . 2 Coord. 2 Coord. 2 Coord. 2

Coord MAD Coord. 2 Coord. 2 Coord. 1 DJI

Coord. 2 Coord. 1 SRS

Coord. 2 Coord. 1 AKA

Coord. 2 NRS CRS GAQ Coord. 1 Coord. 1

Coord. 1 Coord. 1 Coord. 1 Coord. 1 Coord. 1 Figure 3.6: Principal coordinate analysis (PCoA) Coord. 1 for the pairwise (PW) F’ST comparisons among all samples of each sampling Coord. 1 site of Dascyllus aruanus inside and outside the Red Sea (RS). The sites are each represented by a grey diamond. Blue represents the group of seven sites from inside the Red Sea and the western Indian Ocean (IO), where the latter is the righter most diamond of the plot. Green represents all other sites outside the Red Sea and in the western Pacific (WP) and coral reef triangle (CRT). The uppermost diamond/site of the green cluster represents the IND (Indonesia) site. The table in the top left of the plot shows the percentage of variance explained by the first three coordinates (Coord.). The plot displays the distance explained by the first two coordinates (with most explanatory power). For further information on the sampling sites see Table 3.2.

155

Finally, the PCoA plot (Figure 3.6) groups the samples into two clusters. MAD is hereby included into the Red Sea populations but a little bit more distant from the

Red Sea populations at the very left side of the plot (Figure 3.6). The WP populations form the second grouping, where IND is also a bit distant from the other two WP populations (Figure 3.6).

Principal Coordinates (PCoA)

3.3.2.2 Dascyllus aruanus inside the Red SeaPrincipal Coordinates (PCoA)

Principal Coordinates (PCoA) Principal Coordinates (PCoA) D. aruanus (RS) Principal Coordinates (PCoA) Axis 1 Principal Coordinates (PCoA) 2 3 Principal Coordinates (PCoA)

% Coord. 2 65.03 17.60 11.68 Cum % 65.03 82.63 94.31 GAQ Coord. 2 . 2 CRS Coord. 2 Coord Coord. 2 Coord. 1 Coord. 2 AKA Coord. 2 Coord. 2 SRS NRS Coord. 1 DJI

Coord. 1 Coord. 1 Coord. 1 Coord. 1 Coord. 1 Figure 3.7: Principal coordinate analysis (PCoA) Coord. 1 for the pairwise (PW) F’ST comparisons among all samples of each sampling site of Dascyllus aruanus only inside the Red Sea (RS). The sites are each represented by a grey diamond. Blue represents the site of the Gulf of Aqaba (GAQ). Purple represents the site of the central RS (CRS); and in orange all remaining RS site including that off Djibouti (DJI) cluster together. The table in the top left of the plot shows the percentage of variance explained by the first three coordinates (Coord.). The plot displays the distance explained by the first two coordinates (with most explanatory power). For further information on the sampling sites see Table 3.2.

156

After the assessment of the genetic information within the entire data set, I decided to solely analyze the Red Sea sites to assess whether there is genetic differentiation and further population structure within the Red Sea, which is potentially being masked by the much stronger differentiation between sites/populations within vs. outside the Red Sea. For instance, the barplots of Figure 3.5 also display separate clustering of the GAQ population and a presumable genetic differentiation among the other Red Sea populations as you move southwards along the environmental gradient. This can be observed in the barplots of K > 3 (cluster at the left most side of the barplot, in yellow, grey, or black, with increasing K, respectively; Figure 3.5).

Furthermore, the PCoA of the Red Sea populations grouped these into three distinct groups: one for the GAQ, one for the CRS and one for all other Red Sea populations

(Figure 3.7). The CRS samples are however low in numbers due to high missing data

(from 20 samples only five could be included/processed; see Table 3.2). This speaks for low quality DNA and might also be one reason for their separate grouping in the

PCoA and not due to actual genetic differentiation. Thus, even though the PCoA

(Figure 3.7) and some of the barplots (Figure 3.5) hint towards further genetic differentiation within the Red Sea apart from the GAQ, pairwise F’ST calculations between these Red Sea sites, are not significant and therefore reject strong genetic differentiation within the Red Sea (see Table 3.3).

157

Table 3.3: Pairwise (PW) F’ST comparisons over all loci of Dascyllus aruanus from sampling site inside the Red Sea (RS) only GAQ NRS AKA CRS SRS DJI GAQ ---- 0.056 0.100 0.069 0.080 0.065 NRS 0.016 ---- 0.133 0.128 0.091 0.097 AKA 0.012 0.007 ---- 0.154 0.125 0.179 CRS 0.026 0.018 0.012 ---- 0.125 0.113 SRS 0.013 0.009 0.005 0.015 ---- 0.117 DJI 0.015 0.009 0.005 0.017 0.006 ---- The values above the diagonal represent p-values for PW F’ST comparisons, which are displayed below the diagonal. The three- letter abbreviations represent each of the compared sampling sites. More information on the sampling sites can be found in Table 3.2.

158

3.3.2.3 Dascyllus marginatus

GAQ 01 NRS 02 CRS 03 SRS 04 DJI 05 MAG 06 SOC 07

GAQ NRS CRS SRS DJI MAG SOC MUS NRS

GAQ ---- 0.051 0.008 0.002 0.005 0.001 0.001 0.001 02 F’ST 1500 NRS 0.030 ---- 0.015 0.002 0.003 0.001 0.001 0.001 CRS 0.059 0.032 ---- 0.020 0.030 0.001 0.001 0.001 0 SRS 0.078 0.048 0.028 ---- 0.041 0.001 0.001 0.001 DJI 0.084 0.054 0.033 0.020 ---- 0.001 0.001 0.001 CRS MAG 0.607 0.584 0.579 0.565 0.541 ---- 0.111 0.100 03 1500 SOC 0.609 0.585 0.580 0.566 0.542 0.010 ---- 0.094 Overall Pairwise 0 MUS 0.611 0.585 0.581 0.567 0.543 0.010 0.008 ---- SRS 04

1500

0 DJI 05

1500 frequency

Frequency 0 MAG 06

1500

0 SOC 07

1500

0 MUS 08

1500

0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 pairwise Fst Pairwise FST per Locus

Figure 3.8: Pairwise (PW) F’ST comparisons over all loci (Table in the top right corner; values below the diagonal) as well as PW FST comparisons per locus (colored histograms) between each of the sampling sites of Dascyllus marginatus. The sampling sites are abbreviated with three-letter codes, which can be found in Table 3.2. The different colors represent the potential different biogeographic regions along the y-axis of the figure (red: inside the Red Sea (RS); green: at the connection of the Red Sea off Djibouti; and blue: outside the Red Sea and in the Gulf of Aden). The table in the top right also includes p-values for the PW F’ST comparisons above the diagonal. Significantly different F’ST comparisons are marked in bold and italic numbers below the diagonal.

Standardized pairwise (PW) F’STs among all individuals between sampling sites were significantly different between most sampling sites even within the Red Sea

(table inside Figure 3.8). In general, the sites outside the Red Sea were not significantly different from each other or from those within the southern Red Sea 159 and DJI. On the other hand, all sites inside the Red Sea were distinct from those outside. Additionally, and in contrast to D. aruanus, there is also strong and significant differentiation between sampling sites in the north and those in the southern Red Sea indicating the presence of two distinct populations inside the Red

Sea with a genetic break at about 20°N. The profiles of the distributions of PW FSTs per locus also nicely display the genetic differentiation between sampling sites

(histograms in Figure 3.8). For instance, all red and green as well as the last three

(rightmost) blue histograms show similar distribution profiles with most PW FSTs around “0”, indicating low genetic differentiation between those sites (Figure 3.8).

Within this comparison, a slightly different distribution pattern can be found for the northernmost site (GAQ), Figure 3.8: first four histograms (3xred 1xgreen)).

Now, when looking at the histograms of PW comparisons between sites inside vs. outside the Red Sea the distributions show distinct differences with extraordinarily high numbers of fixed loci (FSTs around “1”, Figure 3.8: all remaining blue histograms). The observation of such high numbers of fixed loci between populations inside vs. outside the Red Sea is indicative of selective pressure on the species, potentially leading to speciation. Thus, the SNPs data (Figure 3.9b) and a further mitochondrial DNA (mtDNA) marker COI (under 3.2.3.3) were assessed to phylogenetically investigate the Arabian Sea D. marginatus samples for potential cladogenesis.

160 100

a) 55.5 73.5 81.5 57 59 74.5 71 71 58.5 52.5 52.5 1.0 51 93 56 0.01 52 53

b) 100 0.8

0.6 K=2 0.4 Ancestry coefficients

0.2 DJI GAQ NRS CRS SRS 0.0

Sampled individuals 1.0 Aref 001 Aref 010 Aref 008 Aref 004 Aref 011 MAN 010801 Aref 006 Aref 009 Aref 005 DJI12 076 Aref 013 Aref 007 Aref 016 Aref 003 DJI12 084 DJI12 085 DJI12 078 MAN 010903 Aref 002 MAN 010803 MAN 010901 Aref 015 MAN 020201 MAN 020101 DJI12 067 0.8 DJI12 068 KAUST11 248 HAQ 010602 KAUST11 251 HAQ 010101 HAQ 010503 KAUST11 254 HAQ 010505 KAUST11 240 KAUST 241 KAUST11 245 Aref 012 KAUST11 252 KAUST 183 KAUST 244 KAUST11 239 KAUST 238 FSA 010101 FSA 010102 KAUST 228 HAQ 010401 KAUST 232 KAUST11 247 DJI12 087 KAUST11 255 KAUST11 249 HAQ 010701 KAUST11 246 KAUST11 253 FSA 010104 KAUST 185 FSA 010103 DJI12 083 KAUST 250 0.6

K=3 DJI12 081 0.4 Ancestry coefficients 100 0.2 100 0.0 92 100 Sampled individuals 100 1.0 0.8 0.6

0.4 K=4 Ancestry coefficients MUS SOC MAG 0.2 MUS 010105 0.0 MUS 010102 MAG 010110 MAG 010113 MUS 010103 MAG 010107 MAG 010106 MUS 010118 SOC 010107 MUS 010120 SOC 010117 MUS 010117 MUS 010114 MUS 010113 SOC 010105 MAG 010118 SOC 010111 MAG 010105 SOC 010112 MUS 010109 SOC 010106 SOC 010104 SOC 010119 MAG 010116 MUS 010106 MUS 010104 SOC 010101 SOC 010114 MAG 010103 MUS 010110 MAG 010117 SOC 010118

Sampled individuals MUS 010116 MAG 010104 SOC 010116 MAG 010109 MAG 010101 SOC 010113 MAG 010120 MAG 010115 MUS 010112 SOC 010102 MUS 010101 1.0 MAG 010112 MAG 010102 MUS 010115 SOC 010103

MAG 010114 0.01 MAG 010108 SOC 010110 SOC 010120 SOC 010115 MAG 010111 0.8

0.6 K=5 0.4 Ancestry coefficients 0.2 0.0

Sampled individuals

1.0 c) 0.8 0.6 K=6 0.68 0.4 Ancestry coefficients Ancestry Coeﬃcients 0.66 0.2 0.0 0.64 Sampled individuals 1.0 0.8 Entropy − 0.62 0.6

K=7 0.60 0.4 Ancestry coefficients Minimal Cross 0.2 0.58 0.0

Sampled individuals 1.0 0.56 0.8 0.54 0.6

Minimal Cross-Entropy 2 4 6 8 10

0.4 K=10 Ancestry coefficients Number of ancestral populations 0.2 Number of popula

Sampled individuals Sampled Individuals

Figure 3.9: Genetic clustering analysis in the R package LEA for the SNPs data set of Dascyllus marginatus for all sampling sites. Section a) show the barplots generated in LEA, where each color represents a putative genetic cluster or population. Panel b) shows a genetic phylogenetic tree for all sampling sites with bootstrap values at each branch and the genetic distance at the right corner. The sampling sites are abbreviated with three-letter codes, which can be found in Table 3.2. Panel c) displays the assessment of minimal cross-entropy (MCE) to choose the most likely number of total population clusters K. In this case the smallest MCEs are for K=2 and 3. In a) one can also find one distinct bar for K>4, which is suspected to be offspring of two distinct population clusters inside the Red Sea.

Genetic clustering analysis in LEA (R) and in STRUCTURE also gave comprehensive results for the D. marginatus data set. The total number of putative populations (K) was two (Figure 3.9c). The barplot in Figure 3.9a (K=2) clearly shows the break between these two populations: one inside the Red Sea (incl. DJI) and the other for the remaining three sites outside the Red Sea (MUS, SOC, and MAG). The break is so strong and the previously-mentioned genetic differentiation so high between the two populations that I decided to build a bootstrapped phylogenetic tree from the 161

SNP data (Figure 3.9b), which again clearly shows the split. The tree furthermore, shows one individual inside the Red Sea cluster, which seems to be a bit distant from each of the two lineages/tree branches and is in between (with a bootstrap of

100). I presumed this to be the result of hybridization between the two lineages. The sample is also from DJI, the site that is located at the entrance of the Red Sea and in between the two genetic clusters inside and outside the Red Sea. Additionally, when assessing the genetic clustering for K>4, this sample formed its own cluster, also speaking for its genetic differentiation (different colored line among the samples inside the Red Sea, Figure 3.9a); K>4; individual bar in grey, black and yellow with increasing K, respectively). To further investigate this putative hybrid sample, a

PCoA was done using each individual sample of the entire data set discretely (Figure

3.10) as well as for each sample from inside the Red Sea including DJI (both plots in

Figure 3.11). I was also interested in analyzing the Red Sea samples separately, as once the number of K is increased (K>2), one observes the further differentiation of the Red Sea samples into two distinct cluster (Figure 3.9, K>2, left half of the barplots). Moreover, the difference between the minimal cross entropy between

K=2 and K=3 (Figure 3.9c) is not very big, and thus, also speaks for potential population structure within the Red Sea and putatively three genetic clusters, total.

The fact that we might be dealing with the presence of a two species along the

Arabia Sea, may also mask the presence of further population structure inside the

Red Sea, all of which further justifies a discrete examination of the Red Sea samples.

Principal Coordinates (PCoA) Principal Coordinates (PCoA) Principal Coordinates (PCoA)

Principal Coordinates (PCoA) Principal Coordinates (PCoA) 162Principal Coordinates (PCoA) Principal Coordinates (PCoA) D. marginatus (all) Principal Coordinates (PCoA)

Axis 1 Coord. 2 2 3 Principal Coordinates (PCoA) % 65.03 Coord. 2 17.60 11.68 Principal Coordinates (PCoA) Cum % 65.03 Coord. 2 82.63 94.31 GAQ NRS CRS Coord. 2 Coord. 2

Coord. 2 MAG . 2 SOC Coord. 1 MUS Coord. 1 Coord. 2 Coord. 2

Coord Coord. 1

Coord. 2 DJI

Coord. 2 Coord. 1 DJI Coord. 1 SRS Coord. 1

Coord. 1 Coord. 1 Coord. 1 Coord. 1 Coord. 1 Figure 3.10: Principal coordinate analysis (PCoA) for the genetic distance between each sample of Dascyllus marginatus inside and outside the Red Sea (RS). The individuals are each represented by a grey diamond. Blue represents the group of sites from the northern RS (NRS; including the Gulf of Aqaba: GAQ) and the central RS (CRS). Orange and grey those from the southern RS (SRS) including Djibouti (DJI); and in green all other sites outside the Red Sea and from the Gulf of Aden group together. The grey sample is the putative offspring of the two clusters/groups within the Red Sea (orange and blue), which is a sample from DJI. The table in the top left of the plot shows the percentage of variance explained by the first three coordinates (Coord.). The plot displays the distance explained by the first two coordinates (with most explanatory power). For further information on the sampling sites see Table 3.2.

Once each individual sample was explored in the PCoA of Figure 3.10, the putative hybrid (grey sample within the DJI population, Figure 3.10) between the inside vs. outside Red Sea genetic clusters/species seemed more likely to be a “hybrid” or better worded offspring between the populations inside the Red Sea (North vs.

South Red Sea populations). Conclusively, the complete data set can be interpreted as follows: (1) D. marginatus seems to rather be two different species with the genetic break right at the geographic barrier of the strait of Bab El Mandeb, where 163 similarly to D. aruanus, the DJI individuals still remain genetically more similar to those inside the Red Sea than those outside. To further assess the last mentioned assumption we used eighteen additional nuclear markers for the assessment of hybrids (Appendix). The sequences of these were all the same for samples of both putative populations inside the Red Sea as well as the samples outside the Red Sea, except for two markers (BMP4Cos2FA/BMP4Cos2R (Cooper et al., 2009); and

RAG2F/RAG2R (Dibattista et al., 2012), for which the alleles of this one sample, rather seemed to be a mixture between the two populations inside the Red Sea and not between the two clusters inside vs. outside the Red Sea. A PCoA solely using Red

Sea samples (incl. DJI) was also consulted to further assess the genetic structure within the Red Sea as well as the origin of the genetic differentiation of this one DJI sample (Figure 3.11).

164 Principal Coordinates (PCoA)

3.3.2.4 D. marginatus inside the Red SeaPrincipal Coordinates (PCoA)

Principal Coordinates (PCoA)

Principal Coordinates (PCoA) Principal Coordinates (PCoA) D. marginatus (RS) Principal Coordinates (PCoA) Axis 1 2 3 % 4.98 2.83 2.35 Principal Coordinates (PCoA) Coord. 2 Cum % 4.98 7.81 10.16 CRS Coord. 2

. 2 NRS Coord. 2 Principal Coordinates (PCoA) Coord. 2 Coord DJI Coord. 2 Coord. 2 Coord. 1 DJI Principal Coordinates (PCoA) SRS Coord. 2 Coord. 1 Principal Coordinates (PCoA) GAQ Coord. 1 Principal Coordinates (PCoA) Coord. 1 Coord. 1 CoordPrincipal Coordinates (PCoA) . 1 Principal Coordinates (PCoA) Principal Coordinates (PCoA) Coord. 1 D. marginatus (RS) Coord. 2 Axis 1 2 3 Coord. 1 CRS % 4.98 2.83 2.35 Principal Coordinates (PCoA) Cum % 4.98 Coord. 2 7.81 10.16

NRS Coord. 2 . 2

Coord. 2 Coord. 1 DJI

Coord. 2 Coord Coord. 2 Coord. 2 FSB Coord. 1 SRS DJI Coord. 2

Coord. 1 GAQ Coord. 1 Coord. 1 Coord. 1 Coord. 1 Coord. 1 Figure 3.11: Principal coordinate analysis (PCoA) for the genetic distance between each sample of Dascyllus marginatus inside the Red Sea Coord. 1 (RS). The individuals are each represented by a grey diamond. Blue represents the group of sites from the Gulf of Aqaba (GAQ). Turquoise those from the north RS (NRS); purple from the central RS (CRS). Orange and grey groups individuals from the southern RS (SRS) including the Farasan Banks (FSB) and Djibouti (DJI). The grey sample is the putative offspring of the two clusters/groups within the Red Sea (orange and blue), which is a sample from DJI. The table in the top left of the plot shows the percentage of variance explained by the first three coordinates (Coord.). The plots display the distance explained by the first two coordinates (with most explanatory power). The top PCoA shows how the CRS site needs to be separated into two sites one which is still referred to as CRS and the second one, which is referred to as FSB in the bottom plot. For further information on the sampling sites see Table 3.2. 165

The PCoA of Figure 3.11, revealed four interesting results. (1) When grouping the clustered samples into the site of origin all individuals cluster comprehensively and grouped according to their sampling site, except for the ones from the central Red

Sea (CRS; Figure 3.11 two purple clusters in the top graph). Due to the low numbers of samples available from Thuwal and Al Lith (Manilla Bay, Table 3.2) I decided to use those two sampling sites to jointly represent the CRS site. While it seemed geographically plausible, genetically it was clearly incoherent. As displayed in the upper PCoA of Figure 3.11, the samples from the CRS in purple split into two clusters along the first coordinate (x-axis). This split perfectly separates the individuals of Thuwal from those of Al Lith. Thus, this result luckily revealed (2) the geographic split between the populations found inside the Red Sea for D. marginatus, which is around the area where the Farasan Banks (FSB) start, at about

20°N latitude. Once the CRS samples are separated into the FSB site (off Al Lith) and the CRS site (off Thuwal) the groups of the PCoA are once again comprehensive

(Figure 3.11, bottom graph). This analysis further disclosed (3) the grouping of the sites of the northern and central Red Sea and then those of the south-central Red

Sea (starting at the FSB all the way to DJI) into two distinct populations (also supported by the pairwise F’STs in Table 3.4, below). Nonetheless, the percentage of variance explained by the PCoA is quite low and the one explained by the second coordinate only about 3%. Finally, (4) the potential offspring of the two populations

(Figure 3.11, grey sample) groups into the southern Red Sea population and is not strongly isolated, which further speaks against it being a hybrid between the 166 putative two sister species of D. marginatus of the Arabian Sea (the one inside vs. outside the Red Sea).

Table 3.4: Pairwise (PW) F’STs comparisons over all loci of Dascyllus marginatus from sampling site inside the Red Sea (RS) only HAQ NRS CRS FSB SRS DJI HAQ ---- 0.050 0.016 0.004 0.002 0.005 NRS 0.030 ---- 0.055 0.017 0.001 0.004 CRS 0.066 0.035 ---- 0.050 0.008 0.012 FSB 0.080 0.049 0.044 ---- 0.099 0.092 SRS 0.078 0.048 0.070 0.018 ---- 0.057 DJI 0.084 0.054 0.075 0.020 0.020 ---- The values above the diagonal represent p-values for PW F’ST comparisons, which are displayed below the diagonal. Values in bold below the diagonal are significantly different PW F’ST comparisons. The three-letter abbreviations represent each of the compared sampling sites. More information on the sampling sites can be found in Table 3.2.

To sum up and once more visualize the main outcome of this study, a final PCoA was performed using the pairwise F’STs between all sampling sites (Figure 3.12; see

Table in Figure 3.8 for pairwise F’STs). There, it becomes clear that over 90% of the variance can be explained just by separating the two regions: inside vs. outside the

Red Sea. This PCoA (Figure 3.12) also has the strongest explanatory power among all, higher but comparable to the one for D. aruanus (Figure 3.6).

Principal Coordinates (PCoA) Principal Coordinates (PCoA) Principal Coordinates (PCoA) 167Principal Coordinates (PCoA) Principal Coordinates (PCoA)

D. marginatus (all) Axis 1 2 3 Principal Coordinates (PCoA) Principal Coordinates (PCoA) % 91.07 4.97 1.36 Principal Coordinates (PCoA) Cum % 91.07 96.04 97.39 Principal Coordinates (PCoA) Coord. 2 Coord. 2

Coord. 2 GAQ

Coord. 2 NRS Coord. 2 . 2 CRS SRS DJI Coord Coord. 2 Coord. 2 Coord. 1 Coord. 2 Coord. 1 Coord. 2 MAG Coord. 1 SOC Coord. 1 MUS Coord. 1

Coord. 1 Coord. 1 Coord. 1 Coord. 1 Figure 3.12: Principal coordinate analysis (PCoA) Coord. 1 for the pairwise (PW) F’ST comparisons among all samples of each sampling site of Dascyllus marginatus from inside and outisde the Red Sea (RS). The sites are each represented by a grey diamond. Blue represents the site of within the Red Sea and green those outside the Red Sea and in the Gulf of Aden. The table in the top left of the plot shows the percentage of variance explained by the first three coordinates (Coord.). The plot displays the distance explained by the first two coordinates (with most explanatory power), for which coord. 1 explains over 90% of the variance in the data. For further information on the sampling sites see Table 3.2.

3.3.2.5 Comparative Analysis of D. aruanus and D. marginatus

The data sets for the D. aruanus and D. marginatus were subsampled, taking only specimens from inside the Red Sea to infer population structure within the Red Sea.

This was done because population differentiation within the Red Sea may have been previously masked by the strong genetic differentiation between inside vs. outside

Red Sea specimens. Here, genetic clustering in LEA clearly shows that there is strong genetic differentiation among the endemic D. marginatus species inside the Red Sea

(Figure 3.13b). For this species, samples split into two populations with a genetic 168 break around the south-central Red Sea (Figure 3.13b), at about 20°N. Even though

K with the lowest minimal cross-entropy for D. marginatus is K=1, the difference to

K=2 is not too large and the minimal cross-entropy of the latter also quite small

(Figure 3.13d). This, together with the strongly visual segregation into two genetically distinct clusters in the barplots, suggests the presence of two populations of D. marginatus in the Red Sea. In contrast, for the more widely distributed D. aruanus, the transition between potential clusters is smoother (Figure

3.13a). Also, the proportion of genetic differentiation per sample seems to rather change gradually from north to south within the Red Sea. Interestingly, though, is the fact that the GAQ individuals do seem to be more isolated into a unique cluster in comparison to the remaining D. aruanus samples from inside the Red Sea

(leftmost cluster in Figure 3.13a). Nonetheless, the minimal cross-entropy increases linearly for K>1, representative of panmixia and a lack of population structure among D. aruanus in the Red Sea (Figure 3.13c).

169

a) D. aruanus b) D. marginatus

1.0 1.0 c) 0.8 0.8 0.6 0.6

K=2 0.72 0.4 0.4 0.71 Ancestry coefficients 0.70 0.2 0.2 Entropy − 0.69 0.0 0.0 Minimal Cross

Sampled individuals 0.68 1.0 1.0 0.67 0.8 0.8 0.66

0.6 0.6 K=3 1 2 3 4 5 6

Number of ancestral populations Minimal Cross-Entropy 0.4 0.4 Ancestry coefficients Number of popula

Sampled individuals 1.0 1.0

0.8 0.8 d)

0.6 0.6 K=4 0.4 0.4 Ancestry coefficients 0.68 0.2 0.2 0.0 0.0 Ancestry Coeﬃcients 0.66

Sampled individuals Entropy − 1.0 1.0 0.64 Minimal Cross 0.8 0.8 0.6 0.6 0.62 0.4

0.4 K=5 Ancestry coefficients

1 2 3 4 5 6

Number of ancestral populations 0.2 0.2 Minimal Cross-Entropy Number of popula

Sampled individuals D. marginatus Sampled Individuals Figure 3.13: Genetic clustering analysis in the R package LEA. Section a) and b) show the barplots generated in LEA for samples from the species Dascyllus aruanus and D. marginatus only from inside the Red Sea (RS). Barplots in section c) and d) represent the assessment of minimal cross-entropy (MCE) to choose the most likely number of total population clusters K, which is one under plot c) and either one or putatively two under plot d) for each of the two species.

3.3.3 Phylogenetic Analysis

As the nuclear SNPs marker data set suggested the presence of two distinct species among of Dascyllus marginatus, I decided to do an assessment of the two genetic lineages using the mitochondrial COI marker (Figure 3.14). The phylogenetic tree resulted in a mixture of both lineages with no clear differentiation, which would be against the presence of two different species. Nonetheless, I strongly suggest these are two differentiated species and several explanations for unsorted mitochondrial lineages will be discussed in the discussion of this chapter (under 3.2.4).

MAG_08 MAG_12 MAG_08 SRS_13 MAG_08 MAG_12 NRS_46 MAG_12 SRS_13 MAG_08 MAN_01 SRS_13 NRS_46 MAG_12 CRS_04 MAG_08 NRS_46 MAN_01 SRS_13 GAQ_01 MAG_12 SRS_13 MAN_01 CRS_04 NRS_46 NRS_39 SRS_13 CRS_04 GAQ_01 MAN_01 MUS_16 NRS_46 MAG_08 MAG_08 GAQ_01 NRS_39 MAN_01 MAG_12 MAG_12 CRS_04 SOC_03 MAG_08 NRS_39 MUS_16 SRS_13 170MAG_08 SRS_13 GAQ_01 SRS_03 CRS_04 MAG_12 MUS_16 SOC_03 GAQ_01 SRS_13 NRS_46 MAG_12 NRS_46 NRS_39 MUS_06 SOC_03 SRS_03 MAN_01 SRS_13 MAN_01 MUS_16 SRS_10 NRS_39 NRS_46 MAG_08 SRS_03 MUS_06 SOC_03 DJI_78 MUS_16 SOC_03 MAN_01 CRS_04 NRS_46 MAG_08 CRS_04 MAG_08 MAG_12 MUS_06 SRS_10 SRS_03 DJI_83 SOC_03 SRS_03 CRS_04 GAQ_01 MAN_01 MAG_12 GAQ_01 MAG_12 SRS_13 SRS_10 DJI_78 MAG_15 SRS_03 GAQ_01 NRS_39 CRS_04 SRS_13 NRS_39 SRS_13 MUS_06 NRS_46 DJI_78 DJI_83 SRS_10 MUS_01 MUS_06 SRS_10 NRS_39 MUS_16 GAQ_01 NRS_46 MUS_16 NRS_46 MAN_01 DJI_83 MAG_15 GAQ_01 SRS_10 MUS_16 SOC_03 NRS_39 MAN_01 SOC_03 MAN_01 DJI_78 CRS_04 MAG_15 MUS_01 DJI_83 DJI_81 DJI_78 SOC_03 SRS_03 MUS_16 CRS_04 SRS_03 MAG_08 CRS_04 GAQ_01 MUS_01 GAQ_01 CRS_01 DJI_83 SRS_03 MUS_06 SOC_03 GAQ_01 MAG_08 MUS_06 MAG_12 GAQ_01 MAG_15 NRS_39 DJI_81 MUS_01 NRS_40 MAG_15 MUS_06 SRS_10 MAG_08 SRS_03 NRS_39 MAG_12 SRS_10 SRS_13 NRS_39 GAQ_01 MUS_16 CRS_01 GAQ_07 MUS_01 SRS_10 DJI_78 MAG_12 MUS_06 MUS_16 SRS_13 DJI_78 NRS_46 MUS_16 DJI_81 GAQ_01 SRS_13 SOC_03 CRS_01 NRS_40 DJI_81 SOC_14 GAQ_01 DJI_78 DJI_83 SRS_13 SRS_10 SOC_03 NRS_46 DJI_83 MAN_01 SOC_03 SRS_03 GAQ_07 CRS_01 SOC_13 DJI_81 DJI_83 MAG_15 NRS_46 DJI_78 SRS_03 MAN_01 MAG_15 CRS_04 SRS_03 MAG_08 NRS_40 MAN_01 MUS_06 SOC_14 NRS_40 CRS_01 MAG_15 MUS_01 MAN_01 DJI_83 MUS_06 CRS_04 MUS_01 GAQ_01 MUS_06 MAG_12 GAQ_07 SRS_10 SOC_13 GAQ_07 NRS_40 MUS_01 GAQ_01 CRS_04 MAG_15 SRS_10 GAQ_01 NRS_39 MAG_08 SRS_10 SRS_13 SOC_14 GAQ_01 DJI_78 MAG_12 SOC_14 GAQ_07 SOC_14 GAQ_01 DJI_81 GAQ_01 MUS_01 DJI_78 NRS_39 DJI_81 MUS_16 DJI_78 NRS_46 SOC_13 DJI_83 SRS_13 SOC_13 SOC_14 SOC_13 DJI_81 CRS_01 NRS_39 GAQ_01 DJI_83 MUS_16 MAG_08 CRS_01 SOC_03 DJI_83 MAN_01 MUS_16 MAG_15 NRS_46 SOC_13 CRS_01 NRS_40 MUS_16 DJI_81 MAG_15 SOC_03 MAG_12 NRS_40 SRS_03 MAG_15 CRS_04 MUS_01 MAN_01 NRS_40 GAQ_07 SOC_03 CRS_01 MUS_01 SRS_03 SRS_13 GAQ_07 MUS_06 MUS_01 GAQ_01 SRS_03 GAQ_01 CRS_04 GAQ_07 SOC_14 SRS_03 NRS_40 GAQ_01 MUS_06 NRS_46 SOC_14 SRS_10 GAQ_01 NRS_39 DJI_81 GAQ_01 SOC_14 SOC_13 MUS_06 MAG_08 GAQ_07 DJI_81 SRS_10 MAN_01 SOC_13 DJI_78 DJI_81 MUS_16 CRS_01 SOC_13 SRS_10 MAG_12 SOC_14 CRS_01 DJI_78 CRS_04 DJI_83 NRS_39 CRS_01 SOC_03 NRS_40 MUS_16 MAG_08 DJI_78 SRS_13 SOC_13 NRS_40 DJI_83 GAQ_01 MAG_15 NRS_40 SRS_03 GAQ_07 MAG_12 DJI_83 NRS_46 GAQ_07 MAG_15 NRS_39 MUS_01 SOC_03 GAQ_07 MUS_06 SOC_14 SRS_03 SRS_13 MAG_15 MAN_01 SOC_14 MUS_01 MUS_16 GAQ_01 SOC_14 SRS_10 SOC_13 NRS_46 MUS_01 CRS_04 SOC_13 GAQ_01 SOC_03 DJI_81 MUS_06 SOC_13 DJI_78 MAN_01 GAQ_01 DJI_81 SRS_03 CRS_01 SRS_10 DJI_83 CRS_04 DJI_81 NRS_39 CRS_01 MUS_06 NRS_40 DJI_78 MAG_15 GAQ_01 CRS_01 MUS_16 NRS_40 SRS_10 GAQ_07 DJI_83 MUS_01 Figure 3.14: Phylogenetic tree based on sequences of the mitochondrial COI marker NRS_39 NRS_40 SOC_03 GAQ_07 DJI_78 SOC_14 MAG_15 GAQ_01 for a subset of samples from the entire sampling range of MUS_16 GAQ_07 SRS_03 SOC_14 DJI_83 SOC_13 MUS_01 DJI_81 Dascyllus marginatus. COI sequences from Chromis viridis, another SOC_03 SOC_14 MUS_06 SOC_13 MAG_15 damselfish GAQ_01 CRS_01 species, were used as an out- SOC_13 DJI_81 group. The blue samples are from sites SRS_03 SOC_13 SRS_10 MUS_01 inNRS_40 side the Red Sea (RS), while green are CRS_01 those outside the Red Sea, which putatively represent a different species according MUS_06 DJI_78 GAQ_01 GAQ_07 NRS_40 SOC_14 to the results of the analyses of single nucleotide polymorphism (SNP) libraries. The SRS_10 DJI_83 DJI_81 DJI_78 CRS_01 GAQ_07 SOC_13 grey sample is an individual from Djibouti (DJIMAG_15 ), which seems genetically different DJI_83 MUS_01 NRS_40 SOC_14 and was believed to be either a hybrid between the SOC_13 Red Sea and outside Red Sea MAG_15 GAQ_01 GAQ_07 genetic lineages of D. marginatus or an MUS_01 DJI_81 SOC_14 offspring of the two genetically distinct populations of D. marginatus, which are found inside the GAQ_01 CRS_01 SOC_13 Red Sea. See Table 3.2 for more information on the sampling sites. DJI_81 NRS_40 CRS_01 GAQ_07 NRS_40 SOC_14 GAQ_07 SOC_13 SOC_14 SOC_13 171

3.3.4 Pelagic Larval Duration (PLD) Measurements

Table 3.5: Average pelagic larval durations (PLD in days; mean ± standard deviation) of Dascyllus marginatus, Dascyllus aruanus, and Dascyllus trimaculatus based on replicate counts of daily growth increments from otoliths Regio Locat. Site PLD Dm PLD Da PLD Dt Lat. mean SST [º C] mean CHLA [mg m-3] n RS GAQ HAQ 23.43 ± 2.89 25.10 ± 0.94 26.50 ± 1.17 29.253 24.2 0.20 (NRS) 25.8 0.15 NCRS YRM - - 27.00 23.773 RMA 19.56 ± 1.97 23.77 ± 1.32 - 23.310 AKA 24.55 ± 2.04 24.61 ± 1.89 - 22.938 SNA - 23.45 ± 1.46 26.67 ± 0.94 22.348 FSA 21.67 ± 1.00 - - 22.296 AFA - - 24.92 ± 2.06 22.297 SAK - 23.24 ± 1.33 - 21.671 AMA - 23.73 ± 1.37 - 22.074 Total 20.65 ± 2.33 23.79 ± 1.52 25.72 ± 1.81 27.2 0.14

SCRS MAN 19.29 ± 2.69 21.95 ± 1.77 24.58 ± 1.20 20.135 SAU - 22.52 ± 1.14 24.84 ± 0.23 19.888 LCG 20.67 ± 0.94 - - 20.162 Total 19.59 ± 2.43 22.22 ± 1.48 24.67 ± 0.94 28.8 0.27

SRS GHU - 21.00 - 17.110 DHI - - 22.00 ± 0.95 16.873 ZDU 20.09 ± 0.71 - 20.33 16.834 BAG - 24.00 25.33 16.979 DUR 17.83 ± 1.63 - - 16.860 DUM 20.09 ± 0.71 22.17 ± 1.57 16.564 Total 17.34 ± 1.43 20.63 ± 1.44 22.29 ± 1.76 28.9 1.64

Total 19.82 ± 2.92 23.22 ± 1.96 24.32 ± 2.17 IO - - 24.40 ± 1.25 12.128

IA - 21.05 ± 1.10 - 2.186 All 19.82 ± 2.92 23.09 ± 1.92 24.33 ± 2.02

D R2 [%] m 43.6 39.4 33.0 Da 34.8 30.6 27.6 Dt 46.2 30.1 50.1

Average PLDs for each species (D. marginatus (Dm), D. aruanus (Da), and D. trimaculatus (Dt)) are given among all (All) and within each geographic location: of the Indian Ocean (IO), the Indo-Australian-Archipelago (IA), and the Red Sea (RS); of each region within the RS (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS); and of each site within the Red Sea (three-letter coded reef name). The latitude (Lat., in decimal degrees) is given for each site as used for correlation analysis. Estimates of mean surface sea temperature (mean SST, in °C) and chlorophyll a content (CHLA, in mg m-3) are provided by D. Raitsos and D. Dreano from 10-years satellite remote sensing high-resolution data (in consensus with Raitsos et al., 2013).

172

3.3.4.1 Within the Red Sea

The total mean PLDs of all three species differed significantly from each other

(F(2,144)=50.453, p<0.00, Figure 3.15). Overall, D. trimaculatus showed the highest mean PLD (24.33 ± 2.02), followed by D. aruanus (23.09 ± 1.92), while D. marginatus exhibited the lowest (19.82 ± 2.92) (Table 3.5). Significant interspecific differences in PLDs were found among regions (F(3,144)=31.723, p<0.00, Figure 3.17) and intraspecific differences among all three species: between sampling regions

(F(3,144)=31.723, p<0.00, Figure 3.17) and sites (only assessed for D. marginatus

(F(7,44)=10.099, p< 0.05) and D. aruanus (F(11,76)=7.6954, df= 76, p< 0.05) due to low number of D. trimaculatus specimen per site; see Figure 3.17). The interaction between species and regions was not significant. In addition, a consistent latitudinal decrease in PLD was found to be present from north to south among all of the species between the four regions (see Figure 3.16). PLD ranges were the largest for the endemic species D. marginatus (15–28 d; 13 d difference) and almost double of the ranges found in D. aruanus (19–27 d; 8 d difference) and D. trimaculatus (20–27 d; 7 d difference). Within sampling sites, individuals of the endemic species also showed the highest PLD range (19–28 d, 9 d difference in HAQ) compared to those of the two widespread species (D. aruanus: 6 d difference in MAN; D. trimaculatus: 4 d difference in AFA).

4 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx

Table 1 Sampling sites, including number of samples, reef names and coordinates.

Location Region Site N (Dt) N (Dm) N (Da) Latitude Longitude Reef Name

RS GAQ HAQ 2 7 7 29°15′11.45″N 34°56′20.11″E Haql YRM 1 ––23°46′21.12″N 38°16′34.02″E Yanbu-Ras Majiz RMA – 9 10 23°18′36.12″N 38°26′12.48″E Ras Masturah AKA – 3 11 22°56′15.90″N 38°45′56.58″E Shib Al Karrah SNA 2 – 12 22°20′51.96″N 38°51′09.42″E Shib Nazar FSA – 8 – 22°17′46.80″N 39°04′26.64″E Inner Fsar AFA 4 ––22°17′49.20″N 38°57′32.40″E Al Fahal SAK –– 7 21°40′16.53″N 38°50′36.60″E Shib Al Kabir AMA ––10 22°04′25.75″N 38°46′40.26″E Abu Madaﬁ NCRS Total 7 20 50 MAN 4 7 8 20°08′05.10″N 40°06′04.39″E Manila Bay SAU 2 – 7 19°53′15.43″N 40°09′23.94″E Saut Reef LCG – 2 – 20°09′44.52″N 40°13′36.72″E Coast Guard Reef NSRS Total 6 9 15 GHU –– 1 17°06′37.20″N 42°04′03.10″E Ghurab DAH 2 ––16°52′22.60″N 41°26′24.50″E Dhi Dahaya ZDU 1 7 – 16°50′03.30″N 42°18′38.20″E Zahrat Durakah BAG 1 – 1 16°58′44.00″N 41°23′05.60″E Al Baglah DUR – 13 – 16°51′36.20″N 42°19′18.00″E Durakah DUM 4 – 11 16°33′50.80″N 42°03′30.60″E Dumsuq SRS Total 8 20 13 IA OUT IND –– 7 02°11′11.04″N 118°32′55.33″E Manado Nth Sulawesi IO OUT COC 5 ––12°7′42.24″S 96°55′0.42″E Cocos Keeling Total 28 56 92 173 The three main geographic locations are the Red Sea (RS), the Indo-Australian-Archipelago (IA), and the Indian Ocean (IO). The RS location is subdivided in four major regions (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS). The sites represent the reefs sampled for each region (three letter codes based on the reef names). The IA and IO locations are represented by one sampling site each: Mandando in North Sulawesi,3.3.4.2 Indonesia (IND) andInside vs. Outside the Red Sea the island of Cocos Keeling, Australia (COC) respectively. N represents the number of individuals for which pelagic larval durations (PLDs) were assessed for statistical analyses of each studied species: D. trimaculatus (Dt), D. marginatus (Dm), and D. aruanus (Da). Samples from the RS were collected in 2013, 2014, and 2015; from the IO in 2010 and 2014; and from the IA in 2003.

We report the PLD measurements for D. aruanus and D. trimaculatus from sites variation of the PLD within each species was tested using an indepen- D. trimaculatus (20–27 d; 7 d difference). Within sampling sites, individ- dent one-way ANOVA for each of the three Dascyllus species. A post-outside the ualsRed Sea of the endemic (D. aruanus species: 21.05 ± 1.10 in the IA; also showed the highestD. trimaculatus PLD range (19: 24.40 ± 1.25 –28 hoc test applying the unequal N Tukey's HSD means comparisons was d, 9 d difference in HAQ) compared to those of the two widespread used to detect any signiﬁcant differences within the two aforemen-in the IOspecies) and compare (D. aruanus these : 6 d to difference our mea insurements MAN; D. trimaculatus from inside : 4 the d differ-Red Sea. The tioned analyses. Subsequently, the PLD data was correlated (Pearson ence in AFA). correlation) with latitude, SST, and CHLA (i.e., food availability) mea-mean PLD of both widespread species may be longer outside the Red Sea than surements to assess the impact of the latitudinal environmental gradi- 3.1.2. Inside vs. outside the Red Sea within the Red Sea but are only significantly longer for D. aruanus (D. aruanus: RS vs. ent of the Red Sea. Finally, the PLDs from the Red Sea sites were We report the PLD measurements for D. aruanus and D. trimaculatus compared against those from locations outside the Red Sea for the two IA: t-value = 2.88, df = 86, p = 0.04; from sites outside the Red SeaD. trimaculatus (RS) (D. aruanus: RS vs. IO: t: 21.05 ± 1.10-value = 0.0 in the IA;7, df = 26, widespread species (D. trimaculatus and D. aruanus). For this analysis, D. trimaculatus: 24.40 ± 1.25 in the IO) and compare these to our mea- data from the four Red Sea regions for each of the two species was surements from inside the Red Sea. The mean PLD of both widespread merged and compared independently with the IO (D. trimaculatusp = 0.93; see ) Figure 3.17). and the IA (D. aruanus) using a t-student test. All analyses were conducted with STATISTICA 8.0 (StatSoft, Inc. 2007).

3. Results

3.1. PLD measurements

3.1.1. Within the Red Sea The total mean PLDs of all three species differed significantly from each other (F(2144) = 50.453, p b 0.001, Fig. 3). Overall, D. trimaculatus showed the highest mean PLD (24.33 ± 2.02), followed by D. aruanus (23.09 ± 1.92), while D. marginatus exhibited the lowest (19.82 ± 2.92) (Table 2). Significant interspecific differences in PLDs were found among regions (F(3144) = 31.723, p b 0.001, Fig. 5) and intraspecific differences among all three species: between sampling regions

(F(3144) = 31.723, p b 0.001, Fig. 5) and sites (only assessed for D. marginatus (F(7,44) = 10.099, p b 0.001) and D. aruanus (F(11,76) = 7.6954, p b 0.001 due to low number of D. trimaculatus specimen per site; see Fig. 5). The interaction between species and regions was not signiﬁcant. In addition, a consistent latitudinal decrease in PLD was found to be present from north to south among all of the species be- Figure 3.15: Mean pelagic larval duration (PLD, in days) for Dascyllus marginatus, D. tween the four regions (see Fig. 4). PLD ranges were the largest for the aruanus,Fig. and 3. MeanD. pelagic trimaculatus larval duration averaged (PLD, in days) among for Dascyllus all study marginatus sites , D. aruanus in the , and Red Sea. endemic species D. marginatus (15–28 d; 13 d difference) and almost Whiskers represent 95% confidence intervals around the mean (black circle).D. trimaculatus averaged among all study sites in the Red Sea. Whiskers represent 95% double of the ranges found in D. aruanus (19–27 d; 8 d difference) and conﬁdence intervals around the mean (black circle).

Table 2 Average pelagic larval durations (PLD in days; mean ± standard deviation) of Dascyllus marginatus, Dascyllus aruanus, and Dascyllus trimaculatus based on replicate counts of daily growth increments from otoliths.

Locat. Region Site PLD Dm PLD Da PLD Dt Lat. Mean SST [°C] Mean CHLA [mg m−3]

RS GAQ HAQ 23.43 ± 2.89 25.10 ± 0.94 26.50 ± 1.17 29.253 24.2 0.20 (NRS) 25.8 0.15 NCRS YRM –– 27.00 23.773 RMA 19.56 ± 1.97 23.77 ± 1.32 – 23.310 AKA 24.55 ± 2.04 24.61 ± 1.89 – 22.938 SNA – 23.45 ± 1.46 26.67 ± 0.94 22.348 FSA 21.67 ± 1.00 ––22.296 AFA –– 24.92 ± 2.06 22.297 SAK – 23.24 ± 1.33 – 21.671 AMA – 23.73 ± 1.37 – 22.074 Total 20.65 ± 2.33 23.79 ± 1.52 25.72 ± 1.81 27.2 0.14 SCRS MAN 19.29 ± 2.69 21.95 ± 1.77 24.58 ± 1.20 20.135 SAU – 22.52 ± 1.14 24.84 ± 0.23 19.888 LCG 20.67 ± 0.94 ––20.162 Total 19.59 ± 2.43 22.22 ± 1.48 24.67 ± 0.94 28.8 0.27 SRS GHU – 21.00 – 17.110 DHI – – 22.00 ± 0.95 16.873 ZDU 20.09 ± 0.71 – 20.33 16.834 BAG – 24.00 25.33 16.979 DUR 17.83 ± 1.63 ––16.860 DUM 20.09 ± 0.71 22.17 ± 1.57 16.564 Total 17.34 ± 1.43 20.63 ± 1.44 22.29 ± 1.76 28.9 1.64 Total 19.82 ± 2.92 23.22 ± 1.96 24.32 ± 2.17 IO –– 24.40 ± 1.25 12.128 IA – 21.05 ± 1.10 – 2.186 All 19.82 ± 2.92 23.09 ± 1.92 24.33 ± 2.02 R2 [%] Dm 43.6 39.4 33.0 Da 34.8 30.6 27.6 Dt 46.2 30.1 50.1

Average PLDs for each species (D. marginatus (Dm), D. aruanus (Da), and D. trimaculatus (Dt)) are given among all (All) and within each geographic location: of the Indian Ocean (IO), the Indo-Australian-Archipelago (IA), and the Red Sea (RS); of each region within the RS (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS); and of each site within the RS (three-letter coded reef name). The latitude (Lat., in decimal degrees) is given for each site as used for correlation analysis. Estimates of mean surface sea temperature (mean SST, in °C) and chlorophyll a content (CHLA, in mg m−3) are provided by D. Raitsos and D. Dreano from 10-years satellite remote sensing high-resolution data (in consensus with Raitsos et al., 2013).

species is longer outside the Red Sea than within the Red Sea but only 3.2. Environmental correlations with PLD in the Red Sea significantly longer for D. aruanus (D. aruanus: RS vs. IA: t-value = 2.88, df = 86, p = 0.04; D. trimaculatus: RS vs. IO: t-value = 0.07, Regional means of SST and CHLA concentrations were used as envi- df = 26, p = 0.93; see Fig. 5). ronmental data for Pearson correlations (Table 2). Their respective 174 mean annual profiles are visualized in Fig. 6. The profiles of each region display a similar trend-curve except for SRS and GAQ. The SRS has

Figure 3.16: Mean pelagic larval durations (PLD, in days) of three Dascyllus species in the four regions of the Red Sea (RS) (GAQ: Gulf of Aqaba, NCRS: noFig. 4. Mean pelagic larval durations (PLD, in days) of three Dascyllus species in therth four-central RS, regions of the Red Sea (RS) (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-cen- Fig. 5. Mean pelagic larval duration (PLD, in days) of Dascyllus aruanus and D. trimaculatus SCRS: southtral- RS,cen andtral SRS: RS, south and RS). Average SRS: PLDs south are given RS). by blueAverage triangles PLDs for D. marginatus are given , red by from blue locations inside the Red Sea (RS) and outside the Red Sea (Indo-Australian Archipel- triangles for squaresD. marginatus for D. aruanus, and, red squares for green diamonds for D. aruanusD. trimaculatus, and green diamonds for . Whiskers represent 95% ago (IA)D. and Indian Ocean (IO)). Whiskers represent 95% confidence intervals and boxes trimaculatuscon. Whiskers represent 95% confidence intervals around the mean. (For fidence intervals around the mean. (For interpretation of the references to color in represent one standard error around the mean. Mean PLDs are given by black squares interpretation of the references to color in this figure legend, the reader is referred this figure legend, the reader is referred to the web version of this article.) for D. aruanus and diamonds for D. trimaculatus. to the web version of this article.) Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045

V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 5

Locat. Region Site PLD Dm PLD Da PLD Dt Lat. Mean SST [°C] Mean CHLA [mg m−3]

Average PLDs for each species (D. marginatus (Dm), D. aruanus (Da), and D. trimaculatus (Dt)) are given among all (All) and within each geographic location: of the Indian Ocean (IO), the Indo-Australian-Archipelago (IA), and the Red Sea (RS); of each region within the RS (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS); and of each site within the RS (three-letter coded reef name). The latitude (Lat., in decimal degrees) is given for each site as used for correlation analysis. Estimates of mean surface sea temperature (mean SST, in °C) and chlorophyll a content (CHLA, in mg m−3) are provided by D. Raitsos and D. Dreano from 10-years satellite remote sensing high-resolution data (in consensus with Raitsos et al., 2013).

species is longer outside the Red Sea than within the Red Sea but only 3.2. Environmental correlations with PLD in the Red Sea significantly longer for D. aruanus (D. aruanus: RS vs. IA: t-value = 2.88, df = 86, p = 0.04; D. trimaculatus: RS vs. IO: t-value = 0.07, Regional means of SST and CHLA concentrations were used as envi- df = 26, p = 0.93; see Fig. 5). ronmental data for Pearson correlations (Table 2). Their respective mean annual profiles are visualized in Fig. 6. The profiles of each region display a similar trend-curve except175 for SRS and GAQ. The SRS has

Fig. 4. Mean pelagic larval durations (PLD, in days) of three Dascyllus speciesFigure in the four 3.17: Mean pelagic larval duration (PLD, in days) of Dascyllus aruanus and D. regions of the Red Sea (RS) (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-cen- Fig. 5. Mean pelagic larval duration (PLD, in days) of Dascyllus aruanus and D. trimaculatus tral RS, and SRS: south RS). Average PLDs are given by blue triangles for D. marginatustrimaculatus, red fromfrom locations inside the Red Sea (RS) and outside the locations inside the Red Sea (RS) and outside the Red Sea (Indo-Australian Archipel-Red Sea (Indo- squares for D. aruanus, and green diamonds for D. trimaculatus. Whiskers representAustralian 95% Archipelago (IA) andago Indian (IA) Ocean and (IO)). Whiskers Indian represent Ocean 95% (IO)). confidence Whiskers intervals and boxes represent 95% confidence intervals around the mean. (For interpretation of the referencesconfidence to color in intervals represent one and standard boxes error around represent the mean. one Mean standard PLDs are given error by black aro squaresund the mean. this figure legend, the reader is referred to the web version of this article.) Mean PLDs for are D. aruanus given and by diamonds black for D. squares trimaculatus. for D. aruanus and diamonds for D. trimaculatus. Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045

3.3.4.3 Environmental Correlations With Pelagic Larval Duration (PLD) in the

Red Sea

Regional means of SST and CHLA concentrations were used as environmental data

for Pearson correlations (Table 3.5). Their respective mean annual profiles are

visualized in Figure 3.18. The profiles of each region display a similar trend-curve

except for SRS and GAQ. The SRS has overall much higher CHLA concentrations and

its average annual CHLA profile has an additional peak in the summer. Different

from all other regional CHLA profiles, the GAQ displays a sharp peak later in the

spring and has both the highest concentrations of CHLA over the entire Red Sea

during the spring and the lowest during the summer (Figure 3.18). These unique 176

seasonal changes of CHLA in GAQ result in one of the largest regional ranges (>0.6

mg m−3) in CHLA concentrations in the Red Sea (besides the southernmost SRS

region; Table 3.5 and Figure 3.18, left panel). In contrast, the SST profiles for all

regions have similar ranges of about 6 °C (Figure 3.18, right panel).

6 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx

Figure 3.18: Validated monthly regional averages of chlorophyll a concentrations (CHLA, in mg Fig. 6. Validated monthly regional averages of chlorophyll a concentrations (CHLA, in mg m−3, left panel) and sea surface temperature (SST, in °C, right panel). Each panel shows monthly mean averagesm−3, left panel) and sea for an average calendar year (three-lettersurface temperature (SST, in °C, right panel). Each panel shows monthly coded months, x-axis) for each region (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS). Averagesmean averages for an average calendar year (three were produced from 10-year high-resolution data of satellite remote-sensing CHLA and-letter coded months, x SST estimates provided by D. Raitsos- (foraxis) for each region the regions NRS, NCRS, SCRS and SRS; sensu Raitsos et al. (2013)) and D. Dreano for the Gulf of Aqaba (GAQ). The trend lines are color-coded for each region. (GAQ: Gulf of Aqaba, NCRS: north-central Red Sea (RS), SCRS: south-central RS, and SRS: south RS). Averages were produced from 10-year high-resolution data of satellite remote-sensing CHLA overalland much SST higher estimates CHLA concentrations provided and itsby average D. Raitsos annual CHLA (for the gradual regions and consistent NRS, NCRS, decrease SCRS in mean and PLD SRS; with decreasing sensu latitude profileRaitsos et al. (2013)) and D. Dreano for the Gulf of Aqaba (GAQ). T has an additional peak in the summer. Different from all other re- along the environmentalhe trend lines are color gradient. Moreover, intra-speci-coded fic differences gional CHLA profiles, the GAQ displays a sharp peak later in the spring in PLD between sampling regions along this gradient were higher in D. and hasfor each region. both the highest concentrations of CHLA over the entire Red marginatus than the other two widespread species (D. aruanus and D. Sea during the spring and the lowest during the summer (Fig. 6). trimaculatus), suggesting that the endemic species might be more re- These unique seasonal changes of CHLA in GAQ result in one of the larg- sponsive to environmental differences within its habitat potentially re- est regional ranges3.3.4.4 (N0.6 mg m−3)Correlations With Latitude in CHLA concentrations in the Red lated to a higher degree of specialization. We provide the first PLD Sea (besides the southernmost SRS region; Table 2 and Fig. 6, left estimate for a broad geographic range of the endemic D. marginatus panel). In contrast, the SST profiles for all regions have similar ranges (19.8 ± 2.9 d), which displays a consistently lower PLD at all sites in of about 6 °C (Fig. 6, right panel). comparison to PLDs of D. aruanus and D. trimaculatus. Our results suggest that environmental factors can act as drivers of biogeography (as 3.2.1. CorrelationsPearson correlation tests show a positive with/atitude , highly significant (p

3.3.4.5 Correlations With Sea Surface Temperature (SST)

Temperature (SST) is significantly negatively correlated with PLDs of all three species (longer PLDs at lower SST) and shows the second highest R2 values (except among D. trimaculatus, R2= 30.1%, Table 3.5). This correlation is negative and significant for all species (Figure 3.19).

3.3.4.6 Correlations With Chlorophyll a (CHLA)

CHLA concentration is significantly (p < 0.001) negatively correlated to the PLDs of all three Dascyllus species. The correlation is the strongest for D. trimaculatus (R2 =

50.1%; Table 3.5, Figure 3.19). However, the sample size for this species is low and limits the overall interpretation of this result. Moreover, the R2 values of SST and

CHLA do not differ much and make it difficult to interpret which of the two parameters has the stronger influence on PLDs.

178 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 7

Figure 3.19: Plots of Pearson correlations of pelagic larval durations (PLD, in days, Fig. 7. Plots of Pearson correlations of pelagic larval durations (PLD, in days, y-axis) of three Dascyllus species (Dascyllus marginatus, D. aruanus, D. trimaculatus, from left to right) to: A–C) regional means of seay- surfaceaxis) temperature of three (SST in °C,Dascyllusfirst row), D– F)species regional means ( ofDascyllus chlorophyll a concentrations marginatus (CHLA,, inD. mg m− aruanus3, middle row),, andand G–I)D. latitude (in degrees, last row; values in Tabletrimaculatus 2). The plots are arranged, from left to right) to: A in columns according to species (i.e.,–D.C) regional means of sea surface temperature marginatus shown in panels A, D, G; D. aruanus in panels B, E, H; and D. trimaculatus in panels C, F, I) and regional means of: SST, CHLA, and latitude. The blue circles represent the measured PLD values of each species (y-axis, in days). The regression line is in red and the dotted curves represent the respective(SST in °C, first row), D 95% confidence intervals of the correlation–F) regional means of chlorophyll a concentrations (CHLA, in tests. mg m−3, middle row), and G–I) latitude (in degrees, last row; values in Table 3.5). fishes (Booth andThe plots are arranged in columns according to species (i.e., Parkinson, 2011). We interpret our findings as the all other regions are ratherD. marginatus homogeneous shown in In other words, larvae that synergic effect of greater food availability (higher CHLA concentrations) hatch during the spring/early summer in GAQ have up to six times and higher metabolismpanels A, D, G; (higher SST)D. aruanus on larval growth in panels B, E, H; and rates, subse- moreD. trimaculatus food available compared in panels C, F, I) and to cohorts hatching during other seasons. quently reducingregional means of: SST, PLDs for all species in the warmer CHLA, and latitude. The blue circles represent the measured and more produc- The extreme variability in the food available to these cohorts could be tive southernPLD regions values unique of to each the Red species Sea. Furthermore, (y-axis, in the days). responsible The regression for the large line range is of in PLDs red within and thethe GAQ region. We decrease of PLDsdotted curves represent the respective 95% confidence along the SST and CHLA gradient seems to be stronger therefore hypothesizeintervals of the correlation that if the range of PLD differences within sites and more significant among individuals of the endemic and ecologically is a result of seasonal changes in the pelagic environment, seasonal more specializedtest.D. marginatus in comparison to D. aruanus and D. changes in CHLA (food availability) rather than SST could more likely trimaculatus. We thus propose that the influence of small-scale regional be the driver of differences in PLDs of Dascyllus in the Red Sea. However, changes in the environment might be stronger and positively correlated the proposed hypothesis warrants further investigation in view of the with the degree of ecological specialization and habitat range of correlation between temperature and food availability (CHLA) damselfishes. (McCormick, 1994; Lo-Yat et al., 2011) and the implementation of sta- This is best illustrated in our study by the observation of large differ- tistical models to discretely analyze the influence of these two environ- ences in PLDs among the endemic species, D. marginatus at one specific mental parameters on coral reef fishes. In terms of biological data, site in the Gulf of Aqaba (GAQ) (a 9 d range of difference in HAQ). These recruit collections could be targeted at different times of the year to as- differences in PLDs were the highest among all Dascyllus in the Red Sea. sess whether longer PLDs correlate with seasons of lower productivity. If we then focus on the regional seasonal changes of the two environ- For the more “homogeneously” productive regions of the Red Sea one mental parameters SST and CHLA, the range of seasonal changes of could assess if the differences in PLDs are also more homogeneously dis- CHLA concentrations for the GAQ region is distinctly higher compared tributed throughout the year. Unfortunately, we were unable to reliably to adjacent regions to the south, further suggesting that the GAQ is a back-calculate birthdates for our individual samples to determine spe- variable environment itself. In contrast, the seasonal changes in SST in cific intra-annual patterns. Alternate hypotheses that could explain

3.4 DISCUSSION AND CONCLUSIONS

It is thought that spatial-temporal variation in the environment can select for or against dispersal (Duputié & Massol, 2013), ultimately shaping specialization, evolutionary processes, and the distribution of species (Heinz et al., 2009; Berdahl et al., 2015). Thus, understanding the processes that shape genetic structure and lead to population differentiation across landscapes is one of the primary objectives in the fields of population genetics and biogeography. Here, I assessed which processes are shaping genetics and biological traits in an endemic vs. a widespread coral reef fish species, Dascyllus marginatus and D. aruanus, along their distribution ranges. Three to four main putative barriers along their distribution were explored

(Figure 3.1: b1-4). Additionally, the pelagic larval duration (PLD) of all three

Dascyllus species found in the Red Sea was measured to quantify the correlation between PLD and two environmental parameters: sea surface temperature (SST) and chlorophyll a (CHLA) as a proxy of food availability. In the following, I will first, discuss the effect of the previously mentioned barriers, on geneflow and dispersal, in relation to the biogeography and potential adaptive strategies of D. aruanus and

D. marginatus,. Second, I will present putative reasons for the gradual decreases in

PLD length along the environmental gradient of the Red Sea in the light of biogeography and ecological specialization among the three Red Sea Dascyllus: D. trimaculatus, D. aruanus, and D. marginatus.

180

3.4.1 Contrasting Genetic Structures of the Widespread Dascyllus aruanus vs. the

Endemic D. marginatus

All in all, the best interpretation of the results from the single nucleotide polymorphism (SNPs) data of the widespread D. aruanus is, in my opinion that there are two genetic distinct clusters, which may represent the presence of two species with one of the two additionally displaying some genetic structure among the sampling sites captured in this chapter. These two putative species would be distributed as follows: one in the Red Sea and along the western Indian Ocean and the other, on the eastern side of and in the coral reef triangle. Such a split of D. aruanus was recently also suggested in Borsa et al. (2014). In the SNPs data, strong genetic differentiation is found between the populations of Madagascar plus the Red

Sea and the sites in the west Pacific. While Borsa et al. (2014) refers to the aforementioned differentiation as the “resurrection” of D. abudafur, I am hesitant in calling these genetic clusters different species, due to the still or currently very low divergence between the D. abudafur and D. aruanus based on mitochondrial DNA

(mtDNA) markers (McCafferty et al., 2002; Borsa et al., 2014; Liu et al., 2014) and the results from the this chapter’s SNPs data. I further suggest to rather referring to these as two different populations of D. aruanus. The origin for genetic drift between individuals in these two regions is likely vicariance, where the potential physical barrier was the Indian Ocean/Indo-Australian Archipelago (IO/IA) land barrier

(sensu Cowman and Bellwood, (2013), b4 in Figure 3.1), which arose due to changes in sea level. This land mass has been a barrier to dispersal for several marine 181 species several times through geological history and more recently due to changes in sea level during ice-ages (Cowman & Bellwood, 2013a). This history has given rise to many species pairs with one on each side of the coral reef triangle in several marine taxa. I hypothesize that in the case of D. aruanus the most significant isolation by vicariance was at the latest during the late Miocene (5-7.5 MYA) or as early as up to 25 MYA in the late Oligocene rather than during more recent drops in sea level. The reasons for this assumption is that in comparison to other coral reef fish families, for which recent changes in sea level seem to have had an evolutionary impact (butterflyfishes (Fessler & Westneat, 2007); parrotfishes (Choat et al., 2012);

Labridae and Chaetodontidae (Cowman & Bellwood, 2013a, 2013b), it does not seem to be the case among Pomacentridae (Cowman and Bellwood 2013a, b). The major cladogenesis of the genus Dascyllus was around 12 to 9 MYA (McCafferty et al., 2002). Additionally, the currently assessed genetic differentiation between the two putatively distinct species is very low (in this study as well as in earlier studies by Borsa et al. (2014) and Liu et al. (2014)) in comparison to other divergence among Dascyllus species (McCafferty et al., 2002). The rate of genetic differentiation

I find within my data and the results from the analysis are more typical to those inferring structure in population genetics. Thus, I believe the species has not diverged yet and as dispersal routes are currently presumably open (evidence for gene flow between the Red Sea and Madagascar), gene flow (or hybridization) between the two populations or putative species across the IO/IA may be ongoing, enough to keep them admixed as a single species. Hence, if there was a period of vicariance between the two regions or if speciation was ongoing, I suggest these two 182 will likely have increased geneflow in the coming future, also since the Straight of

Bad El Mandab is further opening and sea levels are rising. Nevertheless, the lack of knowledge on the exact transition zone between the two genetic clusters, the period of isolation that lead to genetic differentiation, and the current degree of gene flow

(or hybridization) between these two putatively differentiated species still warrant further research. As a last point, I suggest that, through the genetic structure I find along the distribution of D. aruanus, isolation by environment (IBE) does not seem to be a significant driver of genetic drift among this species (Figure 3.1 b1 not significant). The environment does not seem to affect geneflow much, a sign for putatively high phenotypic plasticity (see also Chapter 3.4). I also find high genetic richness throughout the Red Sea, which is well represented even in a single portion of one recruitment cohort (see Chapter 3.3), indicative for high and frequent admixture and connectivity within the Red Sea. The dispersive success of D. aruanus also seems evident across putative barriers driving isolation by oceanography (IBO,

Figure 3.1 b2 and b3). While, the species cannot be found within the entire Gulf of

Aden (suggesting that IBO and the b3 barrier in Figure 3.1 may hinder settlement due to lack of habitat), there is no significant sign for restricted gene flow or genetic drift between the populations of the Red Sea and Madagascar. For instance, while the Figure 3.1 b2 barrier leads to some population structure between the Red Sea and Madagascar, visible in the bayesian cluster analysis, genetic differentiation is not significant, indicative for geneflow and dispersal across oceanographic barriers

(low impact of IBE and IBO). Pomacentridae also may ecologically be less susceptible to more geologically soft barriers (Cowman & Bellwood, 2013a). 183

Perhaps more significant for this widespread species, is IBD in combination with the presence longer lasting physical barriers (Figure 3.1 b4 vs. b2), as I find the strongest differentiation between the west Indo-Pacific and the other populations on the Pacific side of the coral reef triangle (the putatively two different species) as well as some genetic structure among the Pacific/coral reef triangle populations

(Indonesia vs. Paracel Islands and Taiwan, visible in the bayesian cluster analysis).

This could be related to the benthic spawning behavior of Pomacentridae in comparison to for example Labridae and Chaetodontidae, which are pelagic spawners and additionally, exhibit longer PLDs. Thus, putatively increased dispersal may also increase the chances of vicariance (Cowman & Bellwood, 2013a). However, the putative differentiation between the west Pacific sampling sites among the study sites of D. aruanus must be interpreted with caution as only two individuals represent the Indonesian population.

More surprisingly was the finding of strong genetic differentiation among the samples of D. marginatus. It did not only reveal significant genetic structure within the Red Sea bur also evidence for the presence of fixed loci in D. marginatus inside vs. outside the Red Sea, in the Gulf of Aden. Building on preliminary analyses of genetic structure, based on 19 nuclear microsatellite (MSAT) loci (Appendix) I was not able to find strong genetic differentiation of the two populations. My conclusions were rather higher genetic richness/diversity in the population of Socotra suggesting that likely, D. marginatus had originated in the Gulf of Aden and around

Socotra, which is plausible since the Red Sea is only ca. 5 MYA while the species is 184 known to be much older (about 9-12 MYA, McCafferty et al., (2002)). From the microsatellite data, I hypothesized that D. marginatus had found its refugia in the

Gulf of Aden during environmental changes along its distribution range (e.g. caused by fluctuations in sea level), which potentially also restricted the access to the Red

Sea and changed the environmental conditions inside the Red Sea drastically.

Therefore, its population core with highest genetic richness could have been established in the Gulf of Aden and Socotra Island, from where it colonized the Red

Sea whenever conditions were favorable. However, the SNPs data (in comparison to the MSAT data) clearly shows a strong genetic break with high divergence between the two regions of the Arabian Sea: the Red Sea and the Gulf of Aden. The phylogenetic tree of the SNPs strongly suggests the presence of two distinct species.

Furthermore, the presence of fixed loci among the two is representative for speciation driven by natural selection. These two species are clearly separated by the Strait of Bab al Mandab at the entrance of the Red Sea since all specimen of

Djibouti (at the very entrance to the Red Sea) cluster together with those inside the

Red Sea, while all other specimen collected outside the Red Sea build the second cluster, without exception. Thus, I suggest that the result of genetic divergence is highly likely a result of vicariance at the strait of Bab El Mandab in combination with disruptive selection (IBD, and IBE). The environment inside vs. outside the Red Sea changes dramatically at this same geographical location at which the potential physical barrier of the Strait of Bab El Mandab is found. Therefore, it is difficult to distinguish, which barrier for geneflow was the strongest for the species: the environmental change or the land barrier. However, I suggest one can infer some 185 discrimination by the following observation. Inside the Red Sea there is also clear population structure: D. marginatus exhibits a strong genetic break at 20°N (the beginning of the Farasan Banks, see b1 in Figure 3.1). Thus, one could hypothesize that the endemic species is generally more sensitive to environmental barriers to gene flow, at least among the sister species in the Red Sea. More general one could say that this Arabian Sea’s Dascyllus marginatus species-complex is more prone to be affected by barriers to geneflow causing IBD and IBE. Additionally, the fact that there is no differentiation within the Gulf of Aden by the suggested b3 barrier

(Figure 3.1) may hint towards IBO being less significant for the endemic species.

Nevertheless, I would more likely interpret the latter irrelevance of barrier b3 as an effect of the fact that the set of environmental conditions present in the Gulf of Aden

(which potentially lead to IBO in other species) are those in which D. marginatus of the Gulf of Aden has evolved and is therefore already adapted to. All in all, I hypothesize that the endemic species is more prone to be affected by environmental barriers (IBE, b1 and b2) than its widespread counterpart. This may be due to its higher susceptibility to natural selection and/or higher mutation rates as an adaptive strategy. In other words, D. marginatus shows higher signs of selective pressure shaping its genetic structure and genetically imprinted adaptation as a response to natural selection, while D. aruanus is more phenotypically and ecologically plastic and thus, exhibits less genetic differentiation across different environmental barriers (see also Chapter 3.4). The widespread species may also have higher dispersal potential or a higher dispersive behavior of its larvae.

Herewith, it may be able to overcome barriers quicker, acting against genetic drift as 186 soon as barriers open up again. In contrast, endemic species may have evolved to have stronger self-recruitment and larval retention, probably also as a result of the stronger effect of selective pressure, faster adaptive response to the environment, and low phenotypic plasticity. This would accelerate genetic drift and isolation between populations among endemics in comparison to widespread species.

3.4.2 Mito-Nuclear Discrepancies in the Genetic Structure of Dascyllus marginatus of the Arabian Sea

The results of genetic structure among D. marginatus in the Arabian Sea were unexpected and highly interesting. Initially, and from preliminary analysis with nuclear microsatellite (MSAT) markers, I intended to focus on the fact that D. marginatus is much older than the Red Sea but has been called a Red Sea endemic for a long time and suggest to refer to it as a Gulf of Aden or Arabian Sea endemic, which opportunistically has colonized the Red Sea and established its Red Sea population potentially about 5 MYA. However, after the analysis of the SNPs data I was convinced of the presence of two different species: a Gulf of Aden endemic and a Red Sea endemic. To further establish evidence for the divergence of two different species I therefore, sequenced the commonly used cytochrome oxidase I (COI) mitochondrial marker using specimen from inside and outside the Red Sea and to quantify the genetic distance between the two. However, this marker did not separate the two species. Nevertheless, I strongly suggest the cladogenesis of an original D. marginatus species due to vicariance and/or disruptive selection at the 187

Strait of Bab El Mandab. The high numbers of fixed loci, high Fst values, and significant genetic differentiation between the two, are hardly to be explained by other means. Additionally, I have observed differences in coloration between the two regions. Even though, different color morphs are no evidence for different species, the specimen from the Gulf of Aqaba are much darker and in general have more intensive pigmentation than the ones from the north and north central Red

Sea. They also appear to grow to much larger sizes. Furthermore, the presence of two species without genetic differentiation in the mitochondrial genome has been recently also been reported in another coral reef fish (e.g. Bernal et al. 2016) and is thus, not mutually exclusive. While the mitochondrial DNA has higher mutation rates than the nuclear DNA (due to smaller effective population size of the former) and should thus, reveal speciation faster and more clearly than nuclear DNA, there are several evolutionary processes, which can be responsible for the herein observed mito-nuclear discrepancies (mito-nuclear discordance Rocha et al., 2008;

Tavera et al., 2012; Toews and Brelsford, 2012). For instance recent hybridization between two species can lead to introgression of the mtDNA homogenizing mitochondrial lineages while having little effect on the diverged nuclear genome

(Toews & Brelsford, 2012; Bernal et al., 2016). Independent lineages can thus emerge despite recurrent hybridization events through the combined effect of mtDNA introgression and limited nuclear DNA introgression. Hybridization has also been increasingly reported in several coral reef fishes (butterflyfishes (Pyle &

Randall, 1994; Montanari et al., 2012), anemonefishes (Litsios & Salamin, 2014;

Gainsford et al., 2015), grunts (Rocha et al., 2008; Bernardi et al., 2013), hamlets 188

(Puebla et al., 2007), angelfishes (Schultz et al., 2007; Gaither et al., 2014), surgeonfishes (Dibattista et al., 2011; DiBattista et al., 2016), wrasses (Pyle &

Randall, 1994; Yaakub et al., 2007), groupers (Harrison et al., 2014) and damselfishes (Coleman et al., 2014). In cases where hybridization is frequent, introgressed lineages may be widespread (e.g. genus Centropyge (Gaither et al.,

2011; DiBattista et al., 2012)). Thus, introgressive hybridization can be easily detected in mtDNA due to the lack of recombination (reviewed by Funk and Omland,

2003). Furthermore, these introgression events can lead to “islands” of genomic differentiation in the recombining nuclear genome (Wu & Ting, 2004; Bird et al.,

2012), which could be represented by the amount of fixed loci observed between the two D. marginatus of the Arabian Sea. Such scenarios are also more frequent under the effect of selective pressure under for example the presence of environmental differences (Gaither et al., 2015), as those present at the transition between the Red Sea and the Gulf of Aden. Putatively induced disruptive selection at even a small number of genes can eventually reduced gene flow and eventual reproductive isolation (Rundle & Nosil, 2005) and speciation.

The Red Sea has been acknowledged as a hotspot for biodiversity among coral reefs worldwide (Roberts et al., 1992; DiBattista et al., 2013, 2015) despite its peripheral location. In DiBattista et al. (2015) the origin of high endemism in the Red Sea is thoroughly discussed. Additionally, they refer to the Gulf of Aden as an extension of the Red Sea biogeographic province, since some species are distributed not only within but also outside the Red Sea, and referred to as the Arabian Sea. However, my data suggests that this should be taken with caution. For many species with 189 distribution ranges along the Arabian Sea, much data and samples are still missing.

Genetic studies like the one herein may reveal that some of these species may actually represent sister species with one inside and the other outside the Red Sea.

Similar to the case of the Marquesas’ Islands (Gaither et al., 2010, 2015), the high endemism rate of endemism at the Red Sea and potentially Gulf of Aden is difficult to explain by geographic isolation alone. Instead, unique and contrasting environmental conditions between the two regions may be fueling the evolutionary pump for endemic species. The Marquesas are known for high endemism (Randall,

2005) and lack fringing reefs and lagoons and are exposed to highly variable sea temperatures (from 26 to 30 °C; and 15–18 °C due to upwelling (Randall & Earle,

2000)), alike the Gulf of Aden (Currie et al., 1973; Smeed, 1997; Kemp, 2000; Jung et al., 2001). Thus, colonists that arrive in the Gulf of Aden and/or the Red Sea and vice-versa are likely undergo strong selection in these differing ecological conditions, leading to differentiation of lineages and speciation (caused by both, selection and genetic drift).

Moreover, and not yet previously discussed for the Arabian Sea (in e.g. DiBattista et al., (2015)) is putative introgressive hybridization between cryptic sister-species of inside and outside the Red Sea, as a driver of the high biodiversity and endemism found in this peripheral region. Most recent studies on fishes have discovered hybridization between divergent lineages increase the source of genetic variation, which through recombination speed up the evolution of many new species (Meier et al., 2017). Thus, hybridization between diverging lineages under certain environmental conditions, which offer diverse ecological opportunities, may 190 increase speciation through adaptive radiation. The Red Sea’s geological history seems like a perfect location for such evolutionary processes.

3.4.3 Pelagic Larval Durations (PLDs) of Dascyllus in the Red Sea

The Red Sea displays a sharp environmental gradient of sea surface temperature

(SST), chlorophyll a (CHLA), and turbidity (Raitsos et al., 2013). Here, the assessment of PLDs amongst the three present damselfishes of the genus Dascyllus revealed that the endemic damselfish species D. marginatus consistently had shorter

PLDs than the two widespread species D. aruanus and D. marginatus. The latter two are described to be more ecologically versatile species and among these, D. trimaculatus showed the longest PLD and the global average for this species is even longer (Dataset S1, Luiz et al., 2013)). D. trimaculatus is also the least ecological specialized thus, suggesting there may be a link between ecology, biogeography, and the length and plasticity of PLDs among Dascyllus.

For my three study species, a gradual and consistent decrease in mean PLD with decreasing latitude was measured, demonstrating a significant link between the pelagic larval duration (PLD) of Dascyllus species and environmental parameters such as mean sea surface temperature (SST) and chlorophyll a (CHLA). Moreover, intra-specific differences in PLD between sampling regions along this gradient were greater in the endemic D. marginatus than the other two widespread species (D. aruanus and D. trimaculatus), suggesting that the endemic species might be more responsive to environmental differences within its habitat, potentially related to a 191 higher degree of specialization. First measurements for a broad geographic range of the endemic D. marginatus (19.8 ± 2.9 d) indicate a consistently lower PLD at all sites in comparison to PLDs of D. aruanus and D. trimaculatus. Additionally, PLDs within species were overall spatially and potentially also temporally highly plastic, which should be considered when evaluating correlations of PLDs and other parameters.

From my data, I find it important to point out the extreme plasticity observed in

PLDs of Dascyllus in relation to the environment and how such variations can radically change the outcome of the study. For instance, in this study, D. marginatus consistently had a shorter PLD than D. aruanus at the same geographic site (i.e., spatially coherent data) and we thus find a positive correlation between distribution range and dispersal potential. However, if our reference PLDs for the two species within the Red Sea were not spatially coherent but we instead use a reference PLD for D. marginatus from the northern Red Sea (GAQ; the only measurement available in previous literature) and a reference PLD for D. aruanus from the southern Red

Sea (SRS), we would conclude the opposite, that is, we would find that the endemic species has a longer PLD than the widespread species. Broad comparisons of PLDs to range sizes may therefore be complicated by large intraspecific variations in PLD among sampling locations (see also Victor, 1991).

The relationship between dispersal potential and biogeographic distribution of marine species has long been an open discussion and many studies have approached this question using PLD data (Victor & Wellington, 2000; Zapata & 192

Herrón, 2002; Bradbury et al., 2008; Weersing & Toonen, 2009; Selkoe & Toonen,

2011; Faurby & Barber, 2012). However, most of these studies have found either weak, non-linear, or no correlation between distribution ranges and PLDs at different geographic scales and for a wide number of species. There can be several reasons for conflicting results among studies on the relationship between dispersal potential and biogeography of marine species.

Appropriately designed studies may still have the potential to provide insight into the unresolved relationship of PLD and biogeography. The cautious collection of adequate temporal and spatial data is crucial. A study focused on long- term/evolutionary processes like biogeography in relation to contemporary demographic data like PLDs, would ideally meet the following criteria: A) address closely related species to eliminate variation related to phylogenetic effects, B) sample from standing (adult) populations to capture as much temporal variation as possible, also to ensure that the data represent the ‘effective’ population's PLD

(from adults from different recruitment cohorts) and/or sample new cohorts over a long temporal period, ideally capturing anomalous years, and C) sample with spatial coherency in order to isolate extrinsic environmental variables from intrinsic biological differences.

The genus Dascyllus may offer an opportunity to address such questions. The genus has an endemic representative almost in every biogeographic province in the Indo-

Pacific, which will allow future comparative studies using this genus to further assess the relationship between PLD and biogeography over a wide geographic 193 range. Other genera (e.g., Chaetodon, Thalassoma, Eviota, etc.) may provide similar opportunities for congeneric comparison of endemic vs. widespread species. I further stress the importance of analyzing PLD data accounting for sources of temporal and spatial variation of PLDs at small scales.

Much work remains to be done to resolve the many potential drivers of plasticity in the PLDs of larval reef fishes at both ecological and evolutionary timescales. The role of various environmental factors and the inclusion of genetic, biological, and ecological data are a prime target for future studies, particularly in connection with understanding how ecological specialization of a species interacts with these factors. 3

V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 3 3

–xxx 194 –xxx 3.5 APPENDIX 3 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx

30 3.5.1 Sample Collection of Dascyllus marginatus30° and D. aruanus Within the Red Sea IA 3

N ° IO (RS) For a First Assessment of Genetic Diversity Using Nuclear Microsatellite xxx 20 – V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 3

Markers (MSATs) V.S.N. Robitzch et al. / Marine3 Pollution Bulletin xxx (2015) xxx

10 10° –xxx

V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx 3

–xxx D. aruanus GAQ V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx GAQGAQ

30° 30 D. marginatus NCRS ● SCRSSCRS SRSSRS NUMV.S.N. RobitzchV.S.N. et al. / Marine Robitzch Pollution Bulletin et xxx al. (2015) / Marine xxx–xxx Pollution Bulletin xxx (2015)3 xxx NUW V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx

MSH

MAT 25 25° 2000 km

RMA AKA

AFA 80 ° 100°100 120120°

AlI

20 20°

DUR CHLA [mg m-3]

KAM 15° 15 0.08 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2.5 10 30 c and Red Sea are entral RS), diamonds (SCRS:ﬁ south-

SOC . The two locations outside the Red Sea ude (°N) and longitude (°E). Additional DJI Fig. 1. Sampling sites of Dascyllus marginatus, D. aruanus, and D. trimaculatus. A total of 20 sites are grouped into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- central RS), or sails (SRS:Raitsos southern et al. (2013) RS) according to the respective region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea 200 km 3 ) concentrations of the Indo-Paci scale approx200 1:20,000,000 km are represented by a white− nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional

200 km information on exact number of samples and coordinates for each site can be found in Table 1. The chlorophyll a (CHLA, in mg m−3) concentrations of the Indo-Paciﬁc and Red Sea are 10° 10 displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations. 0 200 400 km c and Red Sea are a (CHLA, in mg m ﬁ 35° 40° 45° 50° entral RS), diamonds (SCRS: south- . The two locations outside the Red Sea . A total of 20 sites are grouped. The into chlorophyll four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- App. 3.1: Sampling sites inside the central and southern Red Sea (RS) for Table 1 Dascyllus ude (°N) and longitude (°E). Additional

Fig. 1. Sampling sites of Dascyllus marginatus, D. aruanus, and D. trimaculatus. A total of 20 sites are groupedD. trimaculatus into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- aruanus (represented by triangles) and Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by whiteD. marginatus , and circles (GAQ: Gulf of Aqaba), squares (NCRS:and from Socotra and Djibouti north-central RS), diamonds (SCRS: south- central RS), or sails (SRS: southern RS) according to the respective region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea are represented by a white nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago,, D. aruanus respectively). Map axes indicate latitude (°N) and longitude (°E). Additional Raitsos et al. (2013) −3 3 for D. marginatusinformation on exact number(represented of samples and coordinates for each site can by be found incircles). Table 1. The chlorophyll a (CHLA, The in mg m samples ) concentrations of the Indo-Paci were ﬁc and Red Sea are collected for the − ) concentrations of the Indo-Paci displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations. assessment Fig. 1. Sampling of geneti sitesc of diversity DascyllusDascyllus marginatus marginatus and differentiation , D. aruanus, and D. using trimaculatus nuclear . A total microsatellite of 20 sites are grouped into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo-

markers Australian (MSATs). Archipelago; A total andSampling of sites IO: of four Indian and Ocean). five Those regions inside the (at Red several Sea are represented reef sites) a (CHLA, by inwhite mg were m circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- Fig. 1. central RS), or sails (SRS:Australian southern Archipelago; and IO: Indian RS) Ocean). according Those inside the Red Sea are to represented the by respective white circles (GAQ: Gulf of Aqaba), region. squares (NCRS: The north-c regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea sampled for D. aruanus and central RS), orD. marginatus sails (SRS: southern RS) according to the respective respectively. Chlorophyll region. The regions were assigned in consensus with those in a (CHLA, in mg are represented by a white nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latit are represented by a whiteinformation nabla on exact number and of samples a white and coordinatestriangle for each site can be found (IO: in Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional m−3) concentrations are displayed from thedisplayed NASA Giovanni website (http://oceancolor.gsfc.nasa.gov)from to visualize the . approximate A total of differences 20NASA sites between are locations. grouped. Giovanni The into chlorophyll four regions within website the Red Sea (RS) and two outside the Red Sea (IA: Indo- information on exact number of samples and coordinates for each site can be found in Table 1. The chlorophyll a (CHLA, in mg m−3) concentrations of the Indo-Pacific and Red Sea are (http://oceancolor.gsfc.nasa.gov) to visualize approximate Table 1 environmental displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov)D. trimaculatus Fig. 2. Distribution of fish sizes (total length, to invisualize mm) of Dascyllus marginatus approximate(left panel) and D. aruanus (right differences panel) used for PLD measurements between from each site.locations. Sites are given with three letter abbreviations on the x-axis (reef names can be found in Table 1). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles differences between locations. , and represent mean sizes and the whiskers display the maximum and minimum sizes of the fishes. (right panel) used for PLD measurements from each site. Sites are given with three

Please cite this article as: Robitzch,D. aruanus V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of , D. aruanus damselﬁshes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045

ﬁshes. (left panel) and

). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles Table 1 Dascyllus marginatus Dascyllus marginatus http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 Fig. 2. Distribution of ﬁsh sizes (total length, in mm) of Dascyllus marginatus (left panel) and D. aruanus (right panel) used for PLD measurements from each site. Sites are given with three letter abbreviations on the x-axis (reef names can be found in Table 1). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles represent mean sizes and the whiskers display the maximum and minimum sizes of the ﬁshes.

Please cite this articleSampling as: Robitzch, sites of V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045fish sizes (total length, in mm) of Fig. 1. entral RS), diamonds (SCRS: south- Australian Archipelago; and IO: Indian Ocean). Those insideDistribution the Red of Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-c Fig. 2. fic and Red Sea are central RS), or sails (SRS: southern RS) according to theletter respective abbreviations onregion. the x-axis The (reef names regions can be foundwere in assigned in consensus with those in . The two locations outside the Red Sea are represented by a white nabla and a white trianglerepresent (IO: Indian mean sizes Ocean and the whiskers and IA: display Indo-Australian-Archipelago, the maximum and minimum sizes of the respectively). Map axes indicate latit fishes in the Red Sea, Marine Pollution Bulletin (2015), ude (°N) and longitude (°E). Additional information on exact number of samples and coordinatesPlease for cite each thisarticle site can as: Robitzch, be found V.S.N., in et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of entral RS), diamonds (SCRS: south- displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov)damsel to visualize approximate differences between locations. Raitsos et al. (2013)

. The two locations outside the Red Sea −3 ) concentrations of the Indo-Paci c and Red Sea are Raitsos et al. (2013) ude (°N) and longitude (°E).ﬁ Additional a (CHLA, in mg m . A total of 20 sites are grouped. A total into four of regions 20 sites within the are Red grouped Sea (RS) and two into outside four the Red regions Sea (IA: Indo- within the Red Sea (RS) and two outside the Red Sea (IA: Indo- −3 D. trimaculatus (right) panel) concentrations used of for the PLD Indo-Paci measurements from each site. Sites are given with three D.,trimaculatus and . The chlorophyll , and Table 1 D. aruanus(CHLA, in mg m , D., D. aruanus a

. The chlorophyll fishes. Dascyllus marginatus Table(left 1 panel) and Dascyllus marginatus Sampling sites of Fig. 1. ). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles Sampling sites of Table 1 Fig. 1. Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea areDascyllus represented marginatus by white circles (GAQ: Gulf of Aqaba), squareshttp://dx.doi.org/10.1016/j.marpolbul.2015.11.045 (NCRS: north-c Australian Archipelago; andcentral IO: RS), Indian or sails (SRS: Ocean). southern RS)Those according to inside the respective the region. Red The Sea regions are were represented assigned in consensus by with white those in circles (GAQ: Gulf of Aqaba), squares (NCRS: north-c central RS), or sails (SRS: southern RS) according to the respective region. The regions were assigned in consensus with those in are represented by a whiteare nabla represented and by a white white nabla andtriangle a white triangle (IO: (IO: Indian Indian Ocean Ocean and IA: Indo-Australian-Archipelago, and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latit respectively). Map axes indicate latit entral RS), diamonds (SCRS: south- information on exact number of samples and coordinates for each site can be found in informationfi onsh exact sizes number (total of samples length, and in coordinates mm) of for each site can be found in displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations. fic and Red Sea are displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations. . The two locations outside the Red Sea Distribution of Fig. 2. ude (°N) and longitude (°E). Additional letter abbreviations on the x-axis (reef names can be found in represent mean sizes and the whiskers display the maximum and minimum sizes of the Raitsos et al. (2013) Fig. 2. Distribution of fish sizes (total length, in mm) of Dascyllusfi marginatusshes in the Red(left Sea, Marine panel) Pollution and D. Bulletin aruanus (2015),(right panel) used for PLD measurements from each site. Sites are given with three letter abbreviations on the x-axis (reef names can be foundPlease in citeTable this article 1). The as: Robitzch, sites are V.S.N., in latitudinal et al., Productivity order and from sea surface north temperature to south arefrom correlated left to with right, the respectivelypelagic larval duration along of the axis. Black circles damsel −3 represent mean sizes and the whiskers display the maximum and minimum sizes of the fishes. ) concentrations of the Indo-Paci

Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of a (CHLA, in mg m damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045. A total of 20 sites are grouped into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- D. trimaculatus , and . The chlorophyll Table 1 , D. aruanus (right panel) used for PLD measurements from each site. Sites are given with three D. aruanus Fig. 1. Sampling sites of Dascyllus marginatus, D. aruanus, and D. trimaculatus. A total of 20 sites are grouped into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- (left panel) and Dascyllus marginatus fishes. Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ:). The sites Gulf are in latitudinal of Aqaba), order from north squares to south from left(NCRS: to right, respectively north-c along theentral axis. Black circles RS), diamonds (SCRS: south- Dascyllus marginatusTable 1 central RS), or sails (SRS: southern RS) according to theSampling respective sites of region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea are represented by a white nabla and a white triangleFig. 1. (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively).http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 Map axes indicate latitude (°N) and longitude (°E). Additional fish sizes (total length, in mm) of information on exact number of samples and coordinatesAustralian Archipelago; for each site and can IO: Indian be found Ocean). in ThoseTable inside 1. The the Red chlorophyll Sea are representeda (CHLA, by white in mg circles m− (GAQ:3) concentrations Gulf of Aqaba), squares of the (NCRS: Indo-Paci north-cfic and Red Sea are Distribution of displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov)central RS), or sails (SRS: southernFig. 2. toRS) visualize according to approximate the respective region. differences The regions between were assigned locations. in consensus with those in letter abbreviations on the x-axis (reef names can be found in are represented by a white nablarepresent and mean a white sizes and triangle the whiskers display(IO: theIndian maximum Ocean and minimum and sizes IA: ofIndo-Australian-Archipelago, the respectively). Map axes indicate latit information on exact number of samples and coordinates for each site can be found in Please citefishes this inarticle the Red as: Sea,Robitzch, Marine V.S.N., Pollution et al., Bulletin Productivity (2015), and sea surface temperature are(right correlated panel) used for PLD withmeasurements the from pelagic each site. Sites are larval given with three duration of displayed from the NASA Giovannidamsel website (http://oceancolor.gsfc.nasa.gov) to visualize approximateD. aruanus differences between locations. (left panel) and Dascyllus marginatus). The sites are in latitudinalfishes. order from north to south from left to right, respectively along the axis. Black circles Table 1

ﬁsh sizes (total length, in mm) of Distribution of Fig. 2. letter abbreviations on the x-axis (reef names can be found in http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 represent mean sizes and the whiskers display the maximum and minimum sizes of the

Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of ﬁshes in the Red Sea, Marine Pollution Bulletin (2015), damsel

(right panel) used for PLD measurements from each site. Sites are given with three D. aruanus

(left panel) and ﬁshes. ). The sites are in latitudinal order from north to south from left to right, respectively along the axis. Black circles Dascyllus marginatusTable 1

http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 ﬁsh sizes (total length, in mm) of

Distribution of Fig. 2. Distribution of ﬁsh sizes (total length, in mm) of Dascyllus marginatus (left panel)Fig. 2. and D. aruanus (right panel) used for PLD measurements from each site. Sites are given with three letter abbreviations on the x-axis (reef names can be found in Table 1). The sites areletter in latitudinal abbreviations order on the fromx-axis (reef north names to south can be found from in left to right, respectively along the axis. Black circles represent mean sizes and the whiskers display the maximum and minimum sizes ofrepresent the ﬁshes. mean sizes and the whiskers display the maximum and minimum sizes of the

Please cite this article as: Robitzch, V.S.N., et al., Productivity and seaPlease surface citeﬁshes this temperature in article the Red as: Sea, Robitzch, Marine are correlated V.S.N., Pollution et al., Bulletin Productivity with (2015), the pelagic and sea surface larval temperature duration are of correlated with the pelagic larval duration of damsel damselﬁshes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 195

3.5.2 Amplification and Fragment Analysis of MSATs

DNA was extracted using the DNA extraction kit manufactured by MACHEREY-

NAGEL (NucleoSpin®96 Tissue). Fluorescent labeled microsatellite primers were ordered for the published primers designed for D. aruanus (Fauvelot et al., 2009), D. marginatus (Alpermann et al., 2013), D. trimaculatus (Bernardi et al., 2012) and

Abudefduf luridus (Carvalho et al., 2000). PCR conditions for multiplex reactions with over 25 of the former microsatellite markers were adjusted as follows: 95°C for

15 min, continued by 4 touchdown cycles of 94°C for 30 s, 57°C (-0.5°C/cycle) for

90s, 72°C for 60s, and followed by 23 cycles of 94°C for 30 s, 55°C for 90s, 72°C for

60s, and a final extension at 72°C for 20 min to assure 3’ adenylation of PCR products. Multiplexed PCR products were diluted 1 to 30 with Milli-Q water (EMD

Millipore Corporation) prior to fragment analysis. Fragment lengths were detected on a Sanger ABI 3730XL. Geneious 7.1.7 (Biomatters Ltd) was used to genotype alleles and export the genotypes for further analysis. Basic Hardy-Weinberg and statistical test were performed using GENEPOP 4.1 (Raymond & Rousset, 1995) and

GenAlEx 6.5 (Peakall & Smouse, 2006, 2012). STRUCTURE 2.2.3 (Pritchard et al.,

2000; Falush et al., 2003, 2007; Hubisz et al., 2009) was used to assess population structure based on Bayesian multi-locus clustering. The number of putative subpopulations (K) based on the STRUCTURE analysis was calculated with

STRUCTURE HARVESTER’s Evanno Method (Earl & VonHoldt, 2012).

196

3.6 REFERENCES

Alpermann T.J., Wicker F., Nanninga G., Plieske J., & Berumen M.L. (2013) Development of microsatellite markers for the Arabian damselfish, Dascyllus marginatus (Rüppell, 1829) (Actinopterygii, Pomacentridae). Molecular Ecology Resources, 1–9. Barber P.H., Cheng S.H., Erdmann M. V., Tenggardjaja K., & Ambariyanto (2011) Evolution and conservation of marine biodiversity in the Coral Triangle: insights from stomatopod Crustacea. Evolutionary Biology. Berdahl A., Torney C.J., Schertzer E., & Levin S.A. (2015) On the evolutionary interplay between dispersal and local adaptation in heterogeneous environments. Evolution.

Bernal M.A., Gaither M.R., Simison W.B., & Rocha L.A. (2016) Introgression and selection shaped the evolutionary history of sympatric sister-species of coral reef fishes (genus: Haemulon). Molecular Ecology. Bernardi G., Beldade R., Holbrook S.J., & Schmitt R.J. (2012) Full-sibs in cohorts of newly settled coral reef fishes. PloS one, 7, e44953. Bernardi G., Noguchi R., Anderson A.B., Floeter S.R., & Leite Ferreira C.E. (2013) Sargo Amarelo, a traditionally recognized hybrid between two species of Brazilian reef fishes. Marine Biodiversity, 43, 255–256. Berumen M.L., Hoey A.S., Bass W.H., Bouwmeester J., Catania D., Cochran J.E.M., Khalil M.T., Miyake S., Mughal M.R., Spaet J.L.Y., & Saenz-Agudelo P. (2013) The status of coral reef ecology research in the Red Sea. Coral Reefs, 32, 737–748. Bird C.E., Fernandez-Silva I., Skillings D.J., & Toonen R.J. (2012) Sympatric Speciation in the Post “‘Modern Synthesis’” Era of Evolutionary Biology. Evolutionary Biology, 39, 158–180. Bird C.E., Holland B.S., Bowen B.W., & Toonen R.J. (2007) Contrasting phylogeography in three endemic Hawaiian limpets (Cellana spp.) with similar life histories. Molecular Ecology, 16, 3173–3186. Bohonak A.J. (1999) Dispersal, gene flow, and population structure. The Quarterly review of biology, 74, 21–45. Borsa P., Sembiring A., Fauvelot C., & Chen W.-J. (2014) Resurrection of Indian Ocean humbug damselfish , Dascyllus (Forsskal) from synonymy with its Pacific Ocean sibling, Dascyllus aruanus (L.). Comptes Rendus Biologies, 337, 709–716. Bradbury I.R., Laurel B., Snelgrove P.V.R., Bentzen P., & Campana S.E. (2008) Global patterns in marine dispersal estimates: the influence of geography, taxonomic category and life history. Proceedings. Biological sciences / The Royal Society, 275, 1803–1809. Carpenter K.E., Barber P.H., Crandall E.D., Ablan-Lagman M.C., Mahardika G.N., 197

Manjaji-Matsumoto B.M., Juinio-Meñez M.A., Santos M.D., Starger C.J., & Toha A.H. a. (2011) Comparative Phylogeography of the Coral Triangle and Implications for Marine Management. Journal of Marine Biology, 2011, 1–14. Carvalho M.C., Streiff R., Guillemaud T., Afonso P., Santos R.S., & Cancela M.L. (2000) Isolation and characterization of polymorphic microsatellite markers in Abudefduf luridus (Pisces : Pomacentridae). Molecular Ecology, 9, 993–994. Catchen J.M., Amores A., Hohenlohe P., Cresko W., & Postlethwait J.H. (2011) Stacks: Building and Genotyping Loci De Novo From Short-Read Sequences. Genes, Genomes, Genetics, 1, 171–182. Choat J.H., Klanten O.S., Clements K.D., van Herwerden L., Robertson D.R., & Clements K.D. (2012) Pattern and process in the evolutionary history of parrotfishes (Family Labridae). Biological Journal of the Linnean Society, 107, 529–557. Coleman R., Gaither M.R., Kimokeo B., Stanton F.G., Bowen B.W., & Bowen B.W. (2014) Large-scale introduction of the Indo-Pacific damselfish Abudefduf viagiensis into Hawai’i promotes genetic swamping of the endemic congener A. abdominalis. Molecular Ecology, 23, 5552–5565. Cooper J.W., Smith L.L., & Westneat M.W. (2009) Exploring the radiation of a diverse reef fish family: phylogenetics of the damselfishes (Pomacentridae), with new classifications based on molecular analyses of all genera. Molecular phylogenetics and evolution, 52, 1–16. Cowman P.F. & Bellwood D.R. (2013a) Vicariance across major marine biogeographic barriers: temporal concordance and the relative intensity of hard versus soft barriers. Proceeding of the Royal Society Biological Siences, 280, 1–8. Cowman P.F. & Bellwood D.R. (2013b) The historical biogeography of coral reef fishes: global patterns of origination and dispersal. Journal of Biogeography, 40, 209–224. Crispo E., Bentzen P., Reznick D.N., Kinnison M.T., & Hendry A.P. (2006) The relative influence of natural selection and geography on gene flow in guppies. Molecular Ecology, 15, 49–62. Currie R.I., Fisher A.E., & Hargreaves P.M. (1973) Arabian Sea Upwelling. The Biology of the Idian Ocean pp. 37–52. Cushman S.A., Mckelvey K.S., Hayden J., & Schwartz M.K. (2006) Gene Flow in Complex Landscapes: Testing Multiple Hypotheses with Causal Modeling. The American naturalist, 168, 486–499. DiBattista J.D., Berumen M.L., Gaither M.R., Rocha L.A., Eble J.A., Choat J.H., Craig M.T., Skillings D.J., & Bowen B.W. (2013) After continents divide: comparative phylogeography of reef fishes from the Red Sea and Indian Ocean. Journal of Biogeography, 40, 1170–1181. 198

Dibattista J.D., Craig M.T., Rocha L.A., Feldheim K.A., & Bowen B.W. (2012) Phylogeography of the Indo-Pacific butterflyfishes, Chaetodon meyeri and Chaetodon ornatissimus: sister species reveal divergent evolutionary histories and discordant results from mtDNA and microsatellites. Journal of Heredity, 103, 617–629. Dibattista J.D., Feldheim K.A., & Bowen B.W. (2011) Microsatellite DNA markers to resolve population structure and hybridization of two closely related surgeonfish species, Acanthurus nigricans and Acanthurus leucosternon. Conservation Genetics Resources, 3, 159–162. DiBattista J.D., Gaither M.R., Hobbs J.A., Saenz-Agudelo P., Piatek M.J., Bowen B.W., Rocha L.A., Choat J.H., McIlwain J.H., Priest M.A., Sinclair-Taylor T.H., & Berumen M.L. (2017) Comparative phylogeography of reef fishes from the Gulf of Aden to the Arabian Sea reveals two cryptic lineages. Coral Reefs. DiBattista J.D., Randall J.E., & Bowen B.W. (2012) Review of the round herrings of the genus Etrumeus (Clupeidae: Dussumieriinae) of Africa , with descriptions of two new species. Cybium, 36, 447–460. DiBattista J.D., Roberts M.B., Bouwmeester J., Bowen B.W., Coker D.J., Lozano-cort D.F., Choat J.H., Gaither M.R., Hobbs J.A., Maha T., Kochzius M., Myers R.F., Paulay G., & Vanessa S.N. (2015) A review of contemporary patterns of endemism for shallow water reef fauna in the Red Sea. Journal of Biogeography, 1–17. DiBattista J.D., Whitney J., Craig M.T., Hobbs J.A., Rocha L.A., Feldheim K.A., Berumen M.L., & Bowen B.W. (2016) Surgeons and suture zones: Hybridization among four surgeonfish species in the Indo-Pacific with variable evolutionary outcomes. Molecular Phylogenetics and Evolution, 101, 203–215. Doherty P.J., Planes S., & Mather P. (1995) Gene flow and larval duration in seven species of fish from the Great Barrier Reef. Ecology, 76, 2373–2391. Duputié A. & Massol F. (2013) An empiricist’s guide to theoretical predictions on the evolution of dispersal. Interface Focus, 3, 20130028. Earl D.A. & VonHoldt B.M. (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources, 4, 359–361. Falush D., Stephens M., & Pritchard J.K. (2003) Inference of Population Structure Using Multilocus Genotype Data : Linked Loci and Correlated Allele Frequencies. Genetics, 164, 1567–1587. Falush D., Stephens M., & Pritchard J.K. (2007) Inference of population structure using multilocus genotype data: dominant markers and null alleles. Molecular Ecology Notes, 7, 574–578. Faurby S. & Barber P.H. (2012) Theoretical limits to the correlation between pelagic larval duration and population genetic structure. Molecular Ecology, 21, 3419– 3432. 199

Fauvelot C., Smith-Keune C., Jerry D.R., Buston P.M., & Planes S. (2009) Isolation and characterization of 16 microsatellite loci in the humbug damselfish, Dascyllus aruanus (family Pomacentridae). Molecular Ecology Resources, 9, 651–653. Fessler J.L. & Westneat M.W. (2007) Molecular phylogenetics of the butterflyfishes (Chaetodontidae): Taxonomy and biogeography of a global coral reef fish family. Molecular phylogenetics and evolution, 45, 50–68. Fouquet A., Noonan B.P., Rodrigues M.T., Pech N., Gilles A., & Gemmell N.J. (2012) Multiple Quaternary Refugia in the Eastern Guiana Shield Revealed by Comparative Phylogeography of 12 Frog Species. SYSTEMATIC BIOLOGY, 61, 461–489. Froukh T.J. & Kochzius M. (2007) Genetic population structure of the endemic fourline wrasse (Larabicus quadrilineatus) suggests limited larval dispersal distances in the Red Sea. Molecular Ecology, 16, 1359–1367. Funk D.J. & Omland K.E. (2003) Species-Level Paraphyly and Polyphyly: Insights From Animal Mitochondrial DNA. Annual Review of Ecology, Evolution, and Systematics, 34, 397–423. Gainsford A., van Herwerden L., Jones G.P., VanHerwerden L., & Jones G.P. (2015) Hierarchical behaviour, habitat use and species size differences shape evolutionary outcomes of hybridization in a coral reef fish. Journal of Evolutionary Biology, 28, 205–222. Gaither M.R., Bernal M.A., Coleman R.R., Bowen B.W., Jones S.A., Simison W.B., & Rocha L.A. (2015) Genomic signatures of geographic isolation and natural selection in coral reef fishes. Molecular Ecology, 24, 1543–1557. Gaither M.R., Bowen B.W., Bordenave T.-R., Rocha L.A., Newman S.J., Gomez J.A., van Herwerden L., & Craig M.T. (2011) Phylogeography of the reef fish Cephalopholis argus (Epinephelidae) indicates Pleistocene isolation across the Indo-Pacific Barrier with contemporary overlap in The Coral Triangle. BMC Evolutionary Biology, 11, 189. 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. Gaither M.R., Toonen R.J., Robertson D.R., Planes S., & Bowen B.W. (2010) Genetic evaluation of marine biogeographical barriers: perspectives from two widespread Indo-Pacific snappers (Lutjanus kasmira and Lutjanus fulvus). Journal of Biogeography, 37, 133–147. Giles E.C., Saenz-Agudelo P., Hussey N.E., Ravasi T., & Berumen M.L. (2015) Exploring seascape genetics and kinship in the reef sponge Stylissa carteri in the Red Sea. Ecology and Evolution, 5, 2487–2502. Harrison H.B., Feldheim K.A., Perumal S., Williamson D.H., & Berumen M.L. (2014) Validation of microsatellite multiplexes for parentage analysis in a coral reef 200

fish (Lutjanus carponotatus, Lutjanidae). Conservation Genetics Resources, 6, 803–806. Heath M.R. (1992) Field investigations of the early life stages of marine fish. Advances in Marine Biology, 28, 1–174. Heinz S.K., Mazzucco R., & Dieckmann U. (2009) Speciation and the evolution of dispersal along environmental gradients. Evolutionary Ecology, 23, 53–70. Hellberg M.E. (2007) Footprints on water: the genetic wake of dispersal among reefs. Coral Reefs, 26, 463–473. Houde E.D. & Zastrow C.E. (1993) Ecosystem- and taxon-especific dynamic and energetics properties of larval fish assemblages. Bulletin of Marine Science, 53, 290–335. Hubisz M.J., Falush D., Stephens M., & Pritchard J.K. (2009) Inferring weak population structure with the assistance of sample group information. Molecular Ecology Resources, 9, 1322–1332. Jablonski D. (1986) Larval ecology and macroevolution marine invertebrates. Bulletin of Marine Science, 39, 565–587. Jones G.P., Almany G.R., Russ G.R., Sale P.F., Steneck R.S., Oppen M.J.H., & Willis B.L. (2009) Larval retention and connectivity among populations of corals and reef fishes: history, advances and challenges. Coral Reefs, 28, 307–325. Jung S.J.A., Ganssen G.M., & Davies G.R. (2001) Multidecadal variations in the early Holocene outflow of Red Sea Water into the Arabian Sea. Paleoceanography, 16, 658–668. Kemp J.M. (2000) Zoogeography of the coral reef fishes of the north-eastern Gulf of Aden, with eight new records of coral reef fishes from Arabia. Fauna of Arabia, 18, 293–321. Lee C.-R. & Mitchell-Olds T. (2011) Quantifying effects of environmental and geographical factors on patterns of genetic differentiation. Molecular Ecology, 20, 4631–4642. Lischer H.E.L. & Excoffier L. (2012) PGDSpider: an automated data conversion tool for connecting population genetics and genomics programs. Bioinformatics (Oxford, England), 28, 298–9. Litsios G. & Salamin N. (2014) Hybridisation and diversification in the adaptive radiation of clownfishes. BMC Evolutionary Biology, 14, 1–9. Liu S.-Y.V., Chang F.-T., Borsa P., Chen W.-J., & Dai C.-F. (2014) Phylogeography of the humbug damselfish, Dascyllus aruanus (Linnaeus, 1758): evidence of Indo- Pacific vicariance and genetic differentiation of peripheral populations. Biological Journal of the Linnean Society. Lo-Yat A., Simpson S.D., Meekan M., Lecchini D., Martinez E., & Galzin R. (2011) Extreme climatic events reduce ocean productivity and larval supply in a tropical reef ecosystem. Global Change Biology, 17, 1695–1702. 201

Lozano-Cortés D.F. & Berumen M.L. (2016) Colony size-frequency distribution of pocilloporid juvenile corals along a natural environmental gradient in the Red Sea. Marine Pollution Bulletin, 105, 546–552. 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. PNAS, 110, 16498–16502. Mastretta-Yanes A., Arrigo N., Alvarez N., Jorgensen T.H., Piñero D., & Emerson B.C. (2014) Data from: RAD sequencing, genotyping error estimation and de novo assembly optimization for population genetic inference. McCafferty S., Bermingham E., Quenouille B., Planes S., Hoelzer G., & Asoh K. (2002) Historical biogeography and molecular systematics of the Indo-Pacific genus Dascyllus (Teleostei: Pomacentridae). Molecular Ecology, 11, 1377–1392. McCormick M. & Molony B. (1995) Influence of water temperature during the larval stage on size, age and body condition of a tropical reef fish at settlement. Marine Ecology Progress Series, 118, 59–68. Mcleod I.M., Mccormick M.I., & Brooker R.M. (2015) Latitudinal variation in larval development of coral reef fishes: Implications of a warming ocean. Marine Ecology Progress Series, 521, 129–141. Mcleod I.M., Rummer J.L., Clark T.D., Jones G.P., Mccormick M.I., Wenger A.S., & Munday P.L. (2013) Climate change and the performance of larval coral reef shes: the interaction between temperature and food availability. Conservation Physiology, 1, 1–12. Meier J.I., Marques D.A., Mwaiko S., Wagner C.E., Excoffier L., & Seehausen O. (2017) Ancient hybridization fuels rapid cichlid fish adaptive radiations. Nature Publishing Group, 8, 1–11. Montanari S.R., van Herwerden L., Pratchett M.S., Hobbs J.-P.A., & Fugedi A. (2012) Reef fish hybridization: Lessons learnt from butterflyfishes (genus Chaetodon). Ecology and Evolution, 2, 310–28. Nanninga G.B., Saenz-Agudelo P., Manica A., & Berumen M.L. (2014) Environmental gradients predict the genetic population structure of a coral reef fish in the Red Sea. Molecular Ecology, 23, 591–602. Ngugi D.K., Antunes A., Brune A., & Stingl U. (2012) Biogeography of pelagic bacterioplankton across an antagonistic temperature-salinity gradient in the Red Sea. Molecular Ecology, 21, 388–405. Nosil P., Funk D.J., & Ortiz-Barrientos D. (2009) Divergent selection and heterogeneous genomic divergence. Molecular Ecology, 18, 375–402. Peakall R. & Smouse P.E. (2006) genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, 288– 295. Peakall R. & Smouse P.E. (2012) GenAlEx 6.5: genetic analysis in Excel. Population 202

genetic software for teaching and research - an update. Bioinformatics (Oxford, England), 28, 2537–9. Peterson B.K., Weber J.N., Kay E.H., Fisher H.S., & Hoekstra H.E. (2012) Double Digest RADseq: An Inexpensive Method for De Novo SNP Discovery and Genotyping in Model and Non-Model Species. PloS one, 7. Pritchard J.K., Stephens M., & Donnelly P. (2000) Inference of Population Structure Using Multilocus Genotype Data. Genetics, 155, 945–959. Puebla O., Bermingham E., Guichard F., & Whiteman E. (2007) Colour pattern as a single trait driving speciation in Hypoplectrus coral reef fishes? Proceeding of the Royal Society Biology, 274, 1265–1271. Pyle R.L. & Randall J.E. (1994) A review of hybridization in marine angelfishes (Perciformes: Pomacanthidae). Environmental Biology of Fishes, 41, 127–145. Racault M.-F., Raitsos D.E., Berumen M.L., Brewin R.J.W., Platt T., Sathyendranath S., & Hoteit I. (2015) Phytoplankton phenology indices in coral reef ecosystems: Application to ocean-color observations in the Red Sea. Remote Sensing of Environment, 160, 222–234. Raitsos D.E., Pradhan Y., Brewin R.J.W., Stenchikov G., & Hoteit I. (2013) Remote sensing the phytoplankton seasonal succession of the Red Sea. PloS one, 8, e64909. Randall J.E. (2005) Reef and shore fishes of the South Pacific: New Caledonia to Tahiti and the Pitcairn Islands. University of Hawaii Press Honolulu. Randall J.E. & Earle J.L. (2000) Annotated checklist of the shore fishes of the Marquesas Islands. Bishop Museum Press. Raymond M. & Rousset F. (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity. Riginos C. & Liggins L. (2013) Seascape Genetics: Populations, Individuals, and Genes Marooned and Adrift. Geography Compass, 7, 197–216. Roberts C.M.C., Shepherd A.A.R.D.A., & Ormond R.R.F.G. (1992) Large-scale variation in assemblage structure of Red Sea butterflyfishes and angelfishes. Journal of Biogeography, 19, 239–250. Roberts M.B., Jones G.P., Mccormick M.I., Munday P.L., Neale S., Thorrold S., Robitzch V., & Berumen M.L. (2016) Homogeneity of coral reef communities across 8 degrees of latitude in the. Marine Pollution Bulletin, 105, 558–565. Robitzch V., Banguera-Hinestroza E., Sawall Y., Al-Sofyani A., & Voolstra C.R. (2015a) Absence of genetic differentiation in the coral Pocillopora verrucosa along environmental gradients of the Saudi Arabian Red Sea. Frontiers in Marine Science, 2, 1–10. Robitzch V.S.N., Lozano-Cortés D., Kandler N.M., Salas E., & Berumen M.L. (2015b) Productivity and sea surface temperature are correlated with the pelagic larval duration of damsel fi shes in the Red Sea. Marine Pollution Bulletin. 203

Rocha L.A., Rocha C.R., Robertson D.R., & Bowen B.W. (2008) Comparative phylogeography of Atlantic reef fishes indicates both origin and accumulation of diversity in the Caribbean. BMC Evolutionary Biology, 8, 157. Rundle H.D. & Nosil P. (2005) Ecological speciation. Ecology Letters, 8, 336–352. Saenz-agudelo P., Berumen M.L., Jones G.P., Salles C., Maynard J.A., Joannides M., Barbu C.M., Saenz-agudelo P., Almany G.R., Berumen M.L., Thorrold R., Jones G.P., & Planes S. (2015) Coral reef fish populations can persist without immigration. Proceeding of the Royal Society Biology, 282, 1–9. Sawall Y., Al-Sofyani A., Banguera-hinestroza E., Voolstra R C., & Voolstra, Christian R. (2014a) Spatio-temporal analyses of Symbiodinium physiology of the coral Pocillopora verrucosa along large-scale nutrient and temperature gradients in the Red Sea. PloS one, e103179. Sawall Y., Al-Sofyani A., Kuerten B., Al-Aidaroos A.M., Hoang B.X., Marimuthu N., Khomayis H.S., Sommer U., Gharbawi W.Y., & Wahl M. (2014b) Coral Communities, in Contrast to Fish Communities, Maintain a High Assembly Similarity along the Large Latitudinal Gradient along the Saudi Red Sea Coast. Journal of Ecosystem and Ecography, 3, 1–7. Schultz J.K., Pyle R.L., DeMartini E., & Bowen B.W. (2007) Genetic connectivity among color morphs and Pacific archipelagos for the flame angelfish, Centropyge loriculus. Marine Biology, 151, 167–175. Selkoe K.A., Henzler C.M., & Gaines S.D. (2008) Seascape genetics and the spatial ecology of marine populations. Fish and Fisheries, 9, 363–377. Selkoe K.A. & Toonen R.J. (2011) Marine connectivity: a new look at pelagic larval duration and genetic metrics of dispersal. Marine Ecology Progress Series, 436, 291–305. Slatkin M. (1987) Gene Flow and the Geographic Structure of Natural Populations. Science, 236, 787–792. Smeed D. (1997) Seasonal variation of the flow in the strait of Bah al Mandab. Oceanolica Acta, 20, 773–781. Takahashi M. & Watanabe Y. (2005) Effects of temperature and food availability on growth rate during late larval stage of Japanese anchovy (Engraulis japonicus) in the Kuroshio-Oyashio transition region. Fisheries Oceanography, 14, 223– 235. Tavera J.J., Acero A., Balart E.F., & Bernardi G. (2012) Molecular phylogeny of grunts (Teleostei, Haemulidae), with an emphasis on the ecology, evolution, and speciation history of New World species. Evolutionary Biology, 12. Toews D.P.L. & Brelsford A. (2012) The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology. Via S. (2002) The Ecological Genetics of Speciation. The American naturalist, 159, S1–S7. 204

Victor B.C. (1991) Settlement Strategies and Biogeography of Reef Fishes. The Ecology of Fishes on Coral Reefs (ed. by P.F. Sale), pp. 231–260. Academic Press, Inc., San Diego, California. Victor B.C. & Wellington G.M. (2000) Endemism and the pelagic larval duration of reef fishes in the eastern Pacific Ocean. Marine Ecology Progress Series, 205, 241–248. Wang I.J. (2013) Examining the full effects of landscape heterogeneity on spatial genetic variation: a multiple matrix regression approach for quantifying geographic and ecological isolation. Evolution, 67, 3403–3411. Wang I.J. & Bradburd G.S. (2014) Isolation by environment. Molecular Ecology. Wang I.J. & Summers K. (2010) Genetic structure is correlated with phenotypic divergence rather than geographic isolation in the highly polymorphic strawberry poison-dart frog. Molecular Ecology, 19, 447–458. Weersing K. & Toonen R.J. (2009) Population genetics, larval dispersal, and connectivity in marine systems. Marine Ecology Progress Series, 393, 1–12. Wright S. (1943) Isolation by distance. Genetics, 4, 114–138. Wu C.-I. & Ting C. (2004) Genes and speciation. NATURE REVIEWS GENETICS, 5, 114–122. Yaakub S.M., Bellwood D.R., & van Herwerden L. (2007) A rare hybridization event in two common Caribbean wrasses (genus Halichoeres; family Labridae). Coral Reefs, 26, 597–602. Zapata F.A. & Herrón P.A. (2002) Pelagic larval duration and geographic distribution of tropical eastern Pacific snappers (Pisces: Lutjanidae). Marine Ecology Progress Series, 230, 295–300.

205

CHAPTER IV

SIBSHIP ANALYSIS AND GENETIC DIVERSITY OF ONE SINGLE RECRUITMENT

EVENT OF THE CORAL-DWELLING FISH DASCYLLUS ARUANUS

4.1 INTRODUCTION

Most coral reef fishes have a bipartite life cycle with a benthic adult stage, during which movement between reefs is highly restricted and, thus, dispersal essentially occurs during the pelagic larval phase. After completion of the pelagic larval duration (PLD), the larvae return to reefs to settle and are referred to as recruits.

From the moment fish larvae hatch to the day they recruit, they undergo the highest mortality rates of their lifetime and are exposed to a wide range of environmental factors that determine their fate and final location of settlement. The pelagic larval duration (PLD) can last from several days to months, depending on the species, which allows for presumed interspecific differences in dispersal potential

(Tremblay et al., 1994) and local genetic population structures (Eble et al., 2009;

Weersing & Toonen, 2009; Selkoe et al., 2010; Selkoe & Toonen, 2011). Even among species with similar PLDs, population genetic results and the species’ biogeographic ranges can be very different (Weersing & Toonen, 2009). This is hypothesized to be mainly due to the complexity of and differences in behavior throughout the ontogeny of coral reef fishes (e.g. Leis et al., 2009; Victor, 1986). For a long time, larvae of coral reef fishes were considered particle-like objects with little to no significant swimming behavior, and therefore were assumed to be mixed and

206 transported by oceanographic processes until they reached a reef for settlement

(Scheltema, 1971a, 1971b, 1971c, Strathmann, 1974, 1980; Roughgarden et al.,

1985; Warner & Hughes, 1988; Leis, 1991). Indeed, currents can be powerful transport routes for pelagic larvae (Lessios et al., 1999; Muss et al., 2001; Pineda et al., 2007; Treml et al., 2008) as well as physical barriers (Baums et al., 2005;

Thornhill et al., 2008). Nonetheless, more recent studies have found that the PLD is much more complex and unpredictable than previously thought (Thresher et al.,

1989) because many of these larvae can have extraordinary navigation skills and are deliberately perceiving and following chemical cues (Gerlach et al., 2007;

Lecchini & Nakamura, 2013), have strong and efficient vertical and horizontal swimming behavior even against currents (Leis & Mccormick, 2002), perform orientated navigation (following, e.g., sunlight (Mouritsen et al., 2013)), and possibly intentionally travel and/or settle with their kin (e.g. (Shapiro, 1983; Buston et al.,

2009; Madduppa et al., 2014); but see (Avise & Shapiro, 1986)). Thus, larval behavior has increasingly gained attention (reviewed in Kingsford et al., 2002; Leis,

2006; Montgomery et al., 2001) and might contain many answers to differences not only in recruitment patterns but also in dispersal strategies, differences in biogeographic ranges of species, and the genetic structure of populations (e.g.,

Batchelder et al., 2002; Paris and Cowen, 2004; Werner et al., 1993).

Exactly how, when, and at what scale larvae use cues to navigate still remain a mystery, but there is increasing information on the sensory and locomotive abilities of marine larvae (reviewed in (Montgomery et al., 2001; Kingsford et al., 2002; Leis,

2006)Montgomery et al. 2001, Kingsford et al. 2002, Leis 2006). Some larvae are

207 known to adjust their swimming behavior as a response to environmental signals

(Armsworth, 2000; Paris et al., 2007; Willis, 2011). How the behavior of these larvae changes during ontogeny, intra- and interspecifically, is still generally undetermined. Although much is known about some abilities such as vertical migration (e.g. Irisson et al., 2010; Paris and Cowen, 2004), swimming speeds (e.g.

Fisher et al., 2005; Leis and Carson-Ewart, 1997), and orientation behavior (e.g, Leis et al., 2011) of fish larvae at settlement stages, very little information is available on the development of sensory abilities at different and earlier larval stages (Hubbs &

Blaxter, 1986; Leis, 2010). Even at these early stages, organisms can swim actively, migrate vertically, and change their buoyancy (e.g., Cowen, 2002; Fuchs et al., 2007).

Schooling behavior may also play a major role among behavioral traits that influence species’ dispersal potential, foraging strategies, and navigation during larval phases of fishes. Thus, it can also be shaping genetics and biogeography of species of coral reef fishes. Little is known about schooling in coral reef fish larvae but from studies of migratory fishes from temperate regions one could infer that it could also have evolved in and occur during larval stages of tropical fishes to assist navigation through collective movement, especially when it comes to finding their way back from the pelagic to the reef habitat. Schooling could for example increase the ability to find food, as well as chemical and physical cues for navigation (for example, the ‘many wrongs principle’, Codling et al., 2007; Gruenbaum, 1998; Larkin and Walton, 1969; Simons, 2004). It could also decrease risk of predation

(Handegard et al., 2012) and help identify and avoid predators. Additionally, studies have shown that schooling behavior can also change according to the environment

208 and throughout ontogeny (Codling et al., 2007) and that these changes are species- specific responses. Environmental factors that drive these differences could include habitat (near-shore vs offshore), turbidity (some species might school better or worse due to visibility conditions, e.g. Ohata et al., 2014), and/or food availability

(while some species might generally prefer to school they may choose a more solitary behavior to increase successful foraging in less productive waters, e.g., cod- larvae in (Skajaa et al., 2003)). Furthermore, these environmental settings may also vary seasonally, changing the behavior strategies of the larvae depending on when and where they hatch. Thus, to test theories related to larval behavior requires resourceful and creative methods (Staaterman & Paris, 2014) because larvae are hard to track and, although produced by the thousands, are mostly cryptic and very hard to find in the vastness of the ocean’s pelagic waters. The measurement of larval dispersal of marine fauna is also very difficult, particularly if it relies on the direct sampling and/or tracking of the pelagic larval stage. The large numbers of tiny pelagic larva produced by most marine taxa require high sampling efforts resulting in typically relatively low yields and coverage and high error rates in inferences

(Strathmann, 1980, 1990). Therefore, multidisciplinary approaches and data sources (empirical, experimental, and theoretical data from physiological, biological, and most recently, genetic studies) are required to tackle these gaps of knowledge to infer theoretical proxies of larval dispersal strategies and routes (Staaterman &

Paris, 2014).

In this regard, genetic studies have increasingly demonstrated themselves as an effective, straightforward, and rather economical method (relatively low sampling

209 effort, less manpower, and time efficient). For instance, genetic fingerprinting can assist to track even single individuals of marine fish larvae (Planes et al., 2009;

Beldade et al., 2012). Tracking the dispersal potential of species provide crucial information on the behavior of species and help model which factors are establishing connectivity between populations, shaping biogeographic ranges, and ultimately determining the success, survival, distribution, and evolution of species.

Recently, studies on parentage and sibship analysis have provided a means to target such topics more accurately (e.g., (Avise & Shapiro, 1986; Bernardi et al., 2012;

Berumen et al., 2012; Schunter et al., 2013; Nanninga et al., 2015; Herrera et al.,

2016). Moreover, the reconstruction of sibship among recruiting larvae and settled juveniles to assess connectivity pathways can even be a smarter approach than parentage analysis, as it requires less sampling effort and is applicable even for species with large population sizes and adults with higher mobility, which are less site attached (see also Pinsky et al., 2017). Sampling recruits for this purpose instead of earlier larval stages is much more feasible and efficient, as the number of individuals of a cohort decreases throughout ontogeny (due to the aforementioned high mortality rates) and recruitment habitats, such as coral reefs, are spatially a much smaller dimension to investigate in comparison to pelagic waters. About half of the cohort is lost during this transition process of settlement (Caley, 1998;

Doherty et al., 2004; Almany & Webster, 2006) and early-life history mortality is likely to be selective (Perez & Munch, 2010). Thus, and as later ontogenetic stages are assessed the closer one is to assessing processes shaping realized connectivity and the population dynamics. All in all, this approach also decreases sources of error

210 in inference, making the calculation of dispersal and connectivity more accurate.

Here, one distinguishes between two types of connectivity: realized vs. potential connectivity (Watson et al., 2010). Potential connectivity is the probability of larval transport from source to destination. Realized connectivity is the number of larvae that travelled to the final destination (used to estimate e.g. larval settlement patterns), and thus a product of potential connectivity and other factors such as larval production and mortality. The assessment of genetic relatedness of cohorts of settling recruits also gives a good picture of spatial, temporal, and mechanistic patterns of larval dispersal (Planes, 2002; Selkoe et al., 2006; Veliz et al., 2006;

Buston et al., 2009). It has further provided insight into the behavior of fish larvae and the consistency of their supplying source for a certain season, location, and/or species. For instance, employing genetic markers, Bernardi et al. (2012) were the first to detect that siblings of a damselfish were able to stay together during their entire PLD, as they found that related individuals mostly recruited to the same or nearby anemones on the same night. In terms of connectivity of populations, Jones et al. (2005) and Madduppa et al. (2014) found self-recruitment among coral reef fishes of up to 65.2%, which was much higher than expected and previously thought to be unlikely for marine populations. Berumen et al. (2012) also found consistently high self-recruitment over several years, while contrastingly, Schunter et al. (2013)

(6.5% self-recruitment), Nanninga et al. (2015) (0% self-recruitment), and Herrera et al. (2016) (0.4-0.8% depending on the area, and a total of 25% potential sibling among all sites) found no to very little self-recruitment, which further displays the difficulties in assigning marine species to have either ‘open’ or ‘closed’ populations.

211

Due to different methodologies and circumstances in each study, carefully-designed approaches are necessary to compare among findings. The most accurate method to assess connectivity of populations seems lastly to be the assessment of recruitment in a localized manner and for a single species, as each reef location might have other dynamics among the processes that dictate larval supply (see also results in section

3.1, Chapter I). Therefore, it is also beneficial to identify suitable model organisms, ideally omnipresent on a wide range of coral reef ecosystems to be able to compare the results between sites and depict global trends. These model organisms should preferentially be small in size, easy to capture and not under additional fishing pressure and other anthropogenic stresses.

This study focuses on Dascyllus aruanus, a small zooplanktivorous hermaphroditic coral reef damselfish that forms colonies in which all individuals differ in their total body length (personal measurements; Gillespie, 2009) and reside in branching corals for their entire life-span (Fricke & Holzberg, 1974; Coates, 1982). This species is abundant throughout the Indo-Pacific and the Red Sea and has been targeted by many studies. Therefore, there is ample information available regarding its biology and ecology, which will be very helpful for later inferences and interpretation of my results. The individuals use the entire coral colony equally as their home (Sale,

1970), though there is a strict protogynous haremic size-dominated hierarchy, in which the largest specimen is the reproductive male and is usually accompanied by three reproductive females and several other smaller juveniles unless the coral colony is much bigger in size (increased number of mature males) or inhabited by less individuals (less mature females and juveniles) (Fricke & Holzberg, 1974;

212

Coates, 1982; Shpigel & Fishelson, 1986; Schwarz & Smith, 1990). The reproductive females lay egg clutches of 100-1500 eggs (average size of 830 eggs per three females, Wong et al., 2012) at the bottom of the coral colony, which then get fertilized by the mature male, hatch after two to three days (Mizushima et al., 2000), and will complete their PLD in an average of about three weeks. The PLD varies with location, with an average across all published studies of 21.1 d (Luiz et al., 2013) and 23.2 d in the Red Sea (Robitzch et al., 2015), all within the typical range for coral reef damselfishes. Its growth rate is of ~ 5.08 mm per month during the first year, putatively reaching maturity at a size of ~ 38 mm (but probably at a smaller size within the Red Sea, personal observation) and a size of ~ 61mm at the end of its first year (Pillai et al., 1985a). Breeding peaks have been reported from April to January for Minicoy Atoll in the northern Indian Ocean (Pillai et al., 1985a), and spawning peaks from June to September in Japan. In Japan, the species follows a semilunar spawning cycle usually two to four days directly before or around new and full moon, and during this period 33% of the populations’ females spawn in intervals of approximately two weeks (13-59 days; mode=14) one to six times (mode=4) per spawning season (Mizushima et al., 2000). Dascyllus aruanus typically has restricted mobility due to predation pressure and habitat patchiness (Fricke, 1980; Shpigel &

Fishelson, 1986) but exhibits opportunistic behavior when it comes to feeding

(Pillai et al., 1985a, 1985b), which could be one potential reason for its high abundance, success, and wide biogeographic range. Other strategies related to dispersal and recruitment strategies might also promote the nearly global distribution of this species. However, as for many other coral reef fishes, very little

213 is known about its early life stages. In this chapter, the PLD, genetic composition, and relatedness of a subset of 168 recruits of D. aruanus is assessed. These were captured in one single night at one single reef in the north central Saudi Arabian Red

Sea. I was able to evaluate a snapshot of the geneflow for a specific reef during one single recruitment event and ask crucial questions about larval behavior at very early life stages as well as on realized connectivity between populations (Watson et al., 2010) such as: (1) Are larvae travelling and recruiting together with their kin to the new home reef and how likely is it that such an event would occur by chance?

(2) How high or low is the genetic diversity within the recruiting cohort population

(RCP) in comparison to the adults’ standing population (ASP) of the reef of settlement? Finally, (3) how much of the total genetic diversity of the ASP is provided by one single recruitment event and how many recruitment events would it require to provide the effective population size of the ASP of that specific reef (i.e. what are the consequences for population dynamics)?

214

4.2 MATERIALS AND METHODS 30 30° IA

N ° IO 20

4.2.1 Study Site and Sample Collection of Dascyllus aruanus 10 10°

GAQGAQ Al Karrah Reef

30° 30 NCRS NUM ● SCRSSCRS NUW SRS MSH SRS NUM MAT NUW

RMA MSH

AKA

MAT 25 25° AFA 2000 km

scale approx 1:20,000,000 RMA 0 AKA AlI

200 AFA

400 km 80 ° 100°100 120120°

DUR AlI

20 20° KAM

DJI DUR CHLA [mg m-3]

KAM 15° 15 0.08 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1 2.5 10 30

SOC DJI 200 km scale approx200 1:20,000,000 km

200 km SOC

10° 10 0 200 400 km 35° 40° 45° 50° Figure 4.1: Sampling site of Dascyllus aruanus recruits in the Saudi Arabian north central Red Sea. The sampling location is near the coastal city of Yanbu at Al Karrah Reef and indicated by a black triangle.

In the Saudi Arabian Red Sea near the shores of Yanbu at the reef site of Al Karrah

Reef (N22°56.265; E38°45.943), 171 D. aruanus recruits were collected around the first quarter moon on March 5th 2014, using an LED (12 V) powered cylindrical light trap (Bellamare; ~1 m long and ~0.5 m diameter of 500 μm Nitex mesh netting; the prototype of the ones used under 3.1, Chapter 1). The light trap was deployed with a

1 kg weight attached to its cod end from a 35 m research vessel (RV Thuwal) at the sheltered side of the reef. The light trap remained at a depth of ~1m and a bottom depth of ~15 m from ~ 17.00 h (before sunset) until ~06.00 h (before sunrise). The

215 entire catch was immediately preserved in 95% ethanol and then morphologically sorted. A total of 171 recruits of D. aruanus were identified and stored individually for further analysis. This sample is referred to as the recruitment cohort population

(RCP). Additionally, 53 adult specimens were collected from the same reef using clove oil, tweezers, hand nets, and a 1 x 1.5 m plastic bag to cover the targeted coral colonies (Figure 4.1) and are referred to as the adult standing population (ASP).

4.2.2 Single Nucleotide Polymorphisms (SNPs) Library Preparation and

Sequencing

Genomic DNA was extracted from fin or gill tissue preserved in 96% ethanol using a

Nucleospin-96 Tissue Kit (Macherey-Nagel, Düren, Germany). Double digest restriction associated DNA (ddRAD) libraries were prepared using 500 ng of high quality DNA per sample following the protocol described by Peterson et al., (2012) with some modifications. From the 171 samples of the RCP, 3 were excluded because they yielded lower DNA concentrations. In short, genomic DNA of the remaining 224 individuals (RCP+ASP) is digested at 37°C for three hours using the restriction enzymes SphI and MluCI (NEB) followed by a ligation step, where each sample is assigned to one of sixteen unique adaptors. Pools of sixteen individuals are combined and run on a 1.5% agarose gel, from which fragments of ~400 bp are manually excised and purified using a Zymoclean Gel DNA Recovery Kit. Each pool is then amplified adding a unique indexing primer for each pool according to the standard Illumina multiplexed sequencing protocol in a 50μl PCR reaction

216 containing 25μl Kapa Hifi Hotstart Ready Mix Taq, 20μl of pooled library DNA and

2.5μl of the universal Illumina PCR primer and an additional 2.5μl of one of 12 unique indexing primer for each pool. Amplifications are carried out in an

Eppendorf 94-well vapo.protect Mastercycler Pro System (Fisher Scientific) using the following protocol: initial step at 95 ̊C for 3 min, followed by ten cycles of 98 ̊C for 20 sec, 60 ̊C for 30 sec, and 72 ̊C for 30 sec, and a final step at 72 ̊C for 5 min. DNA libraries are then quantified using the High Sensitivity DNA Analysis Kit in a 2100

Bioanalyzer (Agilent Technologies) and by running a qPCR in an ABI 7900HT fast real-time PCR system (Thermo Fisher Scientific) using the KAPA Library

Quantification Kits (Kapa Biosystems). One final time, the length (bp) and quality of the library fragments is measured in the 2200 TapeStation (Agilent Technologies) using the High Sensitivity D1000 ScreenTape Kit. Pools are subsequently combined in equimolar concentration to form a single genomic library. In total two libraries were created and run twice on two separated lanes each of a HiSeq 2000 Illumina sequencer (four lanes total, two lanes for each of the two libraries containing 112 individuals each; single end reads, 1 x 101 bp; v3 reagents) to increase the coverage per locus.

4.2.3 De-novo Assembly

Sequences of 224 individuals were de-multiplexed and filtered by quality using the

‘process_radtags.pl’ pipeline in STACKS version 1.42 (Catchen et al., 2011).

Individual reads with phred-scores below 30 (average on sliding window) or with

217 ambiguous barcodes were discarded. After this, individuals for which less than

500,000 reads were retained were also discarded. For all remaining 211 specimens,

RADSeq loci were assembled de-novo using the ‘denovo_map.pl’ pipeline in STACKS.

Different parameter combinations were evaluated, which resulted in different numbers of loci but gave similar results in genetic comparisons (genetic clustering and pairwise FST among sites). For the main analyses presented here, I used a parameter combination similar to the one recommended by Mastretta-Yanes et al.,

(2014): minimum read depth to create a stack (-m) = 6, number of mismatches allowed between loci within individuals (-M) = 3, number of mismatches allowed between loci within catalog (-n) = 2, and the option to remove or break up highly repetitive RAD-Tags in the ustacks program (-t).

Following de-novo mapping, further data filtering was performed using the population component of STACKS. First, only those loci present in at least 95% of individuals were retained (-r) = 0.95. Second, all loci with minor allele frequencies lower than 0.05 were removed to reduce the number of false polymorphic loci due to sequencing error (--min_maf) = 0.05. Third, the maximum observed heterozygosity required to process a nucleotide site at a locus was set to 60% to filter loci out of Hardy-Weinberg-Equilibrium (HWE) (--max_obs_het) = 0.6. As a last step, the log likelihood (lnl) ratio of all loci was calculated using rxstacks and after looking at its distribution a log likelihood threshold of (--lnl_lim) = -5 was used to filter loci with lnl values below -5. The write_random_snp option produced a vcf file with only one single randomly chosen nucleotide polymorphism (SNP) per stack.

The resulting vcf file was converted to other program specific input files using

218

PGDSPIDER version 2.0.5.2 (Lischer & Excoffier, 2012). The data set was evaluated for linkage disequilibrium (LD, as implemented in Genepop 4.2; (Raymond &

Rousset, 1995; Rousset, 2008) and Hardy-Weinberg-Equilibrium (HWE). However, for this purpose only the remaining 38 adult individuals from the ASP of the reef were used. This was done to prevent false results on LD and HWE deviations in case of the presence of high relatedness or selective pressure among recruits of the RCP.

P-values were tested for false-discovery-rate (FDR; using qvalue R package from

BioLite). GenAlEx 6.501 was used to calculate the multilocus Probability Identity

(PI), which provides an estimate of the average probability that two unrelated individuals drawn from the same population will have the same multilocus genotype (The threshold was set to 7.11*10-36). From the remaining loci, 300 loci with the most equally distributed allele frequencies (above 34 %) were chosen as the final data set. Using this data set the PI for assigning erroneous siblings was very low and below 7.11*10-36. For all following analyses the final data consisted of 300 loci and 193 individuals with less than 20% missing data.

4.2.4 Sibship Assignment, Population Structure, and Statistical Analyses

In order to identify potential genetic differentiation between the ASP of the site and the RCP, pairwise (PW) FST and individual genetic distances were calculated in

GenAlEx 6.501 (Peakall & Smouse, 2006, 2012). Additionally, a multilocus assignment test and a PCoA were performed to see whether individuals were preferentially assigned to one or the other population (ASP vs. RCP). Fixation

219 indices per locus were also calculated with GenAlEx to investigate if there were loci under selection. The number of observed MLGs (multilocus genotypes) was computed using the poppr R package to assess whether one of the two populations was more or less genetically rich due to, for example, strong selective pressure among the individuals of the RCP or of the ASP; or due to a higher source/origin of genetic material amongst the individuals of the ASP vs. the maybe more contained or limited source of the incoming RCP of that particular night. To avoid biases due to sample size only the eMLG (expected MLG) values were compared, which is an approximation of the number of genotypes that would be expected at the largest shared sample size (N=25) based on rarefaction. As a final step to identify any potential genetic structure within the samples, a Bayesian clustering analysis was performed in STRUCTURE v. 2.3.4 (Pritchard et al., 2000). The analyses were carried with 100,000 iterations discarded as burn-in and 200,000 iterations retained and the average of five runs for a total number of populations, from K = 1 to 3.

COLONY was used to identify sibling groups within the recruitment cohort. Sibling assignments were accepted upon a posterior probability exceeding 0.75 (as also in

Schunter et al., 2013 and Herrera et al., 2016). The parameters used were the following: polygamous diploid dioecious species, no inbreeding, unknown population allele frequencies, no information on the maternal or paternal mean sibship size, and using the two most accurate methods: either full likelihood (FL) with high precision or a combination of FL and pairwise-likelihood (PL) (=FPLS; faster with similar accuracy specially for high number of markers and not very big sibship sizes) and medium precision in computation. These parameters were used

220 on three different data sets with 740, 500, or 300 loci using three different random seed numbers per data set to compare the sibling assignment results. As a final step, the 300 loci dataset was run using the most accurate but computationally more demanding full-likelihood method (FL) and the results were once again compared.

4.2.5 Otolith Preparation and Pelagic Larval Duration (PLD) Assessment

Sagittal otoliths of all 171 specimens (preserved in 96% ethanol) were removed and mounted onto a microscopy glass slide with thermoplastic resin to assess the age of the recruit. This is referred to as pelagic larval duration (PLD) since the larvae in the trap most likely had just arrived from the pelagic and were about to settle onto the reef, as well as due to the fact that the ages of the recruits were within the expected

PLD range of the species at that site (~24 d, NCRS, Robitzch et al. 2015). Three to ten pictures were taken of each otolith with the program AxioVision Rel. (V. 4.8.2.0;

Carl Zeiss Micro Imaging GmbH) utilizing a Zeiss AXIO Scope A1 microscope (200× magnification) and a Zeiss AxioCam ICc1. Three independent PLD readings deviating by less than 10% from the mean count were made for each individual. Additionally, and as a trial experiment, otoliths from another collection of recruits were treated with immersion oil or a 5% bleach solution to see if this would improve the visibility of the daily growth increments but after treatment they remained either similarly or even less visible. I reasoned that the best readable otoliths were the least manipulated ones probably due to the fragile nature/susceptibility of the tiny otolith itself. The same person read (Nora Kandler) and counted PLDs for all

221 specimens but lacked any prior information or metadata about the samples to avoid biases. A total of ten adults and 171 recruiting D. aruanus otolith readings were used in the final data set (Table 4.1). The PLD measurements were compared in different ways and the single generated data set assessed for statistical significance/differences to further support the genetic sibship assignments, assuming that siblings travelling and recruiting together should have approximately the same PLD. Additionally, the RCP was randomly subsampled 50 times (with the sample function in R) to compare average squared differences between random pairs of PLDs to the squared difference of the genetically assigned sibling pairs and assess whether associations of PLDs among sibling pairs are significantly different than expected by chance given the PLD distribution in the RCP.

4.2.6 Statistical Significance of Kinship Results and Further Data Mining

To infer the significance of the sibling assignments provided by COLONY, data on the reproductive biology, ecology, abundance, and habitat of D. aruanus was gathered from diverse sources, such as available literature, personal observations, and discussions with colleagues and experts in the field (see literature review/info on the species in this chapter’s introduction). The underlying hypothesis to be tested herein (null hypothesis) is that the sibship results provided by COLONY can be obtained by random, if picking X number of larva out of the (unknown) total recruiting cohort population (TRCP). This hypothesis assumes that a pool of N larvae representing the TRCP at a certain reef is a portion of a completely mixed

222 recruitment cohort of larvae, from which any such larva (which has survived the average PLD) can be arriving at that reef with any other larva from any potential source-reef and/or fish colony. The alternative hypothesis would be that the larvae instead try to stick together among their kin (and are not randomly mixed) while travelling in the pelagic and completing the average PLD. In the case of the alternative hypothesis, the likelihood of sampling siblings by taking a subsample

(herein representative of the RCP) of the TRCP would increase. To test these hypotheses, a program was coded in PYTHON (written by collaborator Liam

Mencel) to randomly draw one million times a subsample of 168 recruits (= RCP) out of the TRCP. Two different values were used as input for the TRCP, which were inferred from data, literature, and open discussions with colleagues (personal observations and anecdotes) on the reproductive biology and ecology of damselfishes and in particular of D. aruanus. One value represented the TRCP under the scenario that larvae belonging to reefs from a radius of 25 km (r25; TRCPr25) can recruit to the study site (Al Karrah Reef). The second value represented the scenario that recruits can come from a radius of 50 km (r50; TRCPr50) around Al Karrah Reef.

With the results from the permutations, an estimated TRCP was calculated to achieve our observed sibling assignments (TRCPrRCP) and the permutations were run again using this value. As a last TRCP value I used the minimum possible TRCP that could correspond to/be representative of our data/results, which is equivalent to the minimum number of colonies (Cmin) required to produce the observed results from COLONY for the studies’ RCP (TRCPCmin).

223

An additional tool was designed to further explore the likelihood of the COLONY results under the TRCPr25 scenario by random chance. The theoretical data set of

TRCPrRCP was subsampled picking 168 individuals (= theoretical RCP) 100,000 times and the results (random picks) were explored for the occurrence of picking a single recruit as well as sibling pairs, triplets, or quadruplets from the TRCPrRCP. These observations were then used as the basis for three further analyses. (1) A total score value was assigned to each of the iterations depending on the frequency of each of the aforementioned events in each random pick of 168 individuals. The scores were indirectly and proportionally assigned to the frequency of pairs. (2) The same scoring system was used to assign a total score value to the empirical data set of 168 individuals (RCP) according to the number of single recruits, sibling pairs, triplets, and quadruplets found in COLONY. (3) With this scoring system random samples were then simulated and their score values computed to assess how frequently the score value of our data set was observed within the random picks and so, the likelihood of the empirical results (i.e., how likely it was to find the assigned score value of the RCP among random picks in the theoretical TRCPrRCP) and the significance/meaning behind the COLONY sibship assignments.

For a final inference, the same tool was used to infer the maximum theoretical TRCP at Al Karrah Reef, which would allow the observation of similar sibship results to those found within this study’s empirical RCP by random (i.e., the average maximum

TRCP size (TRCPRper)). Therefore, the tool inferred score values for different population sizes until the achieved score value was within a 95% confidence interval of the score value calculated for the empirical results (the RCP’s score

224 value) meaning that by picking 168 individuals there is a 95% chance of having a score as high as within our data.

4.2.7 Theoretical Values on the Biology and Population Dynamics of Dascyllus aruanus in the Red Sea

For the first two TRCPs (TRCPr25 and TRCPr50; see Table 4.2 for abbreviations), the following parameters were required: (a) number of potential daily fish recruits produced by an average colony of D. aruanus during the reproductive season in the

Red Sea (RCd), (b) number of D. aruanus colonies per area of habitat (CA), (c) percentage of simultaneously reproductive active females in the population (FR%),

(d) the area of habitat available in a chosen radius of recruitment source (Ar). This results in the following formula: TRCPr = RCd * CA * FR% * Ar.

In order to calculate (a-d) and the first two TRCPrs one needs: For (a), the average number of eggs per colony (E), the hatch rate per clutch (H), the percentage that hatched per day (Hd) the survival rate after the PLD (S); so that RCd = E * H * Hd * S.

For (b), data on the abundance/population density of D. aruanus was measured at three representative mid-shelf central Red Sea reefs using a 50m*2m (At=0.1 km2) transect counting the number of D. aruanus colonies (Da); so that CA= Da/At. The average number of individuals per colony was also calculated for Al Karrah Reef out of a total of four colonies and for the entire surveyed Red Sea colonies (653 specimens total) to get an estimate and see if the average of three reproductive females per colony can be theoretically realized among the Red Sea average number

225 of individuals per colony. For (c), average values reported by Mizushima et al.

(2000) were chosen, so that FR%= 0.33. For (d), the area around the Al Karrah Reef was explored using Google Earth satellite imagery and the trajectory of potential habitat for D. aruanus (L) was measured. The criteria used for this measurement were: the total trajectory of fringing reefs, circumferences of mid-shelf reefs, the diameter of lagoons, and the trajectory of the sheltered backsides of reefs at the shelf-edge found in a radius of either 25 or 50 km. This was then multiplied by the average “width” of D. aruanus habitat (W) following averages of bathymetry measurements from central Red Sea reef and using a depth range from 1-19 m as suitable (bathymetry maps were provided by Maha Khalil); so that Ar = L * W.

4.3 RESULTS

4.3.1 Raw Sequence Filtering and Assembly

A total of 594,692,197 reads of 101 bp each were obtained for 224 individual samples from one location in the Red Sea and Arabian Sea (Fig 1d). As a conservative measure, 29 samples were discarded due to low read recovery (<

500,000), which comprised only 0.4% of all reads. A total of 1,422,802 loci were built into the catalog, with a coverage depth of 4x – 55x and an average of 14.4x.

After further filtering, a total of 19,544 loci were retained, which contained at least one randomly selected SNP present in over 95% of individuals in the population with a minor allele frequency of at least 5% and a maximum heterozygosity of 60%.

226

These gave genotypes for 193 individuals (25 from the standing population and 168 belonging to the recruiting cohort), with less than 20% missing alleles and a minimum coverage depth of 6x per locus. From these and for the analysis of population structure and sibship assignment, I selected three subsets: one with 731, one with 500, and one with 300 loci, all in HWE, with a minimum minor allele frequency of over 30%, and without any significant LD.

4.3.2 Population Genetic Statistics and Bayesian Clustering Analysis

Table 4.1: Summary of Principal Population Genetic Statistics Indiv. (#) FST FIS FIT F’ST Ne Hobs Hexp F ASP 25 ------1.894 0.455 0.470 0.032 RCP 168 ------1.909 0.466 0.480 0.030 ---- 0.000 0.135 0.134 0.003 ------

P-value ---- 0.609 0.001 0.001 0.339 ------The summary statistics of the two analyzed populations (ASP: Adult standing population and RCP: Recruiting cohort population) were calculated in GenAlEx (6.501). The number of individuals (Indiv. (#)) used for the analyses is provided in the first column. Fixation indices (FST, FIS, FIT, and F’ST) were calculated using an AMOVA between the two populations. Effective number of alleles (Ne), observed (Hobs) and expected heterozygosity (Hexp) as well as fixation index (F), were calculated for both populations.

A summary of the principal statistics (number of individuals per site, average number of effective alleles, observed and expected heterozygosity, and fixation indices) is presented in Table 4.1. Generally, the expected heterozygosity was high and equal for both the RCP and ASP samples (He=0.48, Nei 1978), as expected from the criteria chosen for the loci selection to build the dataset. Global FST 0.006 as well as pairwise FST 0.000 (p=0.339) between the two populations was very low

227 and not significant. Briefly, 90% of all FST values per locus were below 0.014, with a mean of 0.005 and a median of 0.002 and a maximum FST of 0.053. When the individuals were grouped into populations the FST values per locus were similar but slightly higher for the ASP (0.032) compared to the RSP (0.028) and in general, those FST had high SE/variability. This might however be an effect of the much lower sample size (25 vs. 168). The distribution of the FSTs per locus was also slightly different between the two populations.

Results from the eMLG also showed no differences in richness, diversity, or evenness between the two populations, probably due to the generally high genetic richness and putative high/persistent/successful gene flow among the species’ RS population, which is also supported by the overall high heterozygosity among all loci. Corrected genotypic richness was also the same within both populations.

With the data set of 300 SNPs from 168 RCP samples and 25 ASP samples, Bayesian clustering analysis (STRUCTURE) was not able to detect any population structure and identified K = 1 to be the most parsimonious grouping of individuals. This result is coherent with the low and insignificant FST values previously mentioned.

Additionally, multilocus population assignment tests also assigned each individual with similar proportions to either of the two putative populations. PCoA of individual genetic distances also did not separate the two populations (RCP and

ASP) and the most powerful first three axes were only able to explain about 5% of the total variation in the data.

228

4.3.3 Sibship Assignment

Siblings within the recruiting cohort population (RCP) were identified using

COLONY. From the 168 genotyped recruits, 111 pairs of half siblings were assigned using the final 300 loci dataset, from which on average 48 assignments (43% of all pairs) were considered as significant with a probability of over 75%. Additionally, three full siblings were found in all data sets always with 100% probability of assignment. The results from the FL and the three PLFL methods with different seeds were overall very similar, thus for subsequent calculations/inferences the FL method results were used since these are considered to be most accurate. From this data, I calculated that the maximum possible number of source colonies providing the RCP is 120, which is 29% fewer colonies than what I would expect if all recruits would be traveling on their own (the average over all methods was 31% fewer colonies). Among these 120 colonies a minimum of 31 had successfully recruited with at least one sibling (minimum 15 sibling pairs, 15 triplet, and 1 quadruplet sibling group), which is equivalent to at least 79 recruits and represents a minimum of 47% of the RCP. The remaining 89 recruits (53% of the total RCP) either arrived alone or their siblings were not captured (e.g., because they did not enter the light trap or were eaten by the many predators surrounding the light trap). Among the 79 recruiting siblings, “super-family” assignments were found as well. For instance five individuals were relatives but only three of them were HS at a time as follows (with each letter representing one recruiting individual and the linking “-“ indicates a HS association): A-B-C-D-E, i.e.: B is HS of A and C; C is HS of B and D; and D is HS of C

229 and E. This means, that the aforementioned HS assignments are most likely the offspring of one male and two females (coherent with the haremic colony structure of the species), while for instance recruits A and D, A and E, or B and E are the progeny of two different males and females (all together equivalent to the progeny of two males and five females) but due to their association through their HSs, certainly belong to the same coral colony and thus, potentially same fish colony as well. This would be evidence for either the offspring coming from a very large coral colony containing more than one fish colony and therefore at least two reproductive males and/or evidence for the presence of “sneaky” males within the same/one single fish colony. To our knowledge this is the first time the recruitment of the progeny of sneaky males and the presence of “super families” has been reported among coral reef fishes. Even though it is yet difficult to infer how significant our observations of recruiting siblings is to the assumption that coral reef fish larvae might intentionally be travelling together with their siblings rather than on their own, the fact that I even find evidence for rare events like the formation of super- families within just one probably relatively small portion of a single recruiting cohort could be consider as a strong hint towards an existing trend of larvae sticking to their kin in the open waters all the way up to their recruitment destination. The other interesting sibling group that was found included one FS assignment among four recruits (F, G, H, and I; FS are represented by a “=” link), which means this group was the progeny of just two females and one male:

230

4.3.4 Pelagic Larval Duration (PLD) Measurements

The PLD of the RCP ranged from 21 to 27 days and had an average PLD of 24.12 and a median of 24.17 days. The PLD of the ASP ranged from 22 to 27 days, with an average of 23.72 and a median of 24.00 days. The distributions of PLD measurements of both populations are shown in Figure 4.2 and were not significantly different from each other (t-test: p=0.35). PLDs of the siblings were also compared to the PLDs of the recruits that arrived without relatives and to the

PLDs of all recruits, which in both comparisons/cases were not significantly different from each (t-test: p=0.18 and p=0.50, respectively). The PLDs of each pair of siblings was compared to one another and was identical in 19% of the siblings- pair, differed by 1 day in 48%, and by 2-3 days maximum in the remaining 33% of the sibling assignments. This can be either by a mistake in the reading. A less likely but possible reason could also be the behavior and recruiting strategies of the larvae: Some may hatch a bit earlier and at a less convenient time (daytime) and remain for some hours close to their home colony before they take up the pelagic journey, which may add a daily increment; Or the observed RCP is a case of strong self-recruitment (defined here as larvae that stay at, or come back to their natal reef after their PLD, see Almany et al., 2007; Planes et al., 2009) and thus, the siblings’

PLDs can differ by a couple days due to different hatch times. However, the high genetic diversity of the population suggests high dispersal and geneflow and less events of self-recruitment. Fluctuations in self-recruitment rates (e.g., Berumen et al.

2012) may enable a given reef to maintain high levels of genetic connectivity while

231 still having episodic self-recruitment events. Additionally, the RCP was randomly subsampled 50 times to compare average squared differences between random pairs of PLDs to the squared difference of the genetically assigned sibling pairs and assess whether it is significantly different than expected by chance given the RCP

PLD distribution. The observed squared difference between the sibling pairs was significantly different from the average squared difference between randomly generated pairs (t-test; p = 0.03), further supporting the significance of the sibship assignments by COLONY.

a) PLD Frequencies - RCP b) PLD Frequencies - ASP

60 4 3 40 2 20 Individuals Individuals 1 0 0 21 22 23 24 25 26 27 21 22 23 24 25 26 27 PLD in days PLD in days

Figure 4.2: Frequency of the pelagic larval durations (PLDs) within the recruiting cohort population (RCP; a)) and the adult standing population (ASP; b)) of Dascyllus aruanus at Al Karrah Reef (north central Red Sea, Saudi Arabia). The y-axis indicates the number of individuals and the x-axis the PLD in number of days counted from growth increments in on sagittal otoliths.

4.3.5 Statistical Significance of Kinship Results and Further Data Mining

The values for the formulas were the following: For (a) the average number of eggs per colony (E)= 830 (Wong et al., 2012); the hatch rate per clutch (H)= 0.5 (it can be anything between 0 and 1, Bernardi’s personal observation for D. aruanus, and after

232 discussions with Giacomo Bernardi, Ricardo Beldade, and Suzanne Mills I decided that 0.5 would be a fair average for damselfishes), the percentage that hatched per day (Hd) = 0.5 (as they hatch after two to three days, Mizushima et al. 2000); the survival rate of the PLD (S) = 0.01 – 0.05 (in agreement with Giacomo Bernardi,

Ricardo Beldade); so that min. RCd = 830 * 0.5 * 0.5 * 0.01 = 2, max. RCd = 830 * 0.5 *

0.5 * 0.05 = 10, and the final averaged RCd = (2+10)/2 = 6. For (b) the average number of D. aruanus colonies (Da) per 2m*50m transect = 10.23, so that CA=

10.23/0.1km2 = 102/km2. The average number of individuals per colony was 17 for

Al Karrah Reef and seven for the entire Red Sea, which are both above four and thus enough to have the assumed three reproductive females per colony. For (c) average values reported by Mizushima et al. (2000) were chosen, so that FR%= 0.33. For (d) the trajectory of potential habitat for D. aruanus (L) measured for a radius of 25 km and 50 km was Lr25= 160 km and Lr50= 387 km respectively and the average “width” of D. aruanus habitat (W) was set to be W = 1.2 km, following averages of bathymetry measurements from central Red Sea reef and using a depth range from

1-19 m as suitable (bathymetry by Maha Khalil); so that Ar25 = 160 km * 1.2 km =

2 2 2 192 km ; and Ar50 = 192 km + 227 km * 1.2 km = 464.4 km .

Therefore, TRCPr25 = RCd * CA * FR% * Ar25 = 19445; and TRCPr50 = RCd * CA * FR% * Ar50

= 47033 at the reef of Al Karrah Reef.

The permutations run for these TRCPrs resulted in an average of 3.49 sibling pairs

(SPr25), 0.04 sibling triplets, and 0 sibling quadruplets at r25 and about half for r50

(i.e., 1.47 sibling pairs (SPr50), 0.01 triplets and 0 quadruplets) when drawing 168

(our sample size; i.e., RCP) individuals randomly out of the TRCP containing six

233

siblings (avg. RCd) per recruiting colony (=CA* FR%). With these results, I can use the following two assumptions ((e) and (f)) to roughly estimate the minimum radius for a TRCP (TRCPrRCP) that could give us the results I obtained with COLONY. (e) ½*r50

= r25; (f) SPr25 = SPr50 * x. (f) would then give us the rate (Rp) at which in theory the

SP assignment increases every time you half the radius providing the TRCP (x = Rp =

3.49/1.47= 2.37). COLONY found 48 SP (SPRCP) within our RCP. So (g) n * Rp * SPr25

= SPRCP gives us the number of how many times I must halve the radius (n) to get 48

SP drawn randomly. Therefore, n = 48/8.26 = 5.811; the theoretical radius (rRCP) for our observed SPRCP would then be 1/n * r25 = 25km/5.8 = 4.31 km radius, which is

2 2 equivalent to an area of 58.37 km (i.e., ARCP= rRCP * π); and the TRCPrRCP= RCd * CA *

FR% * ARCP = 5911 total recruits.

I also know from the COLONY results above that I had a maximum of 120 colonies providing our RCP and if I assume that each one of these ends up with an avg. RCd of

6, I can then assume that if only 120 colonies provided the TRCP the minimum value for it would be an average RCd * 120 = 720 (= TRCPCmin), which is much lower than the TRCPrRCP above and equivalent to an area of ACmin = TRCPCmin/(RCd * CA * FR%) =

2 2 720 / (6 * (102/km ) * 0.33) = 3.57 km (for: TRCPCmin = RCd * CA * FR% * ACmin) and a radius of rCmin = √(ACmin/ π) = 1.066 km. The simulations were run once more with

TRCPrRCP and TRCPCmin, for which the former gave a SPrRCP of 11, 0.41 triplets, and

0.01 quadruplets, which is still far below our observed values. Even for the very unlikely scenario of TRCPCmin, the SP assignments I get by random are below our actual observed values (34 SP, 13.71 triplets, and 3.08 quadruplets, 0.37 quintets, and 0.02 sextets). With the second tool, my data was given a score value of 142 and

234 after 100,000 random picks that score was only observed 0.015% of the times, which shows how rare that score value of the empirical sibship results is and thus highlights the significance of my observations. Furthermore, the tool computed that the TRCP would have to be as little as 750 recruits only (i.e. coming from a max of

125 colonies) to have a 95% chance to find sibship assignments as within our data when picking 168 individuals randomly.

Table 4.2: List of Important Abbreviations and their Respective Value or Result

Abbreviation Meaning Results ASP Subsample of the adult standing population (ASP) in Al Karrah Reef 25 RCP Empirical recruiting cohort population (RCP) sample 168 TRCP Total recruiting cohort population Unknown

TRCPr50 TRCP of Al Karrah Reef under the assumption that larvae from reefs 47033 around a 50km recruit to it.

TRCPr25 TRCP of Al Karrah Reef under the assumption that larvae from reefs 19445 around a 25km recruit to it.

TRCPrRCP TRCP from the maximum possible radius (rRCP = 4.31 km) inferred 5911 by the results of TRCPr25 and TRCPr50 to achieve sibship results similar to those from the RCP TRCPCmin TRCP that results for the minimum number of colonies recruiting to 720 Al Karrah according to the RCP sibship analysis. rCmin = 1.066 km) The most important abbreviations mentioned in the Results section are listed together with their respective meaning (as explained in the Material and Methods section of this chapter). The respective value under the Results column indicates the number of individuals either counted or calculated by inference in that respective population (i.e., the population size). Note that the recruitment estimates are for a given night during the peak recruitment period.

4.4 DISCUSSION AND CONCLUSIONS

The genetic relatedness among recruits of coral reef fish within a settling cohort provides essential knowledge into spatial, temporal, and mechanistic features of larval dispersal in marine organisms (Planes, 2002; Selkoe et al., 2006; Veliz et al.,

235

2006; Buston et al., 2009). For example, one can infer how many adults are actually contributing to one such a recruitment event (Hedgecock, 1994; Hedgecock et al.,

2007b; Beldade et al., 2012) and the genetic diversity transported within the cohort.

In this study, we have clear evidence for a significant presence of siblings within a single recruitment event, strongly suggesting that larvae from Dascyllus aruanus are travelling together throughout their pelagic life stage until they reach their settlement habitat. Within a subsample of the recruiting cohort of one single night at

Al Karrah Reef in the Saudi Arabian Red Sea, I found that at least 47% of these larvae had arrived with at least one relative. There was even the presence of up to four siblings arriving together in the single sample of 168 larval recruits.

Permutations showed that these sibship results were highly significant, meaning that the fishes were actively travelling close to their kin. The sibship assignments were further validated by similar PLDs between members of sibling pairs, meaning they had spent the same time in the pelagic before recruiting together. Other studies on kinship assessment have previously reported the presence of siblings among other coral reef fishes within a recruiting cohort. The percentage of sibling assignments were much lower (6-12%) but were based on the sampling of recently settled larvae or juveniles (Bernardi et al., 2012, Herrera et al., 2016) and not on recruiting larvae. Another aspect to point out is that the sampling device used herein is a light trap, which attracts not only fish larvae but also a wide-range of predators, which opportunistically roam the trap to feed on organisms attracted by its light. In this context, it is highly possible that the rates at which larvae are arriving together with their siblings and the total number of sibling-groups is much

236 higher than calculated but due to high predation around the light traps, the number of siblings is being reduced before they can enter the trap.

4.4.1 Why Are Closely Related Larvae Travelling Together?

In light of the results of this chapter, I suggest that D. aruanus larvae from the Red

Sea are actively remaining together with their siblings as they hatch and begin their

PLD phase. In my opinion, the only other probable explanation to find such high and significant rates of kinship assignments within a small portion of one recruitment event would be that larvae are self-recruiting and/or staying very close to or directly at their natal reef and then self-recruiting together. Self-recruitment has been mostly thought to happen on isolated areas such as distant islands (Sponaugle et al., 2002; Cowen et al., 2006; Berumen et al., 2012) but a clear trigger of self- recruitment is still to be defined and as it has increasingly been reported among various other location (Harrison et al., 2012; Almany et al., 2013) and could also be a species specific trait. The Red Sea has almost no isolated reefs since its entire coast has continuous reef coverage (although it is important to note that determining

“isolation” is difficult and somewhat subjective). Thus, I think it is unlikely that the results found herein are a consequence of self-recruitment. Self-recruitment has been reported to be absent in another pomacentrid (a clownfish) at a reef site in the central Red Sea (Nanninga et al., 2015), even though the reef had the potential to be a relatively isolated reef for the Red Sea. Moreover, and from the genetic data of

Chapter II (including the preliminary analyses based on microsatellite markers in

237 the Appendix) the high heterozygosity, genetic diversity, and the absence of genetic differentiation of D. aruanus throughout the Red Sea speaks for high geneflow and admixture between reef populations and thus, rather against retention of larvae at a particular reef and/or high rates of self-recruitment. This once more speaks against kinship among recruits by chance and rather for active schooling of sibling during the PLD.

But how are these larvae travelling together and moreover, why? Could this be the key to success of widespread species such as D. aruanus?

For larvae, remaining with your close kin means that you can keep up with them, hide among them, and have higher chances of recognizing dangerous predators

(Handegard et al., 2012). Finding the right food source and cues (Handegard et al.,

2012), and navigate to a suitable and species-specific habitat (Couzin, 2007; Torney et al., 2009, 2011; Berdahl et al., 2014) may also become easier. However, it also means that once you find the right type of food and habitat, you have to compete for it. Thus, such schooling behavior might be more favorable in some regions than others (Jonsson & Jonsson, 2011; Quinn, 2011). Hence, is travelling with your sibling then an innate behavior or is it induced by the environment? Behavior can arise as a response to environmental cues, or it can be ontogenetically triggered (i.e., as larvae develop), with eventual schooling behavior emerging as the organisms reach juvenile or adult stages, which, in most cases, go hand in hand with increasing sensory abilities and survival (e.g., Codling et al., 2007). Larval homing behavior is a concept that may help explain observed levels of self-recruitment, geneflow, and connectivity among marine populations. Sensing home, as an important larval

238 behavioral trait, also seems relatively essential when compared with other important larval developmental stages, such as first-feeding and swimming.

Additionally, self-recruitment as an innate behavior could also be explained by the selective pressure of a species’ natural habitat, which has shaped the species’ life- history traits to increase the odds of larvae returning close to their parent populations (Strathmann et al., 2002). I believe coral reef fishes are an ideal group for testing the impact the potential selective pressure of habitat availability on homing behavior since their benthic and highly habitat-specific nature probably demands it. They rely on a habitat type that has been widely modified and impacted by tectonic movements, ice ages, and changes in sea level many times in geological history and in a relatively abrupt manner in comparison to pelagic marine environments. Therefore, I strongly believe that (most) coral reef fish larvae have evolved to self-recruit. It emerges as a suitable innate behavior that could have evolved to succeed and survive when relying on such a small portion of habitat in the vastness of the oceans (only 1% of the ocean’s surface are coral reefs).

Consequently, and depending on the fish’s biology and ontogeny, this behavior will likely result in schooling among close kin because they are nearby and to increase the chances of finding a way back to the correct home. The likelihood of siblings schooling together among coral reef fishes could also depend on the reproductive strategies of the species (depending on whether the species is a benthic brooder or a broadcast spawner), another aspect that warrants further research. Other factors, such as species-specific PLDs, habitat preferences, and seasonal variations in mortality rates, may cause the likelihood of siblings recruiting together to differ

239 taxonomically, spatially, and temporally. As fish larvae travel, their journey, route, and collective decision-making can also be influenced by the environment to different degrees/scales. A benign environment will not mask/alter much of the innate behaviors and result in more consistent connectivity routes. While in the presence of harsh and variable oceanographic conditions, swimming skills, navigation abilities, and orientation cues (chemical, solar, tidal, etc.) will be affected and overall, putatively alter the (innate) outcome of the species PLD. For instance, one could hypothesize that if currents are pushing larvae into the opposite direction to home but are easily avoidable by vertical swimming that might be a good strategy. But if fighting such current will take up too much energy, a different group decision may be taken. Schooling has even been found to have hydrodynamic benefits (Ross et al., 1992; Herskin & Steffensen, 1998; Portugal et al., 2014). Once a larval cohort is ready to settle, if home happens to be the only habitat available (as may occur in isolated areas), these larvae will appear to preferentially settle at home. Near the end of their pelagic phase, if they find another suitable (or maybe even a better) habitat and/or there is not enough food or strong enough cues to make it back to their natal reef, the cohort may spontaneously recruit elsewhere instead. If the region they were born in is too oligotrophic, larvae may also travel further into the pelagic and forage in wider ranges, moving further away from home and potentially losing chemical cues leading back to that home reef. Limited food availability leads to depleted energy reserves and thus, to a trade-off in local habitat selection (Vikebø et al., 2007). The health of the reef can bias the chemical cues that larvae use for orientation (Munday et al., 2009; Dixson et al., 2011, 2014), too, and

240 so, human activity may currently be having a high impact on larval behavior. Thus, whether fish larvae self-recruit is likely the combined result of the innate behavioral trait of homing in coral reef fish larvae biased by habitat, food availability, oceanographic conditions, and nowadays, even the intensity of human impact.

Additionally, the level of relevance that such innate homing traits have taken in larval behavior may also vary inter-specifically or between populations, as behavior is also modified through adaptation (e.g., Giske et al., 2003). There may even be mechanisms that coral reef fish parents can tell their offspring whether they had a good home or not (either chemically and/or epigenetically) and if it is a good strategy to return home or to try and disperse (Nanninga & Berumen, 2014).

How has collective decision-making evolved? Collective behavior (schooling, flocking, herding, etc.) is ubiquitous among animal taxa (Krause & Ruxton, 2002) and is particularly common in fishes (Pitcher & Parrish, 1993). A significant benefit for many organisms that live in groups may be collective intelligence (Couzin, 2007,

2008), and studies have demonstrated that simple interactions among individuals with relative unsophisticated behavior may produce complex outcomes that are adaptive at the group level (Bonabeau et al., 1999; Seeley & Buhrman, 1999; Franks et al., 2002; Camazine, 2003; Sumpter, 2006; Berdahl et al., 2014). Here, I focus on the navigation ability potentially arising from collective intelligence. In theory, collective migratory strategies can evolve under a wide range of ecological scenarios

(Torney et al., 2009, 2010; Guttal & Couzin, 2010; Shaw & Couzin, 2013), and it is likely that migrations and other navigational decisions are being undertaken by large groups (Beauchamp, 2011; Milner-Gulland et al., 2011). Homing pigeons

241 navigate more efficiently when released as groups than individually (Biro et al.,

2006; Dell’Ariccia et al., 2008), and birds with less information can benefit from following others (Flack et al., 2012). Theories, like the ‘many wrongs principle’

(Larkin & Walton, 1969; Gruenbaum, 1998; Simons, 2004; Codling et al., 2007), predict that such groups may average out errors made by single individuals and enhance their navigational plan. In schools of fish, it has also been shown that the ability to sense odor cues increases with group size (Hall Jr et al., 1982; McNicol et al., 1996). Furthermore, empirical (Sumpter et al., 2008; Ward et al., 2008; Arganda et al., 2012) and theoretical (Condorcet, 1785; Ladha, 1992; King & Cowlishaw,

2007) work on voting and binary decision-making shows that larger groups are often able to choose the superior option more often when given two choices.

4.4.2 What Are the Consequences of Schooling With Kin During the Pelagic Larval

Duration (PLD) for Population Dynamics?

Assuming that only a small proportion of the available gene pool is transported to the population by one recruitment event (“sweepstakes effect”, Hedgecock, 1994) a single cohort is expected to have less genetic diversity than that found among cohorts (Hedgecock et al., 2007a). Indeed, significant changes in allele frequencies of larvae and recruits have been documented from one sampling time to the next (Li

& Hedgecock, 1998; Johnson & Wernham, 1999; Moberg & Burton, 2000; Toonen &

Pawlik, 2001; Planes, 2002; Selkoe et al., 2006). Schooling with your kin and recruiting with your siblings may further increase the “sweepstakes effect”. This

242 could eventually lead to population differentiation if the source of recruits is restricted to a single location with low genetic diversity over a couple of generations. However, in population genetics, the exchange and successful reproduction of a single individual from different reefs per generation is already enough to keep connectivity between populations and hinder genetic drift (Hellberg et al., 2002; Thorrold et al., 2002). Even though I find a scenario in which siblings of

D. aruanus are actively schooling and homing together, the genetic diversity of its entire population in the Red Sea is quite high and loci have overall high heterozygosity. Moreover, the genetic diversity is not significantly different and consistently high between both the adult standing population (ASP) as well as that of the subsampled recruiting cohort population (RCP) of the study site. Thus, a single recruitment event was able to provide as much of the genetic richness present in the adult standing population. This can be interpreted as a highly panmictic D. aruanus population, probably due to low self-recruitment, high geneflow, and connectivity between reefs, even if larvae are schooling with their kin until recruitment. It suggests that the species has large habitat availability, is well adapted, and successful throughout the reefs of the Red Sea, all of which have kept the gene pool well-mixed throughout the species’ distribution range. The highly genetically diverse recruitment cohort population may also suggest panmictic sources of larvae and/or the mixture of multiple sources of larvae recruiting to a single reef. Having nonetheless found high relatedness among the recruits indicates that the larvae from a single source are travelling and recruiting together. Thus, the highly panmictic Red Sea population has a large genetic pool among which relatively

243 high and consistent geneflow is occurring. There was also no evidence of genetic pre- or post-settlement selective pressure on the species. However, I did find differences in the average PLD of the RCP and the ASP. When comparing the frequencies in PLD of the two populations, there was a shift towards longer PLDs in the ASP compared to the RCP and while the PLD frequency distributions among the

RCP were normally distributed, those of the ASP were skew towards longer PLDs with high frequency of the longest PLD of 27 d. This could have two foundations: (1)

The RCP may have seasonal variation in PLD due to food availability and/or temperature differences (affecting metabolism and growth rates in the RCP). (2)

There could be a correlation between PLD and the post settlement survival of specimen. In some coral reef fishes, those that had a longer PLD seemed to be more likely to survive the juvenile stages and become part of the ASP (Gagliano et al.,

2007; Mccormick & Gagliano, 2009; Christie et al., 2010; Sponaugle et al., 2011;

Hixon et al., 2014). However, this must be interpreted with caution, as the number of individuals used for the PLD assessment of the ASP is much lower than that for the RCP, which may bias the distribution. Additionally, if larvae are recruiting together, they may also be settling together (e.g., Buston et al., (2009) found close kin within colonies of D. aruanus in Moorea, French Polynesia). Thus, if not many colonies were sampled to assess the PLD of the ASP and several individuals from one colony were assessed, it might have increased the frequency of a certain PLD value. Nonetheless, I believe this last point to be rather unlikely due to the size hierarchy within a fish colony and rather suggest a bias due to lower sampling size, the influence of the season on the PLD, and/or some ontogenetic, biological,

244 physiological and/or behavioral advantage linked to specific optimal length or generally longer PLDs among coral reef fish inducing higher success and survival in later life stages. Maybe fishes with a longer PLD settle at better colonies and thus are more likely to survive, too (i.e., they do not decide too quickly where to settle).

Fishes with longer PLDs may have developed better strategies or are more skilled in the avoidance of predators as they had to survive for a longer while in the most dangerous phase of their life-cycle (with highest mortality rates).

4.4.3 Further Data Mining

I trust that many studies have the potential of reporting much more on their findings than they do. In the light of how time, costs, and resources expensive it is to gather empirical data on coral reef fishes, one should feel obliged to infer as much as possible and reasonable for the data available. Therefore, I dedicated a large portion of the data analysis of this chapter to the inference of a putative recruiting cohort population size for D. aruanus in the central Red Sea. Even if most assumptions and/or inferences herein may be far off the real values, this chapter provides an example of how to proceed when much data is still lacking and illustrates some potential approaches to infer missing data. It furthermore encourages the extensive exploration of literature and other resources available to make better sense of the theoretical significance of results in empirical studies. My empirical data and results were combined with theoretically-derived values to fill gaps of knowledge in recruitment dynamics and biology of D. aruanus. From all inferences, the main

245 outcome was: to observe results similar to my empirical data, using theoretical random picks of a RCP, the TRCP has to be very small, likely much smaller than the actual RCP. This finally gave me confidence to assume that these siblings were travelling together rather than recruiting together by chance. The only other two explanations would be either that (1) my estimates of the TRCP are far off from the real value (i.e., my estimates are much too large) and the source area of the TRCP is much smaller, and/or (2) the mortality rate a lot higher, which would then mean that individuals would likely still need to stick together with their siblings in order to find as many triplets and quadruplets within just a small portion of the TRCP as I find within my RCP subsample.

Additionally, through a thorough exploration of the sibling assignments, this chapter provides first evidence for the existence and survival of offspring of sneaky dads, meaning that the haremic colony structure of D. aruanus might be violated more often than previously thought. This was confirmed by the presence of sibship among individuals coming from different dads and the same mother, which would not be possible if there was only one single reproductively active male per colony.

Finally, it would be very interesting to apply the materials and inference tools used herein at different sites along the environmental gradient of the Red Sea and in a comparative manner to assess the RCP of the endemic D. marginatus as well. Al

Karrah Reef would be a great location as both species are present.

246

4.5 REFERENCES

Almany G.R., Berumen M.L., Thorrold S.R., Planes S., & Jones G.P. (2007) Local replenishment of coral reef fish populations in a marine reserve. Science (New York, N.Y.), 316, 742–744. Almany G.R., Hamilton R.J., Bode M., Matawai M., Potuku T., Saenz-Agudelo P., Planes S., Berumen M.L., Rhodes K.L., Thorrold S.R., Russ G.R., & Jones G.P. (2013) Dispersal of grouper larvae drives local resource sharing in a coral reef fishery. Current Biology, 23, 626–630. Almany G.R. & Webster M.S. (2006) The predation gauntlet: early post-settlement mortality in reef fishes. Coral Reefs, 25, 19–22. Arganda S., Pérez-Escudero A., & Polavieja G.G. (2012) A common rule for decision making in animal collectives across species. PNAS, 109, 20508–20513. Armsworth P.R. (2000) Modelling the swimming response of late stage larval reef fish to different stimuli. Marine Ecology Progress Series, 195, 231–247. Avise J.C. & Shapiro D.Y. (1986) Evaluating Kinship of Newly Settled Juveniles Within Social Groups of the Coral Reef Fish Anthias squamipinnis. Evolution, 40, 1051–1059. Batchelder H.P., Edwards C.A., & Powell T.M. (2002) Individual-based models of copepod populations in coastal upwelling regions: implications of physiologically and environmentally influenced diel vertical migration on demographic success and nearshore retention. Progress in Oceanography, 53, 307–333. Baums I.B., Miller M.W., & Hellberg M.E. (2005) Regionally isolated populations of an imperiled Caribbean coral, Acropora palmata. Molecular Ecology, 14, 1377– 1390. Beauchamp G. (2011) Long-distance migrating species of birds travel in larger groups. Biology Letters, 7, 692–694. Beldade R., Holbrook S.J., Schmitt R.J., Planes S., Malone D., & Bernardi G. (2012) Larger female fish contribute disproportionately more to self- replenishment. Proceeding of the Royal Society Biological Siences. Berdahl A., Westley P.A.H., Levin S.A., Couzin I.D., & Quinn T.P. (2014) A collective navigation hypothesis for homeward migration in anadromous salmonids. Fish and Fisheries, 1–18. Bernardi G., Beldade R., Holbrook S.J., & Schmitt R.J. (2012) Full-sibs in cohorts of newly settled coral reef fishes. PloS one, 7, e44953. Berumen M.L., Almany G.R., Planes S., Jones G.P., Saenz-Agudelo P., & Thorrold S.R. (2012) Persistence of self-recruitment and patterns of larval connectivity in a marine protected area network. Ecology and Evolution, 2, 444–452.

247

Biro D., Sumpter D.J.T., Meade J., & Guilford T. (2006) Report From Compromise to Leadership in Pigeon Homing. Current Biology, 16, 2123–2128. Bonabeau E., Dagorn L., & Freon P. (1999) Scaling in animal group-size distributions. Proceedings of the National Academy of Sciences of the United States of America, 96, 4472–4477. Buston P.M., Fauvelot C., Wong M.Y.L., & Planes S. (2009) Genetic relatedness in groups of the humbug damselfish Dascyllus aruanus: small, similar-sized individuals may be close kin. Molecular Ecology, 18, 4707–4715. Caley M.J. (1998) Age-specific mortality rates in reef fishes: Evidence and implications. Austral Ecology, 23, 241–245. Camazine S. (2003) Self-organization in biological systems. Princeton University Press. Catchen J.M., Amores A., Hohenlohe P., Cresko W., & Postlethwait J.H. (2011) Stacks: Building and Genotyping Loci De Novo From Short-Read Sequences. Genes, Genomes, Genetics, 1, 171–182. Christie M.R., Johnson D.W., Stallings C.D., & Hixon M.A. (2010) Self-recruitment and sweepstakes reproduction amid extensive gene flow in a coral-reef fish. Molecular Ecology, 19, 1042–1057. Coates D. (1982) Some observations on the sexuality of humbug damselfish, Dascyllus aruanus (Pisces, Pomacentridae) in the field. Zeitschrift für Tierpsychologie. Codling E.A., Pitchford J.W., & Simpson S.D. (2007) Group navigation and the “‘many- wrongs principle’” in models of animal movement. Ecology, 88, 1864–1870. Condorcet M. de (1785) Essay on the Application of Analysis to the Probability of Majority Decisions. Paris: Imprimerie Royale. Couzin I. (2007) Collective minds. Nature, 445, 715. Couzin I.D. (2008) Collective cognition in animal groups. Trends in Cognitive Sciences, 13, 36–43. Cowen R.K. (2002) Oceanographic Influences on Larval Dispersal and Retention and Their Consequences for Population Connectivity. Coral Reef Fishes, 149–170. Cowen R.K., Paris C., & Srinivasan A. (2006) Scaling of connectivity in marine populations. Science. Dell’Ariccia G., Dell’Omo G., Wolfer D.P., & Lipp H.-P. (2008) Flock flying improves pigeons’ homing: GPS track analysis of individual flyers versus small groups. Animal Behaviour, 76, 1165–1172. Dixson D.L., Abrego D., & Hay M.E. (2014) Chemically mediated behavior of recruiting corals and fishes: A tipping point that may limit reef recovery. Science, 345, 892–897. Dixson D.L., Jones G.P., Munday P.L., Pratchett M.S., Srinivasan M., Planes S., &

248

Thorrold S.R. (2011) Terrestrial chemical cues help coral reef fish larvae locate settlement habitat surrounding islands. Ecology and Evolution, 1, 586–595. Doherty P.J., Dufour V., Galzin R., Hixon M.A., Meekan M.G., & Planes S. (2004) High mortality during settlement is a population bottleneck for a tropical surgeonfish. Ecology, 85, 2422–2428. Eble J.V.A., Toonen R.J., & Bowen B.W. (2009) Endemism and dispersal: comparative phylogeography of three surgeonfishes across the Hawaiian Archipelago. Marine Biology, 156, 689–698. Fisher R., Leis J.M., Clark D.L., & Wilson S.K. (2005) Critical swimming speeds of late- stage coral reef fish larvae: variation within species, among species and between locations. Marine Biology, 147, 1201–1212. Flack A., Pettit B., Freeman R., Guilford T., & Biro D. (2012) What are leaders made of? The role of individual experience in determining leader e follower relations in homing pigeons. Animal Behaviour, 83, 703–709. Franks N.R., Pratt S.C., Mallon E.B., Britton N.F., & Sumpter D.J.T. (2002) Information flow, opinion polling and collective intelligence in house-hunting social insects. Philosophical Transactions Of The Royal Society B-Biological Sciences, 357, 1567–1583. Fricke H.W. (1980) Control of Different Mating Systems in a Coral Reef Fish by one Environmental Factor. Animal Behaviour, 28, 561–569. Fricke H.W. & Holzberg S. (1974) Social Units and Hermaphroditism in a Pomacentrid Fish. Naturwissenschaftern, 367–368. Fuchs H.L., Neubert M.G., & Mullineaux L.S. (2007) Effects of turbulence-mediated larval behavior on larval supply and settlement in tidal currents. Limnology and Oceanography, 52, 1156–1165. Gagliano M., Mccormick M.I., & Meekan M.G. (2007) Survival against the odds: ontogenetic changes in selective pressure mediate growth-mortality trade-offs in a marine fish. Proceeding of the Royal Society Biology, 274, 1575–1582. Gerlach G., Atema J., Kingsford M.J., Black K.P., & Miller-Sims V. (2007) Smelling home can prevent dispersal of reef fish larvae. Proceedings of the National Academy of Sciences of the United States of America, 104, 858–863. Gillespie A. (2009) Group Dynamics and Interactions between Two Cohabiting Damselfishes, Dascyllus aruanus and Pomacentrus moluccensis. Natural Sciences, 82–86. Gruenbaum D. (1998) Schooling as a strategy for taxis in a noisy environment. Evolutionary Ecology, 12, 503–522. Guttal V. & Couzin I.D. (2010) Social interactions, information use, and the evolution of collective migration. PNAS, 107, 16172–16177. Hall Jr L.W., Burton D.T., Margrey S.L., & Graves W.C. (1982) A comparison of the avoidance responses of individual and schooling juvenile atlantic menhaden,

249

Brevoortia tyrannus subjected to simultaneous chlorine and δt conditions. Journal of Toxicology and Environmental Health, Part A Current Issues, 10, 1017– 1026. Handegard N.O., Boswell K.M., Ioannou C.C., Leblanc S.P., Tjøstheim D.B., & Couzin I.D. (2012) Report The Dynamics of Coordinated Group Hunting and Collective Information Transfer among Schooling Prey. Current Biology, 22, 1213–1217. Harrison H.B., Williamson D.H., Evans R.D., Almany G.R., Thorrold S.R., Russ G.R., Feldheim K.A., van Herwerden L., Planes S., Srinivasan M., Berumen M.L., & Jones G.P. (2012) Larval export from marine reserves and the recruitment benefit for fish and fisheries. Current Biology, 22, 1023–1028. Hedgecock D. (1994) Does variance in reproductive success limit effective population sizes of marine organisms? 122–134. Hedgecock D., Barber P.H., & Edmands S. (2007a) Genetic approaches to measuring connectivity. Oceanography-Washington Dc-Oceanography Society, 20, 70. Hedgecock D., Launey S., Pudovkin A.I., Naciri Y., Lapegue S., & Bonhomme F. (2007b) Small effective number of parents (Nb) inferred for a naturally spawned cohort of juvenile European flat oysters Ostrea edulis. Marine Biology, 150, 1173–1182. Hellberg M.E., Burton R.S., Neigel J.E., & Palumbi S.R. (2002) Genetic assessment of connectivity among marine populations. Bulletin of Marine Science, 70, 273– 290. Herrera M., Nanninga G.B., Planes S., Jones G.P., Thorrold S.R., Saenz-agudelo P., Almany G.R., & Berumen M.L. (2016) Seascape and life-history traits do not predict self-recruitment in a coral reef fish. Biology Letters, 12, 4–7. Herskin J. & Steffensen J.F. (1998) Energy savings in sea bass swimming in a school: measurements of tail beat frequency and oxygen consumption at di ff erent swimming speeds. Journal of Fish Biology, 53, 366–376. Hixon M.A., Johnson D.W., & Sogard S.M. (2014) BOFFFFs: on the importance of conserving old-growth age structure in fishery populations. ICES Journal of Marine Science, 71, 2171–2185. Hubbs C. & Blaxter J.H.S. (1986) Ninth Larval Fish Conference: Development of Sense Organs and Behaviour of Teleost Larvae with Special Reference to Feeding and Predator Avoidance. Transactions of the American Fisheries Society, 115, 98–114. Irisson J.-O., Paris C.B., & Planes S. (2010) Vertical distribution and ontogenetic “‘migration’” in coral reef fish larvae. Limnology and Oceanography, 55, 909– 919. Johnson M.S. & Wernham J. (1999) Temporal variation of recruits as a basis of ephemeral genetic heterogeneity in the western rock lobster Panulirus cygnus. Marine Biology, 135, 133–139.

250

Jonsson B. & Jonsson N. (2011) Habitats as Template for Life Histories. Ecology of Atlantic Salmon and Brown Trout: Habitat as a Template for Life Histories pp. 1– 21. Fish & Fisheries Series 33. King A.J. & Cowlishaw G. (2007) When to use social information: the advantage of large group size in individual decision making. Biology Letters, 3, 137–139. Kingsford M.J., Leis J.M., Shanks A., Lindeman K.C., Morgan S.G., & Pineda J. (2002) Sensory environments, larval abilities and local self-recruitment. Bulletin of Marine Science, 70, 309–340. Krause J. & Ruxton G.D. (2002) Living in groups. Oxford University Press, Ladha K.K. (1992) The Condorcet jury theorem, free speech, and correlated votes. American Journal of Political Science, 617–634. Larkin P.A. & Walton A. (1969) Fish school size and migration. Journal of the Fisheries Board of Canada, 26, 1372–1374. Lecchini D. & Nakamura Y. (2013) Use of chemical cues by coral reef animal larvae for habitat selection. Aquatic Biology, 19, 231–238. Leis J. (1991) Introduction. Dynamics of Reef Fish Populations pp. 121–122. Leis J.M. (2006) Are Larvae of Demersal Fishes Plankton or Nekton? Advances in Marine Biology, 51, 59–141. Leis J.M. (2010) Ontogeny of behaviour in larvae of marine demersal fishes. Ichthyological Research, 57, 325–342. Leis J.M. & Carson-Ewart B.M. (1997) In situ swimming speeds of the late pelagic larvae of some Indo-Pacific coral-reef fishes. Marine Ecology Progress Series, 159, 165–174. Leis J.M., Hay A.C., & Howarth G.J. (2009) Ontogeny of in situ behaviours relevant to dispersal and population connectivity in larvae of coral-reef fishes. Marine Ecology Progress Series, 379, 163–179. Leis J.M. & Mccormick M.I. (2002) The Biology, Behavior, and Ecology of the Pelagic, Larval Stage of Coral Reef Fishes. The Ecology of Fishes on Coral Reefs (ed. by P.F. Sale), pp. 171–200. Academic Press, Inc., San Diego, California. Leis J.M., Siebeck U., & Dixson D.L. (2011) How Nemo Finds Home: The Neuroecology of Dispersal and of Population Connectivity in Larvae of Marine Fishes. Integrative and Comparative Biology, 51, 826–843. Lessios H.A., Kessing B.D., Robertson D.R., & Paulay G. (1999) Phylogeography of the pantropical sea urchin Eucadaris in relation to land barriers and ocean currents. Evolution, 53, 806–817. Li G. & Hedgecock D. (1998) Genetic heterogeneity , detected by PCR SSCP, among samples of larval Pacific oysters (Crassostrea gigas) supports the the hypothesis of large variance in reproductive success. Canadian Journal of Fisheries and Aquatic Sciences, 55, 1025–1033.

251

Lischer H.E.L. & Excoffier L. (2012) PGDSpider: an automated data conversion tool for connecting population genetics and genomics programs. Bioinformatics (Oxford, England), 28, 298–9. 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. PNAS, 110, 16498–16502. Madduppa H.H., Timm J., & Kochzius M. (2014) Interspecific, spatial and temporal variability of self-recruitment in anemonefishes. PloS one, 9, e90648. Mastretta-Yanes A., Arrigo N., Alvarez N., Jorgensen T.H., Piñero D., & Emerson B.C. (2014) Data from: RAD sequencing, genotyping error estimation and de novo assembly optimization for population genetic inference. Mccormick M.I. & Gagliano M. (2009) Carry-over effects - the importance of a good start. Proceedings of the 11th International Coral Reef Symposium. McNicol R.E., Scherer E., & Gee J.H. (1996) Shoaling enhances cadmium avoidance by lake whitefish, Coregonus clupeaformis. Environmental Biology of Fishes, 47, 311–319. Milner-Gulland E.J., Fryxell J.M., & Sinclair A.R.E. (2011) Animal migration: a synthesis. Oxford University Press. Mizushima N., Nakashima Y., & Kuwamura T. (2000) Semilunar spawning cycle of the humbug damselfish Dascyllus aruanus. Journal of Ethology, 18, 105–108. Moberg P.E. & Burton R.S. (2000) Genetic heterogeneity among adult and recruit red sea urchins, Strongylocentrotus franciscanus. Marine Biology, 136, 773–784. Montgomery J.C., Tolimieri N., & Haine O.S. (2001) Active habitat selection by pre- settlement reef fishes. Fish and Fisheries, 2, 261–277. Mouritsen H., Atema J., Kingsford M.J., & Gerlach G. (2013) Sun Compass Orientation Helps Coral Reef Fish Larvae Return to Their Natal Reef. PloS one, 8, e66039. Munday P.L., Dixson D.L., Donelson J.M., Jones G.P., Pratchett M.S., Devitsina G. V., & Døving K.B. (2009) Ocean acidification impairs olfactory discrimination and homing ability of a marine fish. Proceedings of the National Academy of Sciences of the United States of America, 106, 1848–1852. Muss A., Robertson D.R., Stepien C.A., Wirtz P., & Bowen B.W. (2001) Phylogeography of Ophioblennius: the role of ocean currents and geography in reef fish evolution. Evolution, 55, 561–572. Nanninga G.B. & Berumen M.L. (2014) The role of individual variation in marine larval dispersal. Frontiers in Marine Science, 1, 1–17. Nanninga G.B., Saenz-Agudelo P., Zhan P., Hoteit I., & Berumen M.L. (2015) Not finding Nemo: limited reef-scale retention in a coral reef fish. Coral Reefs. Ohata R., Masuda R., Takahashi K., & Yamashita Y. (2014) Moderate turbidity enhances schooling behaviour in fish larvae in coastal waters. ICES Journal of

252

Marine Science, 71, 925–929. Paris C., Chérubin L., & Cowen R.K. (2007) Surfing, spinning, or diving from reef to reef: effects on population connectivity. Marine Ecology Progress Series, 347, 285–300. Paris C.B. & Cowen R.K. (2004) Direct evidence of a biophysical retention mechanism for coral reef fish larvae. Limnology and Oceanography, 49, 1964– 1979. Peakall R. & Smouse P.E. (2006) genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, 288– 295. Peakall R. & Smouse P.E. (2012) GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research--an update. Bioinformatics (Oxford, England), 28, 2537–9. Perez K.O. & Munch S.B. (2010) Extreme selection on size in the early lives of fish. Evolution, 64, 2450–2457. Peterson B.K., Weber J.N., Kay E.H., Fisher H.S., & Hoekstra H.E. (2012) Double Digest RADseq: An Inexpensive Method for De Novo SNP Discovery and Genotyping in Model and Non-Model Species. PloS one, 7. Pillai C.S.G., Mohan M., & Koya K.K.K. (1985a) Ecology and biology of the white tailed humbug Dascyllus aruanus (Pomacentridae, Pisces) from Minicoy Atoll. Journal of the Marine Biological Association of India, 27, 113–123. Pillai C.S.G., Mohan M., & Kunhikoya K.K. (1985b) A critique on the relationship of surface area of live coral with total number of fishes as well as the biomass of fish in a coexisting system of Chromis caeruleus and Dascyllus aruanus (Pomacentridae) at Minicoy Atoll. Journal of the Marine Biological Association of India, 27, 1–8. Pineda J., Hare J.A., & Sponaugle S. (2007) Larval Transport and Dispersal in the Coastal Ocean and Consequences for Population Connectivity. Oceanography, 20, 22–39. Pinsky M.L., Saenz-agudelo P., Salles C., Thorrold S.R., Jones G.P., Almany G.R., Bode M., Berumen M.L., Andréfouët S., & Planes S. (2017) Marine Dispersal Scales Are Congruent over Evolutionary and Ecological Time. Current Biology, 149–154. Pitcher T.J. & Parrish J.K. (1993) Functions of Shoaling Behaviour in Teleosts behaviour in teleosts. Behaviour of Teleost Fishes pp. 363–439. Springer. Planes S. (2002) Biogeography and Larval Dispersal Inferred from Population Genetic Analysis. Coral Reef Fishes pp. 201–220. Elsevier Science (USA). Planes S., Jones G.P., & Thorrold S.R. (2009) Larval dispersal connects fish populations in a network of marine protected areas. Proceedings of the National Academy of Sciences of the United States of America, 106, 5693–5697. Portugal S.J., Hubel T.Y., Fritz J., Heese S., Trobe D., Voelkl B., Hailes S., Wilson A.M., &

253

Usherwood J.R. (2014) Upwash exploitation and downwash avoidance by flap phasing in ibis formation flight Steven. Nature, 505, 399–402. Pritchard J.K., Stephens M., & Donnelly P. (2000) Inference of Population Structure Using Multilocus Genotype Data. Genetics, 155, 945–959. Quinn T.P. (2011) The behavior and ecology of Pacific salmon and trout. UBC press. Raymond M. & Rousset F. (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity. Robitzch V.S.N., Lozano-Cortés D., Kandler N.M., Salas E., & Berumen M.L. (2015) Productivity and sea surface temperature are correlated with the pelagic larval duration of damsel fi shes in the Red Sea. Marine Pollution Bulletin. Ross R.M., Backman T.W.H., & Limburg K.E. (1992) Notes: Group-Size-Mediated Metabolic Rate Reduction in American Shad. Transactions of the American Fisheries Society, 121, 385–390. Roughgarden J., Iwasa Y.O.H., & Baxter C. (1985) Demographic Theory for an Open Marine Population with Space-Limited Recruitment. Ecology, 66, 54–67. Rousset F. (2008) GENEPOP’ 007: a complete re-implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources, 8, 103–106. Sale P. (1970) Behaviour of the humbug fish. Australian Natural History. Scheltema R. (1971a) Dispersal of phytoplanktotrophic shipworm larvae (Bivalvia: Teredinidae) over long distances by ocean currents. Marine Biology. Scheltema R. (1971b) Larval dispersal as a means of genetic exchange between geographically separated populations of shallow-water benthic marine gastropods. The biological bulletin. Scheltema R. (1971c) The dispersal of the larvae of shoal-water benthic invertebrate species over long distances by ocean currents. 4th European marine biology symposium. Cambridge. Schunter C., Garza J.C., Macpherson E., & Pascual M. (2013) SNP development from RNA-seq data in a nonmodel fish: how many individuals are needed for accurate allele frequency prediction? Molecular Ecology Resources, 14, 157–165. Schwarz A. & Smith L. (1990) Sex change in the damselfish Dascyllus reticulatus (Richardson)(Perciformes: Pomacentridae). Bulletin of Marine Science. Seeley T.D. & Buhrman S.C. (1999) Group decision making in swarms of honey bees. Behavioral Ecology and Sociobiology, 45, 19–31. Selkoe K.A., Gaines S.D., Caselle J.E., & Warner R.R. (2006) Current Shifts and Kin Aggregation Explain Genetic Patchiness in Fish Recruits. Ecology, 87, 3082– 3094. Selkoe K.A. & Toonen R.J. (2011) Marine connectivity: a new look at pelagic larval duration and genetic metrics of dispersal. Marine Ecology Progress Series, 436, 291–305.

254

Selkoe K. a, Watson J.R., White C., Horin T. Ben, Iacchei M., Mitarai S., Siegel D., Gaines S.D., & Toonen R.J. (2010) Taking the chaos out of genetic patchiness: seascape genetics reveals ecological and oceanographic drivers of genetic patterns in three temperate reef species. Molecular Ecology, 19, 3708–3726. Shapiro D.Y. (1983) On the possibility of kin groups in coral reef fishes. Ecology of Deep and Shallow Reefs, 39–45. Shaw A.K. & Couzin I.D. (2013) Migration or Residency? The Evolution of Movement Behavior and Information Usage in Seasonal Environments. The American Naturalist, 181, 114–124. Shpigel M. & Fishelson L. (1986) Behavior and physiology of coexistence in two species of Dascyllus (Pomacentridae, Teleostei). Environmental Biology of Fishes, 17, 253–265. Simons A.M. (2004) Many wrongs: the advantage of group navigation. Trends in Ecology & Evolution, 19, 453–455. Skajaa K., Fernö A., & Folkvord A. (2003) Swimming , feeding and predator avoidance in cod larvae (Gadus morhua L.): trade-offs between hunger and predation risk. Proceedings of the 26th Annual Larval Fish Conference. Sponaugle S., Boulay J., & Rankin T. (2011) Growth- and size-selective mortality in pelagic -larvae of a common reef fish. Aquatic Biology, 13, 263–273. Sponaugle S., Cowen R.K., Shanks A., Morgan S.G., Leis J.M., Pineda J., Boehlert G.W., Kingsford M.J., Lindeman K.C., Grimes C., & Munro J.L. (2002) Predicting self- recruitment in marine populations: biophysical correlates and mechanisms. Bulletin of Marine Science, 70, 341–375. Staaterman E. & Paris C.B. (2014) Modelling larval fish navigation: the way forward. ICES Journal of Marine Science, 71, 918–924. Strathmann R. (1974) The spread of sibling larvae of sedentary marine invertebrates. American Naturalist. Strathmann R. (1980) Why does a larva swim so long? Paleobiology. Strathmann R.R. (1990) Why Life Histories Evolve Differently in the Sea. American Zoologist, 30, 197–207. Strathmann R.R., Hughes T.P., Kuris A.M., Lindeman K.C., Morgan S.G., Pandolfi J.M., & Warner R.R. (2002) Evolution of local recruitment and its consequences for marine populations. Bulletin of Marine Science, 70, 377–396. Sumpter D.J.T. (2006) The principles of collective animal behaviour. Philosophical Transactions Of The Royal Society B-Biological Sciences, 361, 5–22. Sumpter D.J.T., Krause J., James R., Couzin I.D., Ward A.J.W., & Hall G. (2008) Report Consensus Decision Making by Fish. Current Biology, 18, 1773–1777. Thornhill D.J., Mahon A.R., Norenburg J.L., & Halanych K.M. (2008) Open-ocean barriers to dispersal: a test case with the Antarctic Polar Front and the ribbon

255

worm Parborlasia corrugatus (Nemertea: Lineidae). Molecular Ecology, 17, 5104–5117. Thorrold S.R., Jones G.P., Hellberg M.E., Burton R.S., Swearer S.E., Neigel J.E., Morgan S.G., & Warner R.R. (2002) Quantifying larval retention and connectivity in marine populations with artificial and natural markers. Bulletin of Marine Science, 70, 291–308. Thresher R.E., Colin P.L., & Bell L.J. (1989) Planktonic Duration, Distribution and Population Structure of Western and Central Pacific Damselfishes (Pomacentridae). Copeia, 1989, 420–434. Toonen R.J. & Pawlik J.R. (2001) Foundations of gregariousness: a dispersal polymorphism among the planktonic larvae. Evolution, 55, 2439–2454. Torney C., Neufeld Z., & Couzin I.D. (2009) Context-dependent interaction leads to emergent search behavior in social aggregates. PNAS, 106, 22055–22060. Torney C.J., Berdahl A., & Couzin I.D. (2011) Signalling and the Evolution of Cooperative Foraging in Dynamic Environments. PLoS Computational Biology, 7, 1–10. Torney C.J., Levin S.A., & Couzin I.D. (2010) Specialization and evolutionary branching within migratory populations. PNAS, 107, 20394–20399. Tremblay M.J., Loder J.W., Werner F.E., Naimie C.E., Page F.H., & Sinclair M.M. (1994) Drift of sea scallop larvae Placopecten magellanus on Georges Bank: a model study of the roles of mean advection, larval behavior and larval origin. Deep Sea Research II, 41, 7–49. Treml E.A., Halpin P.N., Urban D.L., & Pratson L.F. (2008) Modeling population connectivity by ocean currents, a graph-theoretic approach for marine conservation. Landscape Ecology, 23, 19–36. Veliz D., Duchesne P., Bourget E., & Bernatchez L. (2006) Genetic evidence for kin aggregation in the intertidal acorn barnacle (Semibalanus balanoides). Molecular Ecology, 15, 4193–4202. Victor B.C. (1986) Duration of the planktonic larval stage of one hundred species of Pacific and Atlantic wrasses (family Labridae). Marine Biology, 90, 317–326. Vikebø F., Jørgensen C., Kristiansen T., & Fiksen Ø. (2007) Drift, growth, and survival of larval Northeast Arctic cod with simple rules of behaviour. Marine Ecology Progress Series, 347, 207–219. Ward A.J.W., Sumpter D.J.T., Couzin I.D., Hart P.J.B., & Krause J. (2008) Quorum decision-making facilitates information transfer in fish shoals. PNAS, 105, 6948–6953. Warner R.R. & Hughes T.P. (1988) The Population Dynamics of Reef Fishes. Proceedings of the 6th International Coral Reef Symposium, 1, 149–155. Watson J., Mitarai S., Siegel D., Caselle J.E., Dong C., & McWilliams J. (2010) Realized and potential larval connectivity in the Southern California Bight. Marine

256

Ecology Progress Series, 401, 31–48. Weersing K. & Toonen R.J. (2009) Population genetics, larval dispersal, and connectivity in marine systems. Marine Ecology Progress Series, 393, 1–12. Werner F.E., Page F.H., Lynch D.R., Loder J.W., Lough R.G., Perry R.I., Greenberg D.A., & Sinclair M.M. (1993) Influences of mean advection and simple behavior on the distribution of cod and haddock early life stages on Georges Bank. Fishries Oceanography, 2, 43–64. Willis J. (2011) Modelling swimming aquatic animals in hydrodynamic models Modelling swimming aquatic animals in hydrodynamic models. Ecological Modelling, 222, 3869–3887. Wong M.Y.L.Y.L., Fauvelot C., Planes S., & Buston P.M. (2012) Discrete and continuous reproductive tactics in a hermaphroditic society. Animal Behaviour, 84, 897–906.

257

CHAPTER V

INFLUENCES OF HABITAT LOCATION ON FORAGING AND MORPHOLOGICAL

TRAITS OF DASCYLLUS ARUANUS

5.1 INTRODUCTION

Relatively little is known about the eco-morphological variation of coral reef fish populations. Natural selection operates on said phenotypic variation present in a species population, which ultimately may lead to differentiation between populations and the evolution of species (Rundell & Price, 2009). Naturally, populations harbor diverse phenotypes and documenting the amount of variations among those may help understand micro-evolutionary processes.

Highly heterogeneous environments can lead to the differentiation not only of a great number of phenotypes but ultimately a great number of fish species among others, driven by the presence of different trophic niches (Faulks et al., 2015). Coral reefs belong to such highly heterogeneous environments (e.g. Monismith et al.,

2006; Reidenbach et al., 2006), in which a yet unknown but large number of species have found their trophic niche. Hence, coral reefs are (one of) the most species rich ecosystems on the planet (MacArthur & Levins, 1964). In this regard, the assessment of cephalic morphology as well as positioning of fins in relation to the forage preferences of the fishes even within single species can give insight into the plasticity in adaptive strategies of coral reef fishes. Eco-morphological studies reveal associations between those changes in body structures and shapes as a response to 258 the conditions in their habitat (e.g. strength of currents (Webb, 1984; Blake, 2004;

Fulton & Bellwood, 2005; Ohlberger et al., 2006; Langerhans, 2008), or food availability (Aguilar-Medrano et al., 2012; Frédérich et al., 2012). Furthermore, and in the light of genetics, it can also help infer the degree to which such adaptive changes may be shaping speciation processes (either hindering or pushing speciation depending on the phenotypic plasticity of the species) and whether or not there is a correlation between the two components. Genetics can also determine whether the changes are purely phenotypic plasticity or adaptation through natural selection.

This chapter focuses on the morphological and dietary features of different populations of Dascyllus aruanus, a wide-spread Indo-Pacific damselfish species, along a cross-shelf gradient (small spatial scale) in the central Red Sea as well as between two biogeographical regions: the Red Sea and the western Indian Ocean.

Three morphological traits and the trophic ecology have been compared among four populations from the Red Sea and Madagascar. This comparative approach can help understand which forces are driving morphological changes within a population and between populations in different biogeographic regions and putatively environments. It will highlight if the phenotypic responses are region specific or can be globally observed in correlation to the habitat and thus, a more general and species-specific trend. Even though this species is known to be zooplanktivorous

(Randall & Allen, 1977), more recent studies have shown that a substantial portion of its diet can also be comprised of small benthic invertebrates (Frédérich et al.,

2010). When kept in captivity, it also exhibits opportunistic foraging behavior (Pillai 259 et al., 1985). The analyses of the food composition of D. aruanus together with environmental data and the habitat description may also harbor answers regarding adaptive strategies and plasticity.

This chapter was a collaborative work between Bruno Frédérich from the University of Liège and myself. My work consisted of the experimental design, collection of samples and measurement of biological and environmental metadata in the Red Sea, and the co-interpretation of the results. My collaborator in Belgium, together with his MSc student, Chia-Ting Chen, conducted the analyses of geometric- morphometrics, gut content, and isotope ratios (trophic level). Therefore, this chapter is kept relatively short and rather focuses on the interpretation of results in the light of the knowledge and findings gathered in all of the previous chapters.

The main objective of this collaborative work was to assess to which degree habitat location and food source can influence coral reef fish populations by measuring and comparing morphological traits and trophic levels (dietary composition) of D. aruanus populations across a cross-shelf gradient in the Red Sea as well as to a site outside the Red Sea, in Madagascar. This assessment, in the light of genetic findings and further analysis of genetic data, may in the future help define how and where adaptation is occurring and at which geographic scale/geographic areas.

Alternatively, it may reveal if the main differences are purely phenotypic plasticity.

260

5.2 MATERIALS AND METHODS

5.2.1 Study Site and Sample Collection of Dascyllus aruanus

a) b)

Depth (m) Abu Shosha -1,060 - -500

Al Fahal -499 - -200 Central Red Sea -199 - -100 Shib Nazar t -99 - -50 Land Madagascar -49 - -20 1000 km 10 km Reef -19 - -1

Figure 5.1: (a) Sampling sites of Dascyllus aruanus from the central Red Sea (marked by a star) and Madagascar (marked by a triangle). (b) Location of the sites in the central Red Sea along a cross-shelf gradient. A different colored star represents each site, which was chosen according to its location on the reef scape and bathymetric characteristics. The bathymetric map was provided by Maha Khali et al. (in review).

Dascyllus aruanus samples were taken from an inshore (ASU: Abu Shosha), mid- shelf (AFA: Al Fahal), and shelf-edge (SNA: Shib Nazar) reef off Thuwal, Saudi

Arabia, in the Central Red Sea. Six similar-sized (20-30 cm) Pocillopora verrucosa coral colonies were sampled using hand nets, tweezers, a large plastic bag and clove-oil in order to collect about 30 whole specimens of D. aruanus (Table 5.1). For 261 comparative analysis samples were taken from the northern sheltered site of the reef in similar habitats (sheltered patchy and sandy lagoons), as possible at similar depths (10 m (ASU and AFA) to 14 m (SNA)) and from colonies exclusively inhabited by D. aruanus as pomacentrid coral-dwellers. Sampling took place only in the afternoon hours to increase the likelihood of finding undigested prey, which facilitates identification. The samples were immediately stored in 95% ethanol. The metadata from the Red Sea provided for this study (visibility, current speed, distance from coast; Table 5.1), was taken from Chapter II. Whole specimens were shipped to the University of Liege where the analyses under 5.2.1 and 5.2.2 were performed. The samples from Madagascar (TL: Tuléar) were collected by Bruno

Frédérich under similar conditions but from shallower depths (6-8m). Samples were stored in 70% ethanol.

5.2.2 Geometric-Morphometric Analyses

Geometric-morphometric analyses were performed by Chia-Ting Chen (see MSc thesis (Chen 2016) for a detailed description) following Klingenberg et al., (2003) and Frédérich et al. (2012), to assess putative significant variations in morphological features between and within locations. In short, the body shape, attachment angle of pectoral fin, and the eye size were the main characteristics measured. Here fore, each individual was photographed and Figure 5.2 shows the points of comparison of the fish’s body-scape.

Tableau 2. Résumé du nombre d’individus étudiés et de la variance de leur taille (longueur standard). site SL (mm) forme du contenus isotopes corps stomacaux stables nombre 8.8-57.8 141 n= 133 n= 136 d’individus n0= 7 n1= 8 total (N) Al Fahal 13.3-45.8 37 n= 33 n= 34 (AFA) n0= 4 n1= 3 Abu Shosha 11.8-49.4 29 n= 28 n= 26 (ASU) n0= 1 n1= 3 Shi’b Nazar 8.8-57.8 38 n= 35 n= 38 (SNA) n0=2 n1= 2 Tuléar 10.31-52.05 37 n= 37 n= 38 (TL) n0= 0 n1= 0 n est le nombre d’individus pris en compte dans les analyses statistiques, n0 est le nombre d’individus ayant un estomac vide ou qui ne contient peu d’éléments. n1 est le nombre d’individus ayant très peu de matière de muscle donc les compositions isotopiques ne sont pas pris en compte dans les analyses statistiques

262

Fig. 3. 21 Points de repères homologues (LMs) utilisés pour l’analyse de la forme du corps : (1) extrémité antérieure du premaxillaireFigure 5.2 ; (2): Display extrémité postérieur of the du location maxillaire of ; (3) 21 extremité points antérieurefor the de l’oeil; homologous (4) centre de l’oeil; (5–7) extremité inférieure,analysis postérieure, of body et shape supérieureof de Dascyllus l’oeil ; (8) coinaruanus postéro -ventralimplemented du préopercule; in (9) linear extrémité supérieur de l’os préoperculaire ; (10) extrémité postérieure de l’opercule; (11) extrémité supérieure de l’opercule ; (12) insertion de l’opercule sur le profilmodels du corps; (see (13) Chen insertion (2016) de la nageoi for re pelvienne a detailed ; (14-15)description). insertion inférieure The et supérieurepicture de was la nageoire pectorale ; (16) et (18)provided by Chen (2016). insertion antérieure et postérieure de la nageoire dorsale ; (17) et (19) insertion antérieure et postérieure de la nageoire anale ; (20) et (21) base dorsale et ventrale de la nageoire caudale.

5.2.3 Assessment of Trophic Niche

For the assessment of trophic niche and food source the stomach content of each individual was analyzed and muscle tissue was dissected to perform isotopic 6 analyses (13C and 15N). In general, the isotopic signature 13C are used to assess the source of carbon of a fish’s diet, while 15N gives an estimate of the trophic level of the individual (Vander Zanden et al., 1997). Stomach contents provided an insight into the food source and, roughly, whether the fish was feeding on benthic or pelagic organisms. From all collected specimens only two had an empty stomach and were discarded a priori from the analysis (see MSc thesis (Chen 2016) for further details). 263

Table 5.1 includes information of number of fishes collected and analyzed for this

section.

Table 5.1: Number of Samples of Dascyllus aruanus per Site and Site-Specific Information Coast SL VIZ Depth Location Site N Prox. Latitude Longitude Reef Name (mm) (m) (m) (km) RS SNA 42(38) 8.8-57.8 25 35.60 14 22°20'51.96"N 38°51'09.42"E Shib Nazar AFA 37(34) 13.3-45.8 10 22.14 10 22°17'49.20"N 38°57'32.40"E Al Fahal ASU 29(28) 11.8-49.4 3 25.15 10 21°40'16.53"N 38°50'36.60"E Abu Shosha Total 108(100) 8.8-57.8 12.67 27.63 11.33 IO TL 38 10.31-52.05 1 12.00 6-8 23°21'34.4"S 43°39'13.9"E Tuléar

Total 146(138) 8.8-57.8 9.75 23.72 10.25 Sampling locations were SNA, AFA, and ASU (from shore to shelf-edge) inside the Red Sea (RS). Outside the RS and in the Indian Ocean (IO), samples were taken from the lagoon of Tuléar (TL) off Madagascar. Total number of samples are given underneath as N and in parenthesis the number of those which were used for the analyses under 5.3.1 and 5.3.2. Standard length (SL) of the fishes was measured for each site. For statistical analyses of correlations to environmental parameter the proximity to the coast (Coastal Prox.), the visibility (VIZ), and the sampling depths (Depth) of each site were measured, for which the total averages are also given in bold.

5.3 RESULTS

5.3.1 Geometric-Morphometric Analyses

The variation in allometric growth within site was always larger than between sites.

Therefore, our null hypothesis (i.e., that there would be the same pattern in

allometric growth for each population of Dascyllus aruanus) could not be rejected. In

contrast, overall morphological differences were found not only at the large scale

and between the biogeographic provinces (Red Sea and Madagascar, Figure 5.3) but 264 d)d) a) a)

also at the small scale and along the cross-shelf gradient within the central Red Sea

(Figure 5.4). e) e)

d) a) a) b)b) b) c) c)c)

d) a)

FigFig.5 CVA.5 CVA de dela formela forme du ducorps corps chez chez D. D.aruanus aruanus appliquée appliquée aux aux quatre quatre sites sites d’étude d’étude (AFA, (AFA, ASU, ASU, SNA, SNA, TL) TL). (a). D(a)iagramme Diagramme de dispersionde dispersion montrant montrant les variatioles variations den sformes de formes entre entre populations. populations. LesL eellipsess ellipses de defréquence fréquence à 90% à 90% des des observations observations sont sont illustrées illustrées. (b). (b)et (c) et (c)configuration configuration de pointsde points repères repères illustrant illustrant la variation la variation de formede forme le long le long de l’axe de l’axe CV1. CV1. (d) et(d) (e) et configuration(e) configuration de points de points repèresrepères illustrant illustrant la variationla variation de deforme forme le longle long de del’axe l’axe CV2. CV2. La Laconfiguration configuration « bleu « bleu clai clair » rreprésente » représente la forme la forme moyenne moyenne de tousde tous les e)individusles individus (consensus) (consensus) et la et configuration la configuration « bleu « bleu foncé foncé » » e) représentereprésente la formela forme aux aux valeurs valeurs extrême extrêmes des chaquede chaque axe axe CV. CV. d)d) a) e) Canonical Variable 2

12 12 b) c) Canonical Variable 1 b) c)

e) Figure 5.3: Canonical analysis of the variation (CVA) in body shape of Dascyllus aruanus from four sites, three inside the central Red Sea (AFA (red), ASU (green),

and SNA (blue)) and one off the south-western coast of Madagascar (Figb) .5 CVA de la forme du corps chez D.TL aruanus (purple)appliquée aux c)quatre sites). d’étude (AFA, ASU, SNA, TL). (a) Diagramme de dispersion montrant les variations de formes entre populations. Les ellipses de fréquence à 90% des observations sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) et (e) configuration de points repères illustrant la variation de forme le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » (a) shows the dispersion of variation in body shape between populations with each représente la forme aux valeurs extrêmes de chaque axe CV.

population cluster color-coded. (b) and (c) are the configuration points used to Fig.5 CVA de la forme du corps chez D. aruanus appliquée aux quatreillustrate sites d’étude (AFA, the ASU, variation SNA, TL). (a) Diagramme in de body dispersion montrant shape les variatio a nslong de formes entre the populations.canonical variable 1 (CV1) axis, Les ellipses de fréquence à 90% des observations sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) et (e) configuration de points repères illustrant la variation de forme le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » 12 représente la forme aux valeurs extrêmes de chaque axe CV. while (d) and (e) represent those for the CV2 axis. Among these latter figures, the light blue profile is the mean (consensus) body shape of all analyzed individuals, Fig.5 CVA de la forme du corps chez D. aruanus appliquée aux quatre sites d’étude (AFA, ASU, SNA, TL). (a) Diagramme de dispersion montrant les variations de formes entre populations. Les ellipses de fréquence à 90% des observations sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) et (e) configuration de points and the dark blue profile the measurements for extreme body shapes of each axis.repères illustrant la variation de forme le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » représente la forme aux valeurs extrêmes de chaque axe CV. Graphs and figures were provided by Chen (2016). 12

d) d) a) a)

265

e) e)

b) c) a) b) b) c) c) d) a)

d) a)

Fig.6 CVA de Fig.la forme6 CVA du de corps la forme chez D.du aruanuscorps chez appliquée D. aruanus aux troisappliquée sites d’étudeaux trois (AFA, sites d’étudeASU, SNA) (AFA,. (a) ASU, Diagramme SNA). (a) de Ddispersioniagramme montrantde dispersion les variatio montrantns de les formes variatio entrens de populations. formes entre L epopulations.s Les ellipses de fréquenceellipses à de90% fréquence des observations à 90% des sont observations illustrées. sont(b) etillustrées (c) configuration. (b) et (c) deconfiguration points repères de illustrantpoints repères la variation illustrant de laforme variation le long de deforme l’axe le CV1.long de(d) l’axeet (e) CV1. configuration (d) et (e) deconfiguration points de points repères illustrantrepères la variation illustrant de laforme variation le long de deforme l’axe le CV2.long deLa l’axeconfiguration CV2. La «configuration bleu clair » représente « bleu clai lar »forme représente moyenne la forme de tous moyenne les individus de tous (consensus) les individus et (consensus)la configuration et la «configuration bleu foncé » « bleu foncé » représente la formereprésente aux valeurs la forme extrême aux valeurss de chaque extrême axes CV.de chaque axe CV.

d) d) e) e) a) e)

Canonical Variable 2 14 14 b) c)

Canonical Variable 1 b) c) e) Figure 5.4: Canonical analysis of the variation (CVA) in body shape of Dascyllus

aruanus from the three sites inside the central Red Sea (AFA (red), ASU (green), and b) Fig.6 CVA de la forme du corps chez D. aruanus appliquéec) aux trois sites d’étude (AFA, ASU, SNA). (a) Diagramme de dispersion montrant les variations de formes entre populations. Les ellipses de fréquence à 90% des observations sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) et (e) configuration de points repères illustrant la variation de forme le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » SNA (blue)). (a) shows the dispersion of variation in body représente la shape forme aux valeurs extrême between s de chaque axe CV.

populations with each population cluster color-coded. (b) and (c) are the

configuration points used to illustrate the variation in body shape a long the

canonical variable 1 (CV1) axis, while (d) and (e) represent those for the CV2 axis. 14

Fig.6 CVA de la forme du corps chez D. aruanus appliquée aux trois sites d’étude (AFA, ASU, SNA). (a) Diagramme de dispersion montrant les variations de formes entre populations. Les Fig.6 CVA de la forme du corps chez D. aruanus appliquée aux trois sites d’étude (AFA,Among these latter figures ASU, SNA). (a) Diagramme de dispersion montrant, the light blue profile is the mean (consensus) body shape les variations de formesellipses deentre fréquence populations. à 90% des observations Les sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) et (e) configuration de points ellipses de fréquence à 90% des observations sont illustrées. (b) et (c) configuration de points repères illustrant la variation de forme le long de l’axe CV1. (d) etrepères (e) configurationillustrant la variation de de formepoints le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » of all analyzed individuals, and the dark blue profile the measurements for extreme représente la forme aux valeurs extrêmes de chaque axe CV. repères illustrant la variation de forme le long de l’axe CV2. La configuration « bleu clair » représente la forme moyenne de tous les individus (consensus) et la configuration « bleu foncé » représente la forme aux valeurs extrêmes de chaque axe CV. body shapes of each axis. Graphs and figures were provided by Chen (2016).

14 All in all, the most relevant and interesting results in morphology were the 14 differences between sites in angle of attachment of the pectoral fin (1) and the size

of the eyes (2) (Figure 5.5). The effect of the site was significant on these features

(see Chen (2016) for statistical details). 266

Eye Diameter

AYachment Angle

Fig. 4 Illustration des mesures du diamètre de l’œil et l’angle d’insertion de la nageoire pectorale. Pour obtenir l’angle de l’attachementFigure de 5.5 la nageoire: Illustration pectorale, une ligneof the measurement of the diameter of eye (notée a) qui passe par le bout du museau et le milieu de la nageoire - caudasize le. Ensuite, une ligne parallèle (notée b) à la précédente qui passe par l’insertion inférieure de la nageoire pectorale est tracée. Finalement,and the attachment angle of the pectoral fin. To obtain the latter, a line une ligne qui passe par les points d’insertion inférieur et supérieur est tracée créant l’angle de mesure. (a) is drawn, which cuts through the tip of the fish’s mouth and the middle of the caudal fin. Line (b) is drawn along the bottom insertion of the pectoral fin parallel to line a. A last line (c) is placed along the top and bottom insertion points of the pectoral fin, which gives the angle of attachment. The illustration was provided by Chen (2016).

In short, (1) the attachment of the pectoral fin for each site was 67.7°, 74.7°, 76.6°,

80.6° for sites TL, SNA, ASU, and AFA, respectively (Figure 5.6). The most significant differences were for the TL site, which had the most horizontally placed pectoral fin and for AFA, which had the most vertically placed one. Differences between SNA and

ASU were not significant. In terms of (2) eye-sizes, the measured values were 4.3,

4.4, 4.6, and 4.8 mm, for the sites SNA, AFA, ASU, and TL, respectively. There may be a negative correlation between visibility and eye size and the most significant differences in eye-size were measured between SNA and AFA, which are the most extreme sites in terms of differences in visibility within the Red Sea. The site of TL 8 had overall lowest visibility as well as the largest eyes. 267

Figure 5.6: Boxplot of the measurements of the attachment angle of the pectoral Fig. 7 Boxplot de l’angle d’attachement de la nageoire pectorale chez chacune des populations. fins of Dascyllus aruanus from sites inside the Red Sea (AFA, ASU, and SNA populations) vs Madagascar (TL population; x axis). Whiskers represent a 95% confidence interval, each box displays the mean values (bold line) with the upper and lower quartiles. The angles are given in degrees on the y-axis. The figure is 100%modified from Chen (2016).

80%

60% benthos zooplancton 40%

20%

0% AFA ASU SNA RS TL

Fig. 8 Abondances relatives du zooplancton et du benthos chez D. aruanus à chaque site (AFA, ASU, SNA) et régions (RS, TL) ; RS = la moyenne des sites de Mer Rouge.

16 268

5.3.2 Assessment of Trophic Niche

100%

80%

Benthos 60%

40% Plankton

20%

0% AFA ASU SNA RS TL Figure 5.7: Histogram of the relative abundances of zooplankton (blue) and benthic (red) organisms present in the stomach contents of Dascyllus aruanus from sites inside the central Red Sea (AFA, ASU, SNA) and from Madagascar (TL). The column labeled as RS represents the mean values from the central Red Sea sites. The figure is modified from Chen (2016).

Inﬂuences of Habitat Loca2on on Foraging and Morphological Traits of Dascyllus aruanus 2 5.4 DISCUSSION AND CONCLUSIONS

Species descriptions have traditionally been based on the morphological traits of

individuals. However, these traits may be highly plastic for some species and some

diagnostic traits may be relatively cryptic. Currently, genetic analyses are playing an

increasingly important role in studies of speciation and phylogeny. Nonetheless, the

description of evolutionary processes and their documentation remains a challenge.

The combination of both genetics and eco-morphological analyses of coral reef fish

populations may thus enable a holistic view on how micro-evolutionary processes

work on speciation at different temporal, spatial, and ecological scales.

269

This chapter is one of only a few attempts so far to combine the two previously mentioned disciplines (ecomorphology and genetics) for the inference of potential drivers of differentiation of coral reef fish populations (Frédérich et al. 2012). It is also among the first to assess the extent to which the geographic location of a habitat type and food source can influence morphological traits of a wide spread coral reef fish species, Dascyllus aruanus. Furthermore, it is intended to be a first stepping-stone towards comparative studies to assess whether such plasticity is a trait more pronounced in widespread vs. endemic species and eventually help further understanding of forces shaping patterns of biogeography and biodiversity in coral reef fishes.

Here, we found that there are significant differences in morphology among Dascylus aruanus at the small scale of its distribution (along a cross-shelf gradient inside the

Red Sea) as well as between populations along its wider distribution range

(between the Red Sea and Madagascar). The most interesting eco-morphological features in this comparison were eye-size and the attachment angle of the fin. The stomach content of the specimen was dissected to assess the dietary composition at each site and the proportions of benthic vs. planktonic prey.

Within the Red Sea, the eye size seemed to increase with decreasing visibility in the habitat. The offshore site had the highest visibility and was also the site with the smallest eye diameter. Additionally this seemed to be also correlated to the dietary composition. At this offshore site, the main food source of D. aruanus was zooplankton. At the site with the biggest eye-size (mid-shelf) the diet had a significant amount of benthic prey, while the inshore site was in all regards always 270 intermediate. From the results in Chapter I (for which this same sites were analyzed for seasonality in recruitment), the models of biological response variables also tended to categorize the reefs in that manner with mid-shelf and offshore being the extremes in the distribution of results, while the inshore reef was intermediate. I hypothesize the following: the mid-shelf reef is the biggest and most species rich reef along that cross-shelf gradient. It has also the biggest sheltered lagoon from all three reefs, the lowest current speeds, and the lowest visibility. Here, D. aruanus optimizes its feeding strategies. It develops bigger eyes to find prey better, but also does not rely on the harder-to-spot planktonic prey but chooses to feed on benthic organisms. It might also be that habitats with lower visibility and current speeds are more favorable for the abundance and wider diversity of benthos. Thus, D. aruanus opportunistically uses that new food source. Offshore, the currents are faster, and the visibility significantly greater. Here, D. aruanus has better opportunities to catch pelagic prey, and in general the small benthic community may be less abundant and diverse. The angle of the attachment of the fin at this site also demonstrates a phenotypic adjustment as the offshore population has an angle optimized to swim more efficiently in stronger currents. Inshore, I observed that the conditions are more variable and intermediate. Depending on the direction of the wind, current speed and direction changes, and so, the source of water, its visibility and the zooplankton components and abundances may also vary. Therefore, it is a good strategy to maintain an intermediate morphological and dietary plan.

Comparatively, at the site in Madagascar, visibility is very low. The site is very close to the coast and a large shallow lagoon. Here the dietary composition and eye- 271 diameters are similar to those of the Red Sea sheltered mid-shelf site with low visibility. Thus, it is not surprising to find large eyes and a significant proportion of benthic prey. However, the attachment angle of the fin is quite small and smaller than that from the offshore site in the Red Sea. Even though this site off Madagascar is at the shore in a sheltered lagoonal area, current speeds are relatively strong and stronger than those observed at the offshore site in the Red Sea. Therefore, this result is in line with the observations in eco-morphology made within the Red Sea.

The generality of these observations, however, must be tempered by the fact that we only have one sampling site in Madagascar.

Due to the assessment of some environmental parameters of each site and the differences in spatial scale between the comparisons, we were able to resolve that eco-morphological differences in the widespread D. aruanus are highly likely a response to environmental conditions and the location of the species habitat and, thus, a sign of phenotypic plasticity of the species. We believe that adaptation through natural selection is most likely not what is driving these morphological differences. In terms of foraging behavior, these geographical differences in D. aruanus further provide evidence of the adaptive advantage of ecological plasticity of trophic strategies, which may be one key component in species with wide biogeographical distributions. In this context, it will be very helpful to apply a similar analytical approach on the endemic species D. marginatus and compare the results to study differences in ecological adaptation in relation to distribution ranges of coral reef fishes. For instance, D. aruanus often recruits to big conspecific 272 colonies and maintains body conditions potentially due to behavioral plasticity in feeding. Dascyllus aruanus might even benefit from the presence of adults ((Booth,

1995): higher body energy reserves are found in juveniles that share a colony with adults), potentially through observing and learning how to successfully forage in the new home. In contrast, the feeding success of the endemic D. marginatus significantly decreases with group size (Kent et al., 2006). This, among others, could be a factor defining distributions. Temporal variation in food availability (i.e., season) could also influence plasticity and may affect some species more strongly than others.

5.5 REFERENCES

Aguilar-Medrano R., Frédérich B., Balart E.F., & Luna E. (2012) Diversification of the pectoral fin shape in damselfishes (Perciformes, Pomacentridae) of the Eastern Pacific. Zoomorphology, 132, 197–213. Blake R. (2004) Fish functional design and swimming performance. Journal of Fish Biology, 65, 1193–1222. Booth D.J. (1995) Juvenile Groups in a Coral-Reef Damselfish: Density-Dependent Effects on Individual Fitness and Population Demography. Ecology, 76, 91–106. Faulks L., Svanbäck R., Eklöv P., & Östman Ö. (2015) Genetic and morphological divergence along the littoral-pelagic axis in two common and sympatric fishes: perch, Perca fluviatilis (Percidae) and roach, Rutilus rutilus (Cyprinidae). Biological Journal of the Linnean Society, 114, 929–940. Frédérich B., Lehanse O., Vandewalle P., & Lepoint G. (2010) Trophic Niche Width, Shift, and Specialization of Dascyllus aruanus in Toliara Lagoon, Madagascar. Copeia, 2010, 218–226. Frédérich B., Liu S.-Y. V., & Dai C.-F. (2012) Morphological and Genetic Divergences in a Coral Reef Damselfish, Pomacentrus coelestis. Evolutionary Biology, 39, 359–370. Fulton C.J. & Bellwood D.R. (2005) Wave-induced water motion and the functional implications for coral reef fish assemblages. Limnology and Oceanography, 50, 255–264. 273

Kent R., Holzman R., & Genin A. (2006) Preliminary evidence on group-size dependent feeding success in the damselfish Dascyllus marginatus. Marine Ecology Progress Series, 323, 299–303. Klingenberg C.P., Barluenga M., & Meyer A. (2003) Body shape variation in cichlid fishes of the Amphilophus citrinellus species complex. Biological Journal of the Linnean Society, 80, 397–408. Langerhans R.B. (2008) Predictability of phenotypic differentiation across flow regimes in fishes. Integrative and Comparative Biology, 48, 750–768. MacArthur R. & Levins R. (1964) Competition, habitat selection, and character displacement in a patchy environment. Proceedings in Natural Science, 51, 1207–1210. Monismith S.G., Genin A., Reidenbach M.A., Yahel G., & Koseff J.R. (2006) Thermally Driven Exchanges between a Coral Reef and the Adjoining Ocean. Journal of Physical Oceanography, 36, 1332–1347. Ohlberger J., Staaks G., Hoelker, & F (2006) Swimming efficiency and the influence of morphology on swimming costs in fishes. Journal of Comparitive Physiology B, 176, 17–25. Pillai C.S.G., Mohan M., & Koya K.K.K. (1985) Ecology and biology of the white tailed humbug Dascyllus aruanus (Pomacentridae, Pisces) from Minicoy Atoll. Journal of the Marine Biological Association of India, 27, 113–123. Randall H. a. & Allen G.R. (1977) A revision of the damselfish genus Dascyllus (Pomacentridae) with the description of a new species. Records of the Australian Museum, 31, 349–385. Reidenbach M.A., Monismith S.G., & Koseff J.R. (2006) Boundary layer turbulence and flow structure over a fringing coral reef. Limnology and Oceanography, 51, 1956–1968. Rundell R.J. & Price T.D. (2009) Adaptive radiation, nonadaptive radiation, ecological speciation and nonecological speciation. Cell. Webb P.W. (1984) Body Form, Locomotion and Foraging in Aquatic Vertebrates. American Zoologist, 24, 107–120. Vander Zanden M.J., Cabana G., & Rasmussen J.B. (1997) Comparing trophic position of freshwater fish calculated using stable nitrogen isotope ratios (d15 N) and literature dietary data. Canadian Journal of Fisheries and Aquatic Sciences, 54, 1142–1158.

274

CHAPTER VI

FINAL DISCUSSION

This dissertation seeks to link and better understand historical and ecological processes shaping biogeography and genetics of two ecologically and genetically closely related damselfishes, Dascyllus aruanus and D. marginatus. Despite their similarities, these two species have remarkably different biogeographic ranges.

While much information on the endemic D. marginatus is presented, the thesis strongly focuses on the widespread and more commonly studied species, D. aruanus, for which it also provides extensive data on putative dispersal, recruitment, adaptation, and foraging strategies. Additionally, this piece of work provides first measurements of recruitment peaks of coral reef fishes in the Red Sea, baseline knowledge that is still widely lacking in this region.

6.1 BIOGEOGRAPHY

The primary content of this work relates to findings that can help understand putative drivers of biogeography of coral reef fishes. Biogeographic phenomena are not limited to differences found at the borders of biogeographic provinces, but there are differences in species composition between reefs across various quantifiable gradients and at small scales. In the biogeographic province of the Red Sea (as defined by Spalding et al., 2007) we find at least two such gradients: (1) an 275 environmental latitudinal gradient, which is composed by productivity (food availability), temperature, salinity, visibility, and monsoon-driven shifts in currents.

At the center of this gradient, recruitment peaks of coral reef fishes are one or two months after the northern hemisphere’s summer, different than what is observed in most other boreal locations (Chapter II). (2) A longitudinal cross-shelf gradient, along which reefs are exposed to different oceanographic conditions and, thus, have site specific features, which fishes seem to be responding to. For instance, while most Red Sea fish species are found along most of latitudinal gradient of the Red Sea

(Roberts et al., 2016), some may only be inhabiting and/or recruiting to a certain reef-type along the cross-shelf gradient (Chapter II). Moreover, while I found genetic

(SNPs and MSATs) and biological (PLDs; life-history traits) differences along this latitudinal environmental gradient of the Red Sea, I do not find these at the smaller spatial scales of the cross-shelf gradient. (At least this is not the case for the widespread species, D. aruanus.) However, and more interestingly, D. aruanus does show eco-morphological differences between reefs across the shelf. These differences correlate with site-specific traits (such as visibility and current speed).

This also holds true when assessing eco-morphological traits and habitat conditions outside the Red Sea (Chapter V). Thus, while environmental and spatially wider ranging gradients may be driving genetics and biology (and thus adaptation) at a geological time scale of coral reef fishes, differences in the ecosystems at smaller scales may lead to and shape ecological conditions, and thus also be drivers of ecological and phenotypic plasticity. Species for which such plasticity is very low may therefore only inhabit certain reefs with specific ecological features. Others 276 may adjust their phenotype to their home reef, resulting in different eco-morphs at very small spatial scales, as I found in D. aruanus (Chapter V). Given that a species may exhibit more responsiveness through either phenotypic plasticity or through local adaptation, a question still lingers: could such a difference predict whether the species will be endemic or widespread? I hypothesize that widespread species may be more phenotypically plastic (as shown for D. aruanus in terms of eco- morphological differences associated with habitat settings rather than spatial scale) and that this trait is advantageous for a wide distribution range. While endemic and geographically more restricted species, such as D. marginatus, may rather be mutating faster and/or displaying stronger genetic drift as a consequence of local adaptation to their environment (biologically and genetically) through natural selection, but therefore may be less plastic. For instance, while D. aruanus is not significantly affected by group size due to plasticity in its foraging behavior, D. marginatus feeds less well with increasing group size (Kent et al., 2006). Another example can be the fact that while D. aruanus can change diet composition, eye-size, and/or maybe even timing of feeding in response to predominant visibility conditions at a site (Chapter V; Frédérich et al., 2010) D. marginatus seems to adapt to light and visibility in a more complex manner by changing the composition of light receptors in its retina to enhance vision (Brokovich et al., 2010). Also, along the

Red Sea environmental gradient, with increasing food availability and temperature

(increasing metabolic rates), fish larvae of both species were spending less time in their pelagic stage before settling (Chapter III). However, the decrease was much greater for the endemic species, which may also point towards increased local 277 adaptation and lower plasticity.

Further, evidence for differences in adaptation and phenotypic plasticity between species leading to different geographic ranges can also be found in comparing the genetic structure of D. aruanus and D. marginatus (Chapter III). While D. aruanus shows little genetic differentiation between biogeographic provinces and along environmental and ecological gradients, D. marginatus has a strong genetic break in its Red Sea population at the most drastic change in environmental conditions at the

Farasan Banks, in the southern Red Sea. Even more unexpected were the findings of a particularly strong genetic segregation between D. marginatus inside vs. outside the Red Sea (i.e., the Gulf of Aden). This genetic break was so strong that I expect these to be two separate species. This further suggests higher mutation rates and/or genetic drift putatively due to lower plasticity and natural selection in endemic species. One cannot say that therefore one is more successful than the other. In fact, when assessing the southern Red Sea, D. aruanus is very rare, while D. marginatus is highly abundant. In the northern Red Sea, D. marginatus can also be found all the way up to 40 m depths (probably due to its complex adaptation to light conditions, which may be decisive in the highly oligotrophic waters of the Gulf of Aqaba), while

D. aruanus remains within the first 20 meters depths throughout its distribution range. In the Gulf of Aden (outside the Red Sea) D. aruanus is not present at all. Thus,

I conclude that both are highly successful in their own manner. While the widespread species success relies on its plasticity and putatively high dispersal abilities (with geneflow even across the Gulf of Aden to the next nearest population), which allows it to colonize habitats with different environmental and ecological 278 conditions as well as more geographically distant regions, the key to success of the endemic is to be better locally adapted and more niche specific and thus, highly successful (even outmatching the widespread species, as is the case in the southern

Red Sea and Gulf of Aden) in its ecological niche (of putative origin). Lastly, in this study, there is also evidence for lower dispersal potential of endemic vs. widespread species, as the endemic always showed shorter PLDs in comparison to the widespread D. aruanus. However, this can also be an effect of the degree of local adaption and ecological specialization of the endemic species. Additionally, when comparing one single recruitment cohort to the standing population at a north- central mid-shelf Red Sea reef, the genetic diversity of D. aruanus between them was equally high, even though the species seems to be significantly recruiting with its siblings (Chapter IV). This further supports the existence of a highly successful (high genetic diversity) and well-mixed panmictic (high dispersal potential) Red Sea population of D. aruanus.

In general, I suggest that the research presented above should be continued by comprehensively assessing some of the missing aspects of the biology and ecology of the endemic D. marginatus as follows: (1) assess its eco-morphological traits of along the cross-shelf gradient, which may reveal ecological and environmental features related to, first, the species absence at sites on the shelf-edge and, second, its success and high abundance at the southern Red Sea (no “shelf-edge reefs”), which additionally could explain what may be inducing the genetic break I also found coming into that region. (2) Evaluate the recruitment strategies of D. marginatus in terms of genetic diversity and kinship to compare this to the present 279 results in D. aruanus. This can ideally be performed at the same reef (Shib Al

Karrah) as both species can be found here. The sampling method may need to be modified since I have only one record of a D. marginatus recruit from the extensive sampling I did with the light-traps. (3) Further explore the SNPs data and assess the function of the loci under selection along the different gradients and the populations inside and outside the Red Sea for both species.

Last, but not least, in the course of work for this thesis, I came across typical issues that biogeographers face: what exactly is the distribution range of a species and is it even a single species or not? While I was originally certain D. marginatus was a single species across its distribution in the Arabian Seas, it seems it has diverged into a Red Sea and a Gulf of Aden clade. It is yet to be defined whether these are indeed different species or not. The mitochondrial genes do not show lineage sorting. As the major geographical barrier between those two potential species (the narrow and shallow strait of Bad El Mandeb in the south) is expanding, even if these clades represent two distinct species, increased geneflow by hybridization may lead to the coalescence of the two species in the geological future. Along these lines, I suggest that the diversification of the two clades was a recent event caused by vicariance and genetic drift under sea level fluctuation (and high local adaptation and/or mutation rates) and maintained by the contrasting environmental conditions between the regions (Kawecki & Ebert, 2004) and processes such as introgressive hybridization (Gaither et al., 2014, 2015; Bertrand et al., 2017). The same could have happened in D. aruanus but its ecological plasticity and putatively 280 high dispersal potential may have retarded genetic drift. As soon as the sea levels rose again, populations inside and outside the Red Sea were re-connected, allowing geneflow and homogenization.

6.2 BIODIVERSITY

Peripheral environments were long thought to be evolutionary sinks (Briggs 1999).

However, it is now commonly known that some of them, such as the Red Sea, are not sinks and in fact may harbor some of the highest biodiversity in the world. Reasons for such a high number of species have already been well discussed in (e.g.

(DiBattista et al., 2013, 2015). Nonetheless, I propose a further process, introgressive hybridization, as one of the putative strongest sources of increased biodiversity in the Arabian Seas. I base this idea on the incoherent results between mitochondrial and nuclear markers regarding the genetic divergence among D. marginatus inside and outside the Red Sea (Chapter III). The importance and power of introgressive hybridization to speed up speciation and the maintain new species has more recently and increasingly been acknowledged (e.g. Meier et al., 2017).

6.3 CONCLUSIONS

All in all, I believe that the work of this thesis provides robust methods that can be widely applied in other biogeographic provinces in order to gather comparative and comprehensive data to further investigate drivers of distribution ranges and 281 speciation of coral reef fishes in changing environments. To gain a holistic view, the multidisciplinary tools and comparative analyses could be, for instance, applied on populations of D. aruanus and the respective endemic species in sympatry or parapatry across its distribution range. This could also be expanded to other coral reef fish taxa. Finally, and as genomic studies become more feasible, whole genome analyses can be used to assess which loci are under selection in relation to distribution ranges of species to further reveal adaptive strategies and evolutionary trends among coral reef fishes. Epigenetic work to assess differences in gene expression levels will further contribute to our knowledge on how micro- evolutionary processes and adaptive strategies contribute to speciation and boost comprehensive analytical approaches. Ultimately, compiling cross-disciplinary tools to better understand our most threatened coral reef ecosystems may help us decide how and which habitats to protect a priori. It will also lead to more robust modeling and inference of consequences of climate change to enhance decision-making and management strategies.

Under current climatic changes and anthropogenic impact, I also believe that while the habitable geographical ranges for tropical species may increase towards the poles, the barriers to dispersal will rather be linked to heat tolerances and pollution.

In this context, the Red Sea will be of increased relevance, as it is already at the periphery of coral reef distributions and has extreme environmental conditions.

282

6.4 REFERENCES

Bertrand J.A.M., Borsa P., & Chen W.-J. (2017) Phylogeography of the sergeants Abudefduf sexfasciatus and A. vaigiensis reveals complex introgression patterns between two widespread and sympatric Indo-West Pacific reef fishes. Molecular Ecology. Brokovich E., Ben-Ari T., Kark S., Kiflawi M., Dishon G., Iluz D., & Shashar N. (2010) Functional changes of the visual system of the damselfish Dascyllus marginatus along its bathymetric range. Physiology & behavior, 101, 413–421. DiBattista J.D., Berumen M.L., Gaither M.R., Rocha L.A., Eble J.A., Choat J.H., Craig M.T., Skillings D.J., & Bowen B.W. (2013) After continents divide: comparative phylogeography of reef fishes from the Red Sea and Indian Ocean. Journal of Biogeography, 40, 1170–1181. DiBattista J.D., Choat J.H., Gaither M.R., Hobbs J.-P.A., Lozano-Cortes D.F., Myers R.F., Paulay G., Rocha L.A., Toonen R.J., Westneat M.W., & Berumen M.L. (2015) On the origin of endemic species in the Red Sea. Journal of Biogeography. Frédérich B., Lehanse O., Vandewalle P., & Lepoint G. (2010) Trophic Niche Width, Shift, and Specialization of Dascyllus aruanus in Toliara Lagoon, Madagascar. Copeia, 2010, 218–226. Gaither M.R., Bernal M.A., Coleman R.R., Bowen B.W., Jones S.A., Simison W.B., & Rocha L.A. (2015) Genomic signatures of geographic isolation and natural selection in coral reef fishes. Molecular ecology, 24, 1543–1557. 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. Kawecki T.J. & Ebert D. (2004) Conceptual issues in local adaptation. Ecology Letters, 7, 1225–1241. Kent R., Holzman R., & Genin A. (2006) Preliminary evidence on group-size dependent feeding success in the damselfish Dascyllus marginatus. Marine Ecology Progress Series, 323, 299–303. Meier J.I., Marques D.A., Mwaiko S., Wagner C.E., Excoffier L., & Seehausen O. (2017) Ancient hybridization fuels rapid cichlid fish adaptive radiations. Nature Publishing Group, 8, 1–11. Roberts M.B., Jones G.P., Mccormick M.I., Munday P.L., Neale S., Thorrold S., Robitzch V., & Berumen M.L. (2016) Homogeneity of coral reef communities across 8 degrees of latitude in the. Marine pollution bulletin, 105, 558–565. Spalding M.D., Fox H.E., Allen G.R., Davidson N., Ferdaña Z.A., Finlayson M.A.X., Halpern B.S., Jorge M.A., Lourie S.A., Martin K.D., Mcmanus E., Recchia C.A., 283

Robertson J., Spalding M.D., Fox H.E., Allen G.R., Davidson N., Ferdaña Z.A., & Finlayson M.A.X. (2007) Marine Ecoregions of the World: A Bioregionalization of Coastal and Shelf Areas. BioScience, 57, 573–583.

284

APPENDICES

7.1 LIST OF APPENDICES

7.2 List of Publications

7.3 Productivity and sea surface temperature are correlated with the

pelagic larval duration of damselfishes in the Red Sea (2015)

7.4 Homogeneity of coral reef communities across 8 degrees of latitude in

the Saudi Arabian Red Sea (2015)

7.5 A review of contemporary patterns of endemism for shallow water

reef fauna in the Red Sea (2016)

7.6 Coral bleaching in Saudi Arabia affecting both the Red Sea and Arabian

Gulf

285

7.2 LIST OF PUBLICATIONS

1 Robitzch, V; Lozano-Cortés, D; Kandler, NM; Salas, E; Berumen, ML (2015).

Productivity and sea surface temperature are correlated with the pelagic larval

duration of damselfishes in the Red Sea. Marine Pollution Bulletin. doi:

10.1016/j.marpolbul.2015.11.045

2 Roberts, M; Jones, G; McCormick, M; Munday, P; Neale, S; Thorrold, S, Robitzch,

V; Berumen, M. (2015) Homogeneity of coral reef communities across 8 degrees

of latitude in the Saudi Arabian Red Sea. Marine Pollution Bulletin

doi:10.1016/j.marpolbul.2015.11.024

3 DiBattista, JD; Roberts, MB; Bouwmeester, J; Bowen, BW; Coker, DJ; Lozano‐

Cortés, DF; Choat, J; Gaither, MR; Hobbs, J-P A; Khalil, MT; Kochzuis, M; Myers,

RF; Paulay, G; Robitzch, VSN; Saenz-Agudelo, P; Salas, E; Sinclair-Taylor, TH;

Toonen, RJ; Westneat, MW; Williams, ST; & Berumen, ML (2016). A review of

contemporary patterns of endemism for shallow water reef fauna in the Red

Sea. Journal of Biogeography 43:423-439.

4 Lozano-Cortés, D; Robitzch, V; Abdulkader, K; Kattan, Y; Elyas, A; Berumen, ML

(2016). Coral Bleaching in Saudi Arabia Affecting both the Red Sea and Arabian.

Reef Encounters, April, No. 43. ISSN 0225-27987

5 Robitzch, V; Berumen, ML (in prep.). Annual distribution of recruitment of coral

reef fishes along a cross-shelf gradient in the central Red Sea 286

6 Robitzch, V; Berumen, ML; Alpermann, T; Saenz-Agudelo, P (in prep.). Genetics,

biology, and ecology of the endemic Dascyllus marginatus and the cosmopolitan

D. aruanus damselfishes in vs. outside the Red Sea.

7 Robitzch, V; Saenz-Agudelo, P; Berumen, ML (in prep.). Sibship analysis and

genetic diversity of one single recruitment event of the coral-dwelling fish

Dascyllus aruanus.

8 Robitzch, V; Frédérich, B; Chen, CT (in prep.). Influences of habitat location on

foraging and morphological traits of Dascyllus aruanus.

9 Robitzch, V; Berumen, ML (in prep.). A Yearlong Assessment of the Abundances

of Schindleria spp. along a cross-shelf gradient in the central Red Sea

10 Robitzch, V; Saenz-Agudelo, P; Kaplan, D. (in prep.). An Applied Example of

Mathematical and Computational Basics of Sibship Analysis in Protogynous

Haremic Coral Reef Fish.

MPB-07338; No of Pages 9 Marine Pollution Bulletin xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier.com/locate/marpolbul

Productivity and sea surface temperature are correlated with the pelagic larval duration of damselﬁshes in the Red Sea

Vanessa S.N. Robitzch a,⁎, Diego Lozano-Cortés a,b, Nora M. Kandler a, Eva Salas c,d, Michael L. Berumen a a Red Sea Research Center, Division of Biological and Environmental Science and Engineering, King Abdullah University of Science and Technology, Thuwal, 23955-6900, Saudi Arabia b Coral Reef Ecology Research Group, Department of Biology, Universidad del Valle, Apartado Aéreo 25360, Cali, Colombia c Department of Ecology and Evolutionary Biology, University of California, 100 Shaffer Road, Santa Cruz, CA 95060, USA d Section of Ichthyology, California Academy of Sciences, 55 Music Concourse Drive, San Francisco, CA 94118, USA article info abstract

Article history: We examined the variation of pelagic larval durations (PLDs) among three damselfishes, Dascyllus aruanus, D. Received 2 July 2015 marginatus, and D. trimaculatus, which live under the influence of an environmental gradient in the Red Sea. Received in revised form 5 November 2015 PLDs were significantly correlated with latitude, sea surface temperature (SST), and primary production Accepted 20 November 2015 (CHLA; chlorophyll a concentrations). We find a consistent decrease in PLDs with increasing SST and primary Available online xxxx production (CHLA) towards the southern Red Sea among all species. This trend is likely related to higher food availability and increased metabolic rates in that region. We suggest that food availability is a potentially stronger Keywords: Biogeography driver of variation in PLD than temperature, especially in highly oligotrophic regions. Additionally, variations in Endemic PLDs were particularly high among specimens of D. marginatus, suggesting a stronger response to local environ- Environment mental differences for endemic species. We also report the first average PLD for this species over a broad geo- Food availability graphic range (19.82 ± 2.92 days). PLD © 2015 Elsevier Ltd. All rights reserved. Reef fish

1. Introduction 2006). Similarly, higher food availability can increase larval growth (Biktashev et al., 2003; Fontes et al., 2010; Houde and Hoyt, 1987; The time a fish larva spends in the plankton before recruiting onto a Houde, 1989; Landaeta and Castro, 2006; Sponaugle, 2009, 2010) and reef, known as its pelagic larval duration (PLD), is a valuable source of potentially decrease the time a fish larva spends in the pelagic before information on how ecological and oceanographic processes affect a recruiting onto a coral reef (Sponaugle and Grorud-Colvert, 2006). Var- fish's early life. Such information can help to better understand organ- iations in PLDs can be indicative of differences in dispersal potential, isms with complex life cycles, such as many coral reef inhabitants; sub- growth rates, and survival of fish larvae. Furthermore, the conditions a sequently, this fundamental life history information can be used to fish larva experiences during its PLD seem to even influence success, fit- improve management of marine ecosystems and their natural re- ness, and growth in later life stages as juvenile and adult fish (Cushing sources. PLDs of coral reef fishes are known to exhibit variation driven and Horwood, 1994; Houde and Hoyt, 1987; Houde, 1989; Rankin and by a wide range of intrinsic and extrinsic factors (McCormick and Sponaugle, 2011; Sponaugle and Grorud-Colvert, 2006; Sponaugle et Molony, 1995; Searcy and Sponaugle, 2000; Sponaugle, 2010; al., 2011). However, most studies that have assessed the influence of Wellington and Victor, 1992; Wilson and Meekan, 2002). Two of the temperature and food availability on the larval stage of fishes have fo- most relevant extrinsic factors that cause intraspecific variations in lar- cused on larval growth rate (Mcleod et al., 2015), size at settlement val growth and development are temperature and food availability (McCormick and Molony, 1995), or recruitment cohort size (Lo-Yat et (Heath, 1992; Houde and Zastrow, 1993; Mcleod et al., 2011; al., 2011) rather than on the length of PLDs in the standing population, Takahashi and Watanabe, 2005), which can be estimated from remote and still fewer of these studies target coral reef fishes. In our study we sensing satellite data on sea surface temperature (SST) and chlorophyll focus specifically on PLD variations within the standing populations a (CHLA) concentrations respectively. Warmer temperatures increase (i.e., post-settlement and adult fishes, as opposed to recruitment co- metabolic rate and growth rates (Green and Fisher, 2004; Meekan et horts) of three congeneric coral reef damselfishes present in the Red al., 2003; Sponaugle et al., 2006), which can lead to shorter PLDs Sea. (Bergenius et al., 2005; Green and Fisher, 2004; Grorud-Colvert and The Red Sea is an ideal location to study the effect of a variety of Sponaugle, 2011; McCormick and Molony, 1995; Sponaugle et al., environmental parameters on coral reefs (Berumen et al., 2013). The Red Sea harbors thriving and continuous coral reefs along both sides ⁎ Corresponding author. of its narrow basin. Due to its shape, location, and only a single shallow E-mail address: [email protected] (V.S.N. Robitzch). and narrow connection to the Indian Ocean, it displays a unique

Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 2 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx environmental gradient of temperature, salinity, productivity, turbidity, measurements from two sites outside the Red Sea as well as 4) provide and essentially reef-scape (Racault et al., 2015; Raitsos et al., 2013). the first PLD-measurements the first PLDs for D. marginatus from a wide These physical and environmental characteristics have made the Red geographic range. Sea a natural location to assess the environmental impact on several coral reef organisms (e.g., Froukh and Kochzius, 2007; Giles et al., 2. Materials and methods 2015; Lozano-Cortés and Berumen, 2015; Nanninga et al., 2014; Ngugi et al., 2012; Roberts et al., 1992; Robitzch et al., 2015; Sawall et al., 2.1. Fish sampling 2014a, 2014b). Here, we focus on the two seemingly major parameters that may Clove oil, tweezers, hand nets, and spears were used to collect juve- drive PLD variations in the environmental gradient of the Red Sea: tem- nile and adult specimen from the three Dascyllus species (D. aruanus, D. perature and food availability. For this purpose we use the three marginatus, and D. trimaculatus) present in the Red Sea (RS) (Fig. 1). Dascyllus species that live in the Red Sea (D. aruanus, D. marginatus, Fishes of different sizes (Fig. 2) and/or from different sampling years and D. trimaculatus). These species serve as a good model group to (Table 1) were collected for PLD assessment to cover potential temporal study PLD variations in coral reef fishes for several reasons. They are bi- variations as much as possible. These were caught at 18 reef sites off the ologically similar, small in size, and commonly present along the entire Saudi Arabian coastline and at two sites outside the RS (Indian Ocean environmental gradient of the Red Sea at overlapping depth ranges. “IO” and Indo-Australian-Archipelago “IA”; Table 1 and Fig. 1). Samples They are zooplanktivorous, demersal spawners, and have relatively sim- were immediately preserved in 90% ethanol. The sampling range ilar recruitment periods and larval development times (PLD). However, (N1600 km of coastline) covers most of the latitudinal span of the RS. they have subtly different ecological preferences, which make them an Whenever possible, specimens from all three species were collected interesting group to study the effect of the environment on early life his- from the same reef site for coherent interspecific comparisons. In case of tory traits in the context of ecological specialization. We consider D. absence of any of the species, specimens from the nearest possible site trimaculatus as the most ecologically versatile species within this were sampled. The sampling sites were divided into four major Red group. It can be found on exposed reef walls, sheltered backside-reefs, Sea (RS) regions: “GAQ” (Gulf of Aqaba), “NCRS” (north-central RS), and lagoons; it is not dependent on live coral; and it only uses reef “SCRS” (south-central RS); “SRS” (southern RS) (in consensus with structure for shelter, recruitment, and as spawning habitat. While D. Raitsos et al., 2013; Fig. 1 and Table 1). trimaculatus mostly recruits onto anemones in the Red Sea, it can also recruit into branching corals, sea urchins, and other microhabitats. 2.2. Otolith preparation and PLD assessment Once they reach juvenile stages, D. trimaculatus leave their settlement substrate, range throughout the entire reef structure, and spend most PLDs for all three species were estimated from daily increments up of their lifetime foraging in the water column. Intermediate on the eco- to the settlement marks of sagittal otoliths. Sagittae were extracted logical specialization ranking is D. aruanus. This species shows stronger and mounted onto a microscopy glass slide with thermoplastic resin, habitat preferences, only recruiting to and living its entire life in then ground and polished (following Wilson and McCormick, 1997) branching live corals that are located in sheltered sandy reefs and using aluminum oxide lapping films (South Bay Technology, Inc.) of lagoons. Dascyllus aruanus and our third species, D. marginatus, both three different thicknesses (30, 12, and 5 μm). Three to ten pictures show similar habitat preferences and occasionally even co-inhabit the were taken of each polished otolith with the program AxioVision Rel. same coral colony. However, the degree of ecological specialization is (V. 4.8.2.0; copyright 2006–2010 Carl Zeiss Micro Imaging GmbH) uti- stronger in D. marginatus, which cannot be found on very shallow lizing a Zeiss AXIO Scope A1 microscope (200× magnification) and a reef-tops and lagoons, nor can it be found on offshore reefs in most of Zeiss AxioCam ICc1. From the pictures, three independent PLD readings the Red Sea. Therefore, we refer to D. marginatus as the most ecological- were made and if the counts deviated by less than 10% the mean count ly specialized species in our group. for each individual was calculated and included in the analyses. The The Dascyllus model also seems suitable to study variation in PLDs in same reader calculated PLDs for all species and lacked any information the context of biogeography of coral reef associated fishes. Of the three or metadata about the origin of the sample. A total of 28 D. trimaculatus, species that occur in the Red Sea, D. marginatus is an Arabian endemic 56 D. marginatus,and92D. aruanus otolith readings were used for the while D. trimaculatus and D. aruanus are both widespread species, com- statistical analyses (Table 1). mon in most of the Indo-Pacific. In general, pomacentrid species with restricted distributions have shorter PLDs than wide-ranging species 2.3. Chlorophyll a (CHLA) and sea surface temperature (SST) correlations (Cowen and Sponaugle, 1997; Wellington and Victor, 1989), although the opposite is observed for labrids (Cowen and Sponaugle, 1997; Two environmental parameters of the Red Sea were tested for corre- Victor, 1986). The expected trend is that endemic or geographically- lations to PLD data: Chlorophyll a (CHLA) and sea surface temperature constrained species have shorter PLDs potentially restricting dispersal (SST). CHLA content is a good proxy for phytoplankton biomass (Lester and Ruttenberg, 2005; Macpherson and Raventos, 2006; Mora (Håkanson et al., 2007) and is herein used as a proxy of food availability et al., 2003). However, in endemic pomacentrid species the PLDs seem for the pelagic larva of the three studied species. Validated regional av- to have a rather non-linear relationship with range size (Thresher et erages of CHLA concentrations (mg m−3) and SST (°C) were provided al., 1989). Similarly, most studies assessing correlations between PLDs by D. Raitsos, based on Raitsos et al. (2013) for the regions NRS, NCRS, and biogeographic ranges over a number of species find no link SCRS, and SRS; and by D. Dreano for the Gulf of Aqaba (GAQ). These between the two (Macpherson and Raventos, 2006; Victor and estimates were produced from a 10-year high-resolution data set of Wellington, 2000; Zapata and Herrón, 2002). The exact nature of PLD satellite remote-sensing CHLA estimates (see Raitsos et al. (2013) for a vs. biogeography remains an open question. detailed description of data acquisition). These estimates were subse- Thus, the objectives of our study are to use the model species-group quently used for Pearson correlations with PLD for each species. of D. marginatus, D. aruanus, and D. trimaculatus to 1) assess the influence of environmental parameters such as SST and CHLA on the PLDs 2.4. Statistical analyses of coral reef associated damselfishes with different degrees of ecological specialization in a natural environment inside the Red Sea; 2) investi- All data were tested for normality (Kolmogorov–Smirnov test) and gate variations in PLDs related to differences in biogeographic ranges homoscedasticity (Levene's test) prior to analyses. The differences in among closely related species; 3) provide the first PLD-measurements PLD among the three species in the four Red Sea regions (GAQ, NCRS, for D. trimaculatus and D. aruanus from the Red Sea and additional SCRS, and SRS) were explored with a factorial ANOVA. The spatial

Fig. 1. Sampling sites of Dascyllus marginatus, D. aruanus,andD. trimaculatus. A total of 20 sites are grouped into four regions within the Red Sea (RS) and two outside the Red Sea (IA: Indo- Australian Archipelago; and IO: Indian Ocean). Those inside the Red Sea are represented by white circles (GAQ: Gulf of Aqaba), squares (NCRS: north-central RS), diamonds (SCRS: south- central RS), or sails (SRS: southern RS) according to the respective region. The regions were assigned in consensus with those in Raitsos et al. (2013). The two locations outside the Red Sea are represented by a white nabla and a white triangle (IO: Indian Ocean and IA: Indo-Australian-Archipelago, respectively). Map axes indicate latitude (°N) and longitude (°E). Additional information on exact number of samples and coordinates for each site can be found in Table 1. The chlorophyll a (CHLA, in mg m−3) concentrations of the Indo-Paciﬁc and Red Sea are displayed from the NASA Giovanni website (http://oceancolor.gsfc.nasa.gov) to visualize approximate differences between locations.

Table 1 Sampling sites, including number of samples, reef names and coordinates.

Location Region Site N (Dt) N (Dm) N (Da) Latitude Longitude Reef Name

The three main geographic locations are the Red Sea (RS), the Indo-Australian-Archipelago (IA), and the Indian Ocean (IO). The RS location is subdivided in four major regions (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS). The sites represent the reefs sampled for each region (three letter codes based on the reef names). The IA and IO locations are represented by one sampling site each: Mandando in North Sulawesi, Indonesia (IND) and the island of Cocos Keeling, Australia (COC) respectively. N represents the number of individuals for which pelagic larval durations (PLDs) were assessed for statistical analyses of each studied species: D. trimaculatus (Dt), D. marginatus (Dm), and D. aruanus (Da). Samples from the RS were collected in 2013, 2014, and 2015; from the IO in 2010 and 2014; and from the IA in 2003.

variation of the PLD within each species was tested using an indepen- D. trimaculatus (20–27 d; 7 d difference). Within sampling sites, individ- dent one-way ANOVA for each of the three Dascyllus species. A post- uals of the endemic species also showed the highest PLD range (19–28 hoc test applying the unequal N Tukey's HSD means comparisons was d, 9 d difference in HAQ) compared to those of the two widespread used to detect any signiﬁcant differences within the two aforemen- species (D. aruanus:6ddifferenceinMAN;D. trimaculatus: 4 d differ- tioned analyses. Subsequently, the PLD data was correlated (Pearson ence in AFA). correlation) with latitude, SST, and CHLA (i.e., food availability) measurements to assess the impact of the latitudinal environmental gradi- 3.1.2. Inside vs. outside the Red Sea ent of the Red Sea. Finally, the PLDs from the Red Sea sites were We report the PLD measurements for D. aruanus and D. trimaculatus compared against those from locations outside the Red Sea for the two from sites outside the Red Sea (RS) (D. aruanus: 21.05 ± 1.10 in the IA; widespread species (D. trimaculatus and D. aruanus). For this analysis, D. trimaculatus: 24.40 ± 1.25 in the IO) and compare these to our mea- data from the four Red Sea regions for each of the two species was surements from inside the Red Sea. The mean PLD of both widespread merged and compared independently with the IO (D. trimaculatus) and the IA (D. aruanus) using a t-student test. All analyses were conducted with STATISTICA 8.0 (StatSoft, Inc. 2007).

3. Results

3.1. PLD measurements

(F(3144) = 31.723, p b 0.001, Fig. 5) and sites (only assessed for D. marginatus (F(7,44) = 10.099, p b 0.001) and D. aruanus (F(11,76) = 7.6954, p b 0.001 due to low number of D. trimaculatus specimen per site; see Fig. 5). The interaction between species and regions was not signiﬁcant. In addition, a consistent latitudinal decrease in PLD was found to be present from north to south among all of the species between the four regions (see Fig. 4). PLD ranges were the largest for the Fig. 3. Mean pelagic larval duration (PLD, in days) for Dascyllus marginatus, D. aruanus,and – endemic species D. marginatus (15 28 d; 13 d difference) and almost D. trimaculatus averaged among all study sites in the Red Sea. Whiskers represent 95% double of the ranges found in D. aruanus (19–27 d; 8 d difference) and conﬁdence intervals around the mean (black circle).

Table 2 Average pelagic larval durations (PLD in days; mean ± standard deviation) of Dascyllus marginatus, Dascyllus aruanus,andDascyllus trimaculatus based on replicate counts of daily growth increments from otoliths.

Locat. Region Site PLD Dm PLD Da PLD Dt Lat. Mean SST [°C] Mean CHLA [mg m−3]

Average PLDs for each species (D. marginatus (Dm), D. aruanus (Da), and D. trimaculatus (Dt)) are given among all (All) and within each geographic location: of the Indian Ocean (IO), the Indo-Australian-Archipelago (IA), and the Red Sea (RS); of each region within the RS (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS); and of each site within the RS (three-letter coded reef name). The latitude (Lat., in decimal degrees) is given for each site as used for correlation analysis. Estimates of mean surface sea temperature (mean SST, in °C) and chlorophyll a content (CHLA, in mg m−3) are provided by D. Raitsos and D. Dreano from 10-years satellite remote sensing high-resolution data (in consensus with Raitsos et al., 2013).

species is longer outside the Red Sea than within the Red Sea but only 3.2. Environmental correlations with PLD in the Red Sea significantly longer for D. aruanus (D. aruanus:RSvs.IA:t-value= 2.88, df = 86, p = 0.04; D. trimaculatus:RSvs.IO:t-value=0.07, Regional means of SST and CHLA concentrations were used as envi- df = 26, p = 0.93; see Fig. 5). ronmental data for Pearson correlations (Table 2). Their respective mean annual profiles are visualized in Fig. 6. The profiles of each region display a similar trend-curve except for SRS and GAQ. The SRS has

Fig. 4. Mean pelagic larval durations (PLD, in days) of three Dascyllus species in the four regions of the Red Sea (RS) (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-cen- Fig. 5. Mean pelagic larval duration (PLD, in days) of Dascyllus aruanus and D. trimaculatus tral RS, and SRS: south RS). Average PLDs are given by blue triangles for D. marginatus,red from locations inside the Red Sea (RS) and outside the Red Sea (Indo-Australian Archipel- squares for D. aruanus, and green diamonds for D. trimaculatus. Whiskers represent 95% ago (IA) and Indian Ocean (IO)). Whiskers represent 95% confidence intervals and boxes confidence intervals around the mean. (For interpretation of the references to color in represent one standard error around the mean. Mean PLDs are given by black squares this figure legend, the reader is referred to the web version of this article.) for D. aruanus and diamonds for D. trimaculatus.

Fig. 6. Validated monthly regional averages of chlorophyll a concentrations (CHLA, in mg m−3, left panel) and sea surface temperature (SST, in °C, right panel). Each panel shows monthly mean averages for an average calendar year (three-letter coded months, x-axis) for each region (GAQ: Gulf of Aqaba, NCRS: north-central RS, SCRS: south-central RS, and SRS: south RS). Averages were produced from 10-year high-resolution data of satellite remote-sensing CHLA and SST estimates provided by D. Raitsos (for the regions NRS, NCRS, SCRS and SRS; sensu Raitsos et al. (2013)) and D. Dreano for the Gulf of Aqaba (GAQ). The trend lines are color-coded for each region. overall much higher CHLA concentrations and its average annual CHLA gradual and consistent decrease in mean PLD with decreasing latitude profile has an additional peak in the summer. Different from all other re- along the environmental gradient. Moreover, intra-specificdifferences gional CHLA profiles, the GAQ displays a sharp peak later in the spring in PLD between sampling regions along this gradient were higher in D. and has both the highest concentrations of CHLA over the entire Red marginatus than the other two widespread species (D. aruanus and D. Sea during the spring and the lowest during the summer (Fig. 6). trimaculatus), suggesting that the endemic species might be more re- These unique seasonal changes of CHLA in GAQ result in one of the larg- sponsive to environmental differences within its habitat potentially re- est regional ranges (N0.6 mg m−3) in CHLA concentrations in the Red lated to a higher degree of specialization. We provide the first PLD Sea (besides the southernmost SRS region; Table 2 and Fig. 6,left estimate for a broad geographic range of the endemic D. marginatus panel). In contrast, the SST profiles for all regions have similar ranges (19.8 ± 2.9 d), which displays a consistently lower PLD at all sites in of about 6 °C (Fig. 6, right panel). comparison to PLDs of D. aruanus and D. trimaculatus. Our results suggest that environmental factors can act as drivers of biogeography (as 3.2.1. Correlations with/atitude well as endemism) and dispersal among coral reef fishes. We also find Pearson correlation tests show a positive highly significant (p b 0.001) the PLDs within species to be overall spatially and potentially also tem- correlation between latitude and PLD among all species (Fig. 7). Latitude porally highly plastic, which should be considered when evaluating coris also significantly negatively correlated with both CHLA and SST, indicat- relations of PLDs and other parameters. ing the latitudinal character of the Red Sea environmental gradient (p b 0.001). Even though latitude shows the highest correlations 4.1. Pelagic larval durations (PLDs) in the chlorophyll a (CHLA) and sea with PLDs (except among D. trimaculatus,R2 = 46.2%, Table 2), lati- surface temperature (SST) gradient of the Red Sea tude itself is most likely not the driving factor but the correlation is rather a result of the combined effects of the environmental param- The Red Sea is a good living laboratory for examining the effect of eter at each site (e.g. SST and CHLA). strong environmental gradients on various aspects of community ecology (Roberts et al., 1992; Ngugi et al., 2012; Sawall et al., 2014a, 2014b) 3.2.2. Correlations with SST and genetic connectivity (Froukh and Kochzius, 2007; Nanninga et al., Temperature (SST) is significantly negatively correlated with PLDs of 2014; Giles et al., 2015; Reimer et al., in review; Robitzch et al., 2015). all three species (longer PLDs at lower SST) and shows the second However, very few studies have examined variation in life history traits highest R2 values (except among D. trimaculatus,R2 =30.1%,Table 2). or demographics (e.g., growth rates or PLD). In our study, we specifically This correlation is negative and significant for all species (Fig. 7). look at the influence of SST and CHLA levels on early life stages of three Dascyllus species and find both to be consistently negatively correlated 3.2.3. Correlations with CHLA to PLDs of the studied fishes. A reduction in PLDs can be the result of CHLA concentration is significantly (p b 0.001) negatively correlated an increase in the metabolic rate of larvae in warmer waters to the PLDs of all three Dascyllus species. The correlation is the strongest (Atkinson et al., 1996; Bjørnsson et al., 2001; Green and Fisher, 2004; for D. trimaculatus (R2 = 50.1%; Table 2, Fig. 7). However, the sample Hunt et al., 1996) or of higher availability of food (Sponaugle and size for this species is low and limits the overall interpretation of this re- Grorud-Colvert, 2006) or of a combination of the two. In this context, sult. Moreover, the R2 values of SST and CHLA do not differ much and CHLA explained up to 50% of the variation in PLD data (e.g., of D. make it difficult to interpret which of the two parameters has the stron- trimaculatus), suggesting it is a strong environmental driver of PLDs in ger influence on PLDs. Dascyllus. Nevertheless, the coefficients of correlation (R2 values) for both SST and CHLA are quite similar, which makes it difficult to infer 4. Discussion and conclusions which parameter has a stronger effect on PLDs, and a higher metabolic rate in warmer waters also demands more energy and higher food con- Our study shows that there is a significant link between the pelagic sumption (Houde and Zastrow, 1993; Houde, 1989). Since both SST and larval duration (PLD) of Dascyllus species and environmental parame- CHLA co-vary with latitude in the Red Sea, it is not surprising that the ters such as mean sea surface temperature (SST) and chlorophyll a strongest significant correlation was a between PLD and latitude. How- (CHLA) content in the Red Sea. In all three study species, we found a ever, latitude itself is unlikely to be influencing the PLD of coral reef

Fig. 7. Plots of Pearson correlations of pelagic larval durations (PLD, in days, y-axis) of three Dascyllus species (Dascyllus marginatus, D. aruanus, D. trimaculatus, from left to right) to: A–C) regional means of sea surface temperature (SST in °C, ﬁrst row), D–F) regional means of chlorophyll a concentrations (CHLA, in mg m−3, middle row), and G–I) latitude (in degrees, last row; values in Table 2). The plots are arranged in columns according to species (i.e., D. marginatus shown in panels A, D, G; D. aruanus in panels B, E, H; and D. trimaculatus in panels C, F, I) and regional means of: SST, CHLA, and latitude. The blue circles represent the measured PLD values of each species (y-axis, in days). The regression line is in red and the dotted curves represent the respective 95% conﬁdence intervals of the correlation tests.

fishes (Booth and Parkinson, 2011). We interpret our findings as the all other regions are rather homogeneous In other words, larvae that synergic effect of greater food availability (higher CHLA concentrations) hatch during the spring/early summer in GAQ have up to six times and higher metabolism (higher SST) on larval growth rates, subse- more food available compared to cohorts hatching during other seasons. quently reducing PLDs for all species in the warmer and more produc- The extreme variability in the food available to these cohorts could be tive southern regions unique to the Red Sea. Furthermore, the responsible for the large range of PLDs within the GAQ region. We decrease of PLDs along the SST and CHLA gradient seems to be stronger therefore hypothesize that if the range of PLD differences within sites and more significant among individuals of the endemic and ecologically is a result of seasonal changes in the pelagic environment, seasonal more specialized D. marginatus in comparison to D. aruanus and D. changes in CHLA (food availability) rather than SST could more likely trimaculatus. We thus propose that the influence of small-scale regional be the driver of differences in PLDs of Dascyllus in the Red Sea. However, changes in the environment might be stronger and positively correlated the proposed hypothesis warrants further investigation in view of the with the degree of ecological specialization and habitat range of correlation between temperature and food availability (CHLA) damselfishes. (McCormick, 1994; Lo-Yat et al., 2011) and the implementation of sta- This is best illustrated in our study by the observation of large differ- tistical models to discretely analyze the influence of these two environ- ences in PLDs among the endemic species, D. marginatus at one specific mental parameters on coral reef fishes. In terms of biological data, site in the Gulf of Aqaba (GAQ) (a 9 d range of difference in HAQ). These recruit collections could be targeted at different times of the year to as- differences in PLDs were the highest among all Dascyllus in the Red Sea. sess whether longer PLDs correlate with seasons of lower productivity. If we then focus on the regional seasonal changes of the two environ- For the more “homogeneously” productive regions of the Red Sea one mental parameters SST and CHLA, the range of seasonal changes of could assess if the differences in PLDs are also more homogeneously dis- CHLA concentrations for the GAQ region is distinctly higher compared tributed throughout the year. Unfortunately, we were unable to reliably to adjacent regions to the south, further suggesting that the GAQ is a back-calculate birthdates for our individual samples to determine spe- variable environment itself. In contrast, the seasonal changes in SST in cific intra-annual patterns. Alternate hypotheses that could explain

Please cite this article as: Robitzch, V.S.N., et al., Productivity and sea surface temperature are correlated with the pelagic larval duration of damselfishes in the Red Sea, Marine Pollution Bulletin (2015), http://dx.doi.org/10.1016/j.marpolbul.2015.11.045 8 V.S.N. Robitzch et al. / Marine Pollution Bulletin xxx (2015) xxx–xxx the large within-region differences in PLDs include seasonal changes in The genus Dascyllus may offer an opportunity to address such questions. survival strategies and/or competition during the larval stage. For exam- The genus has an endemic representative almost in every biogeographic ple, there may be a benefit in early settlement if there is seasonal vari- province in the Indo-Pacific, which will allow future comparative ability in settlement habitat availability and early settlement increases studies using this genus to further assess the relationship between the opportunity to secure space on the reef. Alternatively, delayed re- PLD and biogeography over a wide geographic range. Other genera cruitment may confer a benefit due to larger size-at-settlement if (e.g., Chaetodon, Thalassoma, Eviota, etc.) may provide similar opportu- post-settlement competition varies seasonally. During seasons that nities for congeneric comparison of endemic vs. widespread species. may have less settlement competition, delaying settlement may confer We further advocate the importance of analyzing PLD data accounting advantages in the form of decreased post-settlement mortality (bigger- for sources of temporal and spatial variation at small scales. is-best hypothesis, Anderson, 1988; Miller et al., 1988). Much work remains to be done to resolve the many potential drivers of plasticity in the PLDs of larval reef fishes at both ecological and evolu- 4.2. Pelagic larval durations (PLDs) of Dascyllyus and the biogeography of tionary timescales. The role of various environmental factors and the in- coral reef fishes clusion of genetic, biological, and ecological data are a prime target for future studies, particularly in connection with understanding how It is thought that spatial-temporal variation in the environment can ecological specialization of a species interacts with these factors. select for or against dispersal (Duputié and Massol, 2013), ultimately shaping specialization, evolutionary processes, and the biogeography Acknowledgments of species (Berdahl et al., 2015; Heinz et al., 2009). In theory, endemic species should have reduced dispersal potentially linked to a shorter We would like to thank fieldwork and logistic support provided by PLD than widespread species, and biogeographic provinces with steep KAUST Coastal and Marine Resources Core Lab. For assistance acquiring environmental gradients may thus have higher rates of endemism samples, we thank A. Gusti, M. Roberts, R. Gatins, A. Kattan, G. Bernardi, than more homogeneous environments. M. Erdmann, and J.-P. Hobbs. For support with environmental data, we The Red Sea has long been recognized as a hotspot of endemism thank D. Raitsos and D. Dreano. We also thank two anonymous referees (Roberts et al., 1992; DiBattista et al., 2015a, 2015b) and in its sharp envi- and S. Agrawal for comments and feedback, as well as J. S. Berdahl for ronmental gradient (Raitsos et al., 2013)theendemicdamselfish species formatting the artwork. Financial support came from KAUST baseline D. marginatus consistently exhibits shorter PLDs than the two widespread research funds and an award from the KAUST Office of Competitive species. Of the two more ecologically versatile species, D. trimaculatus Research Funds (award no. URF/1/1389-01-01) to M.LB. shows the longest PLD and the global average for this species is even longer (Luiz et al., 2013, Dataset S1). From a biogeographical point of view, this might indicate a successive increase of average PLD with spatial dis- References tribution among the three Dascyllus species in our study. The relationship between dispersal potential and biogeographic dis- Anderson, J.T., 1988. A review of size dependent survival during pre-recruit stages of fishes in relation to recruitment. J. Northwest Atl. Fish. Sci. 8, 55–66 (doi:10.1.1.467.4638). tributionofmarinespecieshaslongbeenanopendiscussionandmany Atkinson, D., Johnston, I.A., Bennett, A.F., 1996. Ectotherm life-history responses to devel- studies have approached this question using PLD data. However, most of opmental temperature. In: Johnston, I.A., Bennett, A.F. (Eds.), Anim. Temp. Phenotypic these studies have found either weak, non-linear, or no correlation be- Evol. Adapt. Cambridge University Press, New York, pp. 183–204 http://dx.doi.org/10. 1017/CBO9780511721854.009. tween distribution ranges and PLDs at different geographic scales and Berdahl, A., Torney, C.J., Schertzer, E., Levin, S.A., 2015. On the evolutionary interplay be- for a wide number of species. There can be several reasons for conflict- tween dispersal and local adaptation in heterogeneous environments. Evolution 69 ing results among studies on the relationship between dispersal poten- (6), 1390–1405. http://dx.doi.org/10.1111/evo.12664. fi Bergenius, M.A.J., McCormick, M.I., Meekan, M.G., Robertson, D.R., 2005. Environmental tial and biogeography of marine species. From our data we nd it influences on larval duration, growth and magnitude of settlement of a coral reef important to point out the extreme plasticity observed in PLDs of fish. Mar. Biol. 147 (2), 291–300. http://dx.doi.org/10.1007/s00227-005-1575-z. Dascyllus in relation to the environment and how such variations can Berumen, M.L., Hoey, A.S., Bass, W.H., Bouwmeester, J., Catania, D., Cochran, J.E.M., Khalil, radically change the outcome of the study. For instance, in our study, M.T., Miyake, S., Mughal, M.R., Spaet, J.L.Y., Saenz-Agudelo, P., 2013. The status of coral reef ecology research in the Red Sea. Coral Reefs 32 (3), 737–748. http://dx.doi.org/ D. marginatus consistently has a shorter PLD than D. aruanus at the 10.1007/s00338-013-1055-8. same geographic site (i.e., spatially coherent data) and we thus find a Biktashev, V.N., Brindley, J., Horwood, J.W., 2003. Phytoplankton blooms and fish recruit- – positive correlation between distribution range and dispersal potential. ment rate: effects of spatial distribution. J. Plankton Res. 25 (1), 21 33. http://dx.doi. org/10.1016/j.bulm.2003.08.008. However, if our reference PLDs for the two species within the Red Sea Bjørnsson, B., Steinarsson, A., Oddgeirsson, M., 2001. Optimal temperature for growth and were not spatially coherent but we instead use a reference PLD for D. feed conversion of immature cod (Gadus morhua L.). ICES J. Mar. Sci. 58 (1), 29–38. marginatus from the northern Red Sea (GAQ; the only measurement http://dx.doi.org/10.1006/jmsc.2000.0986. Booth, D.J., Parkinson, K., 2011. Pelagic larval duration is similar across 23° of latitude for available in previous literature) and a reference PLD for D. aruanus two species of butterflyfish (Chaetodontidae) in eastern Australia. Coral Reefs 30, from the southern Red Sea (SRS), we would conclude the opposite, 1071–1075. http://dx.doi.org/10.1007/s00338-011-0815-6. that is, we would find that the endemic species has a longer PLD than Cowen, R.K., Sponaugle, S., 1997. Relationships between early life history traits and recruitment among coral reef fishes. In: Chambers, C.R., Trippel, E.A. (Eds.), Early Life the widespread species. Broad comparisons of PLDs to range sizes may History and Recruitment in Fish Populations. Chapman and Hall, pp. 423–450 therefore be complicated by large intraspecific variations in PLD http://dx.doi.org/10.1007/978-94-009-1439-1_15. among sampling locations (see also Victor, 1991). Cushing, D.H., Horwood, J.W., 1994. The growth and death of fish larvae. J. Plankton Res. 16 (3), 291–300. http://dx.doi.org/10.1093/plankt/16.3.291. We suggest that appropriately designed studies may still have the DiBattista,J.D.,HowardChoat,J.,Gaither,M.R.,Hobbs,J.P.A.,Lozano-Cortés,D.F.,Myers,R.F., potential to provide insight into the unresolved relationship of PLD Paulay, G., Rocha, L.A., Toonen, R.J., Westneat, M., Berumen, M.L., 2015b. On the origin of and biogeography. The cautious collection of adequate temporal and endemic species in the Red Sea. J. Biogeogr. http://dx.doi.org/10.1111/jbi.12631. spatial data is crucial. A study focused on long-term/evolutionary pro- DiBattista, J.D., Roberts, M., Bouwmeester, J., Bowen, B.W., Coker, D., Lozano-Cortés, D.F., Choat, J.H., Gaither, M.R., Hobbs, J.P., Khalil, M.T., Kochzius, M., Myers, R.F., Paulay, cesses like biogeography in relation to contemporary demographic G., Robitzch, V., Rocha, L.A., Saenz-Agudelo, P., Salas, E., Sinclair-Taylor, T.H., Toonen, data like PLDs, would ideally meet the following criteria: A) address R.J., Westneat, M., Williams, S.T., Berumen, M.L., 2015a. A review of contemporary closely related species to eliminate variation related to phylogenetic ef- patterns of endemism for shallow water reef fauna in the Red Sea. J. Biogeogr. http://dx.doi.org/10.1111/jbi.12649. fects, B) sample from standing (adult) populations to capture as much Duputié, A., Massol, F., 2013. An empiricist's guide to theoretical predictions on the evo- temporal variation as possible, also to ensure that the data represent lution of dispersal. Interface Focus 3, 20130028. http://dx.doi.org/10.1098/rsfs.2013. the ‘effective’ population's PLD (from adults from different recruitment 0028. Fontes, J., Afonso, P., Santos, R.S., Caselle, J.E., 2010. Temporal variability of larval growth, cohorts), and C) sample with spatial coherency in order to isolate ex- size, stage duration and recruitment of a wrasse, Coris julis (Pisces: labridae), from the trinsic environmental variables from intrinsic biological differences. Azores. Sci. Mar. 74 (4), 721–729. http://dx.doi.org/10.3989/scimar.2010.74n4721.

Froukh, T.J., Kochzius, M., 2007. Genetic population structure of the endemic fourline Raitsos, D.E., Pradhan, Y., Brewin, R.J.W., Stenchikov, G., Hoteit, I., 2013. Remote sensing wrasse (Larabicus quadrilineatus) suggests limited larval dispersal distances in the the phytoplankton seasonal succession of the Red Sea. PLoS One 8 (6), e64909. Red Sea. Mol. Ecol. 16 (7), 1359–1367. http://dx.doi.org/10.1111/j.1365-294X.2007. http://dx.doi.org/10.1371/journal.pone.0064909. 03236.x. Rankin, T.L., Sponaugle, S., 2011. Temperature influences selective mortality during the Giles, E.C., Saenz-Agudelo, P., Hussey, N.E., Ravasi, T., Berumen, M.L., 2015. Exploring sea- early life stages of a coral reef fish. PLoS One 6 (5), e16814. http://dx.doi.org/10. scape genetics and kinship in the reef sponge Stylissa carteri in the Red Sea. Ecol. Evol. 1371/journal.pone.0016814. 5(13),2487–2502. http://dx.doi.org/10.1002/ece3.1511. Reimer, J., Herrera, M., Gatins, R., Roberts, M., Berumen, M.L., 2015. Latitudinal variation in Green, B.S., Fisher, R., 2004. Temperature influences swimming speed, growth and larval the symbiotic dinoflagellate Symbiodinium of the common reef zoantharian Palythoa duration in coral reef fish larvae. J. Exp. Mar. Biol. Ecol. 299 (1), 115–132. http://dx. tuberculosa on the Saudi Arabian coast of the Red Sea. J. Biogeogr. (in review). doi.org/10.1016/j.jembe.2003.09.001. Roberts, C., Shepherd, A., Ormond, R., 1992. Large-scale variation in assemblage structure Grorud-Colvert, K., Sponaugle, S., 2011. Variability in water temperature affects trait- of Red Sea butterflyfishes and angelfishes. J. Biogeogr. 19 (3), 239–250. http://dx.doi. mediated survival of a newly settled coral reef fish. Oecologia 165 (3), 675–686. org/10.2307/2845449. http://dx.doi.org/10.1007/s00442-010-1748-4. Robitzch, V., Banguera-Hinestroza, E., Sawall, Y., Al-Sofyani, A., Voolstra, C.R., 2015. Ab- Håkanson, L., Bryhn, A.C., Hytteborn, J.K., 2007. On the issue of limiting nutrient and presence of genetic differentiation in the coral Pocillopora verrucosa along environmental dictions of cyanobacteria in aquatic systems. Sci. Total Environ. 379 (1), 89–108. gradients of the Saudi Arabian Red Sea. Front. Mar. Sci. 2 (5), 1–10. http://dx.doi.org/ http://dx.doi.org/10.1016/j.scitotenv.2007.03.009. 10.3389/fmars.2015.00005. Heath, M.R., 1992. Field investigations of the early life stages of marine fish. Adv. Mar. Biol. Sawall, Y., Al-Sofyani, A., Banguera-Hinestroza, E., 2014a. Spatio-temporal analyses of 28, 1–174. http://dx.doi.org/10.1016/S0065-2881(08)60039-5. Symbiodinium physiology of the coral Pocillopora verrucosa along large-scale nutrient Heinz, S.K., Mazzucco, R., Dieckmann, U., 2009. Speciation and the evolution of dispersal and temperature gradients in the Red Sea. PLoS One e103179 http://dx.doi.org/10. along environmental gradients. Evol. Ecol. 23 (1), 53–70. http://dx.doi.org/10.1007/ 1371/journal.pone.0103179. s10682-008-9251-7. Sawall, Y., Al-Sofyani, A., Kuerten, B., Al-Aidaroos, A.M., Hoang, B.X., Marimuthu, N., Houde, E.D., 1989. Comparative growth, mortality, and energetics of marine fish larvae: Khomayis, H.S., Sommer, U., Gharbawi, W.Y., Wahl, M., 2014b. Coral communities, temperature and implied latitudinal effects. Fish. Bull. 87 (3), 471–495 (doi: in contrast to fish communities, maintain a high assembly similarity along the large 10.1.1.475.1985). latitudinal gradient along the Saudi Red Sea coast. J. Ecosyst. Ecography 3 (4), 1–7. Houde, E.D., Hoyt, R.D., 1987. Fish early life dynamics and recruitment variability. Am. http://dx.doi.org/10.4172/2157-7625.S4-003. Fish. Soc. Symp. 2, 17–29. Searcy, S.P., Sponaugle, S., 2000. Variable larval growth in a coral reef fish. Mar. Ecol. Ser. Houde, E.D., Zastrow, C.E., 1993. Ecosystem- and taxon-specific dynamic and energetics 206, 213–226. http://dx.doi.org/10.3354/meps206213. properties of larval fish assemblages. Bull. Mar. Sci. 53 (2), 290–335. Sponaugle, S.U., 2009. Tropical fish otoliths: information for assessment, management Hunt, I., Boutilier, R.G., Miyake, T., Hall, B.K., 1996. Effects of temperature on morpholog- and ecology. In: Green, B.S., Mapstone, B.D., Carlos, G., Begg, G.A. (Eds.), Methods ical landmarks critical to growth and survival in larval Atlantic cod (Gadus morhua). and Technologies in Fish Biology and Fisheries, Reviews: Methods and Technologies Mar. Biol. 124 (4), 593–606. http://dx.doi.org/10.1007/BF00351041. in Fish Biology and Fisheries. Springer Netherlands, Dordrecht, pp. 93–132 http:// Landaeta, M.F., Castro, L.R., 2006. Larval distribution and growth of the rockfish, Sebastes dx.doi.org/10.1007/978-1-4020-5775-5. capensis (Sebastidae, Pisces), in the fjords of southern Chile. ICES J. Mar. Sci. 63 (4), Sponaugle, S., 2010. Otolith microstructure reveals ecological and oceanographic process- 714–724. http://dx.doi.org/10.1016/j.icesjms.2006.01.002. es important to ecosystem-based management. Environ. Biol. Fish 89 (3), 221–238. Lester, S.E., Ruttenberg, B.I., 2005. The relationship between pelagic larval duration and http://dx.doi.org/10.1007/s10641-010-9676-z. range size in tropical reef fishes: a synthetic analysis. Proc. Biol. Sci. 272, 585–591. Sponaugle, S., Grorud-Colvert, K., 2006. Environmental variability, early life-history traits, http://dx.doi.org/10.1098/rspb.2004.2985. and survival of new coral reef fish recruits. Integr. Comp. Biol. 46 (5), 623–633. http:// Lo-Yat, A., Simpson, S.D., Meekan, M., Lecchini, D., Martinez, E., Galzin, R., 2011. Extreme cli- dx.doi.org/10.1093/icb/icl014. matic events reduce ocean productivity and larval supply in a tropical reef ecosystem. Sponaugle, S., Boulay, J., Rankin, T., 2011. Growth- and size-selective mortality in pelagic Glob. Chang. Biol. 17, 1695–1702. http://dx.doi.org/10.1111/j.1365-2486.2010.02355.x. larvae of a common reef fish. Aquat. Biol. 13, 263–273. http://dx.doi.org/10.3354/ Lozano-Cortés, D.F., Berumen, M.L., 2015. Colony size-frequency distribution of ab00370. pocilloporid juvenile corals along a natural environmental gradient in the Red Sea. Sponaugle, S., Grorud-Colvert, K., Pinkard, D., 2006. Temperature-mediated variation in Mar. Pollut. Bull. http://dx.doi.org/10.1016/j.marpolbul.2015.10.051. early life history traits and recruitment success of the coral reef fish Thalassoma Luiz, O.J., Allen, A.P., Robertson, D.R., Floeter, S.R., Kulbicki, M., Vigliola, L., Becheler, R., bifasciatum in the Florida Keys. Mar. Ecol. Prog. Ser. 308, 1–15. http://dx.doi.org/10. Madin, J.S., 2013. Adult and larval traits as determinants of geographic range size 3354/meps308001. among tropical reef fishes. PNAS 110 (41), 16498–16502. http://dx.doi.org/10.1073/ StatSoft, Inc, 2007. STATISTICA (data analysis software system). Version 8.0. pnas.1304074110. www.statsoft.com. Macpherson, E., Raventos, N., 2006. Relationship between pelagic larval duration and geo- Takahashi, M., Watanabe, Y., 2005. Effects of temperature and food availability on growth graphic distribution of Mediterranean littoral fishes. Mar. Ecol. Prog. Ser. 327, rate during late larval stage of Japanese anchovy (Engraulis japonicus)inthe 257–265. http://dx.doi.org/10.3354/meps327257. Kuroshio-Oyashio transition region. Fish. Oceanogr. 14 (3), 223–235. http://dx.doi. McCormick, M., 1994. Variability in age and size at settlement of the tropical goatfish org/10.1111/j.1365-2419.2005.00334.x. Upeneus tragula (Mullidae) in the northern Great Barrier Reef lagoon. Mar. Ecol. Thresher, R.E., Colin, P.L., Bell, L.J., 1989. Planktonic duration, distribution and population Prog. Ser. 103, 1–15. http://dx.doi.org/10.3354/meps103001. structure of western and central Pacificdamselfishes (Pomacentridae). Copeia 1989 McCormick, M., Molony, B., 1995. Influence of water temperature during the larval stage (2), 420–434. http://dx.doi.org/10.2307/1445439. on size, age and body condition of a tropical reef fish at settlement. Mar. Ecol. Prog. Victor, B.C., 1986. Duration of the planktonic larval stage of one hundred species of Pacific Ser. 118, 59–68. http://dx.doi.org/10.3354/meps118059. and Atlantic wrasses (family Labridae). Mar. Biol. 90 (30), 317–326. http://dx.doi.org/ Mcleod, I.M., Mccormick, M.I., Brooker, R.M., 2015. Latitudinal variation in larval develop- 10.1007/BF00428555. ment of coral reef fishes: implications of a warming ocean. Mar. Ecol. Prog. Ser. 521, Victor, B.C., 1991. Settlement strategies and biogeography of reef fishes. In: Sale, P.F. (Ed.), 129–141. http://dx.doi.org/10.3354/meps11136. The Ecology of Fishes on Coral Reefs. Academic Press, Inc., San Diego, California, Mcleod, I.M., Rummer, J.L., Clark, T.D., Jones, G.P., Mccormick, M.I., Wenger, A.S., Munday, pp. 231–260 http://dx.doi.org/10.1016/b978-0-08-092551-6.50014-3. P.L., 2011. Climate change and the performance of larval coral reef fishes: the interac- Victor, B.C., Wellington, G.M., 2000. Endemism and the pelagic larval duration of reef fish- tion between temperature and food availability. Contemp. Phys. 1, 1–12. http://dx. es in the eastern Pacific Ocean. Mar. Ecol. Prog. Ser. 205, 241–248. http://dx.doi.org/ doi.org/10.1093/conphys/cot024. 10.1098/rspb.2004.2985. Meekan, M.G., Carleton, J.H., McKinnon, A.D., Flynn, K., Furnas, M., 2003. What determines Wellington, G.M., Victor, B.C., 1989. Planktonic larval duration of one hundred species of the growth of tropical reef fish larvae in the plankton: food or temperature? Mar. Pacific and Atlantic damselfishes (Pomacentridae). Mar. Biol. 101 (4), 557–567. Ecol. Prog. Ser. 256, 193–204. http://dx.doi.org/10.3354/meps256193. http://dx.doi.org/10.1007/BF00541659. Miller, T.J., Crowder, L.B., Rice, J.A., Marschall, E.A., 1988. Larval size and recruitment Wellington, G.M., Victor, B.C., 1992. Regional differences in duration of the planktonic lar- mechanisms in fishes: toward a conceptual framework. Can. J. Fish. Aquat. Sci. 45 val stage of reef fishes in the eastern Pacific Ocean. Mar. Biol. 113 (3), 491–498. (9), 1657–1670. http://dx.doi.org/10.1139/f88-197. http://dx.doi.org/10.1007/BF00349176. Mora, C., Chittaro, P.M., Sale, P.F., Kritzer, J.P., Ludsin, S.A., 2003. Patterns and processes in Wilson, D.T., McCormick, M.I., 1997. Spatial and temporal validation of settlement-marks reef fish diversity. Nature 421, 933–936. http://dx.doi.org/10.1038/nature01421.1. in the otoliths of tropical reef fishes. Mar. Ecol. Prog. Ser. 153, 259–271. http://dx.doi. Nanninga, G.B., Saenz-Agudelo, P., Manica, A., Berumen, M.L., 2014. Environmental gradi- org/10.3354/meps153259. ents predict the genetic population structure of a coral reef fish in the Red Sea. Mol. Wilson, D.T., Meekan, M.G., 2002. Growth-related advantages for survival to the point of Ecol. 23 (3), 591–602. http://dx.doi.org/10.1111/mec.12623. replenishment in the coral reef fish Stegastes partitus (Pomacentridae). Mar. Ecol. Ngugi, D.K., Antunes, A., Brune, A., Stingl, U., 2012. Biogeography of pelagic Prog. Ser. 231, 247–260. http://dx.doi.org/10.3354/meps231247. bacterioplankton across an antagonistic temperature-salinity gradient in the Red Zapata, F.A., Herrón, P.A., 2002. Pelagic larval duration and geographic distribution of Sea. Mol. Ecol. 21 (2), 388–405. http://dx.doi.org/10.1111/j.1365-294X.2011.05378.x. tropical eastern Pacific snappers (Pisces: Lutjanidae). Mar. Ecol. Prog. Ser. 230, Racault, M.-F., Raitsos, D.E., Berumen, M.L., Brewin, R.J.W., Platt, T., Sathyendranath, S., 295–300. http://dx.doi.org/10.1098/rspb.2004.2985. Hoteit, I., 2015. Phytoplankton phenology indices in coral reef ecosystems: application to ocean-color observations in the Red Sea. Remote Sens. Environ. 160, 222–234. http://dx.doi.org/10.1016/j.rse.2015.01.019.

Contents lists available at ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier.com/locate/marpolbul

Homogeneity of coral reef communities across 8 degrees of latitude in the Saudi Arabian Red Sea☆,☆☆

May B. Roberts a,⁎,GeoffreyP.Jonesb,c,MarkI.McCormickb,c,PhilipL.Mundayb,c, Stephen Neale b, Simon Thorrold d,VanessaS.N.Robitzcha, Michael L. Berumen a a Red Sea Research Center, Division of Biological and Environmental Science and Engineering, King Abdullah University of Science and Technology, Thuwal 23955, Saudi Arabia b Australian Research Council Centre of Excellence for Coral Reef Studies, James Cook University, Townsville, Queensland 4811, Australia c College of Marine and Environmental Sciences, James Cook University, Townsville, Queensland 4811, Australia d Biology Department, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA article info abstract

Article history: Coral reef communities between 26.8°N and 18.6°N latitude in the Saudi Arabian Red Sea were surveyed to pro- Received 30 March 2015 vide baseline data and an assessment of fine-scale biogeography of communities in this region. Forty reefs along Received in revised form 26 October 2015 1100 km of coastline were surveyed using depth-stratified visual transects of fish and benthic communities. Fish Accepted 6 November 2015 abundance and benthic cover data were analyzed using multivariate approaches to investigate whether coral reef Available online 20 November 2015 communities differed with latitude. A total of 215 fish species and 90 benthic categories were recorded on the surveys. There were no significant differences among locations in fish abundance, species richness, or among sev- Keywords: Benthic cover eral diversity indices. Despite known environmental gradients within the Red Sea, the communities remained Biogeography surprisingly similar. The communities do, however, exhibit subtle changes across this span of reefs that likely re- Coral reef fishes flect the constrained distributions of several species of reef fish and benthic fauna. Dissimilarity © 2015 Elsevier Ltd. All rights reserved. Red Sea

1. Introduction (Berumen et al., 2013). The lack of baseline information on fish populations and species ranges within much of the Red Sea hinders attempts to The Red Sea is located in the northwest periphery of the Indian quantify changes in the local ecology due to environmental fluctuations Ocean and has long been recognized as its own biogeographic region or increasing anthropogenic influences. and a hotspot for biodiversity (Goren and Dor, 1994; Randall, 1994; In addition to its unique set of fauna, the Red Sea is also recog- Randall, 1998) with high levels of endemism (Briggs, 1974; Spalding nized as a mostly thriving coral reef ecosystem coexisting within rel- et al., 2007; Briggs and Bowen, 2012; Bowen et al., 2013; Kulbicki atively extreme environmental conditions (Sheppard et al., 1992). et al., 2013, DiBattista et al., in press). Some of the earliest tropical Only the Arabian Gulf supports coral reef environments that experi- marine expeditions were conducted in the Red Sea, where pioneering ence higher temperatures and salinity levels than those located in naturalists described marine fauna which was also representative of the Red Sea (Sheppard et al., 1992). However, across this long and the greater Indian Ocean (Forsskål et al., 1775; Rüppell, 1828; Cuvier, narrow body of water, which spans 17 degrees of latitude, the Red 1828; Ehrenberg, 1834; Klunzinger, 1870). More recently, with the Sea is not homogenous. Sea surface temperatures (SST), salinity, exception of the Gulf of Aqaba, there has been relatively little ecological and nutrient concentrations exhibit latitudinal gradients and fluctu- research in the Red Sea compared to other major tropical reef systems ate seasonally (Acker et al., 2008; Ngugi et al., 2012; Raitsos et al., 2013). Average temperatures increase southward and range from 20–28 °C (north to south) in the winter and 26–32 °C (north to ☆ The RES team (SRT, GPJ, MIM, PLM, SN, MLB) conducted these surveys as part of a south) in summer. The low rainfall and freshwater influx in this fi WHOI-KAUST partnership, with this project speci cally aiming to create some form of hot, arid region and pronounced evaporation rates result in high sa- baseline data for future work at KAUST. The Reef Ecology Lab at KAUST (MBR, VSNR, MLB) is interested in generally understanding Red Sea ecology, particularly in a linity levels (~41 psu) which decrease (to 36 psu) near the Bab al comparative context to other Indo-Pacific reef systems. Mandeb Strait, the only connection to the Indian Ocean (Murray ☆☆ Author contributions: SRT, GPJ, MIM, PLM, SN, MLB collected the data. MBR analyzed and Johns, 1997). Nutrient levels in the Red Sea also increase from the data. SRT and MLB provided funding. MBR and MLB wrote the manuscript. All authors north (chlorophyll-a = 0.03 [mg m−3]) to south (10 [mg m−3]), contributed to manuscript sections or general editing. with the most oligotrophic northern waters characterized by high ⁎ Corresponding author at: Red Sea Research Center, King Abdullah University of Science and Technology, Thuwal 23955, Saudi Arabia. visibility in contrast to the more turbid southern region (Sheppard E-mail address: [email protected] (M.B. Roberts). and Sheppard, 1991).

The Red Sea has recently been described as containing two marine ecoregions (Spalding et al., 2007), with a division in the central Red Sea located near 20°N latitude. This simplified delineation is contentious among some researchers familiar with the region. Nonetheless, it provides a framework for us to test a hypothesis. While levels of endemism are key characteristics for establishing broader biogeographical provinces and realms (Briggs, 1974; Briggs and Bowen, 2012; Spalding et al., 2007), the finer-scale marine ecoregions such as those identified within the Red Sea are defined as “areas of relatively homogeneous species composition, clearly distinct from adjacent systems…[and] determined by a distinct suite of oceanographic or topographic features” (Spalding et al., 2007). However, the available data from this region is focused only on a few taxonomic groups. For example, regional chaetodontid and pomacanthid distributions were explored by Roberts et al. (1992) and Righton et al. (1996) while Sheppard and Sheppard (1991) and DeVantier and Pilcher (2000) have published studies on the distribution of scleractinian assemblages within the Red Sea. While these previous studies provide valuable insight to species- and family-level distributions and patterns, they may not be sufficient to characterize ecoregion boundaries. Large-scale biogeographic trends provide insights into broad ecological processes and relationships to changing environmental conditions; understanding these trends facilitates the establishment of sound management plans. The present study provides an overview of biogeographic patterns of reef communities for this region. The aims of this study were: 1) to determine if and to what degree offshore reef communities change along a latitudinal gradient within our study area, 2) to explore the presence of a within-Red Sea ecological boundary at 20°N as described by Spalding et al. (2007), and 3) to provide baseline data on the biogeography of coral reef communities for fu- Fig. 1. Reef sites surveyed in the Red Sea for fish abundance and benthic cover, with north- ture comparative studies in the Red Sea. To achieve these goals, surveys ern (green triangles), central (dark blue circles), and southern (cyan squares), groupings were conducted on coral reefs in the Saudi Arabian Red Sea spanning of survey sites. Reefs were numbered (shown next to reefs) in order of latitude and these correspond to the numbers in Table 1. 1100 km of latitudinal coastline. Surveys assessed the abundance of reef fish species as well as benthic cover at 40 coral reefs. In addition to providing a valuable dataset of distributions and abundances, this data lays the foundation for investigations of the mechanisms underly- context, while the sections were used to investigate finer-scale spatial ing regional biogeography. variation in community indices. At each reef, four replicate transects were laid at each of four depths: 2. Methods the reef crest (~0 m), 2 m, 6 m, and 10 m, for a total of 16 transects per reef (as per Jones et al., 2004). All species (Appendix S1) were counted 2.1. Ecological survey data collection on 50 m × 4 m belt transects with the exception of the damselfishes (family Pomacentridae) which were surveyed within a 2 m belt tran- Our study area consisted of 40 offshore reefs along the Saudi Arabian sect, as well as the gobies (family Gobidae), blennies (family Blennidae), coastline between 26.8°N and 18.6°N latitude (Fig. 1). Survey sites were and dottybacks (family Pseudochromidae), which were surveyed with- selected to reduce the confounding effects of reef type, reef slope, and in a 1 m belt due to their small size and abundance. For each transect, within-reef location of transects. Reefs were chosen based on their posi- three divers (MIM, PLM, and either GPJ or MLB) observed and recorded tion at the edge of the Arabian shelf and near deep drop-offs, with sur- the abundance of specificgroupsoffishes based on their expertise. Prior vey sites located near the outer reef slope on the leeward side of the reef to other analyses, count densities were standardized to 200 m2 (hereaf- (given predominant northwesterly winds in the Red Sea, this meant ter referred to as ‘abundance’). A fourth diver (SN) conducted point- that our surveys were conducted on the southern ends of the reefs). intercept benthic surveys along the same 50 m transects, recording Reef sites varied between 7 and 81 km from shore, representing the var- the benthos at 100 random points selected by a random number gener- iable width of the continental shelf in the Red Sea (Fig. 1). The lack of ator that allowed for at least two points to be within the bounds of each any significant rainfall (yearly average less than 70 mm (DeVantier meter of the transect to space the points out across the entire transect. and Pilcher, 2000)), the near absence of freshwater runoff, and minimal The substratum under each point was identified to the lowest taxonom- coastal development (especially in the north) greatly reduces the ic group and morphotype where possible (Appendix S2). confounding effects of varying reef distance from shore. We define the terminology used in this study as follows. “Sub-region” refers to the 2.2. Statistical analyses subdivisions of the Red Sea that we surveyed (i.e., the northern, central, or southern sites within the Red Sea), “section” to refer to the groupings 2.2.1. Community indices of reefs, and finally, we refer to each individual reef site as “reef”.We Several community indices were assessed among the three latitudi- surveyed four reefs per section with a total of 10 sections that were fur- nal sub-regions in our study. As each reef had the same number of tran- ther grouped into three sub-regions (Table 1). These sub-regions were sects, total abundance of all fishes surveyed within a reef were used to defined as: “northern” (n = 12 reefs; 26.8°N–24.4°N), “central” (n = calculate the average total abundance of fishes in each of the three 16 reefs; 23.8°N–21.8°N), and “southern” (n = 12 reefs; 19.8°N– sub-regions (n = 12 or 16 reefs per sub-region). Total richness (S) 18.6°N) (see Fig. 1 and Table 1). The groupings by sub-region were was calculated at the reef level by tallying the number of fish species re- used to identify differences within our study area for a biogeographic corded on all of the 16 transects (i.e., if a fish was seen on any transect it 560 M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565

Table 1 Details of 40 Saudi Arabian Red Sea coral reefs surveyed. Surveys were conducted in three general sub-regions of the Red Sea. Ten regions were surveyed, each comprised of four reefs. Reefs were assigned a numerical code (1–40) in order of latitude and used in subsequent ﬁgures. Longitude and latitude indicate survey location on reef. Distance from shore is also reported and shows the straight-line distance to the nearest point on land. The total number of ﬁsh species (S) recorded on 16 belt transects (4 each at the crest, 2 m depth, 6 m depth, and 10 m depth) using visual surveys are reported, along with other community indices: Jʹ, indicating Pielou's evenness; Hʹ, representing Shannon's Diversity index; and 1-λ, representing Simpson's Diversity index.

Sub-region Reef ID # Section Reef name Latitude Longitude Survey year Dist.(km) S Jʹ Hʹ(loge) 1-λʹ

North 1 Wajh Pele 26.80908 35.89095 2011 16.0 68 0.41 1.72 0.62 2 Wajh Skharu Luhs 1 26.62883 36.25481 2011 8.10 93 0.51 2.30 0.72 3 Wajh Skharu Luhs 3 26.40832 36.26557 2011 9.6 87 0.47 2.09 0.71 4 Wajh Skharu Luhs 2 26.37708 36.25453 2011 9.6 86 0.50 2.23 0.78 5 Wajh Bank Wajh Bank 1 25.39082 36.68348 2011 34.3 85 0.58 2.58 0.80 6 Wajh Bank Wajh Bank 2 25.27030 36.85697 2011 25.2 78 0.47 2.07 0.74 7 Wajh Bank Wajh Bank 3 25.24035 36.93472 2011 19.3 82 0.47 2.09 0.70 8 Wajh Bank Wajh Bank 4 25.15465 36.91172 2011 27.5 75 0.44 1.88 0.64 9 Umm Lujj Marker 7–1 24.45313 37.19970 2011 7.0 88 0.53 2.39 0.75 10 Umm Lujj Marker 7–2 24.44277 37.20667 2011 20.5 86 0.46 2.05 0.70 11 Umm Lujj Marker 7–3 24.43110 37.22140 2011 20.7 72 0.43 1.82 0.61 12 Umm Lujj Marker 7–4 24.727 37.151 2011 20.2 82 0.59 2.58 0.81 Central 13 Seven Sisters Abu Galawa 23.86382 37.88830 2008 27.9 85 0.38 1.70 0.56 14 Seven Sisters No Name 23.83428 37.89798 2008 30.1 81 0.42 1.86 0.70 15 Seven Sisters Shi'b Shabarir 23.78768 37.93590 2008 33.1 83 0.31 1.37 0.56 16 Seven Sisters Shib Sufmami 23.75252 37.96917 2008 33.2 72 0.38 1.64 0.69 17 Rabigh Maria's Reef 22.85080 38.72097 2008 16.0 85 0.43 1.89 0.70 18 Rabigh Khamsa 2 22.79837 38.61450 2008 28.6 78 0.46 2.01 0.78 19 Rabigh Noura 22.74988 38.61977 2008 31.2 79 0.47 2.06 0.79 20 Rabigh Bayeda long 22.72068 38.79622 2008 17.8 86 0.42 1.87 0.67 21 Thuwal 3 Stick Reef 22.45928 38.90508 2008 18.2 82 0.74 3.28 0.94 22 Thuwal Al Mutarbej 22.42913 38.94718 2008 13.5 98 0.70 3.19 0.93 23 Thuwal Al Mutarbej South 22.39037 38.91820 2008 16.4 87 0.65 2.91 0.88 24 Thuwal Shi'b Nazar 22.37217 38.89715 2008 18.7 90 0.59 2.67 0.84 25 North Jeddah Madafi 22.05675 38.76688 2008 17.7 85 0.55 2.45 0.79 26 North Jeddah South reef 21.93408 38.86485 2008 9.17 98 0.58 2.65 0.84 27 North Jeddah Coral gardens 21.86748 38.75643 2008 20.9 95 0.63 2.89 0.91 28 North Jeddah Abu Terr 21.86605 38.85972 2008 12.3 94 0.69 3.15 0.91 South 29 Al-Lith Mar Mar 19.84335 39.93358 2009 47.4 101 0.48 2.23 0.75 30 Al-Lith Dohra Island 19.82893 39.89853 2009 51.1 73 0.41 1.76 0.69 31 Al-Lith Al-Jadir 19.78848 39.95683 2009 49.4 87 0.41 1.81 0.63 32 Al-Lith Long Reef 19.76643 39.89223 2009 56.2 78 0.37 1.60 0.61 33 Al-Qunfidhah AQ4 19.15483 40.30113 2009 71.3 71 0.51 2.16 0.75 34 Al-Qunfidhah AQ3 19.10642 40.31775 2009 73.8 89 0.59 2.65 0.81 35 Al-Qunfidhah Murabit 1 19.02432 40.31792 2009 77.6 79 0.57 2.49 0.84 36 Al-Qunfidhah Petit Murabit 19.00238 40.28493 2009 81.5 73 0.45 1.95 0.66 37 Ablo Ablo 4 18.70673 40.65362 2009 57.0 88 0.57 2.56 0.84 38 Ablo Ablo 1 18.67510 40.73922 2009 50.1 85 0.66 2.94 0.89 39 Ablo Ablo 3 18.66772 40.65928 2009 58.9 82 0.50 2.20 0.75 40 Ablo Ablo 2 18.66500 40.81282 2009 41.6 83 0.62 2.75 0.82

was counted for the reef). The Shannon–Weiner diversity index (Hʹ, Warwick, 2001). While there are many options for similarity indices, log(e) scale) (Shannon and Weaver, 1963), Pielou's evenness (Jʹ) the Bray–Curtis method has been shown to be sensitive to differences (Pielou, 1975), and Simpson's diversity (1-λ)(Simpson, 1949)were in community structure when using species abundance data, and is all calculated at the reef level using PRIMER-v6 (Clarke and Gorley, thus commonly used in coral reef community studies (Burt et al., 2006). Values of Jʹ lie between 0–1 where 0 indicates an uneven distri- 2011; Dornelas et al., 2006; Holbrook et al., 2015). Similarity coefficients bution of abundances among species within a site and values closer to are calculated between assemblages of every pair of reefs using the av- 1 show more evenly composed communities. Similarly, Simpson's erage abundances of each species within the reef. Here, the Bray–Curtis index is also on a 0–1 scale, where values near 0 are interpreted as similarity coefficient S, represents the similarity between reefs j and k less diverse and values approaching 1 signify high diversity. All of the where yij represents the average abundance of the species in column i aforementioned indices were compared for significant differences and reef j. among the three sub-regions using one-way Analysis of Variance 8 X 9 (ANOVA) in R v3.03 (RCoreTeam,2014). < p j ‐ = yij yik S ¼ 100 1−X i¼1 : jk : p ; y þy 2.2.2. Multivariate analyses among reefs i¼1 ij ik We also analyzed assemblage and biogeographic patterns in the data using non-metric multi-dimensional scaling (NMDS; Kruskal and Wish, 1978) to display the relative dissimilarity distances based on communi- 2.2.3. Multivariate analyses among sub-regions ty compositions among the 40 reefs. For this and all subsequent multi- To determine if any biogeographic differences were evident, reef variate analyses, data were analyzed at the reef level (treating the 16 sites were analyzed based on the groupings by the previously defined transects as replicates within a reef). Fish abundance data and percent northern, central, and southern sub-regions and assessed to determine benthic cover data were square root transformed to balance the effect if communities were similar among sub-regions. Analysis of similarity of disproportionately abundant species prior to conducting the analy- (ANOSIM; Clarke and Warwick, 2001)wasconductedonboththefish ses. The NMDS plots were calculated using the resemblance matrix of and benthic resemblance matrices to test the null hypothesis that the Bray–Curtis similarity coefficients (Bray and Curtis, 1957; Clarke and community assemblages were similar throughout the three sub-regions. M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565 561

In the ANOSIM significancetest,theresultingglobalR-value lies between −1and1andreflects the degree of similarity between the pairwise tests between the three sub-regions. The more similar reef communities within sub-regions are to each other than to those in other sub-regions, the closer the value of R approaches 1. Positive values nearer to zero indicate that some reefs between sub-regions show more similarity than reefs within the same sub-region, while a negative R value would indicate that reefs between sub-regions are more similar to each other than within the three sub-regions (Clarke and Warwick, 2001). A pairwise comparison of similarity percentages (SIMPER; Clarke and Warwick, 2001) of abundance data between sub-regions was used to determine which species and benthic components contributed most to the differences between sub-regions. This analysis first identifies the species that are most influential in characterizing a sub- region, and then ranks the species that drive average dissimilarity between them. Species with greater relative abundances are more heavily weighted in SIMPER calculations.

2.2.4. Correlations between fish and benthos To test whether relationships between abundances of fish taxa versus benthic components exist, we used the RELATE (Clarke and Warwick, 2001) analysis with Spearman's rank method. This analysis shows how well the fish resemblance matrix correlates with the benthic matrix. Spearman's rank correlation (ρ) is a measure of the degree of correlation between the two resemblance matrices, calculated by matching every variable in one matrix to every variable in the other matrix then creating a ranking based on the similarity of ρ values between benthic categories and fish species across sites. Once patterns were detected, BEST (BVSTEP) analysis (Clarke and Warwick, 2001) used this ranking to identify which of the benthic variables most strongly correlated with fish community compositions. Due to the large number of variables, we used benthic categories that contributed at least 1% of dissimilarity between regions as identified by the SIMPER analysis (41 of 90 categories). These analyses are commonly used to distinguish between different communities and to identify geographic sub-regional breaks (Khalaf and Kochzius, 2002, Burt et al., 2011; Sales et al., 2012). Fig. 2. Community indices for fish assemblages at 40 reefs in the Saudi Arabian Red Sea as All multivariate and correlation analyses were conducted using recorded using ecological survey data. The 40 reefs were grouped by the three sub-regions defined as: “northern” 26.8°–24.4°N (n = 12 reefs), “central” 23.8°–21.8°N (n = 16 reefs), PRIMER 6+ PERMANOVA software (PRIMER-E v6; Clarke and Gorley, and “southern” 19.8°–18.6°N (n = 12). Each reef contained 16 transects. A) Mean abun- 2006). dance of all recorded individuals per reef (all transects totaled at each reef). B) Mean total species richness (S) recorded on each reef. C) Mean Shannon–Weiner diversity 3. Results index (Hʹ) (log(e) basis) values. D) Mean Pielou's evenness (Jʹ) values. All boxplots show the mean (bold line) with the upper and lower quartiles while whiskers indicate the maximum and minimum values found on the reefs within each section of the 3.1. Community indices coastline.

A total of 268,313 individuals were counted representing 215 species from the 40 reefs (Appendix S1). However, nine fish taxa were 3.2. Fish assemblage multivariate analyses across reefs and between subsequently removed from further analysis because they were not sub-regions identified to species and an additional four species, all within the family Mullidae, were also excluded, as they were not counted in The NMDS plot (Fig. 3)forfishes within reefs revealed the clustering all of the survey periods (Appendix S3). The remaining 202 fish species, of northern and southern sites separated by a wide band of central reef representing 110 genera and 26 families were used for subsequent sites, generally showing a gradual progression of community assem- analyses (Appendix S1 and S2). blages from north to south. Pairwise ANOSIM tests by sub-region Total abundance was not significantly different across sub-regions also confirmed significant differences between the three sub-regions

(one-way ANOVA, F2,37 =2.76,p = 0.0764). Similarly, the other com- (p b 0.001, R = 0.321) (Table 2). The benthic NMDS plot (Fig. 3) also munity indices showed no signiﬁcant differences among sub-regions. shows clustering of reef sites based on sub-regions. Benthic ANOSIM

These included mean species richness (one-way ANOVA, F3,37 = results conﬁrmed that sub-regional clusters are signiﬁcant (p b 0.001, 1.319, p =0.280)(Fig. 2b), Shannon's diversity (Hʹ) (one-way R = 0.632) with all pairwise tests yielding p values of 0.001 (Table 3).

ANOVA, F2,37 = 0.585, p = 0.562), Pielou's evenness ( Jʹ) (one-way However, in the SIMPER pairwise comparisons, four species ANOVA, F2,37 = 0.437, p = 0.649), and Simpson's index (1-λʹ) (one- (Pseudanthias squamipinnis, Chromis dimidiata, Chromis flavaxilla, and way ANOVA, F2,37 = 1.487, p = 0.239 (Fig. 2b–d). Pseudochromis fridmani) were consistently identified as being the pri- Among the individual reefs, the highest and lowest values of mary drivers of between-sub-region dissimilarity (Table 2). The three Shannon's diversity, Pielou's evenness, and Simpson's index were with- sub-regions were also characterized by a similar assemblage of fishes in the central sub-region. Abu Terr reef in the region of North Jeddah had with the same 19 species making up the top 25 most influential species the highest indices with Hʹ =3.15,Jʹ = 0.69, and 1-λʹ = 0.91, while the driving the similarities within each of the three sub-regions (Table 4). lowest values were found on Shib Shabarir in the Seven Sisters area with None of the three sub-regions were found to have a notably distinct as- Hʹ =1.37,Jʹ =0.31,and1-λʹ =0.56(Table 2). semblage structure based on pairwise dissimilarity values. SIMPER 562 M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565

Table 2 Similarity results of reef ﬁsh assemblages determined using SIMPER and ANOSIM analyses. The tables show the species cumulatively contributing to the top 25% of the dissimilarityineach pairwise comparison (SIMPER) of three sub-regions from the Saudi Arabian Red Sea. Relative contribution and the cumulative contribution of the top species to sub-regional dissimilarity as well as the mean abundances for each species in each sub-region are shown for comparison. All other species each contributed b2% of differences in assemblages. Percent dissimilarity in addition to overall ANOSIM comparison results (Global R = 0.321, p b 0.001) are included alongside the pairwise results, also presented here.

Sub-regions compared: Species % Contribution to % Cumulative Sub-region 1 mean Sub-region 2 mean dissimilarity contribution abundance abundance

North and Central Pseudoanthias squamipinnis 7.9 7.9 5.2 9.3 Average dissimilarity = 38.4% Chromis dimidiata 7.5 15.3 16.2 16.5 ANOSIM Pseudochromis fridmani 4.8 20.1 5.7 6.1 Global R = 0.212, p = 0.009 Chromis flavaxilla 3.6 23.6 4.7 5.4 Chromis viridis 2.9 26.5 2.1 2.5 North and South Chromis dimidiata 5.1 5.1 16.2 17.9 Average dissimilarity = 39.2% Pseudoanthias squamipinnis 4.9 10.0 5.2 6.6 ANOSIM: Chromis flavaxilla 4.5 14.5 4.7 7.6 Global R = 0.651, p = 0.001 Eviota guttata 3.4 17.9 0.3 3.5 Pseudochromis fridmani 2.8 20.7 5.7 5.8 Chromis viridis 2.5 23.2 2.1 2.2 Chrysiptera unimaculata 2.3 25.4 0.3 2.3 Central and South Chromis dimidiata 7.0 7.0 16.5 17.9 Average dissimilarity = 42.96% Pseudoanthias squamipinnis 7.0 14.0 9.3 6.6 ANOSIM: Pseudochromis fridmani 4.0 18.0 6.1 5.8 Global R = 0.260, p = 0.001 Chromis flavaxilla 3.8 21.8 5.4 7.6 Eviota guttata 2.6 24.3 0.8 3.5 Chromis viridis 2.4 26.7 2.5 2.2 results show that benthic categories driving the top 25% of differences in 3.3. Correlations between fish and benthos the ANOSIM were more diverse than the combinations that drive differences in the fish ANOSIM (Tables 2 and 3). Percent contribution levels The RELATE analysis showed a significant positive correlation be- were also more even and were generally higher among these benthic tween patterns in the fish assemblages and the corresponding benthic factors than contributions found in the fish community (Tables 2 communities (p b 0.001, ρ = 0.386). The SIMPER analysis identified and 3). As in the fish assemblages, low dissimilarity values indicate rel- that 41 of the 90 benthic categories contributed at least 1% to dissimilarity atively low levels of differences among sub-regions. in benthic communities. We therefore used BEST (BVSTEP) to examine

Fig. 3. Cluster analysis and non-metric multidimensional scaling (NMDS) of reef communities at 40 reefs in the Saudi Arabian Red Sea. Each reef contained 16 transects on which abundance of 202 species of fishes were recorded and benthic cover determined. Fish abundance data were square-root transformed prior to creating the resemblance matrix. A) Cluster analysis of the 40 reefs based on fish abundances. B) Cluster analysis of the 40 reefs based on percent cover of benthic categories. C) NMDS of fish assemblages. The two-dimensional distance between sites signifies the relative degree of difference between assemblages. D) NMDS of benthic communities. Colors and shapes denote geographical assignments to one of three sub- regions: Northern sites (1–12, green triangles), central sites (13–29, dark blue circles), and southern sites (30–40, light blue squares). M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565 563

Table 3 Similarity results of benthic communities determined using SIMPER and ANOSIM analyses. The tables show the benthic categories cumulatively contributing to the top 25% of the dissimilarity in each pairwise comparison (SIMPER) of three sub-regions from the Saudi Arabian Red Sea. Relative contribution and the cumulative contribution of the top benthic group to sub- regional dissimilarity as well as the mean percent cover for each in the three sub-regions are shown for comparison. All others contributed b2% of differences in communities. Percent dissimilarity in addition to overall ANOSIM comparison results (Global R = 0.632, p b 0.001) are included alongside the pairwise results, also presented here.

Sub-regions compared: Species % Contribution to % Cumulative Sub-region 1 mean Sub-region 2 mean dissimilarity contribution abundance abundance

North and Central Xeniidae 5.2 5.2 2 3.3 Average dissimilarity = 35.9% ANOSIM: Coralline (encrusting flat) 4.9 10.1 2.5 3.7 Global R = 0.702, p = 0.001 Rubble (turf on rubble) 4.5 14.7 1.3 0.2 Millepora 4.4 19 2.1 0.9 Coralline (turf algae on rock) 3.8 22.8 6.5 5.9 Sand 3.6 26.4 1.1 2.1 North and South Coralline (encrusting flat) 4.9 4.9 2.5 3.7 Average dissimilarity = 36.8% ANOSIM: Coralline (turf algae on rock) 4.1 9 6.5 6 Global R = 0.7, p = 0.001 Rubble (turf on rubble) 3.8 12.8 1.3 0.2 Sponges (encrusting flat) 3.7 16.5 1.9 2.6 Millepora 3.5 20 2.1 1.1 Porites (encrusting columnar) 3.5 23.5 1.1 0 Acropora (digitate) 3.4 27 1.7 0.6 Central and South Coralline (encrusting flat) 4.6 4.6 3.7 3.7 Average dissimilarity = 34.5% ANOSIM: Xeniidae 4.5 9.1 3.3 2.7 Global R = 0.501, p = 0.001 Sinularia 3.9 13 1.7 0.7 Sand 3.9 16.9 2.1 1.3 Sponges (encrusting flat) 3.8 20.6 2.3 2.6 Coralline (turf algae on rock) 3.6 24.3 5.9 6

these 41 benthic variables in order to identify which benthic variables fi Table 4 may have contributed to shifts in the sh assemblages. However, this The most influential coral reef fishes out of the 200 species included in the analysis driving analysis did not find a benthic variable or a combination of benthic vari- similarities (using SIMPER analysis) between three regional sub-regions which are defined ables that correlated significantly (p = 0.07) with the changes in fish as: “northern” 26.8°N–24.4°N (n = 12 reefs), “central” 23.8°N–21.8°N (n = 16 reefs), and assemblagesacrosslatitude. “southern” 19.8°N–18.6°N (n = 12 reefs). This list is a compilation of the top 25 species characteristic of each individual sub-region based on total abundance in the sub-region. Marked cells denote whether that species was identified as a primary driver of similarity 4. Discussion within that sub-region. Previous large-scale studies of biogeographic patterns were con- Top 25 species North Central South ducted primarily on near-shore reefs in the Red Sea and had revealed 1 Acanthurus nigrofuscus xxxa trend of higher diversity in what corresponds to the northern and 2 Acanthurus sohal xxx central sub-regions in our study (Roberts et al., 1992; Sheppard and 3 Amblyglyphidodon indicus xxx 4 Centropyge multispinis xxxSheppard, 1991). These studies also identified overall latitudinal chang- 5 Chaetodon austriacus xxxes in the species composition of selected taxa such as the chaetodontids, 6 Chromis dimidiata xxxpomacanthids and sclerictinians (Roberts et al., 1992; Sheppard and fl 7 Chromis avaxilla xxxSheppard, 1991). Our extensive surveys of 40 reefs along 1100 km of 8 Cirripectes castaneus xxx fi 9 Ctenochaetus striatus xxxthe Saudi Arabia Red Sea coast encompassing 110 genera of shes and 10 Gobiodon rivulatus xxx90 benthic categories, presents a novel dataset representing offshore 11 Gomphosus caeruleus xxxreef communities. Though we do not directly compare our data to the 12 Halichoeres hortulanus xxxfindings in previous literature on inshore reefs, our study revealed 13 Labroides dimidiatus xxx that across latitude, offshore reef fish assemblages were generally 14 Pomacentrus sulfureus xxx 15 Pseudoanthias squamipinnis xxxmore homogenous in nature. While there were subtle assemblage shifts 16 Pseudocheilinus hexataenia xxxalong this gradient, they were not strong enough to be reflected in five 17 Pseudochromis fridmani xxxcommon indices, none of which differed significantly among the three 18 Pygoplites diacanthus xxxsub-regions. Nevertheless, reefs within the same area generally cluster 19 Thalassoma rueppellii xxx 20 Chaetodon paucifasciatus x closely together in an overall latitudinal pattern. Patterns found in the 21 Pseudocheilinus evanidus x fish assemblages also appear to be related to benthic composition, but 22 Naso elegans xx the exact drivers are difficult to identify. We did not find strong evi- 23 Plectroglyphidodon lucozonus xx dence for the current location of the within-Red Sea ecoregion bound- 24 Zebrasoma desjardinii xx ary designated by Spalding et al. (2007) at 20°N. 25 Chaetodon auriga x 26 Chromis viridis x The great deal of homogeneity and relatively small differences in the 27 Heniochus intermedius x species that characterize the reefs throughout our study area are in con- 28 Myripristis murdjan xxtrast to earlier studies on near-shore reefs and contradict the delinea- 29 Paracirrhites forsteri xxtions of distinct bioregions assigned in the MEOWS (Roberts et al., 30 Thalassoma lunare xx 31 Cephalopholis hemistiktos x 1992; Spalding et al., 2007). Subtle shifts are, however, apparent in 32 Chrysiptera unimaculata x the fish assemblage ordination clustering the reefs in a generally latitu- 33 Eviota guttata x dinal order. Northern reefs, as well as some southern reefs, were highly 34 Plectroglyphidodon lacrymatus x clustered within their respective sub-regions while many central and % Cumulative contribution 76% 72% 72% some southern reefs were grouped together also indicating a gradual 564 M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565 latitudinal shift. These patterns can be attributed to the several spe- We found that benthic communities reflected similar patterns to the cies that have restricted ranges within the Red Sea, an effect already fish assemblages, though the ecological relationships between such reported within the chaetodontid family (Roberts et al., 1992). large numbers of variables in the analysis may mask any clear associa- Several butterflyfishes exhibit distributions that are confined to ei- tions. The ordination in the two communities as well as the RELATE ther the northern, southern, or central sub-regions (e.g., Roberts analysis confirm that trends exhibited by the fish assemblages were et al., 1992). Although there were no apparent major habitat changes likely related to patterns found in the benthic communities. Interesting- that would limit these distributions, a more thorough examination of ly, ordination and SIMPER results showed that the northern reefs for specific case studies and habitat associations might reveal further both the fish and benthos show greater uniformity, indicating consis- connections. tency of communities among those reefs. The northern sub-region is ar- There is no standard value for a threshold of difference vs. similarity, guably the most environmentally challenging region in the Red Sea as a however, it is our opinion that the results indicate relatively homoge- result of having the overall lowest productivity (Raitsos et al., 2013)and nous communities (cf. Burt et al., 2011). We did not find evidence highest salinity (Ngugi et al., 2012). It is possible that specificnichespe- supporting the current division of the Red Sea into two bioregions at cialization become necessary as environments become relatively more 20°N latitude, based on the definition used by Spalding et al. (2007). extreme (e.g., Moldenke, 1975). The result of more challenging environ- The lack of separate clusters between sites across this delineation mental conditions may, therefore, be increased homogeneity among the (between central and southern sub-regions) and the uniformity in resident communities. Additionally, it is also interesting that the sub- species driving the similarities within sub-regions indicated a generally regional differences found on inshore reefs in Roberts et al. (1992) ap- homogeneous assemblage along this span of coastline. We also found pear to be more obvious and further north than our findings at offshore that dissimilarities (though significant) were half that of similar studies reef habitats. This may suggest that the forces or gradients that deter- comparing coral reef fish assemblages across other similarly defined mine species distribution in the Red Sea communities are more influen- bioregions that displayed substantially higher dissimilarity percentages tial or stronger in near-shore communities. (e.g., 72–85%) (Burt et al., 2011). This was further corroborated by the Future studies could use the data presented here as a basis for more high resemblance of the fishes (19 of the 25 species identified by in-depth work with the aim of identifying specific mechanisms underly- SIMPER analyses) that contribute to the similarity between each sub- ing the latitudinal gradient in community assemblages. For example, region. Across this stretch of the Red Sea, while the dissimilarities may subsequent studies may further examine reef sites with greater habitat be significant, communities remained relatively similar. Results from variability than the present study (e.g., comparing coastal fringing reefs the ANOSIM analysis indicated that fish assemblages were more similar to offshore reefs, or sheltered vs. exposed sides of a given reef). It is like- between adjacent sub-regions than other pairwise comparisons. The ly that some species exhibiting restricted ranges within the Red Sea, benthic communities formed clear clusters but with less evidence of a such as the butterflyfishes and angelfishes (Roberts et al., 1992), drive latitudinal pattern. However, subsequent BEST analyses and careful ex- the observed shifts in the assemblages along the latitudinal gradient amination of the reef ordination show patterns with similar clustering we explored. There are further reef-scale biological mechanisms between the two data sets. interacting with biogeographic mechanisms to produce community Given the lack of previously available detailed biogeographic infor- variation. For example, local population explosions of Acanthaster planci, mation for much of the Red Sea, and particularly offshore reefs, the Drupella, Echinometra, or coral disease have been known to significantly placement of an ecoregion boundary at 20°N appears to be a slightly alter Red Sea reef fish and coral communities and habitat structure misleading conclusion arising from over-simplification of previous (e.g., Antonius and Riegl, 1998; Khalil et al., 2013; Riegl et al., 2012, studies (although we acknowledge that compromises by Spalding 2013) which likely affect the fish assemblage structure. These types of et al. (2007) were likely necessary to keep the total number of global disturbances have previously been suggested as a potential homogeniz- ecoregions reasonable). Other recent investigations in the southern ing force in Red Sea reefs (Riegl et al., 2012). Future work and repeated Red Sea provide evidence that a more appropriate division may exist observations will be required to confirm this hypothesis. around 17.5°N latitude and southwards to the strait of Bab Al Mandab, where turbidity and productivity levels are much higher than the rest 5. Conclusion of the Red Sea (Raitsos et al., 2013). This shift in the southern Saudi Arabian Red Sea coincides with a distinct habitat change. An extensive Red Sea reef fish assemblages along the northern two-thirds of the network of coral reefs known as the Farasan Banks occupies the region eastern Red Sea are, for the most part, composed of similar assemblages from ~20°N to ~18°N, while to the south from ~17.5°N onwards, lies the of species with no dramatic changes in the general communities along Farasan Islands. The reef communities of the Farasan Islands extend into this latitudinal gradient when comparing reef communities on the Yemeni waters and have been described as unique among Red Sea hab- edge of the continental shelf. Nevertheless, shifts do occur, likely driven itats (Sheppard et al., 1992; Turak et al., 2007) in that they are character- by the respective range limits of several species that are confined to ei- ized by increasingly reduced coral reef development compared to more ther the northern or southern Red Sea. While we recognize that northern Red Sea coral communities. This area has shallow geomor- ecoregion delineation is not a fully quantitative endeavor, we recom- phology that results in high SST, turbidity, and restricted water flow mend that the previously described zonation in Spalding et al.'s Marine (Turak et al., 2007). Although our study did not include sites in the Ecoregions of the World within the Red Sea be reconsidered. This cen- Farasan Islands, other work from this region suggests a major transition tral bioregion lacks clearly distinct species compositions on either side in fauna and assemblage composition. For example, two recent studies of its borders, and as such may not be a reasonable biogeographic de- have identified barriers to gene flow that match this shift in environ- marcation. Due to well-established differences in abiotic conditions as mental conditions (two-band anemonefish, Nanninga et al., 2014; well as habitat structure and geomorphology, we suggest that more Carter's reef sponge, Giles et al., 2015). It may be more appropriate, studies be conducted comparing the fauna between the far southern re- therefore, for a within-Red Sea demarcation of bioregions to be placed gion (below 18° latitude and continuing to the strait of Bab Al Mandab) between the southern end of the Farasan Banks and the beginning of and the rest of the Red Sea including, the western side. While we were the Farasan Islands. In addition, surveys along the Yemeni coast by not able to explore this possibility at the present time, our study pro- Turak et al. (2007) found that the coral assemblages of the northern vides a useful dataset from well-distributed sites along the eastern reefs of Yemen are more similar to the Farasan Islands while the coral Red Sea. Given global challenges associated with “shifting baselines” assemblage in the southern area is likened to the Gulf of Aden, indicat- (Pauly, 1995) and the emerging evidence of overfishing effects in the ing that perhaps the Gulf of Aden bioregion should extend into the Saudi Arabian Red Sea (e.g., Jin et al., 2012; Spaet and Berumen, southern tip of the Red Sea. 2015), some form of recent reference data from this region is needed. M.B. Roberts et al. / Marine Pollution Bulletin 105 (2016) 558–565 565

This dataset could form the basis for later work to investigate finer-scale Holbrook, S.J., Schmitt, R.J., Messmer, V., Brooks, A.J., Srinivasan, M., Munday, P.L., Jones, fi fi G.P., 2015. Reef shes in biodiversity hotspots are at greatest risk from loss of coral relationships between sh, benthos, and abiotic factors to understand species. PLoS One 10, e0124054. the ecological mechanisms driving biogeographic patterns within the Jin, D., Kite-Powell, H., Hoagland, P., Solow, A., 2012. A bioeconomic analysis of traditional Red Sea. fisheries in the Red Sea. Mar. Resour. Econ. 27, 137–148. Jones, G.P., McCormick, M.I., Srinivasan, M., Eagle, J.V., 2004. Coral decline threatens fish biodiversity in marine reserves. Proc. Natl. Acad. Sci. U. S. A. 101, 8251–8253. Acknowledgments Khalaf, M.A., Kochzius, M., 2002. Community structure and biogeography of shore fishes in the Gulf of Aqaba, Red Sea. Helgol. Mar. Res. 55, 252–284. fi This research was supported by a KAUST Competitive Research Khalil, M.T., Cochran, J.E.M., Berumen, M.L., 2013. The abundance of herbivorous sh on an inshore Red Sea reef following a mass coral bleaching event. Environ. Biol. Fish Grant (URF/1/1389-01-01) and baseline research funds to MLB, as 96, 1065–1072. well as WHOI-KAUST Special Partnership Awards (USA-00002 and Klunzinger, C.B., 1870. Synopsis der Fische des Rothen Meeres. I. Theil. Percoiden- – KSA-00011) to SRT. For logistic and fieldwork support, we thank Eric Mugiloiden. Verh. K.-K. Zool.-Bot. Ges. Wien. 20, 669 834. Kruskal, J.B., Wish, M., 1978. Multidimensional Scaling. Sage Publications Inc., Newbury Mason and all the crew from Dream Divers, H. Jahdahli and all the Park, California, USA. KAUST Coastal and Marine Resources Core Lab, Y. Kattan, K.A. Furby, Kulbicki, M., Parravicini, V., Bellwood, D.R., Arias-Gonzalez, E., Chabanet, P., Floeter, S.R., J. Ossolinski, K. Munday, and A.S. Al Kotob. We thank S. Vettori for statis- Friedlander, A., McPherson, J., Myers, R.E., Vigliola, L., Mouillot, D., 2013. Global biogeography of reef fishes: a hierarchical quantitative delineation of regions. PLoS One 8, tical advice. An earlier draft of this manuscript was improved by com- e81847. ments from J. Bouwmeester, B. Bowen, J. Cochran, B. Jones, A. Kattan, Moldenke, A.R., 1975. Niche specialization and species diversity along a California tran- and M. Morton. sect. Oecologia 21, 219–242. Murray, S.P., Johns, W., 1997. Direct observations of seasonal exchange through the Bab el Mandab Strait. Geophys. Res. Lett. 24, 2557–2560. Appendix A. Supplementary data Nanninga, G.B., Saenz-Agudelo, P., Manica, A., Berumen, M.L., 2014. Environmental gradients predict the genetic population structure of a coral reef fish in the Red Sea. Mol. Ecol. 23, 591–602. Supplementary data to this article can be found online at http://dx. Ngugi,D.K.,Antunes,A.,Brune,A.,Stingl,U.,2012.Biogeography of pelagic bacterioplankton doi.org/10.1016/j.marpolbul.2015.11.024. across an antagonistic temperature–salinity gradient in the Red Sea. Mol. Ecol. 21, 388–405. fi References Pauly, D., 1995. Anecdotes and the shifting baseline syndrome of sheries. Trends Ecol. Evol. 10, 430-430. Pielou, E.C., 1975. Ecological Diversity. Wiley, New York. Acker, J., Leptoukh, G., Shen, S., Zhu, T., Kempler, S., 2008. Remotely-sensed chlorophyll a R Core Team, 2014. R: A language and environment for statistical computing. R Founda- observations of the northern Red Sea indicate seasonal variability and influence of tion for Statistical Computing, Vienna, Austria. http://www.R-project.org/. coastal reefs. J. Mar. Syst. 69, 191–204. Raitsos, D.E., Pradhan, Y., Brewin, R.J., Stenchikov, G., Hoteit, I., 2013. Remote sensing the Antonius, A., Riegl, B., 1998. Coral diseases and Drupella cornus invasion in the Red Sea. phytoplankton seasonal succession of the Red Sea. PLoS One 8, e64909. Coral Reefs 17, 48-48. Randall, J.E., 1994. Twenty-two new records of fishes from the Red Sea. Fauna Saudi Arab. Berumen, M.L., Hoey, A.S., Bass, W.H., Bouwmeester, J., Catania, D., Cochran, J.E.M., Khalil, 14, 259–275. M.T., Miyake, S., Mughal, M.R., Spaet, J.L.Y., Saenz-Agudelo, P., 2013. The status of coral Randall, J.E., 1998. Zoogeography of shore fishes of the Indo-Pacific Region. Zool. Stud. 37, reef ecology research in the Red Sea. Coral Reefs 32, 737–748. 227–268. Bowen, B.W., Rocha, L.A., Toonen, R.J., Karl, S.A., the ToBo Laboratory, 2013. The origins of Riegl, B.M., Berumen, M., Bruckner, A., 2013. Coral population trajectories, increased tropical marine biodiversity. Trends Ecol. Evol. 28, 359–366. disturbance and management intervention: a sensitivity analyis. Ecol. Evol. 3, Bray, J.R., Curtis, J.T., 1957. An Ordination of the Upland Forest Communities of Southern 1050–1064. Wisconsin. Ecol. Monogr. 27, 325–349. Riegl, B.M., Bruckner, A.W., Rowlands, G.P., Purkis, S.J., Renaud, P., 2012. Red Sea coral reef Briggs, J.C., 1974. Operation of zoogeographic barriers. Syst. Biol. 23, 248–256. trajectories over 2 decades suggest increasing community homogenization and de- Briggs, J.C., Bowen, B.W., 2012. A realignment of marine biogeographic provinces with cline in coral size. PLoS One 7, e38396. particular reference to fish distributions. J. Biogeogr. 39, 12–30. Righton, D., Kemp, J., Ormond, R., 1996. Biogeography, community structure and diversity Burt, J.A., Feary, D.A., Bauman, A.G., Usseglio, P., Cavalcante, G.H., Sale, P.F., 2011. Biogeo- of Red Sea and Western Indian Ocean butterflyfishes. J. Mar. Biol. Assoc. U. K. 76, graphic patterns of reef fish community structure in the northeastern Arabian Penin- 223–228. sula. ICES J. Mar. Sci. 68, 1875–1883. Roberts, C.M., Dawson-Shepherd, A.R., Ormond, R.F.G., 1992. Large-ScaleVariation in Clarke, K., Warwick, R.M., 2001. Change in marine communities: an approach to statistical assemblage stucture of Red Sea butterflyfishes and angelfishes. J. Biogeogr. 19, analysis and interpretation. 2nd edn. PRIMER-E; Plymouth. 239–250. Clarke, K.R., Gorley, R.N., 2006. PRIMER v6: User Manual/Tutorial, Plymouth. Rüppell, E., 1828. Atlas zur Reise im nördlichen Afrika. Fische des rothen Meeres. Heinrich Cuvier, G.L.C.F.D.V.A. (1828–1850) Histoire Naturelle des Poissons., Paris & Strasbourg. Ludwig Brönner, Frankfurtam Main. DeVantier, L., Pilcher, N., 2000. The status of coral reefs in Saudi Arabia. PERSGA Technical Sales, M., Ballesteros, E., Anderson, M.J., Iveša, L., Cardona, E., 2012. Biogeographical pat- Series Report, Jeddah, pp. 1–45. terns of algal communities in the Mediterranean Sea: Cystoseira crinita-dominated as- DiBattista, J.D., Choat, J.H., Gaither, M.R., Hobbs, J.P., Lozano-Cortés, D.F., Myers, R., Paulay, semblages as a case study. J. Biogeogr. 39, 140–152. G., Rocha, L.A., Toonen, R.J., Westneat, M., Berumen, M.L., 2015. On the origin of Shannon, C.E., Weaver, W., 1963. Mathematical Theory of Communication. University Illi- endemic species in the Red Sea. Journal of Biogeography. http://dx.doi.org/10.1111/ nois Press. jbi.12631 (in press). Sheppard, C., Price, A., Roberts, C., 1992. Marine Ecology of the Arabian Region-Patterns Dornelas, M., Connolly, S.R., Hughes, T.P., 2006. Coral reef diversity refutes the neutral and Processes in Extreme Tropical Environments. Academic Press, Harcourt Brace theory of biodiversity. Nature 440, 80–82. Jovanovich, London, UK. Ehrenberg, C.G., 1834. Beiträge zur physiologischen Kenntnis der Corallenthiere im Sheppard, C.R., Sheppard, A.S., 1991. Corals and coral communities of Arabi. Fauna Saudi Allgemeinen und besonders im rothen Meernebst einem Versuch zur physiologischen Arab. 12, 3–170. Systematik deselben. Abhandlungen der Preußischen Akademie der Wissenschaften, Simpson, E.H., 1949. Measurement of diversity. Nature 163, 688-688. Berlin. Spaet, J.L.Y., Berumen, M.L., 2015. Fish market surveys indicate unsustainable elasmo- Forsskål, P.D., Niebuhr, C., Haas, I., 1775. Amphibiorum, piscium, insectorum, vermium; branch fisheries in the Saudi Arabian Red Sea. Fish. Res. 161, 356–364. quæ in itinere orientali observavit Petrus Forskål. Post mortem auctoris edidit Carsten Spalding, M.D., Fox, H.E., Allen, G.R., Davidson, N., Ferdaña, Z.A., Finlayson, M., Halpern, Niebuhr. Adjuncta Est materia medica Kahirina atque tabula maris rubri geographica. ex B.S., Jorge, M.A., Lombana, A.L., Lourie, S.A., Martin, K.D., McManus, E., Molnar, J., officina Mölleri. Recchia, C.A., Robertson, J., 2007. Marine ecoregions of the world: a bioregionalization Giles, E.C., Saenz-Agudelo, P., Hussey, N.E., Ravasi, T., Berumen, M.L., 2015. Exploring sea- of coastal and shelf areas. Bioscience 57, 573–583. scape genetics and kinship in the reef sponge Stylissa carteri in the Red Sea. Ecology Turak, E., Brodie, J., DeVantier, L., 2007. Reef-building corals and coral communities of the and Evolution 5, pp. 2487–2502. Yemen Red Sea. Fauna Arab. 23, 1–40. Goren, M., Dor, M., 1994. An Updated Check-List of the Fishes of the Red Sea CLOFRES II. Internuniversity Institute for Marine Sciences. Journal of Biogeography (J. Biogeogr.) (2015)

SYNTHESIS A review of contemporary patterns of endemism for shallow water reef fauna in the Red Sea Joseph D. DiBattista1,2*, May B. Roberts1, Jessica Bouwmeester1,3, Brian W. Bowen4, Darren J. Coker1, Diego F. Lozano-Cortes1, J. Howard Choat5, Michelle R. Gaither6, Jean-Paul A. Hobbs2, Maha T. Khalil1, Marc Kochzius7, Robert F. Myers8, Gustav Paulay9, Vanessa S. N. Robitzch1, Pablo Saenz-Agudelo1,10, Eva Salas11,12, Tane H. Sinclair-Taylor1, Robert J. Toonen4, Mark W. Westneat13, Suzanne T. Williams14 and Michael L. Berumen1

1Division of Biological and Environmental ABSTRACT Science and Engineering, Red Sea Research Aim The Red Sea is characterised by a unique fauna and historical periods of Center, King Abdullah University of Science desiccation, hypersalinity and intermittent isolation. The origin and contempo- and Technology, Thuwal 23955, Saudi Arabia, 2Department of Environment and rary composition of reef-associated taxa in this region can illuminate biogeo- Agriculture, Curtin University, Perth, WA graphical principles about vicariance and the establishment (or local 6845, Australia, 3Department of Geology and extirpation) of existing species. Here we aim to: (1) outline the distribution of Carl R. Woese Institute for Genomic Biology, shallow water fauna between the Red Sea and adjacent regions, (2) explore University of Illinois at Urbana-Champaign, mechanisms for maintaining these distributions and (3) propose hypotheses to Urbana, IL 61801, USA, 4Hawai’i Institute of test these mechanisms. Marine Biology, Kane’ohe, HI 96744, USA, Location Red Sea, Gulf of Aden, Arabian Sea, Arabian Gulf and Indian Ocean. 5School of Marine and Tropical Biology, James Cook University, Townsville, Qld 4811, Methods Updated checklists for scleractinian corals, fishes and non-coral Australia, 6School of Biological and invertebrates were used to determine species richness in the Red Sea and the Biomedical Sciences, Durham University, rest of the Arabian Peninsula and assess levels of endemism. Fine-scale diversity Durham DH1 3LE, UK, 7Marine Biology, and abundance of reef fishes within the Red Sea were explored using ecological Vrije Universiteit Brussel (VUB), Brussels survey data. 1050, Belgium, 8Seaclicks/Coral Graphics, Wellington, FL 33411, USA, 9Florida Results Within the Red Sea, we recorded 346 zooxanthellate and azooxanthel- Museum of Natural History, University of late scleractinian coral species of which 19 are endemic (5.5%). Currently 635 Florida, Gainesville, FL 32611-7800, USA, species of polychaetes, 211 echinoderms and 79 ascidians have been docu- 10Instituto de Ciencias Ambientales y mented, with endemism rates of 12.6%, 8.1% and 16.5% respectively. A pre- Evolutivas, Universidad Austral de Chile, liminary compilation of 231 species of crustaceans and 137 species of molluscs Valdivia 5090000, Chile, 11Section of include 10.0% and 6.6% endemism respectively. We documented 1071 shallow Ichthyology, California Academy of Sciences, fish species, with 12.9% endemic in the entire Red Sea and 14.1% endemic in San Francisco, CA 94118, USA, 12Department the Red Sea and Gulf of Aden. Based on ecological survey data of endemic of Ecology and Evolutionary Biology, fishes, there were no major changes in species richness or abundance across University of California, Santa Cruz, CA 1100 km of Saudi Arabian coastline. 95064, USA, 13Department of Organismal Biology and Anatomy, University of Chicago, Main conclusions The Red Sea biota appears resilient to major environmen- Chicago, IL 60637, USA, 14Life Sciences tal fluctuations and is characterized by high rates of endemism with variable Department, Natural History Museum, degrees of incursion into the Gulf of Aden. The nearby Omani and Arabian London SW7 5BD, UK Gulfs also have variable environments and high levels of endemism, but these are not consistently distinct across taxa. The presence of physical barriers does not appear to explain species distributions, which are more likely determined by ecological plasticity and genetic diversity. *Correspondence: Joseph D. DiBattista, Department of Environment and Agriculture, Keywords Curtin University, PO Box U1987, Perth, WA Arabian Peninsula, biodiversity, biogeographical barriers, centre of endemism, 6845, Australia. E-mail: [email protected] coral reef, ecological processes, faunal checklist, marine biogeography

ª 2015 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/jbi 1 doi:10.1111/jbi.12649 J. D. DiBattista et al.

persal, a hypothesis supported by the disjunct distribution of INTRODUCTION some reef fish species (Roberts et al., 1992) and coral genera Biogeographical regions with exceptional taxonomic diversity (F. Benzoni, pers. comm.), as well as genetic differentiation and high levels of endemism are known as biodiversity hot- between populations of coral reef organisms (Froukh & spots and by definition are high conservation priorities Kochzius, 2008; Nanninga et al., 2014; Giles et al., 2015; but (Myers et al., 2000). These hotspots support a disproportion- see Robitzch et al., 2015). ally high percentage of biodiversity including unique species The extent of environmental change within the Red Sea and evolutionary novelty. While the Indo-Malay Archipelago and its effects on shallow water fauna remain controversial. (i.e. Coral Triangle) is the centre of species richness for many Some authors believe that hypersaline conditions, compara- coral reef organisms (Briggs, 2005; Hoeksema, 2007; Veron ble to the present-day Dead Sea (Braithwaite, 1987), extir- et al., 2009), endemism hotspots, as expressed in percentage pated most marine life during glacial maxima (Sheppard of unique fauna, tend to occur in isolated or peripheral et al., 1992), whereas others suggest survival of a decimated regions (Hughes et al., 2002; Roberts et al., 2002). For Indo- fauna within isolated refugia (Goren, 1986; Klausewitz, 1989; Pacific reef fishes, the highest endemism can be found in the Rohling et al., 1998). Hawaiian Islands, Easter Island, Marquesas Islands, Mas- carene Islands and the Red Sea (Mora et al., 2003; Allen, Data limitation in the Red Sea and Arabian 2008; Briggs & Bowen, 2012; Kulbicki et al., 2013). Recent Peninsula research has also demonstrated the importance of peripheral regions, such as the Hawaiian Archipelago, the Marquesas The first step towards understanding the patterns of biodi- Islands and the Red Sea in exporting unique genetic lineages versity is to obtain accurate species checklists and distribu- to other regions (Malay & Paulay, 2010; Eble et al., 2011; tion maps. The seminal works of taxonomists such as Gaither et al., 2011; Bowen et al., 2013; DiBattista et al., Forsskal, Cuvier & Valenciennes, Ruppell,€ Ehrenberg, Heller 2013). and Klunzinger led to the recognition of the Red Sea as a biodiversity hotspot for marine fauna (see Fig. 1). Modern efforts to understand biogeographical processes began with The Red Sea Ekman (1953) and Briggs (1974), both of whom recognised The Red Sea extends 2270 km from 30° N in the Gulf of the Red Sea as an endemism hotspot. Subsequent studies Suez to 12° N in the Gulf of Aden. Based on existing check- have been hindered by a dearth of geographical range infor- lists, 320 zooxanthellate scleractinian corals (Veron et al., mation (Berumen et al., 2013), but recent academic invest- 2009) and 1078 fish species (Golani & Bogorodsky, 2010) ments by several countries that border the Red Sea (Mervis, have been identified in this region, although these values are 2009) has enhanced accessibility and integration of molecular constantly being redefined. The Red Sea harbours one of the and morphological research. highest levels of endemism for marine organisms, with 14% Here we define shallow water (< 200 m) species distribu- for fishes (Randall, 1994), 15% for crabs (Guinot, 1966), up tion patterns of the contemporary Red Sea fauna and com- to 17% for echinoderms (Price, 1982; Campbell, 1987; Dafni, pare these with the rest of the Arabian Peninsula and greater 2008) and as much as 10% for scleractinian corals (Hughes Indian Ocean. Our goals include: (1) outline the distribution et al., 2002). Endemism is even higher for some conspicuous of faunal composition in the Red Sea and adjacent regions, taxa, for example reaching 50% in butterflyfishes (e.g. (2) explore mechanisms for maintaining these distributions Roberts et al., 1992). This endemic region extends to the and (3) propose working hypotheses to test these mecha- Gulf of Aden for many species, and to Oman or Socotra for nisms. fewer species (Winterbottom, 1985; Randall, 1995; Kemp, 1998, 2000; Zajonz et al., 2000). MATERIALS AND METHODS The unique fauna of the Red Sea is coupled with a turbu- lent geological history and unusual environmental conditions, Databases were created from existing checklists for zooxan- including minimal freshwater inflow, high rates of evapora- thellate and, when available, azooxantellate scleractinian cor- tion and latitudinal gradients in environmental variables als (see Appendix S1 for checklist and references), fishes (see (temperature, salinity and nutrients) and a narrow (18 km) Appendix S2 for checklist and references) and non-coral and shallow (137 m) connection with the Indian Ocean at invertebrate species (annelids, arthropods, echinoderms, tuni- the Strait of Bab al Mandab. Water exchange between the cates and molluscs; see Appendix S3 for checklist and refer- Red Sea and Indian Ocean was repeatedly restricted during ences). Pleistocene glacial cycles when sea level lowered as much as Species names for corals and non-coral invertebrates were 140 m (Braithwaite, 1987; Rohling et al., 1998). Isolation of confirmed in the World Register of Marine Species [WoRMS the Red Sea fauna is probably reinforced by cold-water Editorial Board (2014), available from http://www.marine- upwelling off the north-east African and southern Arabian species.org at VLIZ, accessed 2014-09-01]. Fish names were coasts (Smeed, 1997; Kemp, 2000). A turbid-water region confirmed using the Catalog of Fishes (Eschmeyer, 2013) and south of 19–20° N in the Red Sea may also limit larval dis- FishBase (Froese & Pauly, 2014). For corals, we excluded

2 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Figure 1 Number of valid Red Sea endemic scleractinian coral (N = 19), fish (N = 138) or non-coral invertebrate (N = 91) species described from 1741 to 2014 with seminal works noted. reports of nomina nuda and dubia species. All checklists were For non-coral invertebrates we focused on taxa that have updated with recent taxonomic revisions where possible. been studied recently as part of King Abdullah University of For fishes, only those recorded at depths less than 200 m were Science and Technology (KAUST) biodiversity surveys (see included in the checklist (see Appendix S2). We also exclude Appendix S3). Within the crustaceans and molluscs, we waifs, non-neritic pelagic and mesopelagic species that vertically selected families and genera that are well known; polychaetes, migrate to the surface at night, Lessepsian migrants from the echinoderms and ascidians were treated in their entirety. Mediterranean (see Bernardi et al., 2010), as well as most cases of Records of non-coral invertebrates are updated with fishes not identified to species. We include un-named species taxonomic literature, the WoRMS database and our collec- that are clearly identified and await formal description. tions (see Appendix S3 for references).

Journal of Biogeography 3 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Given our interest in Red Sea endemism, we compiled spe- Gulf (see Fig. 2). In cases where data are insufficient to sepa- cies presence–absence records from the seven Marine Ecore- rate the Gulf of Aqaba or Socotra into MEOWs, they were gions of the World (MEOWs) bordering the Arabian combined with the Red Sea or Gulf of Aden respectively. Taxo- Peninsula (modified from Spalding et al., 2007): (1) Gulf of nomic groups that are data-deficient for an entire MEOW are Aqaba, (2) Red Sea, (3) Gulf of Aden, (4) Socotra, (5) South- omitted from regional consideration. The MEOW results are ern Oman, (6) Gulf of Oman and Pakistan and (7) Arabian visualised using Arcgis 10.2 (ESRI, 2014).

Figure 2 Species richness and level of endemism (%) for (a) scleractinian corals, (b) fish, (c) annelids, (d) arthropods, (e) echinoderms, (f) tunicates and (g) molluscs within each of the seven Marine Ecoregions of the World (MEOWs) bordering the Arabian Peninsula (modified from Spalding et al., 2007): (1) Gulf of Aqaba, (2) the Red Sea; (3) Gulf of Aden, (4) Socotra, (5) Southern Oman; (6) Gulf of Oman and Pakistan and (7) Arabian Gulf. In cases where there is no data for a MEOW, the region was coloured white; MEOWs coloured grey have zero values. In cases where data were insufficient to separate the Gulf of Aqaba and Socotra MEOW sub- regions, they were assigned the same colour as their primary MEOW in the Red Sea or Gulf of Aden respectively.

4 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Figure 2 (Continued).

For this review, we define endemism at multiple scales 2008 and 2011. Reefs were grouped into 10 regions from Al using the following terminology: (1) Red Sea endemic: a spe- Wajh (26.8° N) to Ablo (18.6° N). Four replicate belt transects cies only inhabiting the Red Sea (including the Gulf of were made at each of four depths between the reef crest and Aqaba), (2) Red Sea to Gulf of Aden endemic: a species only 10 m. Each belt transect was 50 m 9 4 m with the exception found in the Red Sea and Gulf of Aden (including Socotra) of smaller species (e.g. damselfishes and blennies), which were and (3) Red Sea resident: a species inhabiting the Red Sea surveyed on a 50 9 1 or 2 m transect. One-way ANOVA was but also in regions outside the Red Sea and Gulf of Aden used to resolve latitudinal trends in mean species richness of (i.e. widespread species). For the purposes of the heat maps endemics. Total abundance of these fishes was summed per and discussion, we also estimated the level of endemism for reef and fourth root transformed to balance the effect of very each MEOW individually. As reef fish have been well studied abundant species, such as Chromis dimidiata. All statistical compared to invertebrates, we use survey data from select analyses use the vegan package in R (Oksanen et al., 2014). reef fishes to test patterns of biodiversity and endemism within the Red Sea. These analyses allowed us to more RESULTS broadly assess the role of environmental gradients as barriers to dispersal in the region. Red Sea endemism based on MEOWs of the Arabian Peninsula Ecological survey of fish densities Scleractinian corals Based on reef fish densities (M. Roberts, unpub. data), we assessed the abundance of 33 Red Sea endemics on 45 reefs The Red Sea hosts 346 zooxanthellate and azooxanthellate across 1100 km of Saudi Arabian coastline surveyed between scleractinian coral species, of which 19 are endemic (5.5%;

Journal of Biogeography 5 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Fig. 2a). Within the Red Sea, 307 species were found in the tion of decapods shows 231 Red Sea species, including 23 north/central region and 240 species were found in the (10.0%) endemic to the Red Sea and 31 (13.4%) endemic to southern region. the Red Sea to Gulf of Aden. Although the Red Sea mollus- For comparison, Veron et al. (2009) recorded 289 zooxan- can fauna is represented in museum collections and field thellate coral species in the north/central region and higher guides, sampling of the Arabian Peninsula and northern richness in the southern region with 297 species based on Somalia is limited, and does not allow us to assess Red Sea similar boundaries. Additionally, Hughes et al. (2002) endemism with confidence. Based on the molluscs consid- recognise 10% endemism in the Red Sea versus the 5.5% ered here, we predict 6.0% Red Sea endemism among species endemism identified in our study. The Arabian Gulf hosts 66 that occur within the Arabian Peninsula region. This figure scleractinian coral species and 126 species are recorded in would be higher if the Gulf of Aden were treated in the same the Gulf of Oman. Finally, 95 species are found in the Gulf biogeographical unit as the Red Sea. Exacerbating this lack of of Aden and the Arabian Sea, and 228 species, including one general knowledge is the prevalence of cryptic species among endemic species (0.4% endemism), are found at Socotra. In marine invertebrates, especially in groups that do not use total, 394 scleractinian coral species were recorded in the visual systems for mate recognition (Knowlton, 1993). Inte- Arabian Peninsula (see Appendix S1). grative studies that include field and genetic approaches consistently reveal higher levels of endemism. For example, 36 species (38%) of sea cucumbers from the Red Sea to Gulf of Fish Aden are endemic to the area based on DNA barcodes (G. The Red Sea hosts 1071 recorded fish species (vs. 1760 in the Paulay, unpub. data). For molluscs, molecular data have entire Arabian Peninsula region) of which 138 (12.9%) are identified new species (e.g. nudibranchs; Jorger€ et al., 2012) endemic to the Red Sea and 189 (14.1%) are endemic to the and the resurrection of a historically described species Red Sea and Gulf of Aden (Fig. 2b). Only 1.0%, 1.7% and (Huber & Eschner, 2011). 2.2% of Red Sea fishes have ranges extending to southern Oman, the Gulf of Oman or the Arabian Gulf respectively, Red Sea endemism for reef fish but no further. By comparison, Eschmeyer et al. (2010) recorded 1188 Red Sea fish species, including 159 endemics, Among reef fishes, the proportion of Red Sea endemics per resulting in a comparable endemism rate of 13.6%. Goren & family varies from 0% to 100%. The 14 families with > 50% Dor (1994) listed 1248 species from the Red Sea. Both of these endemism have seven or fewer Red Sea species. Among fami- estimates, however, include all fish species as opposed to our lies with 10 or more Red Sea species, five of these have ende- stricter criteria, and may include unverifiable records for the mism values > 25% (Callionymidae, Pseudochromidae, latter. Similar to Fricke et al. (2014), we note that some of the Tripterygiidae, Monacanthidae and Tetraodontidae). When endemic fish fauna are restricted to the Gulf of Aqaba (4.1%). we consider the Red Sea and Gulf of Aden combined, this This indicates an effective ecological barrier separating the value increases for several families or sub-families including Gulf of Aqaba from the rest of the Red Sea (also see Klause- the Pseudochromidae (from 33.3% to 64.3%), Apogonidae witz, 1989; Sheppard et al., 1992), possibly due to higher sali- (15.3 to 25.3%) and Scarinae (11.1 to 32.0%). Endemism is nity or occasional upwelling in this region. This pattern may apparent for the Chaetodontidae only when the Red Sea and also be explained by sampling bias because 87.5% of the Gulf Gulf of Aden region are considered together (0 to 12.0%, of Aqaba endemics are from a single collection. We also note but 32.0% for the entire Arabian Peninsula region), which that even though the Gulf of Aden and Socotra are not con- contradicts the 50% endemism reported in Roberts et al. sidered centres of endemism (0.7% and 1.4% respectively), (1992). The variable proportion of endemic species across the former has the second highest level of species richness in taxonomic groups indicates that the evolutionary processes the study (Fig. 2b), and the latter appears to be a hotspot for have affected groups of reef fish differently. These results the mixing of Red Sea and Indian Ocean fauna (see DiBattista must be interpreted with caution given that presence–absence et al., 2015). data may be biased for highly dispersive species that appear in locations where they are functionally absent. Non-coral invertebrates Reef fish density data Echinoderms are among the best studied invertebrates, with 211 species recorded from the Red Sea. Of these species, 17 Based on 33 Red Sea to Gulf of Aden endemic reef fish species, (8.1%) are known only from the Red Sea and 21 (10.0%) there are no major changes in species richness or abundance from the Red Sea to Gulf of Aden. Currently 79 ascidian spe- among 10 sub-regions (Figs 3 & 4). One-way ANOVA analyses cies are documented from the Red Sea, with 13 (16.5%) reveal no consistent significant change with latitude or direc- endemic, although the rest of the Arabian Peninsula remains tion across our survey area. Indeed, of 99 comparisons understudied. Among 635 polychaete species recorded from between northern, central and southern regions for all species, the Red Sea, 80 (12.6%) are endemic and 92 (14.5%) are only 16 were significant at P < 0.05. This trend is most appar- Red Sea to Gulf of Aden endemics. An incomplete compila- ent in all numerically dominant species (e.g. Chromis dimidi-

6 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Figure 3 Mean species richness of endemic fishes from (a) the Red Sea based on a maximum of 33 conspicuous species, estimated from sites within the Red Sea from latitude 26.8° N (Al Wajh) to 18.6° N (Ablo). In most cases, there were four reefs surveyed in each of the 10 regions, exceptions include Thuwal (five reefs) and Al Lith (eight reefs). North, central and southern Red Sea partitions defined as Al Wajh to the Seven Sisters (26.8° N to 23.8° N), Rabigh to Jeddah (22.8° N to 21.8° N) and Al Lith to Ablo (19.9° Nto 18.6° N) are shaded light red, light blue and light green respectively. Black horizontal bars on the box plot (b) represent the median of each group. Upper and lower bounds of the boxes represent the 75th and 25th percentiles respectively. Vertical lines extend to the 95th (upper line) and 5th (lower line) percentiles.

ata, Thalassoma rueppellii, Pseudochromis fridmani). Such find- (Fig. 2). The pattern is more complicated for the non-coral ings contradict previous evidence for biogeographical barriers invertebrates, with a trend of highest diversity and endemism in the central Red Sea (Khalaf & Kochzius, 2002; Spalding in the Red Sea, Gulf of Oman and Arabian Gulf (Fig. 2). et al., 2007; but see Kulbicki et al., 2013). This central delin- This confirms the status of the Red Sea as a significant eation may instead represent an ‘average’ boundary for many region of endemism for coral reef biota at the western of the species that show distributional shifts. periphery of the Indo-Pacific. Reef fishes provide the most complete information for investigating the processes that DISCUSSION underlie patterns of endemism. Two features dominate the biogeography of Red Sea reefs. The Red Sea hosts a distinct coral reef fauna with consis- Firstly, the biota has persisted through major environmental tently high endemism for shallow water organisms (> 10% alterations, especially with respect to temperature and salinity in fishes, annelids, arthropods and chordates). Looking (DiBattista et al., 2013). Episodic restrictions of the Strait of across the region, levels of both biodiversity and endemism Bab al Mandab during the Pleistocene produced an environ- are highest in the Red Sea for fishes and scleractinian corals ment that was very different from contemporary conditions,

Journal of Biogeography 7 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Figure 4 Fourth-root transformed average abundance of Red Sea to Gulf of Aden endemic reef species (N = 33) along with standard deviation in the north, central and southern Red Sea partitions defined as Al Wajh to the Seven Sisters (26.8° N to 23.8° N), Rabigh to Jeddah (22.8° N to 21.8° N) and Al Lith to Ablo (19.9° N to 18.6° N) respectively. Bars within the figure were derived from average abundances among transects (area dependent on fish species and thus corrected for) within reefs for each of the three partitions. All fish species have been ordered most to least abundant (with reference to the North partition) and further grouped by family in taxonomic order [Chaetodontidae (red), Pomacentridae (orange), Labridae (yellow), Scaridae (green), Pseudochromidae (teal) and all others considered (blue)]. which in turn would eliminate or extirpate many species nor the presence of physical barriers likely define the distri- throughout the Red Sea. Indeed, we did not detect any dif- butions of reef fishes in the Red Sea. We consider the details ferences in species richness or community composition of of these issues below. the endemic reef fishes across the Red Sea based on our survey data. Secondly, the Red Sea biota are not confined by What are the processes maintaining putative barriers consistent geographical boundaries, with some endemics to dispersal in the Red Sea? penetrating varying distances into the Gulf of Aden and northern Arabian Sea. Pelagic larval duration (PLD) does Environmental gradients not appear to be an important determinant of geographical range size in most instances (Victor & Wellington, 2000; Les- The contemporary Red Sea is a spatially heterogeneous ter & Ruttenberg, 2005; Macpherson & Raventos, 2006; Luiz ecosystem based on gradients in salinity (range: 35–41 ppt), et al., 2013), especially for peripheral regions such as the East temperature (range: 21–34 °C) and primary productivity À Pacific (Robertson, 2001; Zapata & Herron, 2002; Lessios & [Chlorophyll a (chl-a) range: 0.5–4.0 mg m 3] (Sofianos, Robertson, 2006). Thus, neither differences in larval duration 2003; Raitsos et al., 2013) from north to south. Besides spa-

8 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism tial variation there are seasonal differences among regions. tion of recently settled recruits, juveniles and adults. Moni- Temperature variation in the northern (20–30 °N) and toring survivorship of recruits should be included because southern (12–16 °N) Red Sea is much higher (annual range traits that increase survivorship appear important in promot- ~10 °C) than in the central Red Sea (annual range ~5 °C). ing persistence following range extensions (Luiz et al., 2013). Salinity in the Gulf of Suez and Gulf Aqaba also have higher annual ranges (2–4 ppt) than the rest of the Red Sea (< 1 Available resources and recruitment ppt). The oligotrophic waters of the north (chl-a range: 0.1– À 0.35 mg m 3) contrast with the eutrophic waters in the Recruitment failure is a potentially important driver of the south, which vary considerably (chl-a range: 0.5– localised distribution and abundance patterns of Red Sea or À 5.0 mg m 3) due to seasonal influx of nutrient-rich waters regional endemics. For example, distributions may be from the Gulf of Aden. extremely localised in Gulf of Aden and Oman endemic Reef fish species richness, abundance and composition parrotfishes, such as Scarus arabicus and Scarus zufar, appear to be evenly distributed across eight degrees of lati- whereas other regional endemics (Scarus ferrugineus) extend tude and 1100 km of Saudi coastline (Figs 3 & 4), spanning through the entire environmental gradient of the Red Sea a gradient with significant temporal and spatial variation in and northern Arabian Sea (Choat et al., 2012). It is unlikely the physical environment. We lack data, however, from the that dispersal capacity is the limiting factor in these species Gulf of Aqaba in the far north (but see Khalaf & Kochzius, distributions. Testing of recruitment failure hypotheses 2002), and more critically from the Farasan Islands (Saudi requires a capacity to identify recruitment habitats and the Arabia into Yemen) in the far south (Fig. 3). The Farasan age structure and condition of endemic species over their Islands are characterised by shallow sand banks, sparsely dis- distributional range. Genomic and stable isotope analyses tributed reef and eutrophic conditions compared to the slop- provide options to resolve ontogenetic interactions between ing, oligotrophic reefs for the rest of the Red Sea (Sheppard the relevant species and suitable habitats. & Sheppard, 1991; Raitsos et al., 2013). Central and southern Red Sea regions in this study did, however, support a few Phylogenetic community structure in the Red Sea species not recorded from the northern region (Fig. 4). This agrees with previous work that shows some species, such as Phylogenetic hypotheses are now available for a wide range the damselfish Neopomacentrus miryae and the wrasse of reef organisms, including endemic and more widespread Paracheilinus octotaenia are abundant in the northern Red species that occur in the Red Sea (e.g. Fessler & Westneat, Sea, but virtually absent in the southern part (Ormond & 2007). Exploring patterns of phylogenetic community assem- Edwards, 1987; Sheppard et al., 1992; also see Winterbottom, bly at multiple scales (Kooyman et al., 2011) will resolve the 1985). The unique environmental features of the Farasan role of environmental filtering, competition and specific Islands in the far south suggest that fish communities there climatic factors in shaping Red Sea coral reef ecosystems. would also differ from the assemblages to the north and Several families of Red Sea reef fishes are ideal for phylo- should be a focal point for further study. genetic community assembly analysis, including the wrasses and parrotfishes (Labridae), damselfishes (Pomacentridae) and butterflyfishes (Chaetodontidae) (Westneat & Alfaro, Species specific differences in dispersal and colonisation 2005; Fessler & Westneat, 2007; Cooper et al., 2009; Cow- Robertson (2001) found that endemic reef fishes could not man et al., 2009; Choat et al., 2012; Hodge et al., 2014; be differentiated by PLD estimates from similar species with DiBattista et al., in press). The first step is to examine phylo- broad distributions. The conclusion that PLD values are not genetic dispersion of Red Sea reef fishes on their respective reliable indicators of range size is further supported by analy- trees and then examine phylogenetic distance among mem- ses of reef fish taxa with very different larval dispersal char- bers of the community. The endemic Red Sea species appear acteristics that traverse the vast Eastern Pacific Barrier to be derived from many different parts of their family trees, (> 6000 km) in both directions (Lessios & Robertson, 2006). indicating that the factors driving Red Sea endemism impact PLDs as a basis for estimating the dispersal potential in coral multiple clades with different ecologies. Measures of phyloge- reef fishes is also the subject of ongoing debate (Riginos netic under- and over-dispersion can reveal patterns of fau- et al., 2011; Selkoe & Toonen, 2011). nal exchange with the Indian Ocean and the timing of Red To test the hypothesis that dispersal limitation is not driv- Sea endemism among multiple reef organisms (see Hodge ing small range sizes in the Red Sea endemics, larval input et al., 2014). could be quantified in adjacent but divergent environments. This could be tested with light traps, crest nets, the In-Situ Physical barriers to dispersal: One theory to define species Ichthyoplankton Imaging System (ISIIS; Cowen & Guigand, distributions and gene flow 2008) for fish larvae or settlement plates (and complimentary genetics) for corals and non-coral invertebrates (e.g. Plai- Physical barriers to dispersal of marine biota are less evident sance et al., 2011). These methods should be accompanied than among terrestrial ecosystems (Mayr, 1954). In terms of by visual surveys to document the abundance and distribu- habitat patchiness, both the eastern and western coasts of the

Journal of Biogeography 9 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Red Sea are lined with continuous fringing coral reefs from (Acanthaster planci: Vogler et al., 2008; Panulirus pennicila- north to south. Also, the Red Sea is quite narrow, only tus: Iacchei et al., in press; Pocillopora spp., Pinzon et al., 234 km at its widest point, and so this may enhance disper- 2013; Scylla serrata: Fratini & Vannini, 2002; Tridacna max- sal across the deep open centre, which is possibly an effective ima: Nuryanto & Kochzius, 2009; Holothuroids: G. Paulay, barrier only to shallow benthic species with limited dispersal unpub. data) support a genetic distinction of Red Sea versus (Leese et al., 2008; Munday et al., 2009). Indian Ocean populations. These combined results indicate Another physical barrier, albeit ephemeral in nature, is sustained isolation between the Red Sea and WIO popula- the shallow (137 m) Strait of Bab al Mandab in the south, tions for several hundred thousand years. The diversity of which reduces water exchange between the Red Sea and the outcomes is also likely a reflection of taxonomic differences Indian Ocean during glacial maxima (Rohling et al., 1998; in life histories and habitat requirements that have evolved Siddall et al., 2003; Bailey, 2009). This historical barrier may over millions of years. be responsible for some of the observed endemics, although the origination of several reef fish taxa (and their coral reef CONCLUSION AND FUTURE PERSPECTIVES hosts) in the Red Sea pre-dates the Pleistocene (Benzie, 1999; Choat et al., 2012; Duchene et al., 2013; Hodge et al., Since 2000, 58 new endemic species have been described in 2014). the Red Sea indicating that a vast gap remains between rec- Within the Red Sea, the narrow (6 km) and shallow (242– ognised taxonomy and existing biodiversity (Table 1). Most 270 m) Straits of Tiran between the Gulf of Aqaba and Red descriptions are based on morphological data highlighting Sea proper can also act as a physical barrier. The deep and the need for increased sampling in understudied regions of narrow fjord-like Gulf of Aqaba is 180 km long and is the Red Sea (i.e. along its western shores and the far south) 25 km at its widest point, and the depth can reach 1800 m where new species await discovery. Recent genetic tools add but averages 800 m. Hot and dry deserts flank the semi- momentum to the discovery of cryptic species, which can be enclosed basin, which result in a high evaporation rate, high very difficult to distinguish based on morphological charac- salinity (41 ppt) and a thermohaline circulation that drives ters (Knowlton, 1993; Bickford et al., 2007), leading to water exchange with the Red Sea (Reiss & Hottinger, 1984). underestimates of species diversity. Exceptional examples are Water residence time in the upper 300 m of the Gulf of seen in primitive bony fishes such as the round herrings Aqaba varies from only a few months up to two years. The (genus Etrumeus). Recent morphological and phylogenetic amount of Red Sea water reaching the northern tip of the studies reveal seven divergent mitochondrial lineages within Gulf of Aqaba is therefore estimated to be only 1% of that at a single putative species (DiBattista et al., 2012, 2014; Ran- the Straits of Tiran (Wolf-Vecht et al., 1992). dall & DiBattista, 2012), most of which are distributed in Genetics provides one way to examine connectivity and allopatry, and all of which are now described as distinct spe- effective barriers (e.g. Baums et al., 2006; Crandall et al., cies. Phylogenetic frameworks are also fruitful testing 2012; Liggins et al., 2013; Selkoe et al., 2014). Most genetic grounds for biogeographical hypotheses by relating differ- connectivity studies have focused on broad-scale compar- ences in life history, ecology, physiology and behaviour isons between the Red Sea and greater Indo-Pacific. For among closely (and more distantly) related species (see example, Froukh & Kochzius (2008) identified a genetic par- DiBattista et al., in press). tition in the damselfish Chromis viridis between the Red Sea Molecular tools are proving useful for the identification of and Indonesia, whereas studies on lionfish (Pterois spp.) cryptic lineages in endemism hotspots such as Hawai’i (Ran- using similar mtDNA sequence methods find no difference dall et al., 2011), the Marquesas Islands (Szabo et al., 2014) (Kochzius et al., 2003; Kochzius & Blohm, 2005). A study on and the Red Sea (Randall & DiBattista, 2013). In the Red mtDNA sequence divergence between fishes from the Red Sea, only 10% of the new species descriptions listed in Sea and Japan revealed high divergences for Apogon cyano- Table 1 were corroborated using molecular data, but this soma, Gerres oyena, Sargocentron rubrum, Spratelloides delicat- includes three new species of fish (DiBattista et al., 2012; ulus and Terapon jarbua (5.8% to 18.8%), possibly indicating Herler et al., 2013; Koeda et al., 2014) and a coral (Terraneo cryptic species (Tikochinski et al., 2013). The Indo-Pacific et al., 2014). As molecular tools are applied more broadly, damselfish Dascyllus aruanus demonstrated differentiation endemism in the region will continue to rise. But certainly between Red Sea and Western Indian Ocean (WIO) samples broadscale sampling is necessary to discover the cryptic evo- based on mtDNA and microsatellite markers (Liu et al., lutionary lineages hidden in species previously thought to be 2014). However, the goldband goatfish, Upeneus mollucensis, widespread (e.g. Williams & Reid, 2004; Vogler et al., 2008; did not show any mtDNA differentiation on this scale Williams et al., 2011, 2012; Hoareau et al., 2013; Postaire (Tikochinski et al., 2013). Another recent study of Red Sea et al., 2014). Undersampled areas include the Andaman Sea, resident reef fish showed a range of connectivity with the Bangladesh, India, Myanmar, Somalia and much of the Red WIO, from species with almost no differentiation (Halicho- Sea (particularly Eritrea and Yemen). This deficiency is partly eres hortulanus and Lutjanus kasmira) to species with ancient due to long-term political instability, although some regions genetic separations (Neoniphon sammara and Pygoplites dia- like Western Australia are politically stable but undersampled canthus) (DiBattista et al., 2013). Research on invertebrates (e.g. Poore et al., 2014).

10 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Table 1 Valid scleractinian coral, fish and non-coral Table 1 Continued invertebrate endemic species described in the Red Sea from 2000 Species Taxonomic authority to 2014. Montipora pachytuberculata Veron, DeVantier & Turak, 2000 Species Taxonomic authority Montipora saudii Turak, DeVantier & Veron, 2000 Vertebrata Pachyseris inattesa Benzoni & Terraneo, 2014 Acanthoplesiops cappuccino Gill, Bogorodsky & Mal, 2013 Mollusca Adelotremus leptus Smith-Vaniz & Rose, 2012 Turbo (Aspilaturbo) Kreipl & Alf, 2001 Amblyeleotris neglecta Jaafar & Randall, 2009 marisrubri Aseraggodes kruppi Randall, Bogorodsky & Mal, 2013 Tunicata Aseraggodes macronasus Randall, Bogorodsky & Mal, 2013 Boltenia yossiloya Shenkar & Lambert, 2010 Bryaninops spongicolus Suzuki, Bogorodsky & Randall, 2012 Botryllus eilatensis Shenkar & Monniot, 2006 Cabillus nigrostigmus Kovacic & Bogorodsky, 2013 Enneapterygius qirmiz Holleman & Bogorodsky, 2012 Entomacrodus solus Williams & Bogorodsky, 2010 Based on the regional picture of endemism and the under- Etrumeus golanii DiBattista et al., 2012 lying processes that produce them, a primary question is Eviota geminata Greenfield & Bogorodsky, 2014 what prevents Red Sea endemics from spreading eastward. Eviota oculopiperita Greenfield & Bogorodsky, 2014 Indeed, the Red Sea is in contact with the Arabian Sea and Evoxymetopon moricheni Fricke, Golani & Appelbaum-Golani, WIO through the Gulf of Aden. It is unlikely that Red Sea 2014 and regional endemics are confined to particular areas due Gobiodon ater Herler et al., 2013 Gobiodon bilineatus Herler et al., 2013 to either physiological constraints or a limited dispersal Gymnapogon melanogaster Gon & Golani, 2002 capacity. Moreover, the Red Sea reef biota have been and are Gymnothorax baranesi Smith, Brokovich & Einbinder, 2008 currently subject to a demanding and highly variable envi- Gymnoxenisthmus tigrellus Gill, Bogorodsky & Mal, 2014 ronment. A number of taxa display an abrupt southern Heteroeleotris dorsovittata Kovacic, Bogorodsky & Mal, 2014 boundary to their distribution extending only to the Strait of Heteroeleotris psammophila Kovacic & Bogorodsky, 2014 Bab al Mandab, whereas others extend beyond the Gulf of Hippocampus debelius Gomon & Kuiter, 2009 Aden to the northern coast of Oman. In this sense the Hypoatherina golanii Sasaki & Kimura, 2012 Limnichthys marisrubri Fricke & Golani, 2012 southern boundary of the Red Sea is selectively porous, Opisthognathus dipharus Smith-Vaniz, 2010 allowing some species to establish populations in the differ- Pempheris tominagai Koeda, Yoshino & Tachihara, 2014 ent reef environments of the northern Arabian Sea. Both Red Pseudamiops springeri Gon, Bogorodsky & Mal, 2013 Sea and Omani reef environments are highly variable, and Pteragogus clarkae Randall, 2013 for this reason, environmental variation per se in the Gulf of Pteragogus trispilus Randall, 2013 Aden is unlikely to constitute a distributional barrier; rather Siphamia goreni Gon & Allen, 2012 ecological factors may dominate. Soleichthys dori Randall & Munroe, 2008 Symphysanodon disii Khalaf & Krupp, 2008 The geological history and differences in oceanographic Tomiyamichthys dorsostigma Bogorodsky, Kovacic & Randall, regime between the Red Sea, Gulf of Aden, Oman and the 2011 Arabian Gulf have resulted in very different reef ecosystems. Upeneus davidaromi Golani, 2001 This spectrum ranges from sites in the Red Sea dominated Uranoscopus rosette Randall & Arnold, 2012 by corals that have evolved in high temperature and rela- Vanderhorstia opercularis Randall, 2007 tively clear water environments to rocky reefs dominated by Annelida upwelling episodes in the Gulf of Aden and northern Ara- Harmothoe marerubrum Wehe, 2006 Lepidonotus polae Wehe, 2006 bian Sea. Some reef fish taxa, for example, with very different Parahalosydnopsis arabica Wehe, 2006 larval characteristics (e.g. Acanthurus sohal and Scarus ferrug- Arthropoda ineus) are able to extend beyond the Red Sea while others Charybdis omanensis Turkay€ & Spiridonov, 2006 (e.g. Acanthurus gahhm and Chlorurus gibbus) remain septentrionalis restricted to the north of Bab al Mandab. This suggests a € Ethusa thieli Spiridonov & Turkay, 2007 taxon specific capacity to recruit to the distinctive reef sys- Petrolisthes aegyptiacus Werding & Hiller, 2007 tems of Oman and the genetic endowment to respond to the Cnidarians Acropora parapharaonis Veron, 2000 environments encountered there. Thus, present day bound- Anacropora spumosa Veron, Turak & DeVantier, 2000 aries at the southern Red Sea will be porous and determined Cyphastrea hexasepta Veron, DeVantier & Turak, 2000 by differing degrees of ecological plasticity and genetic diver- Echinopora irregularis Veron, Turak & DeVantier, 2000 sity in taxa that penetrate beyond the Red Sea and into the Echinopora tiranensis Veron, Turak & DeVantier, 2000 Gulf of Aden. Goniopora sultani Veron, DeVantier & Turak, 2000 Our primary argument for this ecological filter follows Keith Montipora aspergillus Veron, DeVantier & Turak, 2000 et al. (2015): what appear to be geographical barriers are Montipora echinata Veron, DeVantier & Turak, 2000 Montipora hemispherica Veron, 2000 defined by traits indicative of establishment (i.e. habitat switching) and persistence but not necessarily dispersal (also

Journal of Biogeography 11 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al. see Keith et al., 2011; Luiz et al., 2013). This hypothesis pre- Benzie, J.A.H. (1999) Genetic structure of coral reef organisms: dicts that while a number of species may disperse beyond the ghosts of dispersal past. American Zoologist, 39, 131–145. southern boundary of the Red Sea, the capacity to establish Bernardi, G., Golani, D. & Azzurro, E. (2010) The genetics of populations reflects the extent to which both phenotypic plas- Lessepsian bioinvasions. Fish invasions of the Mediterranean ticity and genetic endowment of the potential colonisers allows Sea: change and renewal (ed. by D. Golani and B. Appel- successful settlement, post-settlement survival and recruitment baum-Golani), pp. 71–84. Pensoft Publishers, Moscow. to novel environments. Individuals successfully colonising reef Berumen, M.L., Hoey, A.S., Bass, W.H., Bouwmeester, J., habitats ecologically distinct from that of the parental popula- Catania, D., Cochran, J.E.M., Khahil, M.T., Miyake, S., tion would be those with the capacity to respond to the novel Mughal, M.R., Spaet, J.L.Y. & Saenz-Agudelo, P. (2013) selective environments. Genetic analyses designed to differenti- The status of coral reef ecology research in the Red Sea. ate between drift and natural selection (i.e. RAD-seq methods; Coral Reefs, 32, 737–748. Willette et al., 2014) in driving differences between parental Bickford, D., Lohman, D.J., Sodhi, N.S., Ng, P.K., Meier, R., and colonising populations would be an appropriate research Winker, K., Ingram, K.K. & Das, I. (2007) Cryptic species design. The prediction is that species that successfully recruit as a window on diversity and conservation. Trends in Ecol- beyond the distributional boundaries of the parental popula- ogy and Evolution, 22, 148–155. tion will display strong signatures of selection. A critical fea- Bowen, B.W., Rocha, L.A., Toonen, R.J., Karl, S.A. & ToBo ture would be to determine if such colonising populations lab (2013) The origins of tropical marine biodiversity. represent an independent evolutionary trajectory driven by Trends in Ecology and Evolution, 28, 359–366. divergent selection in the environment encountered by the Braithwaite, C.J.R. (1987) Geology and paleogeography of colonists. This is the approach taken by Gaither et al. (2015) the Red Sea region. Key environments. Red Sea (ed. by A.J. in a comparative analysis of Indo-Pacific surgeonfish that suc- Edwards and S.M. Head), pp. 22–44. Pergamon Press, cessfully colonised the divergent reef environment of the Mar- Oxford, UK. quesas Islands, and would therefore be appropriate to apply Briggs, J.C. (1974) Marine zoogeography. McGraw-Hill, New more broadly to other reef fauna. York. Briggs, J.C. (2005) Marine East Indies: diversity and speciation. Journal of Biogeography, 32, 1517–1522. ACKNOWLEDGEMENTS Briggs, J.C. & Bowen, B.W. (2012) A realignment of marine The ‘Red Sea and Western Indian Ocean Biogeography biogeographic provinces with particular reference to fish Workshop’ was funded by the King Abdullah University of distributions. Journal of Biogeography, 39,12–30. Science and Technology (KAUST) Office of Competitive Campbell, A.C. (1987) Echinoderms of the Red Sea. Key Research Funds (OCRF) under Award no. 59130357. This environments: red Sea (ed. by A.J. Edwards and S.M. synthesis paper was further supported by the KAUST OCRF Head), pp. 215–232. Pergamon Press, Oxford, UK. under Award No. CRG-1-2012-BER-002 and baseline Choat, J., Klanten, O.S., van Herwerden, L., Robertson, D.R. research funds to M.L.B., as well as a National Geographic & Clements, K.D. (2012) Patterns and processes in the Society Grant 9024-11 to J.D.D. We acknowledge important evolutionary history of parrotfishes (Family Labridae). Bio- data contributions from J. Taylor, J.E. Randall, and F. Ben- logical Journal of the Linnean Society, 107, 529–557. zoni, as well as logistic support from L. Chen, C. Nelson and Cooper, W.J., Smith, L.L. & Westneat, M.W. (2009) Explor- A. Macaulay. The survey data were collected with financial ing the radiation of a diverse reef fish family: phylogenet- support from KAUST Award Nos. USA 00002 and KSA ics of the damselfishes (Pomacentridae), with new 00011 to S.R. Thorrold, logistic support from Dream Divers classifications based on molecular analyses of all genera. and H. Aljahdali, and field assistance from K.A. Furby, J. Molecular Phylogenetics and Evolution, 52,1–16. Ossolinski, K. Munday and A.S. Al Kotob. The manuscript Cowen, R.K. & Guigand, C.M. (2008) In situ ichthyoplank- was improved by comments from D.R. Bellwood and two ton imaging system (ISIIS): system design and preliminary anonymous referees. We apologise to the authors of relevant results. Limnology and Oceanography: Methods, 6, 126–132. research who were not cited due to strict space constraints. Cowman, P.F., Bellwood, D.R. & van Herwerden, L. (2009) Dating the evolutionary origins of wrasse lineages (Labri- dae) and the rise of trophic novelty on coral reefs. Molecu- REFERENCES lar Phylogenetics and Evolution, 52, 621–631. Allen, G.R. (2008) Conservation hotspots of biodiversity and Crandall, E.D., Treml, E.A. & Barber, P.H. (2012) Coalescent endemism for Indo-Pacific coral reef fishes. Aquatic Con- and biophysical models of stepping-stone gene flow in servation: Marine and Freshwater Ecosystems, 18, 541–556. neritid snails. Molecular Ecology, 21, 5579–5598. Bailey, R. (2009) Ecosystem geography: from ecoregions to sites, Crossland, C. (1935) Coral faunas of the Red Sea and Tahiti. 2nd edn. Springer Publishing, New York, USA. Proceedings of the Zoological Society of London, 105, 499– Baums, I.B., Paris, C.B. & Cherubin, L.M. (2006) A bio- 504. oceanographic filter to larval dispersal in a reef-building Cuvier, G. & Valenciennes, A. (1828) Histoire naturelle des coral. Limnology and Oceanography, 51, 1969–1981. poisons, Tome premier. F. G. Levrault, Paris.

12 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Cuvier, G. & Valenciennes, A. (1850) Histoire naturelle des ESRI. (2014) ArcGIS Desktop 10.2. Environmental Systems poissons, Tome vingt-deuxieme. Suite du livre vingt-deux- Research Institute, Redlands, CA, USA. ieme, Suite de la famille des Salmono€ıdes. Ch. Pitois and Fessler, J.L. & Westneat, M.W. (2007) Molecular phylogenet- V. Levrault, Paris and Strasbourg. ics of the butterflyfishes (Chaetodontidae): taxonomy and Dafni, J. (2008) Diversity and recent changes in the echino- biogeography of a global coral reef fish family. Molecular derm fauna of the Gulf of Aqaba with emphasis on the Phylogenetics and Evolution, 45,50–68. regular echinoids. Aqaba-Eilat, the Improbable Gulf. Envi- Forsskal, P. (1775) Descriptiones animalium avium, amphibio- ronment, Biodiversity and Preservation (ed. by F.D. Por), rum, piscium, insectorum, vermium; quae in itinere orientali pp. 225–242. Magnes Press, Jerusalem. observavit Petrus Forsskal. (Post mortem auctoris editit Car- DiBattista, J.D., Randall, J.E. & Bowen, B.W. (2012) Review sten Niebuhr).Molleri,€ Hauniae. of the round herrings of the genus Etrumeus (Clupeidae: Fratini, S. & Vannini, M. (2002) Genetic differentiation in Dussumieriinae) of Africa, with descriptions of two new the mud crab Scylla serrata (Decapoda: Portunidae) within species. Cybium, 36, 447–460. the Indian Ocean. Journal of Experimental Marine Biology DiBattista, J.D., Berumen, M.L., Gaither, M.R., Rocha, L.A., and Ecology, 272, 103–116. Eble, J.A., Choat, J.H., Craig, M.T., Skillings, D.J. & Fricke, R., Bogorodsky, S.V. & Mal, A.O. (2014) Review of Bowen, B.W. (2013) After continents divide: comparative the genus Diplogrammus (Teleostei: Callionymidae) of the phylogeography of reef fishes from the Red Sea and Indian Red Sea, with description of a new species from Saudi Ocean. Journal of Biogeography, 40, 1170–1181. Arabia. Journal of Natural History, 48, 2419–2448. DiBattista, J.D., Randall, J.E., Newman, S.J. & Bowen, B.W. Froese, R. & Pauly, D. Eds. (2014) Fishbase. World wide web (2014) Australian round herring (genus Etrumeus) electronic publication, version (04/2014). Available at: contain cryptic evolutionary partitions concordant with www.fishbase.org. biogeographic boundaries. Marine Biology, 161, 2465– Froukh, T. & Kochzius, M. (2008) Species boundaries and 2477. evolutionary lineages in the blue green damselfishes Chro- DiBattista, J.D., Rocha, L.A., Hobbs, J.P., He, S., Priest, mis viridis and Chromis atripectoralis (Pomacentridae). M.A., Sinclair-Taylor, T.H., Bowen, B.W. & Berumen, Journal of Fish Biology, 72, 451–457. M.L. (2015) When biogeographical provinces collide: Gaither, M.R., Bowen, B.W., Bordenave, T.-R., Rocha, L.A., hybridization of reef fishes at the crossroads of marine Newman, S.J., Gomez, J.A., van Herwerden, L. & Craig, biogeographical provinces in the Arabian Sea. Journal of M.T. (2011) Phylogeography of the reef fish Cephalopholis Biogeography, 42, 1601–1614. argus (Epinephelidae) indicates Pleistocene isolation across DiBattista, J.D., Choat, J.H., Gaither, M.R., Hobbs, J.P., the Indo-Pacific barrier with contemporary overlap in the Lozano-Cortes, D.F., Myers, R., Paulay, G., Rocha, L.A., Coral Triangle. BMC Evolutionary Biology, 11, 189. Toonen, R.J., Westneat, M. & Berumen, M.L. (in press) Gaither, M.R., Bernal, M.A., Coleman, R.R., Bowen, B.W., On the origin of endemic species in the Red Sea. Journal Jones, S.A., Simison, W.B. & Rocha, L.A. (2015) Genomic of Biogeography. signatures of geographic isolation and natural selection in Duchene, D., Klanten, S.O., Munday, P.L., Herler, J. & van coral reef fishes. Molecular Ecology, 24, 1543–1557. Herwerden, L. (2013) Phylogenetic evidence for recent Giles, E.C., Saenz-Agudelo, P., Hussey, N.E., Ravasi, T. & diversification of obligate coral-dwelling gobies compared Berumen, M.L. (2015) Exploring seascape genetics and with their host corals. Molecular Phylogenetics and Evolu- kinship in the reef sponge Stylissa carteri in the Red Sea. tion, 69, 123–132. Ecology and Evolution, 5, 2487–2502. Eble, J.A., Toonen, R.J., Sorenson, L., Basch, L.V., Papasta- Golani, D. & Bogorodsky, S.V. (2010) The fishes of the Red Sea matiou, Y.P. & Bowen, B.W. (2011) Escaping paradise: lar- —reappraisal and updated checklist. Zootaxa, 2463,1–135. val export from Hawaii in an Indo-Pacific reef fish, the Goren, M. (1986) A suggested model for the recolonisation yellow tang (Zebrasoma flavescens). Marine Ecology Progress process of the Red Sea at the post glacial period. Indo- Series, 428, 245–258. Pacific Fish Biology: Proceedings of the Second International Ehrenberg, C.G. (1834) Beitrage zur physiologischen Kenntniss Conference on Indo-Pacific Fishes (ed. by T. Uyeno, R. Arai, der Corallenthiere im Allgemeinen und besunders des Rothen T. Tainuchi and K. Matsura), pp. 648–654. Ichthyological Meeres, nebst einem Versuche zur physiologischen Systematik Society of Japan, Tokyo. derselben. Abhandlungen der Koniglichen€ Akademie der Goren, M. & Dor, M. (1994) An updated check-list of the Wissenschaften, Berlin. fishes of the Red Sea CLOFRES II. Ekman, S. (1953) Zoogeography of the sea. Sidgwick and Jack- Guinot,D.(1966)Lafaunecarcinologique (Crustacea Brachyura) son Ltd, London. de l’ocean Indien occidental et de la Mer Rouge. Catalogue, Eschmeyer, W.N. (2013) Catalog of fishes. Electronic version, remarques biogeographiques et bibliographie. Memoires de updated 4 January 2013. l’Institut fondamental d’Afrique noire, 77,235–352. Eschmeyer, W.N., Fricke, R., Fong, J.D. & Polack, D.A. Heller, C. (1861) Synopsis der im rothen Meere vorkom- (2010) Marine fish diversity: history of knowledge and dis- menden Crustaceen. Verhandlungen der k.k. zoologisch- covery (Pisces). Zootaxa, 2525,19–50. botanischen Gesellschaft in Wien, 11,1–30.

Journal of Biogeography 13 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Herler, J., Bogorodsky, S.V. & Suzuki, T. (2013) Four new Klunzinger, C.B. (1884) Die fische des Rothen Meeres. E. Sch- species of coral gobies (Teleostei: Gobiidae: Gobiodon), weizerbart’sche Verlagshandlung, Stuttgart. with comments on their relationships within the genus. Knowlton, N. (1993) Sibling species in the sea. Annual Zootaxa, 3709, 301–329. Review of Ecology and Systematics, 24, 189–216. Hoareau, T.B., Boissin, E., Paulay, G. & Bruggemann, J.H. Kochzius, M. & Blohm, D. (2005) Genetic population struc- (2013) The Southwestern Indian Ocean as a potential ture of the lionfish Pterois miles (Scorpaenidae, Pteroinae) marine evolutionary hotspot: perspectives from compara- in the Gulf of Aqaba and northern Red Sea. Gene, 347, tive phylogeography of reef brittle-stars. Journal of Biogeog- 295–301. raphy, 40, 2167–2179. Kochzius, M., Soller,€ R., Khalaf, M.A. & Blohm, D. (2003) Hodge, J.R., Herwerden, L. & Bellwood, D.R. (2014) Tempo- Molecular phylogeny of the lionfish genera Dendrochirus ral evolution of coral reef fishes: global patterns and and Pterois (Scorpaenidae, Pteroinae) based on mitochon- disparity in isolated locations. Journal of Biogeography, 41, drial DNA sequences. Molecular Phylogenetics and Evolu- 2115–2127. tion, 28, 396–403. Hoeksema, B.W. (2007) Delineation of the Indo-Malayan cen- Koeda, K., Yoshino, T., Imai, H. & Tachinara, K. (2014) A tre of maximum marine biodiversity: the Coral Triangle. review of the genus Pempheris (Perciformes, Pempheridae) Biogeography, time and place: distributions, barriers and of the Red Sea, with description of a new species. Zootaxa, islands (ed. by W. Renema), pp. 117–178. Springer, London. 3793, 301–330. Huber, M. & Eschner, A. (2011) Tridacna (Chametrachea) Kooyman, R., Rossetto, M., Cornwell, W. & Westoby, M. costata Roa-Quiaoit, Kochzius, Jantzen, Al-Zibdah & Rich- (2011) Phylogenetic tests ofcommunity assembly across ter from the Red Sea, a junior synonym of Tridacna squa- regional to continental scales in tropical and subtropical mosina Sturany, 1899 (Bivalvia, Tridacnidae). Annalen rain forests. Global Ecology and Biogeography, 20, 707–716. Naturhistorischen Museums Wien, 112, 153–162. Kulbicki, M., Parravicini, V., Bellwood, D.R., Arias-Gonzalez, Hughes, T.P., Bellwood, D.R. & Connolly, S.R. (2002) Biodi- E., Chabanet, P., Floeter, S.R., Friedlander, A., McPherson, versity hotspots, centres of endemicity, and the conserva- J., Myers, R.E., Vigliola, L. & Mouillot, D. (2013) Global tion of coral reefs. Ecology Letters, 5, 775–784. biogeography of reef fishes: a hierarchical quantitative Iacchei, M., Gaither, M.R., Bowen, B.W. & Toonen, R.J. delineation of regions. PLoS ONE, 8, e81847. (in press) Testing provincial boundaries for phyllosoma: Leese, F., Kop, A., W€agele, J.-W. & Held, C. (2008) Cryptic Rangewide phylogeography of the pronghorn spiny lobster speciation in a benthic isopod from Patagonian and Falk- Panulirus penicillatus. Journal of Biogeography. land Island waters and the impact of glaciations on its Jorger,€ K.M., Norenburg, J.L., Wilson, N.G. & Schrodl,€ M. population structure. Frontiers in Zoology, 5, 19. (2012) Barcoding against a paradox? Combined molecular Lessios, H.A. & Robertson, D.R. (2006) Crossing the impass- species delineations reveal multiple cryptic lineages in elusive able: genetic connections in 20 reef fishes across the east- meiofaunal sea slugs. BMC Evolutionary Biology, 12, 245. ern Pacific barrier. Proceedings of the Royal Society B: Keith, S.A., Herbert, R.J., Norton, P.A., Hawkins, S.J. & Biological Sciences, 273, 2201–2208. Newton, A.C. (2011) Individualistic species limitations of Lester, S.E. & Ruttenberg, B.I. (2005) The relationship climate-induced range expansions generated by meso-scale between pelagic larval duration and range size in tropical dispersal barriers. Diversity and Distributions, 17, 275–286. reef fishes: a synthetic analysis. Proceedings of the Royal Keith, S.A., Woolsey, P.S., Madin, J.S., Byrne, M. & Baird, A. Society B: Biological Sciences, 272, 585–591. (2015) Differential establishment potential of species pre- Liggins, L., Treml, E.A. & Riginos, C. (2013) Taking the plunge: dicts a shift in coral assemblage structure across a biogeo- an introduction to undertaking seascape genetic studies and graphic barrier. Ecography, 38, 001–010. using biophysical models. Geography Compass, 7,173–196. Kemp, J. (1998) Zoogeography of the coral reef fishes of the Linnæus, C. (1758) Systema naturæ, 10th edn. Holmiæ, Socotra Archipelago. Journal of Biogeography, 25, 919–933. Salvii. Kemp, J. (2000) Zoogeography of the coral reef fishes of the Liu, S.Y.V., Chang, F.T., Borsa, P., Chen, W.J. & Dai, C.F. northeastern Gulf of Aden, with eight new records of coral (2014) Phylogeography of the humbug damselfish, Dascyl- reef fishes from Arabia. Fauna of Arabia, 18, 293–321. lus aruanus (Linnaeus, 1758): evidence of Indo-Pacific Khalaf, M.A. & Kochzius, M. (2002) Community structure vicariance and genetic differentiation of peripheral popula- and biogeography of shore fishes in the Gulf of Aqaba, tions. Biological Journal of the Linnean Society, 113, 931– Red Sea. Helgoland Marine Research, 55, 252–284. 942. Klausewitz, W. (1989) Evolutionary history and zoogeogra- Luiz, O.J., Allen, A.P., Robertson, D.R., Floeter, S.R., Kul- phy of the Red Sea ichthyofauna. Fauna of Saudi Arabia, bicki, M., Vigliola, L., Becheler, R. & Madin, J.S. (2013) 10, 310–337. Adult and larval traits as determinants of geographic range Klunzinger, C.B. (1870) Synopsis der fische des Rothen Meeres. size among tropical reef fishes. Proceedings of the National C. Ueberreuter’shBuchdruckerei, Wien. Academy of Sciences USA, 110, 16498–16502. Klunzinger, C.B. (1877-1879) Die Korallthiere des Rothen Macpherson, E. & Raventos, N. (2006) Relationship between Meeres. Gutmann’schen, Berlin. pelagic larval duration and geographic distribution in

14 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

Mediterranean littoral fishes. Marine Ecology Progress Ser- Red Sea and Gulfs of Aqaba and Suez. Fauna of Saudi ies, 265, 257–265. Arabia, 4,3–21. Malay, M.C.M.D. & Paulay, G. (2010) Peripatric speciation Raitsos, D.E., Pradhan, Y., Brewin, R.J.W., Stenchikov, G. & drives diversification and distributional pattern of reef her- Hoteit, I. (2013) Remote sensing the phytoplankton sea- mit crabs (Decapoda: Diogenidae: Calcinus). Evolution, 64, sonal succession of the Red Sea. PLoS ONE, 8, e64909. 634–662. Randall, J.E. (1955) A revision of the surgeon fish genera Ze- Mayr, E. (1954) Geographic speciation in tropical echinoids. brasoma and Paracanthurus. Pacific Science, 9, 396–412. Evolution, 8,1–18. Randall, J.E. (1994) Twenty-two new records of fishes from Mervis, J. (2009) The big gamble in the Saudi desert. Science, the Red Sea. Fauna of Saudi Arabia, 14, 259–275. 326, 354–357. Randall, J.E. (1995) Coastal fishes of Oman. Crawford House Mora, C., Chittaro, P.M., Sale, P.F., Kritzer, J.P. & Ludsin, Publishing, Bathurst, Australia. S.A. (2003) Patterns and processes in reef fish diversity. Randall, J.E. & DiBattista, J.D. (2012) Etrumeus makiawa,a Nature, 421, 933–936. new species of Round Herring (Clupeidae: Dussumierinae) Munday, P.L., Leis, J.M., Lough, J.M., Paris, C.B., Kingsford, from the Hawaiian Islands. Pacific Science, 66,97–110. M.J., Berumen, M.L. & Lambrechts, J. (2009) Climate change Randall, J.E. & DiBattista, J.D. (2013) A new species of dam- and coral reef connectivity. Coral Reefs, 28,379–395. selfish (Pomacentridae) from the Indian Ocean. Aqua, Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, International Journal of Ichthyology, 19,1–16. G.A. & Kent, J. (2000) Biodiversity hotspots for conserva- Randall, J.E., DiBattista, J.D. & Wilcox, C. (2011) Acanthurus tion priorities. Nature, 403, 853–858. nigros Gunther,€ a valid species of surgeonfish, distinct Nanninga, G.B., Saenz-Agudelo, P., Manica, A. & Berumen, from the Hawaiian A. nigroris Valenciennes. Pacific Science, M.L. (2014) Environmental gradients predict the genetic 65, 265–275. population structure of a coral reef fish in the Red Sea. Reiss, Z. & Hottinger, L. (1984) The Gulf of Aqaba: ecological Molecular Ecology, 23, 591–602. micropaleontology, Vol. 384. Berlin, Springer-Verlag. Nuryanto, A. & Kochzius, M. (2009) Highly restricted gene Riginos, C., Douglas, K.E., Jin, Y., Shanahan, D.F. & Treml, flow and deep evolutionary lineages in the giant clam E.A. (2011) Effects of geography and life history traits on Tridacna maxima. Coral Reefs, 28, 607–619. genetic differentiation in benthic marine fishes. Ecography, Oksanen, J., Guillaume Blanchet, F., Kindt, R., Legendre, P., 34, 566–575. Minchin, P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., Roberts, C.M., Shepherd, A.R.D. & Ormond, R.F.G. (1992) Stevens, H.H. & Wagner, H. (2014) Vegan: community Large-scale variation in assemblage structure of Red Sea ecology package (accessed 1 August 2014). butterflyfishes and angelfishes. Journal of Biogeography, 19, Ormond, R. & Edwards, A. (1987) Red Sea fishes. Key envi- 239–250. ronments. Red Sea (ed. by A.J. Edwards and S.M. Head), Roberts, C.M., McClean, C.J., Veron, J.E., Hawkins, J.P., Al- pp. 251–287. Pergamon Press, Oxford. len, G.R., McAllister, D.E., Mittermeier, C.G., Schueler, Pinzon, J.H., Sampayo, E., Cox, E., Chauka, L.J., Chen, C.A., F.W., Spalding, M., Wells, F., Vynne, C. & Werner, T.B. Voolstra, C.R. & LaJeunesse, T.C. (2013) Blind to mor- (2002) Marine biodiversity hotspots and conservation pri- phology: genetics identifies several widespread ecologically orities for tropical reefs. Science, 295, 1280–1284. common species and few endemics among Indo-Pacific Robertson, D.R. (2001) Population maintenance among tropi- cauliflower corals (Pocillopora, Scleractinia). Journal of cal reef fishes: inferences from small-island endemics. Pro- Biogeography, 40, 1595–1608. ceedings of the National Academy of Sciences, 98, 5667–5670. Plaisance, L., Caley, M.J., Brainard, R.E. & Knowlton, N. Robitzch, V., Banguera-hinestroza, E., Sawall, Y., Al-Sofyani, (2011) The diversity of coral reefs: what are we missing? A. & Voolstra, C.R. (2015) Absence of genetic differentia- PLoS ONE, 6, e25026. tion in the coral Pocillopora verrucosa along environmental Poore, G.C.B., Avery, L., Błazewicz-Paszkowycz,_ M., Browne, gradients of the Saudi Arabian Red Sea. Frontiers in Mar- J., Bruce, N.L., Gerke, S., Glasby, C., Greaves, E., McCal- ine Science, 2,1–10. lum, A.W., Staples, D., Syme, A., Taylor, J., Walker-Smith, Rohling, E.J., Fenton, M., Jorissen, F.J., Bertrand, P., Gnas- G., Warne, M., Watson, C., Williams, A., Wilson, R.S. & sen, G. & Caulet, J.P. (1998) Magnitudes of sea-level low- Woolley, S. (2014) Invertebrate diversity of the unexplored stands of the past 500,000 years. Nature, 394, 162–165. marine western margin of Australia: taxonomy and impli- Ruppell,€ E. (1828) Atlas zu der Reise im nordlichen Afrika. cations for global biodiversity. Marine Biodiversity, 45, Fische des Rothen Meeres, Bronner, Frankfurt. 271–286. Ruppell,€ E. (1835–1838). Neue Wirbelthiere zu der Fauna von Postaire, B., Bruggemann, J.H., Magalon, H. & Faure, B. Abyssinien geh€orig. Fische des Rothen Meeres. Schmerber, (2014) Evolutionary dynamics in the Southwest Indian Frankfurt-am-Main. Ocean marine biodiversity hotspot: a perspective from the Ruppell,€ E. & Leuckart, F.S. (1828) Atlas zu der Reise im rocky shore gastropod genus Nerita. PLoS ONE, 9, e103393. n€ordlichen Afrika von Eduard Ruppell.€ Neue wirbellos Thiere Price, A.R. (1982) Echinoderms of Saudi Arabia. Comparison des Rothen Meeres. Heinrich Ludwig Bronner, Frankfurt an between echinoderm faunas of Arabian Gulf, SE Arabia, Maim.

Journal of Biogeography 15 ª 2015 John Wiley & Sons Ltd J. D. DiBattista et al.

Selkoe, K.A. & Toonen, R.J. (2011) Marine connectivity: a Labridae. Molecular Phylogenetics and Evolution, 36, 370– new look at pelagic larval duration and genetic metrics of 390. dispersal. Marine Ecology Progress Series, 436, 291–305. Willette, D.A., Allendorf, F.W., Barber, P.H., Barshis, D.J., Selkoe, K.A., Gaggiotti, O.E., Laboratory, ToBo, Bowen, B.W. Carpenter, K.E., Crandall, E.D., Cresko, W.A., Fernandez- & Toonen, R.J. (2014) Emergent patterns of population Silva, I., Matz, M.V., Meyer, E., Santos, M.D., Seeb, L.W. genetic structure for a coral reef community. Molecular & Seeb, J.E. (2014) So, you want to use next-generation Ecology, 23, 3064–3079. sequencing in marine systems? Insight from the Pan-Paci- Sheppard, C.R.C. & Sheppard, A.L.S. (1991) Corals and coral fic Advanced Studies Institute. Bulletin of Marine Science, communities of Arabia. Fauna of Saudi Arabia, 12,3–170. 90,79–122. Sheppard, C.R.C., Price, A. & Roberts, C. (1992) Marine ecol- Williams, S.T. & Reid, D.G. (2004) Speciation and diversity ogy of the Arabian region. Academic Press, London. on tropical rocky shores: a global phylogeny of snails of Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., the genus Echinolittorina. Evolution, 58, 2227–2251. Meischner, D., Schmeltzer, I. & Smeed, D.A. (2003) Sea- Williams, S.T., Apte, D., Ozawa, T., Kagilis, F. & Nakano, level fluctuations during the last glacial cycle. Nature, 423, T. (2011) Speciation and dispersal along continental 853–858. coastlines and island arcs in the Indo-West Pacific Smeed, D.A. (1997) Seasonal variation of the flow in the turbinid gastropod genus Lunella. Evolution, 65, 1752– strait of Bab al Mandab. Oceanologica Acta, 20, 773–781. 1771. Sofianos, S.S. (2003) An Oceanic General Circulation Model Williams, S.T., Hall, A. & Kuklinski, P. (2012) Unravelling (OGCM) investigation of the Red Sea circulation: 2. cryptic diversity in the Indo-West Pacific gastropod genus Three-dimensional circulation in the Red Sea. Journal of Lunella (Turbinidae) using elliptic Fourier analysis. Ameri- Geophysical Research, 108, 11-1–11-15. can Malacological Bulletin, 30, 189–206. Spalding, M.D., Fox, H.E., Allen, G.R., Davidson, N., Fer- Winterbottom, R. (1985) Revision of the congrogadid Halio- dana, Z.A., Finlayson, M., Halpern, B.S., Jorge, M.A., phis (Pisces: Perciformes), with the description of a new Lombana, A., Lourie, S.A., Martin, K.D., McManus, E., species from Indonesia, and comments on the endemic Molnar, J., Recchia, C.A. & Robertson, J. (2007) Marine fish fauna of the northern Red Sea. Canadian Journal of ecoregions of the world: a bioregionalization of coastal Zoology, 63, 209–217. and shelf areas. BioScience, 57, 573–583. Wolf-Vecht, A., Paldor, N. & Brenner, S. (1992) Hydro- Szabo, Z., Snelgrove, B., Craig, M.T., Rocha, L.A. & Bowen, graphic indications of advection/convection effects in the B.W. (2014) Phylogeography of the Manybar Goatfish, Paru- Gulf of Eilat. Deep Sea Research Part A. Oceanographic peneus multifasciatus reveals moderate structure between the Research Papers, 39, 1393–1401. Central and North Pacific and a cryptic endemic species in Zajonz, U., Khalaf, M. & Krupp, F. (2000) Coastal fish the Marquesas. Bulletin of Marine Science, 90, 493–512. assemblages of the Socotra Archipelago. Conservation and Terraneo, T.I., Berumen, M.L., Arrigoni, R., Waheed, Z., sustainable use of biodiversity of the Socotra Archipelago. Bouwmeester, J., Caragnano, A., Stefani, F. & Benzoni, F. Marine Habitat, Biodiversity and Fisheries Surveys and (2014) Pachyseris inattesa sp. n. (Cnidaria, Anthozoa, Management. Progress Report of Phase III.–Senckenberg Scleractinia): a new reef coral species from the Red Sea Research Institute, Frankfurt aM. and its phylogenetic relationships. ZooKeys, 433,1–30. Zapata, F.A. & Herron, P.A. (2002) Pelagic larval duration Tikochinski, Y., Friling, M., Harush, N., Lizarovich, R., and geographic distribution of tropical eastern Pacific Manor, N., Horsky, A., Appelbaum-Golani, B. & Golani, snappers (Pisces: Lutjanidae). Marine Ecology Progress Ser- D. (2013) Molecular comparison of geographically ies, 230, 295–300. extreme populations of fish species of wide Indo-Pacific distribution. Israel Journal of Ecology and Evolution, 59, SUPPORTING INFORMATION 197–200. Veron, J.E.N., Devantier, L.M., Turak, E., Green, A.L., Kinin- Additional Supporting Information may be found in the month, S., Stafford-Smith, M. & Peterson, N. (2009) De- online version of this article: lineating the Coral Triangle. Galaxea, Journal of Coral Reef Appendix S1 Updated checklist of corals from the Arabian Studies, 11,91–100. Peninsula region, including source references. Victor, B.C. & Wellington, G.M. (2000) Endemism and the Appendix S2 Updated checklist of shore fish from the Ara- pelagic larval duration of reef fishes in the eastern Pacific bian Peninsula region, including footnotes and source refer- Ocean. Marine Ecology Progress Series, 205, 241–248. ences. Vogler, C., Benzie, J., Lessios, H., Barber, P. & Worheide,€ G. Appendix S3 Updated checklist of non-coral invertebrate (2008) A threat to coral reefs multiplied? Four species of phyla (annelids, arthropods, echinoderms, tunicates and mol- crown-of-thorns starfish. Biology Letters, 4, 696–699. luscs) from the Arabian Peninsula region, including source Westneat, M.W. & Alfaro, M.E. (2005) Phylogenetic references. relationships and evolutionary history of the reef-fish family

16 Journal of Biogeography ª 2015 John Wiley & Sons Ltd Contemporary patterns of Red Sea endemism

BIOSKETCH Author contributions: All authors listed here contributed data, analysed the data or wrote sections of the paper. The authors include researchers with a vast range of expertise including ecological surveys, testing evolutionary models, Editor: David Bellwood resolving life history traits that inﬂuence dispersal, population separations in reef organisms and informing marine conservation initiatives in the greater Indo-Paciﬁc region.

Journal of Biogeography 17 ª 2015 John Wiley & Sons Ltd REEF ENCOUNTER The News Journal of the International Society for Reef Studies Short Communications: Sargassum Seaweed Outbreak

Gower J, Young E, King S (2013) Satellite images suggest a new Sargassum source region in 2011. Remote Sensing Coral Bleaching in Saudi Arabia Letters. 4 (8): 764-773, http://dx.doi.org/10.1080/ Affecting both the Red Sea and 2150704X.2013.796433 Gower J, King S (2011) Distribution of floating Sargassum in Arabian Gulf the Gulf of Mexico and the Atlantic Ocean mapped using MERIS. Int. J. Remote Sens. 32, 1917–1929 Diego Lozano-Cortés1, Vanessa Robitzch2, Khaled Hu C, Montgomery D, Schmitt R, Muller-Karger F (2004) The Abdulkader1, Yasser Kattan1, Alaa Elyas1& dispersal of the Amazon and Orinoco River water in the Michael Berumen2 tropical Atlantic and Caribbean Sea: observation from space and S-PALACE floats. Deep-Sea Research II 51: 1 Environmental Protection Department, Marine 1151-1171 Environment Protection Unit. Saudi Aramco. Dhahran, Knight J, Folland C, Scaife A (2006) Climate impacts of the 31311. Kingdom of Saudi Arabia. 2 Red Sea Research Center, Atlantic Multidecadal Oscillation. Geophysical research Reef Ecology Lab. King Abdullah University of Science and Letters 33: 4p, L17706, doi:10.1029/2006GLO26242 Technology, Thuwal, 23955. Kingdom of Saudi Arabia. Lapointe B (1995) A comparison of nutrient-limited

productivity in Sargassum natans from neritic vs. oceanic waters of the western North Atlantic Ocean. Limnology We are currently experiencing a third global bleaching and Oceanography 40 (3): 625-633 event, officially recognised as such last October by the Marengo J, Borma L, Rodriguez D, Pinho P, Soares W, Alves L US National Oceanic & Atmospheric Administration (2013) Recent extremes of drought and flooding in (NOAA, 2015). NOAA has attributed this global Amazonia: vulnerabilities and human adaptation. bleaching to the combined effects of global warming American Journal of Climate Change 2: 87-96 and a very strong El Niño event, and predicted that it McGrath G (2003) Scientists: seaweed harvesting too risky. Starnewsonline.com, accessed 2-16. may impact over a third of the world’s coral reefs. In NOAA –Teachers at Sea. From: Teacheratsea. this context it is of particular interest that bleaching wordpress.com. Accessed 2-16 has recently occurred on both coasts of the Kingdom Otorongo Expeditions (2012): http://otorongoexpeditions. of Saudi Arabia, since both the Arabian Gulf and Red com/author/otoexadmin/, accessed 2-16 Sea coasts have historically experienced summer peak Oyesiku O, Egunyomi A (2014) Identification and chemical studies of pelagic masses of Sargassum natans temperatures that would normally be expected to (Linnaeus) Gaillon and S. fluitans (Borgessen) Borgesen cause bleaching of corals elsewhere, but have not (brown algae), found offshore in Ondo State, Nigeria. typically done so in these extreme environments. African journal of Biotechnology 13(10): 1188-1193. Doi: 10.5897/AJB2013.12335 In the Arabian Gulf surveys undertaken in September Redmond S, Kim J, Yarish C, Pietrak M, Bricknell I (2014) (2015) by the King Fahd University of Petroleum and Culture of Sargassum in Korea: techniques and potential Minerals (KFUPM) as part of research lead by the for culture in the U.S. Orono, ME: Maine Sea Grant Environmental Protection Department of Saudi College Program. seagrant.umaine.edu/extension/korea- aquaculture. 13p Aramco (Sustaining Project Phase VI) found coral Sargasso Sea Alliance (2014) @SargSeaAlliance.org, bleaching at Manifa, Safaniya, and on Jana Island. accessed 3-16 Surface water temperatures at the time were as high Smetacek V, Zingone A (2013) Green and golden seaweed as 32.5°C. Coral bleaching was noted in both nearshore tides on the rise. Nature 504: 84-88; and offshore locations, and observed to be affecting doi:10.1038/nature12860 both massive and branching corals (Fig. 1), with Porites Voanews.com 2015: gdb.voanews.com/F38835EE-9B97- 4792-40E686F5A55A_mw1024_s_n.jpg, accessed 2-16 being the genus most affected. Mass bleaching events Webster R, T Linton (2013) Development and were previously recorded in the Arabian Gulf during implementation of Sargassum early advisory system 1996 and 1998, when seawater temperature reached (SEAS). Shore & Beach 81 (3): 6p 37.7°C, resulting in the almost total extirpation of the Wikipedia, the free encyclopedia. https://en.wikipedia. most abundant coral genus, Acropora (Sheppard and org/wiki/Langmuir_circulation, accessed 2-16 Loughland 2002, Burt et al. 2011). Acropora colonies remain relatively uncommon on reefs in the Arabian Gulf to this day.

50 | Page VOLUME 31 NUMBER 1 April 2016 REEF ENCOUNTER The News Journal of the International Society for Reef Studies Short Communications: Coral Bleaching in Saudi Arabia

In the Red Sea, scientists of the Reef Ecology Lab at Furby KA, Bouwmeester J, Berumen ML (2013) Susceptibility King Abdullah University of Science and Technology of central Red Sea corals during a major bleaching event. (KAUST) observed the occurrence of bleaching in both Coral Reefs 32: 505-513. NOAA (2015) National Oceanic & Atmospheric South-Central and North-Central regions. In the South- Administration. NOAA declares third ever global coral Central region bleaching was first reported from reefs bleaching event. near Al-Lith during the northern hemisphere summer, http://www.noaanews.noaa.gov/stories2015/100815- and persisted even through December. At least some noaa-declares-third-ever-global-coral-bleaching- offshore reefs (~50km from the coast) were heavily event.html Voolstra CR, Miller, DJ, Ragan MA, Hoffmann A, Hoegh- impacted, as were corals down to depths of 20-30m. In Guldberg O, Bourne D, Ball E, Ying H, Foret S, Takahashi the North-Central region (near Thuwal), preliminary S, Weynberg KD, van Oppen MJH, Morrow K, Chan CX, analyses suggest that the bleaching pattern was more Rosic N, Leggat W, Sprungala S, Imelfort M, Tyson GW, similar to that observed in 2010 (Furby et al. 2013) Kassahn K, Lundgren P, Beeden R, Ravasi T, Berumen M, when deeper and offshore sites appeared much less Abel E, Fyffe T (2015). RefuGe2020 consortium — using ‘omics’ approaches to explore the adaptability and affected. Many different species of hard corals were resilience of coral holobionts to environmental change. affected as were some soft corals and sea anemones Front. Mar. Sci., 2: 2015, p. 68 (Fig. 2). The data are being further analysed and a full http://dx.doi.org/10.3389/fmars.2015.00068 report will be published in due course. The fact that corals in the Kingdom´s waters have been affected by bleaching despite their known greater tolerance of high sea temperatures (Coles & Riegl 2013, Voolstra et al. 2015) suggests that if global warming continues these corals may not be able to adapt further (Baker et al. 2004, 2008, Cantin et al. 2010). It may also suggest that there is less scope than had been hoped for transferring any genetic tolerance to bleaching shown by the Kingdom’s corals to corals elsewhere.

References Baker AC, Glynn PW, Riegl B (2008). Climate change and coral reef bleaching: an ecological assessment of long- term impacts, recovery trends and future outlook. Estuar Coast Shelf Sci 80: 435–471. Baker AC, Starger CJ, McClanahan TR, Glynn PW (2004) Coral reefs: Corals' adaptive response to climate change. Nature 430: 741. doi:10.1038/430741a. Burt J, Al-Harthi S, Al-Cibahy A (2011) Long-term impacts of coral bleaching events on the world’s warmest reefs. Mar Environ Res 72: 225–229. Cantin NE, Cohen AL, Karnauskas KB, Tarrant AM, McCorkle DC (2010) Ocean Warming Slows Coral Growth in the Central Red Sea. Science 329: 322-325. DOI: 10.1126/science.1190182 Coles SL, Riegl B (2013) Thermal tolerances of reef corals in the Gulf: A review of the potential for increasing coral survival and adaptation to climate change through Figure 1. Contrasting coral growth forms affected by assisted translocation. Mar Poll Bull 72: 323–332. bleaching on Jana Island, in the Saudi Arabian Gulf. Douglas AE (2003) Coral bleaching–how and why? Mar Poll Photographs courtesy of Sustaining Project Phase VI, Bull 46: 385–392. Marine Department, KFUPM.

VOLUME 31 NUMBER 1 April 2016 51 | Page REEF ENCOUNTER The News Journal of the International Society for Reef Studies Short Communications: Coral Bleaching in Saudi Arabia

Figure 2. Bleached branching corals and sea anemone at Al Lith, Central Saudi Red Sea. Photographs by Vanessa Robitzch.

52 | Page VOLUME 31 NUMBER 1 April 2016