Open Boulay Thesis Final.Pdf

The Pennsylvania State University

The Graduate School

Eberly College of Science

POPULATION CONNECTIVITY AND SPECIES DIVERSITY

OF PACIFIC CORAL REEFS

A Dissertation in

Biology

Jennifer N. Boulay

 2014 Jennifer N. Boulay

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2014

The dissertation of Jennifer N. Boulay was reviewed and approved* by the following:

Katriona Shea Professor of Biology Chair of Committee

Iliana B. Baums Associate Professor of Biology Dissertation Adviser

Todd C. LaJeunesse Associate Professor of Biology

George Perry Assistant Professor of Anthropology

Jorge Cortés Professor of Biology and Marine Science University of Costa Rica Special Signatory

Douglas R. Cavener Professor of Biology Head of the Department of Biology

*Signatures are on file in the Graduate School

iii

ABSTRACT

Unrecognized genetic and species diversity in biotic interactions involving foundation fauna can impede our understanding of ecological drivers of community structuring. In my thesis,

I show that what has been considered a single species in the Eastern Tropical Pacific (ETP),

Porites lobata, includes a morphologically similar yet ecologically distinct species, Porites evermanni. Porites lobata is replaced by P. evermanni at continental sites and towards higher latitudes along the continental coast, indicating that the two species occupy different ecological niches. Further, rates of asexual fragmentation differed significantly (p=0.01) between the two species as ascertained by genotyping at 11 microsatellite loci from sampling using a spatially explicit random method. Although P. evermanni reproduces extensively via fragmentation, P. lobata does so rarely. This crucial difference in population maintenance may be a driver of population structuring especially in the face of extreme El Niño-Southern Oscillation (ENSO) disturbance events which are frequent in the ETP. At some coastal sites, larger, asexually- produced fragments rather than smaller, sexually-produced larvae appear to have the advantage.

Asexual fragmentation in these corals is thought to be facilitated by presence of endolithic

Lithophaga mussels. Where both Porites species occur, the number of mussels/cm2 is significantly higher in P. evermanni based on field (p<0.001) and photographic (p=0.011) evidence. Additionally, we show that both corals associate primarily with Symbiodinium clade

C15 (identified by DGGE of the ITS2 region); however, P. lobata was found to bleach more readily then P. evermanni when growing in close proximity (p<0.001). Differential bleaching frequency of the coral holobionts thus could be explained by host physiology differences, or by as yet undetected, host-specific Symbiodinium lineages below the C15 resolution level, or a combination of both. To explore the importance of abiotic factors in determining the geographical distribution differences between P. lobata and P. evermanni holobionts, a maximum entropy

iv model was constructed, and indicated that temperature was indeed important in determining

Porites/Symbiodinium species distribution. Thus, niche differentiation in these two closely related coral hosts and their symbionts is likely driven by preferences of the holobionts to abiotic factors explaining the divergent ecological response to thermal stress. Hidden diversity within the coral-Symbiodinium association has until now obscured differences in trophic interactions, reproductive dynamics and bleaching susceptibility, related to differential response to local temperature regimes. Including biotic factors that facilitate asexual reproduction with environmental abiotic factors in a multiple linear regression model to predict the distribution of

Porites species revealed that biotic interactions are valuable for explaining differences in species distributions. Therefore, the variation in biotic interactions crucial to population maintenance across species’ ranges should be considered in light of future climate change.

When confronted with large-scale disturbances brought on by climate change, population maintenance is often dependent on supply of larval recruits from unimpacted locations. In the case of Eastern Pacific Porites, reefs were well connected across the region, likely due to the long-lived, symbiont containing larval stage of the Porites species. In contrast, reef-building

Acropora species have shorter lived larvae and their fragile, branching growth form make them susceptible to physical fragmentation sometimes resulting in large clonal patches. By applying high resolution genetic markers, gene flow and the importance of asexual versus sexual reproduction in population maintenance were investigated in dominant staghorn corals, Acropora cf. pulchra of Guam. High levels of asexual fragmentation were observed at considerable spatial scales and a possible barrier to larval dispersal was identified where strong eddies serve to return larvae to their natal reef. Thus maintenance of coral communities in Guam depends on asexual recruitment by fragmentation possibly perpetuating locally well adapted genotypes while reducing diversity. Management efforts should thus operate on small scales and focus on preserving clonal diversity in this system.

Overall the work presented in this thesis shows how genetic markers allow for investigations of coral population dynamics operating at multiple spatial scales and reveals how population processes impact coral communities locally and regionally.

TABLE OF CONTENTS

List of Figures...... viii

List of Tables ...... xiv

Acknowledgements ...... xviii

Chapter 1 INTRODUCTION ...... 1

Background ...... 1 Study systems and Methods ...... 4 Outline ...... 13 References...... 14

Chapter 2 HIGH GENOTYPIC DIVERSITY OF THE REEF-BUILDING CORAL PORITES LOBATA (SCLERACTINIA: PORITIDAE) IN ISLA DEL COCO NATIONAL PARK, COSTA RICA ...... 18

Abstract ...... 18 Introduction ...... 19 Materials and Methods ...... 22 Results ...... 27 Discussion ...... 30 Acknowledgements ...... 33 Author Contributions ...... 33 References...... 34

Chapter 3 UNRECOGNIZED CORAL SPECIES DIVERSITY MASKS DIFFERENCES IN FUNCTIONAL ECOLOGY ...... 42

Abstract ...... 42 Introduction ...... 42 Results and Discussion ...... 45 Materials and Methods ...... 49 Acknowledgements ...... 56 Data Accessibility ...... 56 Author Contributions ...... 56 References...... 57

Chapter 4 BIOTIC AND ABIOTIC FACTORS EXPLAIN NICHE DIFFERENTIATION BETWEEN PORITES LOBATA AND PORITES EVERMANNI IN THE EASTERN TROPICAL PACIFIC ...... 65

Abstract ...... 65 Introduction ...... 66 Methods ...... 69

vii

Results ...... 76 Discussion ...... 79 Acknowledgements ...... 86 Author Contributions ...... 86 References...... 87

Chapter 5 GENETIC CONNECTIVITY AND SPATIAL CLONAL STRUCTURE IN A DOMINANT STAGHORN CORAL ON REEFS AROUND GUAM ...... 96

Abstract ...... 96 Introduction ...... 97 Methods ...... 100 Results ...... 108 Discussion ...... 113 Acknowledgements ...... 119 Author Contributions ...... 119 References...... 120

Chapter 6 CONCLUSIONS AND SIGNIFICANCE ...... 136

References...... 140

Appendix A SUPPLEMENTAL MATERIAL FOR CHAPTER 2 ...... 141

Appendix B SUPPLEMENTAL MATERIAL FOR CHAPTER 3...... 142

Definitions ...... 142 Supplemental Results ...... 143 Supplemental References ...... 145

Appendix C SUPPLEMENTAL MATERIAL FOR CHAPTER 4...... 151

Appendix D SUPPLEMENTAL MATERIAL FOR CHAPTER 5 ...... 156

Appendix E NO GENE FLOW ACROSS THE EASTERN PACIFIC BARRIER IN THE REEF-BUILDING CORAL PORITES LOBATA ...... 163

Abstract ...... 163 Introduction ...... 164 Materials and Methods ...... 167 Results ...... 171 Discussion ...... 175 Acknowledgements ...... 184 Data Accessibility ...... 184 Author contributions ...... 184 References...... 185

Appendix F SUPPLEMENTAL MATERIAL FOR APPENDIX E ...... 202

viii

LIST OF FIGURES

Figure 1-1. Coral life history and reproduction. Asexual reproduction by fragmentation (Left; green) and sexual reproduction by broadcast spawning (Right; blue) ...... 2

Figure 1-2. Geographical Range Porites lobata in the Pacific Ocean. Sites sampled for Pacific wide analysis of Porites lobata are indicated by red circles. Source: Dr. Iliana Baums ...... 5

Figure 2-1. Porites lobata was sampled randomly in polar plots at oceanic and coastal sites in Costa Rica (A). Three plots each were sampled at the coastal Isla del Caño (B) and the oceanic Isla del Coco (C). Polar plots indicate the genotypic identity and size of P. lobata colonies sampled. Colonies with unique multi-locus genotypes (MLGs) are represented by a solid square. All other symbols indicate a repeated MLG. Ramets of the same genet are indicated by common symbols (except solid squares). Scale bar in (A) indicate distance in km. In polar plots (B, C), the radial axis shows distance (m) in 5m increments and the radial axis shows the angle in degrees in 30° increments. Colony symbols were scaled by estimated colony area (range: 0.0096-4.32 m2). Size distribution of colonies sampled from Isla del Coco (black bars, n=57) and coastal sites (white bars, n=33) is shown in panel (D)...... 39

Figure 2-2. Mean (+1 standard error) of genotypic (A) and genetic (B) diversity indices averaged over all loci (where appropriate) and sites per region (Isla del Coco=black bars; Coastal sites=white bars) in Porites lobata. Genotypic indices (A): clonal richness (NG/N ), genotypic diversity (GO/GE) (Stoddart & Taylor 1988), evenness (GO/NG) (Stoddart & Taylor 1988), Simpson’s diversity (D), Shannon-Weiner diversity (H). Genetic indices (B): effective number of alleles per locus (AE), number of private alleles per locus (AP), observed heterozygosity (HO), expected heterozygosity (HE). Each statistic was tested for signiﬁcant differences between Isla del Coco and the coastal region using t-test. *p<0.05 ...... 40

Figure 2-3. Clonal structure of Porites lobata colonies sampled at six sites in Costa Rica. The relation of genotypic evenness GO/NG to diversity GO/GE combined with K- means clustering procedure resulted in three groups of sites (indicated on graph by distinct symbol shapes). The two purely sexual sites, Punta María and Caño1, have overlapping symbols. Fill of symbols indicate the region to which the sampling sites belong (Isla del Coco=black; coastal sites=white)...... 41

Figure 3-1. Cluster analysis of 11 microsatellite loci amplified in Eastern Tropical Pacific Porites samples. Strong probability of membership to either the Porites lobata or the Porites evermanni cluster is demonstrated by individuals in all locations. (A) Principal coordinates (PCo) analysis on genetic identity by region. North = Mexico and Clipperton Island, Central = Costa Rica, Cocos Island, and Panama, South = Galápagos and Ecuador (B) STRUCTURE plot with probability of membership (PM) to a cluster given on y-axis, samples are on x-axis...... 60

Figure 3-2. Distribution of Porites evermanni and Porites lobata across the Eastern Pacific. Only P. evermanni was found in Mexico or the northern-most site in Costa Rica (upper right insert). P. evermanni was rare offshore (lower left insert). Numbers indicate ramets sampled...... 61

Figure 3-3. Colony growth forms of Porites spp. in the Eastern Pacific. (A) Photos of P. lobata top row: whole colony; bleached colony; bottom: typical ridge-like morphology. (B) Photos of P. evermanni top row: whole colony, endolithic Lithophagea mussels exposed during sampling, “rolling stone” fragment; bottom: typical peak-like nodules with Lithophagea boreholes at the base of or between peaks...... 62

Figure 3-4. Analysis of clonal structure in Costa Rica revealed greater clonal diversity in P. lobata (black) than P. evermanni (red). Panels show average genotypic diversity indices (NG/N and GO/GE) and representative polar plots (3 out of 10) along the Costa Rican coast: Punta Pleito (A), Caño Island: Caño 1 (B), and Cocos Islands: Punta María (C). In bar charts: striped bars = clones; solid bars = unique multi-locus genotypes (MLGs). In polar plots: unique MLGs are represented by unique symbols. Ramets of the same genet share symbols. Radial axis shows distance in 5 m increments and the angular axis is given in 30° increments. Colony symbols were scaled by colony area (range: 35cm2 – 20.4 m2)...... 63

Figure 3-5. Species-level differences were evident in mussel density and amount of bleaching in each colony. (A) Mean mussel density (number of mussels boreholes per cm2 of tissue) ± s.e.m. in Porites evermanni and Porites lobata based on photographic and field analyses. The mean mussel density was higher (t-test, log10 transformed) in P. evermanni than in P. lobata. (B) Polar plot depicting location, species, and bleaching status for each colony sampled at Caño 2. Radial axes (in 30° increments) show distance (m) in 5 m increments. Symbol indicates species. Amount of tissue bleached (%) is represented by fill. (C) Proportions of bleached (white), healthy (green/brown), and pale (yellow) tissue in photos of P. evermanni (Pe) and P. lobata (Pl) based on visual inspections of tissue underneath 36 uniformly spaced points on each image. The mean percent of bleached tissue was higher (t-test, arcsine square root transformed) in P. lobata than P. evermanni...... 64

Figure 4-1. Probability of distribution of Porites lobata (A) and Porites evermanni (B) across the Eastern Tropical Pacific resulting from maximum entropy modeling in MaxEnt of 5 environmental variables: minimum, maximum, and average temperature (divided into El Niño and La Niña year), irradiance (divided into wet and dry seasons) and cholorphyll-a (divided into wet and dry seasons). Probability is given by color of cells (see inset in lower left corner of panel B) with red representing the highest probability and dark green representing the lowest probability of occurrence...... 93

Figure 4-2. (A) The probability of presence of Porites lobata given different minimum temperatures during La Niña compared to Porites evermanni in the background predicted by MAXENT. Black line=mean response of 3 replicates for P. lobata, Blue shading=standard deviation; Red line= mean response of 3 replicates for P. evermanni, Light gray shading= standard deviation; Dark gray horizontal line=expectation under null model. (B). The probability the presence of Porites evermanni given different average temperatures during El Niño compared to Porites lobata in the background predicted by MAXENT. Red line= mean response of 3 replicates for P. evermanni, Blue shading=standard deviation; Black line=mean response of 3 replicates for P. lobata, Dark gray horizontal line=expectation under null model. Note: These graphs indicate the marginal effect of changing exactly one variable, whereas the full model takes advantage of sets of variables changing together. Temperature is given in °C...... 94

Figure 4-3. (A)The residuals verses fitted values predicted by multiple linear regression model using a fragmentation value, depth, latitude, maximum annual temperature, and minimum annual temperature to predict the proportion of P. evermanni verses P. lobata at a sampling site in the ETP. Red line=best fit line (B) A normal Q-Q plot of the residuals. The diagonal represents a normal distribution. Shapiro-Wilk normality test; W=0.97; p-value=0.75...... 95

Figure 5-1. Continuous thicket of staghorn Acropora at Cocos Lagoon (A) Map of Study Sites (B)...... 132

Figure 5-2. Left: 15m radius circular plots mapping the location and genotype of each Acropora cf. pulchra colony randomly sampled in the plot. Symbol and color combinations indicate individual genotypes. One genotype was repeated between plots (indicated by red arrows). Right: Correlograms, resulting from a spatial autocorrelation analysis within plots, which depict the correlation coefficient (r) plotted against geographic distance with 7 distance classes of 3m (solid line). Dashed lines represent 95% confidence intervals about the null hypothesis of no spatial genetic structure (i.e. genotypes are distributed randomly across geographic space). Correlograms are considered to be significant at p<0.01 following (Banks & Peakall 2012)...... 133

Figure 5-3. Correlogram depicting correlation between pairs of individuals within distance bins graphed against geographic distance given in meters (A) or log(1+meter) (B) between individuals (solid line). Dashed lines indicate 95% confidence intervals about the null hypothesis of no spatial genetic structure (i.e. genotypes are distributed randomly across geographic space). Correlograms are considered to be significant at p<0.01 following (Banks & Peakall 2012). Black: with clones; Red: without clones ...... 134

Figure 5-4. Principle coordinates analysis (PCoA) of pairwise FST distance matrix resulting in 3 clusters (A). The separation of Cocos Lagoon from the remaining sites given by PCo1 explains 93% of the variation in the data while the separation of Saipan from Guam given by PCo2 explains 4% of the variation. Solid circles represent clustering based on significant FST while the dashed circle represents most likely clusters based on Bayesian assignment. Results of Structure analyses (B) plotted as bar graphs of the probability of membership (y-axis; PM) to a given cluster for each individual along the x-axis. Top: No admixture model including multiple Acropora spp. gives K=3 as most likely number of clusters with all Acropora cf. pulchra grouped as one cluster. Middle: The most likely number of clusters (K=2) based on Evanno et al. method with Cocos Lagoon having high probability of belonging to a second cluster and Saipan and Tumon samples unresolved. Bottom: Clustering assuming K=3 based on FST results. Saipan has moderate probability of membership to a third cluster and Tumon is largely unresolved...... 135

Figure B-1. Allele frequencies for Porites lobata (Pl) and Porites evermanni (Pe) at 11 microsatellite loci. Circles represent the presence of an allele and are scaled by the frequency of that allele in the data set. Allele size (bp) is given on the x axis. Species are given on the y axis...... 149

Figure B-2. Denaturing gradient gel electrophoresis (DGGE) gel image. Symbiodinium ITS2 sequences from eight representative Porites evermanni (left) and eight Porites lobata samples (right) were run on a denaturing gradient gel with a ladder and two C15 positive controls (+). All samples show subclade C15 as their dominant symbiont as evident by the bands to the left of the arrow. * indicates that the band to the left was excised, re-amplifed, and Sanger sequenced. All sequences aligned with 100% identity with a published C15 ITS2 sequence (GenBank AY 239369.1)...... 150

Figure C-1. Jackknife analysis of gain for Porites evermanni (A) and P. lobata (B) conducted in MAXENT. Comparison of the teal bar to the red bar shows the reduction in gain (y-axis) when each environmental variable (x-axis) is removed from the model. Comparison of the blue bar to 0 depicts the increase in gain when that variable is used in isolation...... 153

Figure C-2. The mean frequency of Porites evermanni in total Porites coral collections sites where the triggerfish, Pseudobalistes naufragium, is absent verses sites where P. naufragium is present. p<0.001; error bars represent standard error...... 154

Figure C-3. Box and whisker plots of sampling depth (m) of sites where P. evermanni can be found in comparison to sites that are exclusively colonized by P. lobata. Y- axis depicts visual cross-section into the water column with 0=surface...... 155

Figure D-1. (A) OBSTRUCT pairwise comparisons of the canonical scores of the correlation coefficients (R2) between sampling sites. Lowercase letters indicate differences between sampling site based on high correlation between ancestry and geography (R2>0.6). (B) Canonical discriminant analysis (CDA) plot generated by OBSTRUCT based on STRUCTURE ancestry results. Percentages on axes indicate the amount of variation in the data explained by each canonical score...... 160

xii

Figure D-2. Average genotypic diversity (GO/GE) verses genotypic evenness among sampling plots for 5 coral species from values reported in literature. APAL=Acropora palmata (Baums et al. 2006a), APUL= Acropora pulchra, PLOB= Porites lobata (Boulay et al. 2012), PEVE= Porites evermanni (Boulay et al. 2014), PDAM=Pocillopora damicornis (Pinzón et al. 2012). P. damicornis plots were divided by habitat type: P=protected E=exposed, because significant differences were found in Pinzón et al. (2012)...... 161

Figure D-3. Current data (A) given as 21-year mean (1993-2013) obtained from NOAA ocean surface current analyses (http://www.oscar.noaa.gov). The predominant current surrounding Guam and Saipan is the North Equatorial Counter Current (red) flowing NW at a rate of approximately 0.1-0.3m/s. Net local currents (B) at 11 sites (A-K) around Guam shown as arrows (in Wolanski et. al 2003). Results of a numerical model (C) predicting eddy formation off the southern tip and leeward side of Guam (in Wolanski et. al 2003)...... 162

Figure E-1. Porites lobata population structure across the central and Eastern Tropical Pacific. Pie charts show the average probability of membership of individuals sampled at that site in one of five clusters as identified by Structure (top bar chart). The size of the pie charts is relative to the sample size (n) collected at each location. Results of clustering assuming K = 5 to K = 2 are shown in descending order...... 195

Figure E-2 Porites lobata: Frequency of inbreeding-adjusted null alleles (+ 1 SD) over 12 loci across the central Pacific (CP), the Hawaiian Archipelago (HI) and the Eastern Tropical Pacific (ETP). A Kruskal-Wallis one-way analysis of variance indicated differences in null allele frequency within loci across regions (asterisk, p <0.05)...... 196

Figure E-3. Porites lobata mean allelic richness (AR(20)) and private allele richness (AP(20)) rarefied to a sample size of three sites per region and 20 genes per site. A Kruskal-Wallis one-way analysis of variance indicated differences across regions (p < 0.01) for both AR(20) and AP(20). CP = Central Pacific, HI = Hawaiian Islands, ETP = Eastern Tropical Pacific...... 197

Figure E-4. Isolation by Distance patterns in Porites lobata. Geographical distance explained 7% of the variation in genetic distance (Fst) across all sampling site, none of the variation in the western Central Pacific (CPW, B), and 35% of the variation in the eastern Central Pacific (CPE,C). Geographical distance explained 43% of the variation in genetic distance among all sites in the Eastern Tropical Pacific (ETP, D, black circles and regression line) and 7% of the variation when excluding Clipperton (D, grey triangles and dotted regression line)...... 198

Figure E-5. Principal component analysis of allele frequency covariance in Porites lobata populations. 31 out of 229 PCA-axes were retained, explaining 100% of the cumulative variance. Plotted are the first two axes explaining 28.9% (p < 0.05) and 47.73% (p<0.05) of the variance, respectively. Central Pacific West (CP (W), circles), Central Pacific East (CP (E), stars), Hawaii (HI, diamonds), Eastern Pacific (EP, triangles)...... 199

xiii

Figure E-6. Mean log-likelihood (A) and Delta K (B) values of K for STRUCTURE analysis of Porites lobata samples pacific-wide...... 200

Figure E-7. When treated as genotypes of unknown origin, four Clipperton genotypes (red arrow) assigned with high probability (mean CP = 0.84 +/- 0.09) to the central Pacific (K = 3) and one genotype appeared admixed between CP (assignment probability = 0.36) and EP (0.62). Structure identified 7 genotypes from the Central Pacific (1 from PH01, 4 from JO01, 2 from LN04) and one genotype from the eastern Pacific (CR03) as first generation migrants (black arrows) with high probability (> 0.9, p < 0.001). Model assumed admixed populations, correlated allele frequencies, and K = 3...... 201

Figure F-1. Porites lobata population structure across the central and Eastern Tropical Pacific assuming no admixture among populations. Graph shows the average probability of membership of individuals sampled at that site in one of five clusters (K = 5) as identified by STRUCTURE...... 206

Figure F-2. Porites lobata population structure across the central and Eastern Tropical Pacific analyzed with eight of twelve loci. Graph shows the average probability of membership of individuals sampled at that site in one of five clusters (K = 5) as identified by STRUCTURE...... 207

Figure F-3. Mean log-likelihood (A) and Delta K (B) values of K for STRUCTURE analysis of Porites lobata samples Pacific-wide using only eight loci...... 208

Figure F-4. Principal component analysis of allele frequency covariance in Porites lobata using only eight of the loci. 147 out of 156 PCA-axes were retained, explaining 100% of the cumulative variance. Plotted are the first two axes explaining 14.67% and 7.91% of the variance, respectively. Central Pacific West (CP (W), circles), Central Pacific East (CP (E), stars), Hawaii (HI, diamonds), Eastern Tropical Pacific (ETP, triangles)...... 209

xiv

LIST OF TABLES

Table 1-1. Published examples of purpose, extent, and organism studied using MAXENT. Source: Elith et al. 2010...... 10

Table 2-1. Genotypic diversity in Porites lobata within and between coastal and oceanic sites (one plot per site) in the Eastern Tropical Pacific...... 37

Table 2-2. Genetic diversity in Porites lobata within and between coastal and oceanic sites (one plot per site) in the Eastern Tropical Pacific...... 38

Table 4-1. Percent contribution and importance of each environmental variable used by MAXENT to model the probable species distributions of Porites lobata and Porites evermanni in the ETP...... 91

Table 4-2. Coefficients (±standard error), significance, and importance of each factor selected in MLR model using SPSS model builder to predict the relative proportion of Porites lobata and Porites evermanni at sampling sites in the ETP...... 92

Table 5-1. GPS coordinates, sampling method and number of Acropora cf. pulchra sampled and genotyped at each sampling site. After genotyping individuals missing data at >3 loci were removed (NM). The number of unique genotypes (NG) by sampling location is also given. GPS locations in decimal degrees...... 125

Table 5-2. Mean and SE sample size of alleles (N), number of alleles (A), number of effective alleles (NE), Shannon diversity index (I), observed heterozygosity (HO), expected and unbiased expected heterozygosity (HE and uHE), and fixation index (F) 2 2 across all loci for each site. NE = 1 / Σpi ; HE = 1 – Σpi ; uHE = (2N / (2N-1)) * HE; F = (HE – HO) / HE); Where pi is the frequency of the ith allele for the population. See supplement for individual locus statistics by site. SE = standard error...... 126

Table 5-3. Mean null allele frequencies by locus and average inbreeding (FI) by site as estimated in INEst v 2.0 (Chybicki & Burczyk 2009)...... 127

Table 5-4. Indices of clonal structure as calculated for each polar plot. Sample size (N), the number of unique MLGs (NG), clonal richness (NG/N), genotypic diversity (GO/GE), genotypic evenness (GO/NG), the average number of ramets (R) per genet (G) is given as well as the average distance between clonemates (Genet Spread), the average maximum linear extent of each genet (Clonal Extent) meters, the average clonal identity for all pairwise comparisons (Psg), the average clonal identity of nearest neighbor (Psp), the aggregation coefficient (Ac), and the reef status. SD = standard deviation...... 128

Table 5-5. Omega values (below diagonal) and associated p-values (above diagonal) from heterogeneity tests to compare spatial autocorrelation in 7 distance classes of 3m between each pair of plots. Following Banks and Peakall (2012), it is recommended that significance of the heterogeneity test is declared when p<0.01 (indicated by underlined italics)...... 129

Table 5-6. Pairwise FST by site. Significantly non-zero values after 420 permutations (p<0.002 after adjustment for multiple comparisons) are given in underlined italic...... 130

Table 5-7. Log Bayes factors (2* difference in log-likelihood) based on Bezier approximation score (a) and the estimated number of migrants per generation (b-d) calculated as the median estimated mutation-scaled population size (Θ) x the median estimated mutation-scaled migration rate / 4. The most likely model (N) for each configuration of populations is underlined. N = northward migration, S = southward migration, N + Cocos L. to NG = a channel of migration from Cocos Lagoon to northern Guam sites in addition to southern Guam, 0 Cocos L. = full model except no migration to and from Cocos Lagoon, 0 Saipan to S = full model but no migration from Saipan to the south...... 131

Table B-1. Porites lobata (Pl) and Porites evermanni (Pe) samples (n = 684) were obtained from three regions (North, Central, and South) and 17 sites in the Eastern Tropical Pacific using haphazard (H) and/or random (R) sampling methods. Sites are arranged in approximately west-to-east and north-to-south order. Given is the ratio of the number of unique multilocus genotypes by species (genets; NgPl, NgPe) over the total sample size (NPl, NPe). GPS locations are in decimal degrees (WGS84). The number of polar plots per site is given where applicable. See Table B-3 for polar plot level information...... 146

Table B-2. Microsatellite loci for Porites lobata (Pl) and Porites evermanni (Pe). The primer sequences are preceded by the name of the fluorescent dye used (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in four multiplex and one singleplex reaction (Plex) using the annealing temperatures (Temp) indicated...... 147

Table B-3. Indices of clonal structure as calculated for each polar plot. Colony sizes were estimated as maximum length multiplied by maximum width from measurements taken in the field. Go = same as Ne, Ge = N. The average number of ramets (R) over the number of genets (G) is given as well as average colony size...... 148

Table D-1. Additional Acroproa spp. samples used in analysis ...... 156

xvi

Table D-2. Microsatellite loci for Acropora cf. pulchra. The fluorescent dye used to label each primer is given (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in two multiplex and three singleplex reactions (Plex). Acr53 and Acr1_4 were run with PET and 6FAM...... 157

Table D-3. Heterozygosity, F-statistics, and deviations from HWE for each locus by site. Significant pHWE after correction for multiple testing given in italics and underlined. Sample size (N), number of alleles (Na), number of effective alleles (Ne), Shannon diversity index (I), observed heterozygosity (Ho), expected heterozygosity (He), and fixation index (F), probability of HWE (pHWE)...... 158

Table D-4. Allele frequencies by locus and site. Frequencies highlighted using underline and italics denote unusual alleles occurring at high frequencies within the Cocos Lagoon samples...... 159

Table E-1. Porites lobata samples (n = 1264) were obtained from three regions (CP = Central Pacific West (W) and East (E), HI = Hawaii, ETP = Eastern Tropical Pacific) and 33 sites. Sites are arranged in approximately west-to-east and north-to- south order. Given are total sample size (N), the number of unique multilocus genotypes (genets; Ng = 139), and the ratio of genets over samples collected (Ng/ Ng).. GPS locations are in decimal degrees (WGS84)...... 192

Table E-2. Summary of per locus statistics based on 12 microsatellite markers for Porites lobata. Na = number of alleles, Neff = number of effective alleles, Ho = observed heterozygosity, Hs = heterozygosity within populations, Ht = total heterozygosity, H’t = corrected total heterozygosity, Gis = inbreeding coefficient, Gst = fixation index, G’st (Nei) = Nei’s corrected fixation index (Nei 1987), all values are significant at p < 0.01, G’st (Hed) = Hedrick’s corrected fixation index (Hedrick & Goodnight 2005), Dest = Jost’s differentiation index (Jost 2008). SE = Standard errors obtained through jackknifing over loci. All values calculated with GENODIVE (Meirmans 2006)...... 193

Table E-3. Population differentiation among 32 sites (A) and 3 biogeographic regions (B) of Porites lobata. Based on an Analysis of Molecular Variance (AMOVA) calculated assuming an infinite allele model (equivalent to Fst). CR04 was excluded due to low sample size (Ng = 1). SE = Standard Error. F’ is a standardized version of Fst (Meirmans 2006)...... 194

Table F-2. Porites lobata null allele frequencies and individual inbreeding values. Given are averages over individuals per site and the lower 0.05 and upper 0.95 Confidence intervals (CI). MS01 and SA03 were the only sites were the Lower CI of Fi was different from zero...... 203

xvii

Table F-3. Pairwise Fst values (lower diagonal) and their significance (upper diagonal) among P.lobata sampling sites. CR03 had only one sample and was therefore excluded...... 204

xviii

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my advisor Dr. Iliana Baums for her support and guidance over these past five and a half years. Next, I want to extend my sincerest gratitude to Dr.

Katriona Shea for acting as the chair of my committee and for always keeping your door open to discuss issues afflicting female scientists. I am incredibly fortunate to have two extraordinary women as the lead members of my committee and as strong role models for my scientific career. I also would like to thank two members of my committee, Drs. Todd LaJeunesse and Jorge Cortés, for lending their expertise to me and allowing me to work in their labs with their students. Both of you were instrumental to me in completing my thesis. Todd, thank you and your lab for all your help with the Symbiodinium portion of my thesis. Don Jorge, thank you for hosting us in Costa

Rica and for coming to Penn State to give a seminar. I will always treasure my time in Costa

Rica. I must also thank Dr. George Perry for serving on my committee. I enjoyed your enthusiasm and it was a pleasure to have you in committee meetings and exams.

The work I present here would not have been possible without my many collaborators. I would like to thank Dr. Michael Hellberg for his help with the Porites project, for quickly responding to my many emails filled with questions, and for your elegant edits to the multiple drafts of the manuscript. It would not have been as successful without you. I must also thank Drs.

Andy Halfourd and Jenny McIlwain for the opportunity to travel to Guam and complete the

Acropora connectivity project. It was the prospect of projects like that which initially got me excited about graduate school. Last, I must thank Jaime Nivia-Ruiz for escorting me all over

Costa Rica and working so hard to help advance my career and our understanding of corals in the

ETP. I will never forget your kindness and all that you risked to help me with my thesis.

I would not have been able to complete this work or present it at national and international conferences without the help of my funding sources. I would like to thank the Penn

xix

State Department of Biology for awarding student travel grants, the Jeanette Ritter Mohnkern

Graduate Student Scholarship in Biology, and a Braddock Supplement to me. The Guam project was only possible due to a NOAA grant to Drs. McIlwain and Baums and a summer fellowship awarded to me from the Penn State Center for Environmental geoChemistry and Genomics. All the work on Porites was made possible by NSF grants awarded to Drs. Baums and Hellberg and undergraduate research grants from the Penn State Eberly College of Science. Last, I want to thank the Society for the Study of Evolution and NSF for student travel grants which allowed me to present the results from the PRSB paper at the Evolution meeting in Ottawa and the Ecological

Genomics Symposium in Kansas City.

I must also thank my fellow Baums’ Lab-ers: Meghann, Dannise, John, Andie, Nick, and

Sharm for always listening, helping to troubleshoot problems, and providing unfettered emotional support. To all the undergraduate students that helped collect data for me: Olivia, Rene, Caitlin,

Kevin, Devon, and the students from BIOL499A, I truly could not have done it without you. You all have a very bright future in science and I hope you enjoyed working on this project as much as

I enjoyed working with you.

My deepest appreciation goes out to the entire Penn State Biology department including my fellow PSU biograds, the faculty, and staff. The incredible women in the business and main office are the most helpful people I have ever met. The department would most certainly cease to operate without them. I must also thank Drs. Chuck Fisher and Doug Crawford for relentlessly striving to make positive changes and improve the working environment for all members of the department. I feel honored to work alongside the best group of faculty at Penn State. Thank you all for creating and encouraging conversation between faculty and students. We learn so much outside of the classroom and lab through these discussions. I also need to thank those department members that helped me pursue my passion for teaching. The professors of the Biology Core

Courses, Drs. Denise Woodward and Carla Hass, not only instruct thousands of undergraduate

xx students each year but they always made time to mentor me and help me grow into the best instructor I could be. I am also incredibly grateful for the opportunities to teach in the field created by Drs. Iliana Baums and Jim Marden.

Last, the love and support that I received from my family throughout this whole process surpassed all expectations. Mom, Dad, and Becky, thank you for being so incredibly understanding during the long periods of neglect when I was “too stressed” or “too busy”. I am aware that I often left you in the dark about what exactly I was doing. Even though I know you were fighting your curiosity, I greatly appreciated the ability to get away from work when I came to visit. You gave me a sanctuary to focus on what really matters in life. That gift is priceless and was exactly what I did want. To my incredible fiancé, Christian, I would not have survived this last year without you. You were at my side for every step of this journey and I am so grateful for your gentle urging forward while also catching me when I fell back. There is no one with whom I would rather walk through life. I cannot forget to thank a great friend, Dr. Lindsay Beck-Johnson, for giving me so much encouragement and guidance and never asking for anything in return.

Thank you for going first and sharing your experience and knowledge with me. Without your incredible footsteps to follow, I would have been lost.

Chapter 1

INTRODUCTION

Corals reefs are among some of the world’s most biologically diverse ecosystems. Built upon the framework of reef building corals, the reef ecosystem contains representatives from many of the known marine taxa. In the face of increasing human populations and the changing global climate, some coral communities may be decimated in a rather short period of time

(Hughes et al. 2003, De’ath et al. 2012). Coral associated invertebrates and vertebrates alike might be lost during disturbance events. Thus, it is of great interest to study the dynamics of these threatened populations of reef corals to bring us closer to understanding how to effectively restore populations and prevent the loss of an economically and ecologically valuable ecosystem.

Background

Life History

Scleractinian (hard) corals build the physical structure of the coral reef ecosystem. As an adult, the coral animal (Phylum: Cnidaria) is sessile or attached to the substrate (Fig. 1-1). The adult releases gametes or larvae into the water column where they drift on ocean currents and have the ability to disperse to more distant reefs before they settle and undergo metamorphosis

(Fig. 1-1) (Fadlallah 1983, Richmond & Hunter 1990). Through asexual reproduction and budding, a single coral polyp develops into a coral colony. The colony deposits a calcium carbonate skeleton as it grows. Through chemical and physical processes the skeleton gets cemented into the structure we know as a coral reef.

Figure 1-1. Coral life history and reproduction. Asexual reproduction by fragmentation (Left; green) and sexual reproduction by broadcast spawning (Right; blue)

In corals, the adult phase is characterized by occasional fragmentation resulting in asexually produced clones known as ramets (Fig. 1-1) (Highsmith 1982, Baums et al. 2006;

Foster et al. 2007). The recombination of genes during sexual reproductions produces a genetically unique individual known as a genet (Fig. 1-1). Because coral larvae are dispersed into the water column, patterns of larval transport are difficult to predict due to physiological capabilities of the larvae and their interaction with the physical constraints of the environment.

This combination of sexually produced pelagic larvae and asexual fragments produces complex patterns of dispersal, distribution, and relatedness in individuals and populations of these species.

These patterns and the relative impacts of each mode of reproduction are not well described in coral systems. However, effective management and conservation strategies for coral reefs require knowledge of the consequences of dispersal and recruitment on ecosystem processes.

Symbiosis

The reef ecosystem comprises a high degree of biodiversity creating a complex network of relationships. One of the most intimate biotic interactions on the reef is between a singled

3 celled alga in the genus Symbiodinium and the coral animal. The coral harbors the Symbiodinium cells in its tissues and keeps them in the photic zone where they can photosynthesize. The alga transfers the photosynthetic products to the coral and the coral provides nutrients for photosynthesis to the symbiont (Muscatine & Hand 1958). This relationship is obligate for the coral and the association is only maintained within a narrow temperature range. A few degree shift in water temperature is enough to cause the breakdown in this relationship and as a result the

Symbiodinium cells may leave the coral tissues in a process called coral bleaching (Glynn 1993,

1996, Brown 1997).

Importance

In addition to providing the foundation for an ecosystem that harbors high biological diversity (Done et al. 1996), coral reefs serve as valuable resources for many basic human needs

(Moberg & Folke 1999). Coral reefs protect shorelines (Cesar 1996) and supply building materials (Dulvy et al. 1995, Cesar 1996) and food (Smith 1978) to coastal populations and are a gold mine for the discovery of new pharmaceuticals (Carté 1996, Haefner 2003). Tourism and fishing provide economic support for people in coastal areas such as the Pacific Islands and the

Florida Keys (Dixon 1993, Pendleton 1995). As the climate changes more rapidly, there is a possibility for range contractions as coral populations currently in suboptimal conditions disappear as a result of further deteriorating conditions (Glynn 1997, Glynn & Ault 2000). Little can be done to mediate these responses to stress without understanding the dominant processes responsible for species distributions and population maintenance within the coral community.

4 Study systems and Methods

Eastern Tropical Pacific Porites

For most of my thesis, I focus on the dominant, Pacific-wide, reef-building massive coral,

Porites lobata, in a marginal environment, the Eastern Tropical Pacific (ETP), where conditions for reef growth are suboptimal (Cortés 1997, Glynn & Ault 2000) and physical isolation from the remainder of the Indo-Pacific results in no input from the center of the species range (Baums et al. 2012). Starting in 2006, researchers at Penn State (Baums) and Louisiana State (Hellberg)

Universities began an extensive field collection, funded by NSF, of Porites lobata to analyze gene flow across its range. Sixteen localities were sampled throughout the Pacific with a focus on the Central and Eastern Pacific (Fig. 1-2). These sites span several latitudinal and longitudinal gradients in abiotic conditions such as temperature, currents, and water chemistry. Over 2000

Porites collections are available. Ecological data and colony measurements were recorded where possible. Each individual was genotyped at 14 microsatellite loci and assigned a multi-locus genotype. The dataset was then analyzed to detect population genetic structure and quantify gene flow among populations (see Appendix E Baums et al. 2012). There has been no recent gene flow across the East Pacific Gap (Fig. 1-2) thereby isolating the Eastern Pacific population of Porites lobata from the remainder of the Pacific (Baums et al. 2012).

Figure 1-2. Geographical Range Porites lobata in the Pacific Ocean. Sites sampled for Pacific wide analysis of Porites lobata are indicated by red circles. Source: Dr. Iliana Baums

Marginal environments at the edges of the species ranges are characterized by low species diversity (Sagarin & Gaines 2002, Johannesson & Andre 2006, Eckert et al. 2008), precarious environmental conditions, and geographic isolation. Climate change might extend these conditions to more central locations and thus marginal habitats are a test bed for how future reefs might look. Environmental conditions are suboptimal for reef growth because of the limited shallow water habitat, fluctuating seawater temperatures due to seasonal upwelling, high sedimentation rates and low aragonite saturation state (Cortés 1997, Glynn & Ault 2000). The combination of suboptimal environmental conditions and isolation has spurred interest in how these reefs persist and if they may provide insights into the fate of reefs elsewhere in the face of a rapidly changing climate.

Diversity is low in the Eastern Tropical Pacific (ETP) compared to the center of coral species diversity in the Indo-West Pacific (Glynn & Wellington 1983, Glynn & Ault 2000, Veron

2000). Plastic morphologies make species identification difficult over wide species ranges and morphologies may converge especially in marginal habitats (Pinzón & LaJeunesse 2011). The taxonomic status of many coral species in the region is unclear. Previous observations of local

6 variation in reproduction and multi-copy ITS (Internal Transcribed Spacer) markers suggests the possibility of cryptic species within Porites in the ETP.

Use of microsatellite markers

Much attention in coral conservation has been given to conserving diversity. A central dogma in ecology is that higher diversity confers high resistance and resilience of communities to stress (Naeem & Li 1997, Tillman et al. 2006, Downing et al. 2012). In my thesis, I use microsatellite genotyping, clustering, and models to describe multiple levels of diversity including functional, clonal, genetic, and species diversity. I examine the differences in genotypic structure and gene flow of reef-building corals, identify cryptic species, and demonstrate functional differences in how unrecognized species interact with their associated fauna and abiotic environment.

Sensitive recognition of population clusters and inference of changes in recent levels of dispersal can be detected with highly polymorphic markers such as microsatellites (Rosenberg et al. 2001). Microsatellites are tandemly repeated DNA sequences 2-4 basepairs in length with high mutation rates resulting in different numbers of repeats. Genotypes at a given locus can thus be scored by size. When several (10-15) markers are employed together to create a multi-locus genotype (MLG), individual-level resolution can be achieved (Queller et al. 1993, Selkoe &

Toonen 2006). These loci have a high number of possible alleles thus there are many combinations that could be seen in the population. This allows us to tell genets apart when using roughly 10 markers. If matches are seen at all loci the colonies are assumed to be ramets.

Although these markers are designed for a given study species, cross amplification over closely related species has been shown to aid in species delimitation when morphological markers are unclear. Diagnostic private alleles at microsatellite markers, monomorphic in one species,

7 correspond with 100% congruence to mitochondrial barcoding genes in fishes (Vanhaecke et al.

2012). However, corals have a very slowly evolving mitochondria genome, thus barcoding using mitochondrial genes is not useful for identification of species. Case studies show that clustering of microsatellite markers using Gaussian and Bayesian methods for species delimitation corresponds well with species delimited based on other data (Hausdorf & Hennig 2010). Here, I used clustering methods to identify cryptic species of corals with no recent reproduction at sympatric sites.

Understanding biotic interactions

As ecosystem engineers, corals have a high number of interactions. To identify specific

ITS-2 type of endosymbiont, I used denaturing gradient gel electrophoresis (DGGE). DGGE works to differentiate between DNA fragments of the same length using a chemical to denature the DNA (Abrams & Stanton 1991). The denaturing process occurs at different denaturing conditions depending on the domains found in the DNA sequence. Therefore, each ITS-2 sequence shows a unique DGGE fingerprint (Fischer & Lerman 1983). When ITS-2 fragments from unknown samples are run on the same gel with previously sequenced standards, DGGE fingerprints can be compared side-by-side on the same gel to identity ITS-2 type (LaJeunesse

2002). Select bands can also be excised for the gel and sequenced using Sanger sequencing.

To identify functional diversity in the complex network of external and endolithic biotic interactions, I used a combination of museum records, field surveys, genetic analysis, photograph analysis, and in situ observations to describe interactions between corals and their associated community. To determine whether biotic interactions play a role in the distribution of their associated coral species, I created a data set comprised of environmental and biotic data at each

8 sampling site and created a linear regression (MLR) model to identify important predictors of species distribution.

Species distribution models

In addition to the MLR model, I utilized species distribution modeling (SDM) to describe differences among coral species in response to environmental variables important for maintaining coral-endosymbiont association. Modern SDM combines site-based ecology and GIS to explore the drivers of species distributions, make predictions to unsampled sites, gain evolutionary insight, and predict distributions under forecasted conditions (see Table 1-1 for examples) (Elith

& Leathwick 2009, Elith et al. 2011). These models have been used in terrestrial, freshwater, and marine environments at varying spatial scales (see Table 1-1 for examples) (Elith & Leathwick

2009, Elith et al. 2011).

SDM requires only presence data and uses these records to compare environmental conditions against background points or pseudo-absences (Elith & Leathwick 2009). The model is affected by the patterning in presence-only records will suggest that area is unsuitable.

However, species are not perfectly detectable. When detectability varies from site to site, the pattern of presence is affected by false absences. Therefore, it is important to note that dispensing with absences by using presence-only data does not address the limitations of absence data.

Organisms may not occupy all suitable habitats. Absences (and pseudo-absences) can be affected by biotic interactions, dispersal constraints and disturbances not necessarily related to environmental conditions alone.

In additional to presence records, SDM requires environmental data across the study area.

The environment is simple to characterize for a sessile organism, such as plants and corals, but not for highly mobile organisms where environmental conditions vary across range and sampling

9 site may not be an accurate reflection of environment (Elith & Leathwick 2009). For sessile organisms, geographically separated quadrats are regarded as statistically independent samples

(Elith & Leathwick 2009). Here, we sampled in circular plots (see Baums et al. 2006) or haphazardly at sampling sites separated by at least roughly 100m. The average distance between sites sampled at the same location or island was 3 km.

For SDM, sampling scale is important to consider (Elith & Leathwick 2009). The grain of sampling should be consistent with environmental data (Levin 1992) although this not always feasible and often overlooked in practice (Elith & Leathwick 2009). Researchers commonly impose scale based on data available and the structure of their chosen model the effect of which depends on the accuracy of the data, the environment, study species, and application (Ferrier &

Watson 1997). Predictors can vary and/or have effects on biotia at different spatial scales. These effects often go untested and undiscussed (Elith & Leathwick 2009).

SDM assumes equilibrium with environment (Elith & Leathwick 2009). Furthermore, environmental factors may limit distribution and biotic interactions may change substantially in new context (Elith & Leathwick 2009). Furthermore, the models are criticized because they cannot take into consideration dispersal pathways (Elith & Leathwick 2009). Incorporation of biotic interactions is also difficult and thus most practitioners use abiotic predictors alone (Elith &

Leathwick 2009). Consequences of ignoring biotic interactions are considered relatively benign except in cases of extrapolations where the effects of competitors, mutualists, and conspecific attractions might have far reaching effects where novel combinations of species are likely to occur. Species distribution could also be influenced by genetic variability, phenotypic plasticity, and evolutionary changes (Elith & Leathwick 2009).

When using SDM, it is important to select functionally/ecologically relevant predictors

(Elith & Leathwick 2009). Indirect predictors (e.g. depth is indirect proxy for temperature, salinity, light, pressure, and resource availability) should be avoided (Elith & Leathwick 2009).

10 Thus, we spent time carefully selecting variables (see Methods: Predictor selection in Chapter 4).

SDM is argued to model the environmental realized niche and are sometimes referred to as ecological niche models. However, a consensus has not been achieved when trying to define what is being modeled and thus, the neutral terminology of SDM is preferred (Elith & Leathwick

2009).

The specific SDM employed in my thesis is a program that utilizes maximum entropy modeling (MAXENT) developed by Phillips (2004, 2008). MAXENT became available 2004 and is used widely in modelling species distributions with diverse aims such as finding correlates of species occurrences, mapping current distributions, and predicting to new times and places (Table

1-1).

Table 1-1. Published examples of purpose, extent, and organisms studied using MAXENT. Source: Elith et al. 2011. Primary Purpose Extent Organism Reference Predict current distributions as input for Andes Humming- Tinoco et al. (2009) conservation planning, risk Global birds Tittensor et al. (2009) assessments or IUCN listing, or new Stony corals surveys seamounts Understand environmental correlates of Norway Macrofungi Wollan et al. (2008) species occurrences, groups of Portugal European Monterroso et al. (2009) species, or other wildcat Predict potential distributions for invasive New Zealand Ants Ward (2007) species, or explore expanding China Nematode Wang et al. (2007) distributions Predict species richness or diversity California Amphibians Graham & Hijmans (2006) and reptiles Brazil Myrtacae Murray-Smith et al. (2009) 19 species Predict current distributions for Global Seaweeds Verbruggen et al. (2009) understanding morphological/ genetic Andes Birds Young et al. (2009) diversity (‘‘phylogeography’’, Madagascar Bats Lamb et al. (2008) ‘‘phyloclimatic studies’’), endemism and evolutionary niche dynamics Hindcast distributions to understand NW Europe Pond snails Cordellier & Pfenninger (2009) patterns of endemism, vicariance, etc Brazilian coast Forests Carnaval & Moritz (2008) Forecast distributions to understand Mediterranean & Cyclamen Yesson & Culham (2006) changes with climate change / land surrounds transformation; includes retrospective Regional W. Banksia Yates et al. (2010) studies Australia Canada Butterflies Kharouba et al. (2009) Test model performance against other Patagonia Insects Tognelli et al. (2009) methods Local region in Rare plants Williams et al. (2009) California Regional to national Many Species Elith et al. (2006)

11 Data are supplied to the program as grids of covariates (predictors/environmental factors) relevant to habitat suitability and covering a pixilation of the geographic area of interest

(landscape) defined by the ecologist (Elith et al. 2011). MAXENT randomly samples 10,000 background locations from the covariate grids (could by chance include presence locations) and assumes that the sample of presence records is also random sample. Sample selection bias occurs when some areas in the landscape are sampled more intensively than others. This has a stronger effect on presence-only models than presence/absence models (Elith et al. 2011). After sampling, the program first estimates the relative suitability of one place verses another and determines what features are important using the conditional density of covariates at presence sites and unconditional density of covariates across study area (Elith et al. 2011). It uses the knowledge of prevalence to calculate a conditional probability of occurrence (Elith et al. 2011). Then it compares the probability densities in covariate space and estimates distribution across geographic space (Elith et al. 2011). It does not take geographic proximity of locations into consideration unless geographic predictors are used so spatially autocorrelated variables should be avoided

(Elith et al. 2011).

The fit of the model is evaluated using by a log likelihood function which describes the log of the probability of an observed outcome. The program considers the tradeoff between fit and complexity and fits a penalized (balances fit and complexity) maximum likelihood model to the data related to other penalties such as AIC (Elith et al. 2011). The fit of the model is given as the area under the receiver operating characteristic (ROC) curve (AUC) (Elith & Leathwick

2009). A gain bar is shown reporting the improvement in penalized average log likelihood compared to a null model.

12 Guam Acropora

Because recovery of coral reefs relies on dispersal and recruitment, reproduction and connectivity in the marine environment is of great interest. I had the opportunity to study clonal structure and gene flow in Acropora on reef flats locally in Guam. Marine populations are not as connected as previously assumed (Cowen et al. 2000, Kinlan & Gaines 2003) but dispersal barriers are often hard to predict and may vary among co-distributed species (Kelly & Palumbi

2010, Selkoe et al. 2014). Large geographic distances such as those that exist among islands in the Paciﬁc, might limit larval exchange among populations leading to regional isolation. Species that build the three-dimensional structure of ecosystems (foundation species) often reproduce asexually as well as sexually. In plants and corals, the contributions of asexual reproduction to population structure decrease genotypic diversity within species while at the same time increasing population size when conditions are not favorable for sexual reproduction. In the face of a rapidly changing environment it becomes increasingly important to understand how reproduction and gene flow act in concert to maintain coral diversity over time.

Again using microsatellite markers to genotype corals, identify clones, and Bayesian clusters to analyze population genetic, I examine gene flow and clonal structure within the branching coral genus Acropora on a local scale. While this study is locally important for the management of reefs on Guam, it also has broader appeal as the study species is found widely across the Pacific Ocean. I sampled using a repeatable spatially explicit method. Results can thus be compared across ranges and species to describe geographic or taxonomic differences in population structure in respect to the geographic range. Although this study has a simple design, it has the potential to not only impact local coral reef conservation and restoration efforts but also allows for comparisons across a wide geographic range and lays the ground-work for future studies on the effects of genotypic diversity on community structure.

13 Outline

In Chapter 2, I use microsatellite markers to describe the clonal diversity of Porites lobata in the peripheral ETP.

In Chapter 3, I test the hypothesis that Porites lobata constitutes one genetic/functional unit in the Eastern Pacific.

In Chapter 4, I use a species distribution model and a multiple linear regression model to evaluate niche differentiation and drivers of species distributions in two Porites holobionts.

In Chapter 5, I conduct an analysis of population genetics and clonal structure in

Acropora pulchra in Guam.

14 References

Abrams, E. S., & V. P. Stanton Jr. 1991. Use of denaturing gradient gel electrophoresis to study conformational transitions in nucleic acids. Method. Enzymol. 212: 71-104. Baums, I.B., M.W. Miller & M.E. Hellberg. 2006. Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr. 76: 503-519. Baums, I.B., J.N. Boulay, N.R. Polato & M. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Brown, B.E. 1997. Coral bleaching: causes and consequences. Coral. Reefs. 16: S129-S138. Carnaval, A.C. & C. Moritz. 2008. Historical climate modelling predicts patterns of current biodiversity in the Brazilian Atlantic forest. J. Biogeogr. 35: 1187–1201. Carté, B.K. 1996. Biomedical potential of marine natural products. BioScience 46: 271–286. Cesar, H. 1996. Economic Analysis of Indonesian Coral Reefs. The World Bank. Cordellier, M. & M. Pfenninger. 2009. Inferring the past to predict the future: climate modelling predictions and phylogeography for the freshwater gastropod Radix balthica (Pulmonata, Basommatophora). Mol. Ecol. 18: 534–544. Cortés, J. 1997. Biology and geology of Eastern Pacific coral reefs. Coral Reefs 16: S39-S46. Cowen, R.K., K.M.M. Lwiza, S. Sponaugle, C.B. Paris & D.B. Olson. 2000. Connectivity of marine populations: Open or closed? Science 287: 857-859. De'ath, G., K.E. Fabricius, H. Sweatman & M. Puotinen. 2012. The 27-year decline of coral cover on the Great Barrier Reef and its causes. Proc. Natl. Acad. Sci. U. S. A. 109: 17995- 17999. Dixon, J. A., Fallon Scura, L., & T. van't Hof. 1993. Meeting ecological and economic goals: marine parks in the Caribbean. Ambio 22:117-125. Done, T.J., Ogden, J.C., Wiebe, W.J., & B.R. Rosen. 1996. Biodiversity and ecosystem function of coral reefs. In:Mooney, H.A., Cushman, J.H., Medina, E., Sala, O.E., Schulze, E.-D. (Eds.), Functional Roles of Biodiversity: A Global Perspective. SCOPE 1996, John Wiley and Sons. Downing, A. S., van Nes, E. H., Mooij, W. M., & M. Scheffer. 2012. The resilience and resistance of an ecosystem to a collapse of diversity. PloS One 7: e46135. Dulvy, N.K., Stanwell-Smith, D., Darwall, W.R.T., & C.J. Horrill. 1995. Coral mining at Mafia Island, Tanzania: a management dilemma. Ambio 24: 358–365. Eckert, C.G., K.E. Samis & S.C. Lougheed. 2008. Genetic variation across species' geographical ranges: the central-marginal hypothesis and beyond. Mol. Ecol. 17: 1170-1188. Elith, J. & J.R. Leathwick. 2009. Species distribution models: ecological explanation and prediction across space and time. Ann. Rev. Ecol. Evol. Syst. 40: 677–697. Elith, J., Graham, C.H., Anderson, R.P. et al. 2006. Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29: 129–151. Elith, J., Phillips, S. J., Hastie, T., Dudík, M., Chee, Y. E., & C. J. Yates. 2011. A statistical explanation of MAXENT for ecologists. Divers. Distrib. 17: 43-57.

15 Fadlallah, Y. H. 1983. Sexual reproduction, development and larval biology in scleractinian corals. Coral Reefs 2: 129-150. Ferrier, S., G. Watson. 1997. An evaluation of the effectiveness of environmental surrogates and modelling techniques in predicting the distribution of biological diversity.Consult. Rep. NSW Natl. Parks Wildl. Serv. Dep. Environ., Sport Territ., Environ. Aust., Canberra. Fischer, S. G. & L. S. Lerman. 1983. DNA fragments differing by single base-pair substitutions are separated in denaturing gradient gels: Correspondence with melting theory. Proc. Natl. Acad. Sci. U.S.A. 80: 1579-1583. Foster, N. L., Baums, I. B., & P. J. Mumby. 2007. Sexual vs. asexual reproduction in an ecosystem engineer: the massive coral Montastraea annularis. J. Anim. Ecol. 76: 384- 391. Glynn, P.W. & G.M. Wellington. 1983. Corals and coral reefs of the Galápagos Islands. University of California Press, Berkeley, CA. Glynn, P.W. 1993. Coral reef bleaching - ecological perspectives. Coral Reefs. 12: 1-17. Glynn, P.W. 1996. Coral reef bleaching: Facts, hypotheses and implications. Glob. Change Biol. 2: 495-509. Glynn, P.W. 1997 East Pacific reef coral biogeography and faunal flux: Durham's dilemma revisited, Proc. 8th Int. Coral Reef Sym.1: 371–378. Glynn, P.W. & J.S. Ault. 2000. A biogeographic analysis and review of the far eastern Pacific coral reef region. Coral Reefs 19: 1-23. Graham, C.H. & R.J. Hijmans. 2006. A comparison of methods for mapping species ranges and species richness. Global Ecol. Biogeogr. 15: 578. Haefner, B. 2003. Drugs from the deep: marine natural products as drug candidates. Drug Discov. Today 8: 536-544. Hausdorf, B. & C. Hennig. 2010. Species delimitation using dominant and codominant multilocus markers. Syst. Biol. 59: 491-503. Highsmith, R. C. 1982. Reproduction by fragmentation in corals. Mar. Ecol. Prog. Ser. 7:207- 226. Hughes, T.P., A.H. Baird, D.R. Bellwood, M. Card, S.R. Connolly, C. Folke, R. Grosberg, O. Hoegh-Guldberg, J.B.C. Jackson, J. Kleypas, J.M. Lough, P. Marshall, M. Nystrom, S.R. Palumbi, J.M. Pandolfi, B. Rosen & J. Roughgarden. 2003. Climate change, human impacts, and the resilience of coral reefs. Science 301: 929-933. Johannesson, K. & C. Andre. 2006. Life on the margin: genetic isolation and diversity loss in a peripheral marine ecosystem, the Baltic Sea. Mol. Ecol. 15: 2013-2029. Kelly, R.P. & S.R. Palumbi. 2010. Genetic structure among 50 species of the northeastern pacific rocky intertidal community. PLoS One 5: e8594. Kharouba, H.M., Algar, A.C. & J.T. Kerr. 2009. Historically calibrated predictions of butterfly species’ range shift using global change as a pseudo-experiment. Ecology 90: 2213– 2222. Kinlan, B.P. & S.D. Gaines. 2003. Propagule dispersal in marine and terrestrial environments: A community perspective. Ecology 84: 2007-2020.

16 LaJeunesse, T.C. 2002. Diversity and community structure of symbiotic dinoflagellates from Caribbean coral reefs. Mar. Biol. 141: 387-400. Lamb, J.M., Ralph, T.M.C., Goodman, S.M., Bogdanowicz, W., Fahr, J., Gajewska, M., Bates, P.J.J., Eger, J., Benda, P. & P.J. Taylor. 2008. Phylogeography and predicted distribution of African-Arabian and Malagasy populations of giant mastiff bats, Otomops spp. (Chiroptera : Molossidae). Acta Chiropterologica 10: 21–40. Levin, S.A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943–67. Moberg, F., & C. Folke. 1999. Ecological goods and services of coral reef ecosystems. Ecol. Econ. 29: 215-233. Monterroso, P., Brito, J.C., Ferreras, P. & P.C. Alves. 2009. Spatial ecology of the European wildcat in a Mediterranean ecosystem: dealing with small radio-tracking datasets in species conservation. J. Zool. 279: 27–35. Murray-Smith, C., Brummitt, N.A., Oliveira-Filho, A.T., Bachman, S., Moat, J., Lughadha, E.M.N. & E.J. Lucas. 2009. Plant diversity hotspots in the Atlantic coastal forests of Brazil. Conserv. Biol. 23: 151–163. Muscatine, L. & C. Hand. 1958. Direct evidence for the transfer of materials from symbiotic algae to the tissues of a Coelenterate. Proc. Natl. Acad. Sci. U. S. A. 44: 1259-1263. Naeem, S. & S. Li. 1997. Biodiversity enhances ecosystem reliability. Nature 390: 507-509. Pendleton, L.H. 1995. Valuing coral reef protection. Ocean Coastal Manag. 26: 119–131. Phillips, S. J., Dudík, M., & R. E. Schapire. 2004. A maximum entropy approach to species distribution modeling. In: Proceedings of the twenty-first international conference on Machine learning, ACM, 83. Phillips, S.J. & M. Dudík. 2008. Modeling of species distributions with MAXENT: new extensions and a comprehensive evaluation. Ecography 31: 161-175. Pinzón, J.H. & T.C. LaJeunesse. 2011. Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Mol. Ecol. 20: 311-325. Queller, D.C., Strassmann, J.E. & C.R. Hughes. 1993. Microsatellites and kinship.Trends Ecol. Evol. 8: 285–288. Richmond, R. H. & C. L. Hunter. 1990. Reproduction and recruitment of corals: Comparisons among the Caribbean, the Tropical Pacific, and the Red Sea. Mar. Ecol. Prog. Ser. 60: 185-203. Rosenberg, N.A., T. Burke, K. Elo, M.W. Feldmann, P.J. Freidlin, M.a.M. Groenen, J. Hillel, A. Maki-Tanila, M. Tixier-Boichard, A. Vignal, K. Wimmers & S. Weigend. 2001. Empirical evaluation of genetic clustering methods using multilocus genotypes from 20 chicken breeds. Genetics 159: 699-713. Sagarin, R.D. & S.D. Gaines. 2002. The 'abundant centre' distribution: to what extent is it a biogeographical rule? Ecol. Lett. 5: 137-147. Selkoe, K. A., & R. J. Toonen. 2006. Microsatellites for ecologists: a practical guide to using and evaluating microsatellite markers. Ecol. Lett. 9: 615-629.

17 Selkoe, K.A., O.E. Gaggiotti, B.W. Bowen & R.J. Toonen. 2014. Emergent patterns of population genetic structure for a coral reef community. Mol. Ecol. 23: 3064-3079. Smith, S.V. 1978. Coral-reef area and the contribution of reefs to processes and resources of the world’s oceans. Nature 273: 225–226. Tilman, D., Reich, P. B., & J. M. Knops. 2006. Biodiversity and ecosystem stability in a decade- long grassland experiment. Nature 441: 629-632. Tinoco, B.A., Astudillo, P.X., Latta, S.C. & C.H. Graham. 2009. Distribution, ecology and conservation of an endangered Andean hummingbird: the Violet-throated Metaltail (Metallura baroni). Bird Conserv. Int. 19: 63–76. Tittensor, D.P., Baco, A.R., Brewin, P.E., Clark, M.R., Consalvey, M., Hall-Spencer, J., Rowden, A.A., Schlacher, T., Stocks, K.I. & A.D. Rogers. 2009. Predicting global habitat suitability for stony corals on seamounts. J. Biogeogr. 36: 1111–1128. Tognelli, M.F., Roig-Junent, S.A., Marvaldi, A.E., Flores, G.E. & J.M. Lobo. 2009. An evaluation of methods for modelling distribution of Patagonian insects. Revista Chilena De Historia Natural 82: 347–360. Vanhaecke, D., C.G. De Leaniz, G. Gajardo, K. Young, J. Sanzana, G. Orellana, D. Fowler, P. Howes, C. Monzon-Arguello & S. Consuegra. 2012. DNA barcoding and microsatellites help species delimitation and hybrid identification in endangered galaxiid fishes. PLoS One 7: e32939. Verbruggen, H., Tyberghein, L., Pauly, K., Vlaeminck, C., Van Nieuwenhuyze, K., Kooistra, W., Leliaert, F. & O. De Clerck. 2009. Macroecology meets macroevolution: evolutionary niche dynamics in the seaweed Halimeda. Global Ecol. Biogeogr. 18: 393–405. Veron, J. 2000. Corals of the World. Australian Institue of Marine Science, Townsville. Wang, Y., Xie, B., Wan, F., Xiao, Q. & L. Dai. 2007. The potential geographic distribution of Radopholus similisin China. Agricultural Sciences in China 6: 1444–1449. Ward, D. 2007. Modelling the potential geographic distribution of invasive ant species in New Zealand. Biological Invasions 9: 723–735. Williams, J.N., Seo, C.W., Thorne, J., Nelson, J.K., Erwin, S., O’Brien, J.M. & M.W. Schwartz. 2009. Using species distribution models to predict new occurrences for rare plants. Diversity and Distributions 15: 565–576. Wollan, A.K., Bakkestuen, V., Kauserud, H., Gulden, G. & R. Halvorsen. 2008. Modelling and predicting fungal distribution patterns using herbarium data. J. Biogeogr. 35: 2298–2310. Yates, C., McNeill, A., Elith, J. & G. Midgley. 2010. Assessing the impacts of climate change and land transformation on Banksia in the South West Australian Floristic Region. Diversity and Distributions 16: 187–201. Yesson, C. & A. Culham. 2006. A phyloclimatic study of Cyclamen. BMC Evol. Biol. 6: 72–95. Young, B.F., Franke, I., Hernandez, P.A., Herzog, S.K., Paniagua, L., Tovar, C. & T. Valqui. 2009. Using spatial models to predict areas of endemism and gaps in the protection of Andean slope birds. Auk 126: 554–565.

Chapter 2

HIGH GENOTYPIC DIVERSITY OF THE REEF-BUILDING CORAL PORITES LOBATA (SCLERACTINIA: PORITIDAE) IN ISLA DEL COCO NATIONAL PARK, COSTA RICA

This chapter is published in Revista de Biologia Tropical (Boulay JN, Cortés, J, Nivia-Riuz, J, Baums, IB, 2012, 60 (Suppl. 3): 279-292)

Abstract

The isolated Isla del Coco experiences periodic, extreme disturbances which devastate coral reefs surrounding the island. Scleractinian corals build the physical structure of the reef therefore ecosystem recovery relies on coral species recovery. Coral recruits can be of sexual or asexual origin, and the relative success of the two recruit types influences the speed and spread of recovery processes. Here we focus on the massive coral, Porites lobata, because it is the main reef-builder around Isla del Coco, to describe the relative contribution of asexual and sexual recruits to population maintenance. P. lobata samples were collected using a spatially explicit random sampling design in three plots at Isla del Coco: Punta Ulloa (n=17), Bahía Weston (n=20) and Punta María (n=20) and samples were genotyped with 11 microsatellite markers. Additional sampling was conducted at three “coastal” sites near the Costa Rican mainland (Isla del Caño

Biological Reserve): Caño1 (n=8), Caño2 (n=10), Caño5 (n=11) to compare the contributions of asexual and sexual recruits at Isla del Coco sites to coastal sites. Isla del Coco sites were characterized by small colony size (>60% of colonies <0.5m2) and high sexual reproduction. Sites were either mostly or entirely sexual, consisting of only unique genotypes (NG/N= 0.90-1.00;

GO/GE=0.83-1.00; D=0.99-1.00). Although there were no significant differences in genetic

19 diversity (number of alleles per locus, number of private alleles) or colony size between Isla del

Coco and the coastal sites, the coastal sites exhibited a greater range of genotypic diversity from moderately asexual (NG/N=0.5; GO/GE=0.36; D=0.8) to purely sexual (NG/N=1.0; GO/GE=1.0;

D=1.0). The mode of asexual reproduction in P. lobata is likely fragmentation of adult colonies rather than asexual larval production because ramets of P. lobata occurred close together and asexually produced larvae have not been reported in gonochoric broadcast spawners like P. lobata. Frequent sexual reproduction at Isla del Coco National Park might represent a resource for rapid recovery following extreme El Niño-Southern Oscillation (ENSO) disturbance events.

In contrast, larger, asexually-produced fragments rather than smaller, sexually-produced larvae appear to have the advantage at some coastal sites. The high frequency of sexual reproduction at

Isla del Coco indicates that not only are sexual partners available but also current conditions are favorable for the delivery of larvae, and the rate of predation on small larval recruits must be moderate.

Introduction

Corals build the three-dimensional habitat of the reef ecosystem and thus act as foundation fauna (Bruno & Bertness 2001). Scleractinian corals often have large geographic distributions and encounter steep environmental gradients across their ranges (Maina et al. 2011).

For example, marginal environments at the edges of coral species’ ranges are characterized by low species diversity and adverse environmental conditions (Veron 2000, Maina et al. 2011,

Polidoro et al. 2012) yet, coral populations persist. In the Eastern Tropical Pacific (ETP), coral species diversity is lower than in the central, Indo-West Pacific region (Wells 1988, Glynn &

Ault 2000) and environmental conditions are suboptimal for reef growth because of the limited shallow water habitat, fluctuating seawater temperatures due to seasonal upwelling, high

20 sedimentation rates and low aragonite saturation state (Cortés 1997, Glynn & Ault 2000, Maina et al. 2011, Polidoro et al. 2012). Further, a 5000-8000km deep water barrier (Dana 1975, Grigg &

Hey 1992) now separates tropical eastern Pacific biotas from the Indo-West Pacific region.

Darwin (1880) regarded this eastern Pacific barrier as “impassable”, and Ekman (1953) concluded that it is the world’s most potent marine barrier to larval dispersal. The combination of suboptimal environmental conditions and isolation has raised interest to understand how these reefs face adverse conditions. Because climate change is expected to increase environmental variability and decrease species diversity in more central locations (Hughes et al. 2003, Hoegh-

Guldberg et al. 2007), investigating how coral populations persist in marginal habitats provides insights for the future of reefs.

Many coral species can maintain their populations via both sexual and asexual reproduction however, the relative importance of asexual and sexual reproduction can vary across coral species ranges (Stoddart 1984a, Baums et al. 2006) and be influenced by biotic and abiotic factors (Glynn et al. 1994, Legoff et al. 2004, Foster et al. 2007). Sexual reproduction in corals occurs via the release of either sperm alone (sperm casting) or the release of eggs and sperm

(broadcast spawning) into the water column. Sexes can be separate (gonochorism) or together

(hermaphroditism). Asexual reproductive modes are just as varied ranging from the release of asexual larvae, gemmae and polyps to fragmentation of adult colonies. Fragmentation is common in branching species such as Acropora but generally thought to be rare in massive species, with notable exceptions such as Montastraea annularis (Foster et al. 2007). The amount of sexual versus asexual recruitment in a population is directly proportional to the genotypic diversity (the number of distinct genets or clones) of a population. Thus, the genotypic diversity in a population gives an indication of the relative importance of asexual versus sexual reproduction to population maintenance.

21 The relative importance of sexual versus asexual reproduction to population maintenance has consequences for the resistance and resilience of populations. While asexual reproduction of potentially well adapted local genotypes allows for the persistence of populations in the absence of sexual partners and/or favorable conditions for larval recruits, genotypically depauperate populations are expected to be less resilient to abiotic and biotic disturbances (red queen hypothesis, Lively et al. 1990, Reusch et al. 2005). It is thus important to determine the levels of genotypic (and genetic) diversity of coral populations. Decreased levels of genotypic diversity in corals have been described in areas of intermediate and increased natural and anthropogenic disturbances (Hunter 1993, Coffroth & Lasker 1998) and regions with lower connectivity among populations of corals (Baums et al. 2006). At the edge of a species’ range, asexually produced coral colonies can dominate local populations of lower density compared to the center of the range (reviewed in Baums 2008), maybe due to the rarity of sexual partners. Concordantly, clonal reproduction was more frequent in marginal than central habitats for at least two species:

Acropora palmata (Baums et al. 2006) and Pocillopora damicornis (Stoddart 1984a, 1984b,

Miller & Ayre 2004). In P. damicornis (Type I), an important reef builder in the Eastern Pacific, there is evidence of both sexual and asexual reproduction in the northern ETP but no information exists for the center of the range (Pinzón et al. 2012). Similarly, the importance of asexual versus sexual reproduction has not been determined for Porites lobata Dana, 1846, despite its prominent role as foundation fauna on Eastern Pacific sites such as the reefs of Isla del Coco.

Isla del Coco National Park (5°32’N, 87°04’W) is located approximately 500 km southwest of the Costa Rican mainland (Fig. 2-1A). The fringing reefs surrounding the island have the highest species richness of zooxanthellate corals among sites on the Pacific coast of

Costa Rica (Guzmán & Cortés 1992) but are dominated by few foundation species (Bakus 1975).

Most of the reefs were constructed by Porites lobata but the agariciids and Pavona spp. are also common (Cortés et al. 2010). Unlike mainland Costa Rica which is affected more regularly by

22 anthropogenic disturbances, the remote location of Isla del Coco and its National Park status renders it less impacted by humans (as reviewed in Cortés et al. 2010). However, natural disturbances in the form of El Niño-Southern Oscillation (ENSO) events are very common in the waters around Isla del Coco (Cortés 2008, Cortés et al. 2010). As a result of the 1982-1983

ENSO event live coral cover was depleted to 2.6 to 3.5% (Guzmán & Cortés 1992). But some recovery has been observed. In 2000, live coral cover had reached 59% at one Isla del Coco site

(Guzmán & Cortés 2007) despite punctuation by an additional severe bleaching event in 1996-

1997. Hundreds of scientific papers have been published on the marine life at Isla del Coco mostly with a focus on taxonomy. In contrast, no known studies have been published on the dynamics of populations, behavior, or genetic makeup of any marine species on the island (Cortés

2008). The periodic natural disturbances and marginal location of Isla del Coco combined with the dominance of few coral foundation species increase the importance of examining the relative importance of sexual versus asexual reproduction to population maintenance of foundation corals at Isla del Coco. Thus, we investigated the genetic and genotypic diversity of a dominant reef builder at Isla del Coco, P. lobata, at Isla del Coco National Park and contrast it with the diversity found at more coastal sites along the Costa Rican mainland.

Materials and Methods

Study species

Porites lobata Dana, 1846 (Scleractinia: Poritidae) is a massive, slow-growing, gonochoric, broadcast-spawning species, that also reproduces asexually by partial mortality or fish-induced fragmentation (Cortés 1997). It has a Pacific-wide distribution and dominates reefs in lower latitudes of the ETP and on oceanic islands such as Isla del Coco (Glynn et al. 1994,

23 Guzmán & Cortés 2007). In Costa Rica, it is the dominant reef-building coral (Cortés & Guzmán

1998) constituting over 90% of the best developed reefs in the country, including Isla del Coco

(Guzmán & Cortés 1992), Isla del Caño (Guzmán & Cortés 1989), and Punta Islotes in Golfo

Dulce (Cortés 1992).

Study sites

P. lobata colonies were sampled in April 2010 at Punta Ulloa, Bahía Weston and Punta

María in Isla del Coco National Park, Costa Rica (“oceanic sites”). Colonies at Bahía Weston and

Punta Ulloa were located 4-7 m deep. Punta María is a deeper site with P. lobata colonies found

10-12 m deep. For comparison, three inshore Costa Rican sites (“coastal sites”) were also sampled in Isla del Caño Biological Reserve. These sites were designated Caño1, Caño2 and

Caño5.

Sampling method

The sampling design followed Baums et al. (2006). Coral fragments were collected under randomly generated coordinates in 15 m radius circular plots (Fig. 2-1B-C). One plot was sampled at each site (see Study Sites) for a total of three plots at Isla del Coco and three coastal plots. Coordinates had a precision of 5° of arc and of 0.5 m along strike and were generated using the random number generation function in Microsoft Excel. Using a compass and a measuring tape secured to the center point of the circle, coordinates were located by a team of SCUBA divers. The center of the plot was diver selected to maximize colony density and therefore sampling feasibility. The colony (defined as a mounding coral skeleton covered by continuous tissue growth) underneath each coordinate was sampled using a hammer and chisel to break off a

24 small fragment of coral tissue (maximum size 1 cm x 1 cm). An underwater photograph was taken of each colony sampled for future reference, and each colony’s maximum length, width, and height was measured to the nearest 10 cm. The same colony was never sampled twice and sampling of a plot ceased when 20 colonies were collected to standardize sampling effort.

However, previously unknown cryptic species diversity within morphologically similar Porites colonies was revealed by genotyping post-sampling (Boulay et al. 2014) and reduced the sample sizes of P. lobata in the Coastal plots and Punta Ulloa (see genotyping results for further discussion). Fragments were placed in individual zip-lock bags underwater and then transferred to vials containing 95% ethanol. Fragments were stored in a -20°C freezer until DNA extraction and genetic analysis could be performed.

Genotyping

Genomic DNA was extracted following the manufacturer’s instructions using the DNeasy

96 Blood and Tissue Kit (Qiagen, CA). Multi-locus genotypes (MLGs) were established for each colony using a total of eleven microsatellite loci (Table A-1) that have been used in previous studies (Polato et al. 2010, Baums et al. 2012). Polymerase chain reactions were run in four multiplex reactions consisting of two to three primer pairs each and one singleplex reaction.

Primer sets were labeled with NED, VIC, PET, or 6FAM (Applied Biosystems, CA). PCR conditions consisted of an initial denaturation step of 95°C for 5 min, followed by 35 cycles

(95°C for 20s; melting temperature of 52°C-56°C depending on the microsatellite for 20s; and

72°C for 30s). A final extension of 30 min at 72°C was used to ensure the addition of a terminal adenine to all products (Brownstein et al. 1996). PCR products were visualized using an ABI

3730 automated sequencer and internal size standards (Genescan LIZ-500; Applied Biosystems,

25 CA) were used to determine the size of the products. Alleles were scored based on amplicon size using Genemapper 4.0 (Applied Biosystems, CA).

Analysis of multi-locus genotype data

Multi-locus genotypes (MLGs) were determined in GenAlEx vers 6.4 by requiring complete matches at all loci (Peakall & Smouse 2006) ignoring missing alleles. Potential genotyping errors were detected with GenClone 2.0 (Arnaud-Haond & Belkhir 2007) and spurious allele calls were corrected. Genotypic and genetic diversity indices were calculated.

Genotypic diversity refers to the number of unique multi-locus genotypes and varies on the level of whole organisms. In contrast, genetic diversity measures allele diversity of individual loci in a population. Genotypic diversity indices such as genotypic richness (NG/N), genotypic diversity

(GO/GE) and genotypic evenness (GO/NG) were calculated (Table 2-1) for each site. Genotypic richness is directly proportional to the frequency of sexual recruitment and is calculated by dividing the number of unique genotypes (genets; NG) by the total number of samples (N).

Genotypic diversity is calculated by dividing the observed (GO) over the expected (GE) genotypic diversity (Stoddart & Taylor 1988). Microsatellites are highly polymorphic markers and thus each sample is expected to be genetically distinct in a sexually reproducing population. The probability of identity (PID) in this system was expected to be low based on previous studies of P. lobata

-8 using this marker system. The greatest PID for any site in this study is 5.6 × 10 therefore, the expected number of genotypes (GE) is equal to the number of colonies genotyped and does not have to be estimated (see Baums et al. 2006). Observed genotypic diversity, GO, is calculated by the inverse of the sum of the square of ni (the number of individuals of genotype i found in the total number of samples) divided by N. If the colonies in the population are all sexually produced

GO will be equal to N. Therefore, genotypic diversity is maximized in a solely sexual population

(GO/GE=1) and approaches zero when all colonies are asexually produced. Lastly, a measure of genotypic evenness is calculated by dividing observed genotypic diversity by the number of unique genotypes (GO/NG). In contrast to richness, evenness is more influenced by genet longevity than recruitment. Evenness will approach zero when the population is dominated by one genotype but will approach 1 when each genotype has an equal number of member colonies

(ramets). Simpson’s corrected diversity (D), and Shannon-Wiener corrected diversity (H), were also calculated for each site using GenoDive (Meirmans & Van Tienderen 2004). Simpson’s diversity is independent of sample size. However, the Shannon-Wiener index is highly biased by sample size and rarely used for genetic studies except where sample sizes are similar. The genetic diversity was estimated by: the number of alleles (NA), number of effective alleles (AE), and number of private alleles (AP) and average observed (HO) and expected (HE) heterozygosity

(Table 2-2). Each richness, diversity, and evenness index averaged over sites within regions was compared between regions (Isla del Coco versus Coastal) using t-tests after checking for equal variance (Levene test, p>0.05) and normality (Shapiro-Wilk, p>0.05). Simpson’s diversity (D) violated the assumptions of normality and equal variance and thus a non-parametric comparison was performed (Mann Whitney U test).

Characterization of sites

The relation between richness and evenness (GO/GE vs GO/NG) characterizes the relative contribution of sexual recruitment and longevity of colonies to population structure and was used to classify each site into three groups (Baums et al. 2006). The first group was characterized by sexual reproduction apparent by the maximization of both indices. The second group was considered to reproduce mainly sexually and the third was dominated by fragmentation as richness and evenness were low. K-means clustering of sites was performed using the squared

27 Euclidian distance of only the uncorrelated (Pearson correlation, corrected p>0.001) diversity estimates: GO/GE, D, and NG/N. Selection of the most likely cluster followed Calinski &

Harabasz' pseudo-F, Akaike Information Criterion, and Bayesian Information Criterion as calculated in GenoDive (Meirmans & Van Tienderen 2004).

Results

Size distribution of colonies

Colony size (m2) was estimated by multiplying length and width measurements and resulting sizes were divided into nine uniform 0.5 m2 size classes ranging from 0-4.5 m2. P. lobata colony size was skewed towards small colonies with few large colonies over 4 m2 observed (Fig. 2-1D). A significantly larger proportion of colonies was found in the smallest size class (0–0.5 m2) compared to other size classes (one way ANOVA, post-hoc Tukey test, p<0.001). The proportion of small colonies at Isla del Coco was 66% (SD = 25%). There was no significant difference (t-test; p>0.05) in the size of colonies between regions (Mean-SizeCOCO=

2 2 0.63 ± 0.85m ; Mean-SizeCOASTAL= 0.94 ± 1.20m ) or between unique colonies and those

2 belonging to a repeated MLG (Mean-SizeUNIQUE=0.64 ± 0.85m ; Mean-SizeREPEAT=1.11 ±

1.38m2).

Genotyping

The total number of samples collected at both sites was comparable (nCOCO=58; nCOASTAL=60). However, genotyping revealed that many of these samples were actually Porites evermanni (Boulay et al. 2014) particularly at Isla del Caño where P. evermanni samples

28 composed over half of the collection (n=31). Because this study was designed to describe population genetic structure of Porites on Isla del Coco, and P. lobata is the dominant Porites on this island, the following results focus on P. lobata only. Genotype assignments were given to all

P. lobata samples collected from Isla del Coco (n=57) and Isla del Caño (n=29). Amplification resulted in an overall average failure of 6% (SD=7%) at Isla del Coco and 3% (4%) at the coastal/mainland sites, and a per locus failure rate of <16% for each locus in the included samples. No individual was missing data at more than two loci. At total of 48 of the 86 individuals had complete multi-locus genotypes at all loci. The probability of identity (PID) for 11 microsatellite loci applied here ranged from 8.9x10-9 at Punta María to 5.6x10-8 at Caño1 (Table

2-2). Thus, the power to distinguish between similar but non-identical multi-locus genotypes was high. Further, the PID when only 9 loci were considered was also sufficiently small to justify

-7 -7 inclusion of individuals with missing data (PIDCOCO = 6.7 × 10 ; PIDCOASTALl = 4.3 × 10 ).

The contribution of fragmentation to population structure

At Isla del Coco, the analysis of 57 samples resulted in 54 unique multi-locus genotypes

(genets); indicating that sexual reproduction is the main reproductive strategy on this island. The three repeated MLGs consisted of only two ramets per genet each and ramets were always confined to a single sampling plot. The range of separation between the clonemates was from 1 to

7 m. Every colony (n=20) at the deepest of the three sites, Punta María, represented a unique

MLG (Fig. 2-1C). At Punta Ulloa (n=17), two ramets were found separated by a distance of less than 0.5 m (Fig. 2-1C). At Bahía Weston (n=20), fragmentation was observed in the largest two colonies in the plot (Fig. 2-1C); one pair of ramets was separated by 6.69 m and another pair separated by a distance of 2.14 m (Fig. 2-1C).

29 At the coastal sites there were 22 unique genotypes out of the 29 Porites lobata samples.

In repeated MLGs, the number of ramets per genet ranged from two at Caño5 (n=11) to four at

Caño2 (n=10) (Fig. 2-1B). At Caño1 (n=8) all colonies were unique (Fig. 2-1B). At Caño5 (Fig.

2-1C) ramets were separated by a distance of less than 3 m but at Caño2, ramets were distributed over distances up to 15.7 m; Fig. 2-1C). At Caño5 fragmentation was observed (Fig. 2-1B) due to partial mortality or fission of a large colony; but, at the shallower Caño2, some of the smallest colonies were asexually produced (Fig. 2-1B) and distributed over greater distances suggesting propagation of small rolling fragments.

In summary, genotypic diversity at Isla del Coco sites was high with indices ranging from

0.83 to 1.0 (Table 2-1). Variability among sites was low with one standard deviation ranging from

1 to 8%. At the coastal sites, genotypic diversity indices were not on average different to those observed at Isla del Coco (between 0.36 and 1.0) but variance was higher (p=0.006, Simpson’s

D). Only Shannon-Weiner diversity was significantly different between Isla del Coco and the coastal sites likely due to its strong dependence on sample size (Fig. 2-2A).

Genetic diversity

Colonies at Isla del Coco carried 4.67 ± 0.32 alleles per locus (2.83±0.10 effective alleles) versus 4.09 ± 0.64 alleles per locus at the coastal sites (Fig. 2-2B). Mean observed heterozygosity across sites was 0.53 ± 0.03 (Fig. 2-2B). Punta María exhibited a relatively high number of private alleles (0.73) per locus while Punta Ulloa, Bahía Weston and the coastal sites had fewer than 0.36 private alleles per locus (Table 2-2). No differences were observed between the coastal island and oceanic Isla del Coco for any genetic diversity index (Fig. 2-2B).

30 Characterization of sites

K-means clustering based on the relationship between genotypic richness and evenness of sites resulted in three clusters (Fig. 2-3). Punta María and Caño1 clustered together as entirely sexual. Bahía Weston, Punta Ulloa, and Caño5 formed a second cluster and are mostly sexual.

Caño2 is distinct from the remaining sites and was dominated by fragmentation (Fig. 2-3).

Discussion

The relative contributions of asexual and sexual recruitment to local population structure of coral species influences how coral populations recover from large scale disturbance events and what consequences such events have on the genetic diversity of the species. Recovery from sexual recruits would increase the genetic and genotypic diversity whereas recovery from regrowth can only preserve existing genotypic diversity. Without genetic markers, it is difficult to distinguish between these two recovery models because small colony sizes are indicative of sexual recruitment of coral larvae as well as regrowth from remnant tissue of colonies that have suffered partial mortality (Miller et al. 2007). Here, we demonstrate a high contribution of sexual reproduction to population maintenance for the framework building coral, Porites lobata, on recovering reefs at coastal and oceanic sites. This result gives a genetic basis to argue that recovery from recent (1982-83 and 1996-97) severe disturbance events proceeded chiefly via sexual recruitment and is concurrent with data demonstrating high sexual recruitment of

Pocillopora in the ETP (Flot et al. 2010, Combosch & Vollmer 2011, Pinzón & LaJeunesse

2011). This is encouraging because it implies adult coral population densities high enough to generate sexual larvae as well as favorable conditions for survival of young recruits leading to the establishment of genetically diverse and thus more resilient populations.

31 Sexual recruits can be of local or distant origin. An analysis of population structure in P. lobata across the entire Pacific basin suggests a biophysical barrier restricting gene flow between central and eastern Pacific populations. However, no further substructure was shown among P. lobata sampled from ten ETP locations, including Isla del Coco, Isla del Caño, mainland Costa

Rica and the Galapágos archipelago (Baums et al. 2012). It is thus likely that the local population at Isla del Coco is recruiting from other ETP sites such as the Galapágos Islands, mainland Costa

Rica, or Panama, in addition to local larval production (Baums et al. 2012). Similarly high connectivity among populations was evident in two Pocillopora species within the ETP (Pinzón

& LaJeunesse 2011). ENSO events in the ETP can cause large-scale disturbances with widespread adult coral mortality but apparently local populations at coastal and remote oceanic sites receive larval subsidies that aid in recovery.

The genetic evidence presented here suggests that fragmentation and reattachment of fragments occurs at low frequency in P. lobata at Isla del Coco even in large colonies (>2.5 m2).

Evidence of limited asexual reproduction at Isla del Coco is congruent with findings for P. lobata populations across the Hawaiian Archipelago (Polato et al. 2010) as well as P. damicornis populations at scales over 5 m along the Panamanian coast (Combosch & Vollmer 2011),

Clipperton Atoll (Flot et al. 2010) and across the ETP (Pinzón & LaJeunesse 2011). In the few instances of clonal reproduction at Bahía Weston and Caño5, large colonies likely survived the recent ENSO events but underwent partial mortality and fragmented as suggested by Guzmán and

Cortés (2007) and supported here by genetic evidence.

While average diversity did not differ between the oceanic and coastal islands, variability in the prevalence of clonal genotypes among sites was increased at the more environmentally variable coastal sites compared to the oceanic sites. Similar high variance in the contribution of asexual reproduction to population maintenance was observed at coastal sites in Mexico for

Pocillopora Type I (Pinzón et al. 2012). In massive corals, the only data for asexual reproduction

32 comes from analyses of clonal structure in Montastrea annularis in the Caribbean. There, we also see variation in the contribution of asexual reproduction to population maintenance with some sites dominated by clones. However this system differs from Porites because fragmentation is driven mainly by abiotic patterns of hurricane frequencies (Foster et al. 2013).

This is in contrast to the mainly biotic factors controlling fragmentation observed in the

ETP. The curious mode of triggerfish-induced fragmentation warrants future studies aimed at comparing the influence of mussels and triggerfish on coral fragmentation between regions in the

ETP. Variation in the contribution of asexual and sexual recruitment to population maintenance may be related to variation in the distributions of the triggerfish species and the boring mussels on which they specialize. One ETP endemic triggerfish species, which specializes on endolithic bivales at Isla del Caño (Guzmán & Cortés 1989), was found in biogeographic surveys at Isla del

Caño to be at moderate abundance (8.4 individuals/hectare; Guzmán 1988) particularly at the base of the reef but, has not been recorded at the major offshore islands such as Isla del Coco

(Berry & Baldwin 1966). To understand the mechanisms controlling diversity in this system, further sampling for coral clonal structure and surveying for abundance of triggerfish and mussels will be required across the range of the coral, triggerfish, and mussels in the ETP and should be augmented with deployment of settlement plates. This knowledge of how coral populations persist in marginal habitats provides insights for conservation of the existing diversity and preserving resilience on coral reefs.

33 Acknowledgements

We thank the government of Costa Rica for facilitating research permits through SINAC and ACMIC. We appreciate the logistical support to facilitate fieldwork provided by CIMAR.

This research was supported by NSF grant OCE- 0550294 to IBB and grants from the

Vicerrectoría de Investigación (UCR) and CONARE to JC.

Author Contributions

JC and JNR collected the samples. JB performed lab and statistical analysis. JB and IB wrote paper. All authors edited and approved the manuscript. IB and JC obtained funding.

34 References

Arnaud-Haond, S. & K. Belkhir. 2007. GENCLONE: a computer program to analyse genotypic data, test for clonality and describe spatial clonal organization. Mol. Ecol. Notes 7: 15-17. Bakus, G.J. 1975. Marine zonation and ecology of Cocos Island, off Central America. Atoll Res. Bull. 179: 1-11. Baums, I.B., M.W. Miller & M.E. Hellberg. 2006. Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr. 76: 503-519. Baums, I.B. 2008. A restoration genetics guide for coral reef conservation. Mol. Ecol. 17: 2796- 2811. Baums, I.B., J.N. Boulay, N.R. Polato & M. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Berry, F.H. & W.J. Baldwin. 1966. Triggerfishes (Balistidae) of the eastern Pacific. Proc. Calif. Acad. Sci. 34: 429-474. Boulay, J.N., M.E. Hellberg, J. Cortés & I.B. Baums. 2014. Unrecognized coral species diversity masks differences in functional ecology. Proc. R. Soc. B-Biol. Sci. 281: 20131580. Brownstein, M.J., J.D. Carpten & J.R. Smith. 1996. Modulation of non-templated nucleotide addition by tag DNA polymerase: Primer modifications that facilitate genotyping. Biotechniques 20: 1004-1010. Coffroth, M.A. & H.R. Lasker. 1998. Population structure of a clonal gorgonian coral: The interplay between clonal reproduction and disturbance. Evolution 52: 379-393. Combosch, D.J. & S.V. Vollmer. 2011. Population genetics of an ecosystem-defining reef coral Pocillopora damicornis in the Tropical Eastern Pacific. PLoS One 6: e2120010 Cortés, J. 1992. The coral reefs of Golfo Dulce, Costa Rica - Ecological aspects. Rev. Biol. Trop. 40: 19-26. Cortés, J. 1997. Biology and geology of eastern Pacific coral reefs. Coral Reefs 16: S39-S46. Cortés, J. & H. Guzmán. 1998. Organisms of Costa Rican coral reefs: description, geographic distribution and natural history of Pacific zooxantellate corals (Anthozoa : Scleractinia). Rev. Biol. Trop. 46: 55-92. Cortés, J. 2008. History of marine research at Cocos Island, Costa Rica. Rev. Biol. Trop. 56: 1- 18. Cortés, J., C.E. Jiménez, A.C. Fonseca & J.J. Alvarado. 2010. Status and conservation of coral reefs in Costa Rica. Rev. Biol. Trop. 58: 33-50. Dana, T.F. 1975. Development of contemporary eastern Pacific coral reefs. Mar. Biol. 33: 355- 374. Darwin, C. 1880. The Origin of Species by Means of Natural Selection or the Preservation of Favored Races in the Struggle for Life. John Murray, London. Ekman, S. 1953. Zoogeography of the Sea. Sidgwick and Jackson Ltd, London.

35 Flot, J.F., A. Couloux & S. Tillier. 2010. Haplowebs as a graphical tool for delimiting species: a revival of Doyle's "field for recombination" approach and its application to the coral genus Pocillopora in Clipperton. BMC Evol. Biol. 10: 372 310.1186/1471-2148-1110- 1372. Foster, N., I. Baums, J. Sanchez, C. Paris, I. Chollett, C. Agudelo, M. Vermeij & P. Mumby. 2013. Hurricane-driven patterns of clonality in an ecosystem engineer: the Caribbean coral Montastraea annularis. Mol. Ecol. 21: 5418-5433. Foster, N.L., I.B. Baums & P.J. Mumby. 2007. Sexual verses asexual reproduction in an ecosystem engineer: the massive coral Montastraea annularis. J. Anim. Ecol. 76: 384- 391. Glynn, P.W., S.B. Colley, C.M. Eakin, D.B. Smith, J. Cortés, N.J. Gassman, H.M. Guzmán, J.B. Delrosario & J.S. Feingold. 1994. Reef coral reproduction in the Eastern Pacific - Costa- Rica, Panama, and Galapágos-Islands (Ecuador) .2. Poritidae. Mar. Biol. 118: 191-208. Glynn, P.W. & J.S. Ault. 2000. A biogeographic analysis and review of the far eastern Pacific coral reef region. Coral Reefs 19: 1-23. Grigg, R.W. & R. Hey. 1992. Paleoceanography of the Tropical Eastern Pacific Ocean. Science 255: 172-178. Guzmán, H.M. 1988. Distribución y abundancia de organismos coralívoros en los arrecifes coralinos de la Isla del Caño, Costa Rica. Rev. Biol. Trop. 36: 191-207. Guzmán, H.M. & J. Cortés. 1989. Coral reef community structure at Caño Island, Pacific Costa Rica. P.S.Z.N.I: Mar. Ecol. 10: 23-41. Guzmán, H.M. & J. Cortés. 1992. Cocos Island (Pacific of Costa-Rica) coral reefs after the 1982- 83 El-Niño disturbance. Rev. Biol. Trop. 40: 309-324. Guzmán, H.M. & J. Cortés. 2007. Reef recovery 20 years after the 1982-1983 El Niño massive mortality. Mar. Biol. 151: 401-411. Hoegh-Guldberg, O., P.J. Mumby, A.J. Hooten, R.S. Steneck, P. Greenfield, E. Gomez, C.D. Harvell, P.F. Sale, A.J. Edwards, K. Caldeira, N. Knowlton, C.M. Eakin, R. Iglesias- Prieto, N. Muthiga, R.H. Bradbury, A. Dubi & M.E. Hatziolos. 2007. Coral reefs under rapid climate change and ocean acidification. 318: 1737-1742. Hughes, T.P., A.H. Baird, D.R. Bellwood, M. Card, S.R. Connolly, C. Folke, R. Grosberg, O. Hoegh-Guldberg, J.B.C. Jackson & J. Kleypas. 2003. Climate change, human impacts, and the resilience of coral reefs. 301: 929 - 933. Hunter, C.L. 1993. Genotypic variation and clonal structure in coral populations with different disturbance histories. Evolution 47: 1213-1228. Legoff, M.C., O.G. Pybus & A.D. Rogers. 2004. Genetic structure of the deep sea coral Lophelia pertusa in the northeast Atlantic revealed by microsatellites and internal transcibed spaces sequences. Mol. Ecol. 13: 537-549. Lively, C.M., C. Craddock & R.C. Vrijenhoek. 1990. Red queen hypothesis supported by parasitism in sexual and clonal fish. 344: 864-866. Maina, J., T.R. Mcclanahan, V. Venus, M. Ateweberhan & J. Madin. 2011. Global gradients of coral exposure to environmental stresses and implications for local management. PLoS One 6: e23064.

Meirmans, P.G. & P.H. Van Tienderen. 2004. GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Mol. Ecol. Notes 4: 792-794. Miller, K.J. & D.J. Ayre. 2004. The role of sexual and asexual reproduction in structuring high latitude populations of the reef coral Pocillopora damicornis. Heredity 92: 557-568. Miller, M.W., I.B. Baums & D.E. Williams. 2007. Visual discernment of sexual recruits is not feasible for Acropora palmata. Mar. Ecol. Prog. Ser. 335: 227-231. Peakall, R. & P.E. Smouse. 2006. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6: 288-295. Pinzón, J.H. & T.C. LaJeunesse. 2011. Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Mol. Ecol. 20: 311-325. Pinzón, J.H., H. Reyes-Bonilla, I.B. Baums & T.C. LaJeunesse. 2012. Contrasting clonal structure among Pocillopora (Scleractinia) communities at two environmentally distinct sites in the Gulf of California. Coral Reefs 31:765-777. Polato, N.R., G.T. Concepcion, R.J. Toonen & I.B. Baums. 2010. Isolation by distance across the Hawaiian Archipelago in the reef-building coral Porites lobata. Mol. Ecol. 19: 4661- 4677. Polidoro, B.A., T. Brooks, K.E. Carpenter, G.J. Edgar, S. Henderson, J. Sanciangco & D.R. Robertson. 2012. Patterns of extinction risk and threat for marine vertebrates and habitat- forming species in the Tropical Eastern Pacific. Mar. Ecol. Prog. Ser. 448: 93-104. Reusch, T.B.H., A. Ehlers, A. Hammerli & B. Worm. 2005. Ecosystem recovery after climatic extremes enhanced by genotypic diversity. Proc. Natl. Acad. Sci. U.S.A. 102: 2826-2831. Stoddart, J.A. 1984a. Genetical structure within populations of the coral Pocillopora damicornis. Mar. Biol. 81: 19-30. Stoddart, J.A. 1984b. Genetic differentiation amongst populations of the coral Pocillopora damicornis off southwestern Australia. Coral Reefs 3: 149-156. Stoddart, J.A. & J.F. Taylor. 1988. Genotypic diversity: estimation and prediction in samples. Genetics 118: 705-711. Veron, J. 2000. Corals of the World. Australian Institue of Marine Science, Townsville. Wells, S.M. 1988. Coral Reefs of the World. Volume 1: Atlantic and Eastern Pacific. United Nations Environment Programme / International Union for Conservation of Nature and Natural Resources, Gland, Switzerland.

08 09 09 09 08 08 09 09

------

× 10 ×

8.9 × 8.910 × 2.810 × 2.6 × 8.610 × 6.010 × 5.6 4.210 × 3.610 ×

1.30 0.93 1.20 1.24 0.90 0.62

1.00 0.96 0.99 0.99 1.00 0.80

1.00 0.90 G 0.95 0.93 1.00 0.71

1.00 0.73 G 0.89 0.83 1.00 0.36

20.00 8.07 15.21 16.67 8.00 3.57

1.00 0.82 N 0.94 0.90 0.95 1.00 0.50 0.76

1 2 MAX 2 2 2 1 4 4

)

20 9 N 16 18 54 8 5 22

within and between coastal within sites oceanic and (one plot per site) in the Eastern

20 11 N 17 20 57 8 10 29

)

Site

Porites Porites lobata

Stoddart & Taylor 1988 StoddartTaylor &

in in

(

Caño5 Caño1 Caño2

Punta Punta María Punta Ulloa Bahía Weston 3 sites 3 sites

Total Total 1988 StoddartTaylor &

(

Weiner diversity Weiner

Genotypic diversity Genotypic

son’s diversity son’s diversity

= evenness = evenness

= genotypic diversity diversity = genotypic

E G

G N

= Probability identity of Probability =

= number of genotypes = number diversity genotypic = observed / /

richness /N = clonal

G G O O O

N = number of samples N = number N per ramets genet = maximum MAX N G G G Simp D = Shannon H =

Tropical Pacific. Pacific. Tropical Table 2 Table Park National Coco del Isla Park National Island Caño 3 8 1 2 4 5 6 7 9 10

Table 2-2. Genetic diversity in Porites lobata within and between coastal and oceanic sites (one plot per site) in the Eastern Tropical Pacific. 1 2 3 4 5 Site AN AE AP HO HE Latitude Longitude (N) (W) Isla del Coco Punta Ulloa 4.36 2.71 0.18 0.55 0.60 05°33’ 87°02’ National Park Bahía Weston 4.64 2.90 0.27 0.54 0.62 05°33’08.2” 87°03’03.2” (oceanic) Punta María 5.00 2.87 0.73 0.50 0.61 05°32’05.4” 87°05’15” Total 3 sites Caño Island Caño1 3.82 2.64 0.09 0.52 0.58 08°42’39” 83°51’54” National Park (coastal) Caño2 3.64 2.82 0.09 0.45 0.57 08°42’43” 83°52’56”

Caño5 4.82 3.17 0.36 0.56 0.62 08°42’32” 83°52’01”

Total 3 sites 1 AN = number of alleles per locus 2 AE =effective number of alleles per locus 3 AP = number of private alleles per locus 4 HO =observed heterozygosity 5 HE =expected heterozygosity

Figure 2-3. Clonal structure of Porites lobata colonies sampled at six sites in Costa Rica. The relation of genotypic evenness GO/NG to diversity GO/GE combined with K-means clustering procedure resulted in three groups of sites (indicated on graph by distinct symbol shapes). The two purely sexual sites, Punta María and Caño1, have overlapping symbols. Fill of symbols indicate the region to which the sampling sites belong (Isla del Coco=black; coastal sites=white).

Chapter 3

UNRECOGNIZED CORAL SPECIES DIVERSITY MASKS DIFFERENCES IN FUNCTIONAL ECOLOGY

This chapter is published in Proceedings of Royal Society B (Boulay, JN, Cortés, J, Hellberg, M, Baums, IB, 2014, 281 (1776): 20131580)

Abstract

Porites corals are foundation species on Pacific reefs but a confused taxonomy hinders understanding of their ecosystem function and responses to climate change. Here we show that what has been considered a single species in the Eastern Tropical Pacific, Porites lobata, includes a morphologically similar yet ecologically distinct species, P. evermanni. While P. lobata reproduces mainly sexually, P. evermanni dominates in areas where triggerfish prey on bioeroding mussels living within the coral skeleton, thereby generating asexual coral fragments.

These fragments proliferate in marginal habitat not colonized by P. lobata. The two Porites species also show a differential bleaching response despite hosting the same dominant symbiont subclade. Thus, hidden diversity within these reef-builders has until now obscured differences in trophic interactions, reproductive dynamics, and bleaching susceptibility, indicative of differential responses when confronted with future climate change.

Introduction

Unrecognized species diversity, especially in foundation species, can impede our understanding of major features of ecosystem functioning and the resilience of communities

43 (Knowlton 1993, Hughes & Stachowicz 2004), thereby complicating projections of the ecological dynamics and future of imperiled coral reefs. Corals are placed in functional groups based on their structural growth forms (e.g. tabular, bushy, massive) (Hughes & Connell 1999). These functional groups show striking differences in life histories and susceptibility to threats

(Bellwood et al.2004), and promote species diversity by providing different habitats for reef dwellers (Idjadi & Edmunds 2006). In addition to these readily apparent morphological differences, marine communities harbor many species not easily resolvable without extensive molecular genetic characterization, but which provide additional diversity (Knowlton 1993) that is typically not perceived by scientists. Here, we concentrate on two coral species with nearly indistinguishable morphologies and test whether unresolved species diversity masks functional diversity (Knowlton 1993), specifically with respect to trophic interactions and stress resistance.

We concentrate on the Eastern Tropical Pacific (ETP), where relatively low species diversity makes the system more tractable compared to richer reefs in the Indo-West Pacific (Glynn &

Wellington 1983, Glynn & Ault 2000).

Environmental conditions for reef growth are suboptimal in the ETP, due mainly to seasonal cold-water upwelling, low aragonite saturation state, and recurrent warm-water events associated with the El Niño Southern Oscillation (ENSO) (Cortés 1997, Glynn & Ault 2000).

Together these factors reduce coral growth (Toth et al.2012) and reef species diversity (Glynn &

Wellington 1983, Glynn & Ault 2000). Reefs in the ETP are built largely by branching

Pocillopora spp. and the massive Porites lobata. However, species identification in these genera is notoriously challenging because colony morphology is plastic (Forsman et al.2009, Pinzón &

LaJeunesse 2011). Based on genetic data, the number of Pocillopora species described from the

ETP has recently been revised (Pinzón & LaJeunesse 2011). Less is known about the taxonomic status of Porites lobata. Seven Porites species are currently recognized in the eastern Pacific

(Veron 2000). Porites lobata and Porites panamensis are thought to be widespread in the region

44 while much of the remaining diversity in this genus (Porites arnaudi, Porites australiensis,

Porites lichen and Porites lutea) is restricted to the higher latitudes (Veron 2000). In lower latitudes P. lobata dominates the reef-building coral community. A survey of genetic variation at the multi-copy internal transcribed spacer (ITS) marker at the genus level found that some colonies from Panama, diagnosed morphologically as P. lobata, clustered genetically with Porites evermanni (Forsman et al.2009) currently thought to be restricted to Hawaii and the Indo-Pacific

(Veron 2000). Combined with previous observations of local variation in reproduction (Glynn et al.1994), this suggests the possibility of unresolved species within P. lobata in the ETP.

Functional differences between morphologically similar coral species have yet to be demonstrated. Although corals in the Orbicella (nee Montastraea) annularis species complex in the Caribbean differ in their reproductive timing and hybridization potential (Levitan et al.2011), no study has shown differences in the way these species interact with other species as we demonstrate here for Porites spp. The ecology and evolution of reproductive mutualisms has received much interest in the terrestrial literature (e.g. Ollerton & Coulthard 2009, Burkle et al.2013) but similar studies are rare in corals partly because corals do not rely on pollinators for sexual reproduction. We show here that asexual reproduction of corals might indeed be dependent on other members of the reef community. Elegant work on Orbicella spp. demonstrated that partial colony bleaching can be attributed to symbiotic algae with varying heat tolerances that occupy different niches within a colony (Rowan et al.1997, Baker et al.2004). Conversely, we present evidence that coral host species harboring the same dominant algal subclade differ in bleaching response pointing to the role of the host in temperature tolerance.

45 Results and Discussion

Samples of massive Porites were collected from seventeen locations in the ETP (see map in Fig. 3-3; see Table B-1 in Electronic supplementary material (ESM)). A principal coordinate analysis (PCoA) based on genetic distance grouped the 448 unique multilocus genotypes (MLGs) into two clusters along the first principle coordinate (PCo1), which explained 67% of the variation (Fig. 3-1A). No geographic structure was evident. Rather, sympatric MLGs assigned with high probability (99.5±2.2% s.d.) to one of two clusters (Fig. 3-1B; see ESM for additional clustering and genotyping results). A network analysis of allelic sequences from five single-copy nuclear genes similarly reveals two clusters in our eastern Pacific collections (n=14): one associated with P. lobata samples from across the Pacific and the other with Hawaiian P. evermanni (Hellberg unpublished).

These two species show a strong geographic distribution gradient between inshore and offshore sites (Fig. 3-2; Analysis of Variance (ANOVA); df = 2, F = 73.383, p<0.001). Insular collections are composed almost entirely of P. lobata (Fig. 3-3A) (Boulay et al.2012). P. evermanni (Fig. 3-3B) constitutes roughly half of the collections at southern sites and gradually increases along a northern gradient. The two most marginal coastal sites in the north are comprised entirely of P. evermanni. (see ESM for additional species distribution results). Thus, the two related species occupy different environmental niches.

On a local scale, we found substantial differences between species in the relative importance of asexual reproduction. Colonies sampled from 10 sites (4 coastal, 6 insular) using a spatially-explicit random method (Baums et al.2006) were genotyped as before, revealing that each species was found in 7 plots (co-occurring in 4). However, standard genotypic diversity estimates (e.g. number of MLGs/number of samples (NG/N); number of observed MLGs/number of expected MLGs (GO/GE); see Table B-3) showed that rates of asexual reproduction differed (t-

46 test; nPlots_Pe=7, nPlots_Pl=7; NG/N, p=0.01; GO/GE, p<0.005) between the species. P. evermanni often reproduces asexually (NG/N=0.47±0.20 s.d., Go/Ge=0.30±0.20 s.d.), whereas P. lobata does so rarely (NG/N=0.81±0.22 s.d., Go/Ge=0.75±0.27 s.d.) (Fig. 3-4). The mode of asexual reproduction here is likely fragmentation of adult colonies rather than asexual larval production because ramets occur close together and asexually produced larvae have not been reported in gonochoric broadcast spawners like these Porites spp. (Baird et al. 2009).

Differences in asexual reproduction appear to be driven by interactions between corals, bivalves, and triggerfish. Endolithic bivalves (Lithophaga spp.) bore into the carbonate skeleton of corals (see Fig. 3-3B), and result in colony fragmentation (Glynn 1974, Glynn & Wellington

1983, Guzmán 1988, Scott & Risk 1988, Guzmán & Cortés 1989). Coral skeletal strength is reduced in colonies containing Lithophaga spp. and colonies fracture along boreholes if present

(Scott & Risk 1988) (see Fig. 3-3B). Two endemic triggerfish (Pseudobalistes naufragium and

Sufflamen verres) common in the ETP prey upon bioeroding Lithophaga spp., which constitute

83% of the diet of P. naufragium at Caño Island (Guzmán 1988). Where the Porites species occur together, mussel density (N cm-2) is higher in P. evermanni than in P. lobata based on direct counts (t-test; p<0.001, nPe=22, nPl=21) and photographic evidence (t-test; p=0.011, nPe=46, nPl=61) (Fig. 3-5A).

To extract their prey, the triggerfish break off coral fragments, resulting in “rolling stones” known as coralliths (Glynn 1974) that can re-attach to the substrate (see Fig. 3-3B). A single fish can produce 3 to 75 fragments in a few minutes from a single colony, with apparently few negative consequences for the coral, as evinced by rapid wound healing of the donor colony

(they regenerate in less than 4 weeks) and high establishment rates of fragments (30-50% of fragments survive) (Guzmán & Cortés 1989). All coralliths sampled randomly in polar plots were

P. evermanni (n=6, mean maximum dimension=10±4.1cm s.d.). Four out of six were fragments from a larger colony sampled in the plot (mean maximum dimension=56.5±32.6 cm s.d.). The

47 fragments were located a mean distance of 3.67±1.69 m s.d. from the likely “parent” colony. The differences between the two species in mode of reproduction may play a large part in their disparate geographical distributions because larger colony fragments might be better able to establish in less hospitable settings (continental margins; northerly latitudes) than sexually produced larvae and fragment production does not require the presence of a sexual partner.

Whereas biting by triggerfish is relatively benign, thermal stress during ENSO events causes coral bleaching and is a major factor in ETP coral mortality (Glynn 1988, Toth et al.

2012). Photographic evidence (t-test; p<0.001) revealed that P. lobata bleached more readily

(nPl=14; mean percent tissue bleached per colony=42.5±35.5% s.d.) than P. evermanni (nPe = 20; mean percent tissue bleached per colony=0.4±1.9% s.d.) at two sites with sympatric P. lobata and

P. evermanni (<5 m apart) (Fig. 3-5C). Such differential bleaching has previously been ascribed to differences in the subclade of algal symbionts hosted in co-occurring coral species (Rowan et al. 1997). However, denaturing gradient gel electrophoresis (DGGE) fingerprinting and sequencing of the ITS2 region of the Symbiodinium algae present in P. lobata (n=18) and P. evermanni (n=19) show that both coral species associate primarily with Symbiodinium subclade

C15 (see Fig. B-2). Therefore, the differential bleaching susceptibility of these coral species points to differences in host physiology, although symbiont density (Cunning & Baker 2013), background (minor) symbionts (Baker 2003), or differences among closely-related strains of the

ITS2-C15 subclade might play an additional role.

Knowlton (1993) proposed that unrecognized species diversity in the sea hinders our understanding of marine ecosystem function and limits our ability to predict how reefs will respond to climate change, but few data have supported this claim. Here we demonstrate that unresolved species diversity has obscured differences in reproduction, ecological dynamics, trophic interactions, and stress resistance between two reef-building coral species.

48 We propose that a three-way interaction between corals, mussels and triggerfish alters the local distribution of the foundation fauna. The species involved have different geographic distributions, varying from the triggerfish restricted to the ETP to the trans-Pacific range of P. lobata. It follows that variation in their co-distribution could alter community dynamics across the Pacific (Thompson 2005). An understanding of such geographic shifts in trophic interactions is important because trophic complexity is one driver of ecosystem diversity (Paine 1966) and marine ecosystem diversity is thought to be linked to increased function (Duarte 2000).

The maintenance and function of ecosystems built by few foundation species rely on the persistence of those species. In P. evermanni, frequent asexual fragmentation allows for local patch reef formation when sexual partners are unavailable (Highsmith 1980, Lasker & Coffroth

1999) and the persistence and spread of potentially locally well-adapted genotypes. However, genetically depauperate populations are more susceptible to non-random (with respect to genotype) stressors than diverse populations (Reusch et al. 2005), an alarming prospect given the increasing frequency of disturbance events affecting ETP coral reefs (Glynn 1990, Collins et al.

2010).

The adaptability of corals in the face of elevated sea surface temperatures and the consequences of more frequent bleaching conditions are not fully understood. Here, however, we have shown that morphological similarity can mask variation in responses to thermal stressors.

Furthermore, ecosystem resilience improves when critical species are functionally redundant in some respects (e.g. the form of the colonies they build) but show differential stress responses

(Bellwood et al. 2004), as observed here. This argues against recent trends of utilizing morphological groups to project the future of reefs (Darling et al. 2012). It follows that as long as coral taxonomy remains unresolved, unrecognized species will continue to obscure our understanding of reef ecosystem function and resilience and our ability to predict the fate of the coral reef ecosystem. The functional differences within morphologically similar species observed

49 in this study lead us to predict that there will be a differential response to climate change among the massive Porites species that compose the foundation of Eastern Tropical Pacific reefs.

Materials and Methods

Study Sites and Sampling

Corals were sampled under randomly generated coordinates in 15m radius polar plots

(Baums et al. 2006) or haphazardly (>5m separating sampled colonies) at sites where random sampling was not feasible due to low colony density (see Table B-1). Random coordinates were generated using the random number generation function in Microsoft Excel 2007 (Microsoft,

WA) with a precision of 5° of arc and of 0.5 m along strike. Coordinates were located by a team of SCUBA divers using a compass and a measuring tape secured to the center point of a circle. If present, the colony underneath a coordinate was sampled using a hammer and chisel to break off a small fragment of coral tissue (maximum size 1 cm x 1 cm). No colony was sampled twice. An underwater photograph was taken of each colony sampled, and colony size was measured as the maximum length, width, and height to the nearest 10 cm. Sampling of a polar plot ceased when

20 colonies had been collected. All colonies within each 15 m radius circle were counted.

Fragments were placed in individual zip-lock bags underwater and then transferred to vials containing 95% ethanol on shore. Fragments were stored in a -20°C freezer prior to genotyping.

Genotyping

DNA from coral tissue was extracted from each sample following the manufacturer's instructions in the DNeasy 96 Blood and Tissue Kit (Qiagen, CA). A total of 11 microsatellite

50 loci were used in the current study (Polato et al. 2010, Baums et al. 2012) (see Table B-2). One additional marker (pl0072) was developed for this study following Polato et al. (2010). DNA was amplified with fluorescently labeled primers in one singleplex and four multiplex reactions consisting of 2-3 primer pairs each (see Table B-2). Thermal cycling was performed in an MJ

Research PT200 (Bio-Rad, CA) or an Eppendorf Mastercycler Gradient (Eppendorf, Germany) with an initial denaturation step of 95°C for 5 min followed by 35 cycles of 95°C for 20s; 52°C-

56°C (see Table B-2) for 20s; and 72°C for 30s. A final extension of 30 min at 72°C ensured the addition of a terminal adenine to all products (Brownstein et al. 1996). Fragments were analyzed using an ABI 3730 with an internal size standard (Genescan LIZ-500, Applied Biosystems, CA;

Penn State Genomics Core Facility - University Park, PA). Electropherograms were visualized and allele sizes were called using GENEMAPPER v 4.0 (Applied Biosystems, CA). Samples that failed to amplify more than 2 of 11 loci were excluded from the analysis. This resulted in an overall average failure of 5±4% s.d. and a per locus failure rate of <12% for each locus in the included samples (n=684).

Analysis of multilocus genotype (MLG) data

MLGs were grouped into genets in GENALEX v 6.4 (Peakall & Smouse 2006) by requiring complete matches at all loci ignoring missing data. Only unique MLGs (n=448) were used in subsequent analyses. Potential genotyping errors were detected with GENCLONE v 2.0

(Arnaud-Haond & Belkhir 2007) and flagged allele calls were corrected when appropriate by re- examining electropherograms. Despite a lower number of alleles in P. evermanni, the power to distinguish unique MLGs was high in both species as calculated by the combined probability of

-10 - identity (PID) (Peakall & Smouse 2006) for each species (PIDlobata = 6.8x10 ; PIDevermanni = 1.7x10

5 ). Given the sample sizes of 377 and 307 samples and the respective PIDs we do not expect to

51 have mistakenly identified colonies as clonemates of another colony when they were in fact the result of independent sexual reproductive events in P. lobata (estimated number of misidentified colonies = 2.5x10-7) or in P. evermanni (estimated number of misidentified colonies = 5.1x10-3).

Clustering analysis

A principal coordinate analysis (PCoA) on standardized codominant genotypic distance was conducted in GENALEX vers 6.4 (Peakall & Smouse 2006) (n=448).We also clustered MLGs using Bayesian assignment methods implemented in STRUCTURE v 2.3.3 (Falush et al. 2003). The software uses a Bayesian clustering algorithm to assign genotypes to clusters that minimize deviation from Hardy-Weinberg expectations. Values of K=1 to 6 were tested where K=1 represents a single panmictic population and K=6 equals the number of subregions (see Table B-

1) by running three replicate simulations with 106 Markov Chain Monte Carlo (MCMC) repetitions each and a burn-in of 100,000 iterations with an admixture model assuming independent allele frequencies among clusters. Under the null hypothesis that P. lobata formed a cohesive genetic unit, we chose the admixture model to prevent masking a signal of gene flow.

However, admixture of individuals was rare (Fig. 3-1B). Futher, we assumed independent allele frequencies because we were looking for strong signals of differentiation. Altering the assumptions (non-admixture, independent allele frequencies) did not change the optimal number of clusters or the outcome of species assignment for any individual. In fact, under a non- admixture model at K = 2, each MLG assigned to its respective cluster with 100% probability of membership giving no evidence of hybridization between species. The Evanno et al. (2005) method implemented in STRUCTURE HARVESTER (Earl & Vonholdt 2012) was used to select the most likely number of clusters. The assignment of individual MLGs to species was congruent between PCoA and STRUCTURE in all cases.

52 Because the markers were developed for Porites lobata, we investigated whether the loci showed differential failure rates between species and if these failures were driving the clustering results. Overall failure rates were low and similar between species (P. lobata: 5%, P. evermanni

4% of samples failed to amplify at any locus). On the locus level, four out of 11 of the microsatellite loci in this analysis (pl1370, pl2258, pl0072, pl905) showed differential failure rates between species (Fishers exact test, nPe = 149, nPl = 299, df = 1, p<0.05) after Bonferonni correction for multiple testing. Including only the 7 non-significant loci does not change the outcome of clustering analyses or the membership of any of the samples in the species clusters identified with all loci (data not shown).

Clonal Structure Analysis

Clonal diversity measures (genotypic richness, genotypic diversity, and genotypic evenness) were calculated using GENODIVE (Meirmans & Van Tienderen 2004) for all randomly sampled polar plots (see Appendix B for definitions of indices and Table B-3 for results). The probability that repeated MLGs were the product of different sexual events (psex) was calculated for each repeated genotype included in the clonal structure analysis using MLGsim (Stenberg et al. 2003). The values of psex for each MLG repeated within a sampling plot were significantly low

-8 -8 -15 -15 (meanPe=4.8x10 ±7.1x10 s.d; meanPl=1.1x10 ±1.7x10 s.d) as to reject origin by sexual reproduction and instead point to an asexual origin (104 simulations, p< 0.05). Colonies were mapped in SIGMAPLOT vers 10.0 (SYSTAT, IL) using polar coordinates from the field with each symbol scaled for colony size (see Fig. 3-4).

53 Geographic distribution of species

The regional difference in the relative numbers of P. evermanni and P. lobata sampled was analyzed with an Analysis of Variance (ANOVA) using each polar plot as a sampling point and a Fisher exact test on the proportions of each species found in a region performed in SYSTAT v 13.1 (SYSTAT, IL). The regions considered for this analysis were: Oceanic Island (Cocos),

Coastal Island (Caño), and Mainland (Mainland Costa Rica). Only the polar plot data (where sampling effort was standardized) were appropriate for this analysis.

Photographic Analysis of Bleaching

Bleaching was observed primarily at two sampling sites, Caño 2 and Tres Hermanas, where P. lobata and P. evermanni co-exist. Caño 2 and Tres Hermanas plots have little topography (depth range= 2.4–5.5 m and 4.6–6.1 m respectively) over the scale sampled. The average distance between a sampled colony and its nearest sampled congener at Caño 2 was

4.44±2.44 m s.d. and at Tres Hermanas was 4.76±3.07 m s.d., thus P. evermanni and P. lobata colonies were exposed to the same water temperatures and UV light conditions.

Photos were taken of each colony sampled in the field as described above (nPl = 14, nPe =

20; nPhotos=81). Thirty-six uniformly spaced points were overlain on each image in Photoshop vers. CS5.1 (Adobe, CA). A categorical score was given for each point based on visual inspections of the health of the tissue underneath. Categories included substrate, healthy coral, bleached coral, pale coral, flag and photo scale. Flags were temporary, numbered markers of standard size (4 cm x 5 cm; Allflex USA, Inc., TX) to identify each colony. Healthy coral tissue was green to brown. Bleached tissue was white. Pale tissue was yellow when compared to a white standard included in each photograph (the photo scale). The percent tissue of each colony that

54 was bleached, healthy, and pale was estimated by dividing the number of points in that category over the total number of points in all three categories. The mean percent of bleached tissue was compared between species using a t-test in SYSTAT v 13.1 (SYSTAT, IL).

Lithophagea surveys

The density of Lithophagea bore holes in coral colonies was surveyed both in the field and photographically at sampling sites where P. evermanni and P. lobata co-occur. At Caño 2 and Drake Bay, colonies of similar size (estimated as the surface area of a 5-sided rectangular prism using max length, width and height as measured in the field) were selected for each species

(nPe=22, nPl=21; colony surface area, t-test, df=42, t=0.759, p=0.45) and the number of mussel holes was counted in situ. At sympatric sampling sites, photos (n=202) of Porites colonies were scaled using a photo scale included in each photograph in the program AXIOVISION 4.8 (Zeiss,

Germany). Visible surface area in each photo was measured using the Trace tool in AXIOVISION.

There was no difference in the average surface area measured between species (nPe = 46, nPl = 61; t-test, df = 105, t = -0.328, p = 0.743). Discernible mussel boreholes (see Fig. 3-2B) were counted using the event tool in AXIOVISION without knowledge of molecular species identification (file names did not reveal species identity). The mean density of mussels (number of mussels per cm2 of tissue) was compared between species using t-tests in SYSTAT v 13.1 (SYSTAT, IL).

Denaturing gradient gel electrophoresis (DGGE)

To determine symbiont diversity, Symbiodinium internal transcribed spacer 2 (ITS2) sequences were fingerprinted with DGGE. By using standards of known unique fingerprints and direct sequencing of dominant bands from those fingerprints the symbiont diversity can be

55 resolved to the subclade level (LaJeunesse 2002). We amplified the Symbiodinium ITS2 region in

3-5 samples of each species from the polar plots where the species co-exist (Caño 1, Caño2,

Caño5, and Tres Hermanas; nPl=18, nPe=19) using primers and amplification protocol previously described (LaJeunesse 2002). The single P. evermanni sample collected at Cocos Island and one

P. evermanni sample from the Galapágos were also run. Of the analyzed colonies, eight were bleached or partially bleached (nPl= 6, nPe= 2); nine were pale (nPl= 5, nPe=4); twelve were healthy

(nPl= 3, nPe= 9); four were healthy with some pale spots (nPl= 2, nPe=2); and four had unrecorded tissue health status (nPl=2, nPe=2). Reaction products and a diagnostic subclade C15 standard

(LaJeunesse et al. 2004) were separated by electrophoresis for 19 hours at 120 V at a constant temperature of 60°C on an 8% polyacrylamide gradient gel containing a gradient of 3.15 M urea/18% deionized formamide to 5.6 M urea/37% deionized formamide. Gels were visualized after staining with SYBRGreen (Molecular Probes, OR) using manufacturer’s specifications and photographed (see Fig. B-2 for example). DGGE profiles were characterized by their prominent bands. Two representative bands from each host species and a C15 positive control were excised, re-amplifed, Sanger sequenced (Penn State Genomics Core Facility - University Park, PA) and aligned against a published C15 ITS2 sequence (GenBank AY239369.1) (LaJeunesse et al. 2003) using GENEIOUS 5.5.6 (Biomatters, New Zealand).

56 Acknowledgements

J. Nivia-Ruiz, S. Luna, D. Wham and T. LaJeunesse provided assistance. A. Lopez, H.

Reyes-Bonilla, A. Baker and Z. Forsman provided samples. C. Fisher and J. Marden commented on drafts. Work was supported by NSF grant no. OCE – 0550294 to IBB and MEH.

Data Accessibility

Coral genotyping data and Photographs are available at Dryad (DOI:

10.5061/dryad.d6108).

Author Contributions

All authors collected samples. JNB performed laboratory and statistical analyses. MEH determined species identity of P. evermanni samples. JNB and IBB wrote the paper. All authors edited and approved the manuscript. IBB and MEH obtained funding.

57 References

Arnaud-Haond, S. & K. Belkhir. 2007. GENCLONE: a computer program to analyze genotypic data, test for clonality and describe spatial clonal organization. Mol. Ecol. Notes 7: 15-17. Baird, A.H., J.R. Guest & B.L. Willis. 2009. Systematic and biogeographical patterns in the reproductive biology of scleractinian corals. Annu. Rev. Ecol. Evol. Syst. 40: 551-571. Baker, A.C. 2003. Flexibility and specificity in coral-algal symbiosis: Diversity, ecology, and biogeography of Symbiodinium. 34: 661-689. Baker, A.C., C.J. Starger, T.R. Mcclanahan & P.W. Glynn. 2004. Corals' adaptive response to climate change. Nature 430: 741-741. Baums, I.B., M.W. Miller & M.E. Hellberg. 2006. Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr. 76: 503-519. Baums, I.B., J.N. Boulay, N.R. Polato & M. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Bellwood, D.R., T.P. Hughes, C. Folke & M. Nystrom. 2004. Confronting the coral reef crisis. Nature 429: 827-833. Boulay, J.N., J. Cortés, J. Nivia-Ruiz & I.B. Baums. 2012. High genotypic diversity of the reef- building coral Porites lobata (Scleractinia: Poritidae) in Isla del Coco National Park, Costa Rica. Rev. Biol. Trop. 60 (Suppl. 3): 279-292. Brownstein, M.J., J.D. Carpten & J.R. Smith. 1996. Modulation of non-templated nucleotide addition by tag DNA polymerase: Primer modifications that facilitate genotyping. Biotechniques 20: 1004-1010. Burkle, L.A., J.C. Marlin & T.M. Knight. 2013. Plant-pollinator interactions over 120 years: Loss of species, co-occurrence, and function. Science 339: 1611-1615. Collins, M., S.I. An, W.J. Cai, A. Ganachaud, E. Guilyardi, F.F. Jin, M. Jochum, M. Lengaigne, S. Power, A. Timmermann, G. Vecchi & A. Wittenberg. 2010. The impact of global warming on the tropical Pacific ocean and El Niño. Nat. Geosci. 3: 391-397. Cortés, J. 1997. Biology and geology of Eastern Pacific coral reefs. Coral Reefs 16: S39-S46. Cunning, R. & A.C. Baker. 2013. Excess algal symbionts increase the susceptibility of reef corals to bleaching. Nat. Clim. Chang. 3: 259-262. Darling, E.S., L. Alvarez-Filip, T.A. Oliver, T.R. Mcclanahan & I.M. Côté. 2012. Evaluating life- history strategies of reef corals from species traits. Ecol. Lett. 15: 1378-1386. Duarte, C.M. 2000. Marine biodiversity and ecosystem services: an elusive link. J. Exp. Mar. Biol. Ecol. 250: 117-131. Earl, D.A. & B.M. Vonholdt. 2012. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv. Genet. Resour. 4: 359-361. Evanno, G., S. Regnaut & J. Goudet. 2005. Detecting the number of clusters of individuals using the software structure: a simulation study. Mol. Ecol. 14: 2611-2620.

58 Falush, D., M. Stephens & J.K. Pritchard. 2003. Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies. Genetics 164: 1567-1587. Forsman, Z.H., D.J. Barshis, C.L. Hunter & R.J. Toonen. 2009. Shape-shifting corals: Molecular markers show morphology is evolutionarily plastic in Porites. BMC Evol. Biol. 9: 45. Glynn, P.W. 1974. Rolling stones among the scleractinian: mobile corallith communities in the Gulf of Panama. Proc. Second Int. Coral Reef Symp. 2: 183-188. Glynn, P.W. & G.M. Wellington. 1983. Corals and Coral Reefs of the Galápagos Islands. University of California Press, Berkeley, CA. Glynn, P.W. 1988. El-Niño Southern Oscillation 1982-1983 - Nearshore population, community, and ecoystem responses. Annu. Rev. Ecol. Syst. 19: U309. Glynn, P.W., S.B. Colley, C.M. Eakin, D.B. Smith, J. Cortés, N.J. Gassman, H.M. Guzmán, J.B. Delrosario & J.S. Feingold. 1994. Reef coral reproduction in the Eastern Pacific - Costa Rica, Panama, and Galapágos Islands (Ecuador) II. Poritidae. Mar. Biol. 118: 191-208. Glynn, P.W. & J.S. Ault. 2000. A biogeographic analysis and review of the far eastern Pacific coral reef region. Coral Reefs 19: 1-23. Guzmán, H.M. 1988. Distribución y abundancia de organismos coralívoros en los arrecifes coralinos de la Isla del Caño, Costa Rica. Rev. Biol. Trop. 36: 191-207. Guzmán, H.M. & J. Cortés. 1989. Coral reef community structure at Caño Island, Pacific Costa Rica. P.S.Z.N.I: Mar. Ecol. 10: 23-41. Highsmith, R.C. 1980. Passive colonization and asexual colony multiplication in the massive coral Porites lutea (Milne Edwards and Haime). J. Exp. Mar. Biol. Ecol. 47: 55-67. Hughes, A.R. & J.J. Stachowicz. 2004. Genetic diversity enhances the resistance of a seagrass ecosystem to disturbance. Proc. Natl. Acad. Sci. U.S.A. 101: 8998-9002. Hughes, T.P. & J.H. Connell. 1999. Multiple stressors on coral reefs: A long-term perspective. Limnol. Oceanogr. 44: 932-940. Idjadi, J.A. & P.J. Edmunds. 2006. Scleractinian corals as facilitators for other invertebrates on a Caribbean reef. Mar. Ecol. Prog. Ser. 319: 117-127. Knowlton, N. 1993. Sibling species in the sea. Annu. Rev. Ecol. Syst. 24: 189-216. LaJeunesse, T.C. 2002. Diversity and community structure of symbiotic dinoflagellates from Caribbean coral reefs. Mar. Biol. 141: 387-400. LaJeunesse, T.C., W.K.W. Loh, R. Van Woesik, O. Hoegh-Guldberg, G.W. Schmidt & W.K. Fitt. 2003. Low symbiont diversity in southern Great Barrier Reef corals, relative to those of the Caribbean. 48: 2046-2054. LaJeunesse, T.C., D.J. Thornhill, E.F. Cox, F.G. Stanton, W.K. Fitt & G.W. Schmidt. 2004. High diversity and host specificity observed among symbiotic dinoflagellates in reef coral communities from Hawaii. Coral Reefs 23: 596-603. Lasker, H.R. & M.A. Coffroth. 1999. Responses of clonal reef taxa to environmental change. Am. Zool. 39: 92-103.

59 Levitan, D.R., N.D. Fogarty, J. Jara, K.E. Lotterhos & N. Knowlton. 2011. Genetic, spatial and temporal components of precise spawning synchrony in reef building corals of the Montastraea annularis species complex. Evolution 65: 1254-1270. Meirmans, P.G. & P.H. Van Tienderen. 2004. GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms. Mol. Ecol. Notes 4: 792-794. Ollerton, J. & E. Coulthard. 2009. Evolution of animal pollination. 326: 808-809. Paine, R.T. 1966. Food web complexity and species diversity. Am. Nat. 100: 65-75. Peakall, R. & P.E. Smouse. 2006. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6: 288-295. Pinzón, J.H. & T.C. LaJeunesse. 2011. Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Mol. Ecol. 20: 311-325. Polato, N.R., G.T. Concepcion, R.J. Toonen & I.B. Baums. 2010. Isolation by distance across the Hawaiian Archipelago in the reef-building coral Porites lobata. Mol. Ecol. 19: 4661- 4677. Reusch, T.B.H., A. Ehlers, A. Hammerli & B. Worm. 2005. Ecosystem recovery after climatic extremes enhanced by genotypic diversity. Proc. Natl. Acad. Sci. U.S.A. 102: 2826-2831. Rowan, R., N. Knowlton, A. Baker & J. Jara. 1997. Landscape ecology of algal symbionts creates variation in episodes of coral bleaching. Nature 388: 265-269. Scott, P.J.B. & M.J. Risk. 1988. The effect of Lithophaga (Bivalvia, Mytilidae) boreholes on the strength of the coral Porites lobata. Coral Reefs 7: 145-151. Stenberg, P., M. Lundmark & A. Saura. 2003. MLGSIM: a program for detecting clones using a simulation approach. Mol. Ecol. Notes 3: 329-331. Thompson, J.N. 2005. The Geographic Mosaic of Coevolution. The University of Chicago Press, Chicago, IL. Toth, L.T., R.B. Aronson, S.V. Vollmer, J.W. Hobbs, D.H. Urrego, H. Cheng, I.C. Enochs, D.J. Combosch, R. Van Woesik & I.G. Macintyre. 2012. ENSO drove 2500-year collapse of Eastern Pacific coral reefs. Science 337: 81-84. Veron, J. 2000. Corals of the World. Australian Institute of Marine Science, Townsville.

Chapter 4

BIOTIC AND ABIOTIC FACTORS EXPLAIN NICHE DIFFERENTIATION BETWEEN PORITES LOBATA AND PORITES EVERMANNI IN THE EASTERN TROPICAL PACIFIC

Authors: Jennifer N. Boulay, Devon McCrossin, Iliana B. Baums

Abstract

Corals build the foundation of highly productive and diverse coral reef ecosystems. In the

Eastern Pacific, two closely related reef-building coral species in the genus Porites have overlapping but not identical distributions. A combination of abiotic and biotic factors likely drives coral species distributions. However, biotic interactions are rarely included in species distribution models. This is surprising as the survival and productivity of corals depend on intimate interactions with intracellular symbionts in the genus Symbiodinium. The mutualism breaks down in a process called bleaching, when temperature changes beyond a narrow tolerance limit. The species differ in bleaching frequency when growing in sympatry despite harboring

Symbiodinium cells belonging to the same ITS-2 clade. Either the host, the symbiont, or the combination of both may play a role in determining bleaching thresholds. Here, the hypothesis that Porites-Symbiodinium holobiont distribution in the Eastern Pacific is determined by bleaching related environmental variables was evaluated using a maximum entropy model. The model suggested that temperature during ENSO (El Niño Southern Oscillation) years and chlorophyll-a concentrations, an indicator of nutrient and light levels, were important in driving

66 P. evermanni species distribution whereas P. lobata was more influenced by minimum temperatures experienced during periods of cold associated with La Niña events. Thus, niche differentiation between P. lobata and P. evermanni can be explained by preferences of host species and their Symbiodinium strains for abiotic factors, explaining the divergent ecological response to thermal stress in two closely related coral hosts. Further, P. evermanni reproduces extensively via fragmentation mediated by the predation of triggerfish on mussels living within the coral’s skeleton. A stepwise multiple linear regression model including a fragmentation value based on the presence of the triggerfish predator and the abundance of their mussel prey suggested that biotic interactions in addition to abiotic factors influenced the relative proportion of P. evermanni to P. lobata at a given sampling site. We conclude that the inclusion of biotic interactions between coral species, their intracellular symbionts and coral associated species across will improve predictions of how coral species distributions will change as a result of a warming, acidifying and overfished ocean.

Introduction

Corals build the physical structure of the coral reef ecosystem and interact with reef- associated species across multiple trophic levels (Glynn, 2004). Climate change threatens coral reef ecosystems (Hughes et al. 2003). As a consequence, it is increasingly important to understand how biotic interactions and abiotic factors drive community structure to make appropriate predictions for effective conservation (Mumby & Van Woesik 2014). Biotic interactions have proven difficult to include in predictive models (Guisan & Thuiller 2005) especially in the marine environment where community level data at numerous sampling sites across a geographic range is markedly challenging to acquire.

67 A mutualism with an endosymbiotic alga (genus Symbiodinium) allows corals to persist in nutrient poor environments through the transfer of photosynthetic products from algal symbiont to host (Muscatine & Hand 1958, Muscatine 1980). However, the coral-algal mutualism is highly sensitive to changes in ocean temperature and UV light levels from the seasonal mean, among other stressors (Glynn 1993, 1996, Brown 1997). The breakdown of the coral-algal relationship resulting in the loss (expulsion) of symbiont is described as coral bleaching (Glynn

1993, 1996, Brown 1997). Without speedy recovery of their Symbiodinium community, corals suffer mortality hence impacting associated reef species (Kokita & Nakazono 2001).

Dinoflagellates in the genus Symbiodinium are a diverse group of algae with a range of temperature and light tolerances, and hence specific algal strains contribute to the stress resistance of the coral holobiont (Howells et al. 2012, LaJeunesse et al. 2014). Within the genus

Symbiodinium, there are nine recognized major taxonomic clades (A-I) (Rowan & Powers 1991,

Rowan & Powers 1992, Knowlton & Rohwer 2003). Diversity within clades is often classified by typing at a single genetic marker. Unique sequences at one such marker, ITS2, have been used to describe subclade lineages (“types”) within numerous clades (LaJeunesse 2002). Although transient or background Symbiodinium types might be present in a coral host (Baker 2003), most colonies harbor a single dominant algal type and often a single individual algal clone (Pettay et al. 2011, Baums et al. 2014).

Thermal stress during El Niño Southern Oscillation events causes coral bleaching and is a major factor in Eastern Tropical Pacific (ETP) coral mortality (Glynn 1988, Toth et al. 2012). In previous work, we showed that Porites lobata bleached more often than Porites evermanni at two sites with sympatric P. lobata and P. evermanni (<5 m apart) (Boulay et al. 2014). Such differential bleaching has previously been ascribed to differences in the subclade of algal symbionts hosted in co-occurring coral species (Rowan et al. 1997). However, DGGE fingerprinting of ITS2 region of the Symbiodinium algae present in P. lobata and P. evermanni

68 showed that both coral species associate primarily with Symbiodinium subclade C15 (Boulay et al. 2014). This suggested that the differential bleaching susceptibility of these coral species can either be attributed to differences in host physiology or distinct associations with closely-related strains of the ITS2-C15 subclade.

The two Porites species not only differ in bleaching frequency in the Eastern Tropical

Pacific but also in how they interact with other members of the reef ecosystem. A unique form of asexual fragmentation is present in massive Porites corals and allows populations to persist when sexual recruitment is unfavorable. Fragmentation occurs as a result of a three-way interaction of coral, mussel, and triggerfish. The coral, Porites, is bored by the endolithic mussel, Lithophaga spp., which weakens the coral’s skeletal strength (Scott & Risk 1988). Triggerfish that specialize on breaking off pieces of the coral skeleton to feed on mussels bite Porites colonies (Guzmán &

Cortés 1989) leading to fragmentation of coral colonies. The fish and coral species involved have very different geographic ranges from being restricted to the coastal region of the ETP

(triggerfish) (Berry & Baldwin 1966, Guzmán 1988) to ranging across the Pacific (P. lobata)

(Veron 2000). We have previously demonstrated that in Costa Rica and on Cocos Island, Porites lobata is not able to take advantage of an asexual strategy for maintaining populations (Boulay et al. 2012); however, its morphologically similar conger, P. evermanni does so frequently. Further,

P. evermanni dominated sample collections in coastal regions and is virtually absent in the

Galapágos where the mussels are abundant but the triggerfish are rarely seen (Boulay et al. 2014).

Variation in the distribution of these coral species and the apparent exclusion of P. lobata from parts of the range is consistent with either barriers to dispersal for P. lobata or unsuitable conditions for P. lobata and/or its Symbiodinium C15 strain. Porites lobata has a Pacific-wide range and previous analysis indicated that there were no barriers to gene flow among P. lobata reefs within the ETP (Baums et al. 2012) making it less likely that pre-settlement dispersal barriers can explain the restricted distribution of P. lobata in the Eastern Pacific. Rather, post-

69 settlement mortality possibly restricts P. lobata. Although the differential bleaching response of the two Porites species pointed to the importance of temperature and light levels, environmental sensitivity of other associated species like the mussels and triggerfish might either directly or indirectly influence the niche distribution of Porites. Thus, we used ecological niche modeling to evaluate the importance of bleaching related environmental variables such as temperature and irradiance to create a species distribution model for each coral-Symbiodinium holobiont and created a multiple linear regression model using abiotic and biotic factors to predict the relative abundance of these closely related coral species in the Eastern Pacific.

Methods

To explore niche differentiation in determining geographic distribution differences of

Porites spp., a species distribution model was constructed. The program MAXENT v 3.3.3

(Phillips et al. 2004, Phillips & Dudík 2008) was used to model the niche distributions of the

Porites holobionts across the ETP. MAXENT uses A) environmental data over a user identified region and B) coordinates of known occurrences to: 1) determine the necessary environmental conditions for survival and 2) extrapolate across the region to determine other suitable habitat.

MAXENT requires only presence data and not true absences. It uses the environmental conditions at the sampling locations selected for training data to predict the probability of occurrence across a wide geographic range.

The MAXENT model was used to evaluate whether the two coral-Symbiodinium holobionts are occupying different abiotic niches. However, we were unable to incorporate biotic interactions that may play a role in structuring these coral populations because we do not have data across such a large geographic range. Therefore, we created a predictive multiple linear

70 regression (MLR) model using geographic variables and environmental variables, but also, included biotic variables at each of our sampling sites across the ETP.

Predictor selection-MAXENT

To prevent over-fitting of the model and to focus on functional relevant rather than indirect predictors, we limited the variables considered in the niche differentiation model to those important to maintaining the coral-algal symbiosis. We thus selected sea surface temperature

(SST) and the light related variables: chlorophyll-a and irradiance.

Chlorophyll-a and irradiance directly influence photosynthesis and may have a stronger influence on the Symbiodinium partner than the coral host. Chlorophyll-a is a measure of phytoplankton abundance in the water related to the amount of nutrients. Irradiance is a measure of light penetration in the water. To account for seasonal variations related to the wet and dry seasons we separated the light related variables into wet and dry seasons. The dry season typically runs from December to April and the wet season from May to November (Arnador et al. 2006).

Large scale weather events related to El Niño Southern Oscillation (ENSO) can have a profound effect on the SST in the ETP leading to mass bleaching events (Glynn et al. 1988).

Prolonged and extreme warming events can occur during El Niño years and cooling occurs during

La Niña years. Therefore, we also divided average monthly SST observations by whether they occurred during El Niño or La Niña events. Because the coral-algal symbiosis is maintained only within a few degrees of the average SST at each site, survival of corals may be limited by SST maxima and minima rather than average temperature per se (Hoegh-Guldberg et al. 2005, Baker et al. 2008). Thus, we included minimum and maximum SST in the model. Temperature, irradiance, and chlorophyll-a measurements were not available across all sampling depths from

71 online downloadable environmental data sources, so depth was not considered in the landscape of the MAXENT model (but see section on depth in MLR model).

Data collection- MAXENT

The abiotic factors tested were maximum, minimum, and average monthly sea surface ocean temperatures for the most recent El Niño and La Niña events of 2009 to 2010 (°C; World

Ocean Atlas Series), average chlorophyll-a concentration (mgm -3) during the dry and wet seasons, and average irradiance levels (kWhm-2day-1) during the dry and wet season months averaged over 12 years (July 1983 - June 2005). All data were downloaded from NOAA’s

National Oceanic Data Center (http://www.nodc.noaa.gov/OC5/SELECT/dbsearch/dbsearch

.html). The data were converted to the necessary format for MAXENT using Excel (Microsoft,

CA) and DIVAGIS v 7.5 (LizardTech, Inc. University of California, US) (Hijmans et al. 2004).

Coordinates from a total 48 sites (Table C-1; see Baums et al. 2012 and Boulay et al.

2014 for sampling methods) where host species is known from genetic data (Boulay et al. 2014) were entered as presence data (n=29 P. evermanni; n=42 P. lobata). Note that when presented with a limited number of sampling sites over a large geographic area, species distribution models should not be used to extrapolate how species distributions might change under changing conditions or to predict occurrences in geographic areas where samples have not been collected

(Elith & Leathwick 2009, Elith et al. 2011; see Chapter 1 for more information on SDM). Thus, caution is warranted when interpreting model results.

MAXENT modeling

The model was run using a maximum of 500 iterations, a 10-5 convergence threshold, and a maximum of 105 background points. Default prevalence was set to 0.5. Data were entered in 3 cross-validated replicate runs (up to 10 additional replicates were run and did not result in changes to model outcome; data not shown), and 20% of presence data was used for training. Due to the limitations of sampling in marine environments, a large number of samples were collected on cruises or field trips to one region. The number of sites sampled in the Galapágos Islands and

Costa Rica together comprises 72% of the total number of sites. Presence only species distribution models assume that presence records are a random sample of the entire geographic area. When a small number of localized sites are utilized to predict over a large geographic area, the power of the model is reduced (Elith & Leathwick 2009, Elith et al. 2011). Therefore, training sites were selected randomly after sorting into broad geographic regions (i.e. Northern Galapágos,

Southern Galapágos, Costa Rica, Mexico, Panama, Equador, Cocos Island, Clipperton Island;

Table C-1) in order to be representative of the geographic range of the species.

In MAXENT, the average area under the receiver operating characteristic (ROC) curve

(AUC) gives an estimate of how well the model performs. High AUC values indicate high predictive power of the model. When AUC is high the probability of presence predicted by the training sites is highly concordant with the field observations from the remaining sites. AUC values equal to 0.5 are no better than random predictions. MAXENT discriminates between a confirmed presence and background points and then treats background points as pseudo-absences for the evaluation procedure (and not for the model fitting). Thus, AUC can be falsely inflated when the occurrence points are localized in a small area and the background samples are a poor representation of true-absences.

73 Each input environmental variable was evaluated for its percent contribution and percent importance to the model. The contribution of a variable was determined by adding the improvement in penalized average log likelihood compared to a null model (gain; see Chapter 1 for more explanation) for each iteration of the training algorithm (Phillips & Dudík 2008).

Importance was calculated by randomly permuting the values of each variable on training presence and background data and re-evaluating the model on the permuted data. Contribution depends on the particular path that MAXENT uses to get to the optimal solution and how much each variable contributes to the increase in gain, while the permutation importance depends only on the final MAXENT model (Phillips & Dudík 2008). A jacknife analysis (Quenouille 1956,

Tukey 1958) was also performed by removing each variable and re-evaluating the model without replacement.

MLR modeling

Using SPSS model builder version 22 (IBM Corp. Armonk, NY), a multiple linear regression (MLR) model was built using a forward stepwise automatic linear method based on optimizing the Akaiki information criterion corrected for finite sample sizes (AICC). The AICC provides a measure that takes into consideration the trade-off between goodness of fit and complexity of the model in contrast to R2 and can be used to compare between models. The lower the AICC value, the better the fit of the model. The response (dependent) variable for the model was the frequency of P. evermanni in the Porites collected (NPe/NTotal) at each sampling site. This value ranges from 0 to 1 where 0 indicates that the site was entirely dominated by P. lobata and 1 indicates that only P. evermanni could be found at the site. At 0.5, both species were represented equally in the collection. Because the response variable is a proportion, it was transformed using a logistic transformation. In total 10 independent variables (factors) were available to the model

74 build program (see below for a description of biotic and abiotic variables). Residuals were examined for pattern and normality.

Biotic interaction variables-MLR

At each sampling site, the presence or absence of the mussel consuming triggerfish,

Pseudobalistes naufragium, was determined using n=43 museum occurrence records reported in

Fish Base (http://www.fishbase.org/museum/OccurrencesList.php?genus=Pseudobalistes& species=naufragium). Where feasible, presence at sampling sites was confirmed post-hoc using visual surveys. Because these presence/absence data are categorical and not continuous abundance records, we did not include them directly as a factor in the model. We chose instead to use the P. naufragium presence/absence data in combination with the prevalence of mussels to create a theoretical variable to serve as a proxy for triggerfish mediated asexual fragmentation to be tested a priori.

Lithophagea spp. mussels were counted in Porites colonies from pictures taken during sampling and averaged by site regardless of host species composition. Presence of both the triggerfish and mussels are required to facilitate the fragmentation. Thus, we created a variable to represent the potential for asexual reproduction by fragmentation generated when triggerfish feed on mussels. This “fragmentation value” was calculated by multiplying triggerfish (0=absence;

1=presence) x average mussel density counts and ranged from 0 to a limitless value at each site where 0 indicates that fragmentation would not occur due to the absence of either the predator or the prey. Low fragmentation values would indicate that the predator was present but mussels were not commonly found and thus fragmentation due to triggerfish predation on mussels would be unlikely to occur. High fragmentation values indicate that mussels were prevalent in the coral skeletons and triggerfish were present to generate asexual fragments. The calculated

75 fragmentation value and mussel densities were highly correlated because they are identical where triggerfish were present. Therefore, we ignored mussel densities in favor of the “fragmentation value”.

Note that the genetic samples from the Galapágos Islands were collected during cruises conducted in 2000 and 2007 but photos were analyzed from a cruise in 2012. Most sites sampled in 2012 were found based on waypoints from the previous cruises. Thus sampling sites at a given island were similar among all years.

Abiotic variables-MLR

Minimum, maximum, and average annual temperature; light related variables: chlorophyll-a and irradiance; geographic variables: latitude and longitude; aragonite saturation state; and sampling depth were available factors in the model in addition to fragmentation value.

For all variables, outliers were examined for errors and corrected but were not removed if valid because they contain biologically relevant data. All variables were investigated for possible improvements to nonlinearity with the response variable using the Box-Cox family of transformations but none were applied.

Values of temperature and light related variables at all sampling coordinates were exported using DIVAGIS from the same gridfiles used in MAXENT. Latitude, longitude, and depth were recording during sampling. Depth was measured as meters below sea surface using dive computers at each sampling site. Coordinates were recording using a handheld Geographical

Positioning System (GPS) unit (Garmin) in the field.

Aragonite saturation state is a measure of the form of calcium carbonate (CaCO3) available in the water column for coral skeleton precipitation. At values greater than 0, the water is saturated; therefore, CaCO3 is available to the coral. At values less than 0, the coral skeleton

76 would dissolve. Aragonite saturation state is directly related to climate change (Kleypas et al.

1999, Hoegh-Guldberg et al. 2007). Atmospheric increases in carbon dioxide interact with the surface of the ocean. Carbon dioxide is naturally buffered in seawater through a series of chemical reactions that results in an increase in pH (known as ocean acidification) and decreased

CaCO3 (Aragonite) (Kleypas et al. 1999, Hoegh-Guldberg et al. 2007). Aragonite saturation states of surface water at atmospheric CO2 stabilization level of 380ppm (2011) were found at

DataBasin (World Resource Institute; Burke et al. 2011; http://databasin.org/datasets/37c3e7e9f81c45a9b60e6c53aa3f509f). Saturation states were divided into numerical ranges (2.5-3.0 and 3.0-3.5) and were not given as continuous data.

Resolution was 200 km x 200 km.

Results

MAXENT Modeling

The MAXENT model indicated that P. lobata was predicted to occur along the upwelling coast of the mainland (Ecuador and Eastern Panama) and offshore (Fig. 4-1A.). Porites evermanni had low probability of success across most of the ETP with a patchy area of high probability along the mainland coast (Fig. 4-1B). These same areas of higher likelihood for P. evermanni coincided with areas of lowest probability for P. lobata (see Mexican coastline Fig. 4-

1). Much of the offshore area is not actually habitable due to a narrow continental shelf and the lack of substrate except for the Galapágos Islands and Cocos Island where P. lobata dominates

(Boulay et al. 2014).

Fit of the MAXENT model was high for Porites lobata and P. evermanni. The model had an average AUC= 0.824 ± 0.149 s.d. for P. evermanni and 0.763 ± 0.047 s.d. for P. lobata.

77 Greater than 50% contribution and importance (Table 4-1) was given to the minimum temperatures in explaining the occurrence of P. lobata. The highest drop in gain was seen when minimum La Niña temperatures were removed and the highest increase in gain was observed when this variable was used in isolation for both the training and test data (Fig. C-1). As temperature decreased, the probability of P. lobata occurrence increased (Fig. 4-2A). For P. evermanni the opposite was predicted by the model. As temperature increased during periods during El Niño warming events, the probability of P. evermanni presence increased (Fig. 4-2B).

Average temperatures during the most recent El Niño events had a 30% contribution to the model and 5.4% importance (Table 4-1). The largest drop in gain was observed when average El Niño temperature was removed (Fig. C-1). The variable with the largest contribution, importance, and gain in isolation to P. evermanni was chlorophyll-a (Table 4-1). Probability of presence was high when chlorophyll was low in the wet season but more variable when chlorophyll increased.

Another strong predictor for P. lobata distribution was irradiance levels (the amount of light penetration) during the dry season (25% contribution, 36% importance). When irradiance was low in the dry season, P. lobata predicted presence increased.

MLR model

The factors that resulted in the best model (R2=0.79; AICC=17.313; F=22.48; df=23; p<0.001) for predicting the relative proportion of Porites evermanni to Porites lobata at a given site based on a maximized/minimized AICC value were: fragmentation value, depth, latitude, maximum temperature, and minimum temperature in descending order of importance (Table 4-2).

All factors were significant (P<0.05) except minimum temperature (Table 4-2). Based on these factors, 19 sites were removed due to missing data (mainly in depth observations). Plot of residuals verses fitted values showed no significant pattern and residuals showed a normal

78 distribution (Fig. 4-3; standard error of residuals=1.152, Shapiro-Wilk normality test of residuals;

W=0.98, p=0.75). The relationship between fragmentation value and proportion P. evermanni in the model was positive with higher fragmentation values associated with increasing predictions of

P. evermanni prevalence and lower fragmentation values predicting higher P. lobata prevalence

(Table 4-2).

According to the musem records, P. naufragium has been recorded off the coast of mainland Mexico, Costa Rica, Panama, Ecuador, El Salvador, and Peru. In the Galapágos Islands,

P. naufragium have been observed following strong El Niño events but no permanent populations establish (F. Rivera, pers. comm.). Based on this information, the absence of P. naufragium from visual surveys on a 2012 Galapágos cruise to 5 islands, and the lack of a museum record from the

Galapágos, we considered the Galapágos Islands to be an absence site. The frequency of P. evermanni at sites where P. naufragium were found (n=26) was significantly higher (Fig. C-2;

PefreqPnabsent=0.09±0.26 s.d; PefreqPnpresent=0.59±0.29 s.d; t-test; p<0.001; df=46) than those where the triggerfish was absent (n=22). Because of this highly significant result, when triggerfish present/absence was included in the regression model and coded as two binary character states, it was found to be a strong predictor of the relative frequency of Porites species (p<0.001; data not shown) as expected. Without including the triggerfish in the model, the number of mussels alone was not found to be a significant factor for predicting of the frequency of P. evermanni vs. P. lobata (p>0.05; data not shown). When triggerfish presence/absence was included in the model with mussel densities, both variables were selected and significant (p<0.05, data not shown). Sites exclusive to P. lobata were distributed over deeper depths than those where P. evermanni occur

(t-test; sites where P. evermanni is present n=19 vs. sites exclusive to P. lobata n=13; df=30; p<0.001; Fig. C-3).

79 Discussion

Abiotic factors strongly predicted the distributions of corals. Relatively less attention has been paid to the importance of biotic factors in determining coral species distribution despite the high species diversity and strong interaction networks on coral reefs (Glynn 2004). Here, we found that including biotic interactions between triggerfish, mussels and corals significantly improves the fit of a regression model that predicted the distribution of two massive Porites species in the Eastern Tropical Pacific. Thus, biotic as well as abiotic factors may be important in determining species distributions.

Temperature

The most important abiotic factor explaining this disparate distribution among those analyzed was temperature. The predicted divergent response of the two Porites species to temperature (Fig. 4-2A vs 4-2B) was in contrast to the species’ homogenous responses to light related environmental variables identifying temperature as the driving variable for niche differentiation between P. lobata and P. evermanni. La Niña minimum temperature was an important predictor for P. lobata and probability of occurrence increased as temperatures decreased. This is not to say that P. lobata prefer cold temperatures but that they are able to survive in periods of cold stress. For example, large cold tolerant P. lobata colonies were found at great depths in the Galapágos (20 m) where temperatures may never increase above 18 °C (I.

Baums, pers. observ.). In contrast, the occurrence of P. evermanni was predicted to increase positively with heat stress in agreement with observed differential bleaching at shallow warm water sites in Costa Rica (Boulay et al. 2014).

80 Light

Maximum entropy modeling of species distribution gave indications of what type of abiotic factors might drive the niche differentiation between P. evermanni and P. lobata holobionts. The importance of the light-level related abiotic factors, chlorophyll-a and irradiance, indicated that the symbiont community harbored within closely related Porites species might have different optimal light levels (Iglesias-Prieto & Trench 1997, Abrego et al. 2008, Thornhill et al. 2008, LaJeunesse et al. 2012).

Chlorophyll was likely influential for P. evermanni due to the proximity of P. evermanni colonies to the mainland coast. Chlorophyll-a concentration is a proxy for the density of phytoplankton near the ocean surface, and therefore, indirectly indicates nutrient richness in the water. Especially in areas with high coastal development and agriculture, nutrients are washed into rivers and flow into the near-shore coastal areas when periods of high rain follow drier conditions (Ryther & Dunstan 1971, Nixon 1995). A high biomass of phytoplankton in the water blocks out light and prevents photosynthesis of benthic organisms ( Nixon 1995,

Burkholder et al. 2007). Further, high nutrient loads (nitrogen and phosphorus) have been directly linked to temperature and UV induced coral bleaching by phosphate starvation as a result of the imbalanced supply of dissolved inorganic nitrogen and elevated phosphatase activity

(Wiedenmann et al. 2012).

When irradiance was low in the dry season, P. lobata predicted presence increased. This is expected for a species with an oceanic island distribution where phytoplankton blooms and sedimentation are less likely to decrease light penetration in the water column. The dry season is typically characterized by low cloud cover thus, irradiance levels can get dangerously high in clear water. Although, experimental manipulation to block ultraviolet radiation (UV) in Panama resulted in no difference in bleaching susceptibility from the ambient UV conditions among

81 Porites lobata colonies (D'Croz et al. 2001), synergistic effects of elevated UV and prolonged periods of high water temperatures have been shown to exacerbate the bleaching response (Coles

& Jokiel 1978, Glynn et al. 1993, Glynn 1996). This is particularly relevant because similarly high temperature and irradiance exposure occurs in the doldrums conditions generated during El

Niño events in the Galapágos (Gleason & Wellington 1993). Further, P. lobata is able to occupy depths where light might be limiting. Sites exclusive to P. lobata were distributed over deeper depths than those where P. evermanni occurred (Fig. C-3 p<0.001).

Bleaching

Field observations showed that the two Porites species differ in their bleaching response

(Boulay et al. 2014). Experimental manipulations of temperature and UV in the ETP on P. lobata implicated thermal stress regardless of UV levels as the main cause of bleaching (D'Croz et al.

2001); however, temperature influences the performance of both host and symbionts independently (e.g. symbiont free larvae (Polato et al. 2010, Polato et al. 2013) and

Symbiodinium in culture (Iglesias-Prieto et al. 1992)) and jointly (Fisher et al. 2012) so it is unclear which partner is limited by this factor.

Although it is difficult to isolate the partner responsible for driving the distributions in these coral-Symbiodinium associations, it is possible that an additional level of unrecognized genetic diversity exists below the C15 clade level. Thus, niche differentiation between the Porites species in the Eastern Tropical Pacific might be driven by either host or symbiont preferences for specific abiotic or biotic conditions. For example, populations of Symbiodinim clade C on the

Great Barrier Reef in Australia exhibited functional variation among populations of the same C1 type sourced from two different thermal environments (Howells et al. 2012). Understanding the ecological and co-evolutionary dynamics of Cnidarian-Symbiodinium mutualisms therefore

82 requires resolution of evolutionary significant units in both host and symbionts (LaJeunesse &

Thornhill 2011, Prada et al. 2014).

Biotic factors

Abiotic factors alone were shown to be strong predictors for distributions of corals.

However, incorporation of a coral fragmentation value into a multiple linear regression model demonstrated that predictive models may be enhanced by incorporating biotic interactions. In the face of a rapidly changing environment it becomes increasingly important to understand how biotic interactions drive the ecology and evolution of the foundation species (e.g. Trapezia crabs defend their coral host species against Acanthaster planci outbreaks; Glynn 1982). Single species based predictive approaches cannot consider the high impact that small changes in competitive interactions have on predicted responses to climate change (Mumby & Van Woesik 2014). Sexual reproduction was found to be high in P. lobata on offshore islands (Boulay et al. 2012) where ecological niche modeling predicts high probability of presence for this species. However, evidence for asexual reproduction facilitated by biotic interactions is also present in this system

(Boulay et al. 2014). Porites evermanni persists along the coastline through asexual fragmentation. Biotic interactions, both facultative and competitive, should be considered in future modeling based approached to determine how geographic shifts in the distributions of associated species alter the community structure.

Additional data/Future experiments

This study demonstrated the need for higher resolution data in marine environments.

Although we took measures to limit the effect of sampling bias, our predictive abilities were

83 limited because some areas in the landscape were sampled more intensively than others due to the nature of the marine environment and sampling limitations. At each sampling location or island

(n=20) there were between 1 and 6 sites sampled. The average distance between sites sampled at the same location or island was 3 km (n=59 pairs of sites; range 0.09 km to 19.6 km). Ideally, this analysis could be performed on a smaller scale where sampling sites are representative of entire geographic area. However, coarse resolution of the environmental data precludes a more focused investigation within Costa Rica or the Galapágos. Newly released data sets such as Bio-ORACLE

(23 variables; c. 9.3km grid) (Tyberghein et al. 2012) and NOAA’s Pathfinder (SST only; 4 km grid) (Casey et al. 2010) provide finer resolution of environmental variables.

Aragonite saturation state was not shown to be an important variable in predicting the distribution of Porites. However, we would like to explore the role that aragonite saturation (and thus ocean acidification) may play in skeletal strength, fragmentation and the distribution of mussels and their coral hosts in the ETP. The resolution of the available aragonite saturation state data in the region could only divide our study area into two ranges of aragonite saturation states

(high=3.0-3.5; low=2.5-3.0). To further explore the role that ocean acidification and climate change may play in this system in the future we must reduce the differences in the grain of presence data (sampling sites) and environmental data. There is some evidence that mussel counts may be higher at sites with the lower aragonite saturation state (Baums, unpublished data) and, therefore, less available calcium carbonate for skeleton deposition. Skeletal cores of Porites and finer resolution ocean chemistry data have been collected in the Galapágos Islands where a natural gradient in pH results in different aragonite saturation states in the northern Galapágos

Islands (Darwin and Wolf) in comparison to the southern islands (e.g. Marchena, Isabella,

Florena). Further, depth was shown to be an important predictor of species distribution; however, this result is likely due to differences in other environmental variables experiences at different depths, for example: light and nutrient availability and temperature. To enhance our

84 understanding of this system, depth specific data should be included in the MAXENT and MLR models rather than the actual depth.

At sites where P. lobata and P. evermanni were found simultaneously, there were significantly more mussel boreholes per cm2 of coral tissue in P. evermanni (Boulay et al. 2014) than in its congener. However, mussel abundance varies across the ETP. In the Galapágos, where

P. lobata was found without P. evermanni, mussel densities were high in the former species in absentia of the alternative host P. evermanni and triggerfish predator. Thus, Lithophagea spp. are not dependent on P. evermanni, however, may exhibit host preference. Additional experiments could be conducted to test hypotheses regarding predator removal and settlement preferences when both coral species are present. This could be explained by the variation in distributions of the three interacting community members.

Porites colonies were cored in the Galapágos Islands and tissue samples have been collected to identify species. The skeletal density of both species could be compared and incorporated into the fragmentation value to account for differences in skeletal strength between congeners. Further, the relation of this fragmentation value to the actual rate that fragments are generated in the field should be confirmed using empirical studies. Galapágos genetic samples from the 2012 cruise coinciding with photographs analyzed for mussel densities in this study are currently being analyzed and could help improve the accuracy of the model.

We are aware that many of the environmental variables utilized in this study are correlated and thus violate the assumptions of both models (e.g. maximum temperature and average temperature, temperature and latitude; depth and temperature). However, what remains unclear is the correlation between biotic variables and abiotic factors (e.g. the effect of Aragonite saturation state and temperature on mussels and triggerfish distributions). When the successes of closely interacting organisms are dependent on each other, co-evolution and interacting variations in response to selection pressures within species result in unique biogeographic patterns that

85 would not be predicted in isolation (Thompson 2005). Future work should focus on empirical tests of the response of triggerfish and mussels to climate change in isolation and in interaction to better our understanding of how the distributions of these interacting species will change across the landscape of varying selection pressures.

This study represents a first foray into exploring how environmental variables in combination with biotic interactions allow for niche differentiation in closely related coral species. Given that the two Porites species are geographically isolated without evidence of dispersal limitation, we previously suspected that both abiotic and biotic factors are important for predicting species distributions. Despite the ETP being data limited, we were able to use rudimentary modeling to confirm a number of influential biological and environmental factors such as temperature, depth, latitude, and biologically mediated fragmentation potential which follows the patterns previously observed in Boulay et al. (2014).

86 Acknowledgements

We would like to thank M. Sanders for assistance with GIS and MAXENT. Photograph analysis was conducted with help from C. Kupp and R. Clarke. Additional thanks to D. Miller for his advice on modeling biotic interactions in the system using MLR. Work was supported by NSF grant no. OCE – 0550294 to IB and the Khaled bin Sultan Living Oceans Foundation.

Author Contributions

JB and IB collected sample data. DM downloaded environmental data and ran MAXENT program. JB performed analysis and wrote the paper. IB read, edited, and approved the manuscript and obtained funding.

87 References

Abrego, D., K.E. Ulstrup, B.L. Willis & M.J. Van Oppen. 2008. Species–specific interactions between algal endosymbionts and coral hosts define their bleaching response to heat and light stress. Proc. R. Soc. B-Biol. Sci. 275: 2273-2282. Amador, J. A., Alfaro, E. J., Lizano, O. G. & Magaña, V. O. 2006. Atmospheric forcing in the Eastern Tropical Pacific: A review. Prog. Oceanogr. 69: 101-142. Baker, A.C. 2003. Flexibility and specificity in coral-algal symbiosis: Diversity, ecology, and biogeography of Symbiodinium. Annu. Rev. Ecol. Syst. 34: 661-689. Baker, A.C., P.W. Glynn & B. Riegl. 2008. Climate change and coral reef bleaching: An ecological assessment of long-term impacts, recovery trends and future outlook. Estuar. Coast. Shelf S. 80: 435-471. Baums, I.B., J. Boulay, N.R. Polato & M.E. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Baums, I.B., M.K. Devlin-Durante & T.C. LaJeunesse. 2014. New insights into the dynamics between reef corals and their associated dinoflagellate endosymbionts from population genetic studies. Mol. Ecol. 23: 4203-4215 Berry, F.H. & W.J. Baldwin. 1966. Triggerfishes (Balistidae) of the eastern Pacific. Proc. Calif. Acad. Sci. 34: 429-474. Boulay, J.N., J. Cortés, J. Nivia-Ruiz & I.B. Baums. 2012. High genotypic diversity of the reef- building coral Porites lobata (Scleractinia: Poritidae) in Isla del Coco National Park, Costa Rica. Rev. Biol. Trop. 60 (Suppl. 3): 279-292. Boulay, J.N., M.E. Hellberg, J. Cortés & I.B. Baums. 2014. Unrecognized coral species diversity masks differences in functional ecology. Proc. R. Soc. B-Biol. Sci. 281: 20131580. Brown, B.E. 1997. Coral bleaching: causes and consequences. Coral. Reefs. 16: S129-S138. Burke, L., K. Reytar, M. Spalding, & A. Perry. 2011. Reefs at Risk Revisited. Washington, DC, USA: World Resources Institute. Burkholder, J.M., D.A. Tomasko & B.W. Touchette. 2007. Seagrasses and eutrophication. J. Exp. Mar. Biol. Ecol. 350: 46-72. Casey, K. S., Brandon, T. B., Cornillon, P., & Evans, R. 2010. The past, present, and future of the AVHRR Pathfinder SST program. In: Oceanography from Space. Springer, Netherlands, 273-287. Coles, S. & P. Jokiel. 1978. Synergistic effects of temperature, salinity and light on the hermatypic coral Montipora verrucosa. Mar. Biol. 49: 187-195. D'Croz, L., J.L. Maté & J.E. Oke. 2001. Responses to elevated sea water temperature and UV radiation in the coral Porites lobata from upwelling and non-upwelling environments on the Pacific coast of Panama. Bull. Mar. Sci. 69: 203-214. Elith, J., & Leathwick, J. R. 2009. Species distribution models: ecological explanation and prediction across space and time. Ann. Rev. Ecol. Evol. Syst. 40: 677. Elith, J., Phillips, S. J., Hastie, T., Dudík, M., Chee, Y. E., & Yates, C. J. 2011. A statistical explanation of MAXENT for ecologists. Divers. Distrib. 17: 43-57.

88 Fisher, P., M. Malme & S. Dove. 2012. The effect of temperature stress on coral–Symbiodinium associations containing distinct symbiont types. Coral Reefs 31: 473-485. Gleason, D.F. & G.M. Wellington. 1993. Ultraviolet-radiation and coral bleaching. Nature 365: 836-838. Glynn, P.W. 1982. Crustacean symbionts and the defense of corals: coevolution on the reef? In: Nitecki MH (ed) Coevolution. University of Chicago Press, Chicago, 111–178. Glynn, P.W. 1988. El-Niño Southern Oscillation 1982-1983 - Nearshore population, community, and ecoystem responses. Annu. Rev. Ecol. Syst. 19: U309. Glynn, P.W. 1993. Coral reef bleaching - ecological perspectives. Coral Reefs. 12: 1-17. Glynn, P.W., R. Imai, K. Sakai, Y. Nakano & K. Yamazato. 1993. Experimental responses of Okinawan (Ryukyu Island, Japan) reef corals to hugh sea temperature and UV radiation. Proceedings of the 7th International Coral Reef Symposium1: 27-37. Glynn, P.W. 1996. Coral reef bleaching: Facts, hypotheses and implications. Glob. Change Biol. 2: 495-509. Glynn, P.W. 2004. High complexity food webs in low-diversity eastern Pacific reef-coral communities. Ecosystems 7: 358-367. Guisan, A., & Thuiller, W. 2005. Predicting species distribution: offering more than simple habitat models. Ecol. Lett. 8: 993-1009. Guzmán, H.M. 1988. Distribución y abundancia de organismos coralívoros en los arrecifes coralinos de la Isla del Caño, Costa Rica. Rev. Biol. Trop. 36: 191-207. Guzmán, H.M. & J. Cortés. 1989. Coral reef community structure at Caño Island, Pacific Costa Rica. P.S.Z.N.I: Mar. Ecol. 10: 23-41. Hijmans, R.J., L. Guarino, C. Bussink, P. Mathur, M. Cruz, I. Barrentes & E. Rojas. 2004. DIVA- GIS. v 5.0. A geographic information system for the analysis of species distribution data. Hoegh-Guldberg, O., M. Fine, W. Skirving, R. Johnstone, S. Dove & A. Strong. 2005. Coral bleaching following wintry weather. Limnol. Oceanogr. 50: 265-271. Hoegh-Guldberg, O., P.J. Mumby, A.J. Hooten, R.S. Steneck, P. Greenfield, E. Gomez, C.D. Harvell, P.F. Sale, A.J. Edwards, K. Caldeira, N. Knowlton, C.M. Eakin, R. Iglesias- Prieto, N. Muthiga, R.H. Bradbury, A. Dubi & M.E. Hatziolos. 2007. Coral reefs under rapid climate change and ocean acidification. Science 318: 1737-1742. Howells, E.J., V.H. Beltran, N.W. Larsen, L.K. Bay, B.L. Willis & M.J.H. Van Oppen. 2012. Coral thermal tolerance shaped by local adaptation of photosymbionts. Nature Climate Change 2: 116-120. Hughes, T.P., A.H. Baird, D.R. Bellwood, M. Card, S.R. Connolly, C. Folke, R. Grosberg, O. Hoegh-Guldberg, J.B.C. Jackson, J. Kleypas, J.M. Lough, P. Marshall, M. Nystrom, S.R. Palumbi, J.M. Pandolfi, B. Rosen & J. Roughgarden. 2003. Climate change, human impacts, and the resilience of coral reefs. Science 301: 929-933. Iglesias-Prieto, R., J.L. Matta, W.A. Robins & R.K. Trench. 1992. Photosynthetic response to elevated temperature in the symbiotic dinoflagellate Symbiodinium microadriaticum in culture. Proc. Natl. Acad. Sci. U. S. A. 89: 10302-10305.

89 Iglesias-Prieto, R. & R.K. Trench. 1997. Acclimation and adaptation to irradiance in symbiotic dinoflagellates. II. Response of chlorophyll-protein complexes to different photon-flux densities. Mar. Biol. 130: 23-33. Kleypas, J. A., Buddemeier, R. W., Archer, D., Gattuso, J. P., Langdon, C., & Opdyke, B. N. 1999. Geochemical consequences of increased atmospheric carbon dioxide on coral reefs. Science 284: 118-120. Knowlton, N. & F. Rohwer. 2003. Multispecies microbial mutualisms on coral reefs: The host as a habitat. Am. Nat. 162: S51-S62. Kokita, T. & A. Nakazono. 2001. Rapid response of an obligately corallivorous filefish Oxymonacanthus longirostris (Monacanthidae) to a mass coral bleaching event. Coral Reefs 20: 155-158. LaJeunesse, T.C. 2002. Diversity and community structure of symbiotic dinoflagellates from Caribbean coral reefs. Mar. Biol. 141: 387-400. LaJeunesse, T.C. & D.J. Thornhill. 2011. Improved resolution of reef-coral endosymbiont (Symbiodinium) species diversity, ecology, and evolution through psbA non-coding region genotyping. PloS One 6: e29013. LaJeunesse, T.C., J.E. Parkinson & J.D. Reimer. 2012. A genetics‐based description of Symbiodinium minutum sp. nov. and S. psygmophilum sp. nov.(Dinophyceae), two dinoflagellates symbiotic with cnidaria. J. Phycol. 48: 1380-1391. LaJeunesse, T.C., D.C. Wham, D.T. Pettay, J.E. Parkinson, S. Keshavmurthy & C.A. Chen. 2014. Ecologically differentiated stress-tolerant endosymbionts in the dinoflagellate genus Symbiodinium (Dinophyceae) Clade D are different species. Phycologia 53.4: 305-319. Mumby, P.J. & R. Van Woesik. 2014. Consequences of ecological, evolutionary and biogeochemical uncertainty for coral reef responses to climatic stress. Curr. Biol. 24: R413-R423. Muscatine, L. & C. Hand. 1958. Direct evidence for the transfer of materials from symbiotic algae to the tissues of a Coelenterate. Proc. Natl. Acad. Sci. U. S. A. 44: 1259-1263. Muscatine, L. 1980. Productivity of zooxanthellae. In: Primary productivity in the sea, Springer US, 381-402. Nixon, S.W. 1995. Coastal marine eutrophication: a definition, social causes, and future concerns. Ophelia 41: 199-219. Pettay, D.T., D.C. Wham, J.H. Pinzon & T.C. LaJeunesse. 2011. Genotypic diversity and spatial– temporal distribution of Symbiodinium clones in an abundant reef coral. Mol. Ecol. 20: 5197-5212. Phillips, S. J., Dudík, M., & Schapire, R. E. 2004. A maximum entropy approach to species distribution modeling. In: Proceedings of the twenty-first international conference on Machine learning, ACM, 83. Phillips, S.J. & M. Dudík. 2008. Modeling of species distributions with MAXENT: new extensions and a comprehensive evaluation. Ecography 31: 161-175. Polato, N.R., C.R. Voolstra, J. Schnetzer, M.K. Desalvo, C.J. Randall, A.M. Szmant, M. Medina & I.B. Baums. 2010. Location-specific responses to thermal stress in larvae of the reef- building coral Montastraea faveolata. PLoS One 5: e11221.

90 Polato, N.R., N.S. Altman & I.B. Baums. 2013. Variation in the transcriptional response of threatened coral larvae to elevated temperatures. Mol. Ecol. 22: 1366-1382. Prada, C., S.E. Mcilroy, D.M. Beltrán, D.J. Valint, S.A. Ford, M.E. Hellberg & M.A. Coffroth. 2014. Cryptic diversity hides host and habitat specialization in a gorgonian‐algal symbiosis. Mol. Ecol. 23: 3330-3340. Quenouille, M. H. 1956. Notes on bias in estimation. Biometrika 43: 353–360. Rowan, R. & D.A. Powers. 1991. A Molecular Genetic Classification of Zooxanthellae and the Evolution of Animal-Algal Symbioses. Science 251: 1348-1351. Rowan, R. & D.A. Powers. 1992. Ribosomal-Rna Sequences and the Diversity of Symbiotic Dinoflagellates (Zooxanthellae). Proc. Natl. Acad. Sci. U. S. A. 89: 3639-3643. Rowan, R., N. Knowlton, A. Baker & J. Jara. 1997. Landscape ecology of algal symbionts creates variation in episodes of coral bleaching. Nature 388: 265-269. Ryther, J.H. & W.M. Dunstan. 1971. Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science 171: 1008-1013. Scott, P.J.B. & M.J. Risk. 1988. The effect of Lithophaga (Bivalvia, Mytilidae) boreholes on the strength of the coral Porites lobata. Coral Reefs 7: 145-151. Thompson, J.N. 2005. The Geographic Mosaic of Coevolution. The University of Chicago Press, Chicago, IL. Thornhill, D.J., D.W. Kemp, B.U. Bruns, W.K. Fitt & G.W. Schmidt. 2008. Correspondence between cold tolerance and temperate biogeography in a western Atlantic Symbiodinium (Dinophyta) lineage. J. Phycol. 44: 1126-1135. Toth, L.T., R.B. Aronson, S.V. Vollmer, J.W. Hobbs, D.H. Urrego, H. Cheng, I.C. Enochs, D.J. Combosch, R. Van Woesik & I.G. Macintyre. 2012. ENSO drove 2500-year collapse of Eastern Pacific coral reefs. Science 337: 81-84. Tukey, J. W. 1958. Bias and confidence in not quite large samples. Ann. Math. Stat. 29:614. Tyberghein, L., H. Verbruggen, K. Pauly, C. Troupin, F. Mineur & O. De Clerck. 2012. Bio‐ ORACLE: a global environmental dataset for marine species distribution modelling. Global Ecology and Biogeography 21: 272-281. Veron, J. 2000. Corals of the World. Australian Institue of Marine Science, Townsville. Wiedenmann, J., C. D’angelo, E.G. Smith, A.N. Hunt, F.-E. Legiret, A.D. Postle & E.P. Achterberg. 2012. Nutrient enrichment can increase the susceptibility of reef corals to bleaching. Nat. Clim. Chang. 3: 160-164.

Table 4-1. Percent contribution and importance of each environmental variable used by MAXENT to model the probable species distributions of Porites lobata and Porites evermanni in the ETP. Porites lobata Porites evermanni Percent Percent Percent Percent Variable Season Contribution Importance Contribution Importance Temperature maximum El Niño - - - - average El Niño - - 27.9 5.4 minimum El Niño - - - - maximum La Niña - - 0.2 0 average La Niña - - 7.0 0 minimum La Niña 55.5 58.4 - - Chlorophyll-a wet - - 46.1 61.6 dry 1.4 4.7 0.3 2.2 Irradiance wet 18.6 0.9 18.5 30.7 dry 24.5 36 0 0.1

92 Table 4-2. Coefficients (±standard error), significance, and importance of each factor selected in MLR model using SPSS model builder to predict the relative proportion of Porites lobata and Porites evermanni at sampling sites in the ETP. Factor Coefficient (±SE) Significance (p-value) Importance Fragmentation value 23.38±7.87 0.007 0.28 Depth -0.14±0.05 0.011 0.24 Latitude 0.84±0.34 0.024 0.18 Maximum -0.61±0.26 0.027 0.18 Temperature Minimum temperature -0.55±0.29 0.065 0.12

94 A

(°C) B

(°C) Figure 4-2. (A) The probability of presence of Porites lobata given different minimum temperatures during La Niña compared to Porites evermanni in the background predicted by MAXENT. Black line=mean response of 3 replicates for P. lobata, Blue shading=standard deviation; Red line= mean response of 3 replicates for P. evermanni, Light gray shading= standard deviation; Dark gray horizontal line=expectation under null model. (B). The probability the presence of Porites evermanni given different average temperatures during El Niño compared to Porites lobata in the background predicted by MAXENT. Red line= mean response of 3 replicates for P. evermanni, Blue shading=standard deviation; Black line=mean response of 3 replicates for P. lobata, Dark gray horizontal line=expectation under null model. Note: These graphs indicate the marginal effect of changing exactly one variable, whereas the full model takes advantage of sets of variables changing together. Temperature is given in °C.

ETP. Red line=best fit line (B) A normal Q-Q plot of the residuals. The diagonal represents a normal distribution. Shapiro-Wilk normality test; W=0.97; p-value=0.75.

Chapter 5

GENETIC CONNECTIVITY AND SPATIAL CLONAL STRUCTURE IN A DOMINANT STAGHORN CORAL ON REEFS AROUND GUAM

Authors: Boulay JN, Burdick D, Halford A, McIllwain J, Baums IB

Abstract

Ocean warming threatens the survival of many coral species. On Guam, Staghorn

Acropora communities provide critical habitat to key fisheries species, yet they are highly susceptible to thermal and light stress. Nothing is known about the population structure of these coral communities on Guam, hampering design of effective conservation efforts. Corals reproduce asexually by fragmentation as well as sexually. Therefore, knowledge about both larval dispersal and clonal propagation is required to describe the population dynamics of Guam’s

Acropora communities. The clonal diversity of staghorn patches and their population genetic structure was assessed by genotyping n=399 colonies of Acropora cf. pulchra from multiple sites within Guam and from a neighboring island (Saipan, n = 30) using 8 microsatellite loci. The analysis revealed high levels of asexual fragmentation (NG/N ranged from 0.13 to 0.46) at considerable spatial scales (>30m) and was one of the highest compared to four other corals species analyzed with similar methods in previous studies. Data also indicate localized disturbance patterns. Bayesian clustering analysis and migration modeling illustrated patterns of dispersal and connectivity between patches of staghorn coral. Importantly, a putative dispersal barrier was discovered between Cocos Lagoon at the southern tip of Guam and Acropora patches along Guam’s entire West coast. The most likely migration model similarly suggested that the

97 Cocos Lagoon population does not contribute enough migrants per generation to the remainder of

Guam to maintain genetic homogeneity. Current data from drifter buoys detected strong eddies off the southwestern tip of the island, consistent with an oceanographic larval dispersal barrier.

Comparisons of all Acropora cf. pulchra genotypes with other Acroporids from Guam and the

Caribbean showed that the Cocos Lagoon and Saipan samples were indeed divergent populations of A. pulchra and not an unidentified species. These findings suggested that Guam’s staghorn populations should be managed as two populations with low sexual recruitment rates.

Introduction

The continued survival of many coral species is uncertain as increasing sea surface temperatures are expected to cause more frequent and intense bleaching of corals, increased prevalence of disease, and a greater frequency of destructive storm events (Hoegh-Guldberg et al.

2007). These disturbance events are exacerbated by chronic anthropogenic stress due to growing coastal human populations and increased coastal development. Globally, 19% of the original area of coral reefs has been lost, with another 15% seriously threatened within the next 10-20 years and a further 20% under threat of loss in the next 20-40 years (Wilkinson 2004). In 2012 a time series analysis across 214 reefs revealed 50% coral cover loss on the Great Barrier Reef since

1985 (De'ath et al. 2012).

Corals disperse and exchange genes primarily via planktonic larvae. High dispersal ability and the absence of biophysical barriers can lead to high genetic connectivity across a species range (Klanten et al. 2007). Large geographic distances, such as those among islands in the central Paciﬁc, might limit larval exchange among populations and lead to regional isolation

(Baums et al. 2012). Furthermore, larvae can recruit locally despite the potential to travel vast distances (Swearer et al. 1999, Ayre & Hughes 2000, Taylor & Hellberg 2003, Almany et al.

98 2007) because the presence of larval retention mechanisms and cryptic barriers might prevent coral larvae from reaching their maximum dispersal potential (Baums et al. 2005b, Baums et al.

2006b). Knowledge of dispersal breakpoints is essential to designing effective marine protected areas (Palumbi 2004). Reefs with a high degree of connectedness through regular exchange of larvae are genetically more diverse and hypothesized to be more resilient to disturbance events

(Hughes & Stachowicz 2004, Hughes et al. 2008).

While sexual larval dispersal and genetic connectivity patterns are important in determining the long-term fate of coral populations after large scale disturbance events such as mass coral bleaching (Glynn 1996), recovery from physical disturbance events such as storms and ship groundings is often a local process via asexual fragmentation (Lirman 2003, Foster et al.

2013). Significant contributions of asexual reproduction by fragmentation to population maintenance has been previously observed in both branching (Ayre & Hughes 2000, Baums et al.

2006a) and massive corals (Foster et al. 2007); asexual fragmentation allows individuals to persist in time and space until sexual reproduction via larval dispersal is possible (sensu Honnay

& Bossuyt 2005). Genotypic diversity indices provide a direct measure of the amount of sexual vs. asexual recruitment to the population. Limited larval exchange in the face of ongoing asexual reproduction through fragmentation may result in the persistence of locally well-adapted genotypes but limited diversity of genotypes. In turn, limited genotypic diversity may limit the rate of adaptation to changing conditions (Baums 2008).

In the Indo-Pacific, the branching corals of the genus Acropora, form large dense patches

(see Fig. 5-1A), provide critical habitat, and are often indicative of healthy, stable reef systems due to their high sensitivity to environmental stress (e.g. Glynn et al. 1992, Salvat 1992, Peters

1993, Johnstone & Kahn 1995). These species are commonly confined to the inner reef-flat zone, moats, and lagoons where water is retained at low tide (Veron 2000). This narrow habitat type experiences highly variable environmental conditions, exposing these corals to high UV,

99 bleaching inducing temperatures, coral diseases, sedimentation, eutrophication, other land-based sources of pollution, cyclonic waves, and even trampling at low tide. Sexual recruitment has decreased markedly on Guam’s reefs in the last few decades and is generally low throughout the

Marianas island chain compared to reefs in other parts of the Pacific (Birkeland 1997). However, the relative amount of sexual and asexual recruitment to populations of Acroporids on Guam is unknown.

Here we aim to characterize the population genetic patterns and spatial extent of clonality within dominant Acroporids in Guam by using microsatellite markers. We test the following hypotheses: (Ho) Samples of Acropora sp. from reefs around Guam show no evidence of population structure because of high levels of ongoing gene ﬂow; (Hi) Samples of Acropora sp. from Guam show population structure shaped by limited gene ﬂow across patches followed by long-term isolation. Both hypotheses can be examined using microsatellite markers at multiple spatial scales within Guam and among Guam and other Pacific islands where Acropora samples have been collected. Sensitive recognition of population clusters and inference of changes in recent levels of dispersal can be detected with several (10-15) highly polymorphic markers such as microsatellites (Rosenberg et al. 2001). This knowledge will enable better predictions for the future of Guam’s staghorn patches, especially in response to climate change pressures, and enable comparisons of the importance of asexual versus sexual reproduction to population maintenance across several coral species that were analyzed with comparable methods.

100 Methods

Species Selection and Identification

Locations of staghorn Acroporid species around Guam had been previously identified by benthic mapping and ground-truthing (D. Burdick, pers. comm.). The staghorn Acropora spp. found on Guam include A. muricata (formerly A. formosa), A. aspera, A. cf. pulchra, and less commonly A. acuminate, A. austera, A. teres, A. virgata, and A. vaughani. Although another non- target species may be locally dominant at a given site (e.g. A. muricata at Agat Bay and A. aspera at Cocos Lagoon), we focus on A. cf. pulchra for the purpose of this analysis because it occurs in sufficient densities at the greatest number of sites around the island. Species were identified by morphology in the field based on the most recent species classification for Guam (Randall &

Burdick, in prep.). Voucher specimen were collected from A. muricata, A. aspera, and A. cf. pulchra and soaked in a bleach solution for 24 hrs to remove all tissue. Skeletal structure from voucher specimen and in situ photographs were examined for species identification confirmation.

Observations of the presence of gametes, which present as pinkish septal contents, were made on broken fragments. Only colonies identified as A. aspera contained evidence of gametes.

Additional tissue samples were collected from all encountered staghorn species, genotyped, and clustered using Bayesian methods to confirm morphological species identification via molecular methods (see Sampling, Genotyping, and Bayesian Analysis; see Table D-1 for list of samples).

Sampling

In order to answer both questions of reef connectivity and clonal structure, Acropora cf. pulchra was sampled from reef flats around Guam and around Saipan between July 20, 2013 and

January 16, 2014 (Fig. 5-1B). On Guam, the sampling design followed Baums et al.(2006a) by

101 sampling Acropora cf. pulchra thickets randomly in 15m radius circular plots (see Fig. 5-2, see below), when feasible (Table 5-1). Additional samples were collected haphazardly from one site at Saipan, a neighboring island in the Marianas archipelago located 200 km to the northeast of

Guam.

Random coordinates for locations within circular plots were generated using the random number generation function in Microsoft Excel (Microsoft 2010). Coordinates consist of a heading with a precision of 5° of arc and a distance with precision 0.5 m along strike (e.g. 15° and 7.5m). Using a compass and a measuring tape secured to a point in the middle of a high density stand, coordinates were located by divers. Only colonies located under randomly generated coordinates were sampled. Once a coordinate was found, the colony underneath that coordinate was sampled by breaking off a small fragment of coral tissue (maximum size 5 cm x 1 cm) and given an identifying number to link the sample with the location within the circle. If there was no colony under a coordinate, it was crossed out and the next random number was sampled. No colony was sampled twice and the sampling of a particular plot ceased after 30 colonies were collected. A colony was defined as a continuous individual attached to the substrate by a stalk with no skeletal connections to another colony; but, large continuous thickets (Fig. 5-

1A) can obscure colony margins. An underwater photograph was taken of each colony sampled for future reference and colony size was recorded using maximum length, width, and height.

Because fragmentation in branching corals is common and each multilocus genotypes is included only once in population genetic analyses, multiple circular plots separated by a distance of at least

450m (mean dist. = 1267 ± 733m) were sampled whenever possible in an attempt to capture adequate sample sizes (>30 unique sample per site).

At sites where samples were taken for species identification or colonies did not occur in high density, haphazard sampling was conducted. There was a minimum distance of 5 m between

102 sampled colonies to minimize the number of clones. Haphazard sampling also ceased after 30 colonies were sampled.

Fragments were placed in individual zip-lock bags underwater (with the identifying number) and then transferred to vials containing 95% ethanol on the surface. Wherever possible, fragments were stored in a -20°C freezer.

Genotyping

DNA from coral tissue was extracted from each sample (n=399) following the manufacturer's instructions in the DNeasy 96 Blood and Tissue Kit (Qiagen, CA). For each colony, a multi-locus genotype (MLG) was determined using 8 microsatellite markers previously developed for the Acropora genus (Baums et al. 2005a, Tang et al. 2010; Table D-2). DNA was amplified with fluorescently labeled primers in two 10 µl multiplex polymerase chain reactions

(PCR, I and II, Table D-2). Plex I consisted of 0.1 µM each of primer pairs 166 and 192, 0.06 µM

181 primer, 1×PCR Reaction Buffer (Promega), 2mM MgCl2 (Promega), 0.2 mM dNTPs

(Bioline), 0.1 U µl–1 Taq-Polymerase (Promega GoTaq). Plex II consisted of 0.1 µM of primer pair 182, 0.06 µM 180 primer, 1×PCR Reaction Buffer (Promega), 3mM MgCl2 (Promega), 0.2 mM dNTPs (Bioline), and 0.1 U µl–1 Taq-Polymerase (Promega GoTaq). Thermal cycling was performed in an Eppendorf Mastercycler Gradient (Eppendorf, Germany) with an initial denaturation step of 94°C for 5 min followed by 35 cycles of 94°C for 20s; 54°C for 20s; and

72°C for 30s. A final extension of 30 min at 72°C ensured the addition of a terminal adenine to all products. Three singleplex reactions using primers (Ac53, Acr 1_4 and Acr1_60; Table D-2) with labeled tail sequences were conducted on the same machines with 30 cycles of 94°C for 45s;

50°C for 30s; and 72°C for 45s and a final extension of 10 min at 72°C. Plex III-V each consisted of 0.3 µM each of the reverse primer and the labeled tail, 0.03 µM forward primer with tail

103 sequence, 1×PCR Reaction Buffer (Promega), 2mM MgCl2 (Promega), 0.8 mM dNTPs (Bioline),

0.02% BSA, and 0.03 U µl–1 Taq-Polymerase (Promega GoTaq). Fragments were analyzed using an ABI 3730 with an internal size standard (Genescan LIZ500_3730, Applied Biosystems, CA).

Electropherograms were visualized and allele sizes were called using GENEMAPPER V 5.0

(Applied Biosystems, CA). Samples that failed to amplify more than 2 of 8 loci were excluded from all further analyses. This resulted in an overall average failure rate per locus of 2 ± 2% s.d. and all loci failing in less than 6% of the included A. cf. pulchra samples.

Analysis of multilocus genotype (MLG) data

POWSIM V 4.1 (Ryman & Palm 2006) was used to estimate the statistical power for testing genetic differentiation using this marker set. Type I errors (α) occurred at a frequency less than or equal to 0.05 in 1000 runs. Power (ß) was estimated to be greater than 0.95. The

-6 combined probability of identity by chance rather than descent was low (PID= 1.1x10 ).

Therefore, multi-locus genotypes (MLGs) were grouped into genets in GENALEX V 6.501

(Peakall & Smouse 2006) by requiring complete matches at all loci and ignoring missing data.

Only unique MLGs (n=127) were used in genetic connectivity analyses. Potential genotyping errors were detected with GENCLONE V 2.0 (Arnaud-Haond & Belkhir 2007) and flagged allele calls were corrected when appropriate by re-examining electropherograms. Deviations from

Hardy Weinberg Equilibrium (HWE) were examined in GENEPOP V 4.2 (Raymond & Rousset

1995). MICROCHECKER V 2.2.3 (Van Oosterhout et al. 2006) and INEST V 2.0 (Chybicki &

Burczyk 2009) were used to look for the presence of null alleles with and without accounting for inbreeding, respectively. INEST was further used to evaluate causes of HWE deviations by estimating inbreeding levels under a model including null alleles and genotyping errors. Selfing rates were estimated using RMES (David et al. 2007). Population differentiation (FST) estimates

104 were calculated using FSTAT V 2.9.3 (Goudet 1995). Input files were generated using CREATE V

1.33 (Coombs et al. 2008).

Bayesian Analysis of Genetic Structure

Once genets were delineated, a population genetic analysis was performed using the program STRUCTURE V 2.3.4 (Pritchard et al. 2000) to determine the connectivity and genetic structure of Acropora cf. pulchra around Guam. The optimal number of populations K was estimated for values between 1 and 8 by running five replicate simulations with 106 Markov

Chain Monte Carlo (MCMC) repetitions each and a burn-in of 200,000 iterations with an admixture model assuming independent allele frequencies among clusters. K=1 represents a single panmictic population and K=8 equals one greater than the number of sites or species. The results were analyzed using the Evanno et al. (2005) method implemented in STRUCTURE

HARVESTER V 0.6.93 (Earl & Vonholdt 2012) to select the most likely number of clusters.

Replicate runs were summarized using CLUMPP V 1.1.2 (Jakobsson & Rosenberg 2007) and graphs were drawn using DISTRUCT V 1.1 (Rosenberg 2004). A multispecies analysis including A. palmata (Caribbean) was also conducted with the same simulation parameters under a no admixture model with morphological species assignments as a prior to confirm the absence of misidentified genets from Acropora spp. other than A. cf. pulchra.

The program OBSTRUCT (Gayevskiy et al. 2014) was used to provide a statistical analysis of STRUCTURE results. OBSTRUCT is a software to objectively analyze ancestry profiles produced by the Bayesian population genetics software package STRUCTURE by applying a correlation coefficient (R2) statistic to determine whether a fixed factor of interest (e.g. geographic origin) correlates with inferred structure. By removing each predefined population and inferred population individually and recalculating R2 with each iteration, the program was

105 also used to determine the amount that each sampled and inferred population contributed to the overall structure in the data. Further, it tests significance by permuting ancestry profiles for the overall data set and for pairwise combinations of sampled populations. Finally, OBSTRUCT produces canonical discriminant analysis (CDA) plots to visualize the clustering of individuals by sampled population within the data similar to a principal components analysis (PCA) plot.

Clonality

Because sampling effort was constant among plots, genet (clonal) diversity measures such as genotypic richness (NG/N), diversity (GO/GE), evenness (GO/NG), ramets per genet (R/G) could be calculated per plot (see Baums et al. 2006a for explanation of indicies, Arnaud-Haond et al. 2007, Boulay et al. 2014). Clonal richness and diversity are maximized at a value of 1, where each sample represents a sexual event, and approaches 0 when the collection is dominated by a single genet (clone). These indices directly estimate the frequency of sexual recruitment verses clonal input. Using these standardized indices allowed for comparisons among species, sites, and regions (ANOVA, GO/GE). MLGSIM V 2.0 (Stenberg et al. 2003) was used to calculate the likelihood that a repeated MLG is the result of sexual reproduction (Psex) for each repeated MLG, and therefore a product of asexual reproduction. Using 1000 simulated populations under the FIS model, which takes into account deviations from HWE (following Arnaud-Haond et al. 2007), and allele frequencies from all samples (including repeated MLGs, n=387) the program generated a distribution of Psex values against which the observed Psex values were tested for significance.

Under this model, all repeated MLGs were found to be significant and thus are assumed to be the product of asexual reproduction.

106 Spatial Structure

The center points for each circular plot were recorded using a handheld GPS (Garmin,

KS); thus, each randomly sampled colony had a known geographic coordinate. For all plots average distance between clones, the maximum distance between ramets, and a standard aggregation index (AC) was calculated. AC is the probability of clonal identity among nearest neighbors (Psp) compared to the probability of clonal identity among all pairwise samples within a plot (global; Psg) following Arnaud-Haound et al. (2007). When all nearest neighbors preferentially share the same MLG, AC is maximized (at 1.0). At Ac values close to 0, the probability among nearest neighbors does not differ on average from the global one and distribution of MLGs is intermingled.

Spatial autocorrelation analyses were performed to compare the spatial structuring of genetic distance among those colonies sampled randomly (n=291; n=81 excluding clonemates) using GENALEX v 6.501 (Peakall & Smouse 2006). Pairwise linear genetic and geographic distances among all samples were calculated within binned distance classes (Smouse & Peakall

1999). To guarantee a statistically large sample size in all bins (~30 pairwise comparisons), we utilized 7 even distance classes with a size of 3m within individual plots. At the island scale, geographic distance was computed as the log(1+pairwise geographic distance in meters) and 12 variable distance classes were selected based on relevant sampling distances. For all analyses

9999 random permutation of the data were generated to create a distribution of correlation coefficients under random spatial structure. The observed correlation between genetic and geographic distance for each distance class was then compared to the distribution and considered to be significant spatial structuring if outside of the 95% confidence intervals. Heterogeneity tests to compare correlograms across sampling plots was also conducted in GENALEX v 6.501

107

(Peakall & Smouse 2006). Following Banks and Peakall (2012) significance of heterogeneity tests was declared when p<0.01.

Migration Rates

The program MIGRATE v 3.6.4 (Beerli & Felsenstein 2001) was used to estimate migration patterns among sites under multiple scenarios. Based on the results of the STRUCTURE run, FST values, and local differences in current regimes, 3 configurations of populations were considered with either 2 (Saipan/Guam and Cocos Lagoon), 3 (Saipan, Guam, Cocos Lagoon), or

4 (Saipan, Tumon/Tanguisson, Hagatna/Asan/Luminao/Agat, Cocos Lagoon) groups. For each configuration we ran multiple models. The first model placed no limitations on the amount or direction of gene flow (full model resulting in a 2 x 2, 3 x 3, or 4 x 4 migration matrix where all parameters are allowed to vary). Additional models were also run under each configuration assuming unidirectional migration (both N and S). Under the 3 and 4 population configurations, additional models were run to allow for complicated patterns of migration. In total, 11 different models were run (the full 4x4 model did not converge) and compared using log Bayes factor calculations of the raw thermodynamic and Bezier approximation scores. Each model was run using a Brownian motion mutation model for microsatellite data, a Bayesian inference analysis strategy, a slice sampler, and assuming uniform priors for effective population size (Θ) and

Migration rate (M) with upper boundary limits of 100 as appropriate for microsatellite data (P.

Beerli, pers. comm.). One long Markov Chain with 1.5x106 iterations and a burn-in of 2x104 proceeded by 4 chains under a static heating scheme were implemented for each run with 10 replicate runs performed per scenario. Posterior distributions of parameters were examined to determine if upper bounds for parameters were suitable. Acceptance ratio of genealogies and effective MCMC sample size (ESS) were examined to determine if runs were long enough to

108 reach convergence. To be satisfactory, acceptance ratios should be 0.44 and ESS >1000 (P.

Beerli, pers. comm.). For all runs ESS values were >1000 and acceptance ratios varied from 0.37 to 0.48.

Results

Genotyping

A total of 399 A. cf. pulchra adults were sampled from 12 sites around Guam and 30 individuals were sampled from Saipan (Table 5-1). Failures in amplification resulted in the removal of 12 samples from the analysis and reduced the sample size to n=387. The average number of alleles (A) ranged from 4.1 ± 0.6 to 5.6 ± 0.7 at all sites except Cocos Lagoon where allelic diversity was lower (A=2.4 ± 0.3; Diversity Index= 0.68 ± 0.12; Table 5-2; see Table D-3 for results by locus). Although tests of HWE revealed that 4 of 8 loci were in HWE for all sites, significant deviations from HWE were evident in 10 out of the 54 tests (after correction for multiple testing) mainly at locus 166 and locus 180 (heterozygote excess). One site each deviated from HWE at locus 181 and locus 182 (Table D-3).

Null allele testing in MICROCHECKER suggested that locus 166 and 181 contained null alleles for all sites. However, the program INEST, which accounts for inbreeding and genotyping errors when estimating null alleles, found significant null alleles at locus 166 in 5 of the 7 sites, at locus 181 only at Saipan, and locus 1_4 in Tumon. At locus 166, the additional 2 sites exhibited non-significant but high null allele frequencies (>0.1). Similarly, null allele frequencies were evident at some sites at locus 181, locus 53, and locus 1_4 (Table 5-3).

Estimated inbreeding levels were low (Avg(Fi)<0.035 for all sites) and all 95% confidence intervals (CI) included 0 when using the full model in INEST (nfb; including null

109 alleles (n), inbreeding (f), and genotyping errors(b)). At all sites, the model assuming null alleles and genotyping errors outperformed the full model including inbreeding (DIC values nb

(DIC values fb>nfb). At Cocos Lagoon both the “nb” and the “fb” model outperformed the “nfb” model but no deviations from HWE were observed. Under a maximum likelihood model in

RMES, no site showed evidence for a significantly non-zero selfing rate based on both the 95% CI and hypothesis testing comparing the likelihood (lnl) value of the estimated selfing rate under an unconstrained model to the lnl under a constrained selfing rate =0 (chi-square, df=1,p>0.2 al sites).

Clonal structure

High levels of fragmentation were observed within our spatially explicit random sample collection (mean NG/N = 0.28 ± 0.1 SD). Fragmentation rates varied from 0.13 to 0.46 among sites. The three plots sampled within Tumon Bay had the highest clonal richness (mean NG/N =

0.4 ± 0.07 SD). The lone plot at Asan Bay had the lowest. On average, each genet comprised 4 ramets ± 1.5 SD within plots and therefore markedly reduced our sample size for further analysis

(overall N=387; NG=127). Even where sampling was conducted haphazardly and intentionally biased against clones, repeated sampling of genotypes was common (i.e. Luminao NG/N =15/30; see discussion of spatial extent of clonality). Between plots within the same site only one colony

(Tumon N) shared a MLG (Psex p-value<0.001) with colonies from a separate plot (Tumon S). In this case, ramets were separated by a mean distance of 1.44 ± 0.005km. No other MLGs were shared between plots within or between sites. Excluding the previous case, maximum clonal extent ranged from 4.3 to 26.4m with a global average of 16.3 ± 6.3m.

110 The spatial structure within a given plot was not significantly autocorrelated except at

Agat S, where the distribution of clones exhibited high spatially aggregation (Table 5-4. Ac=0.5;

Fig. 5-2B), and W. Hagatna, where the majority of sampling occurred within 6m from the center point (Fig. 5-2D). Pairwise heterogeneity tests (Table 5-5) thus indicated that the spatial structure of genetic distance at Agat S was significantly different from the other 9 plots but no other significant differences were found between plots. The lack of spatial autocorrelation within plots was likely due to the random distributions of ramets (i.e. Asan and Tumon S) or a prevalent clone with a large spatial extent (i.e. E. Hagatna, Cocos Lagoon; see Fig. 5-2).

When pairwise comparisons from all plots were analyzed together, genetic distance was significantly autocorrelated with geographic distance for pairs of samples within 0-30 m, corresponding to the maximum distance within one circular sampling plot (radius =15m; Fig. 5-3

<1.5 log(1+m)). The strength of correlation fell drastically at distances comparable to repeated plots within one site (Fig. 5-3 classes 2.7 and 3; r=0.031 and -0.017 respectively; p<0.001 and

0.003) consistent with a well-mixed non-clonal population. At distances comparable to those between sampling sites, spatial autocorrelation was also negligible (r=-0.053 to -0.009; p<0.001).

The same analysis excluding clones (shown in red Fig 5-3) resulted in significantly higher correlation at distances between 10 and 30m (r=0.039 and 0.061; p<0.001). However, the overall pattern of high correlation at the plot level in the analysis with clones was not mirrored indicating that the clonal subrange, or the linear distance over which clones affected the genetic structure of the population (described in Arnaud-Haond et al. 2007), extended over the entire range of the plot.

111 Population Structure

Pairwise FST values revealed significant population structure (Table 5-6; Fig 5-4A) between Saipan and Guam as well as between Cocos Lagoon at the southern tip of Guam and all other sites. Allele identity was similar overall in the Cocos Lagoon samples compared to the rest of the population (no private alleles) but unusual alleles occurred at high frequencies (Table D-4) and were likely responsible for the observed high FST values. Note, that the sample size was small

(Cocos Lagoon MLG=7; low allelic diversity) so these differences in allele frequencies were possibly exaggerated.

Results of Bayesian clustering including additional species indicated that despite high FST the Cocos Lagoon samples were most likely A. cf pulchra. The most likely number of clusters for the species analysis was K=3, with all A. cf pulchra samples clustering together (Fig. 5-4B, top) and thus appropriate for population level analyses. At K=2 (most likely via the Evanno method),

MLGs from Cocos Lagoon separated with high probability of membership to a second cluster, but Saipan and Tumon MLGs are not well resolved (Fig. 5-4B, middle). At K=3, Saipan samples had high probability of membership into a third cluster but samples from Tumon assigned with equal probability to the first and third cluster (Fig. 5-4B, bottom). Additionally, no isolation by distance pattern was observed (Mantel test; data not shown). OBSTRUCT to assessed the statistical significance of the STRUCTURE results at K=2, with a high R2 value of 0.88 (p< 0.001) indicating strong diversification and/or population structure and thus rejecting the null hypothesis that ancestry is randomly scattered among the predetermined populations (Gayevskiy et al. 2014).

After removal of each predefined population and recalculation of R2, only removal of the Cocos

Lagoon population resulted in a decrease in R2 (from 0.88 to 0.77) indicating that Cocos Lagoon alone contributed to an increase in structure and the rest of the data set was well mixed. This same pattern was followed when three inferred clusters were considered. Further, removal of the

112 third inferred population resulted in an increase in R2 (+0.03) indicating that the third cluster was more homogenized than average and thus contributed less than average to the structure in the data

(Gayevskiy et al. 2014). See the Supplemental material for this chapter for canonical plots of these data (Fig. D-1).

Migration modeling in MIGRATE

For all configurations of populations (2, 3 or 4), the greatest support was shown for the unidirection model where migrants were passed to the next most northern site based on LogBayes

Factor calculations (Table 5-7) consistent with the prevailing regional current (see Discussion).

The number of migrants per generation (Nm), calculated as Θ*(M/4), estimated by this model are given in Table 5-7. Less than one migrant per generation was estimated from Cocos Lagoon (Nm

<0.8) under any population configuration and little support was shown for models allowing migration to Cocos Lagoon. Under the two population scenario, grouping Guam and Saipan together, Nm from Cocos Lagoon to Guam/Saipan was estimated to be 0.2. When Guam and

Saipan were split into separate populations, there was a high amount of migration estimated from

Guam to Saipan (Nm = 84.7) consistent with on-going gene flow detected with STRUCTURE.

Migration rates from Cocos Lagoon to Guam remained the same. When sampling sites were grouped into four population groupings: Cocos Lagoon; mid-Guam –Agat, Luminao, Hagatna, and Asan; north-Guam – Tumon and Tanguisson; and Saipan, the model gave the greatest likelihood score. Migration was strong in the northward direction with the exception of little migration from Cocos Lagoon (Nm =0.8).

113 Discussion

Sessile coral populations are connected via their planktonic larval stages, but local patch dynamics are often more influenced by fragmentation of adult colonies than sexual larval production. This is true for branching and massive reef building species (Baums et al. 2006a,

Foster et al. 2007, Pinzón et al. 2012, Boulay et al. 2014). Understanding of coral population dynamics thus requires knowledge about larval dispersal as well as frequency of fragmentation.

High resolution genetic markers can provide insights into both processes. Acropora cf. pulchra is a dominant habitat builder in shallow lagoons around Guam and achieves local dominance mainly via fragmentation with little sexual larval recruitment (Birkeland 1997), even when compared with other highly fragmenting Acroporids in the Caribbean (Baums et al. 2006a). Local populations are connected via larvae along the entire west coast of Guam with the exception of

Cocos Lagoon at the southernmost tip of Guam. Dispersal breakpoints were consistent with current patterns around Guam leading us to hypothesize that weakly swimming larvae and relatively short pelagic larval duration for this species result in dispersal patterns dominated by physical processes.

Genet diversity indicates local variation in disturbance events and re-colonization history

We found that fragmentation contributes significantly to genetic structure on the local patch scale (<30m) but not on the bay and island scale. Full occupation of space by multiple genets, seen in the more intermingled plots (low Ac; no significant spatial structure, e.g. Fig. 5-

2F) suggests a long history of clonal growth following colonization, and weak competitive interactions among genets. Only occasionally, high aggregation of genets was observed (Agat S, significant spatial structure; Fig. 5-2B; AC=0.5) consistent with either recent expansion into

114 empty space or high competition among clones resulting in exclusion (Arnaud-Haond et al.

2007). However, because the majority of plots sampled around Guam conformed to an intermingled pattern, it is unlikely that competition between clones had a strong segregating effect. Rather fragmentation and thus genet diversity is likely influenced by local variation in disturbance events and recolonization history (Lasker & Coffroth 1999, Baums et al. 2006a,

Foster et al. 2007, Pinzón et al. 2012).

Our understanding of the maximum spatial extent of fragmentation in this system was limited by the sampling plot size. Ramets of Acropora palmata in the Caribbean exhibited a maximum spatial extent of 75m (Baums et al. 2006a). One clone of Pocillopora damicornis in the Eastern Tropical Pacific was found over a linear distance of 90m (Pinzón et al. 2012).

However, as in this study, sampling did not occur over distances greater than 90m and less than

10km so full spatial clonal extent could not be resolved. Here, we observed one clone at a distance of 1.4km from ramets of the same genotype in Tumon Bay. Future investigations should increase sampling between 30 and 500m to fully resolve the clonal subrange.

Sexual recruitment of Acropora cf pulchra is low on the local patch scale

Clonal structure indices provide a direct measurement for the input of sexual recruits to the population. However, direct comparisons of clone diversity, similar to species diversity, are sensitive to sampling effort and scale. Baums et al. (2006a) were the first to use random circular sampling plots to determine reef coral clone diversity on a scale of 15 m and they now have become the standard method allowing for across species comparisons. Prior studies using random circular plots determined clonal structure in branching and massive corals in the Eastern

Tropical Pacific (ETP)(Pinzón et al. 2012, Boulay et al. 2014) and the Caribbean (Baums et al.

2006b). Genotypic diversity of A. pulchra was among the lowest measured; it was significantly

115 less than that of the massive P. lobata and the thin branching P. damicornis in a high energy environment (Fig. D-2; SPSS; ANOVA; p<0.05) but comparable to genotypic diversity of P. damicornis growing in a protected environment. Although not statistically significant, the thick- branching Caribbean A. palmata, and the massive P. evermanni from the ETP, had almost double the mean clonal diversity than A. pulchra on a scale of a 15m radius plot. Asexual recruits are likely highly successful in A. pulchra due to the combination of a calm environment and low input of sexual recruits. Indeed, deployment of 800 PVC plates in 2005 found only 16 coral recruits (mainly Pocillopora spp.; some Acropora spp. and Porites spp.) in Guam’s Asan Bay

(Minton & Lundgren 2006) These sexual recruitment rates were orders of magnitude lower than reports from 1979 (0.53 coral recruits per plate) corroborating the genetic evidence.

Although asexual reproduction may allow for the survival of genets as conditions for sexual recruitment become increasingly unfavorable with increasing temperature (Randall &

Szmant 2009, Baums et al. 2013, Woolsey et al. 2013), limited sexual recruitment and high asexual fragmentation may pose a threat to the resilience of the system (Baums 2008). Although the consequences of increased clonal reproduction have not been well examined in corals, the plant literature suggests that low clonal diversity confers higher susceptibility to environmental volatility on both the population and community level (Reusch et al. 2005). Low genotypic diversity has been shown to result in higher vulnerability to pathogens and parasites in plant communities (Booth & Grime 2003). This is of particular concern for coral populations as coral infectious diseases and bleaching events are on the rise contributing to the collapse of coral reef ecosystems (Harvell et al. 2002, Hughes et al. 2003, Jones et al. 2004). Communities dominated by a few foundation species might be particularly influenced by their genotypic diversity as these species are vital to the persistence of the ecosystem (Duffy 2006).

116 Connectivity within Guam and between Guam and the Northern Marianas Islands consistent with physical forcing

Dispersal breaks are hard to predict and often vary among co-distributed species (Baums et al. 2006b, Kelly & Palumbi 2010, Foster et al. 2012, Selkoe et al. 2014). Larval duration and oceanographic patterns have proven to be poor predictors of genetic connectivity (see Bradbury et al. 2008, Toonen et al. 2011). Here we found an unexpected, putative genetic break in the shallow reef-flat coral Acropora pulchra between Cocos Lagoon and the remainder of Guam using Bayesian clustering. Further, migration modeling suggested that migration occurs in a northward direction and fewer than 1 migrant from Cocos Lagoon is recruited into the

Guam/Saipan population each generation. It is typically assumed that >1 migrant per generation is required to maintain genetic connections (Spieth 1974).

The only other species examined in the area is a rabbitfish species that inhabits reef-flat zones (Priest et al. 2012). Among sites at Guam and Saipan, no genetic structure was observed in this strong swimming species with a pelagic larval duration of 32 days and large settlement size.

In contrast, the maximum competency period of Acropora pulchra is 14 days with peak settlement occurring at 10 days post fertilization (Baird 2001). In contrast to the rabbitfish, A. pulchra larvae are not strong swimmers and are likely more influenced by oceanographic currents. Thus, a combination of larval swimming ability and current patterns may explain the difference in dispersal break-points among these co-distributed species.

The prevailing current along the southern Marianas Islands and Guam is the Northern

Equatorial counter current which flows northwest at a rate of 0.1-0.3 ms-1 (Fig. D-3A).

Oceanographic mooring data from 1971 and drifters deployed between 1987 and 1996 (Fig. D-

3B; Wolanski et al. 2003) indicated a predominantly northward flow from Tumon and

Tanguisson along the coast of Guam toward the northern tip of the island where currents drive to the northwest. Between Tumon and Hagatna the data suggested an oceanic inflow that diverges

117 into two coastal currents upon meeting the shore. The predominant directionality of the coastal currents around Hagatna, Asan, and Luminao is westward, resulting in convergence with a northward current offshore. At Agat, the southern-most sampling location closest to Cocos

Lagoon, the flow is northwestward, away from Cocos Lagoon. Drifter data and numerical models suggested that strong eddies can form off the southwestern tip of the island with complete rotation in 4-5 days. A weak island-sized eddy also forms on the northwest side of the island with inflow toward Tumon (Fig. D-3C).

The patterns reported by Wolanski et al. (2003) are reflected in the genetic data presented here. Circulation on the lee side of the island with onshore flow at Tumon would allow for a high degree of genetic exchange among these sites. Separation of the onshore current between Tumon and Hagatna may explain why FST between these two sites was higher than expected (0.035) given close proximity. Larvae released from corals at Cocos Lagoon would travel west/northwest with a strong potential of being returned in 5 days, and thus allow for the local retention of propagules and lack of connectivity with the remainder of the island.

Thus, differences in early life history, in combination with local current patterns, presumably drive the disparity in population structure between rabbitfish and A. pulchra. Closely related species and those with similar life histories often exhibit different patterns of connectivity.

A multispecies analysis of 27 species of corals, invertebrates, fishes, and mammals similarly revealed variability among species in connectivity along the Hawaiian Archipelago (Toonen et al.

2011). Common dispersal breakpoints could be statistically identified when >50% of the sampled species shared a similar breakpoint. The data presented here argue against generalizing management implications from single-species studies. Ecosystem based surveys could be utilized to look for concurrent breaks for proper management of Guam reefs.

A bio-physical dispersal model has been developed for rabbitfish that inhabit narrow reef-flats where A. pulchra is found (Halford and McIllwain, unpublished). The existing model

118 should be modified using information gained from the genetic connectivity and migration modeling presented here to be applicable to corals. The linkage of biological and ecological parameters would allow for investigations of population-level responses under a range of future impact scenarios, thus providing local coral reef managers the foresight necessary to prepare adaptive management strategies for coping with future change. A formal combined analysis is forthcoming. Regardless of the modeling outcome, A. cf. pulchra presents as a species with low genotypic diversity and low sexual recruitment on the local patch scale. This combined with dispersal breakpoints over small spatial scales indicate that populations should be managed on local scales with the goal to maintain clone diversity.

119 Acknowledgements

We would like to thank The University of Guam and L. Raymundo for facilitating field collections and providing collection permits, M. Durante for laboratory assistance, and the

Research Computing and Cyber infrastructure Unit of Information Technology Services at The

Pennsylvania State University for providing advanced computing resources and services that have contributed to the research results reported in this paper. Additional thanks to D. Benavente and

R. Miller for assisting in sample collection. This research was supported by a NOAA Coral Reef

Conservation grant awarded to JM and AH and a Penn State CECG summer fellowship awarded to JB.

Author Contributions

JB, AH and DB collected samples. JB conducted laboratory analysis, statistical analysis, and wrote the paper. IB read, edited and approved the manuscript. IB and JM received funding from a NOAA Coral Reef Conservation grant. JB acquired funding from The Center for

Environmental geoChemistry and Genomics at Penn State University. JM provided logistical support for field work.

120 References

Almany, G. R., M. L. Berumen, S. R. Thorrold, S. Planes, & G. P. Jones. 2007. Local replenishment of coral reef fish populations in a marine reserve. Science 316:742-744. Arnaud-Haond, S. & K. Belkhir. 2007. GENCLONE: a computer program to analyse genotypic data, test for clonality and describe spatial clonal organization. Mol. Ecol. Notes 7: 15-17. Arnaud-Haond, S., C.M. Duarte, F. Alberto & E.A. Serrao. 2007. Standardizing methods to address clonality in population studies. Mol. Ecol. 16: 5115-5139. Ayre, D.J. & T.P. Hughes. 2000. Genotypic diversity and gene flow in brooding and spawning corals along the Great Barrier Reef, Australia. Evolution 54: 1590-1605. Banks, S.C. & R. Peakall. 2012. Genetic spatial autocorrelation can readily detect sex-biased dispersal. Mol. Ecol. 21: 2092-2105. Baums, I.B., C.R. Hughes & M.H. Hellberg. 2005a. Mendelian microsatellite loci for the Caribbean coral Acropora palmata. Mar. Ecol. Prog. Ser. 288: 115-127. Baums, I.B., M.W. Miller & M.E. Hellberg. 2005b. Regionally isolated populations of an imperiled Caribbean coral, Acropora palmata. Mol. Ecol. 14: 1377-1390. Baums, I.B., M.W. Miller & M.E. Hellberg. 2006a. Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr. 76: 503-519. Baums, I.B., C.B. Paris & L.M. Cherubin. 2006b. A bio-oceanographic filter to larval dispersal in a reef-building coral. Limnol. Oceanogr. 51: 1969-1981. Baums, I.B. 2008. A restoration genetics guide for coral reef conservation. Mol. Ecol. 17: 2796- 2811. Baums, I.B., J. Boulay, N.R. Polato & M.E. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Baums, I.B., M.K. Devlin-Durante, N.R. Polato, D. Xu, S. Giri, N.S. Altman, D. Ruiz, J.E. Parkinson & J.N. Boulay. 2013. Genotypic variation influences reproductive success and thermal stress tolerance in the reef building coral, Acropora palmata. Coral Reefs 32: 703-717. Beerli, P. & J. Felsenstein. 2001. Maximum likelihood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc. Nat. Acad. Sci. U. S. A. 98: 4563-4568. Booth, R.E. & J.P. Grime. 2003. Effects of genetic impoverishment on plant community diversity. J. Ecol. 91: 721-730. Boulay, J.N., J. Cortés, J. Nivia-Ruiz & I.B. Baums. 2012. High genotypic diversity of the reef- building coral Porites lobata (Scleractinia: Poritidae) in Isla del Coco National Park, Costa Rica. Rev. Biol. Trop. 60 (Suppl. 3): 279-292. Boulay, J.N., M.E. Hellberg, J. Cortés & I.B. Baums. 2014. Unrecognized coral species diversity masks differences in functional ecology. Proc. R. Soc. B-Biol. Sci. 281: 20131580. Bradbury, I.R., B. Laurel, P.V.R. Snelgrove, P. Bentzen & S.E. Campana. 2008. Global patterns in marine dispersal estimates: the influence of geography, taxonomic category and life history. Proc. R. Soc. B-Biol. Sci. 275: 1803-1809.

121 Chybicki, I.J. & J. Burczyk. 2009. Simultaneous estimation of null alleles and inbreeding coefficients. J. Hered. 100: 106-113. Coombs, J.A., B.H. Letcher & K.H. Nislow. 2008. CREATE: a software to create input files from diploid genotypic data for 52 genetic software programs. 8: 578-580. David, P., B. Pujol, F.R. Viard, V. Castella & J. Goudet. 2007. Reliable selfing rate estimates from imperfect population genetic data. Mol Ecol 16: 2474-2487. De'ath, G., K.E. Fabricius, H. Sweatman & M. Puotinen. 2012. The 27-year decline of coral cover on the Great Barrier Reef and its causes. Proc. Natl. Acad. Sci. U. S. A. 109: 17995- 17999. Duffy, J.E. 2006. Biodiversity and the functioning of seagrass ecosystems. Mar. Ecol. Prog. Ser. 311: 233-250. Earl, D.A. & B.M. Vonholdt. 2012. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv. Genet. Resour. 4: 359-361. Evanno, G., S. Regnaut & J. Goudet. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol. Ecol. 14: 2611-2620. Foster, N.L., I.B. Baums & P.J. Mumby. 2007. Sexual verses asexual reproduction in an ecosystem engineer: the massive coral Montastraea annularis. J. Anim. Ecol. 76: 384- 391. Foster, N.L., C.B. Paris, J.T. Kool, I.B. Baums, J.R. Stevens, J.A. Sanchez, C. Bastidas, C. Agudelo, P. Bush, O. Day, R. Ferrari, P. Gonzalez, S. Gore, R. Guppy, M.A. Mccartney, C. Mccoy, J. Mendes, A. Srinivasan, S. Steiner, M.J.A. Vermeij, E. Weil & P.J. Mumby. 2012. Connectivity of Caribbean coral populations: complementary insights from empirical and modelled gene flow. Mol. Ecol. 21: 1143-1157. Foster, N.L., I.B. Baums, J.A. Sanchez, C.B. Paris, I. Chollett, C.L. Agudelo, M.J.A. Vermeij & P.J. Mumby. 2013. Hurricane-driven patterns of clonality in an ecosystem engineer: the Caribbean coral Montastraea annularis. PLoS One. 8: e53283. Gayevskiy V., Klaere S., Knight S., & Goddard M.R. 2014. OBSTRUCT: A method to objectively analyse factors driving population structure using Bayesian ancestry profiles. PLoS One 9: e85196. Glynn, P.W., R. Imai, K. Sakai, Y. Nakano & K. Yamazato. 1992. Experimental responses of Okinawan (Ryukyu Islands, Japan) reef corals to high sea temperatrues and UV radiation. Proceedings of the 7th International Coral Reef Symposium 1: 27-37. Glynn, P.W. 1996. Coral reef bleaching: facts, hypotheses and implications. Glob. Change Biol. 2: 495-509. Goudet, J. 1995. FSTAT (Version 1.2): A computer program to calculate F-statistics. J. Hered. 86: 485-486. Harvell, C.D., C.E. Mitchell, J.R. Ward, S. Altizer, A.P. Dobson, R.S. Ostfeld & M.D. Samuel. 2002. Climate warming and disease risks for terrestrial and marine biota. Science 296: 2158-2162.

122 Hoegh-Guldberg, O., P.J. Mumby, A.J. Hooten, R.S. Steneck, P. Greenfield, E. Gomez, C.D. Harvell, P.F. Sale, A.J. Edwards, K. Caldeira, N. Knowlton, C.M. Eakin, R. Iglesias- Prieto, N. Muthiga, R.H. Bradbury, A. Dubi & M.E. Hatziolos. 2007. Coral reefs under rapid climate change and ocean acidification. Science 318: 1737-1742. Honnay, O. & B. Bossuyt. 2005. Prolonged clonal growth: escape route or route to extinction? Oikos 108: 427-432. Hughes, A.R. & J.J. Stachowicz. 2004. Genetic diversity enhances the resistance of a seagrass ecosystem to disturbance. Proc. Natl. Acad. Sci. U.S.A. 101: 8998-9002. Hughes, A.R., B.D. Inouye, M.T.J. Johnson, N. Underwood & M. Vellend. 2008. Ecological consequences of genetic diversity. Ecol. Lett. 11: 609-623. Hughes, T.P., A.H. Baird, D.R. Bellwood, M. Card, S.R. Connolly, C. Folke, R. Grosberg, O. Hoegh-Guldberg, J.B.C. Jackson, J. Kleypas, J.M. Lough, P. Marshall, M. Nystrom, S.R. Palumbi, J.M. Pandolfi, B. Rosen & J. Roughgarden. 2003. Climate change, human impacts, and the resilience of coral reefs. Science 301: 929-933. Jakobsson, M. & N.A. Rosenberg. 2007. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801-1806. Jones, R.J., J. Bowyer, O. Hoegh-Guldberg & L.L. Blackall. 2004. Dynamics of a temperature- related coral disease outbreak. Mar. Ecol. Prog. Ser. 281: 63-77. Kelly, R.P. & S.R. Palumbi. 2010. Genetic structure among 50 species of the northeastern pacific rocky intertidal community. PLoS One 5: e8594. Klanten, O.S., J.H. Choat & H. Lynne Van. 2007. Extreme genetic diversity and temporal rather than spatial partitioning in a widely distributed coral reef fish. Mar. Biol. V150: 659-670. Lasker, H.R. & M.A. Coffroth. 1999. Responses of clonal reef taxa to environmental change. Am. Zool. 39: 92-103. Lirman, D. 2003. A simulation model of the population dynamics of the branching coral Acropora palmata - Effects of storm intensity and frequency. Ecol. Model. 161: 169-182. Mills, L.S. & F.W. Allendorf. 1996. The one-migrant-per-generation rule in conservation and management. Conserv. Biol. 10: 1509-1518. Palumbi, S.R. 2004. Marine reserves and ocean neighborhoods: the spatial scale of marine populations and their management. Annual Review of Environment and Resources 29: 31-68. Peakall, R. & P.E. Smouse. 2006. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6: 288-295. Pinzón, J.H., H. Reyes-Bonilla, I.B. Baums & T.C. LaJeunesse. 2012. Contrasting clonal structure among Pocillopora (Scleractinia) communities at two environmentally distinct sites in the Gulf of California. Coral Reefs 31:765-777. Priest, M.A., A.R. Halford & J.L. Mcilwain. 2012. Evidence of stable genetic structure across a remote island archipelago through self-recruitment in a widely dispersed coral reef fish. Ecol. Evol. 2: 3195-3213.

123 Pritchard, J.K., M. Stephens & P. Donnelly. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945-959. Randall, C.J. & A.M. Szmant. 2009. Elevated temperature affects development, survivorship, and settlement of the elkhorn coral, Acropora palmata (Lamarck 1816). Biol. Bull. 217: 269- 282. Raymond, M. & F. Rousset. 1995. GENEPOP (Version-1.2) - Population genetics software for exact tests and ecumenicism. J. Hered. 86: 248-249. Reusch, T.B.H., A. Ehlers, A. Hammerli & B. Worm. 2005. Ecosystem recovery after climatic extremes enhanced by genotypic diversity. Proc. Natl. Acad. Sci. U.S.A. 102: 2826-2831. Rosenberg, N.A., T. Burke, K. Elo, M.W. Feldmann, P.J. Freidlin, M.a.M. Groenen, J. Hillel, A. Maki-Tanila, M. Tixier-Boichard, A. Vignal, K. Wimmers & S. Weigend. 2001. Empirical evaluation of genetic clustering methods using multilocus genotypes from 20 chicken breeds. Genetics 159: 699-713. Rosenberg, N.A. 2004. DISTRUCT: a program for the graphical display of population structure. Mol. Ecol. Notes 4: 137-138. Ryman, N. & S. Palm. 2006. POWSIM: a computer program for assessing statistical power when testing for genetic differentiation. Mol. Ecol. Notes 6: 600-602. Salvat, B. 1992. The 1991 bleaching event in the Society Islands, French Polynesia. Proceedings of the 7th International Coral Reef Symposium 1: 73. Selkoe, K.A., O.E. Gaggiotti, B.W. Bowen & R.J. Toonen. 2014. Emergent patterns of population genetic structure for a coral reef community. Mol. Ecol. 23: 3064-3079. Smouse, P.E. & R. Peakall. 1999. Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity 82: 561-573. Spieth, P.T. 1974. Gene flow and genetic differentiation. Genetics 78: 961-965. Stenberg, P., M. Lundmark & A. Saura. 2003. MLGSIM: a program for detecting clones using a simulation approach. Mol. Ecol. Notes 3: 329-331. Swearer, S.E., J.E. Caselle, D.W. Lea & R.R. Warner. 1999. Larval retention and recruitment in an island population of a coral-reef fish. Nature 402: 799-802. Tang, P.C., N.V. Wei, C.W. Chen, C.C. Wallace & C.A. Chen. 2010. Comparative study of genetic variability of AAT and CT/GT microsatellites in staghorn coral, Acropora (Scleractinia: Acroporidae). Zool. Stud. 49: 657-668. Taylor, M.S. & M.E. Hellberg. 2003. Larvae retention: genes or oceanography? Response. Science 300: 1657-1658. Toonen, R., K. Andrews, I. Baums, C. Bird, G. Concepcion, T. Daly-Engel, J. Eble, A. Faucci, M. Gaither, M. Iacchei, J. Puritz, J. Schultz, D. Skillings, M. Timmers & B. Bowen. 2011. Deﬁning boundaries for ecosystem-based management: a multispecies case study of marine connectivity across the Hawaiian Archipelago. J Mar. Bio. 2011: 460173 Van Oosterhout, C., D. Weetman & W.F. Hutchinson. 2006. Estimation and adjustment of microsatellite null alleles in nonequilibrium populations. Mol. Ecol. Notes 6: 255-256. Veron, J. 2000. Corals of the World. Australian Institue of Marine Science, Townsville.

124 Vucetich, J.A. & T.A. Waite. 2000. Is one migrant per generation sufficient for the genetic management of fluctuating populations? Anim. Conserv. 3: 261-266. Wolanski, E., R.H. Richmond, G. Davis, E. Deleersnijder & R.R. Leben. 2003. Eddies around Guam, an island in the Mariana Islands group. Cont. Shelf Res. 23: 991-1003. Woolsey, E., M. Byrne & A. Baird. 2013. The effects of temperature on embryonic development and larval survival in two scleractinian corals. 493: 179-184.

125 Table 5-1. GPS coordinates, sampling method and number of Acropora cf. pulchra sampled and genotyped at each sampling site. After genotyping individuals missing data at >3 loci were removed (NM). The number of unique genotypes (NG) by sampling location is also given. GPS locations in decimal degrees.

Site GPS Sampling Method N NM NG Agat N 13.38328, E 144.65231 Random 30 7 Haphazard 9 1 4 N 13.37637, E 144.64642 Random 30 9 *Asan N 13.47692, E 144.72046 Random 30 4 N 13.47953, E 144.72806 Haphazard 30 9 Cocos Lagoon N 13.25112, E 144.67667 Random 30 7 *E. Hagatna N 13.48538, E 144.76701 Random 30 8 *W. Hagatna N 13.47948, E 144.74541 Random 30 7 Tanguisson N 13.54803, E 144.81023 Random 30 7 Luminao N 13.46463, E 144.64723 Haphazard 30 15 Tumon N 13.50605, E 144.78826 Random 30 2 9 N 13.51274, E 144.79962 Random 30 4 11 N 13.50943, E 144.79683 Random 30 3 11 Saipan N 15.17779, E 145.75097 approximate Haphazard 30 2 19 total: 9 sites 10 random plots 399 12 127 *grouped together for population genetic analyses

126 Table 5-2. Mean and SE sample size of alleles (N), number of alleles (A), number of effective alleles (NE), Shannon diversity index (I), observed heterozygosity (HO), expected and unbiased expected heterozygosity (HE and uHE), and fixation index (F) across all loci for each site. NE = 1 / 2 2 Σpi ; HE = 1 – Σpi ; uHE = (2N / (2N-1)) * HE; F = (HE – HO) / HE); Where pi is the frequency of the ith allele for the population. See supplement for individual locus statistics by site. SE = standard error.

Site N A NE I Ho He uHe F Saipan Mean 18.4 4.3 3.0 1.07 0.49 0.53 0.55 0.09 SE 0.5 0.6 0.6 0.20 0.11 0.09 0.09 0.11 Tanguisson Mean 7.0 4.1 2.7 1.06 0.57 0.55 0.60 -0.01 SE 0.0 0.6 0.5 0.16 0.11 0.06 0.07 0.15 Tumon Mean 29.0 5.5 2.9 1.19 0.49 0.59 0.60 0.19 SE 0.7 0.8 0.5 0.16 0.09 0.06 0.06 0.11 Hagatna/Asan Mean 27.9 5.6 2.6 1.15 0.50 0.57 0.58 0.16 SE 0.1 0.7 0.3 0.14 0.09 0.06 0.06 0.09 Luminao Mean 15.0 5.5 3.0 1.24 0.53 0.61 0.63 0.12 SE 0.0 0.8 0.5 0.16 0.12 0.06 0.07 0.16 Agat Mean 20.0 5.1 2.8 1.19 0.60 0.60 0.61 -0.01 SE 0.0 0.9 0.4 0.15 0.09 0.05 0.05 0.12 Cocos Lagoon Mean 6.1 2.4 1.9 0.68 0.50 0.43 0.47 -0.11 SE 0.5 0.3 0.2 0.12 0.15 0.07 0.08 0.23 Total Mean 17.6 4.6 2.7 1.08 0.53 0.56 0.58 0.06 SE 1.1 0.3 0.2 0.06 0.04 0.02 0.03 0.05

127 Table 5-3. Mean null allele frequencies by locus and average inbreeding (FI) by site as estimated in INEst v 2.0 (Chybicki & Burczyk 2009).

Null allele frequency FI Site Acr166 Acr181 Acr192 Acr180 Acr182 Acr53 Acr1_4 Acr1_60

Agat 0.206 0.023 0.011 0.005 0.006 0.012 0.024 0.007 0.013 lower CI 0.043 0 0 0 0 0 0 0 0 upper CI 0.377 0.117 0.058 0.027 0.032 0.071 0.127 0.039 0.045 Cocos Lagoon 0.144 0.162 0.030 0.030 0.038 0.067 0.177 0.014 0.035 lower CI 0 0 0 0 0 0 0 0 0 upper CI 0.450 0.523 0.151 0.152 0.194 0.292 0.689 0.072 0.128 Hagatna/Asan 0.194 0.119 0.036 0.009 0.044 0.046 0.151 0.027 0.032 lower CI 0.049 0 0 0 0 0 0 0 0 upper CI 0.343 0.233 0.121 0.036 0.127 0.148 0.296 0.090 0.102 Luminao 0.311 0.090 0.061 0.011 0.010 0.055 0.040 0.018 0.021 lower CI 0.103 0 0 0 0 0 0 0 0 upper CI 0.507 0.227 0.186 0.047 0.044 0.194 0.165 0.078 0.072 Saipan 0.219 0.345 0.024 0.019 0.013 0.103 0.049 0.025 0.023 lower CI 0.066 0.104 0 0 0 0 0 0 0 upper CI 0.392 0.570 0.097 0.074 0.056 0.263 0.187 0.107 0.074 Tanguisson 0.155 0.026 0.022 0.010 0.011 0.019 0.026 0.023 0.028 lower CI 0 0 0 0 0 0 0 0 0 upper CI 0.447 0.144 0.119 0.056 0.064 0.104 0.142 0.119 0.103 Tumon 0.261 0.196 0.080 0.014 0.068 0.104 0.303 0.046 0.031 lower CI 0.122 0 0 0 0 0 0.079 0 0 upper CI 0.400 0.379 0.194 0.054 0.158 0.226 0.495 0.138 0.092

128 Table 5-4. Indices of clonal structure as calculated for each polar plot. Sample size (N), the number of unique MLGs (NG), clonal richness (NG/N), genotypic diversity (GO/GE), genotypic evenness (GO/NG), the average number of ramets (R) per genet (G) is given as well as the average distance between clonemates (Genet Spread), the average maximum linear extent of each genet (Clonal Extent) meters, the average clonal identity for all pairwise comparisons (Psg), the average clonal identity of nearest neighbor (Psp), the aggregation coefficient (Ac), and the reef status. SD = standard deviation.

GO/ Genet Clonal 1 1 1 Reef Site N NG NG/N GO/GE R/G Psg Psp Ac 2 NG Spread (sd) Extent (sd) Status Agat Agat N 30 7 0.23 0.11 0.49 4.29 11.39(5.99) 21.25(4.37) 0.73 0.50 0.32 fair Agat S 30 9 0.30 0.15 0.49 3.33 7.42(5.98) 12.97(10.36) 0.79 0.40 0.50 fair Tanguisson 30 7 0.23 0.14 0.58 4.29 11.53(5.6) 18.94(3.92) 0.78 0.70 0.10 good *Hagatna W Hagatna 30 7 0.23 0.17 0.71 4.29 7.75(5.56) 14.02(5.56) 0.83 0.57 0.32 varied E Hagatna 30 8 0.27 0.12 0.45 3.75 7.5(4.38) 13.76(4.98) 0.75 0.57 0.24 varied *Asan heavily 30 4 0.13 0.09 0.71 7.50 11.14(4.99) 21.47(0.94) 0.67 0.6 0.11 impacted Cocos Lagoon 30 7 0.23 0.06 0.24 4.29 11.98(5.87) 16.54(13.5) 0.42 0.47 -0.12 best Tumon Ypao Beach (S) 28 9 0.32 0.18 0.56 3.11 10.55(5.33) 17.86(4.6)† 0.83 1.00 -0.20 recovering Fujita Rd. (N) 26 12 0.46 0.18 0.38 2.17 11.07(5.2) 18.99(6.48) 0.81 0.88 -0.09 recovering Matapang (C) 27 11 0.41 0.15 0.38 2.45 8.34(4.05) 12.37(5.14) 0.79 0.78 0.01 recovering Total Avg 29.1 8.1 0.28 0.13 0.50 3.95 9.96 16.31 SD 1.5 2.3 0.10 0.04 0.15 1.48 3.77 6.34 *Grouped together for population genetic analyses † excludes clone found in Tumon N plot which changes the clonal extent of that clone from 24.9 to 1446 m resulting in an average clonal extent of 302.51(640.91 sd) m. 1(Arnaud-Haond et al. 2007) 2(Porter et al. 2005, Burdick et al. 2008)

129 Table 5-5. Omega values (below diagonal) and associated p-values (above diagonal) from heterogeneity tests to compare spatial autocorrelation in 7 distance classes of 3m between each pair of plots. Following Banks and Peakall (2012), it is recommended that significance of the heterogeneity test is declared when p<0.01 (indicated by underlined italics). Agat Agat W. E. Cocos Tumon Tumon Tumon Tanguisson Asan N S Hagatna Hagatna L. S N C Agat N - <0.01 0.47 0.42 0.13 0.26 0.55 0.29 0.56 0.38 Agat S 35.37 - <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 Tanguisson 13.72 38.70 - 0.58 0.94 0.28 0.75 0.39 0.56 0.76 W. 14.33 33.19 12.41 - 0.17 0.37 0.28 0.19 0.66 0.44 Hagatna E. Hagatna 19.93 43.15 6.81 18.80 - 0.02 0.55 0.16 0.18 0.55 Asan 16.90 51.92 16.47 15.07 27.44 - 0.51 0.62 0.85 0.75 Cocos L. 12.68 47.91 10.26 16.48 12.69 13.19 - 0.99 0.93 0.87 Tumon S 16.52 50.74 14.82 18.30 19.28 11.80 4.40 - 0.92 0.72 Tumon N 12.68 38.32 12.61 11.41 18.57 8.67 7.00 7.32 - 0.90 Tumon C 14.88 49.43 10.09 14.13 12.66 10.19 8.25 10.51 7.85 -

130

Table 5-6. Pairwise FST by site. Significantly non-zero values after 420 permutations (p<0.002 after adjustment for multiple comparisons) are given in underlined italic. Saipan Tanguisson Tumon Hagatna/Asan Luminao Agat Cocos L. -- Saipan 0.039 -- Tanguisson 0.032 0.015 -- Tumon 0.042 0.004 0.035 -- Hagatna/Asan 0.034 -0.025 0.021 0.006 -- Luminao 0.044 0.005 0.011 -0.003 0.002 -- Agat Cocos L. 0.147 0.160 0.127 0.153 0.170 0.162 --

131 Table 5-7. Log Bayes factors (2* difference in log-likelihood) based on Bezier approximation score (a) and the estimated number of migrants per generation (b-d) calculated as the median estimated mutation-scaled population size (Θ) x the median estimated mutation-scaled migration rate / 4. The most likely model (N) for each configuration of populations is underlined. N = northward migration, S = southward migration, N + Cocos L. to NG = a channel of migration from Cocos Lagoon to northern Guam sites in addition to southern Guam, 0 Cocos L. = full model except no migration to and from Cocos Lagoon, 0 Saipan to S = full model but no migration from Saipan to the south. a) Number of Populations Migration Model 2 3 4 Full -33427 -55608 - N -6341 -4742 0 S -11027 -5814 -679 N+Cocos L. to NG - - -869 0 Cocos L. - -120730 - 0 Saipan to S - -39542 -

b) Two populations Number of Migrants per Generation Receiving Population Θ Source Population Saipan/Guam Cocos L. 7.7 Saipan/Guam - - 0.8 Cocos L. 0.2 -

c) Three populations Number of Migrants per Generation Receiving Population Θ Source Population Saipan Guam Cocos L. 1.6 Saipan - - - 6.8 Guam 84.7 - - 0.8 Cocos L. - 0.2 - d) Four populations Number of Migrants per Generation Receiving Population Θ Source Population Saipan North Guam Mid Guam Cocos L. 1.6 Saipan - - - - 39.6 North Guam 556.9 - - - 6.1 Mid Guam - 57.7 - - 2.23 Cocos L. - - 0.8 -

132

Figure 5-1. Continuous thicket of staghorn Acropora at Cocos Lagoon (A) Map of Study Sites (B).

133

mean distance = 1.44 ± 0.005km

134

135

1.0

0.0

136

Chapter 6

CONCLUSIONS AND SIGNIFICANCE

Almost 20 years ago, Knowlton (1993) proposed that hidden species diversity in the sea hinders our understanding of marine ecosystem function and limits our ability to predict how reefs will respond to climate change. Although cited over 600 times since, few studies have directly addressed even the first aspect of this hypothesis, much less the implications for understanding coral's responses to climate change. In this thesis, I demonstrated that unresolved diversity within a foundation species (Chapter 3) has not only obscured an understanding of community interactions, but also masked differences in reproduction, susceptibility to climate change, and niche differentiation (Chapter 4) along the margins of the species ranges.

Here, I showed genetically that what has been considered a single species of reef coral in the Eastern Tropical Pacific (Porites lobata) includes a second unrecognized species, P. evermanni, and that despite their similar appearance the two differ ecologically in ways critical to reef function and persistence (Chapter 3). Despite harboring the same Symbiodinium ITS2-type, their response to elevated temperatures and importance of light related variables differs determining their distribution across the range (Chapter 4). I further demonstrate that they differ in reproductive strategies. P. lobata is mainly sexual (Chapter 2), and P. evermanni is largely clonal (Chapter 3). Clonal reproduction of these massive Porites colonies proceeds via an interesting three-way biotic interaction in contrast to the more common physical breakage of branches in other coral species. The bioeroding mussel, Lithophaga spp., is the main food source for predatory triggerfish. To access their prey, the fish expectorate “rolling stone” coral fragments which can re-attach to the substrate. At sites sympatric for the coral host, mussels occur in higher

137

densities in P. evermanni. These differences argue against recent trends of utilizing broad functional groups based on morphology or Symbiodinium types to project the future of reefs and for the inclusion of biological interaction into ecological models.

No previous study has demonstrated functional differences between morphologically similar coral species. Corals in the Orbicella (nee Montastraea) annularis species complex in the

Caribbean differ in their reproductive timing and hybridization potential (Levitan et al. 2011), but no study has shown differences in the way these species interact with other invertebrates and vertebrate species as we demonstrate here for Porites species. The ecology and evolution of reproductive mutualisms has received a lot of interest in the terrestrial literature (e.g. Ollerton &

Coulthard 2009, Burkle et al. 2013) but similar studies are rare in corals. Here I show that asexual reproduction of corals might indeed be dependent on other members of the reef community.

Previous work has concentrated on analyzing speciation, clonal structure, gene flow and dispersal patterns of marine invertebrates in the Caribbean and Pacific (Dennis & Hellberg 2010,

Baums et al. 2012, Pinzón et al. 2012, Foster et al. 2013) but the combination of genetics with ecological interactions between corals and associated invertebrates and vertebrates is novel.

These findings are crucial for understanding past and future research in the ETP and on coral reefs globally. The observed functional differences presented in this thesis provoke additional hypotheses for continued research on resilience of ecosystems in marginal environments, maintenance and diversity of symbioses, asexual reproduction, and community interactions across geographic ranges of interacting species.

The multi-interaction system described in this thesis provides an opportunity to evaluate the hypothesis of niche construction and ecological inheritance recently discussed by Mumby and van Woesik (2014). Organisms that modify their environment alter the selection gradient on the

138

rest of the community. It follows that the presence of a nice constructor provides an evolutionary driver for an interacting niche recipient in the form of a non-genetic, ecological inheritance. The interaction between mussels, P. evermanni, and triggerfish reveals the way in which the presence of one species can alter reproduction of another and thus has the potential to result in evolutionary change in the system. Biological interactions (both positive e.g. facilitation and negative e.g. competition) have been shown to influence patterns of colonization and primary succession at hydrothermal vents (Shea et al. 2008). Future studies should formally test the niche construction hypothesis in the shallow water corals in the ETP and evaluate it under a changing climate.

Large-scale disturbances due to warming sea-surface temperatures and increasing storm activity have resulted in dramatic population declines of reef communities. Population maintenance of coral communities relies on a sufficient supply of young corals, be they of sexual or asexual origin. Because corals are sessile as adults, connectivity among reefs is dependent on the movement of planktonic larvae. The development of single-copy, high-resolution molecular markers (microsatellites) has invigorated research on connectivity, dispersal and reproduction in corals. A long standing hypothesis predicts that the time larvae spend in the water column

(planktonic larval duration, or PLD) should be related to the distance over which larvae disperse successfully. Although this assumption holds for the most extreme cases (larvae that crawl out of the mother colony and settle with in centimeters or those that spend many weeks in the water column), species with intermediate PLDs do not seem to follow a predictable pattern. However, location specific, shared dispersal breakpoints among reef species do emerge when large numbers of co-distributed species are analyzed (Lubchenco et al. 2003, Toonen et al. 2011, Selkoe et al.

2014). The unpredictability of shared dispersal breakpoints from life history data implies that each vulnerable reef communities will need to be investigated individually. The reefs of Guam

139

are one such vulnerable reef community and I show here that a dispersal breakpoint exists for staghorn Acropora along the west coast of Guam (Chapter 5). The presence of a barrier to dispersal in this system is contrary to unrestricted gene flow around Guam observed in a reef associate fish species, but consistent with oceanic currents and life-history characteristics within this Acropora species. Further, asexual recruitment was found to be the dominant process involved in population maintenance across large spatial scales in Guam resulting in low levels of genotypic diversity (Chapter 5). Although the consequences of increased clonal reproduction have not been well examined in corals, plant literature suggests that low genotypic diversity within ecosystem engineers confers higher susceptibility to environmental volatility on both the population and community level (Reusch et al. 2005).

Current genetic markers make it possible to investigate coral communities to populations and back. The work presented in this thesis uncovered an interaction network among closely related coral species and their associates suitable for thinking about processes pertinent to the survival of coral populations and their dependent communities. That missing information may help us make the critical links to understanding the future of coral reefs. As the climate changes and temperatures get warmer P. evermanni and P. lobata dominated communities will possibly respond differently. Without resolving their ecological differences we would not be able to make such predictions about the coral community. This thesis is one of the few bodies of work that clearly demonstrates that species, genetic, and functional diversity are important to our understanding of coral populations and communities.

140

References

Baums, I.B., J. Boulay, N.R. Polato & M.E. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Burkle, L.A., J.C. Marlin & T.M. Knight. 2013. Plant-pollinator interactions over 120 years: Loss of species, co-occurrence, and function. Science. 339: 1611-1615. Dennis, A.B. & M.E. Hellberg. 2010. Ecological partitioning among parapatric cryptic species. Molo. Ecol. 19: 3206-3225. Foster, N.L., I.B. Baums, J.A. Sanchez, C.B. Paris, I. Chollett, C.L. Agudelo, M.J.A. Vermeij & P.J. Mumby. 2013. Hurricane-Driven Patterns of Clonality in an Ecosystem Engineer: The Caribbean Coral Montastraea annularis. PLoS One 8: e53283. Knowlton, N. 1993. Sibling species in the sea. Annu. Rev. Ecol. Syst. 24: 189-216. Levitan, D.R., N.D. Fogarty, J. Jara, K.E. Lotterhos & N. Knowlton. 2011. Genetic, spatial and temporal components of precise spawning synchrony in reef building corals of the Montastraea annularis species complex. Evolution 65: 1254-1270. Lubchenco, J., S.R. Palumbi, S.D. Gaines & S. Andelman. 2003. Plugging a hole in the ocean: The emerging science of marine reserves. Ecol. Appl. 13: S3-S7. Mumby, P.J. & R. Van Woesik. 2014. Consequences of ecological, evolutionary and biogeochemical uncertainty for coral reef responses to climatic stress. Curr. Biol. 24: R413-R423. Ollerton, J. & E. Coulthard. 2009. Evolution of animal pollination. Science 326: 808-809. Pinzón, J., H. Reyes-Bonilla, I. Baums & T. LaJeunesse. 2012. Contrasting clonal structure among Pocillopora (Scleractinia) communities at two environmentally distinct sites in the Gulf of California. Coral Reefs 3: 765-777. Reusch, T.B.H., A. Ehlers, A. Hammerli & B. Worm. 2005. Ecosystem recovery after climatic extremes enhanced by genotypic diversity. Proc. Natl. Acad. Sci. U.S.A. 102: 2826-2831. Selkoe, K.A., O.E. Gaggiotti, B.W. Bowen & R.J. Toonen. 2014. Emergent Patterns of Population Genetic Structure for a Coral Reef Community. Mol. Ecol. 23:3064-3079. Shea, K., Metaxas, A., Young, C. R., & Fisher, C. R. 2008. Processes and interactions in macrofaunal assemblages at hydrothermal vents: a modeling perspective. Magma to Microbe 259-274. Toonen, R., K. Andrews, I. Baums, C. Bird, G. Concepcion, T. Daly-Engel, J. Eble, A. Faucci, M. Gaither, M. Iacchei, J. Puritz, J. Schultz, D. Skillings, M. Timmers & B. Bowen. 2011. Deﬁning boundaries for ecosystem-based management: a multispecies case study of marine connectivity across the Hawaiian Archipelago. J Mar. Bio. 2011: 460173

141

Appendix A

SUPPLEMENTAL MATERIAL FOR CHAPTER 2

Table A-1. Microsatellite loci for Porites lobata. The primer sequences are preceded by the name of fluorescent dye used (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in four multiplex and one singleplex reaction (Plex). Annealing Marker Size Reference Primer Sequence Repeat Temp Plex Name (bp) (°C) F: 6FAM-GTTTGCCTCTCTTCTGTTCATT (ATCC)6 ATT Polato et PL0340 (CGTT)4 TGTT 216-275 52 A al., 2010 R: AACATTATGGCTAGTTCTTTGAACG (CATT)3 Polato et F: VIC-GCCAGTAGGTGGATACACTGTT PL0780 (ATT)4 (GTT)7 136-163 52 A al., 2010 R: CAAGTACGTTGACGTCGTTG Baums et F: NED-GGTCCAAAGTCCACCATCA (ATC)9 ACC PL0905 126-183 52 A al., 2012 R: TGGTGGAAATAAGTGGTCGA (ATC)9 Polato et F:PET- ATGTCCCTGAAACGGAAGTA (ACC)7… PL1357 252-300 52 D al., 2010 R: GATGATGATGTTGTTGATGGTG (ATC)4…(ACC)7 Baums et F: PET-GCACTGTCTGTAACAAGCGAA PL1370 (GTT)8 189-246 54 E al., 2012 R: CATATTGGAAGGAGGGCTC Baums et F: 6FAM-AAACGTTCCCTATCCCATCC PL1483 (GTT)10 143-173 54 E al., 2012 R: GCAAAGCTGCTACATTTCACTAA Polato et F: PET-TGTTTCTGAGTGGCTGTGCT PL1551 (GTT)8 178-196 52 A al., 2010 R: GGTTGGAAAGGGTCCTTCAT Polato et F: PET-CGTTGACGTAACCTTCACCA PL1556 (ATC)10 153-168 56 B al., 2010 R: CACAGGGTAACCTTCCTTGC Polato et F: 6FAM-CCTTGGTTAATTTGCCCTTG PL1629 (GCT)8 168-180 52 D al., 2010 R: ACCAGTCCGGAGTCAAGCTA Baums et F: VIC-TAAGCCACAGCAGGTGTACG PL1868 (AAC)10 179-206 52 D al., 2012 R: AAACGTTCCCTATCCCATCC F: PET-CGCAGTTCCTTTGATTTGGT Polato et PL2069 R: (GTT)8 249-267 52 C al., 2010 GTTTCTTTAGCGGTTGATGGCTTGTTAC Polato et F: NED-ATTAGCGGATGAAGCGAAGA PL2258 (GAT)10 217-250 56 B al., 2010 R: TCCAATGTAACGCCAAATCA

142

Appendix B

SUPPLEMENTAL MATERIAL FOR CHAPTER 3

Definitions

Clonal Structure Indices

Genotypic richness was calculated by dividing the number of unique genotypes by the total number of samples (Ng/N) (Stoddart 1983). Genotypic diversity was calculated by dividing the observed genotypic diversity over the expected (Go/Ge) (Stoddart 1983). This measure is equivalent to evenness corresponding to Nei’s index (eve) as calculated in GENODIVE. Because each individual is expected to be genetically distinct, Ge is equal to the number of colonies genotyped. Observed genotypic diversity, Go, was calculated as:

where gi is the relative frequency of the ith of k genotypes estimated as ni (the number of samples of genotype i found in the total number of samples) divided by N. If colonies in the population are all sexually produced, Go will be equal to N. Therefore, genotypic diversity is maximized in a solely sexual population (Go/Ge=1) and approaches 0 when all colonies are asexually produced. Lastly, a measure of genotypic evenness was calculated by dividing observed genotypic diversity by the number of unique genotypes (Go/Ng) (Stoddart 1983). In contrast to richness, evenness is more influenced by genet longevity than recruitment. Evenness will approach 0 when the population is dominated by one genotype but will approach 1 when each

143 genotype is equally represented in number of colonies. However, this statistic is meaningless when populations consist of only one genotype.

Supplemental Results

Clustering Results

The first three principal coordinates explained >80% of the variation. The first principal component captured the differentiation between the two Porites species, explaining > 60% of the variation (Fig. 3-1A).

STRUCTURE HARVESTER supported K = 2 as the most likely number of clusters. Support for K = 2 clusters was strong; adding a second cluster to the STRUCTURE analysis resulted in a

26% increase in likelihood from assuming just one cluster, while further increases in the number of clusters resulted in <2% increases in likelihood using this marker set.

Genotyping results

Strong clustering of MLGs into two groups in the PCA and STRUCTURE analysis can be attributed to the presence of fixed alleles at three loci and low allelic diversity in P. evermanni.

(Fig. B-1). Rarefied allelic richness estimates based on a standardized sample size (n=15) resulted in a mean number of alleles per locus per sampling location of 4.07±0.23 s.d. alleles in P. lobata and 2.40±0.17 s.d. alleles in P. evermanni. These differences in allelic diversity between species are likely due to ascertainment bias rather than biology because the markers were developed to be highly variable in P. lobata (Polato et al. 2010, Baums et al. 2012).

Species distributions

No difference was found in the number of colonies sampled per plot per region

(ANOVA, square root transformed, df = 2, F = 0.574, p = 0.5; post-hoc pairwise tests (Tukey), p

144 > 0.5 for all comparisons) or in the area sampled (standardized by design). The proportion of P. evermanni differed among all regions tested (ANOVA, arcsine square root transformed, df = 2, F

= 73.383, p<0.001; post-hoc pairwise tests (Tukey), p < 0.005 for all comparisons; Fishers exact tests, Bonferonni correction, p<0.001). Mainland Costa Rica had the highest proportion of P. evermanni, followed by the coastal island and then the oceanic island where P. evermanni was nearly absent (Fig. 3-3).

145 Supplemental References

Baums, I.B., J.N. Boulay, N.R. Polato & M. Hellberg. 2012. No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418-5433. Polato, N.R., G.T. Concepcion, R.J. Toonen & I.B. Baums. 2010. Isolation by distance across the Hawaiian Archipelago in the reef-building coral Porites lobata. Mol. Ecol. 19: 4661- 4677. Stoddart, J.A. 1983. A genotypic diversity measure. J. Hered. 74: 489-490.

146 Table B-1. Porites lobata (Pl) and Porites evermanni (Pe) samples (n = 684) were obtained from three regions (North, Central, and South) and 17 sites in the Eastern Tropical Pacific using haphazard (H) and/or random (R) sampling methods. Sites are arranged in approximately west-to- east and north-to-south order. Given is the ratio of the number of unique multilocus genotypes by species (genets; NgPl, NgPe) over the total sample size (NPl, NPe). GPS locations are in decimal degrees (WGS84). The number of polar plots per site is given where applicable. See Table B-3 for polar plot level information. Sampling Region Subregion Site Site Name Ng /N Ng /N Latitude Longitude Pl Pl Pe Pe Method(s) North Clipperton CL01 Clipperton 5/51 2/2 10.29989 -109.216 H Mexico ME01 Ixtapa 0/0 3/17 17.65436 -101.6234 H Central Costa Rica CR01* San Juanillo 0/0 5/50 10.31117 -85.74408 H; R (n=1)

CR02* Marino 1 22/29 23/34 9.104583 -83.7068 H; R (n=1) Ballena CR03* Manuel 1 3/3 17/27 9.381867 -84.1438 H Antonio CR04* Caño Island 60/791 24/43 8.71067 -83.8911 H; R (n=3) CR05* Drake Bay 1/81 6/12 8.6713 -83.7267 CR06* Gulfo Dulce 29/301 43/61 8.727433 -83.3863 H; R (n=2) CR07* Cocos Island 53/561 1/1 5.534834 -87.0875 R (n=3) † 2 Panama PA01* N Panama 13/131 9/11 7.8168 -81.75 H 3 PA02* S Panama 4/41 7/7 8.63 -79.03 H † South Galapágos GA01 Darwin 36/401 4/5 1.616525 -91.9733 H GA02 Wolf 35/411 0/0 1.336935 -91.806 H GA03 Marchena 24/361 0/0 0.318369 -90.4691 H GA04 Central 12/131 0/0 -0.72846 -90.059 H Galapágos4 GA05* Espanola 0/0 4/10 -1.3501 -89.618 H Ecuador EC01 La Llorona 2/201 1/27 1.476383 -80.7937 H Total 299/377 149/307

1 Samples from Baums et al. 2012 2 Uva, Coibita 3 Contadora, Saboga 4 Santiago, Baltra, Floreana *Note that site numbers differ from sites described in Baums et al. 2012 †Collection does not include samples from Forsman et al. 2009

147 Table B-2. Microsatellite loci for Porites lobata (Pl) and Porites evermanni (Pe). The primer sequences are preceded by the name of the fluorescent dye used (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in four multiplex and one singleplex reaction (Plex) using the annealing temperatures (Temp) indicated.

Marker Temp Primer Sequence Repeat Size (bp) Plex Name (°C) F: NED-CACGCCTTTCTATTACGTTGA PL0072 (AACG)10 205-407 54 E R: CACCCTCTTACACTTCATTCATT F: VIC-GCCAGTAGGTGGATACACTGTT PL0780* (ATT)4 (GTT)7 136-163 52 A R: CAAGTACGTTGACGTCGTTG

† F: NED-GGTCCAAAGTCCACCATCA (ATC)9 ACC PL0905 126-183 52 A R: TGGTGGAAATAAGTGGTCGA (ATC)9 F:PET- ATGTCCCTGAAACGGAAGTA (ACC)7…(ATC) Pl: 252-300 PL1357* 52 D R: GATGATGATGTTGTTGATGGTG 4…(ACC)7 Pe: 100-111 F: PET-GCACTGTCTGTAACAAGCGAA PL1370† (GTT)8 189-246 54 E R: CATATTGGAAGGAGGGCTC F: PET-TGTTTCTGAGTGGCTGTGCT PL1551* (GTT)8 178-196 52 A R: GGTTGGAAAGGGTCCTTCAT F: PET-CGTTGACGTAACCTTCACCA PL1556* (ATC)10 153-168 56 B R: CACAGGGTAACCTTCCTTGC F: 6FAM-CCTTGGTTAATTTGCCCTTG PL1629* (GCT)8 168-180 52 D R: ACCAGTCCGGAGTCAAGCTA F: VIC-TAAGCCACAGCAGGTGTACG Pl: 185-206 PL1868† (AAC)10 52 D R: AAACGTTCCCTATCCCATCC Pe: 179 F: PET-CGCAGTTCCTTTGATTTGGT Pl:249-267 PL2069* (GTT)8 52 C R: GTTTCTTTAGCGGTTGATGGCTTGTTAC Pe: 264 F: NED-ATTAGCGGATGAAGCGAAGA PL2258* (GAT)10 217-250 56 B R: TCCAATGTAACGCCAAATCA * Polato et al. 2010 † Baums et al. 2012

148 Table B-3. Indices of clonal structure as calculated for each polar plot. Colony sizes were estimated as maximum length multiplied by maximum width from measurements taken in the field. Go = same as Ne, Ge = N. The average number of ramets (R) over the number of genets (G) is given as well as average colony size. Size 2 Species Site Plot N Ng Ng/N Go Go/Ge Go/Ng R/G (m ) ± s.d. Caño Island Caño1 12 4 0.33 1.71 0.14 0.43 3.0 0.9±1.2 Caño2 10 5 0.50 2.50 0.25 0.50 2.0 0.3±0.3

Caño5 9 4 0.44 3.00 0.33 0.75 2.25 2.9±2.5 Marino Ballena Tres Hermanas 10 4 0.40 2.38 0.24 0.60 2.5 1.3±2.6 San Juanillo Punta Pleito 27 5 0.19 1.48 0.05 0.30 5.4 1.1±3.9

P. evermanni P. Gulfo Dulce Aguja 25 17 0.68 11.79 0.47 0.69 1.47 0.3±0.5 Sandalo 25 19 0.76 16.03 0.64 0.84 1.32 0.3±0.3 Cocos Island *Punta Ulloa 1 1 1.00 1.00 1.00 1.00 1.0 - Total 118 58 0.49 38.90 0.33 0.67 2.0 0.8±2.2

Mean* 16.85 8.29 0.47 5.56 0.30 0.59 2.6 1.0 s.d. 8.32 6.68 0.20 5.86 0.20 0.19 1.3 1.4 Caño Island Caño1 8 8 1.00 8.00 1.00 1.00 1.0 1.2±1.1 Caño2 10 5 0.50 3.57 0.36 0.71 2.0 0.2±0.2

Caño5 11 9 0.82 8.07 0.73 0.90 1.2 1.4±1.5 Marino Ballena Tres Hermanas 4 2 0.50 1.60 0.40 0.80 2.0 0.2±0.2

P. lobata P. Cocos Island Punta Ulloa 17 16 0.94 15.21 0.89 0.95 1.1 0.1±0.3 Bahia Weston 20 18 0.90 16.67 0.83 0.93 1.1 0.7±0.8 Punta Maria 20 20 1.00 20.00 1.00 1.00 1.0 0.9±1.0 Total 90 78 0.87 73.12 0.81 0.94 1.15 0.7±1.0

Mean 12.85 11.14 0.81 10.45 0.75 0.90 1.3 0.7

s.d. 6.23 6.89 0.22 6.95 0.27 0.11 0.5 0.5

* P. evermanni means exclude Punta Ulloa

149

150

151

Appendix C

SUPPLEMENTAL MATERIAL FOR CHAPTER 4

Table C-1. GPS coordinates given in decimal degrees of sampling sites from Baums et al. 2012 and Boulay et al. 2014 included in MAXENT and MLR models, Porites species composition (L= P. lobata; E= P. evermanni), and Pseudobalistes naufragium presence (P)/absence (A) at each site. Island/ Latitude Longitude Porites P. naufragium Region Location Site Name °N °E spp. presence Clipperton Clipperton Clipperton 10.299889 -109.216363 L/E unknown Mexico Ixtapa IZM 17.622485 -101.521356 E P IZZ 17.654359 -101.623395 E P San Juanillo Costa Rica San Juanillo Beach 10.31117 -85.744083 E P Punta Pleito 10.38183 -85.747 E P Marino Tres Ballena Hermanas 9.104583 -83.706817 L/E P Roca Bellena 9.107033 -83.728017 L/E P Tombolo 9.143033 -83.757117 L/E P BTW 3H and RB 9.108633 -83.713667 L/E P Manuel Manuel Antonio Antonio 9.381867 -84.143867 L/E P Drake Bay San Josecito 8.6713 -83.727 L/E P Caño Island Caño 1 8.71097 -83.86517 L/E P Caño 2 8.71205 -83.882433 L/E P Caño 3 8.7107 -83.89113 L/E P Caño 4 8.698383 -83.88155 L/E P Caño 5 8.70905 -83.871 L/E P Caño 6 8.7097 -83.91515 L P Gulfo Dulce Sandalo 8.578033 -83.346667 L/E P Aguja 8.578617 -83.355 E P Punta Camibar 8.722383 -83.410183 L/E P Bajo Camibar 8.727433 -83.386317 L/E P Cocos Cocos Island Island Punta María 5.534833 -87.0875 L A Punta Ulloa 5.550281 -87.032408 L/E A Bahía Weston 5.552278 -87.050889 L A

152

Table C-1. Cont’d. Island/ Latitude Longitude Porites P. naufragium Region Location Site Name °N °E spp. presence Panama Heliport 7.64692 -81.7043 L/E P Coral Collection Point 7.81686 -81.75 L/E P WestWall 7.816 -81.75 L/E P PlayaSueca 8.62675 -79.0303 L/E P Saboga 8.633064 -79.066831 L/E P Northern Galapágos Darwin Darwin 1.616525 -91.9733 L A Site1 1.336935 -91.806 L A Site2 0.318369 -90.4691 L/E A Site3 -0.72846 -90.059 L A Site4 -1.3501 -89.618 L A Site6 1.67598 -92.0074 L A Wolf Site 1 1.38497 -91.81294 L A Site 2 1.39155 -91.82101 L A Anchorage 1.67505 -91.99302 L A Southern Itabaca -0.478166 -90.260189 Galapágos Balta L A North -0.393741 -90.2761 Seymour L A Florena Champion -1.221036 -90.393983 L A Devils Crown -1.21643 -90.4222 L A Marchena Pt. Espejo 0.30779 -90.40228 L A Site 2 0.32348 -90.40108 L A Site 3 0.32447 -90.40167 L A Espanola Bajo Gardner -1.34918 -89.6398 E A Saboga Cousins -0.236963 -90.57309159 L A Ecuador La Llorona La Llorona 1.476383 -80.7937 L/E A

153

154

0.7

0.6 in totalin 0.5

0.4

P. evermanni evermanni P. 0.3

0.2

0.1

coral collection (#Pe/#Pe+#Pl) collection coral Frequency of Frequency 0 P. naufragium absent P. naufragium present

155

Figure C-3. Box and whisker plots of sampling depth (m) of sites where P. evermanni can be found in comparison to sites that are exclusively colonized by P. lobata. Y-axis depicts visual cross-section into the water column with 0=surface.

156 Appendix D

SUPPLEMENTAL MATERIAL FOR CHAPTER 5

Table D-1. Additional Acroproa spp. samples used in analysis

Species Origin N NM NG A. aspera Cocos Lagoon 15 7 6 A. muricata Agat N 10 4 A. acuminata Agat N 1 1 A. palmata Florida* 10 10 * From Baums et al. 2005

157 Table D-2. Microsatellite loci for Acropora cf. pulchra. The fluorescent dye used to label each primer is given (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in two multiplex and three singleplex reactions (Plex). Acr53 and Acr1_4 were run with PET and 6FAM.

Marker Annealing Reference Primer Sequence Repeat Dye Size (bp) Plex Name Temp (°C)

Baums et F: CGCTCTCCTATGTTCGATTG Acr166 AAT PET 126-152 54 I al., 2005 R: TCTACCCGCAATTTTCATCA

Baums et F: TTTCTCAGTGGGTTCCATCA Acr180 AAT VIC 113-122 54 II al., 2005 R: CCTTTCGTTGCTGCAATTTT

Baums et F: TTCTCCACATGCAAACAAACA Acr181 AAT VIC 156-162 54 I al., 2005 R: GCCAGGATAGCGGATAATGA

Baums et F: TCCCACAACTCACACTCTGC Acr182 AAT 6FAM 136-162 54 II al., 2005 R: ACGCGGAAATAGTGATGCTC

Baums et F: TTTGAGCATTTAAGGAGCAACA Acr192 AAT 6FAM 110-128 54 I al., 2005 R: CAGCAGACTCAACAGCAGGA

Tang et F: GGTGAGTTTCTTCGCTGACT PET Acr53 GT 181-187 50 III al.,2010 (NED) R: ATCTAGAATCACGCGCAAGGT

Tang et F: GGCGACCAGACAGGCTCTTA Acr1_60 GT NED 201-223 50 IV al.,2010 R: TTGGCATGAAGTTGAATACGA

Tang et F: GAAGGGAGAGAATCATGTCA 6FAM Acr1_4 GT 177-187 50 V al.,2010 R: TGTGGCAAGTTGTTCGGCTA (NED)

158 Table D-3. Heterozygosity, F-statistics, and deviations from HWE for each locus by site. Significant pHWE after correction for multiple testing given in italics and underlined. Sample size (N), number of alleles (Na), number of effective alleles (Ne), Shannon diversity index (I), observed heterozygosity (Ho), expected heterozygosity (He), and fixation index (F), probability of HWE (pHWE). Site Acr166 Acr181 Acr192 Acr180 Acr182 Acr53 Acr1_4 Acr1_60 Saipan N 19 15 19 19 19 18 19 19 Na 5 3 4 6 7 3 2 4 Ne 3.97 1.51 2.77 4.66 6.22 1.66 1.11 1.76 I 1.47 0.63 1.15 1.66 1.88 0.69 0.21 0.85 Ho 0.32 0.13 0.68 0.89 0.95 0.39 0.11 0.47 He 0.75 0.34 0.64 0.79 0.84 0.40 0.10 0.43 F 0.58 0.61 -0.07 -0.14 -0.13 0.02 -0.06 -0.10 pHWE <0.0001 0.01 0.87 0.04 0.15 >0.99 >0.99 0.63 Tanguisson N 7 7 7 7 7 7 7 7 Na 3 4 4 5 8 3 2 4 Ne 1.81 1.85 3.38 4.08 4.90 2.28 1.51 1.58 I 0.80 0.90 1.30 1.49 1.83 0.90 0.52 0.75 Ho 0.00 0.43 0.71 1.00 0.86 0.71 0.43 0.43 He 0.45 0.46 0.70 0.76 0.80 0.56 0.34 0.37 F 1.00 0.07 -0.01 -0.32 -0.08 -0.27 -0.27 -0.17 pHWE 0.01 0.16 0.81 0.84 0.93 >0.99 >0.99 >0.99 Tumon N 30 25 30 29 31 30 27 30 Na 6 8 5 6 9 3 2 5 Ne 2.85 2.32 2.43 3.63 6.12 2.30 1.43 1.93 I 1.26 1.33 1.08 1.47 1.97 0.91 0.48 0.97 Ho 0.23 0.40 0.50 1.00 0.65 0.47 0.15 0.53 He 0.65 0.57 0.59 0.72 0.84 0.57 0.30 0.48 F 0.64 0.30 0.15 -0.38 0.23 0.18 0.51 -0.11 pHWE <0.0001 0.02 0.02 <0.0001 <0.0001 0.32 0.02 >0.99 Hagatna N 28 28 28 28 28 28 27 28 /Asan Na 5 7 6 6 9 3 3 6 Ne 2.34 2.52 2.57 3.52 3.96 2.09 1.25 2.57 I 1.18 1.25 1.13 1.44 1.73 0.90 0.40 1.22 Ho 0.25 0.39 0.54 0.96 0.61 0.46 0.15 0.61 He 0.57 0.60 0.61 0.72 0.75 0.52 0.20 0.61 F 0.56 0.35 0.12 -0.35 0.19 0.11 0.26 0.01 pHWE <0.0001 0.01 0.56 0.002 0.14 0.57 0.26 0.50 Luminao N 15 15 15 15 15 15 15 15 Na 4 9 5 5 8 4 2 7 Ne 2.42 3.19 3.06 3.54 6.25 2.18 1.30 2.28 I 1.11 1.65 1.31 1.38 1.93 0.92 0.39 1.22 Ho 0.00 0.40 0.53 1.00 1.00 0.40 0.27 0.67 He 0.59 0.69 0.67 0.72 0.84 0.54 0.23 0.56 F 1.00 0.42 0.21 -0.39 -0.19 0.26 -0.15 -0.19 pHWE <0.0001 <0.0001 0.05 0.001 0.63 0.21 >0.99 0.84 Agat N 20 20 20 20 20 20 20 20 Na 5 6 6 4 10 3 2 5 Ne 2.97 2.58 2.90 2.96 5.37 1.94 1.47 2.32 I 1.30 1.21 1.34 1.21 1.96 0.83 0.50 1.15 Ho 0.25 0.50 0.65 1.00 0.85 0.55 0.30 0.70 He 0.66 0.61 0.66 0.66 0.81 0.48 0.32 0.57 F 0.62 0.18 0.01 -0.51 -0.04 -0.14 0.06 -0.23 pHWE 0.0001 0.32 0.82 0.0004 0.48 0.81 >0.99 0.95 Cocos Lagoon N 7 6 7 6 6 7 3 7 Na 2 1 3 3 3 2 2 3 Ne 1.69 1.00 2.39 2.32 2.32 1.96 1.38 2.28 I 0.60 0.00 0.98 0.92 0.92 0.68 0.45 0.90 Ho 0.00 0.00 0.57 1.00 0.83 0.29 0.33 1.00 He 0.41 0.00 0.58 0.57 0.57 0.49 0.28 0.56 F 1.00 N/A 0.02 -0.76 -0.46 0.42 -0.20 -0.78 pHWE 0.02 N/A 0.50 0.09 0.40 0.44 0.73 0.05

159 Table D-4. Allele frequencies by locus and site. Frequencies highlighted using underline and italics denote unusual alleles occurring at high frequencies within the Cocos Lagoon samples. Locus Allele Saipan Tanguisson Tumon Hagatna/Asan Luminao Agat Cocos L. Acr166 121 0.00 0.00 0.02 0.00 0.00 0.00 0.00 126 0.21 0.14 0.08 0.07 0.13 0.05 0.71 127 0.21 0.00 0.47 0.09 0.00 0.25 0.29 130 0.37 0.71 0.35 0.63 0.60 0.50 0.00 133 0.00 0.14 0.05 0.11 0.13 0.08 0.00 136 0.16 0.00 0.03 0.11 0.13 0.13 0.00 152 0.05 0.00 0.00 0.00 0.00 0.00 0.00 Acr181 136 0.07 0.00 0.00 0.04 0.07 0.00 0.00 156 0.80 0.71 0.64 0.57 0.53 0.55 1.00 159 0.00 0.07 0.06 0.25 0.07 0.28 0.00 162 0.13 0.00 0.00 0.00 0.00 0.00 0.00 174 0.00 0.00 0.00 0.00 0.00 0.03 0.00 176 0.00 0.00 0.02 0.00 0.00 0.00 0.00 193 0.00 0.00 0.04 0.02 0.07 0.00 0.00 202 0.00 0.14 0.06 0.05 0.03 0.05 0.00 211 0.00 0.00 0.06 0.04 0.07 0.08 0.00 214 0.00 0.07 0.08 0.04 0.07 0.00 0.00 217 0.00 0.00 0.04 0.00 0.03 0.03 0.00 220 0.00 0.00 0.00 0.00 0.07 0.00 0.00 Acr192 107 0.00 0.00 0.00 0.00 0.00 0.05 0.00 110 0.29 0.43 0.55 0.45 0.47 0.53 0.57 113 0.50 0.21 0.32 0.43 0.30 0.23 0.21 119 0.16 0.21 0.08 0.07 0.07 0.08 0.21 122 0.00 0.00 0.03 0.02 0.00 0.00 0.00 125 0.00 0.00 0.02 0.02 0.07 0.03 0.00 128 0.05 0.14 0.00 0.02 0.10 0.10 0.00 Acr180 104 0.32 0.00 0.19 0.07 0.03 0.08 0.00 107 0.26 0.36 0.43 0.43 0.40 0.48 0.08 110 0.13 0.21 0.12 0.20 0.20 0.18 0.50 113 0.08 0.14 0.19 0.23 0.27 0.28 0.00 116 0.11 0.21 0.05 0.05 0.10 0.00 0.42 122 0.11 0.00 0.02 0.02 0.00 0.00 0.00 131 0.00 0.07 0.00 0.00 0.00 0.00 0.00 Acr182 136 0.21 0.07 0.06 0.14 0.07 0.15 0.00 139 0.18 0.07 0.24 0.13 0.10 0.08 0.50 141 0.13 0.36 0.24 0.45 0.17 0.35 0.42 142 0.18 0.21 0.13 0.05 0.20 0.08 0.00 144 0.00 0.00 0.02 0.05 0.00 0.00 0.00 145 0.00 0.00 0.00 0.00 0.00 0.03 0.00 147 0.05 0.07 0.06 0.02 0.03 0.03 0.00 152 0.00 0.07 0.11 0.02 0.13 0.05 0.00 161 0.13 0.00 0.06 0.09 0.23 0.10 0.00 164 0.11 0.07 0.06 0.05 0.00 0.13 0.08 167 0.00 0.07 0.00 0.00 0.07 0.03 0.00 Acr53 181 0.06 0.07 0.08 0.20 0.03 0.10 0.57 183 0.75 0.50 0.52 0.64 0.57 0.68 0.43 185 0.19 0.43 0.40 0.16 0.37 0.23 0.00 187 0.00 0.00 0.00 0.00 0.03 0.00 0.00 Acr1_4 177 0.05 0.21 0.19 0.09 0.13 0.20 0.17 179 0.95 0.79 0.81 0.89 0.87 0.80 0.83 181 0.00 0.00 0.00 0.02 0.00 0.00 0.00 Acr1_60 201 0.05 0.00 0.08 0.00 0.03 0.05 0.00 203 0.00 0.07 0.02 0.05 0.03 0.00 0.00 211 0.00 0.00 0.00 0.02 0.07 0.10 0.00 213 0.11 0.00 0.13 0.14 0.03 0.08 0.43 215 0.74 0.79 0.70 0.57 0.63 0.63 0.50 217 0.00 0.00 0.00 0.02 0.03 0.00 0.07 223 0.11 0.07 0.07 0.20 0.17 0.15 0.00 229 0.00 0.07 0.00 0.00 0.00 0.00 0.00

160 A

161

PDAM - E

PLOB

APAL

PEVE

APUL

PDAM - P

162

A B

163 Appendix E

NO GENE FLOW ACROSS THE EASTERN PACIFIC BARRIER IN THE REEF-BUILDING CORAL PORITES LOBATA

Published in Molecular Ecology (Baums, IB, Boulay, JN, Polato, NR, Hellberg, M, 2012, 21 (22): 5418-5433)

Abstract

The expanse of deep water between the central Pacific islands and the continental shelf of the Eastern Tropical Pacific is regarded as the world’s most potent marine biogeographic barrier.

During recurrent climatic fluctuations (ENSO, El Niño Southern Oscillation), however, changes in water temperature and the speed and direction of currents become favourable for trans-oceanic dispersal of larvae from central Pacific to marginal eastern Pacific reefs. Here, we investigate the population connectivity of the reef-building coral Porites lobata across the Eastern Pacific Barrier

(EPB). Patterns of recent gene flow in samples (n=1173) from the central Pacific and the Eastern

Tropical Pacific (ETP) were analyzed with 12 microsatellite loci. Results indicated that P. lobata from the ETP are strongly isolated from those in the central Pacific and Hawaii (F’ct=0.509;

P<0.001). However, samples from Clipperton Atoll, an oceanic island on the eastern side of the

EPB, grouped with the central Pacific. Within the central Pacific, Hawaiian populations were strongly isolated from three co-occurring clusters found throughout the remainder of the central

Pacific. No further substructure was evident in the ETP. Changes in oceanographic conditions during ENSO over the past several thousand years thus appear insufficient to support larval deliveries from the central Pacific to the ETP or strong post-settlement selection acts on ETP settlers from the central Pacific. Recovery of P. lobata populations in the frequently disturbed

ETP thus must depend on local larval sources.

164 Introduction

The geographic isolation of shallow-water tropical corals living in the eastern Pacific has stimulated interest in their origin and evolution. As elsewhere in tropical seas, habitats formed by these corals harbour a rich diversity of associated species and contribute to local economies via fisheries and reef-related tourism (Jameson & McManus 1995; Davidson et al.2003). However, coral communities in the eastern Pacific often occur where environmental conditions for reef growth are marginal (Glynn 1984; Guzmán & Cortés 1993; Cortés 1997). These precarious conditions have spurred interest in how these reef corals persist (Richmond 1985; Glynn et al.1991; Glynn & Colgan 1992). Here, we assess one component of this persistence, the extent of gene flow between coral populations in the eastern Pacific and populations further west, to answer long-standing questions for marine biogeographers and provide critical information for the management of coral reefs.

The Eastern Tropical Pacific biogeographic zone (Fig. E-1) stretches from the Sea of

Cortez to the northern Pacific coast of Peru (Cortés 1997) and became isolated from the

Caribbean c. 3 Mya with the closure of the Central American Portal (Duque-Caro 1990; Coates &

Obando 1996). A 5000 to 8000-km deep-water barrier (Dana 1975; Grigg & Hey 1992) now separates Eastern Tropical Pacific biotas from the Indo-West Pacific region. Darwin (Darwin

1880, p. 317) regarded this eastern Pacific barrier (EPB) as ‘impassable’, and Ekman (Ekman

1953) concluded that it is the world’s most potent [soft] marine barrier to larval dispersal.

The present-day eastern Pacific coral fauna has been viewed as a relict derived from pan-

Tethyan, western Atlantic (Caribbean) species formerly connected via the shallow Central

American corridor (McCoy & Heck 1976; Heck & McCoy 1978). After the closure of the Central

American Portal, the eastern Pacific communities were modified by extinctions and evolutionary

165 changes mediated by unfavourable climatic conditions during the late Pliocene and Pleistocene

(Budd 1989, 1994).

In contrast, Dana (1975) [and later Glynn & Wellington (1983) and Cortés (1997)] argued that the eastern Pacific coral reef biota was established more recently (since Pleistocene low sea level stands) by dispersal from the other side of the EPB, chiefly via the North Equatorial

Counter Current (NECC). These conclusions are based on the taxonomic affinities of reef- building corals inhabiting the eastern Pacific and on their potential for dispersal, inferred from a combination of larval durations, rafting capabilities and trans-Pacific current patterns.

Glynn & Ault (2000) defined three main biogeographic provinces in the modern eastern

Pacific based on presence/absence data of reef-building coral species. The Equatorial province, including mainland Ecuador to Costa Rica, the Galapágos Archipelago and Cocos Island, is the most species-rich with 17–26 species, followed by the Northern province (which includes mainland Mexico and the Revillagigedo Islands) with 18–24 species. The Island Group province

(including Malpelo Island and Clipperton Atoll) is relatively species poor (7–10 species) and extends across the EPB to include some islands/atolls in the central Pacific.

In terms of ongoing connectivity, gene flow between central and eastern Pacific populations has been inferred in fish (Rosenblatt & Waples 1986; Lessios & Robertson 2006), sea urchins (Lessios et al. 2003) and seastars (Nishida & Lucas 1988), but little is known about the extent of gene flow across the EPB in reef-building corals. Comparisons of P. lobata between

South Pacific Islands and the Galapágos detected moderate levels of genetic differentiation in the

ITS-1 and ITS-2 regions (Forsman 2003). Restricted dispersal between the central and eastern

Pacific was also evident based on ITS sequence data of Pocillopora spp. (Combosch et al. 2008); however, the taxonomy of the eastern Pacific pocilloporids is in flux (Pinzon & LaJeunesse 2011) complicating the interpretation of these results beyond problems inherent to interpretation of multi-copy markers like ITS.

166 The severe 1982-1983 El Niño Southern Oscillation (ENSO) event (Glynn 1988) forced the recognition that changes in Pacific circulation patterns and transport rates could greatly influence west-to-east dispersal routes. The 1982-83 and 1997-98 ENSOs resulted in extensive mortality of reef-building corals (Glynn 1997, Glynn & Ault 2000). Soon after, however, some

Indo-West Pacific colonists arrived (Lessios et al. 1996, Reid and Kaiser 2001), Because these classic eastern Pacific ENSO events accelerate the rate and latitudinal extent of eastward flow along the North Equatorial Countercurrent (thus halving the transport time across the EPB), they should enhance the eastward transport of larvae across the EPB (Richmond 1990, Glynn et al.

1996).

The escalating magnitude and frequency of ENSO events since the mid-1970s

(Rajagopalan et al. 1997; Trenberth & Hoar 1996) further suggests that the pattern of trans-

Pacific gene flow between corals populations may have undergone recent changes. However this change may be driven by the emergence of a new type of El Niño, the Central Pacific (CP) El

Niño (Kao & Yu 2009; Kug et al. 2009; Lee & McPhaden 2010), in which the warm water anomaly associated with the sea surface warming event is shifted westward to the Central Pacific.

Current models of global warming predict that the ratio of CP- to EP (Eastern Pacific) - ENSO will continue to increase (Yeh et al. 2009). Thus, predictions regarding the effects of ENSO events on trans-EPB dispersal by corals remain unclear.

Porites lobata is an ecosystem engineer that builds the framework of reefs throughout the

Pacific (Glynn et al. 1994). Porites spp. can become large (one giant measured 7 m tall and 41 m in circumference, Brown et al. 2009) and old (approaching 1000 years, Potts et al. 1985), and skeleton cores provide long-term temperature records, akin to tree ring data (Cole et al. 1993). P. lobata produces planktonic larvae via gonochoric broadcast spawning (Glynn et al. 1994). Eggs contain symbiotic algae (Glynn et al. 1994), thus larvae can obtain nutrition during their planktonic lives, thereby extending their dispersal potential (Richmond 1987). Phylogenetic and

167 morphological analyses provide evidence for unrecognized species diversity within the genus

(Forsman 2003; Forsman & Birkeland 2009; Forsman et al. 2010), however only limited information is available on the population genetic structure of Pacific Porites species. Polato et al. (2010) showed that Porites lobata follows an isolation-by-distance pattern along the Hawaiian

Archipelago. Little gene flow connected the Hawaiian Islands and their closest neighbor,

Johnston Atoll, 2500 km away.

Here we test the following hypotheses using multi-locus genotypes generated from a set of polymorphic microsatellite markers: Ho) Samples of P. lobata from the central and eastern

Pacific show evidence of population differentiation due to low levels of ongoing gene flow; Hi)

Isolated Eastern Pacific atolls/islands in the Island Group of Glynn & Ault (2000, Clipperton) are connected to Central Pacific atolls/islands (Line Islands, Johnston, Hawaii) in accordance with coral biogeographic patterns; Hii) Populations within the Eastern Tropical Pacific are subdivided due to limited gene flow between oceanic island and continental shelf populations.

Materials and Methods

Sample collection

Samples were collected from locations in two regions, the central Pacific (CP) and the

Eastern Tropical Pacific (ETP, Table E-1, Fig. E-1). Small fragments (~1 cm2) were broken from colonies using a hammer and chisel and stored in 70% ethanol at -20°C until DNA extraction could be performed. Genomic DNA was extracted using the Qiagen DNeasy 96 blood and tissue kit. Data for Hawaii and Johnston (n = 318) were published in Polato et al. (2010) but were generated in the same laboratory as data presented here.

168 Microsatellite analysis

A total of 12 microsatellite loci were used (Table F-1): eight from Polato et al. (2010) and four additional loci (pl1370, pl1483, pl1868, and pl905) developed for this study (Table F-1).

Briefly, multiplex PCR reactions using fluorescently labeled primers were performed in four multiplex reactions consisting of 2-4 primer pairs each and in one single-plex reaction (Table F-

1). Thermal cycling was performed in an MJ Research PT200 or an Eppendorf Mastercycler

Gradient cycler with an initial denaturation step of 95°C for 5 min followed by 35 cycles of 95°C for 20s; 52°C-56°C (see Table F-1) for 20s; and 72°C for 30s. A final extension of 30 min at

72°C ensured the addition of a terminal adenine (Brownstein et al. 1996). Fragments were analyzed using an ABI 3730 sequencer with an internal size standard (Genescan LIZ-500,

Applied Biosystems, CA). Electropherograms were visualized and allele sizes were called using

GENEMAPPER 4.0 (Applied Biosystems, CA). An allele calling error rate of 0.01 was determined based on repeated runs of 100 samples.

Samples that failed to amplify for more than 2 of 12 loci (n =290 of 1554 or 18%) were excluded from all further analysis. The remaining individuals (n = 1264) thus sometimes contained missing data at one (n = 259) or two (n = 182) loci. In this dataset, There was an overall average failure of 4% (SD 3%) and a per locus failure rate of <10% for each locus in the included samples.

Analysis of multi-locus genotype data

Multi-locus genotype (MLG) data were assigned in GENALEX 6.4 by requiring complete matches at all loci. Considering missing data in the assignment resulted in the same number of unique MLG (n = 1173) as ignoring the missing data. Only unique MLGs were used in

169 subsequent analyses. Potential genotyping errors were detected with GENCLONE 2.0 (Arnaud-

Haond & Belkhir 2007) and spurious allele calls were corrected by re-examining the allele calls in GENEMAPPER 4.0.

Unique MLGs were tested for conformation to Hardy-Weinberg equilibrium (HWE) and

GENEPOP ON THE WEB was used to test genotypic disequilibrium (Raymond & Rousset 1995).

We used the R-package FDRtool to adjust p-values for multiple testing (Strimmer 2008). Large heterozygote deficits are common in marine invertebrates (Addison & Hart 2005) including corals (Baums 2008). We attempted to distinguish among some of the possible causes by estimating null allele frequencies while accounting for inbreeding using INEST (Chybicki &

Burczyk 2009) and distinguished between inbreeding and self-fertilization using RMES (David et al. 2007).

Because observed allelic diversity can be proportional to sample size (Leberg 2002), the program HP-RARE (Kalinowski 2005) was used to compute rarified allelic richness controlled for sample size. Non-parametric Kruskal Wallis ANOVA was used to test for significant differences among diversity measures because of slight deviations from normality.

To inspect for a relationship between Fst and geographic distance, we used Mantel’s test for isolation by distance (IBD) was run in GENODIVE with 999 bootstrap permutations. Principal component analysis (PCA) was performed on a matrix of covariance values calculated from population allele frequencies (GENODIVE). CR04 was excluded from IBD and PCA analysis based on its small sample size (Ng = 1).

Population clustering

STRUCTURE V2.3.3 (Pritchard et al. 2000) was used to estimate the number of population clusters (K). Here, genotypes are assigned to clusters by minimizing linkage disequilibrium and

170 deviations from Hardy-Weinberg within clusters. Location priors did not increase resolution of population clustering in our data, so only runs without location priors are reported. Correlated allele frequencies and admixed populations were assumed based on previous work on this (Polato et al. 2010) and other broadcast spawning corals (Baums et al. 2010; Baums et al. 2006; Foster et al. 2012). Changing these assumptions did not alter the outcome of the clustering (Fig. F-1).

Values of K=1 to 33 were tested by running replicate simulations (≥ 3) with 106 Markov Chain

Monte Carlo (MCMC) repetitions each, and a burn-in of 10,000 iterations on the Bioportal of the

University of Oslo (Kumar et al. 2009). The most likely value for K based on STRUCTURE output was determined by plotting the log probability (L(K)) of the data over multiple runs and comparing that with ΔK (Evanno et al. 2005) as implemented in STRUCTURE HARVESTER (Earl

2009). Results of the three STRUCTURE runs were merged with CLUMPP (Jakobsson & Rosenberg

2007) and results were visualized with DISTRUCT (Rosenberg 2004).

The robustness of results from STRUCTURE results were tested using two other clustering programs with different algorithms and assumptions: INSTRUCT, which models inbreeding (see above) and does not assume Hardy-Weinberg equilibrium (Gao et al. 2007) and GENELAND

(Guillot et al. 2005; Guillot et al. 2008), which can account for null alleles but assumes uncorrelated allele frequencies (Gao et al. 2011), a condition likely violated by admixture here. In

INSTRUCT, the model to infer population structure and inbreeding coefficients was run in three parallel chains with 5 x 105 MCMC repetitions and a burn-in of 105 iterations each. We estimated the number of clusters in GENELAND without specifying any priors. The number of iterations was

106, the thinning interval was 103, and the maximum number of populations K = 20. Convergence of the Markov Chain was checked by inspecting the log-likelihood posterior densities and by comparing the log-likelihood values of multiple independent runs (n = 3).

Analysis of molecular variance (AMOVA) (Excoffier et al. 1992) as implemented in

GENODIVE was used to test hypotheses based on both biogeography and on the clusters identified

171 by STRUCTURE. Statistical analyses were performed at multiple scales (Table E-1): by site, within regions (Hawaii, Central Pacific, Eastern Tropical Pacific), and among regions. Site CR04 was excluded because of its low sample size (Ng = 1).

We were particularly interested in the genetic connection of Clipperton Atoll to the

Central and Eastern Pacific. To inspect Clipperton's clustering, we assigned all multi-locus genotypes to their region of origin (Central Pacific, Hawaii or Eastern Tropical Pacific) while treating the 5 Clipperton samples as unknowns. We did these runs using three replicates with 107 repetitions (106 discarded as burn-in), assuming correlated allele frequency and admixture.

STRUCTURE returned the assignment probability of each of the five Clipperton samples to one of the three regions. Further, multilocus genotypes that had a significantly lower assignment probability for the cluster they were sampled in then for one of the other two clusters were flagged as potential migrants in this analysis.

Results

Multi-locus genotyping

Our genetic analysis of 1264 sampled ramets yielded 1173 unique multi-locus genotypes

(Table E-1). The median combined probability of identity was 8.7 x 10-11. Thus, samples with identical multi-locus genotypes can be confidently ascribed to clonal reproduction. The high proportion of unique MLGs over the total number of samples collected (mean Ng/N = 0.91 +-

0.20, Table E-1) confirms earlier findings of limited asexual reproduction in P. lobata (Polato et al. 2010). Populations ranged from almost entirely clonal (site CR03 in Costa Rica) to entirely sexual (n = 14 sites, Table E-1), although detailed comparisons of clonal structure among sites

172 would be misleading due to variation in sampling effort. Repeated multi-locus genotypes (i.e. exact matches at all loci) were always confined to a single sampling location (i.e. within <2 km).

Tests of LD and deviation from HWE

Only 7.9% of 2,178 tests rejected the null hypothesis of independence among loci after

FDR correction when testing the 33 sites as defined by geography (Table E-1), indicating that loci are largely in linkage equilibrium. 18% of 396 tests showed significant deviation from HWE after

FDR correction (Strimmer 2008) when testing the 33 sites as defined by geography.

Conventional algorithms to estimate the frequencies of null alleles require a priori information on the level of inbreeding (Chybicki & Burczyk 2009; van Oosterhout et al. 2006). In the absence of such information, we used INEST (Chybicki & Burczyk 2009) to estimate the contribution of inbreeding and null alleles to heterozygote deficits (Table F-2, Fig. E-2).

Individual inbreeding values were generally low (Fi mean = 0.03; CI95 = 0.00 – 0.13, Table F-1) and null allele frequencies ranged between 0.08 and 0.24 across loci and populations (Fig. E-2).

Interestingly, loci showed significant differences in the frequency of null alleles across the three regions (Fig. E-2): Each region had at least one locus with lower null allele frequency then other regions. In accordance with the findings of INEST, selfing rates were not significantly different from zero overall (sall = 0.01, 95% CI95 = 0 - 0.26, chi-square = -160.66, DF = 1, p = 1) and selfing rates did not differ among populations (saverage = 0.03, CI95 = 0.01 - 0.06, chi-square =

12.80, DF = 16, p = 0.69).

The high frequency of null alleles at some loci and locations might lead to overestimation of the number of clusters K when running STRUCTURE (STRUCTURE documentation v. 2.3.3). We thus re-ran STRUCTURE using the same settings as above on a dataset excluding the four loci with the highest frequencies of nulls (PL1483, PL1556, PL1968, PL340). The remaining 8 loci had

173 null allele frequencies below 15%. The results (Supplement) are consistent with the findings using all 12 loci with respect to the most likely number of population clusters and patterns of population differentiation and isolation by distance (see below).

Genetic diversity and population structure

Sites in the central Pacific had higher average allelic richness (AR(20)) and private allele richness (AP(20)) compared to Hawaii and the Eastern Tropical Pacific (p < 0.001, DF 2, One-

Way Kruskal Wallis ANOVA) when rarefied to a sample size of three sites per region and 20 genes per site (Fig. E-3). Similarly, the number of effective alleles (AE) averaged across sites was highest in the central Pacific (AEmean = 6.24, 1.83 SD) followed by Hawaii (AEmean = 5.81,

0.76 SD) and the Eastern Tropical Pacific (AEmean = 4.34, 1.38 SD).

The signature of isolation by distance (IBD) was small when the entire dataset was considered (Fig. E-4A, r2=0.07; P=0.001). Regionally, IBD was non-significant in the western portion of the central Pacific (Fig. E-4B, r2=0.00; P>0.1), moderate in the eastern portion of the central Pacific (Fig. E-4C, r2=0.35; P<0.001) and strong in the ETP (Fig. E-4D, r2=0.43; P<

0.01), but only when the most distant site (Clipperton) was included (Fig. E-4D, r2=0.07; P< 0.01 without Cl01). Polato et al. (2010) reported r2=0.32 along the Hawaiian Island Chain (2500 km; comparable to the ETP in geographic scale (Fig. E-4)).

Principal components analysis separated the Hawaiian Islands and the ETP from the reminder of the Pacific (Fig. E-5). Interestingly, Clipperton Atoll grouped with the central

Pacific, even though geographically it lies east of the Eastern Pacific Barrier. Johnston Atoll occupied a position equidistant from the center of the other clusters (Hawaii, central Pacific and

ETP).

174

Plots of ΔK (Evanno et al. 2005) and LnP(K) from STRUCTURE indicate that five is the most likely number of population clusters present in the full dataset (Fig. E-6 A,B). At K = 2,

Hawaii formed a separate cluster from the reminder of the samples. At K = 3, the ETP and the northern Line Islands separated from the reminder of the CP and HI. At K = 4, a cluster with few members appeared in the Western CP. At K = 5, samples from the northern Line Islands that previously grouped with the ETP, now formed a separate cluster confined to the west of the

Eastern Pacific Barrier (Fig. E-1). The ETP appeared homogenous at this level of analysis (Fig.

E-1), with the exception of samples from Clipperton, five samples from the Galapágos and one sample from Costa Rica. Consistent with the PCA analysis (Fig. E-5), Johnston appeared admixed between Hawaii and the central Pacific.

INSTRUCT results which consider potential inbreeding agreed with those from

STRUCTURE that 5 was the most likely value of K (not shown). GENELAND which considers null alleles estimated the number of clusters as 7 under a model without prior information on spatial location or population membership of samples (not shown). One of the additional clusters in

Geneland separated the Marquesas (MQ1 and MQ2) from the Line Islands (also observable in

STRUCTURE runs on central Pacific samples only at K = 9, not shown). The second additional cluster occurred at low frequency in the Line Islands, Phoenix, Marshalls, Fiji, Samoa and

Moorea.

FST values were significant in 58% of all pairwise comparisons (after FDR correction), with larger values observed between more distant sites (Table F-3). FST ranged from 0 to 0.27 (0 to 0.47 with Meirman’s F’ST; -0.14 – 0.76 with Jost’s D), with maximum values detected between the Hawaiian sites and the reminder of the central Pacific (Online supplements). There was significant among population differentiation based on an AMOVA considering 32 populations

(Fst = 0.157, SE 0.018, p<0.001, F’st = 0.459, Table E-3A). Adding the level of “region” to the

AMOVA resulted in strong differentiation among regions (Fct = 0.145, 0.02 SE, p <0.001, F’ct=

175

0.464, Table E-3B) and a lower amount of differentiation among populations within regions (Fsc

= 0.063, 0.00 SE, p<0.001, F’sc =0.197, Table E-3B) consistent with PCA and Structure analysis.

In STRUCTURE runs where all but the Clipperton samples were assigned a priori to their region of origin (HI, CP or ETP, Fig. E-7), the Clipperton samples assigned with a higher probability to the Central (average assignment probability CP = 0.74 +/- 0.23 SD) than to the

Eastern Tropical Pacific (average assignment probability ETP = 0.24 +/- 0.23 SD). As expected, none of the Clipperton samples assigned to Hawaii (average assignment probability HI = 0.02 +/-

0.00 SD). Considering the a priori assigned samples (all but the samples from Clipperton),

STRUCTURE identified 7 genotypes from the Central Pacific (1 from PH01, 4 from JO01, 2 from

LN04, Fig. E-7) and one genotype from the eastern Pacific (CR03, Fig. E-7) as first generation migrants with high probability (> 0.9, p < 0.001). All of JO01 migrants had likely ancestry in HI whereas the most likely origin of the PH01 and the 2 LN04 migrants was the ETP (Fig. E-7).

None of the genotypes from the Galapágos assigned to the CP in this analysis (Fig. E-7). The origin of the migrant CR03 sample appeared to be the CP (Fig. E-7).

Discussion

Our genetic data corroborate previous biogeographic hypotheses on the Eastern Pacific

Barrier in the broadest sense: most populations of Porites lobata are presently isolated from those in the central Pacific. However, the data also suggest that Clipperton Atoll is genetically similar to populations in the central Pacific despite residing to the east of the Eastern Pacific Barrier, and that gene flow between insular and continental populations within the ETP is quite high.

176 Is the EPB a barrier to corals?

The depauperate coral reef fauna in the Eastern Tropical Pacific experiences frequent large-scale disturbances in the form of ENSO warming events (Glynn & Colgan 1992; McPhaden

1999; Wyrtki 1975). ENSO can lead to widespread bleaching and mortality of corals (Glynn

1984; Glynn & Deweerdt 1991; Jimenez & Cortés 2001). However, recovery of reefs from ENSO events can be rapid, at least in some locations (Glynn & Colgan 1992; Glynn et al. 2009). Were recovering reefs reseeded via long-distance dispersal or were recruits derived from local sources?

The vibrant coral reefs of the Central Pacific (Edmunds et al. 2010; Sandin et al. 2008; Veron

1995) support large populations of the major Eastern Pacific reef-builders, including Porites lobata, and thus might be a source for recruits. However, the broad stretch of deep water between the Central and Eastern Pacific (Darwin 1880) and the mostly westward current flow of the North

Equatorial Current (NEC) (reviewed in Kessler 2006; Wyrtki et al. 1981) are formidable barriers to dispersal. During classic (EP) ENSO years, eastward flow in the North Equatorial Counter

Current (NECC) is warmer and faster (reviewed in Bonjean & Lagerloef 2002; Kessler 2006), providing a potential bridge to the Eastern Pacific (Richmond 1990).

Our results suggest that the Central and most Eastern Pacific locations are presently isolated (Figs. E-1, D-5, F-1, F-2, F-4, Table E-3). The one exceptional location is Clipperton

Atoll. Situated to the east of the EPB, it nonetheless groups genetically with the Central Pacific

(Table E-3, Figs E-1, E-5, E-7). The most likely route for dispersing larvae or rafting adult corals from the Central Pacific (specifically the Line Islands) to Clipperton Atoll is the North Equatorial

Counter Current (NECC), which skirts Clipperton Atoll even in non-ENSO years (Kessler 2006).

Clipperton samples had a 0.74 +/- 0.23 SD assignment probability to the Central Pacific compared to a 0.24 +/- 0.23 SD probability of belonging to the Tropical Eastern Pacific.

Increased sample sizes from Clipperton would help confirm these findings, although STRUCTURE

177 should deliver robust assignments for the five samples in hand given the comprehensive sampling of potential source populations (Falush et al. 2003).

The genetic clustering of P. lobata samples from Clipperton with the Central Pacific concurs with biogeographical clustering based on the distribution of coral species (Glynn and

Ault (2000). Their island group includes Clipperton Atoll and the central Pacific islands of

Hawaii, Johnston and Fanning. While P. lobata samples from Clipperton group genetically with the Central Pacific islands to which Fanning belongs (Figs. E-1, E-5, E-7), Hawaiian P. lobata are differentiated from both the central Pacific and the ETP (Figs. E-1, E-5). Our sampling does not allow for complete overlap with the Glynn and Ault predictions: we were unable to secure samples from Malpelo (another ETP island biogeographically grouped with the CP), and P. lobata does not occur in their northern province (where it is replaced by P. evermanni, Boulay et al. 2014). However, the geographic restriction of P. lobata to the southern province indicates a lack of successful recruitment to the northern province, supporting Glynn and Ault’s biogeographic clusters.

In model runs where each genotype (with the exception of Clipperton, see above) was assigned a priori to originate from the location it was sampled (Fig. E-7), seven genotypes sampled in the Central Pacific and one genotype sampled in the eastern Pacific (CR03) were identified as first generation migrants with high probability (> 0.9, p < 0.001) confirming very low levels of migration among regions consistent with the high among-region Fst value (0.145 +/-

0.02 SE; F’st = 0.465, Table E-3 B). Preliminary analysis showed that the flagged CR03 genotype harbored an unusual ITS – sequence (Forsman et al. 2009) indicating possible introgression from

Porites evermanni. Thus, introgression within the EP and not migration from the CP might be the cause for the unusual genetic composition of this individual. Only 2 of the CP migrants were assigned to the EP and ITS sequences of CP migrants grouped within the P. lobata clade identified by (Forsman et al. 2009). Future work will explore the extent of introgression between

178 P. lobata and P. evermanni in the ETP and elsewhere with the expectation that introgression rates between species will vary across their geographic range (Fukami et al. 2004; Ladner & Palumbi

2012).

Similar to findings for P. lobata, populations in the central/western Pacific and the

Eastern Pacific were differentiated in Conus snails (Duda & Lessios 2009), soldierfish (Craig et al. 2007), and lobsters (Chow et al. 2011). Limited gene flow was also reported between the

Galapágos and South Pacific Island populations of P. lobata (Forsman 2003) and between Central and Eastern Pacific populations of Pocillopora damicornis (Combosch et al. 2008), in contrast to ongoing gene flow between urchin populations in Clipperton/Cocos Island and the Central Pacific

(Lessios et al. 1998). Further, of 18 fish species found on either side of the EBP, only 2 showed significant divergence between the Central and Eastern Pacific (Lessios & Robertson 2006).

The strong divergence among Hawaii, the Central Pacific, and the Eastern Pacific, as well as the occurrence of a well-supported but rare cluster within the western Central Pacific, raise questions about the level of taxonomic resolution addressed here. While we cannot exclude the possibility that each of the highly supported clusters constitutes a different species (Ladner &

Palumbi 2012), we think this is unlikely for several reasons. First, markers designed for P. lobata often failed to amplify when used on other Porites species. We determined this by applying our markers to samples identified as other Porites species (n = 37 ; P. latistella, P. compressa, P. duerdeni, P. lutea, P. panamensis) by independent expert morphological analysis (Z. Forsman) and ITS sequencing (Forsman et al. 2009). In fact, we initially discovered that Porites samples from the northern EP are P. evermanni and not P. lobata based on patterns of amplification failure and fixed alleles with non-overlapping size range at three loci in P. evermanni that are otherwise polymorphic in P. lobata. We have since substantiated this finding by describing habitat and ecological differences between the species (Boulay et al. 2014). Based on this finding, we conducted analyses on patterns of amplification failure across loci and found no further signal,

179 i.e. knowing that one locus failed did not help to predict amplification failure at any of the other

11 loci (see also Fig. E-2). Second, phylogenetic analysis of sequences from six nuclear markers for samples of P. lobata from Hawaii, the type locality, and other closely related Porites agree with our microsatellite and field identifications (Hellberg et al. in prep). Furthermore, preliminary analysis of ITS sequences of representative samples from clusters identified here fall within the clade previously described as Porites lobata by Forsman (2009). Regardless of the level of taxonomic differentiation, the conclusion of a general lack of gene flow across the EPB and isolation of Hawaii holds.

Patterns of gene flow within the Central Pacific

Within the central Pacific, Hawaiian populations were strongly isolated from the remainder of the region, including their nearest neighbor Johnston Atoll, as in Polato et al.

(2010). The near-linear arrangement of the Line Islands, the Marquesas, Moorea, and Johnston

Atoll (Fig. E-1) lends itself to tests for isolation by distance and indeed the correlation between genetic and geographic distance was moderately strong in this region (Fig. E-4 C).

General patterns of population genetic differentiation among reef dwellers in this region of the Central Pacific are yet to emerge. Restricted gene flow has been reported for corals

(Magalon et al. 2005), oysters (Arnaud-Haond et al. 2004), and some reef fish (Gaither et al.

2010) (Planes & Fauvelot 2002), with turbinid gastropods revealing endemic genetic clades in each archipelago (Meyer et al. 2005). In contrast, some other reef fish, even congeners of those mentioned above (Eble et al. 2011; Gaither et al. 2010), show little structure across the central

Indo-Pacific, and no population differentiation was observed between populations of the urchin

Diadema savignyi from Moorea and Kiribati (Lessios et al. 2001).

180 Patterns of gene flow within the Eastern Tropical Pacific

With the exception of the differentiation of Clipperton Atoll from in the reminder of the

ETP, population differentiation was weak in this region (Fig. E-1). Data on population genetic structure of corals in the ETP is sparse and complicated by difficult morphological species identification (Pinzon & LaJeunesse 2010). Genetically, three types (Type I – III) of Pocillopora spp. can be distinguished in the ETP (Toonen unpubl. data, Pinzon & LaJeunesse 2010). P. damicornis Type I, the only type with sufficient samples sizes across the region to allow for population level analysis, shows panmixia in the ETP (including the Mexican mainland,

Revillagigedo Island, Clipperton Atoll, the Galapágos, and Panama) at seven microsatellite loci

(Pinzon & LaJeunesse 2010). In contrast, Combosh and Vollmer (2011) found five distinct but co-occurring genetic clusters in varying proportions along the Panama coast. It is not clear whether those clusters correspond to any of the types identified in the Pinzon and LaJeunesse study. Using six allozyme loci (Chávez-Romo et al. 2009) found three genetically distinct clusters along the Mexican coast, but again it is not clear whether those samples represented just one or multiple types described by Pinzon and LaJeunesse (2010).

Several other marine organisms show little population genetic structure within the ETP.

Some that evince population genetic differences between the central/western Pacific and the

Eastern Pacific show no further structure within ETP samples (Duda & Lessios 2009; Chow et al.

2011; Craig et al. 2007). Similarly, no population structure was found among ETP sites in rocky intertidal snails and sea urchins (Hurtado et al. 2007; McCartney et al. 2000). In contrast, significant population structure often occurs between the Gulf of California and populations to the south (Riginos & Nachman 2001; Hurtado et al. 2007; Saarman et al. 2010).

181 Is the Eastern Tropical Pacific marginal?

Many reef-building corals occur over large geographic ranges and experience suboptimal and variable conditions at the margins of their distributions. Such marginal populations can provide insights into how corals might respond to climate change (Goodkin et al. 2011; Guinotte et al. 2003; Hennige et al. 2010; Lirman & Manzello 2009). For example, coral communities in the Eastern Tropical Pacific (ETP) already experience seasonal cold upwelling, El Niño Southern

Oscillation warm events, and reduced aragonite saturation states (Fong & Glynn 2000; Glynn &

Colgan 1992). In fact, the Eastern Pacific experiences some of the most severe stress exposures of any coral province worldwide (Maina et al. 2011). The conditions in edge habitats have spurred interest in how coral populations persist there, how they will react to a rapidly changing climate, and what role they play in the evolution of coral species.

Physical isolation is expected to increase and population size is expected to decrease with increasing distance from the geographic center of a species’ range often accompanied by losses in allelic diversity due to lack of gene flow and increased levels of inbreeding (reviewed in Eckert et al. 2008; Sagarin & Gaines 2002). For P. lobata, sites in the central Pacific had almost twice as much allelic richness then sites in Hawaii and the ETP (Fig. E-3) and a higher number of effective alleles, and inbreeding was generally low (Table F-2). Because corals can reproduce locally by asexual means (Baums et al. 2006; Foster et al. 2007; Highsmith 1982), reduced gene flow into marginal populations can result in increased clonality. Because sampling effort was not constant across site, genotypic diversity cannot be directly compared across sites. Generally, we find little evidence of asexual reproduction in P. lobata across its range, however, congruent with the above predictions, two of the eastern Pacific sites (CR03 and EC01) showed low genotypic diversity (Table E-1, Polato et al. 2010).

182 Microsatellite heterozygosities generally decrease with increasing distance from the centers of coral diversity in the Pacific and Atlantic (Baums 2008). Examples of low heterozygosity in marginal locales include P. damicornis from Lord Howe Island (Miller & Ayre

2004), Seriatopora hystrix from an isolated site in Australia (Scott Reef; Underwood et al. 2007), and coral species from Japan (Adjeroud & Tsuchiya 1999; Ayre & Hughes 2004). In taxa that are connected via gene flow across the EPB, including 18 species of reef fish (Lessios & Robertson

2006) and a sea urchins (Lessios et al. 1998), no such reduction in genetic diversity across the

EBP is apparent.

Conclusions

The Eastern Pacific Barrier isolates populations of the important ecosystem engineer,

Porites lobata, in the Central and Eastern Tropical Pacific. The possible exception to this generality comes from Clipperton Atoll, which we found to be most genetically similar to populations in the Line Islands and the Marquesas. Dispersal from the Central Pacific to

Clipperton Atoll likely occurs via the North Equatorial Counter Current (NECC), which reaches

Clipperton even during non-ENSO years when the NECC is relatively weak. We had hypothesized that recurrent strengthening of the NECC during increasingly intense ENSO events may result in gene flow between the Central and Eastern Tropical Pacific, however very little exchange is evident in the data, nor did we find support for genetic differentiation between the oceanic island and continental shelf in accordance with biogeographic patterns (Glynn & Ault

2000).

Climate change is threatening coral reefs world-wide (Hoegh-Guldberg et al. 2007;

Hughes et al. 2003). Coral populations already growing in marginal habitats (Maina et al. 2011) can provide insights into how corals might respond to climate change (Cooper et al. 2012;

183 Goodkin et al. 2011; Guinotte et al. 2003; Hennige et al. 2010; Lirman & Manzello 2009).

Mounting evidence indicates that marginal coral populations harbor less neutral genetic diversity then more central populations (Fig. E-3; Adjeroud & Tsuchiya 1999; Ayre & Hughes 2004;

Baums 2008; Miller & Ayre 2004; Underwood et al. 2007), but little is known about the distribution of functional genetic diversity across the range of coral species. This should be a major focus of future research.

184 Acknowledgements

We thank M. Durante and D. Almeida for help in the lab. Field assistance and/or samples were provided by A Baker, P Barber, D Barshis, B Bowen, L Castillo, A Chiriboga, G

Concepcion, J Cortés, M Craig, J Eble, J-F Flot, Z Forsman, E Franklin, M Gaither, S Godwin, B

Greene, P LaFemina, T Lison de Loma, S Luna, J Maragos, T McDole, J Nivia-Ruiz, D Obura, S

Planes, L Rocha, R Rotjan, J Salerno, S Sandin, A Simoes Correa , C Starger, M Stat, M

Timmers, R Toonen, M Vera, J-S White and J Williams. Z. Forseman generously shared samples of other Porites species with us. We thank the governments of French Polynesia, Kiribati,

Ecuador, Costa Rica, Panama, and the US for facilitating research permits. This research was supported by NSF grant OCE- 0550294 to IBB and MEH.

Data Accessibility

Sampling locations, Structure input files and microsatellite data: DRYAD entry doi:10.5061/dryad.7gp1f.

Author contributions

IB performed the statistical analysis and wrote the paper. JB and NP developed methods and performed lab analysis. All authors collected samples, read, edited and approved the manuscript. IB and ME obtained funding and designed the study.

185 References

Addison, J.A. & M.W. Hart. 2005. Spawning, copulation and inbreeding coefficients in marine invertebrates. Biol. Lett. 1: 450-453. Adjeroud, M. & M. Tsuchiya. 1999. Genetic variation and clonal structure in the scleractinian coral Pocillopora damicornis in the Ryukyu Archipelago, southern Japan. Mar. Biol. 134: 753-760. Arnaud-Haond, S., V. Vonau, F. Bonhomme, P. Boudry, F. Blanc, J. Prou, T. Seaman & E. Goyard. 2004. Spatio-temporal variation in the genetic composition of wild populations of pearl oyster (Pinctada margaritifera cumingii) in French Polynesia following 10 years of juvenile translocation. Mol. Ecol. 13: 2001-2007. Arnaud-Haond, S. & K. Belkhir. 2007. GENCLONE: a computer program to analyse genotypic data, test for clonality and describe spatial clonal organization. Mol. Ecol. Notes 7: 15-17. Ayre, D.J. & T.P. Hughes. 2004. Climate change, genotypic diversity and gene flow in reef- building corals. Ecol. Lett. 7: 273-278. Baums, I.B., M.W. Miller & M.E. Hellberg. 2006. Geographic variation in clonal structure in a reef building Caribbean coral, Acropora palmata. Ecol. Monogr. 76: 503-519. Baums, I.B. 2008. A restoration genetics guide for coral reef conservation. Mol. Ecol. 17: 2796- 2811. Baums, I.B., M.E. Johnson, M.K. Devlin-Durante & M.W. Miller. 2010. Host population genetic structure and zooxanthellae diversity of two reef-building coral species along the Florida Reef Tract and wider Caribbean. Coral Reefs 29: 835-842. Bonjean, F. & G.S.E. Lagerloef. 2002. Diagnostic model and analysis of the surface currents in the tropical Pacific Ocean. J. Phys. Oceanogr. 32: 2938-2954. Brown, D.P., L. Basch, D. Barshis, Z. Forsman, D. Fenner & J. Goldberg. 2009. American Samoa's island of giants: massive Porites colonies at Ta'u island. Coral Reefs. 28: 735- 735. Brownstein, M.J., J.D. Carpten & J.R. Smith. 1996. Modulation of non-templated nucleotide addition by tag DNA polymerase: Primer modifications that facilitate genotyping. Biotechniques 20: 1004-1010. Chávez-Romo, H., F. Correa-Sandoval, D. Paz-García, H. Reyes-Bonilla, R. López-Pérez, P. Medina-Rosas & M. Hernández-Cortés. 2009. Genetic structure of a scleractinia coral, Pocillopora damicornis, in the Mexican Pacific. Proc. 11th Int. Coral Reef Symp. 429- 433. Chow, S., A. Jeffs, Y. Miyake, K. Konishi, M. Okazaki, N. Suzuki, M.F. Abdullah, H. Imai, T. Wakabayasi & M. Sakai. 2011. Genetic Isolation between the Western and Eastern Pacific Populations of Pronghorn Spiny Lobster Panulirus penicillatus. PLoS One 6: e29280. Chybicki, I.J. & J. Burczyk. 2009. Simultaneous estimation of null alleles and inbreeding coefficients. J. Hered. 100: 106-113.

186 Cole, J.E., R.G. Fairbanks & G.T. Shen. 1993. Recent variability in the Southern Oscillation - isotopic results from a Tarawa Atoll coral. Science 260: 1790-1793. Combosch, D.J., H.M. Guzmán, H. Schuhmacher & S.V. Vollmer. 2008. Interspecific hybridization and restricted trans-Pacific gene flow in the Tropical Eastern Pacific Pocillopora. Mol. Ecol. 17: 1304-1312. Combosch, D.J. & S.V. Vollmer. 2011. Population genetics of an ecosystem-defining reef coral Pocillopora damicornis in the Tropical Eastern Pacific. PLoS One 6: e21200. Cooper, T.F., R.A. O’leary & J.M. Lough. 2012. Growth of western Australian corals in the Anthropocene. Science 335: 593-596. Cortés, J. 1997. Biology and geology of Eastern Pacific coral reefs. Coral Reefs 16: S39-S46. Craig, M.T., J.A. Eble, B.W. Bowen & D.R. Robertson. 2007. High genetic connectivity across the Indian and Pacific Oceans in the reef fish Myripristis berndti (Holocentridae). Mar. Ecol. Prog. Ser. 334: 245-254. Dana, T.F. 1975. Development of contemporary eastern Pacific coral reefs. Mar. Biol. 33: 355- 374. Darwin, C. 1880. The Origin of Species by Means of Natural Selection or the Preservation of Favored Races in the Struggle for Life. John Murray, London. David, P., B.I.T. Pujol, F. Viard, V. Castella & J. Goudet. 2007. Reliable selfing rate estimates from imperfect population genetic data. Mol. Ecol. 16: 2474-2487. Duda, T.F. & H.A. Lessios. 2009. Connectivity of populations within and between major biogeographic regions of the tropical Pacific in Conus ebraeus, a widespread marine gastropod. Coral Reefs 28: 651-659. Duque-Caro, H. 1990. The Choco block in the NW corner of South America: structural tectonostratigraphic and paleogeographic implications. J. S. Am. Earth Sci. 3: 71-84. Eble, J.A., L.A. Rocha, M.T. Craig & B.W. Bowen. 2011. Not all larvae stay close to home: Insights into marine population sonnectivity with a focus on the brown surgeonfish (Acanthurus nigrofuscus). J. Mar. Biol. 2011: 518512-518516. Eckert, C.G., K.E. Samis & S.C. Lougheed. 2008. Genetic variation across species' geographical ranges: the central-marginal hypothesis and beyond. Mol. Ecol. 17: 1170-1188. Edmunds, P.J., J.J. Leichter & M. Adjeroud. 2010. Landscape-scale variation in coral recruitment in Moorea, French Polynesia. Mar. Ecol. Prog. Ser. 414: 75-89. Ekman, S. 1953. Zoogeography of the Sea. Sidgwick and Jackson Ltd, London. Evanno, G., S. Regnaut & J. Goudet. 2005. Detecting the number of clusters of individuals using the software structure: a simulation study. Mol. Ecol. 14: 2611-2620. Excoffier, L., P.E. Smouse & J.M. Quattro. 1992. Analysis of Molecular Variance Inferred from Metric Distances among DNA Haplotypes - Application to Human Mitochondrial-DNA Restriction Data. Genetics 131: 479-491. Falush, D., M. Stephens & J.K. Pritchard. 2003. Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies. Genetics 164: 1567-1587.

187 Fong, P. & P.W. Glynn. 2000. A regional model to predict coral population dynamics in response to El Niño-Southern Oscillation. Ecol. Appl. 10: 842-854. Forsman, Z.H., D.J. Barshis, C.L. Hunter & R.J. Toonen. 2009. Shape-shifting corals: Molecular markers show morphology is evolutionarily plastic in Porites. Evol. Biol. 9: 45. Forsman, Z.H. & C. Birkeland. 2009. Porites randalli: a new coral species (Scleractinia, Poritidae) from American Samoa. Zootaxa 2244: 51-68. Forsman, Z.H., J.A. Martinez, J.E. Maragos & R.J. Toonen. 2010. Resurrection of Porites hawaiiensis Vaughan, 1907; a Hawaiian coral obscured by small size, cryptic habitat, and confused taxonomy. Zootaxa 67-68. Foster, N.L., I.B. Baums & P.J. Mumby. 2007. Sexual verses asexual reproduction in an ecosystem engineer: the massive coral Montastraea annularis. J. Anim. Ecol. 76: 384- 391. Foster, N.L., C.B. Paris, J.T. Kool, I.B. Baums, J.R. Stevens, J.A. Sanchez, C. Bastidas, C. Agudelo, P. Bush, O. Day, R. Ferrari, P. Gonzalez, S. Gore, R. Guppy, M. Mccartney, C. Mccoy, J.M. Mendes, A. Srinivasan, S.C.C. Steiner, M.J.A. Vermeij, E. Weil & P.J. Mumby. 2012. Connectivity of Caribbean coral populations: complementary insights from empirical and modelled gene flow. Mol. Ecol. 21.5:1143-1157. Gaither, M.R., R.J. Toonen, D.R. Robertson, S. Planes & B.W. Bowen. 2010. Genetic evaluation of marine biogeographical barriers: perspectives from two widespread Indo-Pacific snappers (Lutjanus kasmira and Lutjanus fulvus). J. Biogeogr. 37: 133-147. Gao, H., S. Williamson & C.D. Bustamante. 2007. A Markov chain Monte Carlo approach for joint inference of population structure and inbreeding rates from multilocus genotype data. Genetics 176: 1635. Gao, H., K. Bryc & C.D. Bustamante. 2011. On identifying the optimal number of population clusters via the Deviance Information Criterion. PLoS One 6: e21014. Glynn, P.W. & G.M. Wellington. 1983. Corals and Coral Reefs of the Galápagos Islands. University of California Press, Berkeley, CA. Glynn, P.W. 1984. Widespread coral mortality and the 1982-83 El Niño warming event. Environ. Conserv. 11: 133-146. Glynn, P.W. & W.H. Deweerdt. 1991. Elimination of 2 reef-building hydrocorals following the 1982-83 El-Niño warming event. Science 253: 69-71. Glynn, P.W., N.J. Gassman, C.M. Eakin, J. Cortés, D.B. Smith & H.M. Guzmán. 1991. Reef Coral Reproduction in the Eastern Pacific - Costa-Rica, Panama, and Galapágos-Islands (Ecuador) .1. Pocilloporidae. Mar. Biol. 109: 355-368. Glynn, P.W. & M.W. Colgan. 1992. Sporadic disturbances in fluctuating coral-reef environments - El Niño and coral-reef development in the eastern Pacific. Am. Zool. 32: 707-718. Glynn, P.W., S.B. Colley, C.M. Eakin, D.B. Smith, J. Cortés, N.J. Gassman, H.M. Guzmán, J.B. Rosario & J.S. Feingold. 1994. Reef coral reproduction in the eastern Pacific: Costa Rica, Panamá, and Galápagos Islands (Ecuador). II. Poritidae. Mar. Biol. 118: 191-208. Glynn, P.W. & J.S. Ault. 2000. A biogeographic analysis and review of the far eastern Pacific coral reef region. Coral Reefs 19: 1-23.

188 Glynn, P.W., B. Riegl, A.M.S. Correa & I.B. Baums. 2009. Rapid recovery of a coral reef at Darwin Island, Galapágos Islands. Galapágos Research 66: 6-13. Goodkin, N.F.G.N.F., A.D. Switzer, D. Mccorry, L. Devantier, J.D. True, K.A. Hughen, N. Angeline & T.T. Yang. 2011. Coral communities of Hong Kong: long-lived corals in a marginal reef environment. Mar. Ecol. Prog. Ser. 426: 185-196. Grigg, R.W. & R. Hey. 1992. Paleoceanography of the Tropical Eastern Pacific-Ocean. Science 255: 172-178. Guillot, G., F. Mortier & A. Estoup. 2005. GENELAND: a computer package for landscape genetics. Mol. Ecol. Notes 5: 712-715. Guillot, G., F. Santos & A. Estoup. 2008. Analysing georeferenced population genetics data with GENELAND: a new algorithm to deal with null alleles and a friendly graphical user interface. Bioinformatics 24: 1406-1407. Guinotte, J.M., R.W. Buddemeier & J.A. Kleypas. 2003. Future coral reef habitat marginality: temporal and spatial effects of climate change in the Pacific basin. Coral Reefs 22: 551- 558. Guzmán, H.M. & J. Cortés. 1993. Coral reefs of the tropical eastern pacific - review and perspectives. Rev. Biol. Trop. 41: 535-557. Hedrick, P.W. & C. Goodnight. 2005. A standardized genetic differentiation measure. Evolution 59: 1633-1638. Hennige, S.J., D.J. Smith, S.J. Walsh, M.P. Mcginley, M.E. Warner & D.J. Suggett. 2010. Acclimation and adaptation of scleractinian coral communities along environmental gradients within an Indonesian reef system. J. Exp. Mar. Biol. Ecol. 391: 143-152. Highsmith, R.C. 1982. Reproduction by fragmentation in corals. Mar. Ecol. Prog. Ser. 7: 207- 226. Hoegh-Guldberg, O., P.J. Mumby, A.J. Hooten, R.S. Steneck, P. Greenfield, E. Gomez, C.D. Harvell, P.F. Sale, A.J. Edwards & K. Caldeira. 2007. Coral reefs under rapid climate change and ocean acidification. Science 318: 1737 - 1742. Hughes, T.P., A.H. Baird, D.R. Bellwood, M. Card, S.R. Connolly, C. Folke, R. Grosberg, O. Hoegh-Guldberg, J.B.C. Jackson & J. Kleypas. 2003. Climate change, human impacts, and the resilience of coral reefs. Science 301: 929 - 933. Hurtado, L.A., M. Frey, P. Gaube, E. Pfeiler & T.A. Markow. 2007. Geographical subdivision, demographic history and gene flow in two sympatric species of intertidal snails, Nerita scabricosta and Nerita funiculata, from the Tropical Eastern Pacific. Mar. Biol. 151: 1863-1873. Jakobsson, M. & N.A. Rosenberg. 2007. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801. Jameson, S. & J. Mcmanus. 1995. State of the Reefs: Regional and Global Perspectives. International Coral Reef Initiative. US Department of State, Washington, DC. Jiménez, C. & J. Cortés. 2001. Effects of the 1991-92 El Niño on scleractinian corals of the Costa Rican central Pacific coast. Rev. Biol. Trop. 49: 239-250.

189

Jost, L. 2008. GST and its relatives do not measure differentiation. Mol. Ecol. 17: 4015-4026. Kalinowski, S.T. 2005. HP-RARE 1. 0: a computer program for performing rarefaction on measures of allelic richness. Mol. Ecol. Notes 5: 187-189. Kao, H.Y. & J.Y. Yu. 2009. Contrasting Eastern-Pacific and Central-Pacific types of ENSO. J. Climate 22: 615-632. Kessler, W.S. 2006. The circulation of the eastern tropical Pacific: A review. Prog. Oceanogr. 69: 181-217. Kug, J.S., F.F. Jin & S.I. An. 2009. Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. J. Climate 22: 1499-1515. Kumar, S., A. Skjaeveland, R.J.S. Orr, P. Enger, T. Ruden, B.H. Mevik, F. Burki, A. Botnen & K. Shalchian-Tabrizi. 2009. AIR: A batch-oriented web program package for construction of supermatrices ready for phylogenomic analyses. BMC Bioinformatics10: 357 Leberg, P.L. 2002. Estimating allelic richness: effects of sample size and bottlenecks. Mol. Ecol. 11: 2445-2449. Lee, T. & M.J. Mcphaden. 2010. Increasing intensity of El Niño in the central-equatorial Pacific. Geophys. Res. Lett. 37: 14. Lessios, H.A., B.D. Kessing & D.R. Robertson. 1998. Massive gene flow across the world's most potent marine biogeographic barrier. Proc. R. Soc. B.-Biol. Sci. 265: 583-588. Lessios, H.A., B.D. Kessing & J.S. Pearse. 2001. Population structure and speciation in tropical seas: Global phylogeography of the sea urchin Diadema. Evolution. 55: 955-975. Lessios, H.A., J. Kane & D.R. Robertson. 2003. Phylogeography of the pantropical sea urchin Tripneustes: Contrasting patterns of population structure between oceans. Evolution. 57: 2026-2036. Lessios, H.A. & D.R. Robertson. 2006. Crossing the impassable: genetic connections in 20 reef fishes across the eastern Pacific barrier. Proc. R. Soc. B.-Biol. Sci. 273: 2201-2208. Lirman, D. & D. Manzello. 2009. Patterns of resistance and resilience of the stress-tolerant coral Siderastrea radians (Pallas) to sub-optimal salinity and sediment burial. J. Exp. Mar. Biol. Ecol. 369: 72-77. Magalon, H., M. Adjeroud & M. Veuille. 2005. Patterns of genetic variation do not correlate with geographical distance in the reef-building coral Pocillopora meandrina in the South Pacific. Mol. Ecol. 14: 1861-1868. Maina, J., T.R. Mcclanahan, V. Venus, M. Ateweberhan & J. Madin. 2011. Global gradients of coral exposure to environmental stresses and implications for local management. PLoS One 6: e23064. McCartney, M.A., G. Keller & H.A. Lessios. 2000. Dispersal barriers in tropical oceans and speciation of Atlantic and eastern Pacific Echinometra sea urchins. Mol. Ecol. 9: 1391- 1400. McPhaden, M.J. 1999. Genesis and evolution of the 1997-98 El Niño. Science 283: 950-954. Meirmans, P.G. 2006. Using the AMOVA framework to estimate a standardized genetic differentiation measure. Evolution 60: 2399-2402.

190 Meyer, C.P., J.B. Geller & G. Paulay. 2005. Fine scale endemism on coral reefs: Archipelagic differentiation in turbinid gastropods. Evolution 59: 113-125. Miller, K.J. & D.J. Ayre. 2004. The role of sexual and asexual reproduction in structuring high latitude populations of the reef coral Pocillopora damicornis. Heredity 92: 557-568. Nei, M. 1987. Molecular Evolutionary Genetics. Columbia University Press, New York. Nishida, M. & J.S. Lucas. 1988. Genetic-Differences between Geographic Populations of the Crown-of-Thorns Starfish Throughout the Pacific Region. Mar. Biol. 98: 359-368. Pinzón, J.H. & T.C. LaJeunesse. 2011. Species delimitation of common reef corals in the genus Pocillopora using nucleotide sequence phylogenies, population genetics and symbiosis ecology. Mol. Ecol. 20: 311-325. Planes, S. & C. Fauvelot. 2002. Isolation by distance and vicariance drive genetic structure of a coral reef fish in the Pacific Ocean. Evolution 56: 378-399. Polato, N.R., G.T. Concepcion, R.J. Toonen & I.B. Baums. 2010. Isolation by distance across the Hawaiian Archipelago in the reef-building coral Porites lobata. Mol. Ecol. 19: 4661- 4677. Potts, D.C., T.J. Done, P.J. Isdale & D.A. Fisk. 1985. Dominance of a coral community by the genus Porites (Scleractinia). Mar. Ecol. Prog. Ser. 23: 79-84. Pritchard, J.K., M. Stephens & P. Donnelly. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945-959. Rajagopalan, B., U. Lall & M.A. Cane. 1997. Anomalous ENSO occurrences: An alternate view. J. Climate 10: 2351-2357. Raymond, M. & F. Rousset. 1995. GENEPOP (Version-1.2) population genetics software for exact tests and ecumenicism. J. Hered. 86: 248-249. Richmond, R.H. 1985. Origin and evolution of Eastern Pacific coral reefs. Am. Zool. 25: A90- A90. Richmond, R.H. 1987. Energetics, competence, and long-distance dispersal of planula larvae of the coral Pocillopora damicornis. Mar. Biol. 93: 527-533. Riginos, C. & M.W. Nachman. 2001. Population subdivision in marine environments: the contributions of biogeography, geographical distance and discontinuous habitat to genetic differentiation in a blennioid fish, Axoclinus nigricaudus. Mol. Ecol. 10: 1439-1453. Rosenberg, N.A. 2004. DISTRUCT: a program for the graphical display of population structure. Mol. Ecol. Notes 4: 137-138. Rosenblatt, R.H. & R.S. Waples. 1986. A genetic comparison of allopatric populations of shore fish species from the Eastern and Central Pacific Ocean - dispersal or vicariance. Copeia 275-284. Saarman, N.P., K.D. Louie & H. Hamilton. 2010. Genetic differentiation across eastern Pacific oceanographic barriers in the threatened seahorse Hippocampus ingens. Conserv. Genetics 11: 1989-2000. Sagarin, R.D. & S.D. Gaines. 2002. The 'abundant centre' distribution: to what extent is it a biogeographical rule? Ecol. Lett. 5: 137-147.

191 Sandin, S.A., J.E. Smith, E.E. Demartini, E.A. Dinsdale, S.D. Donner, A.M. Friedlander, T. Konotchick, M. Malay, J.E. Maragos, D. Obura, O. Pantos, G. Paulay, M. Richie, F. Rohwer, R.E. Schroeder, S. Walsh, J.B.C. Jackson, N. Knowlton & E. Sala. 2008. Baselines and degradation of coral reefs in the Northern Line Islands. PLoS One 3: e1548. Strimmer, K. 2008. FDRTOOL: a versatile R package for estimating local and tail area-based false discovery rates. Bioinformatic 24: 1461-1462. Trenberth, K.E. & T.J. Hoar. 1996. The 1990-1995 El Niño Southern Oscillation event: Longest on record. Gephys. Res. Lett. 23: 57-60. Underwood, J.N., L.D. Smith, M.J.H. Van Oppen & J.P. Gilmour. 2007. Multiple scales of genetic connectivity in a brooding coral on isolated reefs following catastrophic bleaching. Mol. Ecol. 16: 771-784. Van Oosterhout, C., D. Weetman & W.F. Hutchinson. 2006. Estimation and adjustment of microsatellite null alleles in nonequilibrium populations. Mol. Ecol. Notes 6: 255-256. Veron, J.E.N. 1995. Corals in Space and Time. The Biogeography and Evolution of the Scleractinia. Cornell University Press, Ithaca, London. Wyrtki, K. 1975. El Niño - Dynamic response of the equatorial Pacific Ocean to atmospheric forcing. J. Phys. Oceanogr. 5: 572-584. Wyrtki, K., E. Firing, D. Halpern, R. Knox, G.J. Mcnally, W.C. Patzert, E.D. Stroup, B.A. Taft & R. Williams. 1981. The Hawaii to Tahiti shuttle experiment. Science 211: 22-28.

192 Table E-1. Porites lobata samples (n = 1264) were obtained from three regions (CP = Central Pacific West (W) and East (E), HI = Hawaii, ETP = Eastern Tropical Pacific) and 33 sites. Sites are arranged in approximately west-to-east and north-to-south order. Given are total sample size (N), the number of unique multilocus genotypes (genets; Ng = 139), and the ratio of genets over samples collected (Ng/ Ng).. GPS locations are in decimal degrees (WGS84). Region Subregion Site Site Name N Ng Ng/N GPS N GPS W CP (W) Indonesia IN01 Kalimantan1 20 20 1.00 -1.10612 114.1439 Marshalls MS01 Kwajalein 30 30 1.00 9.200792 167.4228 MS02 Majuro 20 19 0.95 7.115578 171.184 Fiji FI01 Fiji 33 25 0.76 -16.5782 179.4144 Samoa SA01 American Samoa 9 9 1.00 NA NA SA02 Ofu/Olosega 78 69 0.88 -14.1528 -169.647 SA03 Tutuila 46 41 0.89 -14.2928 -170.699 Phoenix Islands PH01 Enderbury 23 22 0.96 -3.13264 -171.089 Hawaii North HN01 Hawaii North2 84 84 1.00 28.14504 -177.006 HI Hawaii Central HC01 Hawaii Central3 149 140 0.94 23.74846 -166.158 Hawaii Middle HM01 Hawaii Main4 53 50 0.94 20.82216 -156.341 CP (E) Johnston Atoll JO01 Johnston Atoll 58 56 0.97 16.74463 -169.526 Line Islands LN01 Kingman Reef 22 22 1.00 6.396564 -162.416 LN02 Palmyra 19 19 1.00 5.881678 -162.085 LN03 Teraina 10 10 1.00 4.683889 -160.38 LN04 Tabuaeran 7 6 0.86 3.867286 -159.324 LN05 Christmas 49 49 1.00 1.982039 -157.265 LN06 Jarvis 12 12 1.00 -0.37941 -160.015 Moorea MO01 Moorea 50 50 1.00 -17.5261 -149.818 Marquesas MQ01 Hiva Oa 22 22 1.00 -9.76595 -139.008 MQ02 Motane 82 81 0.99 -9.98589 -138.829 ETP Galapágos CL01 Clipperton 5 5 1.00 10.29989 -109.216 GA01 Darwin 46 45 0.98 1.616525 -91.9733 GA02 Wolf 45 44 0.98 1.336935 -91.806 GA03 Marchena 38 36 0.95 0.318369 -90.4691 GA04 Southern Galapágos5 14 14 1.00 -0.72846 -90.059 Costa Rica CR01 Marino Ballena 32 26 0.81 9.104583 -83.7068 CR02 Cano 79 64 0.81 8.71067 -83.8911 CR03 Drake Bay 8 1 0.13 8.6713 -83.7267 CR04 Gulfo Dulce 29 28 0.97 8.727433 -83.3863 CR05 Cocos Island 55 52 0.95 5.534834 -87.0875 Panama PA01 Panama6 17 17 1.00 8.223265 -80.3817 Ecuador EC01 LaLlorona 20 5 0.25 1.476383 -80.7937 Total 1264 1173 Mean 38.30 35.55 0.91 SD 30.39 29.10 0.20

1 Krakatau, Bankga, Lembeh Strait, Komodo, Bali 2 Samples from Polato et al. 2010: Pearl and Hermes, Midway, Kure 3 Samples from Polato et al. 2010: Maro, Necker, French Frigate Shoals, Nihoa, Gardener Pinnacles 4 Samples from Polato et al. 2010: Oahu, Hawai'i 5 Santiago, Baltra, Champion 6 Uva, Contadora, Coibita

193 Table E-2. Summary of per locus statistics based on 12 microsatellite markers for Porites lobata. Na = number of alleles, Neff = number of effective alleles, Ho = observed heterozygosity, Hs = heterozygosity within populations, Ht = total heterozygosity, H’t = corrected total heterozygosity, Gis = inbreeding coefficient, Gst = fixation index, G’st (Nei) = Nei’s corrected fixation index (Nei 1987), all values are significant at p < 0.01, G’st (Hed) = Hedrick’s corrected fixation index (Hedrick & Goodnight 2005), Dest = Jost’s differentiation index (Jost 2008). SE = Standard errors obtained through jackknifing over loci. All values calculated with GENODIVE (Meirmans 2006).

Locus Na Neff Ho Hs Ht H't Gis Gst G'st(Nei) G'st(Hed) Dest PL 0340 18 2.562 0.253 0.657 0.726 0.728 0.614 0.095 0.097 0.282 0.207 PL 0780 18 3.035 0.636 0.702 0.828 0.832 0.094 0.152 0.156 0.521 0.435 PL 0905 27 4.469 0.624 0.819 0.907 0.910 0.237 0.097 0.100 0.550 0.501 PL 1357 31 3.211 0.534 0.726 0.845 0.849 0.264 0.141 0.145 0.526 0.449 PL 1370 21 2.531 0.508 0.637 0.774 0.778 0.203 0.177 0.182 0.498 0.390 PL 1483 19 2.933 0.446 0.699 0.799 0.802 0.361 0.126 0.129 0.426 0.344 PL 1551 11 2.569 0.513 0.642 0.762 0.765 0.201 0.157 0.161 0.447 0.344 PL 1556 18 2.049 0.211 0.551 0.716 0.721 0.618 0.230 0.236 0.521 0.378 PL 1629 10 2.046 0.494 0.535 0.606 0.608 0.076 0.118 0.121 0.258 0.158 PL 1868 18 3.311 0.414 0.742 0.804 0.806 0.442 0.078 0.080 0.307 0.249 PL 2069 10 2.937 0.52 0.696 0.804 0.808 0.252 0.135 0.138 0.453 0.368 PL 2258 28 2.595 0.474 0.651 0.814 0.819 0.272 0.200 0.205 0.584 0.480 Mean 19.08 2.85 0.47 0.67 0.78 0.79 0.30 0.14 0.14 0.44 0.35 SE 1.99 0.19 0.04 0.02 0.02 0.02 0.05 0.013 0.01 0.03 0.03

194 Table E-3. Population differentiation among 32 sites (A) and 3 biogeographic regions (B) of Porites lobata. Based on an Analysis of Molecular Variance (AMOVA) calculated assuming an infinite allele model (equivalent to Fst). CR04 was excluded due to low sample size (Ng = 1). SE = Standard Error. F’ is a standardized version of Fst (Meirmans 2006). A) Source of Variation Nested in %var F-stat F-value SE p-value F'-value Within Individual -- 0.612 F_it 0.388 0.032 -- -- Among Individual Population 0.231 F_is 0.274 0.037 0.001 -- Among Population 0.157 F_st 0.157 0.018 0.001 0.459

Source of Variation Nested in %var F-stat F-value SE p-value F'-value Within Individual -- 0.583 F_it 0.417 0.041 -- -- Among Individual Population 0.219 F_is 0.274 0.041 0.001 -- Among Population Region 0.054 F_sc 0.063 0.007 0.001 0.197 Among Region -- 0.145 F_ct 0.145 0.020 0.001 0.464

195

196

197

Figure E-3. Porites lobata mean allelic richness (AR(20)) and private allele richness (AP(20)) rarefied to a sample size of three sites per region and 20 genes per site. A Kruskal-Wallis one- way analysis of variance indicated differences across regions (p < 0.01) for both AR(20) and AP(20). CP = Central Pacific, HI = Hawaiian Islands, ETP = Eastern Tropical Pacific.

198

199

200

Figure E-6. Mean log-likelihood (A) and Delta K (B) values of K for STRUCTURE analysis of Porites lobata samples pacific-wide.

201

202

Appendix F

SUPPLEMENTAL MATERIAL FOR APPENDIX E

Table F-1. Microsatellite loci for Porites lobata. The primer sequences are preceded by the name of fluorescent dye used (6FAM, VIC, NED or PET; Applied Biosystems, CA). The type of repeat and the size of the polymerase chain reaction (PCR) product is given (in basepairs, bp). Loci were amplified in four multiplex and one singleplex reaction (Plex). Annealing Locus Primer Sequence Size Reference Repeat Temp Plex Name (bp) (°C) (ATCC)6 ATT Polato et F: 6FAM-GTTTGCCTCTCTTCTGTTCATT 216- PL0340 (CGTT)4 TGTT 52 A al., 2010 275 R: AACATTATGGCTAGTTCTTTGAACG (CATT)3 Polato et F: VIC-GCCAGTAGGTGGATACACTGTT 136- PL0780 (ATT)4 (GTT)7 52 A al., 2010 R: CAAGTACGTTGACGTCGTTG 163 Current F: NED-GGTCCAAAGTCCACCATCA (ATC)9 ACC 126- PL0905 52 A study R: TGGTGGAAATAAGTGGTCGA (ATC)9 183 Polato et F:PET- ATGTCCCTGAAACGGAAGTA (ACC)7…(ATC)4 252- PL1357 52 D al., 2010 R: GATGATGATGTTGTTGATGGTG …(ACC)7 300 Current F: PET-GCACTGTCTGTAACAAGCGAA 189- PL1370 (GTT)8 54 E study R: CATATTGGAAGGAGGGCTC 246 Current F: 6FAM-AAACGTTCCCTATCCCATCC 143- PL1483 (GTT)10 54 E study R: GCAAAGCTGCTACATTTCACTAA 173 Polato et F: PET-TGTTTCTGAGTGGCTGTGCT 178- PL1551 (GTT)8 52 A al., 2010 R: GGTTGGAAAGGGTCCTTCAT 196 Polato et F: PET-CGTTGACGTAACCTTCACCA 153- PL1556 (ATC)10 56 B al., 2010 R: CACAGGGTAACCTTCCTTGC 168 Polato et F: 6FAM-CCTTGGTTAATTTGCCCTTG 168- PL1629 (GCT)8 52 D al., 2010 R: ACCAGTCCGGAGTCAAGCTA 180 Current F: VIC-TAAGCCACAGCAGGTGTACG 179- PL1868 (AAC)10 52 D study R: AAACGTTCCCTATCCCATCC 206 Polato et F: PET-CGCAGTTCCTTTGATTTGGT 249- PL2069 (GTT)8 52 C al., 2010 R: GTTTCTTTAGCGGTTGATGGCTTGTTAC 267 Polato et F: NED-ATTAGCGGATGAAGCGAAGA 217- PL2258 (GAT)10 56 B al., 2010 R: TCCAATGTAACGCCAAATCA 250

203

0.25 0.08 0.11 0.16 0.10 0.22 0.51 0.21 0.10 0.13 0.38 0.22 0.06 0.08 0.18 0.14 0.12 0.09 0.14 0.11 0.07 0.07 0.13 0.11 0.09 0.04 0.05 0.11 0.12 0.08 0.06 0.11

Upper CI Upper

0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Lower CI Lower

0.05 0.01 0.02 0.03 0.02 0.05 0.21 0.05 0.02 0.03 0.12 0.04 0.01 0.01 0.04 0.02 0.02 0.01 0.03 0.02 0.01 0.01 0.02 0.02 0.02 0.01 0.01 0.02 0.02 0.01 0.01 0.02

Fi Mean

0.24 0.08 0.12 0.07 0.19 0.22 0.14 0.12 0.11 0.08 0.11 0.13 0.27 0.18 0.13 0.18 0.14 0.21 0.14 0.15 0.15 0.09 0.21 0.09 0.02 0.03 0.05 0.23 0.04 0.04 0.03 0.07

0905

0.06 0.03 0.05 0.09 0.16 0.09 0.06 0.15 0.16 0.05 0.05 0.14 0.12 0.04 0.09 0.12 0.04 0.06 0.13 0.08 0.04 0.05 0.09 0.27 0.02 0.02 0.02 0.22 0.08 0.03 0.05 0.07

0780

0.32 0.33 0.06 0.23 0.28 0.20 0.36 0.36 0.25 0.29 0.21 0.38 0.04 0.29 0.27 0.27 0.28 0.40 0.14 0.35 0.39 0.08 0.11 0.27 0.26 0.28 0.29 0.00 0.25 0.24 0.16 0.19

0340

.08

0.27 0.18 0.06 0.33 0.22 0.16 0.17 0.21 0.14 0.19 0.15 0.19 0.05 0.22 0.18 0.10 0.09 0.13 0.17 0 0.13 0.06 0.08 0.00 0.11 0.09 0.22 0.66 0.09 0.05 0.07 0.10

2258

0.17 0.03 0.07 0.31 0.19 0.20 0.16 0.10 0.38 0.18 0.20 0.13 0.03 0.07 0.09 0.23 0.08 0.07 0.08 0.24 0.26 0.06 0.15 0.00 0.19 0.15 0.07 0.31 0.05 0.03 0.07 0.19

2069

.24

0.18 0.30 0.13 0.24 0.22 0.20 0.17 0.23 0.11 0.13 0.18 0.19 0.07 0.12 0.25 0.30 0 0.22 0.16 0.14 0.11 0.14 0.21 0.00 0.30 0.26 0.15 0.11 0.16 0.16 0.20 0.38

1868

0.11 0.05 0.03 0.12 0.10 0.24 0.08 0.25 0.09 0.06 0.06 0.10 0.07 0.14 0.14 0.20 0.09 0.15 0.08 0.14 0.04 0.03 0.06 0.00 0.13 0.03 0.06 0.36 0.02 0.03 0.02 0.07

1629

.38

0.09 0.28 0.06 0.32 0.27 0.33 0.15 0.09 0.22 0.23 0.10 0.40 0.31 0 0.41 0.36 0.30 0.27 0.27 0.19 0.18 0.04 0.09 0.00 0.32 0.29 0.31 0.00 0.07 0.19 0.11 0.35

1556

0.27 0.06 0.03 0.10 0.28 0.15 0.07 0.03 0.16 0.06 0.08 0.07 0.06 0.03 0.09 0.18 0.11 0.17 0.11 0.13 0.13 0.03 0.04 0.00 0.07 0.12 0.11 0.16 0.14 0.07 0.10 0.13

1551

0.21 0.07 0.06 0.25 0.23 0.23 0.09 0.14 0.11 0.11 0.09 0.17 0.06 0.07 0.27 0.28 0.06 0.29 0.19 0.06 0.08 0.07 0.17 0.00 0.25 0.26 0.13 0.02 0.17 0.17 0.15 0.24

1483

null null allele frequencies and individual inbreeding values. Given are averages over individuals

0.12 0.19 0.20 0.26 0.16 0.05 0.07 0.08 0.08 0.07 0.23 0.12 0.06 0.08 0.13 0.10 0.06 0.06 0.02 0.04 0.02 0.21 0.12 0.00 0.15 0.26 0.11 0.46 0.13 0.13 0.14 0.20

1370

0.11 0.02 0.02 0.19 0.41 0.09 0.07 0.17 0.29 0.10 0.20 0.13 0.07 0.12 0.23 0.16 0.09 0.23 0.08 0.06 0.07 0.03 0.05 0.00 0.11 0.10 0.18 0.04 0.08 0.05 0.18 0.21

Mean Null Allele Frequency Per Locus Frequency Allele Null Mean 1357

Porites Porites lobata

MS02 JO01 HC01 GA04 EC01 Site IN01 MS01 FI01 SA01 SA02 SA03 PH01 LN01 LN02 LN03 LN04 LN05 LN06 MO01 MQ01 MQ02 HN01 HM01 CL01 GA01 GA02 GA03 CR01 CR02 CR03 CR04 PA01

I of Fi was different from zero. from different was Fi I of able able F

C T and site per the (CI). MS01upper 0.05 lower the 0.95intervals SA03 and and sites were Confidence only the Lowerwere Region CP HI ETP

204

0.00 0.02 0.00 0.13 0.08 0.18 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.11 0.14 0.11 0.12 0.16 0.10 0.13 0.14 0.18 0.20 0.19

LN0

0.00 0.00 0.00 0.12 0.08 0.13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.11 0.12 0.08 0.11 0.13 0.10 0.10 0.12 0.16 0.15 0.21

LN05

CR03 had onlyCR03 had

0.13 0.30 0.37 0.08 0.06 0.07 0.20 0.00 0.01 0.02 0.14 0.01 0.00 0.00 0.0 0.00 0.03 0.46 0.06 0.10 0.05 0.09 0.11 0.08 0.07 0.11 0.06 0.06 0.10 0.08 0.22

LN04

0.06 0.30 0.03 0.10 0.09 0.13 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.05 0.06 0.11 0.07 0.12 0.15 0.14 0.12 0.13 0.17 0.17 0.22

LN03

sampling sites. sampling

0.00 0.00 0.09 0.09 0.14 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.06 0.07 0.07 0.10 0.12 0.12 0.12 0.11 0.11 0.12 0.15 0.15 0.19

LN02

P.lobata

0.00 0.00 0.03 0.15 0.15 0.18 0.00 0. 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.07 0.03 0.15 0.16 0.09 0.15 0.20 0.17 0.17 0.18 0.16 0.15 0.19 0.19 0.25

LN01

0.00 0.00 0.11 0.13 0.12 0.19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.17 0.11 0.13 0.09 0.09 0.13 0.14 0.19 0.15 0.18 0.13 0.16 0.16 0.20 0.20 0.22

JO01

0.00 0.00 0.31 0.29 0.34 0.36 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.20 0.32 0.30 0.33 0.31 0.35 0.26 0.27 0.41 0.32 0.34 0.34 0.35 0.31 0.35 0.39 0.44

HM01

0.00 0.00 0.32 0.30 0.34 0.36 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.21 0.34 0.31 0.33 0.33 0.35 0.26 0.29 0.41 0.32 0.35 0.35 0.35 0.32 0.35 0.39 0.42

HC01

0.00 0.00 0.29 0.28 0.31 0.33 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.20 0.31 0.29 0.30 0.30 0.32 0.24 0.27 0.38 0.30 0.32 0.32 0.32 0.30 0.33 0.36 0.40

HN01

.00 .13

0.00 0.04 0.06 0.06 0.10 0.04 0.00 0.00 0.00 0 0.00 0.26 0.29 0.28 0.10 0.09 0.01 0.01 0.04 0.04 0.04 0.07 0.04 0.09 0.09 0.09 0.08 0.10 0.12 0 0.16

PH01

0.00 0.03 0.05 0.08 0.10 0.15 0.00 0.00 0.00 0.00 0.00 0.22 0.24 0.23 0.11 0.10 0.05 0.04 0.09 0.09 0.07 0.09 0.08 0.13 0.13 0.13 0.12 0.14 0.16 0.17 0.17

SA03

0.00 0.02 0.03 0.06 0.09 0.12 0.03 0.00 0.00 0.00 0.03 0.22 0.25 0.23 0.11 0.07 0.04 0.02 0.08 0.09 0.05 0.07 0.09 0.10 0.10 0.1 0.10 0.11 0.13 0.13 0.18

SA02

0.03 0.06 0.10 0.11 0.13 0.15 0.00 0.00 0.00 0.06 0.07 0.30 0.32 0.33 0.16 0.14 0.09 0.09 0.14 0.14 0.06 0.13 0.19 0.14 0.12 0.14 0.12 0.13 0.16 0.18 0.18

SA01

0.00 0.06 0.08 0.09 0.15 0.18 0.00 0.00 0.08 0.04 0.03 0.28 0.30 0.29 0.16 0.12 0.07 0.06 0.13 0.13 0.08 0.10 0.13 0.17 0.16 0.18 0.15 0.16 0.19 0.19 0.22

FI01

values (lower diagonal) and their significance (upper diagonal) among among (upper diagonal) their and significance (lower diagonal) values

0.00 0.13 0.15 0.20 0.21 0.25 0.00 0.11 0.13 0.12 0.08 0.35 0.37 0.37 0.24 0.19 0.14 0.15 0.20 0.19 0.15 0.21 0.19 0.22 0.23 0.23 0.21 0.22 0.25 0.26 0.25

MS02

0.03 0.04 0.09 0.09 0.13 0.00 0.05 0.05 0.04 0.03 0.03 0.2 0.28 0.27 0.11 0.07 0.02 0.02 0.07 0.07 0.06 0.11 0.09 0.11 0.12 0.11 0.11 0.12 0.15 0.15 0.16

MS01

Pairwise F Pairwise

0.04 0.02 0.03 0.05 0.11 0.14 0.15 0.04 0.08 0.01 0.03 0.26 0.29 0.27 0.13 0.07 0.03 0.01 0.08 0.10 0.05 0.05 0.09 0.13 0.13 0.14 0.12 0.13 0.17 0.17 0.20

IN01

Q01

able able

R02

T excluded. and therefore sample was one MS01 PH01 LN02 M GA04 PA01 IN01 MS02 FI01 SA01 SA02 SA03 HN01 HC01 HM01 JO01 LN01 LN03 LN04 LN05 LN06 MO01 MQ02 CL01 GA01 GA02 GA03 CR01 C CR03 CR04 CR05 EQ01

205

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00

EQ01

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.13 0.04 0.00 0.20 0.04 0.03 0.02 0.21

PA01

0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.10 0.00 0.00 0.21

CR05

0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.08 0.04 0.03 0.17

CR04

CR03

0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.23 0.01 0.02 0.16

CR02

0.00 0.00 0.00 0.00 0. 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.05 0.15 0.00 0.00 0.01 0.01 0.15

CR01

0.00 0.00 0.00 0.00 0.06 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.22 0.00 0.04 0.03 0.06 0.06 0.09 0.15

GA04

.00

0.00 0.00 0.00 0.00 0.03 0.08 0.00 0.00 0.00 0.00 0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.06 0.08 0.09 0.13

GA03

0.00 0.00 0.00 0.00 0.05 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.05 0.05 0.01 0.02 0.01 0.03 0.15

GA02

0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.03 0.01 0.02 0.03 0.05 0.15

GA01

0.00 0.05 0.00 0.00 0.09 0.18 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.02 0.01 0.00 0.00 0.00 0.00 0.13 0.16 0.15 0.14 0.17 0.19 0.21 0.30

CL01

0.00 0.00 0.00 0.94 0.14 0.17 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.15 0.15 0.14 0.14 0.15 0.15 0.18 0.18 0.22

MQ02

. .

.00

0.01

0.00 0.00 0.00 0.13 0.15 0.00 0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.15 0.13 0.12 0.13 0.14 0.13 0.16 0.17 0.22

’d

MQ01

. cont

0.00 0.00 0.00 0.07 0.12 0.13 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.09 0.15 0.13 0.13 0.14 0.13 0.13 0.16 0.16 0.19

MO01

able F able

T MS01 PH01 LN02 MQ01 GA04 PA01 IN01 MS02 FI01 SA01 SA02 SA03 HN01 HC01 HM01 JO01 LN01 LN03 LN04 LN05 LN06 MO01 MQ02 CL01 GA01 GA02 GA03 CR01 CR02 CR03 CR04 CR05 EQ01

206

207

208

Figure F-3. Mean log-likelihood (A) and Delta K (B) values of K for STRUCTURE analysis of Porites lobata samples Pacific-wide using only eight loci.

209

210

VITA Jennifer N. Boulay

Education

The Pennsylvania State University State College, Pennsylvania 2009-2014 Ph. D. in Biology Advisor: Dr. Iliana Baums

University of Miami Coral Gables, Florida 2005-2009 B.S. in Biology and Marine Science with minor in Chemistry Magna Cum Laude Honors Thesis Advisor: Dr. Su Sponaugle

Professional Experience

Research Assistant The Pennsylvania State University 2009-2014 Dr. Iliana Baums’ Molecular Ecology Lab

Teaching Assistant The Pennsylvania State University BIOL 110 - Biology: Basic Concepts and Biodiversity Lab Fall 2010/2011/2012 BIOL 499A - Tropical Field Biology Dec. 2010/2011/2012 BIOL 482 - Coastal Biology Spring 2013/2014

Publications

Boulay, JN, Cortés, J, Hellberg, M, Baums, IB, (2014) Unrecognized coral species diversity masks differences in functional ecology. Proc. R. Soc. B. 281: 20131580

Baums IB, Devlin-Durante MK, Polato NR, Xu D, Giri S, Altman NS, Ruiz D, Parkinson JE, Boulay JN, (2013) Genotypic variation influences reproductive success and thermal stress tolerance in the reef building coral, Acropora palmata. Coral Reefs. 1-15.

Mangubhai, S, Rotjan, R, Obura, D, McMahon, K, Lohmann, P, Hodges, B, Shamberger, K, Boulay, JN (2012) Phoenix Islands Protected Area Assessment 2012 Preliminary Expedition Report.

Boulay JN, Cortés, J, Nivia-Riuz, J, Baums, IB, (2012) High genotypic diversity of the reef-building coral Porites lobata (Dana, 1846) (Scleractinia: Poritidae) at Coco Island National Park, Costa Rica. Rev. Biol. Trop. 60 (Suppl. 3): 279-292.

Baums, IB, Boulay, JN, Polato, NR, Hellberg, M, (2012) No gene flow across the Eastern Pacific Barrier in the reef-building coral Porites lobata. Mol. Ecol. 21: 5418- 5433

Sponaugle, S, Boulay, JN, Rankin, T, (2011) Growth- and size-selective mortality in pelagic larvae of a common reef fish. Aquatic Biology. 13(3), 263-273.

Selected Honors and Awards

Jeanette Ritter Mohnkern Graduate Student Scholarship in Biology, PSU 2013-2014 NSF GRFP Honorable Mention April 2011 Braddock Supplement, Penn State University, Department of Biology 2009-2010